5GNR:TheNextGenerationWirelessAccessTechnology

ErikDahlman

StefanParkvall

JohanSköld

TableofContents

Coverimage

Titlepage

Copyright

Preface

Acknowledgments

AbbreviationsandAcronyms

Chapter1.WhatIs5G?

Abstract

1.13GPPandtheStandardizationofMobileCommunication

1.2TheNextGeneration—5G/NR

Chapter2.5GStandardization

Abstract

2.1OverviewofStandardizationandRegulation

2.2ITU-RActivitiesFrom3Gto5G

2.35GandIMT-2020

2.43GPPStandardization

Chapter3.Spectrumfor5G

Abstract

3.1SpectrumforMobileSystems

3.2FrequencyBandsforNR

3.3RFExposureAbove6GHz

Chapter4.LTE—AnOverview

Abstract

4.1LTERelease8—BasicRadioAccess

4.2LTEEvolution

4.3SpectrumFlexibility

4.4Multi-AntennaEnhancements

4.5Densification,SmallCells,andHeterogeneousDeployments

4.6DeviceEnhancements

4.7NewScenarios

Chapter5.NROverview

Abstract

5.1Higher-FrequencyOperationandSpectrumFlexibility

5.2Ultra-LeanDesign

5.3ForwardCompatibility

5.4TransmissionScheme,BandwidthParts,andFrameStructure

5.5DuplexSchemes

5.6Low-LatencySupport

5.7SchedulingandDataTransmission

5.8ControlChannels

5.9Beam-CentricDesignandMulti-AntennaTransmission

5.10InitialAccess

5.11InterworkingandLTECoexistence

Chapter6.Radio-InterfaceArchitecture

Abstract

6.1OverallSystemArchitecture

6.2Quality-Of-ServiceHandling

6.3RadioProtocolArchitecture

6.4User-PlaneProtocols

6.5Control-PlaneProtocols

Chapter7.OverallTransmissionStructure

Abstract

7.1TransmissionScheme

7.2Time-DomainStructure

7.3Frequency-DomainStructure

7.4BandwidthParts

7.5Frequency-DomainLocationofNRCarriers

7.6CarrierAggregation

7.7SupplementaryUplink

7.8DuplexSchemes

7.9AntennaPorts

7.10Quasi-Colocation

Chapter8.ChannelSounding

Abstract

8.1DownlinkChannelSounding—CSI-RS

8.2DownlinkMeasurementsandReporting

8.3UplinkChannelSounding—SRS

Chapter9.Transport-ChannelProcessing

Abstract

9.1Overview

9.2ChannelCoding

9.3RateMatchingandPhysical-LayerHybrid-ARQFunctionality

9.4Scrambling

9.5Modulation

9.6LayerMapping

9.7UplinkDFTPrecoding

9.8Multi-AntennaPrecoding

9.9ResourceMapping

9.10DownlinkReservedResources

9.11ReferenceSignals

Chapter10.Physical-LayerControlSignaling

Abstract

10.1Downlink

10.2Uplink

Chapter11.Multi-AntennaTransmission

Abstract

11.1Introduction

11.2DownlinkMulti-AntennaPrecoding

11.3NRUplinkMultiantennaPrecoding

Chapter12.BeamManagement

Abstract

12.1InitialBeamEstablishment

12.2BeamAdjustment

12.3BeamRecovery

Chapter13.RetransmissionProtocols

Abstract

13.1Hybrid-ARQWithSoftCombining

13.2RLC

13.3PDCP

Chapter14.Scheduling

Abstract

14.1DynamicDownlinkScheduling

14.2DynamicUplinkScheduling

14.3SchedulingandDynamicTDD

14.4TransmissionWithoutaDynamicGrant

14.5DiscontinuousReception

Chapter15.UplinkPowerandTimingControl

Abstract

15.1UplinkPowerControl

15.2UplinkTimingControl

Chapter16.InitialAccess

Abstract

16.1CellSearch

16.2RandomAccess

Chapter17.LTE/NRInterworkingandCoexistence

Abstract

17.1LTE/NRDual-Connectivity

17.2LTE/NRCoexistence

Chapter18.RFCharacteristics

Abstract

18.1SpectrumFlexibilityImplications

18.2RFRequirementsinDifferentFrequencyRanges

18.3ChannelBandwidthandSpectrumUtilization

18.4OverallStructureofDeviceRFRequirements

18.5OverallStructureofBase-StationRFRequirements

18.6OverviewofConductedRFRequirementsforNR

18.7ConductedOutputPowerLevelRequirements

18.8TransmittedSignalQuality

18.9ConductedUnwantedEmissionsRequirements

18.10ConductedSensitivityandDynamicRange

18.11ReceiverSusceptibilitytoInterferingSignals

18.12RadiatedRFRequirementsforNR

18.13OngoingDevelopmentsofRFRequirementsforNR

Chapter19.RFTechnologiesatmm-WaveFrequencies

Abstract

19.1ADCandDACConsiderations

19.2LOGenerationandPhaseNoiseAspects

19.3PowerAmplifierEfficiencyinRelationtoUnwantedEmission

19.4FilteringAspects

19.5ReceiverNoiseFigure,DynamicRange,andBandwidthDependencies

19.6Summary

Chapter20.BeyondtheFirstReleaseof5G

Abstract

20.1IntegratedAccess-Backhaul

20.2OperationinUnlicensedSpectra

20.3Non-orthogonalMultipleAccess

20.4Machine-TypeCommunication

20.5Device-To-DeviceCommunication

20.6SpectrumandDuplexFlexibility

20.7ConcludingRemarks

References

Index

Copyright

AcademicPressisanimprintofElsevier125LondonWall,LondonEC2Y5AS,UnitedKingdom525BStreet,Suite1650,SanDiego,CA92101,UnitedStates50HampshireStreet,5thFloor,Cambridge,MA02139,UnitedStatesTheBoulevard,LangfordLane,Kidlington,OxfordOX51GB,UnitedKingdomCopyright©2018ElsevierLtd.Allrightsreserved.

Nopartofthispublicationmaybereproducedortransmittedinanyformorbyanymeans,electronicormechanical,includingphotocopying,recording,oranyinformationstorageandretrievalsystem,withoutpermissioninwritingfromthepublisher.Detailsonhowtoseekpermission,furtherinformationaboutthePublisher’spermissionspoliciesandourarrangementswithorganizationssuchastheCopyrightClearanceCenterandtheCopyrightLicensingAgency,canbefoundatourwebsite:www.elsevier.com/permissions.

ThisbookandtheindividualcontributionscontainedinitareprotectedundercopyrightbythePublisher(otherthanasmaybenotedherein).

NoticesKnowledgeandbestpracticeinthisfieldareconstantlychanging.Asnewresearchandexperiencebroadenourunderstanding,changesinresearchmethods,professionalpractices,ormedicaltreatmentmaybecomenecessary.

Practitionersandresearchersmustalwaysrelyontheirownexperienceandknowledgeinevaluatingandusinganyinformation,methods,compounds,orexperimentsdescribedherein.Inusingsuchinformationormethodstheyshouldbemindfuloftheirownsafetyandthesafetyofothers,includingpartiesforwhomtheyhaveaprofessionalresponsibility.

Tothefullestextentofthelaw,neitherthePublishernortheauthors,contributors,oreditors,assumeanyliabilityforanyinjuryand/ordamageto

contributors,oreditors,assumeanyliabilityforanyinjuryand/ordamagetopersonsorpropertyasamatterofproductsliability,negligenceorotherwise,orfromanyuseoroperationofanymethods,products,instructions,orideascontainedinthematerialherein.

BritishLibraryCataloguing-in-PublicationDataAcataloguerecordforthisbookisavailablefromtheBritishLibraryLibraryofCongressCataloging-in-PublicationDataAcatalogrecordforthisbookisavailablefromtheLibraryofCongressISBN:978-0-12814323-0

ForInformationonallAcademicPresspublicationsvisitourwebsiteathttps://www.elsevier.com/books-and-journals

Publisher:MaraConnerAcquisitionEditor:TimPittsEditorialProjectManager:JoshuaMearnsProductionProjectManager:KameshRamajogiCoverDesigner:GregHarris

TypesetbyMPSLimited,Chennai,India

Preface

Long-Term Evolution (LTE) has become themost successful wirelessmobilebroadband technology across the world, serving billions of users. Mobilebroadband is, and will continue to be, an important part of future cellularcommunication,but futurewirelessnetworksare toa largeextentalsoaboutasignificantly wider range of use cases and a correspondingly wider range ofrequirements. Although LTE is a very capable technology, still evolving andexpected tobeusedformanyyears tocome,anew5GradioaccessknownasNewRadio(NR)hasbeenstandardizedtomeetfuturerequirements.ThisbookdescribesNR,developed in3GPP (ThirdGenerationPartnership

Project)asoflateSpring2018.Chapter 1 provides a brief introduction, followed by a description of the

standardization process and relevant organizations such as the aforementioned3GPP and ITU in Chapter 2. The frequency bands available for mobilecommunication are covered in Chapter 3 together with a discussion on theprocessforfindingnewfrequencybands.An overview ofLTE and its evolution is found inChapter 4.Although the

focusofthebookisNR,abriefoverviewofLTEasabackgroundtothecomingchapters is relevant. One reason is that both LTE and NR are developed by3GPP and hence have a common background and share several technologycomponents.Many of the design choices inNR are also based on experiencefromLTE.Furthermore,LTEcontinuestoevolveinparallelwithNRandisanimportantcomponentin5Gradioaccess.Chapter5providesanoverviewofNR.Itcanbereadonitsowntogetahigh-

levelunderstandingofNR,orasanintroductiontothesubsequentchapters.Chapter 6 outlines the overall protocol structure in NR, followed by a

descriptionoftheoveralltime–frequencystructureofNRinChapter7.Multiantenna processing and beamforming are integral parts of NR. The

channel sounding tools to support these functions are outlined in Chapter 8,followed by the overall transport-channel processing in Chapter 9 and the

associated control signaling in Chapter 10. How the functions are used tosupportdifferentmulti-antennaschemesandbeamformingfunctionsisthetopicofChapters11and12.RetransmissionfunctionalityandschedulingarethetopicsofChapters13and

14,followedbypowercontrolinChapter15andinitialaccessinChapter16.CoexistenceandinterworkingwithLTEisanessentialpartofNR,especially

inthenonstandaloneversionwhichreliesonLTEformobilityandinitialaccess,andiscoveredinChapter17.Radio-frequency (RF) requirements, taking into account spectrum flexibility

across largefrequencyrangesandmultistandardradioequipment,are the topicofChapter18.Chapter19discusses theRF implementationaspects forhigherfrequencybandsinthemm-waverange.Finally,Chapter20concludesthebookwithanoutlooktofutureNRreleases.

Acknowledgments

WethankallourcolleaguesatEricssonforassisting in thisprojectbyhelpingwith contributions to the book, giving suggestions and comments on thecontents,andtakingpartinthehugeteameffortofdevelopingNRandthenextgenerationofradioaccessfor5G.Thestandardizationprocess involvespeoplefromallpartsof theworld,and

weacknowledgetheeffortsofourcolleaguesinthewirelessindustryingeneraland in 3GPP RAN in particular.Without their work and contributions to thestandardization,thisbookwouldnothavebeenpossible.Finally, we are immensely grateful to our families for bearingwith us and

supportingusduringthelongprocessofwritingthisbook.

AbbreviationsandAcronyms

3GPPThirdGenerationPartnershipProject5GCN5GCoreNetworkAASActiveAntennaSystemACIRAdjacentChannelInterferenceRatioACKAcknowledgment(inARQprotocols)ACLRAdjacentChannelLeakageRatioACSAdjacentChannelSelectivityADCAnalog-to-DigitalConverterAFApplicationFunctionAGCAutomaticGainControlAMAcknowledgedMode(RLCconfiguration)AMAmplitudeModulationAMFAccessandMobilityManagementFunctionA-MPRAdditionalMaximumPowerReductionAMPSAdvancedMobilePhoneSystemARIAcknowledgmentResourceIndicatorARIBAssociationofRadioIndustriesandBusinessesARQAutomaticRepeat-reQuestASAccessStratumATISAllianceforTelecommunicationsIndustrySolutionsAUSFAuthenticationServerFunctionAWGNAdditiveWhiteGaussianNoiseBCBandCategoryBCCHBroadcastControlChannelBCHBroadcastChannelBiCMOSBipolarComplementaryMetalOxideSemiconductorBPSKBinaryPhase-ShiftKeyingBSBaseStationBWBandwidthBWPBandwidthpartCACarrieraggregationCACLRCumulativeAdjacentChannelLeakageRatioCBGCodeblockgroupCBGFICBGflushinformationCBGTICBGtransmitindicatorCCComponentCarrierCCCHCommonControlChannelCCEControlChannelElementCCSAChinaCommunicationsStandardsAssociationCDMCodeDivisionMultiplexingCDMACode-DivisionMultipleAccessCEPTEuropeanConferenceofPostalandTelecommunicationsAdministrationCITELInter-AmericanTelecommunicationCommissionC-MTCCriticalMachine-TypeCommunicationsCMOSComplementaryMetalOxideSemiconductorCNCoreNetworkCoMPCoordinatedMulti-PointTransmission/ReceptionCORESTControlresourcesetCPCyclicPrefixCPCompressionPointCQIChannel-QualityIndicatorCRBCommonresourceblockCRCCyclicRedundancyCheckC-RNTICellRadio-NetworkTemporaryIdentifierCSCapabilitySet(forMSR

basestations)CSIChannel-StateInformationCSI-IMCSIInterferenceMeasurementCSI-RSCSIReferenceSignalsCS-RNTIConfiguredschedulingRNTICWContinuousWaveD2DDevice-to-DeviceDACDigital-to-AnalogConverterDAIDownlinkAssignmentIndexD-AMPSDigitalAMPS

DCDualConnectivityDCDirectCurrentDCCHDedicatedControlChannelDCHDedicatedChannelDCIDownlinkControlInformationDFTDiscreteFourierTransformDFTS-OFDMDFT-SpreadOFDM(DFT-precodedOFDM,seealsoSC-FDMA)DLDownlinkDL-SCHDownlinkSharedChannelDM-RSDemodulationReferenceSignalDRDynamicRangeDRXDiscontinuousReceptionDTXDiscontinuousTransmissionEDGEEnhancedDataRatesforGSMEvolution,EnhancedDataRatesforGlobalEvolutionECCElectronicCommunicationsCommittee(ofCEPT)eIMTAEnhancedInterferenceMitigationandTrafficAdaptationEIRPEffectiveIsotropicRadiatedPowerEISEquivalentIsotropicSensitivityeMBBenhancedMBB

EMFElectromagneticFieldeNBeNodeBEN-DCE-UTRANRDual-ConnectivityeNodeBE-UTRANNodeBEPCEvolvedPacketCoreETSIEuropeanTelecommunicationsStandards

InstituteE-UTRAEvolvedUTRAEVMErrorVectorMagnitudeFCCFederalCommunicationsCommissionFDDFrequencyDivisionDuplexFDMFrequencyDivisionMultiplexingFETField-EffectTransistorFDMAFrequency-DivisionMultipleAccessFFTFastFourierTransformFoMFigure-of-MeritFPLMTSFuturePublicLandMobileTelecommunicationsSystemsFR1FrequencyRange1

FR2FrequencyRange2GaAsGalliumArsenideGaNGalliumNitrideGERANGSM/EDGERadio

AccessNetworkgNBgNodeBgNodeBgeneralizedNodeBGSAGlobalmobileSuppliersAssociationGSMGlobalSystemforMobile

CommunicationsGSMAGSMAssociationHARQHybridARQHBTHeterojunctionBipolarTransistorHEMTHighElectron-Mobility

TransistorHSPAHigh-SpeedPacketAccessICIntegratedCircuitICNIRPInternationalCommissiononNon-IonizingRadiationICSIn-Channel

SelectivityIEEEInstituteofElectricalandElectronicsEngineersIFFTInverseFastFourierTransformILInsertionLossIMDInterModulationDistortionIMT-2000InternationalMobileTelecommunications2000(ITU’snameforthefamilyof3Gstandards)IMT-2020InternationalMobileTelecommunications2020(ITU’snameforthefamilyof5Gstandards)IMT-AdvancedInternationalMobileTelecommunicationsAdvanced(ITU’snameforthefamilyof4Gstandards)InGaPIndiumGalliumPhosphideIOTInternetofThingsIPInternetProtocolIP33rdorderInterceptPointIRIncrementalRedundancyITRSInternationalTelecomRoadmapforSemiconductorsITUInternationalTelecommunicationsUnionITU-RInternationalTelecommunicationsUnion-RadiocommunicationsSectorKPIKeyPerformanceIndicatorL1-RSRPLayer1ReferenceSignalReceiverPowerLCInductor(L)-CapacitorLAALicense-AssistedAccessLCIDLogicalChannelIndexLDPCLow-DensityParityCheckCodeLOLocalOscillatorLNALow-NoiseAmplifierLTCCLowTemperatureCo-firedCeramicLTELong-TermEvolutionMACMediumAccessControlMAC-CEMACcontrolelementMANMetropolitanAreaNetworkMBBMobileBroadbandMB-MSRMulti-BandMultiStandardRadio(basestation)MCGMasterCellGroupMCSModulationandCodingSchemeMIBMasterInformationBlockMMICMonolithicMicrowaveIntegratedCircuitMIMOMultiple-InputMultiple-OutputmMTCmassiveMachineTypeCommunicationMPRMaximumPowerReductionMSRMulti-StandardRadioMTCMachine-TypeCommunicationMU-MIMOMulti-UserMIMO

NAKNegativeAcknowledgment(inARQprotocols)NB-IoTNarrow-BandInternet-of-ThingsNDINew-DataIndicatorNEFNetworkexposurefunctionNFNoiseFigureNGTheinterfacebetweenthegNBandthe5GCN

NG-cThecontrol-planepartofNGNGMNNextGenerationMobileNetworksNG-uTheuser-planepartofNGNMTNordiskMobilTelefon(NordicMobileTelephony)NodeBNodeB,a

logicalnodehandlingtransmission/receptioninmultiplecells.Commonly,butnotnecessarily,correspondingtoabasestationNOMANonorthogonalMultipleAccessNRNewRadioNRFNRrepositoryfunctionNSNetworkSignalingNZP-CSI-RSNon-zero-powerCSI-RS

OBUEOperatingBandUnwantedEmissionsOCCOrthogonalCoverCode

OFDMOrthogonalFrequency-DivisionMultiplexingOOBOut-Of-Band(emissions)OSDDOTASensitivityDirectionDeclarationsOTAOver-The-AirPAPowerAmplifierPAEPower-AddedEfficiencyPAPRPeak-to-AveragePowerRatioPARPeak-to-AverageRatio(sameasPAPR)PBCHPhysicalBroadcastChannelPCBPrintedCircuitBoardPCCHPagingControlChannelPCFPolicycontrolfunctionPCGProjectCoordinationGroup(in3GPP)PCHPagingChannelPCIPhysicalCellIdentityPDCPersonalDigitalCellularPDCCHPhysicalDownlinkControlChannelPDCPPacketDataConvergenceProtocolPDSCHPhysicalDownlinkSharedChannelPDUProtocolDataUnitPHSPersonalHandy-phoneSystemPHYPhysicalLayerPLLPhase-LockedLoopPMPhaseModulationPMIPrecoding-MatrixIndicatorPNPhaseNoisePRACHPhysicalRandom-AccessChannelPRBPhysicalResourceBlockP-RNTIPagingRNTIPSDPowerSpectralDensityPSSPrimarySynchronizationSignalPUCCHPhysicalUplinkControlChannelPUSCHPhysicalUplinkSharedChannelQAMQuadratureAmplitudeModulationQCLQuasiCo-LocationQoSQuality-of-ServiceQPSKQuadraturePhase-ShiftKeyingRACHRandomAccessChannelRANRadioAccessNetworkRA-RNTIRandomAccessRNTIRATRadioAccessTechnologyRBResourceBlockREResourceElementRFRadioFrequencyRFICRadioFrequencyIntegratedCircuitRIRankIndicatorRIBRadiatedInterfaceBoundaryRITRadioInterfaceTechnologyRLCRadioLinkControlRMSIRemainingMinimumSystemInformationRNTIRadio-NetworkTemporaryIdentifierRoAoARangeofAngleofArrivalROHCRobustHeaderCompressionRRCRadioResourceControlRRMRadioResourceManagementRSReferenceSymbolRSPCIMT-2000RadioInterface

SpecificationsRSRPReferenceSignalReceivedPowerRVRedundancyVersionRXReceiverSCGSecondaryCellGroupSCSSub-CarrierSpacingSDLSupplementaryDownlinkSDMASpatialDivisionMultipleAccessSDOStandardsDevelopingOrganizationSDUServiceDataUnitSEMSpectrumEmissionsMaskSFISlotformatindicatorSFI-RNTISlotformatindicatorRNTISFNSystemFrameNumber(in3GPP).

SISystemInformationMessageSIBSystemInformationBlockSIB1SystemInformationBlock1

SiGeSiliconGermaniumSINRSignal-to-Interference-and-NoiseRatioSIRSignal-to-InterferenceRatioSiPSystem-in-PackageSI-RNTISystemInformationRNTISMFSessionmanagementfunctionSNDRSignaltoNoise-and-DistortionRatioSNRSignal-to-NoiseRatioSoCSystem-on-ChipSRSchedulingRequestSRISRSresourceindicatorSRITSetofRadioInterfaceTechnologiesSRSSoundingReferenceSignalSSSynchronizationSignalSSBSynchronizationSignalBlockSSSSecondarySynchronizationSignalSMTSurface-MountassemblySULSupplementaryUplinkSU-MIMOSingle-UserMIMO

TABTransceiver-ArrayBoundaryTACSTotalAccessCommunicationSystem

TCITransmissionconfigurationindicationTCPTransmissionControlProtocolTC-RNTITemporaryC-RNTITDDTime-DivisionDuplexTDMTimeDivisionMultiplexingTDMATime-DivisionMultipleAccessTD-SCDMATime-Division-SynchronousCode-DivisionMultipleAccessTIATelecommunicationIndustryAssociationTRTechnicalReportTRPTotalRadiatedPowerTSTechnicalSpecificationTRSTrackingReferenceSignalTSDSITelecommunicationsStandardsDevelopmentSociety,IndiaTSGTechnicalSpecificationGroupTTATelecommunicationsTechnologyAssociationTTCTelecommunicationsTechnologyCommitteeTTITransmissionTimeIntervalTXTransmitterUCIUplinkControlInformationUDMUnifieddatamanagementUEUserEquipment,the3GPPnameforthemobileterminalUEMUnwantedEmissionsMaskULUplinkUMTSUniversalMobileTelecommunicationsSystemUPFUserplanefunctionURLLCUltra-reliablelow-latencycommunicationUTRAUniversalTerrestrialRadioAccessV2XVehicular-to-AnythingV2VVehicular-to-VehicularVCOVoltage-ControlledOscillatorWARCWorldAdministrativeRadioCongressWCDMAWidebandCode-DivisionMultipleAccessWGWorkingGroupWiMAXWorldwideInteroperabilityforMicrowaveAccessWP5DWorkingParty5D

WRCWorldRadiocommunicationConferenceXnTheinterfacebetweengNBsZCZadoff-ChuZP-CSI-RSZero-powerCSI-RS

CHAPTER1

WhatIs5G?

Abstract

Thechaptergivesbackgroundto5Gmobilecommunication,describingtheearlier generations and the justification for a newgeneration. It describesthehigh-level5Gusecases,eMBB,mMTC,andURLLC.Italsodescribesthe3GPPprocessfordevelopingthenew5G/NRradio-accesstechnology.

Keywords5G;NR;3GPP;eMBB;URLLC;mMTC;machine-typecommunication

Over the last 40 years, the world has witnessed four generations of mobilecommunication(seeFig.1.1).

FIGURE1.1 Thedifferentgenerationsofmobilecommunication.

The first generation ofmobile communication, emerging around 1980, wasbased on analog transmission with the main technologies being AMPS(Advanced Mobile Phone System) developed within North America, NMT(NordicMobileTelephony)jointlydevelopedbythe,at thattime,government-controlled public-telephone-network operators of the Nordic countries, and

TACS(TotalAccessCommunicationSystem)usedin,forexample,theUnitedKingdom. The mobile-communication systems based on first-generationtechnologywere limited tovoice services and, for the first time,mademobiletelephonyaccessibletoordinarypeople.The second generation of mobile communication, emerging in the early

1990s, saw the introductionofdigital transmissionon the radio link.Althoughthe target service was still voice, the use of digital transmission allowed forsecond-generationmobile-communication systems to also provide limited dataservices. Therewere initially several different second-generation technologies,includingGSM (Global System forMobile communication) jointly developedby a large number of European countries, D-AMPS (Digital AMPS), PDC(PersonalDigitalCellular)developedandsolelyusedinJapan,and,developedata somewhat later stage, theCDMA-based IS-95 technology.As timewent by,GSM spread fromEurope to other parts of theworld and eventually came tocompletelydominateamong the second-generation technologies.Primarilydueto the success of GSM, the second-generation systems also turned mobiletelephonyfromsomethingstillbeingusedbyonlyarelativelysmallfractionofpeopletoacommunicationtoolbeinganecessarypartoflifeforalargemajorityoftheworld'spopulation.EventodaytherearemanyplacesintheworldwhereGSMisthedominating,andinsomecaseseventheonlyavailable, technologyfor mobile communication, despite the later introduction of both third-andfourth-generationtechnologies.The thirdgenerationofmobile communication,often just referred toas3G,

wasintroducedintheearly2000.With3Gthetruesteptohigh-qualitymobilebroadbandwastaken,enablingfastwirelessinternetaccess.Thiswasespeciallyenabledbythe3GevolutionknownasHSPA(HighSpeedPacketAccess)[21].In addition, while earlier mobile-communication technologies had all beendesigned for operation in paired spectrum (separate spectrum for network-to-device and device-to-network links) based on the Frequency-Division Duplex(FDD), see Chapter 7, 3G also saw the first introduction of mobilecommunicationinunpairedspectrumbasedonthechina-developedTD-SCDMAtechnologybasedonTimeDivisionDuplex(TDD).Wearenow, andhavebeen for several years, in the fourth-generation (4G)

eraofmobilecommunication,representedbytheLTEtechnology[28]LTEhasfollowedinthestepsofHSPA,providinghigherefficiencyandfurtherenhancedmobile-broadbandexperienceintermsofhigherachievableend-userdatarates.This is provided by means of OFDM-based transmission enabling wider

transmission bandwidths and more advanced multi-antenna technologies.Furthermore,while3Gallowedformobilecommunicationinunpairedspectrumby means of a specific radio-access technology (TD-SCDMA), LTE supportsboth FDD and TDD operation, that is, operation in both paired and unpairedspectra, within one common radio-access technology. By means of LTE theworld has thus converged into a single global technology for mobilecommunication,usedbyessentiallyallmobile-networkoperatorsandapplicableto both paired and unpaired spectra.As discussed in somewhatmore detail inChapter4,thelaterevolutionofLTEhasalsoextendedtheoperationofmobile-communicationnetworksintounlicensedspectra.

1.13GPPandtheStandardizationofMobileCommunicationAgreeingonmultinationaltechnologyspecificationsandstandardshasbeenkeyto the successofmobile communication.Thishas allowed for thedeploymentand interoperability of devices and infrastructure of different vendors andenableddevicesandsubscriptionstooperateonaglobalbasis.As already mentioned, already the first-generation NMT technology was

createdonamultinationalbasis,allowingfordevicesandsubscriptiontooperateover the national borders between the Nordic countries. The next step inmultinationalspecification/standardizationofmobile-communicationtechnologytook place when GSM was jointly developed between a large number ofEuropean countries within CEPT, later renamed ETSI (EuropeanTelecommunications Standards Institute). As a consequence of this, GSMdevicesandsubscriptionswerealreadyfromthebeginningabletooperateoveralargenumberofcountries,coveringaverylargenumberofpotentialusers.Thislargecommonmarkethadaprofound impactondeviceavailability, leading toanunprecedentednumberofdifferentdevicetypesandsubstantialreductionindevicecost.However, the final step to true global standardization of mobile

communication camewith the specification of the 3G technologies, especiallyWCDMA.Workon3G technologywas initiallyalsocarriedoutona regionalbasis, that is, separately within Europe (ETSI), North America (TIA, T1P1),Japan(ARIB),etc.However,thesuccessofGSMhadshowntheimportanceofalargetechnologyfootprint,especiallyintermsofdeviceavailabilityandcost.Italso become clear that although work was carried out separately within the

different regional standard organizations, there were many similarities in theunderlying technologybeingpursued.Thiswas especially true forEurope andJapanwhichwerebothdevelopingdifferentbutverysimilarflavorsofwidebandCDMA(WCDMA)technology.As a consequence, in 1998, the different regional standardization

organizations came together and jointly created the Third-GenerationPartnershipProject (3GPP)with the taskof finalizing thedevelopmentof3GtechnologybasedonWCDMA.Aparallelorganization(3GPP2)wassomewhatlater created with the task of developing an alternative 3G technology,cdma2000,asanevolutionofsecond-generationIS-95.Foranumberofyears,thetwoorganizations(3GPPand3GPP2)withtheirrespective3Gtechnologies(WCDMAandcdma2000)existed inparallel.However,over time3GPPcameto completely dominate and has, despite its name, continued into thedevelopment of 4G (LTE, and 5G) technologies. Today, 3GPP is the onlysignificant organization developing technical specifications for mobilecommunication.

1.2TheNextGeneration—5G/NRDiscussionsonfifth-generation(5G)mobilecommunicationbeganaround2012.In many discussions, the term 5G is used to refer to specific new 5G radio-accesstechnology.However,5Gisalsooftenusedinamuchwidercontext,notjustreferringtoaspecificradio-accesstechnologybutrathertoawiderangeofnewservicesenvisionedtobeenabledbyfuturemobilecommunication.

1.2.1The5GUseCasesIn thecontextof5G,one isoften talkingabout threedistinctiveclassesofusecases: enhanced mobile broadband (eMBB), massive machine-typecommunication (mMTC), and ultra-reliable and low-latency communication(URLLC)(seealsoFig.1.2).

•eMBBcorrespondstoamoreorlessstraightforwardevolutionofthemobile-broadbandservicesoftoday,enablingevenlargerdatavolumesandfurtherenhanceduserexperience,forexample,bysupportingevenhigherend-userdatarates.

•mMTCcorrespondstoservicesthatarecharacterizedbyamassive

numberofdevices,forexample,remotesensors,actuators,andmonitoringofvariousequipment.Keyrequirementsforsuchservicesincludeverylowdevicecostandverylowdeviceenergyconsumption,allowingforverylongdevicebatterylifeofuptoatleastseveralyears.Typically,eachdeviceconsumesandgeneratesonlyarelativelysmallamountofdata,thatis,supportforhighdataratesisoflessimportance.

•URLLCtype-of-servicesareenvisionedtorequireverylowlatencyandextremelyhighreliability.Exampleshereofaretrafficsafety,automaticcontrol,andfactoryautomation.

FIGURE1.2 High-level5Guse-caseclassification.

Itisimportanttounderstandthattheclassificationof5Gusecasesintothesethreedistinctiveclassesissomewhatartificial,primarilyaimingtosimplifythedefinitionofrequirementsforthetechnologyspecification.Therewillbemanyuse cases that donot fit exactly into oneof these classes. Just as an example,theremaybeservicesthatrequireveryhighreliabilitybutforwhichthelatencyrequirements are not that critical. Similarly, theremay be use cases requiringdevicesofvery lowcostbutwhere thepossibility forvery longdevicebatterylifemaybelessimportant.

1.2.2EvolvingLTEto5GCapabilityThe first release of the LTE technical specifications was introduced in 2009.Sincethen,LTEhasgonethroughseveralstepsofevolutionprovidingenhancedperformanceandextendedcapabilities.Thishasincludedfeaturesforenhanced

mobilebroadband,includingmeansforhigherachievableend-userdataratesaswell as higher spectrum efficiency. However, it has also included importantsteps to extend the set of use cases towhichLTE can be applied. Especially,therehavebeenimportantstepstoenabletrulylow-costdeviceswithverylongbatterylife,inlinewiththecharacteristicsofmassiveMTCapplications.Therehave recently also been some significant steps taken to reduce the LTE air-interfacelatency.With these finalized, ongoing, and future evolution steps, the evolution of

LTEwill be able to support awide rangeof the use cases envisioned for 5G.Takingintoaccountthemoregeneralviewthat5Gisnotaspecificradio-accesstechnologybutratherdefinedbytheusecasestobesupported,theevolutionofLTE should thus be seen as an important part of the overall 5G radio-accesssolution, see Fig. 1.3. Although not being the main aim of this book, anoverviewofthecurrentstateoftheLTEevolutionisprovidedinChapter4.

FIGURE1.3 EvolutionofLTEandNRjointlyprovidingtheoverall5Gradio-accesssolution.

1.2.3NR—TheNew5GRadio-AccessTechnologyDespite LTE being a very capable technology, there are requirements notpossible to meet with LTE or its evolution. Furthermore, technologydevelopment over themore than 10 years that have passed since thework onLTEwasinitiatedallowsformoreadvancedtechnicalsolutions.Tomeettheserequirementsandtoexploitthepotentialofnewtechnologies,3GPPinitiatedthedevelopment of a new radio-access technologyknown asNR (NewRadio).Aworkshop setting the scope was held in the fall of 2015 and technical work

began in the spring of 2016. The first version of the NR specifications wasavailable by the end of 2017 to meet commercial requirements on early 5Gdeploymentsalreadyin2018.NRreusesmanyofthestructuresandfeaturesofLTE.However,beinganew

radio-access technology means that NR, unlike the LTE evolution, is notrestrictedbyaneedtoretainbackwardscompatibility.TherequirementsonNRarealsobroaderthanwhatwasthecaseforLTE,motivatingapartlydifferentsetoftechnicalsolutions.Chapter2discussesthestandardizationactivitiesrelatedtoNR,followedbya

spectrumoverviewinChapter3andabriefsummaryofLTEanditsevolutioninChapter4.Themainpartofthisbook(Chapters5–19)thenprovidesanin-depthdescriptionofthecurrentstageoftheNRtechnicalspecifications,finishingwithanoutlookofthefuturedevelopmentofNRinChapter20.

1.2.45GCN—TheNew5GCoreNetworkIn parallel to NR, that is, the new 5G radio-access technology, 3GPP is alsodeveloping a new 5G core network referred to as 5GCN. The new 5G radio-accesstechnologywillconnecttothe5GCN.However,5GCNwillalsobeabletoprovideconnectivityfortheevolutionofLTE.Atthesametime,NRmayalsoconnect via the legacy core network EPC when operating in so-called non-standalonemodetogetherwillLTE,aswillbefurtherdiscussedinChapter6.

CHAPTER2

5GStandardization

Abstract

Thischapterpresentstheregulationandstandardizationactivitiesrelatedto5G NR, including all the relevant regulation and standards bodies. TheITU-R IMT-2020 process for 5G is presented together with thecorersponding3GPPprocessthatledto5GNR.

Keywordsstandardization;regulation;ITU-R;IMT-2020;3GPP;TSGRAN;5G;NR;usagescenarios;keycapabilities;technicalperformancerequirements

The research, development, implementation, and deployment of mobile-communicationsystems isperformedby thewireless industry inacoordinatedinternational effort by which common industry specifications that define thecompletemobile-communicationsystemareagreed.Theworkdependsheavilyonglobal and regional regulation, in particular for the spectrumuse that is anessential component for all radio technologies. This chapter describes theregulatoryandstandardizationenvironment thathasbeen,andcontinues tobe,essentialfordefiningthemobile-communicationsystems.

2.1OverviewofStandardizationandRegulationThereareanumberoforganizationsinvolvedincreatingtechnicalspecificationsand standards aswell as regulation in themobile-communications area.Thesecan looselybedivided into threegroups:StandardsDevelopingOrganizations,regulatorybodiesandadministrations,andindustryforums.StandardsDevelopingOrganizations(SDOs)developandagreeontechnical

standards formobilecommunicationssystems, inorder tomake itpossible forthe industry to produce and deploy standardized products and provide

interoperability between those products. Most components of mobile-communication systems, including base stations and mobile devices, arestandardizedtosomeextent.Thereisalsoacertaindegreeoffreedomtoprovideproprietary solutions in products, but the communications protocols rely ondetailed standards for obvious reasons. SDOs are usually nonprofit industryorganizationsandnotgovernmentcontrolled.Theyoftenwritestandardswithinacertainareaundermandatefromgovernments(s)however,givingthestandardsahigherstatus.There are nationals SDOs, but due to the global spread of communications

products,mostSDOsare regional and also cooperateon aglobal level.As anexample, the technical specificationsofGSM,WCDMA/HSPA,LTE, andNRareallcreatedby3GPP(ThirdGenerationPartnershipProject)whichisaglobalorganization from seven regional and national SDOs in Europe (ETSI), Japan(ARIBandTTC), theUnitedStates(ATIS),China(CCSA),Korea(TTA),andIndia(TSDSI).SDOstendtohaveavaryingdegreeoftransparency,but3GPPisfully transparentwith all technical specifications,meeting documents, reports,ande-mailreflectorspubliclyavailablewithoutchargeevenfornonmembers.Regulatorybodiesandadministrationsaregovernment-ledorganizationsthat

set regulatory and legal requirements for selling, deploying, and operatingmobile systems and other telecommunication products. One of their mostimportanttasksistocontrolspectrumuseandtosetlicensingconditionsforthemobileoperatorsthatareawardedlicensestousepartsoftheRadioFrequency(RF)spectrumformobileoperations.Anothertaskistoregulate“placingonthemarket” of products through regulatory certification, by ensuring that devices,base stations, and other equipment is type-approved and shown to meet therelevantregulation.Spectrum regulation is handled both on a national level by national

administrations, but also through regional bodies in Europe (CEPT/ECC), theAmericas(CITEL),andAsia(APT).Onagloballevel,thespectrumregulationis handled by the International Telecommunications Union (ITU). Theregulatory bodies regulatewhat services the spectrum is to be used for and inaddition setmore detailed requirements such as limits on unwanted emissionsfrom transmitters.They are also indirectly involved in setting requirementsonthe product standards through regulation. The involvement of ITU in settingrequirementsonthetechnologiesformobilecommunicationisexplainedfurtherinSection2.2.Industry forumsare industry-ledgroupspromotingand lobbyingforspecific

technologies or other interests. In the mobile industry, these are often led byoperators, but there are also vendors creating industry forums.An example ofsuch a group is GSMA (GSM Association) which is promoting mobile-communication technologies based on GSM,WCDMA, LTE, and NR. Otherexamples of industry forums areNextGenerationMobile Networks (NGMN),which is an operator group defining requirements on the evolution of mobilesystems,and5GAmericas,whichisaregionalindustryforumthathasevolvedfromitspredecessor4GAmericas.Fig.2.1illustratestherelationshipbetweendifferentorganizationsinvolvedin

setting regulatoryand technicalconditions formobile systems.The figurealsoshowsthemobileindustryview,wherevendorsdevelopproducts,placethemonthemarketandnegotiatewithoperatorswhoprocureanddeploymobilesystems.This process relies heavily on the technical standards published by the SDOs,while placing products on the market relies on certification of products on aregional or national level. Note that, in Europe, the regional SDO (ETSI) isproducing the so-called harmonized standards used for product certification(through the“CE”-mark),basedonamandate fromthe regulators, in thiscasethe European Commission. These standards are also used for certification inmany countries outside of Europe. In Fig. 2.1, full arrows indicate formaldocumentation such as technical standards, recommendations, and regulatorymandatesthatdefinethetechnologiesandregulation.Dashedarrowsshowmoreindirectinvolvementthrough,forexample,liaisonstatementsandwhitepapers.

FIGURE2.1 Simplifiedviewoftherelationshipbetweenregulatorybodies,standardsdevelopingorganizations,industryforums,andthemobileindustry.

2.2ITU-RActivitiesFrom3Gto5G2.2.1TheRoleofITU-RITU-R is the radio communications sector of the InternationalTelecommunications Union. ITU-R is responsible for ensuring efficient andeconomical use of theRF spectrum by all radio communication services. Thedifferent subgroups andworkingpartiesproduce reports and recommendationsthat analyze and define the conditions for using the RF spectrum. The quiteambitious goal of ITU-R is to “ensure interference-free operations of radiocommunicationsystems,”by implementing theRadioRegulationsandregionalagreements.TheRadioRegulationsisaninternationalbindingtreatyforhowRFspectrum is used. AWorld Radio-communication Conference (WRC) is heldevery 3–4 years. At WRC the Radio Regulations are revised and updated,resultinginrevisedandupdateduseoftheRFspectrumacrosstheworld.Whilethetechnicalspecificationofmobile-communicationtechnologies,such

asNR,LTE,andWCDMA/HSPAisdonewithin3GPP,thereisaresponsibility

for ITU-R in the process of turning the technologies into global standards, inparticular for countries that are not covered by the SDOs that are partners in3GPP. ITU-R defines the spectrum for different services in the RF spectrum,includingmobile services, and someof that spectrum is particularly identifiedfor so-called InternationalMobileTelecommunications (IMT) systems.WithinITU-R,itisWorkingParty5D(WP5D)thathastheresponsibilityfortheoverallradio system aspects of IMT systems, which, in practice, corresponds to thedifferent generations of mobile-communication systems from 3G onwards.WP5D has the prime responsibility within ITU-R for issues related to theterrestrial component of IMT, including technical, operational, and spectrum-relatedissues.WP5D does not create the actual technical specifications for IMT, but has

kept therolesofdefiningIMTincooperationwiththeregionalstandardizationbodiesandmaintainingasetofrecommendationsandreportsforIMT,includingasetofRadioInterfaceSpecifications(RSPCs).Theserecommendationscontain“families”ofRadioInterfaceTechnologies(RITs)foreachIMTgeneration,allincluded on an equal basis. For each radio interface, the RSPC contains anoverviewofthatradiointerface,followedbyalistofreferencestothedetailedspecifications.The actual specifications aremaintainedby the individualSDOand the RSPC provides references to the specifications transposed andmaintained by each SDO. The following RSPC recommendations are inexistenceorplanned:

•ForIMT-2000:ITU-RRecommendationM.1457[49]containingsixdifferentRITsincludingthe3GtechnologiessuchasWCDMA/HSPA.

•ForIMT-Advanced:ITU-RRecommendationM.2012[45]containingtwodifferentRITswherethemostimportantis4G/LTE.

•ForIMT-2020:AnewITU-RRecommendation,containingtheRITsfor5Gtechnologies,plannedtobedevelopedin2019–20.

Each RSPC is continuously updated to reflect new developments in thereferenceddetailedspecifications,suchasthe3GPPspecificationsforWCDMAand LTE. Input to the updates is provided by the SDOs and the PartnershipProjects,nowadaysprimarily3GPP.

2.2.2IMT-2000andIMT-Advanced

Workonwhatcorrespondstothirdgenerationofmobilecommunicationstartedin the ITU-R in the 1980s. First referred to as Future Public Land MobileSystems(FPLMTS)itwaslaterrenamedIMT-2000.Inthelate1990s,theworkinITU-RcoincidedwiththeworkindifferentSDOsacrosstheworldtodevelopanewgenerationofmobilesystems.AnRSPCforIMT-2000wasfirstpublishedin2000andincludedWCDMAfrom3GPPasoneoftheRITs.ThenextstepforITU-RwastoinitiateworkonIMT-Advanced,thetermused

for systems that include new radio interfaces supporting new capabilities ofsystemsbeyond IMT-2000.Thenewcapabilitiesweredefined ina frameworkrecommendationpublishedby the ITU-R[41]andweredemonstratedwith the“vandiagram”shown inFig.2.2.Thestep into IMT-AdvancedcapabilitiesbyITU-R coincided with the step into 4G, the next generation of mobiletechnologiesafter3G.

FIGURE2.2 IllustrationofcapabilitiesofIMT-2000andIMT-Advanced,basedontheframeworkdescribedinITU-RRecommendationM.1645[41].

AnevolutionofLTEasdevelopedby3GPPwassubmittedasonecandidatetechnologyforIMT-Advanced.Whileactuallybeinganewrelease(Release10)oftheLTEspecificationsandthusanintegralpartofthecontinuousevolutionofLTE, the candidate was named LTE-Advanced for the purpose of ITU-Rsubmission and this name is alsoused in theLTE specifications fromRelease10. In parallel with the ITU-R work, 3GPP set up its own set of technical

requirementsforLTE-Advanced,withtheITU-Rrequirementsasabasis[10].The target of the ITU-R process is always harmonization of the candidates

throughconsensusbuilding.ITU-Rdeterminedthat twotechnologieswouldbeincludedinthefirstreleaseofIMT-Advanced,thosetwobeingLTE-Advancedand WirelessMAN-Advanced [37] based on the IEEE 802.16m specification.Thetwocanbeviewedasthe“family”ofIMT-Advancedtechnologiesasshownin Fig. 2.3. Note that, of these two technologies, LTE has emerged as thedominating4Gtechnologybyfar.

FIGURE2.3 RadioInterfaceTechnologiesIMT-Advanced.

2.2.3IMT-2020ProcessinITU-RWP5DStarting in 2012, ITU-RWP5D set the stage for the next generation of IMTsystems, named IMT-2020. It is a further development of the terrestrialcomponentofIMTbeyondtheyear2020and,inpractice,correspondstowhatismorecommonlyreferredtoas“5G,”thefifthgenerationofmobilesystems.Theframeworkandobjective for IMT-2020 isoutlined in ITU-RRecommendationM.2083 [47], often referred to as the “Vision” recommendation. Therecommendation provides the first step for defining the new developments ofIMT,lookingatthefuturerolesofIMTandhowitcanservesociety,lookingatmarket,userandtechnologytrends,andspectrumimplications.Theusertrends

forIMTtogetherwiththefutureroleandmarketleadtoasetofusagescenariosenvisioned for both human-centric and machine-centric communication. Theusage scenarios identified are Enhanced Mobile Broadband (eMBB), Ultra-Reliable and Low Latency Communications (URLLC), andMassiveMachine-TypeCommunications(mMTC).The need for an enhanced mobile broadband experience, together with the

newandbroadenedusagescenarios,leadstoanextendedsetofcapabilitiesforIMT-2020. TheVision recommendation [47] gives a first high-level guidancefor IMT-2020 requirements by introducing a set of key capabilities, withindicative targetnumbers.Thekeycapabilities and the relatedusage scenariosarediscussedfurtherinSection2.3.Asaparallelactivity,ITU-RWP5Dproducedareporton“Futuretechnology

trendsofterrestrialIMTsystems”[43],withafocusonthetimeperiod2015–20.ItcoverstrendsoffutureIMTtechnologyaspectsbylookingatthetechnicalandoperationalcharacteristicsofIMTsystemsandhowtheyareimprovedwiththeevolution of IMT technologies. In this way, the report on technology trendsrelates to LTE in 3GPP Release 13 and beyond, while the Visionrecommendation looks furtheraheadandbeyond2020.Anewaspecton IMT-2020 is that itwillbecapableofoperating inpotentialnew IMTbandsabove6 GHz, including mm-wave bands. With this in mind, WP5D produced aseparate report studying radiowavepropagation, IMTcharacteristics, enablingtechnologies,anddeploymentinfrequenciesabove6GHz[44].AtWRC-15,potentialnewbandsforIMTwerediscussedandanagendaitem

1.13 was set up for WRC-19, covering possible additional allocations to themobile services and for future IMT development. These allocations areidentified in a number of frequency bands in the range between 24.25 and86GHz.ThespecificbandsandtheirpossibleusegloballyarefurtherdiscussedinChapter3.AfterWRC-15, ITU-RWP5Dcontinued theprocessof setting requirements

and defining evaluation methodologies for IMT-2020 systems, based in theVisionrecommendation[47]andtheotherpreviousstudyoutcomes.Thisstepoftheprocesswascompletedinmid-2017,asshownintheIMT-2020workplaninFig. 2.4. The result was three documents published late in 2017 that furtherdefinetheperformanceandcharacteristicsthatareexpectedfromIMT-2020andthatwillbeappliedintheevaluationphase:

•Technicalrequirements:ReportITU-RM.2410[51]defines13

minimumrequirementsrelatedtothetechnicalperformanceoftheIMT-2020radiointerface(s).TherequirementsaretoalargeextentbasedonthekeycapabilitiessetoutintheVisionrecommendation(ITU-R,2015c).ThisisfurtherdescribedinSection2.3.

•Evaluationguideline:ReportITU-RM.2412[50]definesthedetailedmethodologytouseforevaluatingtheminimumrequirements,includingtestenvironments,evaluationconfigurations,andchannelmodels.MoredetailsaregiveninSection2.3.

•Submissiontemplate:ReportITU-RM.2411[52]providesadetailedtemplatetouseforsubmittingacandidatetechnologyforevaluation.Italsodetailstheevaluationcriteriaandrequirementsonservice,spectrum,andtechnicalperformance,basedonthetwopreviouslymentionedITU-RreportsM.2410andM.2412.

FIGURE2.4 WorkplanforIMT-2020inITU-RWP5D[40].

ExternalorganizationsarebeinginformedoftheIMT-2020processthroughacircular letter.After aworkshop on IMT-2020was held inOctober 2017, theIMT-2020processisopenforreceivingcandidateproposals.Theplan,asshowninFig.2.4,istostarttheevaluationofproposalsin2018,

aiming at an outcomewith the RSPC for IMT-2020 being published early in2020.

2.35GandIMT-2020ThedetailedITU-RtimeplanforIMT-2020waspresentedabovewiththemost

important steps summarized in Fig. 2.4. The ITU-R activities on IMT-2020startedwithdevelopmentof the“vision”recommendationITU-RM.2083[47],outlining theexpectedusescenariosandcorrespondingrequiredcapabilitiesofIMT-2020.Thiswas followed by definition ofmore detailed requirements forIMT-2020, requirements that candidate technologies are then to be evaluatedagainst, as documented in the evaluation guidelines. The requirements andevaluationguidelineswerefinalizedmid-2017.With the requirements finalized, candidate technologies canbe submitted to

ITU-R. The proposed candidate technology/technologies will be evaluatedagainst the IMT-2020 requirementsand the technology/technologies that fulfillthe requirements will be approved and published as part of the IMT-2020specifications in thesecondhalfof2020.Furtherdetailson theITU-RprocesscanbefoundinSection2.2.3.

2.3.1UsageScenariosforIMT-2020Withawide rangeofnewusecasesbeingoneprincipaldriver for5G, ITU-Rhas defined three usage scenarios that form a part of the IMT Visionrecommendation[47].InputsfromthemobileindustryanddifferentregionalandoperatororganizationsweretakenintotheIMT-2020processinITU-RWP5D,andweresynthesizedintothethreescenarios:

•EnhancedMobileBroadband(eMBB):Withmobilebroadbandtodaybeingthemaindriverforuseof3Gand4Gmobilesystems,thisscenariopointsatitscontinuedroleasthemostimportantusagescenario.Thedemandiscontinuouslyincreasingandnewapplicationareasareemerging,settingnewrequirementsforwhatITU-RcallsEnhancedMobileBroadband.Becauseofitsbroadandubiquitoususe,itcoversarangeofusecaseswithdifferentchallenges,includingbothhotspotsandwide-areacoverage,withthefirstoneenablinghighdatarates,highuserdensity,andaneedforveryhighcapacity,whilethesecondonestressesmobilityandaseamlessuserexperience,withlowerrequirementsondatarateanduserdensity.TheEnhancedMobileBroadbandscenarioisingeneralseenasaddressinghuman-centriccommunication.

•Ultra-reliableandlow-latencycommunications(URLLC):Thisscenarioisintendedtocoverbothhuman-andmachine-centriccommunication,

wherethelatterisoftenreferredtoascriticalmachinetypecommunication(C-MTC).Itischaracterizedbyusecaseswithstringentrequirementsforlatency,reliability,andhighavailability.Examplesincludevehicle-to-vehiclecommunicationinvolvingsafety,wirelesscontrolofindustrialequipment,remotemedicalsurgery,anddistributionautomationinasmartgrid.Anexampleofahuman-centricusecaseis3Dgamingand“tactileinternet,”wherethelow-latencyrequirementisalsocombinedwithveryhighdatarates.

•Massivemachinetypecommunications(mMTC):Thisisapuremachine-centricusecase,wherethemaincharacteristicisaverylargenumberofconnecteddevicesthattypicallyhaveverysparsetransmissionsofsmalldatavolumesthatarenotdelay-sensitive.Thelargenumberofdevicescangiveaveryhighconnectiondensitylocally,butitisthetotalnumberofdevicesinasystemthatcanbetherealchallengeandstressestheneedforlowcost.DuetothepossibilityofremotedeploymentofmMTCdevices,theyarealsorequiredtohaveaverylongbatterylifetime.

The usage scenarios are illustrated in Fig. 2.5, togetherwith some exampleuse cases.The three scenarios above are not claimed to cover all possible usecases, but they provide a relevant grouping of a majority of the presentlyforeseenusecasesandcanthusbeusedto identify thekeycapabilitiesneededfor the next-generation radio interface technology for IMT-2020. There willmost certainly be new use cases emerging,whichwe cannot foresee today ordescribeinanydetail.Thisalsomeansthatthenewradiointerfacemusthaveahighflexibilitytoadapttonewusecasesandthe“space”spannedbytherangeof the key capabilities supported should support the related requirementsemergingfromevolvingusecases.

FIGURE2.5 IMT-2020usecasesandmappingtousagescenarios.FromITU-R,RecommendationITU-RM.2083[47],usedwithpermissionfromtheITU.

2.3.2CapabilitiesofIMT-2020As part of developing the framework for the IMT-2020 as documented in theIMTVisionrecommendation[47],ITU-Rdefinedasetofcapabilitiesneededforan IMT-2020 technology to support the 5G use cases and usage scenariosidentified throughthe inputsfromregionalbodies, researchprojects,operators,administrations, and other organizations. There are a total of 13 capabilitiesdefinedinITU-R[47],whereeightwereselectedaskeycapabilities.Thoseeightkeycapabilitiesareillustratedthroughtwo“spiderweb”diagrams(seeFigs.2.6and2.7).

FIGURE2.6 KeycapabilitiesofIMT-2020.FromITU-R,RecommendationITU-RM.2083[47],usedwithpermissionfromtheITU.

FIGURE2.7 RelationbetweenkeycapabilitiesandthethreeusagescenariosofITU-R.FromITU-R,RecommendationITU-RM.2083[47],usedwithpermissionfromtheITU.

Fig.2.6illustratesthekeycapabilitiestogetherwithindicativetargetnumbersintended to give a first high-level guidance for the more detailed IMT-2020requirementsthatarenowunderdevelopment.AscanbeseenthetargetvaluesarepartlyabsoluteandpartlyrelativetothecorrespondingcapabilitiesofIMT-Advanced.Thetargetvaluesforthedifferentkeycapabilitiesdonothavetobereached simultaneously and some targets are to a certain extent evenmutuallyexclusive. For this reason, there is a second diagram shown inFig. 2.7whichillustrates the“importance”ofeachkeycapability for realizing the threehigh-levelusagescenariosenvisionedbyITU-R.Peakdatarateisanumberonwhichthereisalwaysalotoffocus,butitisin

factquiteanacademicexercise.ITU-Rdefinespeakdataratesasthemaximumachievabledatarateunderidealconditions,whichmeansthattheimpairmentsinan implementation or the actual impact from a deployment in terms ofpropagation, etc. does not come into play. It is a dependent key performanceindicator (KPI) in that it is heavily dependent on the amount of spectrumavailable for an operator deployment. Apart from that, the peak data ratedependsonthepeakspectralefficiency,whichisthepeakdataratenormalizedbythebandwidth:

Since large bandwidths are really not available in any of the existing IMTbandsbelow6GHz,itisexpectedthatreallyhighdatarateswillbemoreeasilyachievedathigherfrequencies.Thisleadstotheconclusionthatthehighestdatarates can be achieved in indoor and hotspot environments, where the lessfavorablepropagationpropertiesathigherfrequenciesareoflessimportance.Theuser experienced data rate is the data rate that can be achieved over a

largecoverageareaforamajorityoftheusers.Thiscanbeevaluatedasthe95thpercentilefromthedistributionofdataratesbetweenusers.Itisalsoadependentcapability, not only on the available spectrum but also on how the system isdeployed.Whileatargetof100Mbit/sissetforwideareacoverageinurbanandsuburban areas, it is expected that 5G systems could give 1 Gbit/s data rateubiquitouslyinindoorandhotspotenvironments.SpectrumefficiencygivestheaveragedatathroughputperHzofspectrumand

per“cell,”orratherperunitofradioequipment(alsoreferredtoasTransmissionReceptionPoint,TRP). It isanessentialparameter fordimensioningnetworks,but the levelsachievedwith4Gsystemsarealreadyveryhigh.The targetwasset to three times the spectrum efficiency target of 4G, but the achievableincreasestronglydependsonthedeploymentscenario.Areatrafficcapacityisanotherdependentcapability,whichdependsnotonly

onthespectrumefficiencyandthebandwidthavailable,butalsoonhowdensethenetworkisdeployed:

Byassumingtheavailabilityofmorespectrumathigherfrequenciesandthatverydensedeploymentscanbeused,atargetofa100-foldincreaseover4GwassetforIMT-2020.Networkenergyefficiency is,asalreadydescribed,becominganincreasingly

important capability. The overall target stated by ITU-R is that the energyconsumptionoftheradioaccessnetworkofIMT-2020shouldnotbegreaterthanIMTnetworksdeployed today,while still delivering theenhancedcapabilities.The target means that the network energy efficiency in terms of energyconsumedperbitofdatathereforeneedstobereducedwithafactoratleastasgreatastheenvisagedtrafficincreaseofIMT-2020relativetoIMT-Advanced.These first five key capabilities are of highest importance for theEnhanced

Mobile Broadband usage scenario, although mobility and the data ratecapabilitieswould not have equal importance simultaneously. For example, in

hotspots,averyhighuser-experiencedandpeakdatarate,butalowermobility,wouldberequiredthaninwideareacoveragecase.Latency isdefinedas thecontributionbytheradionetworkto the timefrom

when the source sendsapacket towhen thedestination receives. Itwillbeanessential capability for theURLLCusage scenarioand ITU-Renvisions that a10-foldreductioninlatencyfromIMT-Advancedisrequired.Mobilityisinthecontextofkeycapabilitiesonlydefinedasmobilespeedand

the target of 500 km/h is envisioned in particular for high-speed trains and isonly a moderate increase from IMT-Advanced. As a key capability, it will,however,alsobeessentialfortheURLLCusagescenariointhecaseofcriticalvehicle communication at high speed and will then be of high importancesimultaneouslywith lowlatency.Note thatmobilityandhighuser-experienceddataratesarenottargetedsimultaneouslyintheusagescenarios.Connection density is defined as the total number of connected and/or

accessible devices per unit area. The target is relevant for the mMTC usagescenariowithahighdensityof connecteddevices,but aneMBBdense indoorofficecanalsogiveahighconnectiondensity.InadditiontotheeightcapabilitiesgiveninFig.2.6therearefiveadditional

capabilitiesdefinedin[47]:

•SpectrumandbandwidthflexibilitySpectrumandbandwidthflexibilityreferstotheflexibilityofthesystemdesigntohandledifferentscenarios,andinparticulartothecapabilitytooperateatdifferentfrequencyranges,includinghigherfrequenciesandwiderchannelbandwidthsthantoday.

•ReliabilityReliabilityrelatestothecapabilitytoprovideagivenservicewithaveryhighlevelofavailability.

•ResilienceResilienceistheabilityofthenetworktocontinueoperatingcorrectlyduringandafteranaturalorman-madedisturbance,suchasthelossofmainspower.

•SecurityandprivacySecurityandprivacyreferstoseveralareassuchasencryptionandintegrityprotectionofuserdataandsignaling,aswellasend-userprivacy,preventingunauthorizedusertracking,andprotectionofnetworkagainsthacking,fraud,denialofservice,maninthemiddle

attacks,etc.•OperationallifetimeOperationallifetimereferstooperationtimeperstoredenergycapacity.Thisisparticularlyimportantformachine-typedevicesrequiringaverylongbatterylife(forexamplemorethan10years),whoseregularmaintenanceisdifficultduetophysicaloreconomicreasons.

Note that these capabilities are not necessarily less important than thecapabilities of Fig. 2.6, despite the fact that the latter are referred to as “keycapabilities.”Themaindifferenceis that the“keycapabilities”aremoreeasilyquantifiable, while the remaining five capabilities are more of qualitativecapabilitiesthatcannoteasilybequantified.

2.3.3IMT-2020PerformanceRequirementsandEvaluationBased on the usage scenarios and capabilities described in the Visionrecommendation(ITU-R,2015c),ITU-RdevelopedasetofminimumtechnicalperformancerequirementsforIMT-2020.ThesearedocumentedinITU-RreportM.2410 [51] and will serve as the baseline for the evaluation of IMT-2020candidate technologies (see Fig. 2.4). The report describes 14 technicalparameters and the corresponding minimum requirements. These aresummarizedinTable2.1.

Table2.1

The evaluation guideline of candidate radio interface technologies for IMT-2020isdocumentedinITU-RreportM.2412[50]andfollowsthesamestructureas thepreviousevaluationdone for IMT-Advanced. Itdescribes theevaluationmethodologyforthe14minimumtechnicalperformancerequirements,plustwoadditional requirements: support of a wide range of services and support ofspectrumbands.Theevaluationisdonewithreferencetofivetestenvironmentsthatarebased

on the usage scenarios from the Vision recommendation [47]. Each testenvironmenthasanumberofevaluationconfigurationsthatdescribethedetailedparametersthataretobeusedinsimulationsandanalysisfortheevaluation.Thefivetestenvironmentsare:

•IndoorHotspot-eMBB:Anindoorisolatedenvironmentatofficesand/orinshoppingmallsbasedonstationaryandpedestrianuserswithveryhighuserdensity.

•DenseUrban-eMBB:Anurbanenvironmentwithhighuserdensityandtrafficloadsfocusingonpedestrianandvehicularusers.

•Rural-eMBB:Aruralenvironmentwithlargerandcontinuouswidearea

coverage,supportingpedestrian,vehicular,andhigh-speedvehicularusers.

•UrbanMacro-mMTC:Anurbanmacro-environmenttargetingcontinuouscoveragefocusingonahighnumberofconnectedmachinetypedevices.

•UrbanMacro-URLLC:Anurbanmacro-environmenttargetingultra-reliableandlow-latencycommunications.

There are three fundamentalways that requirementswill be evaluated for acandidatetechnology:

•Simulation:Thisisthemostelaboratewaytoevaluatearequirementanditinvolvessystem-orlink-levelsimulations,orboth,oftheradiointerfacetechnology.Forsystem-levelsimulations,deploymentscenariosaredefinedthatcorrespondtoasetoftestenvironments,suchasindoor,denseurban,etc.Requirementsthatwillbeevaluatedthroughsimulationareaverageandfifthpercentilespectrumefficiency,connectiondensity,mobilityandreliability.

•Analysis:Somerequirementscanbeevaluatedthroughacalculationbasedonradiointerfaceparametersorbederivedfromotherperformancevalues.Requirementsthatwillbeevaluatedthroughanalysisarepeakspectralefficiency,peakdatarate,user-experienceddatarate,areatrafficcapacity,controlanduserplanelatency,andmobilityinterruptiontime.

•Inspection:Somerequirementscanbeevaluatedbyreviewingandassessingthefunctionalityoftheradiointerfacetechnology.Requirementsthatwillbeevaluatedthroughsimulationarebandwidth,energyefficiency,supportofawiderangeofservices,andsupportofspectrumbands.

Once candidate technologies are submitted to ITU-R and have entered theprocess,theevaluationphasewillstart.Evaluationcanbedonebytheproponent(“self-evaluation”)orbyanexternalevaluationgroup,doingpartialorcompleteevaluationofoneormorecandidateproposals.

2.43GPPStandardization

Witha framework for IMTsystemssetupby the ITU-R,withspectrummadeavailable by the WRC and with an ever-increasing demand for betterperformance, the task of specifying the actual mobile-communicationtechnologies falls on organizations like 3GPP.More specifically, 3GPPwritesthetechnicalspecificationsfor2GGSM,3GWCDMA/HSPA,4GLTE,and5GNR.3GPP technologiesare themostwidelydeployed in theworld,withmorethan 95% of theworld’s 7.8 billionmobile subscriptions inQ4 2017 [30]. Inorder to understand how 3GPP works, it is important to also understand theprocessofwritingspecifications.

2.4.1The3GPPProcessDevelopingtechnicalspecificationsformobilecommunicationisnotaone-timejob; it isanongoingprocess.Thespecificationsareconstantlyevolving, tryingtomeetnewdemands for servicesand features.Theprocess isdifferent in thedifferentfora,buttypicallyincludesthefourphasesillustratedinFig.2.8:

1.Requirements,whereitisdecidedwhatistobeachievedbythespecification.

2.Architecture,wherethemainbuildingblocksandinterfacesaredecided.3.Detailedspecifications,whereeveryinterfaceisspecifiedindetail.4.Testingandverification,wheretheinterfacespecificationsareprovento

workwithreal-lifeequipment.

FIGURE2.8 Thestandardizationphasesanditerativeprocess.

Thesephasesareoverlappinganditerative.Asanexample,requirementscanbeadded,changed,ordroppedduringthelaterphasesifthetechnicalsolutionscall for it. Likewise, the technical solution in the detailed specifications canchangeduetoproblemsfoundinthetestingandverificationphase.Thespecificationstartswiththerequirementsphase,whereitisdecidedwhat

shouldbeachievedwiththespecification.Thisphaseisusuallyrelativelyshort.

Inthearchitecturephase,thearchitectureisdecided—thatis,theprinciplesofhowtomeet the requirements.Thearchitecturephase includesdecisionsaboutreference points and interfaces to be standardized. This phase is usually quitelongandmaychangetherequirements.Afterthearchitecturephase,thedetailedspecificationphasestarts.Itisinthis

phase that thedetails foreachof the identified interfacesarespecified.Duringthe detailed specification of the interfaces, the standards body may find thatpreviousdecisionsinthearchitectureorevenintherequirementsphasesneedtoberevisited.Finally,thetestingandverificationphasestarts.Itisusuallynotapartofthe

actual specification, but takes place in parallel through testing byvendors andinteroperability testing between vendors. This phase is the final proof of thespecification. During the testing and verification phase, errors in thespecificationmay still be found and those errorsmay change decisions in thedetailedspecification.Albeitnotcommon,changesmayalsoneedtobemadetothe architecture or the requirements. To verify the specification, products areneeded.Hence, the implementation of the products starts after (or during) thedetailedspecificationphase.Thetestingandverificationphaseendswhenthereare stable test specifications that can be used to verify that the equipment isfulfillingthetechnicalspecification.Normally, it takes approximately one year from the time when the

specificationiscompleteduntilcommercialproductsareoutonthemarket.3GPPconsistsofthreeTechnicalSpecificationsGroups(TSGs)(seeFig.2.9)

where TSG RAN (Radio Access Network) is responsible for the definition offunctions,requirements,andinterfacesoftheRadioAccess.TSGRANconsistsofsixworkinggroups(WGs):

1.RANWG1,dealingwiththephysicallayerspecifications.2.RANWG2,dealingwiththelayer2andlayer3radiointerface

specifications.3.RANWG3,dealingwiththefixedRANinterfaces—forexample,

interfacesbetweennodesintheRAN—butalsotheinterfacebetweentheRANandthecorenetwork.

4.RANWG4,dealingwiththeradiofrequency(RF)andradioresourcemanagement(RRM)performancerequirements.

5.RANWG5,dealingwiththedeviceconformancetesting.6.RANWG6,dealingwithstandardizationofGSM/EDGE(previouslyina

separateTSGcalledGERAN)andHSPA(UTRAN).

FIGURE2.9 3GPPorganization.

Thework in 3GPP is carried out with relevant ITU-R recommendations inmindandtheresultoftheworkisalsosubmittedtoITU-RasbeingpartofIMT-2000,IMT-Advanced,andnowalsoasacandidateforIMT-2020intheformofNR. The organizational partners are obliged to identify regional requirementsthatmayleadtooptionsinthestandard.Examplesareregionalfrequencybandsand special protection requirements local to a region. The specifications aredevelopedwithglobalroamingandcirculationofdevicesinmind.Thisimplies

thatmany regional requirements in essencewill beglobal requirements for alldevices, since a roaming device has to meet the strictest of all regionalrequirements.Regionaloptionsinthespecificationsarethusmorecommonforbasestationsthanfordevices.The specifications of all releases can be updated after each set of TSG

meetings,whichoccurfourtimesayear.The3GPPdocumentsaredividedintoreleases,whereeachreleasehasasetoffeaturesaddedcomparedtothepreviousrelease.The features are defined inWork Items agreed andundertakenby theTSGs.LTEisdefinedfromRelease8andonwards,whereRelease10ofLTEisthe first version approved by ITU-R as an IMT-Advanced technology and istherefore also the first release named LTE-Advanced. From Release 13, themarketingnameforLTEischangedtoLTE-AdvancedPro.AnoverviewofLTEisgiveninChapter4.FurtherdetailsontheLTEradiointerfacecanbefoundin[28].ThefirstreleaseforNRisin3GPPRelease15.AnoverviewofNRisgivenin

Chapter5withfurtherdetailsthroughoutthisbook.The3GPPTechnicalSpecifications(TS)areorganizedinmultipleseriesand

arenumberedTSXX.YYY,whereXXdenotesthenumberofthespecificationseries and YYY is the number of the specification within the series. Thefollowingseriesofspecificationsdefinetheradioaccesstechnologiesin3GPP:

•25-series:RadioaspectsforUTRA(WCDMA/HSPA);•45-series:RadioaspectsforGSM/EDGE;•36-series:RadioaspectsforLTE,LTE-AdvancedandLTE-AdvancedPro;

•37-series:Aspectsrelatingtomultipleradioaccesstechnologies;•38-series:RadioaspectsforNR.

2.4.2Specificationof5Gin3GPPasanIMT-2020CandidateIn parallel with the definition and evaluation of the next-generation accessinitiated in ITU-R, 3GPP started to define the next-generation 3GPP radioaccess.Aworkshopon5Gradioaccesswasheldin2014andaprocesstodefinethe evaluation criteria for 5G was initiated with a second workshop in early2015. The evaluationwill follow the same process that was usedwhen LTE-Advanced was evaluated and submitted to ITU-R and approved as a 4G

technology as part of IMT-advanced. The evaluation and submission of NRfollowstheITU-RtimelinedescribedinSection2.2.3.3GPPTSGRANdocumentedscenarios,requirements,andevaluationcriteria

forthenew5GradioaccessinreportTR38.913[10]whichisingeneralalignedwith the corresponding ITU-R reports [50,51]. As for the case of the IMT-Advancedevaluation,thecorresponding3GPPevaluationofthenext-generationradioaccesscouldhavealargerscopeandmayhavestricterrequirementsthantheITU-RevaluationofcandidateIMT-2020radiointerfacetechnologiesthatisdefinedbyITU-RWP5D.ThestandardizationworkforNRstartedwithastudy itemphase inRelease

14 and continued with development of a first set of specifications through awork item inRelease 15.A first set of theRelease 15NR specificationswaspublishedinDecember2017andthefullspecificationsareduetobeavailableinmid-2018.FurtherdetailsonthetimeplanandthecontentoftheNRreleasesisgiveninChapter5.3GPPmadeafirstsubmissionofNRasanIMT-2020candidatetotheITU-R

WP5Dmeeting inFebruary2018.NRwas submittedboth as anRITby itselfandasanSRIT(setofcomponentRITs)togetherwithLTE.Thefollowingthreecandidatesweresubmitted,allincludingNRasdevelopedby3GPP:

•3GPPsubmittedacandidatenamed“5G,”containingtwosubmissions:thefirstsubmissionwasanSRITcontainingtwocomponentRITs,thesebeingNRandLTE.ThesecondsubmissionwasaseparateRITbeingNR.

•KoreasubmittedNRasaRIT,withreferenceto3GPP.•ChinasubmittedNRasaRIT,withreferenceto3GPP.

FurthersubmissionstoITU-Rwillbemadeby3GPP,givingmoredetailsofNRas an IMT-2020candidate, according to theprocessdescribed inFig. 2.4.Simulations for the self-evaluations have also started in 3GPP, targeting theevaluationphasein2019.

CHAPTER3

Spectrumfor5G

Abstract

This chapter describes the internation al process for regulating spectrumandhowthepresentIMTspectrumhasbeenassingedbytheITU-R.Basedon the outcome of the most recent WRC, the candidate 5G spectrum ispresentedtogetherwithalltheoperatingbandsspecifiedforNRin3GPP.

KeywordsSpectrum;WRC;IMT;allocation;frequencyband;operatingband;RFexposure

3.1SpectrumforMobileSystemsHistorically, the bands for the first and second generation of mobile serviceswereassignedatfrequenciesaround800–900MHz,butalsoinafewlowerandhigher bands.When 3G (IMT-2000)was rolled out, focuswas on the 2GHzbandandwith thecontinuedexpansionof IMTserviceswith3Gand4G,newbandswereaddedatbothlowerandhigherfrequencies,presentlyspanningfrom450MHztoaround6GHz.Whilenew,previouslyunexploited,frequencybandsare continuously defined for new mobile generations, the bands used forpreviousgenerationsareusedforthenewgenerationaswell.Thiswasthecasewhen3Gand4Gwereintroducedanditwillalsobethecasefor5G.Bands at different frequencies have different characteristics. Due to the

propagation properties, bands at lower frequencies are good for wide-areacoveragedeployments,inurban,suburban,andruralenvironments.Propagationpropertiesofhigher frequenciesmake themmoredifficult touse forwide-areacoverage and, for this reason, higher-frequency bands have to a larger extentbeenusedforboostingcapacityindensedeployments.

Withtheintroductionof5G,thedemandingeMBBusagescenarioandrelatednew services will require even higher data rates and high capacity in densedeployments.Whilemanyearly5Gdeploymentswillbe inbandsalreadyusedfor previous mobile generations, frequency bands above 24 GHz are beinglookedatasacomplement to the frequencybandsbelow6GHz.With the5Grequirements for extreme data rates and localized areas with very high areatrafficcapacitydemands,deploymentusingevenhigherfrequencies,evenabove60 GHz, are considered. Referring to the wavelength, these bands are oftencalledmm-wavebands.New bands are defined continuously by 3GPP, mainly for the LTE

specification,butnowalsoforthenewNRspecifications.Manynewbandsaredefined forNR operation only.Both paired bands,where separated frequencyrangesareassignedforuplinkanddownlink,andunpairedbandswithasingleshared frequency range for uplink and downlink, are included in the NRspecifications. Paired bands are used for Frequency Division Duplex (FDD)operation, while unpaired bands are used for Time Division Duplex (TDD)operation.TheduplexmodesofNRaredescribedfurtherinChapter7.Notethatsome unpaired bands are defined as Supplementary Downlink (SDL) orSupplementaryUplink (SDL)bands.Thesebandsarepairedwith theuplinkordownlink of other bands through carrier aggregation, as described in Section7.6.

3.1.1SpectrumDefinedforIMTSystemsbytheITU-RTheITU-RidentifiesfrequencybandstouseformobileserviceandspecificallyforIMT.ManyofthesewereoriginallyidentifiedforIMT-2000(3G)andnewones camewith the introduction of IMT-Advanced (4G). The identification ishowevertechnologyandgeneration“neutral,”sincetheidentificationisforIMTingeneral, regardlessofgenerationorRadio InterfaceTechnology.TheglobaldesignationsofspectrumfordifferentservicesandapplicationsaredonewithintheITU-RandaredocumentedintheITURadioRegulations[48]andtheuseofIMTbandsgloballyisdescribedinITU-RRecommendationM.1036[46].ThefrequencylistingsintheITURadioRegulations[48]donotdirectlylista

bandforIMT,butratherallocateabandforthemobileservicewithafootnotestating that the band is identified for use by administrations wishing toimplementIMT.Theidentificationismostlybyregion,butisinsomecasesalso

specifiedonaper-countrylevel.Allfootnotesmention“IMT”only,sothereisno specific mentioning of the different generations of IMT. Once a band isassigned, it is thereforeupto theregionalandlocaladministrationstodefineabandforIMTuseingeneralorforspecificgenerations.Inmanycases,regionaland localassignmentsare“technologyneutral”andallowforanykindof IMTtechnology.ThismeansthatallexistingIMTbandsarepotentialbandsforIMT-2020(5G)deploymentinthesamewayastheyhavebeenusedforpreviousIMTgenerations.TheWorld Administrative Radio Congress WARC-92 identified the bands

1885–2025and2110–2200MHzasintendedforimplementationofIMT-2000.Out of these 230 MHz of 3G spectrum, 2× 30 MHz were intended for thesatellitecomponentofIMT-2000andtherestfortheterrestrialcomponent.Partsofthebandswereusedduringthe1990sfordeploymentof2Gcellularsystems,especiallyintheAmericas.Thefirstdeploymentsof3Gin2001–2byJapanandEuropeweredoneinthisbandallocation,andforthatreasonitisoftenreferredtoastheIMT-2000“coreband.”Additional spectrum for IMT-2000 was identified at the World Radio-

communication Conference1 WRC-2000, where it was considered that anadditionalneedfor160MHzofspectrumfor IMT-2000wasforecastedby theITU-R. The identification includes the bands used for 2G mobile systems at806–960and1710–1885MHz,and“new”3Gspectrum in thebandsat2500–2690MHz.The identificationof bandspreviously assigned for 2Gwas also arecognitionoftheevolutionofexisting2Gmobilesystemsinto3G.AdditionalspectrumwasidentifiedatWRC’07forIMT,encompassingbothIMT-2000andIMT-Advanced. The bands added were 450–470, 698–806, 2300–2400, and3400–3600MHz, but the applicability of the bands varies on a regional andnational basis. At WRC’12 there were no additional spectrum allocationsidentifiedforIMT,buttheissuewasputontheagendaforWRC’15.Itwasalsodetermined to study the use of the band 694–790MHz formobile services inRegion1(Europe,MiddleEast,andAfrica).WRC’15wasanimportantmilestonesettingthestagefor5G.Firstanewset

of bands were identified for IMT, where many were identified for IMT on aglobal,orclosetoglobal,basis:

•470–694/698MHz(600MHzband):IdentifiedforsomecountriesinAmericasandtheAsia-Pacific.ForRegion1,itisconsideredforanewagendaitemforIMTatWRC-23.

•694–790MHz(700MHzband):ThisbandisnowalsoidentifiedfullyforRegion1andistherebyaglobalIMTband.

•1427–1518MHz(L-band):Anewglobalbandidentifiedinallcountries.•3300–3400MHz:Globalbandidentifiedinmanycountries,butnotinEuropeorNorthAmerica.

•3400–3600MHz(C-band):Nowaglobalbandidentifiedforallcountries.ThebandwasalreadyallocatedinEurope.

•3600–3700MHz(C-band):Globalbandidentifiedinmanycountries,butnotinAfricaandsomecountiesinAsia-Pacific.InEurope,thebandhasbeenavailablesinceWRC’07.

•4800–4990MHz:NewbandidentifiedforafewcountriesinAsia-Pacific.

Especiallythefrequencyrangefrom3300to4990MHzisofinterestfor5G,sinceitisnewspectruminhigherfrequencybands.Thisimpliesthatitfitswellwith thenewusage scenarios requiringhighdata rates and is also suitable formassive MIMO implementation, where arrays with many elements can beimplementedwithreasonablesize.Sinceitisnewspectrumwithnowidespreaduseformobilesystemstoday,itwillbeeasiertoassignthisspectruminlargerspectrumblocks,therebyenablingwiderRFcarriersandultimatelyhigherend-userdatarates.The second major outcome from WRC’15 concerning IMT was the new

agenda item (1.13) appointed for the next WRC, to identify high-frequencybands above 24GHz for 5Gmobile services. These bandswill be studied byITU-R until 2019 and be considered for IMT identification at WRC’19. Theprimary target for the bands is deployment of IMT-2020. A majority of thebandstobestudiedarealreadytodayassignedtothemobileserviceonaprimarybasis, inmostbands togetherwith fixedand satellite services.Theyconsistofthefollowingbandranges:

•24.25–27.5GHz;•37–40.5GHz;•42.5–43.5GHz;•45.5–47GHz;•47.2–50.2GHz;•50.4–52.6GHz;•66–76GHz;

•81–86GHz.

TherearealsobandstobestudiedforIMTthatarepresentlynotallocatedtothemobileserviceonaprimarybasisandwhereitwillbeinvestigatedwhethertheallocationcanbechangedtoincludemobile:

•31.8–33.4GHz;•40.5–42.5GHz;•47–47.2GHz.

ThecompletesetofbandsisillustratedinFig.3.1.

FIGURE3.1 NewIMTbandsunderstudyinITU-RTG5/1.

ITU-Rhas formed a special taskgroupTG5/1,whichwill conduct sharingand compatibility studies for the new bands and prepare input for WRC’19agendaitem1.13.Thetaskgroupwilldocumentspectrumneeds,technicalandoperational characteristics including protection criteria for existing servicesallocatedinoradjacenttothebandsstudied,basedonthestudies.Asaninputtothestudies,technicalandoperationalcharacteristicsofIMT-2020wereneeded.Thesecharacteristicswereprovidedfrom3GPPascharacteristicsofNR,givenatanearlystageofstandardizationinJanuary2017.Itshouldbenotedthattherearealsoalargenumberofotherfrequencybands

identified formobile services, but not specifically for IMT. These bands areoftenusedalsoforIMTonaregionalornationalbasis.AtWRC’15,therewassomeinterest toalsostudy27.5–29.5GHzfor IMT,but itwasnot included instudies of 5G/IMT-2020 bands. Still, the band is planned for 5G mobilesservices in at least the US and Korea. There was also support for studies of5G/IMT-2020 in the frequency bands below 20 GHz, but those bands wereultimatelynotincluded.Itisexpectedthatseveralbandsintherange6–20GHzwillbeconsidered formobileservices including IMT, inaddition to thebandsstudiedwithinITU-R.Oneexample isanFCCinquiry intonewuse, includingnext-generation wireless broadband services, in the frequency range 5925–

7125MHz.Thesomewhatdivergingarrangementbetweenregionsofthefrequencybands

assigned to IMTmeans that there is not one single band that can be used forroaming worldwide. Large efforts have, however, been put into defining aminimumsetofbandsthatcanbeusedtoprovidetrulyglobalroaming.Inthisway, multiband devices can provide efficient worldwide roaming for devices.With many of the new bands identified at WRC’15 being global or close toglobal,globalroamingismadepossiblefordevicesusingfewerbandsanditalsofacilitateseconomyofscaleforequipmentanddeployment.

3.1.2GlobalSpectrumSituationfor5GThere is a considerable interest globally to make spectrum available for 5Gdeployments.ThisisdrivenbyoperatorsandindustryorganizationssuchastheGlobalmobile SuppliersAssociation [35] andDIGITALEUROPE [29], but isalso supported by regulatory bodies in different countries and regions. Anoverviewof the spectrumsituation for5G is given in [56]. In standardization,3GPPhasfocuseditsactivitiesonbandswhereahighinterestisevident(thefulllist of bands is in Section 3.2). The spectrum of interest can be divided intobandsatlow,medium,andhighfrequencies:Low-frequencybandscorrespondtoexistingLTEbandsbelow2GHz,which

are suitable as a coverage layer, providingwide and deep coverage, includingindoor.Thebandswith highest interest here are the 600 and700MHzbands,whichcorrespond to3GPPNRbandsn71andn28(seeSection3.2 for furtherdetails). Since the bands are not very wide, a maximum of 20MHz channelbandwidthisexpectedinthelow-frequencybands.For early deployment, the 600MHz band is considered for NR in the US,

while the 700MHz band is defined as one of the so-called pioneer bands forEurope. In addition, a number of additional LTE bands in the below 3 GHzrangeareidentifiedforpossible“re-farming”andhavebeenassignedNRbandnumbers. Since the bands are in general already deployed with LTE, NR isexpectedtobedeployedgraduallyatalaterstage.Medium-frequencybandsareintherange3–6GHzandcanprovidecoverage,

capacity, as well as high data rates through the wider channel bandwidthpossible.The highest interest globally is in the range 3300–4200MHz,where3GPPhasdesignatedNRbandsn77andn78.Due to thewiderbands,channelbandwidthsup to100MHzarepossible.Up to200MHzperoperatormaybe

assigned in this frequency range in the longer term,where carrier aggregationcouldthenbeusedtodeploythefullbandwidth.The range 3300–4200MHz is of global interest,with some variations seen

regionally;and3400–3800MHzisapioneerbandinEurope,whileChinaandIndia are planning for 3300–600MHz and in Japan 3600–4200MHz is beingconsidered. Similar frequency ranges are considered inNorthAmerica (3550–3700MHzand initial discussions about 3700–4200MHz),LatinAmerica, theMiddleEast,Africa,India,Australia,etc.Atotalof45countriessigneduptotheIMT identification of the 3300–3400MHz band inWRC-15. There is also alargeamountofinterestforahigherbandinChina(primarily4800–5000MHz)andJapan(4400–4900MHz).Inaddition, thereareanumberofpotentialLTEre-farmingbandsinthe2–6GHzrangethathavebeenidentifiedasNRbands.High-frequencybandsareinthemm-Waverangeabove24GHz.Theywillbe

bestsuitedforhotspotcoveragewithlocallyveryhighcapacityandcanprovideveryhighdatarates.Thehighest interest is in therange24.25–29.5GHz,with3GPPNRbandsn257andn258assigned.Channelbandwidthsupto400MHzare defined for these bands, with even higher bandwidths possible throughcarrieraggregation.The mmWave frequency range is new for IMT deployment, as discussed

above.The band 27.5–28.35was identified at an early stage in theUS,while24.25–27.5GHz,alsocalledthe“26GHzband,”isapioneerbandforEurope,noting that not all of itmay bemade available for 5G.Different parts of thelarger range 24.25–29.5 GHz are being considered globally. The range 27.5–29.5 GHz is the first range planned for Japan and 26.5–29.5 GHz in Korea.Overall,thisbandcanbeseenasglobalwithregionalvariations.Therange37–40 GHz is also planned for the US and similar ranges around 40 GHz areconsideredinmanyotherregionstoo,includingChina.

3.2FrequencyBandsforNRNRcanbedeployedbothinexistingIMTbandsandinfuturebandsthatmaybeidentified atWRC,or in regional bodies.Thepossibility of operating a radio-accesstechnologyindifferentfrequencybandsisafundamentalaspectofglobalmobileservices.Most2G,3G,and4Gdevicesaremultibandcapable,coveringbandsusedinthedifferentregionsoftheworldtoprovideglobalroaming.Fromaradio-accessfunctionalityperspective,thishaslimitedimpactandthephysical-layerspecificationssuchasthoseforNRdonotassumeanyspecificfrequency

band.SinceNRhoweverspanssuchavastrangeoffrequencies,therearecertainprovisionsthatareintendedonlyforcertainfrequencyranges.ThisincludeshowthedifferentNRnumerologiescanbeapplied(seeChapter7).ManyRFrequirementsarespecifiedwithdifferentrequirementsacrossbands.

ThisiscertainlythecaseforNR,butalsoforpreviousgenerations.Examplesofband-specific RF requirements are the allowed maximum transmit power,requirements/limits on out-of-band (OOB) emission and receiver blockinglevels. Reasons for such differences are varying external constraints, oftenimposed by regulatory bodies, in other cases differences in the operationalenvironmentthatareconsideredduringstandardization.ThedifferencesbetweenbandsaremorepronouncedforNRduetothevery

widerangeoffrequencybands.ForNRoperation in thenewmm-Wavebandsabove24GHz,bothdevicesandbasestationswillbeimplementedwithpartlynovel technologyand therewillbeamorewidespreaduseofmassiveMIMO,beam forming, and highly integrated advanced antenna systems. This createsdifferences in how RF requirements are defined, how they are measured forperformanceassessmentandultimatelyalsowhatthelimitsfortherequirementsare set. Frequency bands within the scope of the present Release 15 work in3GPPareforthisreasondividedintotwofrequencyranges:

•Frequencyrange1(FR1)includesallexistingandnewbandsbelow6GHz.

•Frequencyrange2(FR2)includesnewbandsintherange24.25–52.6GHz.

Thesefrequencyrangesmaybeextendedorcomplementedwithnewrangesin future 3GPP releases. The impact of the frequency ranges on the RFrequirementsisfurtherdiscussedinChapter18.ThefrequencybandswhereNRwilloperateareinbothpairedandunpaired

spectra, requiring flexibility in the duplex arrangement. For this reason, NRsupportsbothFDDandTDDoperation.SomerangesarealsodefinedforSDLorSUL.ThesefeaturesarefurtherdescribedinSection7.7.3GPP defines operating bands, where each operating band is a frequency

range for uplink and/or downlink that is specified with a certain set of RFrequirements. The operating bands each have a number, where NR bands arenumbered n1, n2, n3, etc. When the same frequency range is defined as anoperatingbandfordifferentradioaccesstechnologies,thesamenumberisused,

butwritteninadifferentway.4GLTEbandsarewrittenwithArabicnumerals(1,2,3,etc.),while3GUTRAbandsarewrittenwithRomannumerals(I,II,II,etc.).LTEoperatingbandsthatareusedwiththesamearrangementforNRareoftenreferredtoas“LTEre-farmingbands.”Release15ofthe3GPPspecificationsforNRincludes26operatingbandsin

frequency range 1 and three in frequency range 2. Bands for NR have anumberingschemewithassignednumbersfromn1ton512usingthefollowingrules:

1.ForNRinLTEre-farmingbands,theLTEbandnumbersarereusedforNR,justaddingan“n.”

2.NewbandsforNRareassignedthefollowingnumbers:–Therangen65ton256isreservedforNRbandsinfrequencyrange1(someofthesebandscanbeusedforLTEinaddition).

–Therangen257ton512isreservedfornewNRbandsinfrequencyrange2.

The scheme “conserves” band numbers and is backwards compatible withLTE(andUTRA)anddoesnotleadtoanynewLTEnumbersabove256,whichisthepresentmaximumpossible.AnynewLTE-onlybandscanalsobeassignedunusednumbersbelow65.Inrelease15,theoperatingbandsinfrequencyrange1areintherangen1ton84asshowninTable3.1.Thebandsinfrequencyrange2areintherangefromn257ton260,asshowninTable3.2.AllbandsforNRare summarized in Figs. 3.2, 3.3, and 3.4,which also show the correspondingfrequencyallocationdefinedbytheITU-R.

Table3.1

Table3.2

FIGURE3.2 Operatingbandsspecifiedin3GPPrelease15forNRbelow1GHz(inFR1),shownwiththecorrespondingITU-Rallocation.Notfullydrawntoscale.

FIGURE3.3 Operatingbandsspecifiedin3GPPrelease15forNRbetween1GHzand6GHz(inFR1),shownwiththecorrespondingITU-Rallocation.Notfullydrawntoscale.

FIGURE3.4 Operatingbandsspecifiedin3GPPrelease15forNRabove24GHz(inFR2),shownwiththecorrespondingITU-Rallocation,alsoindicatingwhichpartsareforstudyforIMTunderagendaitem1.13.Notfullydrawntoscale.

Some of the frequency bands are partly or fully overlapping. Inmost casesthisisexplainedbyregionaldifferencesinhowthebandsdefinedbytheITU-Rareimplemented.Atthesametime,ahighdegreeofcommonalitybetweenthebands isdesired to enableglobal roaming.Originating inglobal, regional, andlocal spectrum developments, a first set of bands was specified as bands forUTRA. The complete set of UTRA bands later transferred to the LTEspecifications in 3GPP Release 8. Additional bands have been added in laterreleases. In release15,manyof theLTEbandsarenow transferred to theNRspecifications.

3.3RFExposureAbove6GHzWith theexpansionof the frequency ranges for5Gmobile communications tobands above 6 GHz, existing regulations on human exposure to RFelectromagnetic fields (EMFs)mayrestrict themaximumoutputpowerofuserdevicestolevelssignificantlylowerthanwhatareallowedforlowerfrequencies.International RF EMF exposure limits, for example those recommended by

the International Commission on Non-Ionizing Radiation (ICNIRP) and thosespecified by theFederalCommunicationsCommission (FCC) in theUS, havebeensetwithwidesafetymarginstoprotectagainstexcessiveheatingoftissueduetoenergyabsorption.Inthefrequencyrangeof6–10GHz,thebasiclimitschangefrombeingspecifiedasspecificabsorptionrate(W/kg)toincidentpower

density(W/m2).Thisismainlybecausetheenergyabsorptionintissuebecomesincreasinglysuperficialwithincreasingfrequency,andtherebymoredifficulttomeasure.Ithasbeenshownthatforproductsintendedtobeusedincloseproximityto

thebody,therewillbeadiscontinuityinmaximumallowedoutputpowerasthetransition ismade from specific absorption rate to power density-based limits[27].TobecompliantwithICNIRPexposurelimitsatthehigherfrequencies,thetransmitpowermighthave tobeup to10dBbelow thepower levelsused forcurrentcellular technologies.Theexposure limitsabove6GHzappear tohavebeen setwith safetymarginseven larger than thoseusedat lower frequencies,andwithoutanyobviousscientificjustification.Forthelower-frequencybands,largeeffortshavebeenspentovertheyearsto

characterizetheexposureandtosetrelevantlimits.Withagrowinginterestforutilizing frequency bands above 6 GHz for mobile communications, researchefforts are likely to increase which eventually may lead to revised exposurelimits. In the most recent RF exposure standards published by IEEE (C95.1-2005,C95.1-2010a),theinconsistencyatthetransitionfrequencyislessevident.However,theselimitshavenotyetbeenadoptedinanynationalregulationanditisimportantalsothatotherstandardizationorganizationsandregulatorsworktoaddressthisissue.Ifnot,thismighthavealargenegativeimpactoncoverageathigherfrequencies,inparticularforuserequipmentintendedtobeusednearthebody, such aswearables, tablets, andmobile phones, forwhich themaximumtransmitpowermightbeheavilylimitedbythecurrentRFexposureregulations.

1TheWorldAdministrativeRadioConference(WARC)wasreorganizedin1992andbecametheWorldRadio-communicationConference(WRC).

CHAPTER4

LTE—AnOverview

Abstract

In this chapter, an overview of the 4G standardLTE and its evolution isprovided in order to give a background and set the scene for thedesccriptionof5GNR.

KeywordsLTE;LTEAdvanced;LTEAdvancedPro;release8;LTEevolution;License-assistedaccess;LAA;V2VmVX;D2D;sTTI

The focus of this book isNR, the new 5G radio access.Nevertheless, a briefoverviewofLTEasbackgroundtothecomingchaptersisrelevant.Onereasonis that both LTE and NR have been developed by 3GPP and hence have acommon background and share several technology components. Many of thedesignchoicesinNRarealsobasedonexperiencefromLTE.Furthermore,LTEcontinues to evolve in parallelwithNRand is an important component in 5Gradioaccess.ForadetaileddescriptionofLTEsee[28].TheworkonLTEwasinitiatedinlate2004withtheoverallaimofproviding

anewradio-accesstechnologyfocusingonpacket-switcheddataonly.Thefirstrelease of the LTE specifications, release 8, was completed in 2008 andcommercialnetworkoperationbeganinlate2009.Release8hasbeenfollowedbysubsequentLTEreleases,introducingadditionalfunctionalityandcapabilitiesindifferentareas,as illustrated inFig.4.1.Releases10and13areparticularlyinteresting. Release 10 is the first release of LTE-Advanced, and release 13,finalizedinlate2015,isthefirstreleaseofLTE-AdvancedPro.Currently,asofthis writing, 3GPP is working on release 15 which, in addition to NR, alsocontainsafurtherevolutionofLTE.

FIGURE4.1 LTEanditsevolution.

4.1LTERelease8—BasicRadioAccessRelease8isthefirstLTEreleaseandformsthebasisforallthefollowingLTEreleases.InparallelwiththeLTEradioaccessscheme,anewcorenetwork,theEvolvedPacketCore(EPC)wasdeveloped[63].One important requirement imposedon theLTEdevelopmentwas spectrum

flexibility. A range of carrier bandwidths up to and including 20 MHz issupportedforcarrier frequencies frombelow1GHzup toaround3GHz.Oneaspect of spectrum flexibility is the support of both paired and unpairedspectrum using Frequency-Division Duplex (FDD) and Time-Division Duplex(TDD), respectively, with a common design, albeit two different framestructures. The focus of the development work was primarily macronetworkswith above-rooftop antennas and relatively large cells. For TDD, the uplink–downlinkallocationisthereforeinessencestaticwiththesameuplink–downlinkallocationacrossallcells.The basic transmission scheme in LTE is orthogonal frequency-division

multiplexing(OFDM).Thisisanattractivechoiceduetoitsrobustnesstotimedispersion and ease of exploiting both the time and frequency domain.Furthermore, it also allows for reasonable receiver complexity also incombination with spatial multiplexing (MIMO) which is an inherent part ofLTE. Since LTE was primarily designed with macronetworks in mind withcarrierfrequenciesuptoafewGHz,asinglesubcarrierspacingof15kHzandacyclicprefixofapproximately4.7µs1was found tobeagoodchoice. In total1200subcarriersareusedina20MHzspectrumallocation.Fortheuplink,wheretheavailabletransmissionpowerissignificantlylower

thanforthedownlink,theLTEdesignsettledforaschemewithalowpeak-to-average ratio to provide a high power-amplifier efficiency. DFT-precodedOFDM,with the samenumerologyas in thedownlink,was chosen to achievethis. A drawback with DFT-precoded OFDM is the larger complexity on thereceiverside,butgiventhatLTErelease8doesnotsupportspatialmultiplexing

intheuplinkthiswasnotseenasamajorproblem.In the time domain, LTE organizes transmissions into 10-ms frames, each

consisting of ten 1-ms subframes. The subframe duration of 1 ms, whichcorrespondsto14OFDMsymbols,isthesmallestschedulableunitinLTE.Cell-specific reference signals is a cornerstone in LTE. The base station

continuouslytransmitsoneormorereferencesignals(oneperlayer),regardlessofwhethertherearedownlinkdatatotransmitornot.ThisisareasonabledesignforthescenarioswhichLTEwasdesignedfor—relativelylargecellswithmanyuserspercell.Thecell-specificreferencesignalsareusedformanyfunctionsinLTE: downlink channel estimation for coherent demodulation, channel-statereporting for scheduling purposes, correction of device-side frequency errors,initialaccess,andmobilitymeasurements tomention justa few.Thereferencesignaldensitydependsonthenumberoftransmissionlayerssetupinacell,butfor the common case of 2×2MIMO, every third subcarrier in four out of 14OFDMsymbolsinasubframeareusedforreferencesignals.Thus,inthetimedomaintherearearound200µsbetweenreferencesignaloccasions,whichlimitsthepossibilitiestoswitchoffthetransmittertoreducepowerconsumption.Data transmissioninLTEisprimarilyscheduledonadynamicbasis inboth

uplinkanddownlink.Toexploit the typically rapidlyvaryingradioconditions,channel-dependent scheduling can be used. For each 1-ms subframe, thescheduler controls which devices are to transmit or receive and in whatfrequency resources.Differentdata ratescanbeselectedbyadjusting thecoderateoftheTurbocodeaswellasvaryingthemodulationschemefromQPSKupto64-QAM.Tohandletransmissionerrors,fasthybridARQwithsoftcombiningisusedinLTE.Upondownlinkreceptionthedeviceindicatestheoutcomeofthedecoding operation to the base station, which can retransmit erroneouslyreceiveddatablocks.The scheduling decisions are provided to the device through the Physical

DownlinkControlChannel(PDCCH).Iftherearemultipledevicesscheduledinthe same subframe,which is a common scenario, there aremultiplePDCCHs,oneperscheduleddevice.ThefirstuptothreeOFDMsymbolsofthesubframeare used for transmission of downlink control channels. Each control channelspans the full carrier bandwidth, thereby maximizing the frequency diversity.Thisalso implies thatalldevicesmustsupport thefullcarrierbandwidthup tothemaximumvalueof20MHz.Uplinkcontrolsignalingfromthedevices,forexample hybrid-ARQ acknowledgments and channel state information fordownlink scheduling, is carried on the Physical Uplink Control Channel

(PUCCH),whichhasabasicdurationof1ms.Multiantenna schemes, and in particular single-userMIMO, are an integral

partofLTE.AnumberoftransmissionlayersaremappedtouptofourantennasbymeansofaprecodermatrixofsizeNA×NL,where thenumberof layersNL,also known as the transmission rank, is less than or equal to the number ofantennasNA.Thetransmissionrank,aswellastheexactprecodermatrix,canbeselectedby thenetworkbasedonchannel-statusmeasurementscarriedoutandreportedbytheterminal,alsoknownasclosed-loopspatialmultiplexing.Thereis also a possibility to operate without closed-loop feedback for precoderselection. Up to four layers is possible in the downlink although commercialdeployments typically use only two layers. In the uplink only single-layertransmissionispossible.Incaseofspatialmultiplexing,byselectingrank-1transmission,theprecoder

matrix,whichthenbecomesanNA×1precodervector,performsa(single-layer)beamforming function. This type of beamforming can more specifically bereferred to as codebook-based beamforming as the beamforming can only bedoneaccordingtoalimitedsetofpredefinedbeamforming(precoder)vectors.Usingthebasicfeaturesdiscussedabove,LTErelease8is intheorycapable

ofprovidingpeakdata ratesup to150Mbit/s in thedownlinkusing two-layertransmissionin20MHzand75Mbit/sintheuplink.Latency-wiseLTEprovides8msroundtrip timein thehybrid-ARQprotocoland(theoretically) less than5ms one-way delay in the LTE RAN. In practical deployments, includingtransport and core network processing, an overall end-to-end latency of some10msisnotuncommoninwell-deployednetworks.

4.2LTEEvolutionReleases 8 and 9 form the foundation of LTE, providing a highly capablemobile-broadband standard. However, to meet new requirements andexpectations, the releases following the basic ones provide additionalenhancements and features in different areas. Fig. 4.2 illustrates some of themajorareas inwhichLTEhasevolvedover the10yearssince its introductionwithdetailsprovidedinthefollowing.

FIGURE4.2 LTEevolution.

Release10marks thestartof theLTEevolution.Oneof themain targetsofLTE release 10was to ensure that theLTE radio-access technologywould befully compliant with the IMT-Advanced requirements, thus the name LTE-AdvancedisoftenusedforLTErelease10andlater.However,inadditiontotheITUrequirements,3GPPalsodefineditsowntargetsandrequirementsforLTE-Advanced[10].Thesetargets/requirementsextendedtheITUrequirementsbothintermsofbeingmoreaggressiveaswellasincludingadditionalrequirements.Oneimportantrequirementwasbackwardscompatibility.Essentiallythismeansthatanearlier-releaseLTEdeviceshouldbeabletoaccessacarriersupportingLTErelease-10functionality,althoughobviouslynotbeingabletoutilizealltherelease-10 features of that carrier.The principle of backwards compatibility isimportant and has been kept for all LTE releases, but also imposes somerestrictionsontheenhancementspossible;restrictionsthatarenotpresentwhendefininganewstandardsuchasNR.LTE release 10 was completed in late 2010 and introduced enhanced LTE

spectrum flexibility through carrier aggregation, further extendedmultiantennatransmission, support for relaying, and improvements around intercellinterferencecoordinationinheterogeneousnetworkdeployments.Release11furtherextendedtheperformanceandcapabilitiesofLTE.Oneof

themostnotable featuresofLTErelease11, finalized in late2012,was radio-interface functionality for coordinated multipoint (CoMP) transmission andreception. Other examples of improvements in release 11 were carrier-aggregation enhancements, a new control-channel structure (EPDCCH), andperformancerequirementsformoreadvanceddevicereceivers.

Release12wascompleted in2014and focusedon small cellswith featuressuchasdualconnectivity,small-cellon/off,and(semi-)dynamicTDD,aswellason new scenarios with introduction of direct device-to-device communicationandprovisioningofcomplexity-reducedmachine-typecommunication.Release13,finalizedattheendof2015,marksthestartofLTEAdvancedPro.

It is sometimes in marketing dubbed 4.5G and seen as an intermediatetechnologystepbetween4GdefinedbythefirstreleasesofLTEandthe5GNRair interface. License-assisted access to support unlicensed spectra as acomplement to licensed spectra, improved support for machine-typecommunication,andvariousenhancements incarrieraggregation,multiantennatransmission, and device-to-device communication are some of the highlightsfromrelease13.Release14wascompletedinthespringof2017.Apartfromenhancementsto

someofthefeaturesintroducedinearlierreleases,forexampleenhancementstooperation in unlicensed spectra, it introduced support for vehicle-to-vehicle(V2V) and vehicle-to-everything (V2X) communication, as well as wide-areabroadcastsupportwithareducedsubcarrierspacing.Release 15 will be completed in themiddle of 2018. Significantly reduced

latency through the so-called sTTI feature, as well as communication usingaerialsaretwoexamplesofenhancementsinthisrelease.In general, expanding LTE to new use cases beyond traditional mobile

broadbandhasbeeninfocusforthelaterreleasesandtheevolutionwillcontinuealso in thefuture.This isalsoan importantpartof5GoverallandexemplifiesthatLTEremainsimportantandavitalpartoftheoverall5Gradioaccess.

4.3SpectrumFlexibilityAlreadythefirstreleaseofLTEprovidesacertaindegreeofspectrumflexibilityin terms of multibandwidth support and a joint FDD/TDD design. In laterreleasesthisflexibilitywasconsiderablyenhancedtosupporthigherbandwidthsand fragmented spectra using carrier aggregation and access to unlicensedspectraasacomplementusinglicense-assistedaccess(LAA).

4.3.1CarrierAggregationAsmentionedearlier,thefirstreleaseofLTEalreadyprovidedextensivesupportfor deployment in spectrum allocations of various characteristics, with

bandwidths ranging from roughly 1 MHz up to 20 MHz in both paired andunpairedbands.WithLTErelease10thetransmissionbandwidthcanbefurtherextended by means of carrier aggregation (CA), where multiple componentcarriersareaggregatedandjointlyusedfortransmissionto/fromasingledevice.Up to five component carriers, possibly each of different bandwidth, can beaggregatedinrelease10,allowingfortransmissionbandwidthsupto100MHz.Allcomponentcarriersneedtohavethesameduplexschemeand,inthecaseofTDD, uplink–downlink configuration. In later releases, this requirement wasrelaxed.Thenumberofcomponentcarrierspossibletoaggregatewasincreasedto32,resultinginatotalbandwidthof640MHz.Backwardscompatibilitywasensured as each component carrier uses the release-8 structure. Hence, to arelease-8/9 device each component carrier will appear as an LTE release-8carrier, while a carrier-aggregation-capable device can exploit the totalaggregatedbandwidth,enablinghigherdatarates.Inthegeneralcase,adifferentnumberof component carriers canbeaggregated for thedownlinkanduplink.This is an important property from a device complexity point of view whereaggregation can be supported in the downlink where very high data rates areneededwithoutincreasingtheuplinkcomplexity.Componentcarriersdonothavetobecontiguousinfrequency,whichenables

exploitation of fragmented spectra; operatorswith a fragmented spectrum canprovide high-data-rate services based on the availability of a wide overallbandwidth even though they do not possess a single wideband spectrumallocation.Fromabasebandperspective,thereisnodifferencebetweenthecasesinFig.

4.3 and they are all supported by LTE release 10. However, the RF-implementationcomplexityisvastlydifferent,withthefirstcasebeingtheleastcomplex. Thus, although carrier aggregation is supported by the basicspecifications,notalldeviceswillsupportit.Furthermore,release10hassomerestrictionsoncarrieraggregationintheRFspecifications,comparedtowhathasbeen specified for physical layer and related signaling, while in later releasesthere is support for carrier-aggregation within and between a much largernumberofbands.

FIGURE4.3 Carrieraggregation.

Release 11 provided additional flexibility for aggregation of TDD carriers.Priortorelease11,thesamedownlink–uplinkallocationwasrequiredforalltheaggregated carriers. This can be unnecessarily restrictive in the case ofaggregationofdifferentbandsastheconfigurationineachbandmaybegivenbycoexistence with other radio access technologies in that particular band. Aninteresting aspect of aggregating different downlink–uplink allocations is thatthe device may need to receive and transmit simultaneously in order to fullyutilizebothcarriers.Thus,unlikepreviousreleases,aTDD-capabledevicemay,similarlytoaFDD-capabledevice,needaduplexfilter.Release11alsosawtheintroduction of RF requirements for interband and noncontiguous intrabandaggregation, aswell as support for aneven larger setof interbandaggregationscenarios.Release 12 defined aggregations between FDD and TDD carriers, unlike

earlier releases thatonly supportedaggregationwithinoneduplex type.FDD–TDD aggregation allows for efficient utilization of an operator’s spectrumassets.ItcanalsobeusedtoimprovetheuplinkcoverageofTDDbyrelyingonthepossibilityforcontinuousuplinktransmissionontheFDDcarrier.Release13 increased thenumberofcarrierspossible toaggregate from5 to

32,resultinginamaximumbandwidthof640MHzandatheoreticalpeakdatarate around25Gbit/s in thedownlink.Themainmotivation for increasing thenumberofsubcarriersistoallowforverylargebandwidthsinunlicensedspectraaswillbefurtherdiscussedinconjunctionwithlicense-assistedaccessbelow.Carrier aggregation is one of the most successful enhancements of LTE to

datewithnewcombinationsoffrequencybandaddedineveryrelease.

4.3.2License-AssistedAccess

Originally, LTE was designed for licensed spectra where an operator has anexclusivelicenseforacertainfrequencyrange.Alicensedspectrumoffersmanybenefits since the operator can plan the network and control the interferencesituation, but there is typically a cost associated with obtaining the spectrumlicense and the amount of licensed spectra is limited. Therefore, usingunlicensedspectraasacomplementtoofferhigherdataratesandhighercapacityin localareas isof interest.Onepossibility is tocomplement theLTEnetworkwith Wi-Fi, but higher performance can be achieved with a tighter couplingbetween licensed and unlicensed spectra. LTE release 13 therefore introducedlicense-assisted access, where the carrier aggregation framework is used toaggregate downlink carriers in unlicensed frequency bands, primarily in the5GHzrange,withcarriersinlicensedfrequencybandsasillustratedinFig.4.4.Mobility, critical control signaling, and services demanding high quality-of-service relyoncarriers in the licensed spectrawhile (partsof) lessdemandingtraffic can be handled by the carriers using unlicensed spectra. Operator-controlled small-cell deployments are the target. Fair sharing of the spectrumresourceswithothersystems, inparticularWi-Fi, isanimportantcharacteristicofLAAwhich therefore incudesa listen-before-talkmechanism. In release14,license-assisted access was enhanced to address also uplink transmissions.Although the LTE technology standardized in 3GPP supports license-assistedaccess only, where a licensed carrier is needed, there has been work outside3GPP in the MulteFire alliance resulting in a standalone mode-of-operationbasedonthe3GPPstandard.

FIGURE4.4 License-assistedaccess.

4.4MultiAntennaEnhancementsMultiantennasupporthasbeenenhancedover thedifferent releases, increasingthe number of transmission layers in the downlink to eight and introducinguplinkspatialmultiplexingofuptofourlayers.Full-dimensionMIMOandtwo-dimensional beamforming are other enhancements, as is the introduction ofcoordinatedmultipointtransmission.

4.4.1ExtendedMultiAntennaTransmissionInrelease10,downlinkspatialmultiplexingwasexpandedtosupportuptoeighttransmissionlayers.Thiscanbeseenasanextensionoftherelease-9dual-layerbeamforming to support up to eight antenna ports and eight correspondinglayers.Togetherwith thesupport forcarrieraggregation thisenablesdownlinkdata rates up to 3 Gbit/s in 100 MHz of spectra in release 10, increased to

25Gbit/s in release13using32carriers,eight layers spatialmultiplexing,and256QAM.Uplinkspatialmultiplexingofuptofourlayerswasalsointroducedaspartof

LTErelease10.Togetherwiththepossibilityforuplinkcarrieraggregationsthisallows for uplink data rates up to 1.5Gbit/s in 100MHzof spectrum.Uplinkspatialmultiplexingconsistsof a codebook-based schemeunder the controlofthe base station, which means that the structure can also be used for uplinktransmitter-sidebeamforming.Animportantconsequenceof themultiantennaextensionsinLTErelease10

was the introduction of an enhanced downlink reference-signal structure thatmoreextensivelyseparated thefunctionofchannelestimationand thefunctionof acquiring channel-state information. The aim of this was to better enablenovelantennaarrangementsandnewfeaturessuchasmoreelaboratemultipointcoordination/transmissioninaflexibleway.In release 13, and continued in release 14, improved support for massive

antennaarrayswasintroduced,primarilyintermsofmoreextensivefeedbackofchannel-state information. The larger degrees of freedom can be used for, forexample, beamforming in both elevation and azimuth and massive multiuserMIMO where several spatially separated devices are simultaneously servedusing the same time-frequency resource. These enhancements are sometimestermedfull-dimensionMIMOandformastepintomassiveMIMOwithaverylargenumberofsteerableantennaelements.

4.4.2MultipointCoordinationandTransmissionThe first release of LTE included specific support for coordination betweentransmissionpoints,referredtoasInterCellInterferenceCoordination(ICIC),tocontrol the interference between cells. However, the support for suchcoordinationwassignificantlyexpandedaspartofLTErelease11,includingthepossibilityformuchmoredynamiccoordinationbetweentransmissionpoints.In contrast to release 8 ICIC,whichwas limited to the definition of certain

messagesbetweenbasestationstoassistcoordinationbetweencells,therelease11activitiesfocusedonradio-interfacefeaturesanddevicefunctionalitytoassistdifferent coordinationmeans, including the support for channel-state feedbackfor multiple transmission points. Jointly these features and functionality gounder the name Coordinated MultiPoint (CoMP) transmission/reception.Refinement to the reference-signal structurewas also an important part of the

CoMPsupport,aswastheenhancedcontrol-channelstructureintroducedaspartofrelease11,seebelow.Support for CoMP includes multipoint coordination—that is, when

transmissiontoadeviceiscarriedoutfromonespecifictransmissionpointbutwhereschedulingandlinkadaptationarecoordinatedbetweenthetransmissionpoints,aswellasmultipointtransmissioninwhichcasetransmissiontoadevicecanbe carriedout frommultiple transmissionpoints either in such away thatthat transmissioncanswitchdynamicallybetweendifferent transmissionpoints(DynamicPointSelection)orbecarriedout jointly frommultiple transmissionpoints(JointTransmission)(seeFig.4.5).

FIGURE4.5 DifferenttypesofCoMP.

A similar distinction can be made for uplink where one can distinguishbetween (uplink)multipoint coordination andmultipoint reception. In general,uplink CoMP is mainly a network implementation issue and has very littleimpact on the device and very little visibility in the radio-interfacespecifications.TheCoMPworkinrelease11assumed“ideal”backhaul,inpracticeimplying

centralizedbasebandprocessingconnectedtotheantennasitesusinglow-latencyfiberconnections.Extensionstorelaxedbackhaulscenarioswithnon-centralizedbasebandprocessingwereintroducedinrelease12.Theseenhancementsmainlyconsisted of defining newX2messages between base stations for exchanginginformation about so-calledCoMPhypotheses, essentially a potential resourceallocation,andtheassociatedgain/cost.

4.4.3EnhancedControlChannelStructureInrelease11,anewcomplementarycontrolchannelstructurewasintroducedtosupport intercell interference coordination and to exploit the additionalflexibilityof thenew reference-signal structurenot only fordata transmission,

whichwasthecaseinrelease10,butalsoforcontrolsignaling.Thenewcontrol-channel structure can thusbe seen as a prerequisite formanyCoMP schemes,although it is also beneficial for beamforming and frequency-domaininterference coordination as well. It is also used to support narrow-bandoperationforMTCenhancementsinreleases12and13.

4.5Densification,SmallCells,andHeterogeneousDeploymentsSmall cells and dense deployment has been in focus for several releases asmeanstoprovideveryhighcapacityanddatarates.Relaying,small-cellon/off,dynamic TDD, and heterogeneous deployments are some examples ofenhancements over the releases. License-assisted access, discussed in Section4.3.2,isanotherfeatureprimarilytargetingsmallcells.

4.5.1RelayingIn thecontextofLTE,relaying implies that thedevicecommunicateswith thenetworkviaarelaynode that iswirelesslyconnected toadonorcellusing theLTEradio-interfacetechnology(seeFig.4.6).Fromadevicepointofview,therelaynodewillappearasanordinarycell.Thishastheimportantadvantageofsimplifying the device implementation and making the relay node backwardscompatible—thatis,LTErelease-8/9devicescanalsoaccessthenetworkviatherelay node. In essence, the relay is a low-power base station wirelesslyconnectedtotheremainingpartofthenetwork.

FIGURE4.6 Exampleofrelaying.

4.5.2HeterogeneousDeployments

Heterogeneous deployments refer to deployments with a mixture of networknodeswithdifferenttransmitpowerandoverlappinggeographicalcoverage(Fig.4.7). A typical example is a pico node placed within the coverage area of amacrocell. Although such deployments were already supported in release 8,release10introducednewmeanstohandletheinterlayer interferencethatmayoccur between, for example, a pico layer and the overlaid macro. Themultipoint-coordination techniques introduced in release 11 further extend theset of tools for supporting heterogeneous deployments. Enhancements toimprovemobilitybetweenthepicolayerandthemacrolayerwereintroducedinrelease12.

FIGURE4.7 Exampleofheterogeneousdeploymentwithlow-powernodesinsidemacrocells.

4.5.3Small-CellOn/OffIn LTE, cells are continuously transmitting cell-specific reference signals andbroadcasting system information, regardless of the traffic activity in the cell.One reason for this is to enable idle-mode devices to detect the presence of acell; if thereareno transmissionsfromacell there isnothingfor thedevice tomeasure upon and the cellwould therefore not be detected. Furthermore, in alargemacrocelldeployment there isa relativelyhigh likelihoodofat leastonedevice being active in a cell motivating continuous transmission of referencesignals.However, in a dense deployment with many relatively small cells, the

likelihoodofnotallcellsserving thedeviceat thesametimecanberelativelyhigh in some scenarios. The downlink interference scenario experienced by adevicemayalsobemore severewithdevicesexperiencingvery lowsignal-to-interferenceratiosduetointerferencefromneighboring,potentiallyempty,cells,

especiallyifthereisalargeamountofline-of-sightpropagation.Toaddressthis,release 12 introduced mechanisms for turning on/off individual cells as afunctionof the traffic situation to reduce theaverage intercell interferenceandreducepowerconsumption.

4.5.4DualConnectivityDualconnectivityimpliesadeviceissimultaneouslyconnectedtotwocells,seeFig.4.8,asopposed to thebaselinecasewith thedeviceconnected toa singledevice only. User-plane aggregation, where the device is receiving datatransmission from multiple sites, separation of control and user planes, anduplink–downlink separation where downlink transmissions originate from adifferentnodethantheuplinkreceptionnodearesomeexamplesofthebenefitswith dual connectivity. To some extent it can be seen as carrier aggregationextendedtothecaseofnonidealbackhaul.Thedualconnectivityframeworkhasalsoturnedouttobeverypromisingforintegratingotherradio-accessschemessuchasWLANinto3GPPnetworks.ItisalsoessentialforNRwhenoperatinginnon-standalonemodewithLTEprovidingmobilityandinitialaccess.

FIGURE4.8 Exampleofdualconnectivity.

4.5.5DynamicTDDInTDD,thesamecarrierfrequencyissharedinthetimedomainbetweenuplinkand downlink. The fundamental approach to this in LTE, as well as inmanyotherTDDsystems,istostaticallysplittheresourcesintouplinkanddownlink.Havingastaticsplitisareasonableassumptioninlargermacrocellsastherearemultiple users and the aggregated per-cell load in uplink and downlink isrelativelystable.However,withanincreasedinterestinlocal-areadeployments,

TDDisexpectedtobecomemoreimportantcomparedtothesituationforwide-area deployments to date. One reason is unpaired spectrum allocations beingmore common in higher-frequency bands not suitable forwide-area coverage.Another reason is that many problematic interference scenarios in wide-areaTDDnetworksarenotpresentwithbelow-rooftopdeploymentsofsmallnodes.An existing wide-area FDD network could be complemented by a local-arealayerusingTDD,typicallywithlowoutputpowerpernode.Tobetterhandlethehightrafficdynamicsinalocal-areascenario,wherethe

numberofdevicestransmittingto/receivingfromalocal-areaaccessnodecanbevery small, dynamic TDD is beneficial. In dynamic TDD, the network candynamicallyuseresourcesforeitheruplinkordownlinktransmissionstomatchthe instantaneous traffic situation,which leads to an improvement of the end-userperformancecomparedtotheconventionalstaticsplitofresourcesbetweenuplinkanddownlink.Toexploitthesebenefits,LTErelease12includessupportfordynamicTDD,orenhancedInterferenceMitigationandTrafficAdaptation(eIMTA)asistheofficialnameforthisfeaturein3GPP.

4.5.6WLANInterworkingThe 3GPP architecture allows for integrating non-3GPP access, for exampleWLAN, but also cdma2000 [12].Essentially, these solutions connect the non-3GPP access to the EPC and are thus not visible in the LTE radio-accessnetwork. One drawback of this way of WLAN interworking is the lack ofnetwork control; the device may selectWi-Fi even if staying on LTE wouldprovide abetter user experience.One exampleof such a situation iswhen theWi-Fi network is heavily loaded while the LTE network enjoys a light load.Release12thereforeintroducedmeansforthenetworktoassistthedeviceintheselection procedure. Basically, the network configures a signal-strengththresholdcontrollingwhenthedeviceshouldselectLTEorWi-Fi.Release13providedfurtherenhancementsinWLANinterworkingwithmore

explicit control from the LTE RAN on when a device should useWi-Fi andwhen to use LTE. Furthermore, release 13 also includes LTE–WLANaggregationwhereLTEandWLANare aggregated at thePDCP level using aframeworkverysimilartodualconnectivity.

4.6DeviceEnhancements

Fundamentally,adevicevendorisfreetodesignthedevicereceiverinanywayas longas it supports theminimumrequirementsdefined in the specifications.Thereisanincentiveforthevendorstoprovidesignificantlybetterreceiversasthiscouldbedirectlytranslatedintoimprovedend-userdatarates.However,thenetwork may not be able to exploit such receiver improvements to their fullextentasitmightnotknowwhichdeviceshavesignificantlybetterperformance.Networkdeploymentsthereforeneedtobebasedontheminimumrequirements.Defining performance requirements formore advanced receiver types to someextentalleviatesthisastheminimumperformanceofadeviceequippedwithanadvanced receiver is known. Both releases 11 and 12 saw a lot of focus onreceiverimprovementswithcancellationofsomeoverheadsignalsinrelease11andmoregenericschemesinrelease12,includingnetwork-assistedinterferencecancellation (NAICS), where the network can provide the devices withinformationassistingintercellinterferencecancellation.

4.7NewScenariosLTEwasoriginallydesignedasamobilebroadbandsystem,aimingatprovidinghigh data rates and high capacity overwide areas. The evolution of LTE hasaddedfeaturesimprovingcapacityanddatarates,butalsoenhancementsmakingLTEhighlyrelevantalsofornewusecases.Operationinareaswithoutnetworkcoverage,forexampleinadisasterarea,isoneexample,resultinginsupportfordevice-to-devicecomminationbeingincludedintheLTE.Massivemachine-typecommunication,wherealargenumberoflow-costdevices,forexamplesensors,areconnectedtoacellularnetworkisanotherexample.V2V/V2Xandremote-controlleddronesareyetotherexamplesofnewscenarios.

4.7.1Device-To-DeviceCommunicationCellularsystems,suchasLTE,aredesignedassumingthatdevicesconnecttoabasestation tocommunicate. Inmostcases this isanefficientapproachas theserverwith thecontentof interest is typicallynot in thevicinityof thedevice.However, if the device is interested in communicating with a neighboringdevice,orjustdetectingwhetherthereisaneighboringdevicethatisofinterest,thenetwork-centriccommunicationmaynotbethebestapproach.Similarly,forpublicsafety,suchasafirstresponderofficersearchingforpeopleinneedinadisaster situation, there is typically a requirement that communication should

alsobepossibleintheabsenceofnetworkcoverage.Toaddressthesesituations,release12introducednetwork-assisteddevice-to-

device communication using parts of the uplink spectrum (Fig. 4.9). Twoscenarioswereconsideredwhendevelopingthedevice-to-deviceenhancements,incoverageaswellasout-of-coveragecommunicationforpublicsafety,andincoveragediscoveryofneighboringdevicesforcommercialusecases.Inrelease13, device-to-device communication was further enhanced with relayingsolutionsforextendedcoverage.Thedevice-to-devicedesignalsoservedasthebasisfortheV2VandV2Xworkinrelease14.

FIGURE4.9 Device-to-devicecommunciation.

4.7.2Machine-TypeCommunicationMachine-typecommunication(MTC)isaverywideterm,basicallycoveringalltypesofcommunicationbetweenmachines.Althoughspanningawiderangeofdifferent applications,manyofwhichareyetunknown,MTCapplications canbedividedintotwomaincategories,massiveMTCandultrareliablelow-latencycommunication(URLLC).ExamplesofmassiveMTCscenariosaredifferenttypesofsensors,actuators,

and similar devices. These devices typically have to be of very low cost andhaveverylowenergyconsumption,enablingverylongbatterylife.Atthesametime, the amountofdatageneratedbyeachdevice isnormallyvery small andvery low latency is not a critical requirement. URLLC, on the other hand,correspondstoapplicationssuchastrafficsafety/controlorwirelessconnectivityforindustrialprocesses,andingeneralscenarioswhereveryhighreliabilityand

availabilityisrequired,combinedwithlowlatency.TobettersupportmassiveMTC,severalenhancementshavebeenintroduced,

startingwithrelease12andtheintroductionofanew,low-enddevicecategory,category0,supportingdataratesupto1Mbit/s.Apower-savemodeforreduceddevice power consumptionwas also defined.Release 13 further improved theMTC support by defining category-M1 with further extended coverage andsupportfor1.4MHzdevicebandwidth,irrespectiveofthesystembandwidth,tofurtherreducedevicecost.FromanetworkperspectivethesedevicesarenormalLTEdevices,albeitwithlimitedcapabilities,andcanbefreelymixedwithmorecapableLTEdevicesonacarrier.Narrow-band Internet-of-Things (NB-IoT) isaparallelLTE trackcompleted

in release 13. It targets even lower cost and data rates than category-M1,250 kbit/s or less, in a bandwidth of 180 kHz, and even further enhancedcoverage.ThankstotheuseofOFDMwith15-kHzsubcarrierspacing,itcanbedeployed inband on top of an LTE carrier, outband in a separate spectrumallocation,or intheguardbandsofLTE,providingahighdegreeofflexibilityfor an operator. In the uplink, transmission on a single tone is supported toobtain very large coverage for the lowest data rates. NB-IoT uses the samefamily of higher-layer protocols (MAC, RLC, and PDCP) as LTE, withextensionsforfasterconnectionsetupapplicabletobothNB-IoTandcategory-M1,andcanthereforeeasilybeintegratedintoexistingdeployments.Both eMTC and NB-IoT will play an important role in 5G networks for

massivemachine-typecommunication.SpecialmeansfordeployingNRontopofanalready-existingcarrierusedformassivemachine-typecommunicationhasthereforebeenincluded(seeChapter17).Improved support for URLLC has been added in the later LTE releases.

ExampleshereofarethesTTIfeatureinrelease15(seebelow)andthegeneralworkonthereliabilitypartofURLLCinrelease15.

4.7.3LatencyReduction—sTTIIn release 15, work on reducing the overall latency has been carried out,resultingintheso-calledshortTTI(sTTI)feature.Thetargetwiththisfeatureistoprovidevery low latency forusecaseswhere this is important, forexamplefactory automation. It uses similar techniques as used in NR, such as atransmissiondurationofa fewOFDMsymbolsand reduceddeviceprocessingdelay,butincorporatedinLTEinabackwards-compatiblemanner.Thisallows

for low-latency services to be included in existing networks, but also impliescertainlimitationscomparedtoaclean-slatedesignsuchasNR.

4.7.4V2VandV2XIntelligent transportation systems (ITSs) refer to services to improve trafficsafetyand increaseefficiency.Examplesarevehicle-to-vehiclecommunicationforsafety,forexampletotransmitmessagestovehiclesbehindwhenthecarinfront breaks. Another example is platooning where several trucks drive veryclosetoeachotherandfollowthefirsttruckintheplatoon,therebysavingfueland reducing CO2 emissions. Communication between vehicles andinfrastructureisalsouseful,forexampletoobtaininformationaboutthetrafficsituation, weather updates, and alternative routes in case of congestion (Fig.4.10).

FIGURE4.10 IllustrationofV2VandV2X.

Inrelease14,3GPPspecifiedenhancementsinthisarea,basedonthedevice-to-device technologies introduced in release 12 and quality-of-serviceenhancements in the network. Using the same technology for communicationbothbetweenvehiclesandbetweenvehiclesandinfrastructureisattractive,bothtoimprovetheperformancebutalsotoreducecost.

4.7.5AerialsTheworkonaerialsinrelease15coverscommunicationviaadroneactingasarelay to provide cellular coverage in an otherwise noncovered area, but alsoremote control of drones for various industrial and commercial applications.

Sincethepropagationconditionsbetweenthegroundandanairbornedronearedifferentthaninaterrestrialnetwork,newchannelmodelsaredevelopedaspartofrelease15.Theinterferencesituationforadroneisdifferentthanforadeviceon the groups due to the larger number of base stations visible to the drone,calling for interference-mitigation techniques such as beamforming, aswell asenhancementstothepower-controlmechanism.

1Thereisalsoapossibilityfor16.7µsextendedcyclicprefixbutthatoptionisrarelyusedinpractice.

CHAPTER5

NROverview

Abstract

This chapter provides an overview of NR, its design principles, and themostimportanttechnologycomponents.

KeywordsNR;ultra-lean;forwardcompatibility;beam-centric

Fig. 5.1 outlines the timeline for the NR development within 3GPP. Thetechnical work on NRwas initiated in the spring of 2016 as a study item in3GPPrelease14,basedonakick-offworkshopin thefallof2015.Duringthestudyitemphase,different technicalsolutionswerestudied,butgiventhetighttime schedule, some technical decisionswere taken already in this phase.Theworkcontinuedintoaworkitemphaseinrelease15,resultinginthefirstversionof the NR specifications available by the end of 2017, before the closure of3GPP release 15 in mid-2018. The reason for the intermediate release of thespecifications,beforetheendofrelease-15,istomeetcommercialrequirementsonearly5Gdeployments.

FIGURE5.1 3GPPtimeline.

ThefirstspecificationfromDecember2017,whichisthefocusofthisbook,is limited to non-standalone NR operation (see Chapter 6), implying that NR

devices rely on LTE for initial access and mobility. The final release-15specificationssupportstandaloneNRoperationaswell.Thedifferencebetweenstandaloneandnon-standaloneprimarilyaffectshigher layersand the interfacetothecorenetwork;thebasicradiotechnologyisthesameinbothcases.Duringthedevelopmentofrelease15,thefocuswasoneMBBand(tosome

extent) URLLC type of services. For massive machine-type communication(mMTC), LTE-based technologies such as eMTC andNB-IoT [28,58] can beused with excellent results. The support for LTE-based massive MTC on acarrieroverlappingwithanNRcarrierhasbeenaccountedfor in thedesignofNR (see Chapter 17), resulting in an integrated overall system capable ofhandlingaverywiderangeofservices.NativeNRsupportforextendedmMTC,as well as special technology features such as direct device-to-deviceconnectivity, in3GPPreferred toassidelink transmission,willbeaddressed inlaterreleases.InparalleltotheworkontheNRradio-accesstechnologyin3GPP,anew5G

core network has been developed, responsible for functions not related to theradio access but needed for providing a complete network. However, it ispossible to connect theNR radio-access network also to the legacyLTE corenetworkknownastheEvolvedPacketCore(EPC).Infact,thisisthecasewhenoperatingNRinnon-standalonemodewhereLTEandEPChandlefunctionalitylike connection set-up and paging and NR primarily provides a data-rate andcapacity booster. Later releases will introduce standalone operation with NRconnectingtothe5Gcore.Theremainingpartof thischapterprovidesanoverviewofNRradioaccess

includingbasicdesignprinciplesandthemostimportanttechnologycomponentsofNRrelease15.Thechaptercaneitherbereadonitsowntogetahigh-leveloverviewofNR,orasan introductionto thesubsequentChapters6–19,whichprovideadetaileddescriptionoftheNR.ComparedtoLTE,NRprovidesmanybenefits.Someofthemainonesare:

•exploitationofmuchhigher-frequencybandsasameantoobtainadditionalspectratosupportverywidetransmissionbandwidthsandtheassociatedhighdatarates;

•ultra-leandesigntoenhancenetworkenergyperformanceandreduceinterference;

•forwardcompatibilitytoprepareforfuture,yetunknown,usecasesandtechnologies;

•lowlatencytoimproveperformanceandenablenewusecases;and•abeam-centricdesignenablingextensiveusageofbeamformingandamassivenumberofantennaelementsnotonlyfordatatransmission(whichtosomeextentispossibleinLTE)butalsoforcontrol-planeproceduressuchasinitialaccess.

Thefirst threecanbeclassifiedasdesignprinciples (or requirementson thedesign) and will be discussed first, followed by a discussion of the keytechnologycomponentsappliedtoNR.

5.1Higher-FrequencyOperationandSpectrumFlexibilityOnekeyfeatureofNRisasubstantialexpansionintermsoftherangeofspectrain which the radio-access technology can be deployed. Unlike LTE, wheresupportforlicensedspectraat3.5GHzandunlicensedspectraat5GHzarejustbeing introduced,NRsupports licensed-spectrumoperationfrombelow1GHzup to 52.6 GHz1 already from its first release, with extension to unlicensedspectraalsoalreadybeingplannedfor.Operationatmm-wavefrequenciesoffersthepossibilityforlargeamountsof

spectrum and associated verywide transmission bandwidths, thereby enablingveryhigh trafficcapacityandextremedata rates.However,higher frequenciesare also associatedwith higher radio-channel attenuation, limiting the networkcoverage.Although this can partly be compensated for bymeans of advancedmulti-antennatransmission/reception,whichisoneofthemotivatingfactorsforthe beam-centric design in NR, a substantial coverage disadvantage remains,especially in non-line-of-sight and outdoor-to-indoor propagation conditions.Thus, operation in lower-frequency bands will remain a vital component forwirelesscommunicationalsointhe5Gera.Especially,jointoperationinlowerand higher spectra, for example 2 GHz and 28 GHz, can provide substantialbenefits.Ahigher-frequencylayer,withaccesstoalargeamountofspectracanprovideservicetoalargefractionoftheusersdespitethemorelimitedcoverage.Thiswill reduce the load on themore bandwidth-constrained lower-frequencyspectrum,allowingtheuseofthistofocusontheworst-caseusers[66].Anotherchallengewithoperationinhigher-frequencybandsistheregulatory

aspects. For non-technical reasons, the rules defining the allowed radiationchangesat6GHz,fromaSAR-basedlimitationtoamoreEIRP-likelimitation.

Depending on the device type (handheld, fixed, etc.), this may result in areducedtransmissionpower,makingthelinkbudgetmorechallengingthanwhatpropagation conditions alonemay indicate and further stressing the benefit ofcombinedlow-frequency/high-frequencyoperation.

5.2Ultra-LeanDesignAn issue with current mobile-communication technologies is the amount oftransmissionscarriedbynetworknodesregardlessoftheamountofusertraffic.Such signals, sometimes referred to as “always-on” signals, include, forexample,signalsforbase-stationdetection,broadcastofsysteminformation,andalways-on reference signals for channel estimation. Under the typical trafficconditions for which LTEwas designed, such transmissions constitute only aminorpartoftheoverallnetworktransmissionsandthushavearelativelysmallimpactonthenetworkperformance.However,inverydensenetworksdeployedfor high peak data rates, the average traffic load per network node can beexpected to be relatively low, making the always-on transmissions a moresubstantialpartoftheoverallnetworktransmissions.Thealways-ontransmissionshavetwonegativeimpacts:

•theyimposeanupperlimitontheachievablenetworkenergyperformance;and

•theycauseinterferencetoothercells,therebyreducingtheachievabledatarates.

The ultra-lean design principle aims at minimizing the always-ontransmissions, therebyenablinghighernetworkenergyperformanceandhigherachievabledatarates.In comparison, the LTE design is heavily based on cell-specific reference

signals,signalsthatadevicecanassumearealwayspresentanduseforchannelestimation, tracking, mobility measurements, etc. In NR, many of theseprocedureshavebeenrevisitedandmodifiedtoaccountfortheultra-leandesignprinciple.Forexample, thecell-searchprocedureshavebeenredesigned inNRcompared to LTE to support the ultra-lean paradigm. Another example is thedemodulation reference-signal structure where NR relies heavily on referencesignalsbeingpresentonlywhendataaretransmittedbutnototherwise.

5.3ForwardCompatibility

5.3ForwardCompatibilityAn important aim in thedevelopmentof theNRspecificationwas to ensure ahigh degree of forward compatibility in the radio-interface design. In thiscontext, forward compatibility implies a radio-interface design that allows forsubstantial future evolution, in terms of introducing new technology andenablingnewserviceswithyetunknownrequirementsandcharacteristics,whilestillsupportinglegacydevicesonthesamecarrier.Forwardcompatibilityisinherentlydifficulttoguarantee.However,basedon

experience from the evolution of previous generations, 3GPP agreed on somebasicdesignprinciplesrelatedtoNRforwardcompatibilityasquotedfrom[3]:

•Maximizingtheamountoftimeandfrequencyresourcesthatcanbeflexiblyutilizedorthatcanbeleftblankwithoutcausingbackwardcompatibilityissuesinthefuture;

•Minimizingtransmissionofalways-onsignals;•Confiningsignalsandchannelsforphysicallayerfunctionalitieswithinaconfigurable/allocabletime/frequencyresource.

According to the thirdbulletone should, asmuchaspossible, avoidhavingtransmissionsontime/frequencyresourcesfixedbythespecification.Inthiswayoneretainsflexibilityforthefuture,allowingforlaterintroductionofnewtypesoftransmissionswithlimitedconstraintsfromlegacysignalsandchannels.Thisdiffers from the approach taken in LTE where, for example, a synchronoushybrid-ARQprotocolisused,implyingthataretransmissionintheuplinkoccursat a fixedpoint in time after the initial transmission.The control channels arealsovastlymoreflexibleinNRcomparedtoLTEinordernottounnecessarilyblockresources.Note that these design principles partly coincide with the aim of ultra-lean

design as described above. There is also a possibility in NR to configurereservedresources,thatis,time-frequencyresourcesthat,whenconfigured,arenotusedfortransmissionandthusavailableforfutureradio-interfaceextensions.Thesamemechanismcanalsobeused forLTE-NRcoexistence in thecaseofoverlappingLTEandNRcarriers.

5.4TransmissionScheme,BandwidthParts,andFrameStructure

SimilartoLTE[28],OFDMwasfoundtobeasuitablewaveformforNRduetoits robustness to time dispersion and ease of exploiting both the time andfrequencydomainwhendefiningthestructurefordifferentchannelsandsignals.However, unlike LTE where DFT-precoded OFDM is the sole transmissionschemeintheuplink,NRusesconventional,thatis,non-DFT-precodedOFDM,asthebaselineuplinktransmissionschemeduetothesimplerreceiverstructuresincombinationwithspatialmultiplexingandanoveralldesiretohavethesametransmissionschemeinbothuplinkanddownlink.Nevertheless,DFT-precodingcanbeusedasacomplementintheuplinkforsimilarreasonsasinLTE,namelyto enable high power-amplifier efficiency on the device side by reducing thecubicmetric[60].Cubicmetricisameasureoftheamountofadditionalpowerback-offneededforacertainsignalwaveform.Tosupportawiderangeofdeploymentscenarios,fromlargecellswithsub-

1GHzcarrierfrequencyuptomm-wavedeploymentswithverywidespectrumallocations,NRsupportsaflexibleOFDMnumerologywithsubcarrierspacingsrangingfrom15kHzupto240kHzwithaproportionalchangeincyclicprefixduration. A small subcarrier spacing has the benefit of providing a relativelylong cyclic prefix in absolute time at a reasonable overhead while highersubcarrierspacingsareneededtohandle,forexample,theincreasedphasenoiseat higher carrier frequencies. Up to 3300 subcarriers are used although themaximum total bandwidth is limited to 400 MHz, resulting in the maximumcarrier bandwidths of 50/100/200/400 MHz for subcarrier spacings of15/30/60/120kHz, respectively. Ifeven largerbandwidthsare tobesupported,carrieraggregationcanbeused.Although the NR physical-layer specification is band-agnostic, not all

supportednumerologiesarerelevantforall frequencybands(seeFig.5.2).Foreachfrequencyband,radiorequirementsarethereforedefinedforasubsetofthesupported numerologies as illustrated in Fig. 5.2. The frequency range 0.45–6GHziscommonlyreferredtoasfrequencyrange1(FR1)inthespecifications,while the range 24.25–52.6GHz is known as FR2.Currently, there is noNRspectrum identified between 6 GHz and 24.25 GHz. However, the basic NRradio-access technology is spectrum agnostic and the NR specifications caneasilybeextendedtocoveradditionalspectra,forexample,spectrafrom6GHzupto24.25GHz.

FIGURE5.2 SpectraidentifiedforNRandcorrespondingsubcarrierspacings.

In LTE, all devices support the maximum carrier bandwidth of 20 MHz.However,giventheverywidebandwidthspossibleinNR,itisnotreasonabletorequire all devices to support the maximum carrier bandwidth. This hasimplications on several areas and requires a design different from LTE, forexample the design of control channels as discussed later. Furthermore, NRallowsfordevice-sidereceiver-bandwidthadaptationasameans to reduce thedevice energy consumption. Bandwidth adaptation refers to the use of arelatively modest bandwidth for monitoring control channels and receivingmediumdata rates, and dynamically opens up awideband receiver onlywhenneededtosupportveryhighdatarates.To handle these two aspects NR defines bandwidth parts that indicate the

bandwidthoverwhichadeviceiscurrentlyassumedtoreceivetransmissionsofa certain numerology. If a device is capable of simultaneous reception ofmultiplebandwidthsparts,itisinprinciplepossibleto,onasinglecarrier,mixtransmissionsofdifferentnumerologiesforasingledevice,althoughrelease15onlysupportsasingleactivebandwidthpartatatime.TheNR time-domain structure is illustrated in Fig. 5.3 with a 10-ms radio

framedividedintoten1-mssubframes.Asubframeisinturndividedintoslotsconsisting of 14 OFDM symbols each, that is, the duration of a slot inmillisecondsdependsonthenumerology.Forthe15-kHzsubcarrierspacing,anNR slot has a structure that is identical to the structure of an LTE subframe,which isbeneficial fromacoexistenceperspective.Sinceaslot isdefinedasafixednumberofOFDMsymbols,ahighersubcarrierspacingleadstoashorterslot duration. In principle this could be used to support lower-latencytransmission,butasthecyclicprefixalsoshrinkswhenincreasingthesubcarrierspacing,itisnotafeasibleapproachinalldeployments.Therefore,NRsupportsamore efficient approach to low latency by allowing for transmission over afraction of a slot, sometimes referred to as “mini-slot” transmission. Suchtransmissions can also preempt an already ongoing slot-based transmission to

anotherdevice,allowingforimmediatetransmissionofdatarequiringverylowlatency.

FIGURE5.3 Framestructure(TDDassumedinthisexample).

Having the flexibility of starting a data transmission not only at the slotboundaries is also useful when operating in unlicensed spectra. In unlicensedspectrathetransmitteristypicallyrequiredtoensurethattheradiochannelisnotoccupied by other transmissions prior to starting a transmission, a procedurecommonlyknownas“listen-before-talk.”Clearly,once thechannel is foundtobe available it is beneficial to start the transmission immediately, rather thanwaituntilthestartoftheslot,inordertoavoidsomeothertransmitterinitiatingatransmissiononthechannel.Operation in themm-wave domain is another example of the usefulness of

“mini-slot” transmissions as the available bandwidth in such deployments isoftenvery largeandevena fewOFDMsymbolscanbesufficient tocarry theavailable payload. This is of particular use in conjunction with analogbeamforming, discussed below, where transmissions to multiple devices indifferentbeamscannotbemultiplexedinthefrequencydomainbutonlyinthetimedomain.

Unlike LTE, NR does not include cell-specific reference signals but solelyrelies on user-specific demodulation reference signals for channel estimation.Notonlydoesthisenableefficientbeamformingandmulti-antennaoperationasdiscussedbelow,it isalsoinlinewiththeultra-leandesignprincipledescribedabove. In contrast to cell-specific reference signals, demodulation referencesignalsarenot transmittedunless therearedata to transmit, thereby improvingnetworkenergyperformanceandreducinginterference.The overall NR time/frequency structure, including bandwidth parts, is the

topicofChapter7.

5.5DuplexSchemesTheduplexschemetouseistypicallygivenbythespectrumallocationathand.For lower-frequency bands, allocations are often paired, implying frequency-division duplex (FDD) as illustrated in Fig. 5.4. At higher-frequency bands,unpaired spectrum allocations are increasingly common, calling for time-division duplex (TDD). Given the significantly higher carrier frequenciessupportedbyNRcomparedtoLTE,efficientsupportforunpairedspectraisanevenmorecriticalcomponentofNR,comparedtoLTE.

FIGURE5.4 Spectrumandduplexschemes.

NRcanoperateinbothpairedandunpairedspectrausingonecommonframestructure,unlikeLTEwere twodifferent framestructureswereused (and laterexpandedtothreewhensupportforunlicensedspectrawasintroducedinrelease13).ThebasicNRframestructureisdesignedsuchthatitcansupportbothhalf-duplexandfull-duplexoperation.Inhalfduplex,thedevicecannottransmitandreceive at the same time. Examples hereof are TDD and half-duplex FDD. Infull-duplex operation, on the other hand, simultaneous transmission andreceptionispossiblewithFDDasatypicalexample.Asalreadymentioned,TDDincreasesinimportancewhenmovingtohigher-

frequencybandswhereunpairedspectrumallocationsaremorecommon.These

frequencybandsarelessusefulforwide-areacoveragewithverylargecellsdueto their propagation conditions but are highly relevant for local-area coveragewith smaller cell sizes. Furthermore, some of the problematic interferencescenarios in wide-area TDD networks are less pronounced in local areadeployments with lower transmission power and below-rooftop antennainstallations. In such denser deployments with smaller cell sizes, the per-celltrafficvariationsaremorerapidcomparedtolarge-celldeploymentswithalargenumberofactivedevicespercell.Toaddresssuchscenarios,dynamicTDD,thatis, the possibility for dynamic assignment and reassignment of time-domainresourcesbetweenthedownlinkanduplinktransmissiondirections,isakeyNRtechnology component. This is in contrast toLTEwhere the uplink–downlinkallocation does not changeover time.2DynamicTDDenables following rapidtrafficvariationswhichareparticularlypronouncedindensedeploymentswitharelativelysmallnumberofuserspercell.Forexample,ifauseris(almost)aloneinacellandneedstodownloadalargeobject,mostoftheresourcesshouldbeutilized in the downlink direction and only a small fraction in the uplinkdirection.Atalaterpointintime,thesituationmaybedifferentandmostofthecapacityisneededintheuplinkdirection.ThebasicapproachtodynamicTDDisforthedevicetomonitorfordownlink

controlsignalingandfollowtheschedulingdecisions.Ifthedeviceisinstructedto transmit, it transmits in the uplink, otherwise itwill attempt to receive anydownlink transmissions. The uplink–downlink allocation is then completelyunderthecontroloftheschedulerandanytrafficvariationscanbedynamicallytracked. There are deployment scenarios where dynamic TDD may not beuseful,but it ismuchsimpler to restrict thedynamicsofadynamicscheme inthose scenarios when needed rather than trying to add dynamics to afundamentally semistatic design as LTE. For example, in a wide-area macronetworkwithabove-rooftopantennas,theintercellinterferencesituationrequirescoordination of the uplink–downlink allocation between the cells. In suchsituations,asemistaticallocationisappropriatewithoperationalongthelinesofLTE.Thiscanbeobtainedbytheappropriateschedulingimplementation.Thereis also the possibility to semistatically configure the transmission direction ofsome or all of the slots, a feature that can allow for reduced device energyconsumptionas it isnotnecessary tomonitorfordownlinkcontrolchannels inslotsthatareaprioriknowntobereservedforuplinkusage.

5.6Low-LatencySupport

5.6Low-LatencySupportThepossibilityforverylowlatencyisanimportantcharacteristicofNRandhasimpacted many of the NR design details. One example is the use of “front-loaded” reference signals and control signaling, as illustrated in Fig. 5.3. Bylocating the reference signals and downlink control signaling carryingschedulinginformationatthebeginningofthetransmissionandnotusingtime-domain interleaving across OFDM symbols, a device can start processing thereceived data immediately without prior buffering, thereby minimizing thedecoding delay. The possibility for transmission over a fraction of a slot,sometimesreferredtoas“mini-slot”transmission,isanotherexample.Therequirementsonthedevice(andnetwork)processingtimesaretightened

significantly inNRcompared toLTE.Asanexample,adevicehas torespondwith a hybrid-ARQ acknowledgment of approximately one slot (or even lessdependingondevicecapabilities)afterreceivingadownlinkdata transmission.Similarly, the time from grant reception to uplink data transfer is in the samerange.Thehigher-layerprotocolsMACandRLChavealsobeendesignedwithlow

latency in mind with header structures chosen to enable processing withoutknowing the amount of data to transmit (see Chapter 6). This is especiallyimportant in the uplink direction as the device may only have a few OFDMsymbolsafterreceivingtheuplinkgrantuntilthetransmissionshouldtakeplace.Incontrast,theLTEprotocoldesignrequirestheMACandRLCprotocollayersto know the amount of data to transmit before any processing can take place,whichmakessupportforaverylowlatencymorechallenging.

5.7SchedulingandDataTransmissionOnekeycharacteristicofmobileradiocommunicationisthelargeandtypicallyrapid variations in the instantaneous channel conditions stemming fromfrequency-selective fading, distance-dependent path loss, and randominterferencevariationsduetotransmissionsinothercellsandbyotherdevices.Instead of trying to combat these variations, they can be exploited throughchannel-dependent scheduling where the time-frequency resources aredynamically shared between users (see Chapter 14 for details). Dynamicscheduling is used in LTE as well and on a high level, the NR schedulingframework is similar to the one in LTE. The scheduler, residing in the basestation, takes scheduling decisions based on channel-quality reports obtainedfrom the devices. It also takes different traffic priorities andquality-of-service

requirements into account when forming the scheduling decisions sent to thescheduleddevices.Eachdevicemonitorsseveralphysicaldownlinkcontrolchannels(PDCCHs),

typically once per slot, although it is possible to configure more frequentmonitoring to support traffic requiring very low latency. Upon detection of avalid PDCCH, the device follows the scheduling decision and receives (ortransmits)oneunitofdataknownasatransportblockinNR.Inthecaseofdownlinkdata transmission, thedeviceattempts todecodethe

downlink transmission. Given the very high data rates supported by NR,channel-codingdata transmissionisbasedonlow-densityparity-check(LDPC)codes [68]. LDPC codes are attractive from an implementation perspective,especially at higher code rates where they can offer a lower complexity thanTurbocodesasusedinLTE.Hybrid automatic repeat-request (ARQ) retransmission using incremental

redundancy is used where the device reports the outcome of the decodingoperation to the base station (see Chapter 13 for details). In the case oferroneously received data, the network can retransmit the data and the devicecombines the soft information frommultiple transmission attempts. However,retransmitting thewhole transport block could in this case become inefficient.NR therefore supports retransmissions on a finer granularity known as code-block group (CBG). This can also be useful when handling preemption. Anurgent transmission to a second device may use only one or a few OFDMsymbolsandthereforecausehighinterferencetothefirstdeviceinsomeOFDMsymbolsonly.InthiscaseitmaybesufficienttoretransmittheinterferedCBGsonlyandnotthewholedatablock.Handlingofpreemptedtransmissioncanbefurther assisted by the possibility to indicate to the first device the impactedtime-frequency resources such that it can take this information into account inthereceptionprocess.AlthoughdynamicschedulingisthebasicoperationofNR,operationwithout

a dynamic grant can be configured. In this case, the device is configured inadvance with resources that can be used for uplink data transmission (ordownlinkdata reception).Onceadevicehasdata available it can immediatelycommence uplink transmissionwithout going through the scheduling request–grantcycle,therebyenablinglowerlatency.

5.8ControlChannels

Operation ofNR requires a set of physical-layer control channels to carry theschedulingdecisionsinthedownlinkandtoprovidefeedbackinformationintheuplink. A detailed description of the structure of these control channels isprovidedinChapter10.DownlinkcontrolchannelsareknownasPDCCHs(physicaldownlinkcontrol

channels). One major difference compared to LTE is the more flexible time-frequency structure of downlink control channels where PDCCHs aretransmitted in one or more control resource sets (CORESETs) which, unlikeLTEwherethefullcarrierbandwidthisused,canbeconfiguredtooccupyonlypart of the carrier bandwidth. This is needed in order to handle devices withdifferentbandwidthcapabilitiesandalsoinlinewiththeprinciplesforforwardcompatibilityasdiscussedabove.AnothermajordifferencecomparedtoLTEisthe support for beamforming of the control channels, which has required adifferent reference signal design with each control channel having its owndedicatedreferencesignal.Uplinkcontrolinformation,suchashybrid-ARQacknowledgments,channel-

state feedback for multi-antenna operation, and scheduling request for uplinkdata awaiting transmission, are transmitted using the physical uplink controlchannel (PUCCH).There are several different PUCCH formats, depending onthe amount of information and the duration of the PUCCH transmission. Theshort PUCCH is transmitted in the last one or two symbols of a slot and cansupportveryfastfeedbackofhybrid-ARQacknowledgmentsinordertorealizeso-called self-contained slots where the delay from the end of the datatransmission to the receptionof the acknowledgment from thedevice is in theorder of an OFDM symbol, corresponding to a few tens of microsecondsdepending on the numerology used. This can be compared to almost 3ms inLTEandisyetanotherexampleonhowthefocusonlowlatencyhasimpactedtheNRdesign.ForsituationswhenthedurationoftheshortPUCCHistooshortto provide sufficient coverage, there are also possibilities for longer PUCCHdurations.Forthephysical-layercontrolchannels,forwhichtheinformationblocksare

small compared todata transmissionandhybrid-ARQ isnotused,polar codes[17]havebeenselected.For thesmallestcontrolpayloads,Reed–Mullercodesareused.

5.9Beam-CentricDesignandMulti-AntennaTransmission

TransmissionSupport for a for a large number of steerable antenna elements for bothtransmissionand reception is akey featureofNR.Athigher-frequencybands,the large number of antenna elements are primarily used for beamforming toextend coverage,while at lower-frequency bands they enable full-dimensionalMIMO,sometimesreferredtoasmassiveMIMO,andinterferenceavoidancebyspatialseparation.NR channels and signals, including those used for control and

synchronization, have all been designed to support beamforming (Fig. 5.5).Channel-stateinformation(CSI)foroperationofmassivemulti-antennaschemescan be obtained by feedback of CSI reports based on transmission of CSIreference signals in the downlink, as well as using uplink measurementsexploitingchannelreciprocity.

FIGURE5.5 BeamforminginNR.

To provide implementation flexibility, NR is deliberately supportingfunctionality to support analog beamforming a well as digitalprecoding/beamforming (see Chapter 11). At high frequencies, analogbeamforming,wherethebeamisshapedafterdigital-to-analogconversion,maybe necessary from an implementation perspective, at least initially. Analogbeamformingresultsintheconstraintthatareceiveortransmitbeamcanonlybeformed in one direction at a given time instant and requires beam-sweepingwhere the same signal is repeated inmultipleOFDMsymbols but in differenttransmit beams. By having beam-sweeping possibility, it is ensured that anysignal can be transmitted with a high gain, narrow beam to reach the entireintendedcoveragearea.

Signaling to support beam-management procedures is specified, such as anindication to the device to assist selection of a receive beam (in the case ofanalog receive beamforming) to be used for data and control reception. For alarge number of antennas, beams are narrow and beam tracking can fail,thereforebeam-recoveryprocedureshavealsobeendefinedwhereadevicecantrigger a beam-recovery procedure. Moreover, a cell may have multipletransmission points, each with beams and the beam-management proceduresallowfordevicetransparentmobilityforseamlesshandoverbetweenthebeamsof different points. Additionally, uplink-centric and reciprocity-based beammanagementispossiblebyutilizinguplinksignals.With the use of amassivenumber of antenna elements for lower-frequency

bands, the possibility to separate users spatially increases both in uplink anddownlink, but requires that the transmitter has channel knowledge. For NR,extendedsupportforsuchmulti-userspatialmultiplexingisintroduced,eitherbyusing a high-resolution channel-state-information feedback using a linearcombinationofDFTvectors,oruplinksoundingreferencesignalstargetingtheutilizationofchannelreciprocity.Twelveorthogonaldemodulationreferencesignalsarespecifiedformulti-user

MIMOtransmissionpurposes,whileanNRdevicecanmaximallyreceiveeightMIMO layers in the downlink and up to four layers in the uplink.Moreover,additionalconfigurationofaphasetrackingreferencesignalisintroducedinNRsincetheincreasedphasenoisepowerathighcarrierfrequencybandsotherwisewilldegradedemodulationperformanceforlargermodulationconstellations,forexample64QAM.In addition, NR is prepared to support distributed MIMO, although the

supportisnotcompleteinrelease15.DistributedMIMOimpliesthatthedevicecan receivemultiple independent physical data shared channels (PDSCHs)perslottoenablesimultaneousdatatransmissionfrommultipletransmissionpointsto thesameuser. Inessence,someMIMOlayersare transmittedfromonesitewhileotherlayersaretransmittedfromanothersite.Multi-antenna transmission ingeneral, aswell asamoredetaileddiscussion

on NR multi-antenna precoding, is described in Chapter 11 with beammanagementbeingthesubjectofChapter12.

5.10InitialAccessInitial access is the procedures allowing a device to find a cell to camp on,

receive thenecessary system information, and to request a connection throughrandomaccess.ThebasicstructureofNRinitialaccess,describedinChapter16,issimilartothecorrespondingfunctionalityofLTE[28]:

•Thereisapairofdownlinksignals,theprimarysynchronizationsignal(PSS)andthesecondarysynchronizationsignal(SSS),thatisusedbydevicestofind,synchronizeto,andidentifyanetwork;

•Thereisadownlinkphysicalbroadcastchannel(PBCH)transmittedtogetherwiththePSS/SSS.ThePBCHcarriesaminimumamountofsysteminformationincludinganindicationwheretheremainingbroadcastsysteminformationistransmitted.InthecontextofNR,thePSS,SSS,andPBCHarejointlyreferredtoasasynchronizationsignal(SS)block;

•Thereisafour-stagerandom-accessprocedure,commencingwiththeuplinktransmissionofarandom-accesspreamble.

However,therearesomeimportantdifferencesbetweenLTEandNRintermsofinitialaccess.Thesedifferencescomemainlyfromtheultra-leanprincipleandthebeam-centricdesign,bothofwhichimpacttheinitialaccessproceduresandpartlyleadtodifferentsolutionscomparedtoLTE.InLTE,thePSS,SSS,andPBCHarelocatedatthecenterofthecarrierand

are transmitted once every 5 ms. Thus, by dwelling on each possible carrierfrequency during at least 5ms, a device is guaranteed to receive at least onePSS/SSS/PBCH transmission if a carrier exists at the specific frequency.Without any a priori knowledge a device must search all possible carrierfrequenciesoveracarrierrasterof100kHz.ToenablehigherNRnetworkenergyperformanceinlinewiththeultra-lean

principle,theSSblockis,bydefault,transmittedonceevery20ms.Duetothelonger period between consecutive SS blocks, compared to the correspondingsignals/channelsinLTE,adevicesearchingforNRcarriersmustdwelloneachpossible frequency for a longer time. To reduce the overall search timewhilekeeping the device complexity comparable to LTE, NR supports a sparsefrequencyraster forSSblock.Thisimpliesthat thepossiblefrequency-domainpositionsoftheSSblockcouldbesignificantlysparser,comparedtothepossiblepositionsofanNRcarrier(thecarrierraster).Asaconsequence,theSSblockwilltypicallynotbelocatedatthecenteroftheNRcarrier,whichhasimpactedtheNRdesign.

ThesparseSS-blockrasterenablesasignificantlyreducedtimeforinitialcellsearch,atthesametimeasthenetworkenergyperformancecanbesignificantlyimprovedduetothelongerSS-blockperiod.Network-side beam-sweeping is supported for both downlink SS-slock

transmission and uplink random-access reception as a means to improvecoverage,especiallyinthecaseofoperationathigherfrequencies.ItisimportanttorealizethatbeamsweepingisapossibilityenabledbytheNRdesign.Itdoesnot imply that it must be used. Especially at lower carrier frequencies, beamsweepingmaynotbeneeded.

5.11InterworkingandLTECoexistenceAsitisdifficulttoprovidefullcoverageathigherfrequencies,interworkingwithsystems operating at lower frequencies is important. In particular, a coverageimbalancebetweenuplinkanddownlinkisacommonscenario,especiallyiftheyareindifferentfrequencybands.Thehighertransmitpowerforthebasestationcompared to the mobile device results in the downlink achievable data ratesoftenbeingbandwidthlimited,makingitmorerelevanttooperatethedownlinkinahigherspectrumwherewiderbandwidthmaybeavailable. Incontrast, theuplink is more often power-limited, reducing the need for wider bandwidth.Instead,higherdata ratesmaybeachievedon lower-frequencyspectra,despitetherebeinglessavailablebandwidth,duetolessradio-channelattenuation.Through interworking, ahigh-frequencyNRsystemcancomplement a low-

frequencysystem(seeChapter17fordetails).Thelower-frequencysystemcanbe eitherNRorLTE, andNR supports interworkingwith either of these.Theinterworking can be realized at different levels, including intra-NR carrieraggregation, dual connectivity3 with a common packet data convergenceprotocol(PDCP)layer,andhandover.However, the lower-frequency bands are often already occupied by current

technologies, primarily LTE. Furthermore, an additional low-frequencyspectrum is planned to be deployed with LTE in the relatively near future.LTE/NRspectrumcoexistence, that is, thepossibilityforanoperator todeployNRinthesamespectrumasanalreadyexistingLTEdeploymenthasthereforebeen identified as a way to enable early NR deployment in lower-frequencyspectrawithoutreducingtheamountofspectrumavailabletoLTE.Two coexistence scenarios were identified in 3GPP and guided the NR

design:

•Inthefirstscenario,illustratedintheleftpartofFig.5.6,thereisLTE/NRcoexistenceinbothdownlinkanduplink.Notethatthisisrelevantforbothpairedandunpairedspectraalthoughapairedspectrumisusedintheillustration.

•Inthesecondscenario,illustratedintherightpartofFig.5.6,thereiscoexistenceonlyintheuplinktransmissiondirection,typicallywithintheuplinkpartofalower-frequencypairedspectrum,withNRdownlinktransmissiontakingplaceinthespectrumdedicatedtoNR,typicallyathigherfrequencies.Thisscenarioattemptstoaddresstheuplink–downlinkimbalancediscussedabove.NRsupportsasupplementaryuplink(SUL)tospecificallyhandlethisscenario.

FIGURE5.6 ExampleofNR–LTEcoexistence.

The possibility for an LTE-compatible NR numerology based on 15-kHzsubcarrierspacing,enablingidenticaltime/frequencyresourcegridsforNRandLTE, is one of the fundamental tools for such coexistence. The flexible NRscheduling with a scheduling granularity as small as one symbol can then beusedtoavoidscheduledNRtransmissionstocollidewithkeyLTEsignals,suchascell-specificreferencesignals,CSI-RS,andthesignals/channelsusedforLTEinitial access. Reserved resources, introduced for forward compatibility (seeSection 5.3), can also be used to further enhance NR–LTE coexistence. It ispossible to configure reserved resources matching the cell-specific referencesignalsinLTE,therebyenablinganenhancedNR–LTEoverlayinthedownlink.

1Theupperlimitof52.6GHzisduetosomeveryspecificspectrumsituations.2InlaterLTEreleases,theeIMTAfeaturesallowssomedynamicsintheuplink–downlinkallocation.3IntheDecemberversionofrelease15,dualconnectivityisonlysupportedbetweenNRandLTE.DualconnectivitybetweenNRandNRispartofthefinalJune2018release15.

CHAPTER6

Radio-InterfaceArchitecture

Abstract

This chapter described the overall NR architecture. DIfferent alternativesfor connecting the NR RAN to the core netwrok (EPC or 5GCN) arediscussed.Theoverallprotocolstructureandthedifferentchanneltypesarealsooutlined.

Keywords5GCN;EPC;architecture;dualconnectivity;gNB;RAN;userplane;controlplane;protocolarchitecture;paging

ThischaptercontainsabriefoverviewoftheoverallarchitectureofanNRradio-accessnetworkandtheassociatedcorenetwork,followedbydescriptionsoftheradio-accessnetworkuser-planeandcontrol-planeprotocols.

6.1OverallSystemArchitectureInparalleltotheworkontheNR(NewRadio)radio-accesstechnologyin3GPP,theoverall systemarchitectures of both theRadio-AccessNetwork (RAN) andthe Core Network (CN) were revisited, including the split of functionalitybetweenthetwonetworks.The RAN is responsible for all radio-related functionality of the overall

network including, for example, scheduling, radio-resource handling,retransmission protocols, coding, and various multi-antenna schemes. Thesefunctionswillbediscussedindetailinthesubsequentchapters.The 5G core network is responsible for functions not related to the radio

accessbutneededforprovidingacompletenetwork.Thisincludes,forexample,authentication, charging functionality, and setup of end-to-end connections.Handlingthesefunctionsseparately,insteadofintegratingthemintotheRAN,is

beneficialasitallowsforseveralradio-accesstechnologiestobeservedbythesamecorenetwork.However, it is possible to connect theNR radio-access network also to the

legacyLTE(Long-TermEvolution)corenetworkknownastheEvolvedPacketCore(EPC).Infact,thisisthecasewhenoperatingNRinnon-standalonemode,where LTE and EPC handle functionality like connection set-up and paging.LaterreleaseswillintroducestandaloneoperationwithNRconnectingtothe5Gcore,aswellasLTEconnecting to the5Gcore.Thus, theLTEandNRradio-accessschemesandtheircorrespondingcorenetworksarecloselyrelated,unlikethetransitionfrom3Gto4Gwherethe4GLTEradio-accesstechnologycannotconnecttoa3Gcorenetwork.Although thisbook focuseson theNRradioaccess, abriefoverviewof the

5G core network, as well as how it connects to the RAN, is useful as abackground.

6.1.15GCoreNetworkThe5GcorenetworkbuildsupontheEPCwiththreenewareasofenhancementcompared toEPC: service-based architecture, support for network slicing, andcontrol-plane/user-planesplit.Aservice-basedarchitectureisthebasisforthe5Gcore.Thismeansthatthe

specification focuses on the services and functionalities provided by the corenetwork,ratherthannodesassuch.Thisisnaturalasthecorenetworktodayisalreadyoftenhighlyvirtualizedwith thecorenetworkfunctionalityrunningongenericcomputerhardware.Network slicing is a term commonly seen in the context of 5G.A network

slice is a logical network serving a certain business or customer need andconsists of the necessary functions from the service-based architectureconfigured together. For example, one network slice can be set up to supportmobile broadband applications with full mobility support, similar to what isprovided by LTE, and another slice can be set up to support a specific non-mobile,latency-criticalindustry-automationapplication.Thesesliceswillallrunonthesameunderlyingphysicalcoreandradionetworks,but,fromtheend-userapplicationperspective,theyappearasindependentnetworks.Inmanyaspectsitis similar to configuring multiple virtual computers on the same physicalcomputer.Edgecomputing,wherepartsoftheend-userapplicationrunclosetothecorenetworkedgetoprovidelowlatency,canalsobepartofsuchanetwork

slice.Control-plane/user-plane split is emphasized in the 5G core network

architecture, including independent scaling of the capacity of the two. Forexample,ifmorecontrolplanecapacityisneed,itshouldbestraightforwardtoadditwithoutaffectingtheuser-planeofthenetwork.Onahighlevel,the5GcorecanbeillustratedasshowninFig.6.1.Thefigure

usesaservice-basedrepresentation,wheretheservicesandfunctionalitiesareinfocus. In the specifications there is also an alternative, reference-pointdescription,focusingonthepoint-to-pointinteractionbetweenthefunctions,butthatdescriptionisnotcapturedinthefigure.

FIGURE6.1 High-levelcorenetworkarchitecture(service-baseddescription).

Theuser-planefunctionconsistsoftheUserPlaneFunction(UPF)whichisagateway between the RAN and external networks such as the Internet. Itsresponsibilities include packet routing and forwarding, packet inspection,quality-of-servicehandlingandpacketfiltering,andtrafficmeasurements.Italsoservesasananchorpointfor(inter-RAT)mobilitywhennecessary.Thecontrol-planefunctionsconsistofseveralparts.TheSessionManagement

Function (SMF) handles, among other functions, IP address allocation for thedevice(alsoknownasUserEquipment,UE),controlofpolicyenforcement,andgeneral session-management functions. TheAccess andMobilityManagementFunction(AMF)isinchargeofcontrolsignalingbetweenthecorenetworkandthe device, security for user data, idle-state mobility, and authentication. Thefunctionality operating between the core network,more specifically theAMF,

and the device is sometimes referred to as theNon-Access Stratum (NAS), toseparateitfromtheAccessStratum(AS),whichhandlesfunctionalityoperatingbetweenthedeviceandtheradio-accessnetwork.In addition, the core network can also handle other types of functions, for

example, thePolicy Control Function (PCF) responsible for policy rules, theUnifiedDataManagement(UDM)responsibleforauthenticationcredentialsandaccessauthorization,theNetworkExposureFunction(NEF),theNRRepositoryFunction (NRF), the Authentication Server Function (AUSF) handingauthenticationfunctionality,andtheApplicationFunction(AF).Thesefunctionsarenotdiscussedfurtherinthisbookandthereaderisreferredto[13]forfurtherdetails.It should be noted that the core network functions can be implemented in

many ways. For example, all the functions can be implemented in a singlephysical node, distributed across multiple nodes, or executed on a cloudplatform.The description above focused on the new 5G core network, developed in

paralleltotheNRradioaccessandcapableofhandlingbothNRandLTEradioaccesses. However, to allow for an early introduction of NR in existingnetworks,itisalsopossibletoconnectNRtoEPC,theLTEcorenetwork.Thisis illustrated as “option 3” in Fig. 6.2 and is also known as “non-standaloneoperation”asLTEisusedforcontrol-planefunctionalitysuchasinitialaccess,paging, and mobility. The nodes denoted eNB and gNB will be discussed inmoredetailinthenextsection;forthetimebeingeNBandgNBcanbethoughtofasbasestationsforLTEandNR,respectively.

FIGURE6.2 Differentcombinationsofcorenetworksandradio-accesstechnologies.

Inoption3,theEPCcorenetworkisconnectedtotheeNB.Allcontrol-planefunctionsarehandledbyLTE,andNRisusedonlyfortheuser-planedata.ThegNB is connected to the eNB and user-plane data from the EPC can beforwardedfromtheeNBto thegNB.Therearealsovariantsof this:option3aandoption3x.Inoption3a, theuser-planepartsofboththeeNBandgNBare

directly connected to the EPC. In option 3x, only the gNB user plane isconnectedtotheEPCanduser-planedatatotheeNBareroutedviathegNB.For standalone operation, the gNB is connected directly to the 5G core as

showninoption2.Bothuser-planeandcontrol-planefunctionsarehandledbythegNB.Options4,5,and7showvariouspossibilitiesforconnectinganLTEeNBtothe5GCN.

6.1.2Radio-AccessNetworkTheradio-accessnetworkcanhavetwotypesofnodesconnectedtothe5Gcorenetwork:

•AgNB,servingNRdevicesusingtheNRuser-planeandcontrol-planeprotocols;or

•Anng-eNB,servingLTEdevicesusingtheLTEuser-planeandcontrol-planeprotocols.1

Aradio-accessnetworkconsistingofbothng-eNBsforLTEradioaccessandgNBsforNRradioaccessisknownasanNG-RAN,althoughthetermRANwillbeusedinthefollowingforsimplicity.Furthermore,itwillbeassumedthattheRANisconnectedtothe5Gcoreandhence5Gterminology,suchasgNB,willbeused.Inotherwords,thedescriptionwillassumea5GcorenetworkandanNR-based RAN as shown in option 2 in Fig. 6.2. However, as alreadymentioned,thefirstversionofNRoperatesinnon-standalonemodewhereNRisconnected to the EPC using option 3. The principles are in this case similar,althoughthenamingofthenodesandinterfacesdiffersslightly.ThegNB(orng-eNB)isresponsibleforall radio-relatedfunctions inoneor

several cells, for example, radio resource management, admission control,connection establishment, routing of user-plane data to the UPF and control-plane information to theAMF, andQoS flowmanagement. It is important tonote that an gNB is a logical node and not a physical implementation. OnecommonimplementationofangNBisathree-sectorsite,whereabasestationishandling transmissions in three cells, although other implementations can befound aswell, such as one baseband processing unit towhich several remoteradioheadsareconnected.Examplesof thelatterarea largenumberof indoorcells,orseveralcellsalongahighway,belongingtothesamegNB.Thus,abasestationisapossibleimplementationof,butnotthesameas,agNB.

AscanbeseeninFig.6.3, thegNBisconnectedtothe5Gcorenetworkbymeansof theNG interface,more specifically to theUPFbymeansof theNGuser-planepart(NG-u),andtotheAMFbymeansoftheNGcontrol-planepart(NG-c).OnegNBcanbeconnectedtomultipleUPFs/AMFsforthepurposeofloadsharingandredundancy.

FIGURE6.3 Radio-accessnetworkinterfaces.

TheXn interface, connectinggNBs to eachother, ismainlyused to supportactive-modemobilityanddualconnectivity.ThisinterfacemayalsobeusedformulticellRadio ResourceManagement (RRM) functions. The Xn interface isalso used to support losslessmobility between neighboring cells bymeans ofpacketforwarding.ThereisalsoastandardizedwaytosplitthegNBintotwoparts,acentralunit

(gNB-CU)andoneormoredistributedunits(gNB-DU)usingtheF1interface.In thecaseofasplitgNB, theRRC,PDCP,andSDAPprotocols,describedinmore detail below, reside in the gNB-CU and the remaining protocol entities(RLC,MAC,PHY)inthegNB-DU.TheinterfacebetweenthegNB(orthegNB-DU)andthedeviceisknownas

theUuinterface.Foradevicetocommunicate,atleastoneconnectionbetweenthedeviceand

the network is required. As a baseline, the device is connected to one cellhandlingall theuplinkaswellasdownlink transmissions.Alldata flows,userdata aswell as RRC signaling, are handled by this cell. This is a simple androbustapproach, suitable forawide rangeofdeployments.However, allowingthedevicetoconnecttothenetworkthroughmultiplecellscanbebeneficialinsome scenarios. One example is user-plane aggregation, where flows frommultiplecellsareaggregatedinordertoincreasethedatarate.Anotherexampleiscontrol-plane/user-planeseparationwherethecontrolplanecommunicationishandled by one node and the user plane by another. The scenario of a deviceconnectedtotwocells2isknownasdualconnectivity.DualconnectivitybetweenLTEandNRisofparticularimportanceasitisthe

basisfornon-standaloneoperationusingoption3asillustratedinFig.6.4.TheLTE-based master cell handles control-plane and (potentially) user-planesignaling,andtheNR-basedsecondarycellhandlesuser-planeonly,inessenceboostingthedatarates.

FIGURE6.4 LTE–NRdualconnectivityusingoption3.

Dual connectivity between NR and NR is not part of the December 2017versionofrelease15butispossibleinthefinalJune2018versionofrelease15.

6.2Quality-Of-ServiceHandlingHandlingofdifferentquality-of-service (QoS) requirements ispossiblealreadyinLTE,andNRbuildsuponandenhancesthisframework.ThekeyprinciplesofLTEarekept,namelythatthenetworkisinchargeoftheQoScontrolandthatthe5Gcore networkbut not the radio-access network is awareof the service.QoShandlingisessentialfortherealizationofnetworkslicing.Foreachconnecteddevice,thereisoneormorePDUsessions,eachwithone

ormoreQoS flows anddata radio bearers.The IPpackets aremapped to theQoSflowsaccordingtotheQoSrequirements,forexampleintermsofdelayorrequired data rate, as part of theUDF functionality in the core network.Eachpacket can bemarkedwith aQoS Flow Identifier (QFI) to assist uplinkQoShandling.Thesecondstep,mappingofQoSflowstodataradiobearers,isdonein the radio-access network. Thus, the core network is aware of the servicerequirements,whiletheradio-accessnetworkonlymapstheQoSflowstoradiobearers.TheQoS-flow-to-radio-bearermapping isnotnecessarilyaone-to-onemapping;multipleQoSflowscanbemappedtothesamedataradiobearer(Fig.6.5).

FIGURE6.5 QoSflowsandradiobearersduringaPDUsession.

Therearetwowaysofcontrollingthemappingfromquality-of-serviceflowstodataradiobearersintheuplink:reflectivemappingandexplicitconfiguration.

In the case of reflective mapping, which is a new feature in NR whenconnectedtothe5Gcorenetwork,thedeviceobservestheQFIinthedownlinkpackets for the PDU session. This provides the device with knowledge aboutwhichIPflowsaremappedtowhichQoSflowandradiobearer.Thedevicethenusesthesamemappingfortheuplinktraffic.In the case of explicit mapping, the quality-of-service flow to data radio

bearermappingisconfiguredinthedeviceusingRRCsignaling.

6.3RadioProtocolArchitectureWiththeoverallnetworkarchitectureinmind,theRANprotocolarchitectureforthe user and control planes can be discussed. Fig. 6.6 illustrates the RANprotocolarchitecture(theAMFis,asdiscussedintheprevioussection,notpartoftheRANbutisincludedinthefigureforcompleteness).

FIGURE6.6 User-planeandcontrolplaneprotocolstack.

In the following, the user-plane protocols will be described in Section 6.4,followedbythecontrolplaneprotocolsinSection6.5.AsseeninFig.6.6,manyof the protocol entities are common to the user and control planes and hencePDCP,RLC,MAC,andPHYwillonlybedescribedintheuser-planesection.

6.4User-PlaneProtocolsAgeneraloverviewoftheNRuser-planeprotocolarchitectureforthedownlinkisillustratedinFig.6.7.ManyoftheprotocollayersaresimilartothoseinLTE,althoughtherearesomedifferencesaswell.Oneofthedifferencesisthequality-of-service handling in NR when connected to a 5G core network, where the

SDAP protocol layer accepts one or more QoS flows carrying IP packetsaccording to theirQuality-of-Service requirements. In the caseof theNRuserplaneconnectedtotheEPC,theSDAPisnotused.

FIGURE6.7 NRdownlinkuser-planeprotocolarchitectureasseenfromthedevice.

As will become clear in the subsequent discussion, not all the entitiesillustrated inFig.6.7areapplicable inall situations.Forexample,ciphering isnotusedforbroadcastingof thebasicsysteminformation.TheuplinkprotocolstructureissimilartothedownlinkstructureinFig.6.7,althoughtherearesomedifferences with respect to, for example, transport-format selection and thecontroloflogical-channelmultiplexing.The different protocol entities of the radio-access network are summarized

belowanddescribedinmoredetailinthefollowingsections.

•ServiceDataApplicationProtocol(SDAP)isresponsibleformappingQoSbearerstoradiobearersaccordingtotheirquality-of-servicerequirements.ThisprotocollayerisnotpresentinLTEbutintroducedinNRwhenconnectingtothe5Gcorenetworkduetothenewquality-of-servicehandling.

•PacketDataConvergenceProtocol(PDCP)performsIPheadercompression,ciphering,andintegrityprotection.Italsohandlesretransmissions,in-sequencedelivery,andduplicateremoval3inthecaseofhandover.Fordualconnectivitywithsplitbearers,PDCPcanprovideroutingandduplication.ThereisonePDCPentityperradiobearerconfiguredforadevice.

•Radio-LinkControl(RLC)isresponsibleforsegmentationandretransmissionhandling.TheRLCprovidesservicestothePDCPintheformofRLCchannels.ThereisoneRLCentityperRLCchannel(andhenceperradiobearer)configuredforadevice.ComparedtoLTE,theNRRLCdoesnotsupportin-sequencedeliveryofdatatohigherprotocollayers,achangemotivatedbythereduceddelaysasdiscussedbelow.

•Medium-AccessControl(MAC)handlesmultiplexingoflogicalchannels,hybrid-ARQretransmissions,andschedulingandscheduling-relatedfunctions.TheschedulingfunctionalityislocatedinthegNBforbothuplinkanddownlink.TheMACprovidesservicestotheRLCintheformoflogicalchannels.TheheaderstructureintheMAClayerhasbeenchangedinNRtoallowformoreefficientsupportoflow-latencyprocessingthaninLTE.

•PhysicalLayer(PHY)handlescoding/decoding,modulation/demodulation,multi-antennamapping,andothertypicalphysical-layerfunctions.ThephysicallayeroffersservicestotheMAClayerintheformoftransportchannels.

To summarize the flowofdownlinkdata throughall theprotocol layers, anexample illustrationwith threeIPpackets, twoononeradiobearerandoneonanother radio bearer, is given inFig. 6.8. In this example, there are two radiobearers and one RLC SDU is segmented and transmitted in two differenttransports.Thedataflowinthecaseofuplinktransmissionissimilar.

FIGURE6.8 Exampleofuser-planedataflow.

TheSDAPprotocolmapstheIPpacketstothedifferentradiobearers;inthisexampleIPpacketsnandn+1aremappedtoradiobearerxandIPpacketmismapped to radiobearery. Ingeneral, thedata entity from/to ahigherprotocollayer is known as a Service Data Unit (SDU) and the corresponding entityto/from a lower protocol layer entity is called a Protocol Data Unit (PDU).Hence, the output from the SDAP is an SDAP PDU, which equals an PDCPSDU.ThePDCPprotocolperforms(optional) IP-headercompression, followedby

ciphering,foreachradiobearer.APDCPheaderisadded,carryinginformationrequired for deciphering in the device aswell as a sequence number used forretransmission and in-sequence delivery, if configured. The output from thePDCPisforwardedtotheRLC.TheRLCprotocolperformssegmentationofthePDCPPDUsifnecessaryand

adds an RLC header containing a sequence number used for handingretransmissions.UnlikeLTE,theNRRLCisnotprovidingin-sequencedeliveryofdatatohigherlayers.Thereasonisadditionaldelayincurredbythereorderingmechanism, a delay thatmight be detrimental for services requiring very lowlatency. If needed, in-sequence delivery can be provided by the PDCP layerinstead.TheRLCPDUsareforwardedtotheMAClayer,whichmultiplexesanumber

ofRLCPDUsandattachesaMACheadertoformatransportblock.Notethatthe MAC headers are distributed across the MAC PDU, such that the MACheader related to a certainRLCPDU is located immediatelyprior to theRLCPDU.ThisisdifferentcomparedtoLTE,whichhasalltheheaderinformationatthe beginning of the MAC PDU and is motivated by efficient low-latencyprocessing.WiththestructureinNR,theMACPDUcanbeassembled“onthefly”asthereisnoneedtoassemblethefullMACPDUbeforetheheaderfields

can be computed. This reduces the processing time and hence the overalllatency.TheremainderofthischaptercontainsanoverviewoftheSDAP,RLC,MAC,

andphysicallayers.

6.4.1ServiceDataAdaptationProtocol(SDAP)The Service Data Adaptation Protocol (SDPA) is responsible for mappingbetween a quality-of-service flow from the 5G core network and a data radiobearer,aswellasmarkingthequality-of-serviceflowidentifier(QFI)inuplinkand downlink packets. The reason for the introduction of SDAP inNR is thenew quality-of-service handling compared to LTE when connected to the 5Gcore.InthiscasetheSDAPisresponsibleforthemappingbetweenQoSflowsand radio bearers as described in Section 6.2. If the gNB is connected to theEPC,asisthecasefornon-standalonemode,theSDAPisnotused.

6.4.2Packet-DataConvergenceProtocol(PDCP)The PDCP performs IP header compression to reduce the number of bits totransmitover the radio interface.Theheader-compressionmechanism isbasedon robust header compression (ROHC) framework [38], a set of standardizedheader-compression algorithms also used for several other mobile-communication technologies.PDCPisalsoresponsible forciphering toprotectagainst eavesdroppingand, for the controlplane, integrityprotection to ensurethatcontrolmessagesoriginatefromthecorrectsource.Atthereceiverside,thePDCPperformsthecorrespondingdecipheringanddecompressionoperations.The PDCP is also responsible for duplicate removal and (optional) in-

sequence delivery, functions useful, for example, in the case of intra-gNBhandover.Uponhandover,undelivereddownlinkdatapacketswillbeforwardedbythePDCPfromtheoldgNBtothenewgNB.ThePDCPentityinthedevicewillalsohandleretransmissionofalluplinkpacketsnotyetdeliveredtothegNBasthehybrid-ARQbuffersareflusheduponhandover.Inthiscase,somePDUsmaybereceivedinduplicate,bothover theconnectiontotheoldgNBandthenewgNB.ThePDCPwill in this case remove any duplicates.ThePDCP canalsobeconfiguredtoperformreorderingtoensurein-sequencedeliveryofSDUstohigher-layerprotocolsifdesirable.DuplicationinPDCPcanalsobeusedforadditionaldiversity.Packetscanbe

duplicatedandtransmittedonmultiplecells,increasingthelikelihoodofatleastonecopybeingcorrectlyreceived.Thiscanbeusefulforservicesrequiringveryhighreliability.Atthereceivingend,thePDCPduplicateremovalfunctionalityremovesanyduplicates.Inessence,thisresultsinselectiondiversity.DualconnectivityisanotherareawherePDCPplaysanimportantrole.Indual

connectivity,adeviceisconnectedtotwocells,oringeneral,twocellgroups,4theMasterCellGroup (MCG)andtheSecondaryCellGroup (SCG).The twocellgroupscanbehandledbydifferentgNBs.Aradiobeareristypicallyhandledbyone of the cell groups, but there is also the possibility for split bearers, inwhich case one radio bearer is handled by both cell groups. In this case thePDCP is inchargeofdistributing thedatabetween theMCGand theSCG,asillustratedinFig.6.9.

FIGURE6.9 Dualconnectivitywithsplitbearer.

The June 2018 version of release 15 supports dual connectivity in general,while theDecember2017version is limited todualconnectivitybetweenLTEandNR. This is of particular importance as it is the basis for non-standaloneoperation using option 3 as illustrated in Fig. 6.4. TheLTE-basedmaster cellhandles control-plane and (potentially) user-plane signaling, and theNR-basedsecondarycellhandlesuser-planeonly,inessenceboostingthedatarates.

6.4.3Radio-LinkControlTheRLCprotocolisresponsibleforsegmentationofRLCSDUsfromthePDCPinto suitably sized RLC PDUs. It also handles retransmission of erroneouslyreceivedPDUs,aswellasremovalofduplicatePDUs.Dependingonthetypeof

service, theRLCcanbeconfigured inoneof threemodes—transparentmode,unacknowledged mode, and acknowledged mode—to perform some or all ofthesefunctions.Transparentmodeis,asthenamesuggests,transparent,andnoheadersareadded.Unacknowledgedmodesupportssegmentationandduplicatedetection, while acknowledged mode in addition supports retransmission oferroneouspackets.Onemajordifferencecompared toLTE is that theRLCdoesnot ensure in-

sequence delivery of SDUs to upper layers. Removing in-sequence deliveryfromtheRLCreducestheoveralllatencyaslaterpacketsdonothavetowaitforretransmission of an earlier missing packet before being delivered to higherlayersbutcanbeforwarded immediately.Anotherdifference is theremovalofconcatenation from theRLCprotocol to allowRLCPDUs to be assembled inadvance,prior to receiving theuplinkschedulinggrant.Thisalsohelps reducetheoveralllatency,asdiscussedinChapter13.Segmentation, one of the main RLC functions, is illustrated in Fig. 6.10.

Included in the figure is also the correspondingLTE functionality,which alsosupportsconcatenation.Dependingon theschedulerdecision,acertainamountof data, that is, certain transport-block size, is selected.As part of the overalllow-latency design of NR, the scheduling decision in case of an uplinktransmission isknown to thedevice justbefore transmission, in theorderof afewOFDMsymbolsbefore.InthecaseofconcatenationinLTE,theRLCPDUcannotbeassembleduntiltheschedulingdecisionisknown,whichresultsinanadditionaldelayuntil theuplink transmissionandcannotmeet the low-latencyrequirementofNR.ByremovingtheconcatenationfromRLC,theRLCPDUscan be assembled in advance and upon receipt of the scheduling decision thedeviceonlyhastoforwardasuitablenumberofRLCPDUstotheMAClayer,thenumberdependingonthescheduledtransportblocksize.Tocompletelyfillup the transport block size, the last RLC PDUmay contain a segment of anSDU. The segmentation operation is simple. Upon receiving the schedulinggrant,thedeviceincludestheamountofdataneededtofillupthetransportblockandupdatestheheadertoindicateitisasegmentedSDU.

FIGURE6.10 RLCsegmentation.

TheRLC retransmissionmechanism is also responsible for providing error-free delivery of data to higher layers. To accomplish this, a retransmissionprotocoloperatesbetween theRLCentities in the receiver and transmitter.Bymonitoring the sequence numbers indicated in the headers of the incomingPDUs,thereceivingRLCcanidentifymissingPDUs(theRLCsequencenumberisindependentofthePDCPsequencenumber).StatusreportsarefedbacktothetransmittingRLCentity, requesting retransmissionofmissingPDUs.Basedonthe received status report, the RLC entity at the transmitter can take theappropriateactionandretransmitthemissingPDUsifneeded.Although the RLC is capable of handling transmission errors due to noise,

unpredictable channel variations, etc., error-free delivery is in most caseshandledby theMAC-basedhybrid-ARQprotocol.Theuseofa retransmissionmechanismintheRLCmaythereforeseemsuperfluousatfirst.However,aswillbe discussed inChapter 13, this is not the case and the use of bothRLC-andMAC-based retransmission mechanisms is in fact well motivated by thedifferencesinthefeedbacksignaling.ThedetailsofRLCarefurtherdescribedinSection13.2.

6.4.4Medium-AccessControlThe MAC layer handles logical-channel multiplexing, hybrid-ARQretransmissions, and scheduling and scheduling-related functions, includinghandling of different numerologies. It is also responsible formultiplexing/demultiplexing data across multiple component carriers whencarrieraggregationisused.

6.4.4.1LogicalChannelsandTransportChannelsThe MAC provides services to the RLC in the form of logical channels. Alogicalchannel isdefinedby the typeof information itcarriesand isgenerallyclassified as a control channel, used for transmission of control andconfigurationinformationnecessaryforoperatinganNRsystem,orasatrafficchannel,usedfortheuserdata.Thesetoflogical-channeltypesspecifiedforNRincludes:

•TheBroadcastControlChannel(BCCH),usedfortransmissionofsysteminformationfromthenetworktoalldevicesinacell.Priortoaccessingthesystem,adeviceneedstoacquirethesysteminformationtofindouthowthesystemisconfiguredand,ingeneral,howtobehaveproperlywithinacell.Notethat,inthecaseofnon-standaloneoperation,systeminformationisprovidedbytheLTEsystemandthereisnoBCCH.

•ThePagingControlChannel(PCCH),usedforpagingofdeviceswhoselocationonacelllevelisnotknowntothenetwork.Thepagingmessagethereforeneedstobetransmittedinmultiplecells.Notethat,inthecaseofnon-standaloneoperation,pagingisprovidedbytheLTEsystemandthereisnoPCCH.

•TheCommonControlChannel(CCCH),usedfortransmissionofcontrolinformationinconjunctionwithrandomaccess.

•TheDedicatedControlChannel(DCCH),usedfortransmissionofcontrolinformationto/fromadevice.Thischannelisusedforindividualconfigurationofdevicessuchassettingvariousparametersindevices.

•TheDedicatedTrafficChannel(DTCH),usedfortransmissionofuserdatato/fromadevice.Thisisthelogicalchanneltypeusedfortransmissionofallunicastuplinkanddownlinkuserdata.

TheabovelogicalchannelsareingeneralpresentalsoinanLTEsystemandused for similar functionality. However, LTE provides additional logicalchannels for features not yet supported byNR (but likely to be introduced inupcomingreleases).Fromthephysicallayer,theMAClayerusesservicesintheformoftransport

channels.A transportchannel isdefinedbyhow andwithwhatcharacteristicsthe information is transmitted over the radio interface. Data on a transportchannelareorganizedintotransportblocks.IneachTransmissionTimeInterval

(TTI),atmostonetransportblockofdynamicsizeistransmittedovertheradiointerfaceto/fromadevice(inthecaseofspatialmultiplexingofmorethanfourlayers,therearetwotransportblocksperTTI).AssociatedwitheachtransportblockisaTransportFormat (TF),specifying

how the transport block is to be transmitted over the radio interface. Thetransport format includes information about the transport-block size, themodulation-and-coding scheme, and the antenna mapping. By varying thetransport format, theMAClayercan thusrealizedifferentdatarates,aprocessknownastransport-formatselection.Thefollowingtransport-channeltypesaredefinedforNR:

•TheBroadcastChannel(BCH)hasafixedtransportformat,providedbythespecifications.ItisusedfortransmissionofpartsoftheBCCHsysteminformation,morespecificallytheso-calledMasterInformationBlock(MIB),asdescribedinChapter16.

•ThePagingChannel(PCH)isusedfortransmissionofpaginginformationfromthePCCHlogicalchannel.ThePCHsupportsdiscontinuousreception(DRX)toallowthedevicetosavebatterypowerbywakinguptoreceivethePCHonlyatpredefinedtimeinstants.

•TheDownlinkSharedChannel(DL-SCH)isthemaintransportchannelusedfortransmissionofdownlinkdatainNR.ItsupportskeyNRfeaturessuchasdynamicrateadaptationandchannel-dependentschedulinginthetimeandfrequencydomains,hybridARQwithsoftcombining,andspatialmultiplexing.ItalsosupportsDRXtoreducedevicepowerconsumptionwhilestillprovidinganalways-onexperience.TheDL-SCHisalsousedfortransmissionofthepartsoftheBCCHsysteminformationnotmappedtotheBCH.EachdevicehasaDL-SCHpercellitisconnectedto.InslotswheresysteminformationisreceivedthereisoneadditionalDL-SCHfromthedeviceperspective.

•TheUplinkSharedChannel(UL-SCH)istheuplinkcounterparttotheDL-SCH—thatis,theuplinktransportchannelusedfortransmissionofuplinkdata.

In addition, the Random-Access Channel (RACH) is also defined as atransportchannel,althoughitdoesnotcarrytransportblocks.Part of theMAC functionality is multiplexing of different logical channels

andmappingof the logicalchannels to theappropriate transportchannels.Themappingbetween logical-channel typesand transport-channel types isgiven inFig.6.11.ThisfigureclearlyindicateshowDL-SCHandUL-SCHarethemaindownlink and uplink transport channels, respectively. In the figures, thecorrespondingphysicalchannels,describedfurtherbelow,arealsoincludedandthemappingbetweentransportchannelsandphysicalchannelsisillustrated.

FIGURE6.11 Mappingbetweenlogical,transport,andphysicalchannels.

To support priority handling, multiple logical channels, where each logicalchannelhas itsownRLCentity,canbemultiplexed intoone transportchannelby theMAC layer.At the receiver, theMAC layer handles the correspondingdemultiplexingandforwardstheRLCPDUstotheirrespectiveRLCentity.Tosupportthedemultiplexingatthereceiver,aMACheaderisused.Theplacementof the MAC headers has been improved compared to LTE, again with low-latencyoperationinmind.InsteadoflocatingalltheMACheaderinformationatthebeginningofaMACPDU,which implies thatassemblyof theMACPDUcannot start until the scheduling decision is available, the subheadercorrespondingtoacertainMACSDUisplacedimmediatelybeforetheSDU,asshown in Fig. 6.12. This allows the PDUs to be preprocessed before havingreceivedtheschedulingdecision.Ifnecessary,paddingcanbeappendedtoalignthetransportblocksizewiththosesupportedinNR.

FIGURE6.12 MACSDUmultiplexingandheaderinsertion(uplinkcase).

Thesubheadercontainstheidentityofthelogicalchannel(LCID)fromwhichtheRLCPDUoriginatedandthelengthofthePDUinbytes.Thereisalsoaflagindicatingthesizeofthelengthindicator,aswellasareservedbitforfutureuse.Inaddition tomultiplexingofdifferent logicalchannels, theMAClayercan

alsoinsertMACcontrolelementsintothetransportblockstobetransmittedoverthe transport channels. A MAC control element is used for inband controlsignalingandidentifiedwithreservedvaluesintheLCIDfield,wheretheLCIDvalue indicates the typeof control information.Both fixed-andvariable-lengthMACcontrol elements are supported, dependingon their usage.For downlinktransmissions,MACcontrolelementsare locatedat thebeginningof theMACPDU,while for uplink transmissions theMACcontrol elements are located atthe end, immediately before the padding (if present). Again, the placement ischoseninordertofacilitatelow-latencyoperationinthedevice.MAC control elements are, as mentioned above, used for inband control

signaling.ItprovidesafasterwaytosendcontrolsignalingthanRLC,withouthaving to resort to the restrictions in terms of payload sizes and reliabilityofferedbyphysical-layerL1/L2control signaling (PDCCHorPUCCH).TherearemultipleMACcontrolelements,usedforvariouspurposes,forexample:

•Scheduling-relatedMACcontrolelements,suchasbufferstatusreportsandpowerheadroomreportsusedtoassistuplinkschedulingasdescribedinChapter14,andtheconfiguredgrantconfirmationMACcontrolelementusedwhenconfiguringsemipersistentscheduling;

•Random-access-relatedMACcontrolelementssuchastheC-RNTIandcontention-resolutionMACcontrolelements;

•Timing-advanceMACcontrolelementstohandletimingadvanceasdescribedinChapter15;

•Activationanddeactivationofpreviouslyconfiguredcomponents;•DRX-relatedMACcontrolelements;•Activation/deactivationofPDCPduplicationdetection;and•Activation/deactivationofCSIreportingandSRStransmission(seeChapter8).

TheMACentityisalsoresponsiblefordistributingdatafromeachflowacrossthedifferentcomponentcarriers,orcells,inthecaseofcarrieraggregation.The

basic principle for carrier aggregation is independent processing of thecomponentcarriersinthephysicallayer,includingcontrolsignaling,scheduling,andhybrid-ARQretransmissions,whilecarrieraggregationisinvisibleabovetheMAClayer.Carrieraggregation is thereforemainlyseen in theMAClayer,asillustrated in Fig. 6.13, where logical channels, including any MAC controlelements, aremultiplexed to form transport blocksper component carrierwitheachcomponentcarrierhavingitsownhybrid-ARQentity.

FIGURE6.13 Carrieraggregation.

Both carrier aggregation and dual connectivity result in the device beingconnected tomore thanonecell.Despite this similarity, thereare fundamentaldifferences, primarily related to how tightly the different cells are coordinatedandwhethertheyresideinthesameorindifferentgNBs.Carrier aggregation implies very tight coordination, with all the cells

belonging to the same gNB. Scheduling decisions are taken jointly for all thecellsthedeviceisconnectedtobyonejointscheduler.Dualconnectivity,on theotherhand,allows foramuch loosercoordination

between the cells.The cells canbelong todifferent gNBs, and theymay evenbelong to different radio-access technologies as is the case for NR-LTE dualconnectivityincaseofnon-standaloneoperation.Carrier aggregationanddual connectivity canalsobe combined.This is the

reasonforthetermsmastercellgroupandsecondarycellgroup.Withineachofthecellgroups,carrieraggregationcanbeused.

6.4.4.2Scheduling

OneofthebasicprinciplesofNRradioaccessisshared-channeltransmission—that is, time–frequency resources are dynamically shared between users. Thescheduler ispartof theMAClayer(althoughoftenbetterviewedasaseparateentity)andcontrolstheassignmentofuplinkanddownlinkresourcesintermsofso-calledresourceblocksinthefrequencydomainandOFDMsymbolsandslotsinthetimedomain.Thebasicoperationof the scheduler isdynamic scheduling,where thegNB

takes a scheduling decision, typically once per slot, and sends schedulinginformation to the selected set of devices. Although per-slot scheduling is acommoncase,neithertheschedulingdecisions,northeactualdatatransmissionisrestrictedtostartorendattheslotboundaries.Thisisusefultosupportlow-latencyoperationaswellasfutureextensionstounlicensedspectrumoperationasmentionedinChapter6.Uplink and downlink scheduling are separated in NR, and uplink and

downlinkschedulingdecisionscanbetakenindependentlyofeachother(withinthelimitssetbytheduplexschemeinthecaseofhalf-duplexoperation).The downlink scheduler is responsible for (dynamically) controlling which

device(s)totransmittoand,foreachofthesedevices,thesetofresourceblocksupon which the device’s DL-SCH should be transmitted. Transport-formatselection (selection of transport-block size, modulation scheme, and antennamapping) and logical-channel multiplexing for downlink transmissions arecontrolledbythegNB,asillustratedintheleftpartofFig.6.14.

FIGURE6.14 Transport-formatselectionin(a)downlinkand(b)uplink.

The uplink scheduler serves a similar purpose, namely to (dynamically)controlwhichdevicesaretotransmitontheirrespectiveUL-SCHandonwhichuplinktime–frequencyresources(includingcomponentcarrier).Despitethefactthat the gNB scheduler determines the transport format for the device, it isimportant to point out that the uplink scheduling decision does not explicitlyscheduleacertainlogicalchannelbutratherthedeviceassuch.Thus,althoughthe gNB scheduler controls the payload of a scheduled device, the device isresponsibleforselectingfromwhichradiobearer(s)thedataaretakenaccordingtoasetofrules,theparametersofwhichcanbeconfiguredbythegNB.Thisisillustrated in the right part ofFig. 6.14,where thegNBscheduler controls thetransportformatandthedevicecontrolsthelogical-channelmultiplexing.Althoughtheschedulingstrategyisimplementationspecificandnotspecified

by3GPP,theoverallgoalofmostschedulersistotakeadvantageofthechannelvariationsbetweendevicesandpreferablyscheduletransmissionstoadeviceonresourceswithadvantageouschannelconditionsinboththetimeandfrequencydomain,oftenreferredtoaschannel-dependentscheduling.Downlink channel-dependent scheduling is supported through channel-state

information (CSI), reported by the device to the gNB and reflecting theinstantaneousdownlink channelquality in the timeand frequencydomains, aswellasinformationnecessarytodeterminetheappropriateantennaprocessinginthe case of spatial multiplexing. In the uplink, the channel-state informationnecessaryforuplinkchannel-dependentschedulingcanbebasedonasoundingreference signal transmitted from each device for which the gNB wants toestimatetheuplinkchannelquality.Toaidtheuplinkschedulerinitsdecisions,the device can transmit buffer-status and power-headroom information to thegNBusingMACcontrolelements.This informationcanonlybe transmitted ifthedevicehasbeengivenavalid schedulinggrant.For situationswhen this isnot thecase,an indicator that thedeviceneedsuplinkresources isprovidedaspartoftheuplinkL1/L2control-signalingstructure(seeChapter10).Althoughdynamicschedulingisthebaselinemode-of-operation,thereisalso

a possibility for transmission/receptionwithout a dynamic grant to reduce thecontrol-signalingoverhead.Thedetailsdifferbetweendownlinkanduplink.Inthedownlink,aschemesimilartosemipersistentschedulinginLTEisused.

A semistatic scheduling pattern is signaled in advance to the device. Uponactivation byL1/L2 control signaling,which also includes parameters such asthe time–frequency resources and coding-and-modulation scheme to use, thedevice receives downlink data transmissions according to the preconfigured

pattern.In the uplink, there are two slightly different schemes, type 1 and type 2,

differing on how to activate the scheme. In type 1, RRC configures allparameters, including the time–frequency resources and the modulation-and-codingschemetouse,andalsoactivatestheuplinktransmissionaccordingtotheparameters. Type 2, on the other hand, is similar to semipersistent schedulingwhereRRCconfigurestheschedulingpatternintime.ActivationisdoneusingL1/L2signaling,which includes thenecessary transmissionparameters (exceptthe periodicitywhich is provides throughRRC signaling). In both type 1 andtype2,thedevicedoesnottransmitintheuplinkunlesstherearedatatoconvey.

6.4.4.3HybridARQWithSoftCombiningHybrid ARQ with soft combining provides robustness against transmissionerrors.Ashybrid-ARQretransmissionsarefast,manyservicesallowforoneormultiple retransmissions, and the hybrid-ARQ mechanism therefore forms animplicit(closedloop)rate-controlmechanism.Thehybrid-ARQprotocolispartoftheMAClayer,whilethephysicallayerhandlestheactualsoftcombining.5HybridARQisnotapplicableforalltypesoftraffic.Forexample,broadcast

transmissions, where the same information is intended for multiple devices,typicallydonotrelyonhybridARQ.Hence,hybridARQisonlysupportedforthe DL-SCH and the UL-SCH, although its usage is up to the gNBimplementation.Thehybrid-ARQprotocolusesmultipleparallelstop-and-waitprocessesina

similar way to LTE. Upon receipt of a transport block, the receiver tries todecodethetransportblockandinformsthetransmitterabouttheoutcomeofthedecodingoperationthroughasingleacknowledgmentbitindicatingwhetherthedecodingwassuccessfulorifaretransmissionofthetransportblockisrequired.Clearly, the receiver must know to which hybrid-ARQ process a receivedacknowledgment is associated. This is solved by using the timing of theacknowledgmentforassociationwithacertainhybrid-ARQprocessorbyusingthe position of the acknowledgment in the hybrid-ARQ codebook in case ofmultiple acknowledgments transmitted at the same time (see Section 13.1 forfurtherdetails).Anasynchronoushybrid-ARQprotocolisusedforbothdownlinkanduplink

—that is, an explicit hybrid-ARQ process number is used to indicate whichprocess is being addressed. In an asynchronous hybrid-ARQ protocol, theretransmissionsare inprinciplescheduledsimilarly to the initial transmissions.

Theuseofanasynchronousuplinkprotocol,insteadofasynchronousoneasinLTE, is necessary to support dynamic TDD where there is no fixeduplink/downlink allocation. It also offers better flexibility in terms ofprioritization between data flows and devices and is beneficial for futureextensiontounlicensedspectrumoperation.6Up to 16 hybrid-ARQ processes are supported. Having a larger maximum

numberofhybrid-ARQprocesses than inLTE7 ismotivatedby thepossibilityforremoteradioheads,whichincursacertainfront-hauldelay,togetherwiththeshorterslotdurationsathighfrequencies.Itisimportantthough,thatthelargernumberofmaximumhybrid-ARQprocessesdoesnot implya longer roundtriptimeasnotallprocessesneedtobeused,itisonlyanupperlimitofthenumberofprocessespossible.Theuseofmultipleparallelhybrid-ARQprocesses,illustratedinFig.6.15,for

adevicecanresultindatabeingdeliveredfromthehybrid-ARQmechanismoutof sequence. For example, transport block 3 in the figure was successfullydecoded before transport block 2, which required retransmissions. For manyapplicationsthisisacceptableand,ifnot,in-sequencedeliverycanbeprovidedthroughthePDCPprotocol.Thereasonfornotprovidingin-sequencedeliveryintheRLCprotocolistoreducelatency.Ifin-sequencedeliverywouldbeenforcedinFig.6.15,packetnumbers3,4,and5wouldhavetobedelayeduntilpacketnumber 2 is correctly received before delivering them to higher layers, whilewithout in-sequence delivery each packet can be forwarded as soon as it iscorrectlyreceived.

FIGURE6.15 Multipleparallelhybrid-ARQprocesses.

One additional feature of the hybrid-ARQ mechanism in NR compared toLTEisthepossibilityforretransmissionofcodeblockgroups,afeaturethatcanbebeneficialforverylargetransportblocksorwhenatransportblockispartiallyinterfered by another preempting transmission. As part of the channel-codingoperation in the physical layer, a transport block is split into one or morecodeblockswitherror-correctingcodingappliedtoeachofthecodeblocksofatmost 8448 bits8 in order to keep the channel-coding complexity reasonable.Thus,evenformodestdataratestherecanbemultiplecodeblockspertransportblockandatGbpsdataratestherecanbehundredsofcodeblockspertransportblock. Inmany cases, especially if the interference is bursty and hits a smallnumberofOFDMsymbols in the slot, only a fewof these codeblocks in thetransportblockmaybecorrupted,whilethemajorityofcodeblocksarecorrectlyreceived.Tocorrectlyreceivethetransportblock,itissufficienttoretransmittheerroneouscodeblocks.Atthesametime,thecontrolsignalingoverheadwouldbe too large if individual code blocks can be addressed by the hybrid-ARQmechanism. Therefore, codeblock groups (CBGs) are defined. If per-CBGretransmission is configured, feedback is provided per CBG and only theerroneously receivedcodeblockgroupsare retransmitted (Fig.6.16).This canconsume less resource than retransmitting the whole transport block. CBGretransmissionsare invisible to theMAClayerandarehandled in thephysicallayer,despitebeingpartof thehybrid-ARQmechanism.The reason for this isnot technical but purely related to the specification structure. From a MACperspective, the transportblockisnotcorrectlyreceiveduntilall theCBGsarecorrectly received. It is not possible, in the samehybrid-ARQprocess, tomixtransmission of new CBGs belonging to another transport block withretransmissionsofCBGsbelongingtotheincorrectlyreceivedtransportblock.

FIGURE6.16 Codeblockgroupretransmission.

The hybrid-ARQmechanismwill rapidly correct transmission errors due to

noiseorunpredictablechannelvariations.Asdiscussedabove, theRLCisalsocapable of requesting retransmissions, which at first sight may seemunnecessary.However,thereasonforhavingtworetransmissionmechanismsontopofeachothercanbeseeninthefeedbacksignaling—hybridARQprovidesfast retransmissions but due to errors in the feedback the residual error rate istypicallytoohighfor,forexample,goodTCPperformance,whileRLCensures(almost) error-free data delivery but slower retransmissions than the hybrid-ARQ protocol. Hence, the combination of hybridARQ and RLC provides anattractivecombinationofsmallroundtriptimeandreliabledatadelivery.

6.4.5PhysicalLayerThe physical layer is responsible for coding, physical-layer hybrid-ARQprocessing,modulation,multi-antennaprocessing,andmappingofthesignaltothe appropriate physical time–frequency resources. It also handlesmapping oftransportchannelstophysicalchannels,asshowninFig.6.11.Asmentioned in the introduction, thephysical layerprovidesservices to the

MAC layer in the form of transport channels.Data transmissions in downlinkanduplinkusetheDL-SCHandUL-SCHtransport-channeltypes,respectively.Thereisatmostonetransportblock(twotransportblocksinthecaseofspatialmultiplexingofmorethanfourlayersinthedownlink)toasingledeviceperTTIonaDL-SCHorUL-SCH.In thecaseofcarrieraggregation, there isoneDL-SCH(orUL-SCH)percomponentcarrierseenbythedevice.Aphysical channel corresponds to the setof time–frequency resourcesused

for transmissionofaparticular transportchannelandeach transportchannel ismappedtoacorrespondingphysicalchannel,asshowninFig.6.11.Inadditionto thephysical channelswithacorresponding transport channel, therearealsophysical channels without a corresponding transport channel. These channels,known as L1/L2 control channels, are used for downlink control information(DCI),providingthedevicewiththenecessaryinformationforproperreceptionanddecodingofthedownlinkdatatransmission,anduplinkcontrolinformation(UCI) used for providing the scheduler and the hybrid-ARQ protocol withinformationaboutthesituationatthedevice.Thefollowingphysical-channeltypesaredefinedforNR:

•ThePhysicalDownlinkSharedChannel(PDSCH)isthemainphysicalchannelusedforunicastdatatransmission,butalsofortransmissionof,

forexample,paginginformation,random-accessresponsemessages,anddeliveryofpartsofthesysteminformation.

•ThePhysicalBroadcastChannel(PBCH)carriespartofthesysteminformation,requiredbythedevicetoaccessthenetwork.

•ThePhysicalDownlinkControlChannel(PDCCH)isusedfordownlinkcontrolinformation,mainlyschedulingdecisions,requiredforreceptionofPDSCH,andforschedulinggrantsenablingtransmissiononthePUSCH.

•ThePhysicalUplinkSharedChannel(PUSCH)istheuplinkcounterparttothePDSCH.ThereisatmostonePUSCHperuplinkcomponentcarrierperdevice.

•ThePhysicalUplinkControlChannel(PUCCH)isusedbythedevicetosendhybrid-ARQacknowledgments,indicatingtothegNBwhetherthedownlinktransportblock(s)wassuccessfullyreceivedornot,tosendchannel-statereportsaidingdownlinkchannel-dependentscheduling,andforrequestingresourcestotransmituplinkdataupon.

•ThePhysicalRandom-AccessChannel(PRACH)isusedforrandomaccess.

Notethatsomeofthephysicalchannels,morespecificallythechannelsusedfordownlinkanduplinkcontrolinformation(PDCCHandPUCCH)donothaveacorrespondingtransportchannelmappedtothem.

6.5Control-PlaneProtocolsThecontrol-planeprotocolsare,amongotherthings,responsibleforconnectionsetup,mobility,andsecurity.TheNAScontrol-plane functionality operates between theAMF in the core

network and thedevice. It includes authentication, security, anddifferent idle-mode procedures such as paging (described below). It is also responsible forassigninganIPaddresstoadevice.The Radio Resource Control (RRC) control-plane functionality operates

betweentheRRClocatedinthegNB.RRCisresponsibleforhandlingtheRAN-relatedcontrol-planeprocedures,including:

•Broadcastofsysteminformationnecessaryforthedevicetobeabletocommunicatewithacell.Acquisitionofsysteminformationis

describedinChapter16.•TransmissionofpagingmessagesoriginatingfromtheMMEtonotifythedeviceaboutincomingconnectionrequests.PagingisusedintheRRC_IDLEstate(describedfurtherbelow)whenthedeviceisnotconnectedtoacell.Indicationofsystem-informationupdatesisanotheruseofthepagingmechanism,asispublicwarningsystems.

•Connectionmanagement,includingsettingupbearersandmobility.ThisincludesestablishinganRRCcontext—thatis,configuringtheparametersnecessaryforcommunicationbetweenthedeviceandtheradio-accessnetwork.

•Mobilityfunctionssuchascell(re)selection.•Measurementconfigurationandreporting.•Handlingofdevicecapabilities;whenconnectionisestablishedthedevicewillannounceitscapabilitiesasnotalldevicesarecapableofsupportingallthefunctionalitydescribedinthespecifications.

RRCmessages are transmitted to the device using signaling radio bearers(SRBs),usingthesamesetofprotocollayers(PDCP,RLC,MAC,andPHY)asdescribed in Section 6.4. The SRB ismapped to the common control channel(CCCH) during establishment of connection and, once a connection isestablished, to thededicated control channel (DCCH).Control-plane anduser-planedatacanbemultiplexedintheMAClayerandtransmittedtothedeviceinthesameTTI.TheaforementionedMACcontrolelementscanalsobeusedforcontrol of radio resources in some specific cases where low latency is moreimportantthanciphering,integrityprotection,andreliabletransfer.

6.5.1RRCStateMachineInmostwireless communication systems, the device can be in different statesdependingonthetrafficactivity.ThisistruealsoforNRandanNRdevicecanbe in one of three RRC states, RRC_IDLE, RRC_ACTIVE, andRRC_INACTIVE (see Fig. 6.17). The first two RRC states, RRC_IDLE andRRC_CONNECTED, are similar to the counterparts in LTE, whileRRC_INACTIVEisanewstateintroducedinNRandnotpresentintheoriginalLTE design. There are also core network states not discussed further herein,CN_IDLE and CN_CONNECTED, depending on whether the device hasestablishedaconnectionwiththecorenetworkornot.

FIGURE6.17 RRCstates.

InRRC_IDLE,thereisnoRRCcontext—thatis,theparametersnecessaryforcommunication between the device and the network—in the radio-accessnetworkandthedevicedoesnotbelongtoaspecificcell.Fromacorenetworkperspective,thedeviceisintheCN_IDLEstate.Nodatatransfermaytakeplaceas the device sleeps most of the time to reduce battery consumption. In thedownlink,devicesinidlestateperiodicallywakeuptoreceivepagingmessages,if any, from the network. Mobility is handled by the device through cellreselection (see Section 6.5.2). Uplink synchronization is not maintained andhencetheonlyuplinktransmissionactivitythatmaytakeplaceisrandomaccess,discussed inChapter16, tomove toaconnectedstate.Aspartofmoving toaconnected state, the RRC context is established in both the device and thenetwork.In RRC_CONNECTED, the RRC context is established and all parameters

necessary forcommunicationbetween thedeviceand the radio-accessnetworkareknowntobothentities.Fromacorenetworkperspective,thedeviceisintheCN_CONNECTEDstate.Thecelltowhichthedevicebelongsisknownandanidentityof thedevice, theCellRadio-NetworkTemporaryIdentifier (C-RNTI),used for signaling purposes between the device and the network, has beenconfigured.Theconnectedstateisintendedfordatatransferto/fromthedevice,butdiscontinuous reception (DRX) can be configured to reduce device powerconsumption(DRXisdescribedinfurtherdetailinSection14.5).SincethereisanRRCcontextestablishedinthegNBintheconnectedstate,leavingDRXandstartingtoreceive/transmitdataisrelativelyfastasnoconnectionsetupwithitsassociatedsignalingisneeded.Mobilityismanagedbytheradio-accessnetwork,thatis,thedeviceprovidesneighboring-cellmeasurementstothenetworkwhichcommands the device to perform a handover when relevant. Uplink timealignmentmayormaynotexistbutneedtobeestablishedusingrandomaccessandmaintainedasdescribedinSection16.2fordatatransmissiontotakeplace.In LTE, only idle and connected states are supported. A common case in

practice is to use the idle state as the primary sleep state to reduce the devicepower consumption. However, as frequent transmission of small packets is

commonformanysmartphoneapplications,theresultisasignificantamountofidle-to-activetransitionsinthecorenetwork.Thesetransitionscomeatacostintermsofsignalingloadandassociateddelays.Therefore,toreducethesignalingload and in general reduce the latency, a third state is defined in NR, theRRC_INACTIVEstate.InRRC_INACTIVE,theRRCcontextiskeptinboththedeviceandthegNB.

The core network connection is also kept, that is, the device is inCN_CONNECTED from a core network perspective. Hence, transition toconnected state for data transfer is fast.No core network signaling is needed.TheRRCcontextisalreadyinplaceinthenetworkandidle-to-activetransitionscan be handled in the radio-access network. At the same time, the device isallowed to sleep in a similarway as in the idle state andmobility is handledthrough cell reselection, that is, without involvement of the network. Thus,RRC_INACTIVEcanbeseenasamixoftheidleandconnectedstates.9As seen from the discussion above, one important difference between the

differentstatesisthemobilitymechanismsinvolved.Efficientmobilityhandlingis a key part of anymobile communication system. For the idle and inactivestates,mobility ishandledby thedevice throughcell reselection,while for theconnected mode, mobility is handled by the radio-access network based onmeasurements.Thedifferentmobilitymechanismsaredescribedbelow,startingwithidle-andinactive-modemobility.

6.5.2Idle-StateandInactive-StateMobilityThepurposeof themobilitymechanism in idleand inactivestates is toensurethatadeviceisreachablebythenetwork.Thenetworkdoesthisbynotifyingthedevice by means of a paging message. The area over which such a pagingmessageistransmittedisakeyaspectofthepagingmechanismandinidleandinactivemodes,thedeviceisincontrolonwhentoupdatethisinformation.Thisissometimes referred toascell reselection. Inessence, thedevicesearches forandmeasuresoncandidatecellssimilartotheinitialcellsearchasdescribedinChapter16.Oncethedevicediscoversacellwithareceivedpowersufficientlyhigher than itscurrentone, itconsidered thisas thebestcelland, ifnecessary,contactsthenetworkthroughrandomaccess.

6.5.2.1TrackingtheDeviceInprinciple, the network could transmit thepage to the device over the entire

coverageof thenetwork,bybroadcasting thepagingmessage fromeverycell.However,thatwouldobviouslyimplyaveryhighoverheadintermsofpaging-message transmissions as the vastmajority of the paging transmissionswouldtakeplace incellswhere the targetdevice isnot located.Ontheotherhand, ifthepagingmessage isonly tobe transmitted in thecell inwhich thedevice islocated,thereisaneedtotrackthedeviceonacelllevel.Thiswouldimplythatthe device would have to inform the network every time it moves out of thecoverageofonecellandintothecoverageofanothercell.Thiswouldalsoleadtoveryhighoverhead,inthiscaseintermsofthesignalingneededtoinformthenetwork about the updated device location. For this reason, a compromisebetweenthesetwoextremesistypicallyused,wheredevicesareonlytrackedonacell-grouplevel:

•Thenetworkonlyreceivesnewinformationaboutthedevicelocationifthedevicemovesintoacelloutsideofthecurrentcellgroup;

•Whenpagingthedevice,thepagingmessageisbroadcastoverallcellswithinthecellgroup.

ForNR, the basic principle for such tracking is the same for idle state andinactivestate,althoughthegroupingissomewhatdifferentinthetwocases.AsillustratedinFig.6.18,NRcellsaregroupedintoRANAreas,whereeach

RANArea is identifiedbyanRANArea Identifier (RAI).TheRANAreas, inturn, are grouped into even larger Tracking Areas, with each Tracking AreabeingidentifiedbyaTrackingAreaIdentifier(TAI).Thus,eachcellbelongstooneRANAreaandoneTrackingArea, the identitiesofwhichareprovidedaspartofthecellsysteminformation.

FIGURE6.18 RANAreasandTrackingAreas.

TheTrackingAreasare thebasis fordevice trackingoncore-network level.EachdeviceisassignedaUERegistrationAreabythecorenetwork,consistingofalistoftrackingareaidentifiers.WhenadeviceentersacellthatbelongstoaTrackingAreanotincludedintheassignedUERegistrationAreaitaccessesthenetwork,includingthecorenetwork,andperformsaNASRegistrationUpdate.The core network registers the device location and updates the device UERegistration Area, in practice providing the device with a new TAI list thatincludesthenewTAI.The reason the device is assigned a set of TAIs, that is, a set of Tracking

Areas, is to avoid repeatedNASRegistrationUpdates if a devicemoves backandforthover theborderof twoneighborTrackingAreas.Bykeeping theoldTAIwithin the updatedUERegistrationArea no new update is needed if thedevicemovesbackintotheoldTAI.TheRANAreaisthebasisfordevicetrackingonradio-access-networklevel.

UEs in inactive statecanbeassignedaRANNotificationArea thatconsistsofeitherofthefollowing:

•Alistofcellidentities;•AlistofRAIs,inpracticealistofRANAreas;or•AlistofTAIs,inpracticealistofTrackingAreas.

NotethefirstcaseisessentiallythesameashavingeachRANAreaconsistofasinglecell,whilethelastcaseisessentiallythesameashavingtheRANAreascoincidewiththeTrackingAreas.

TheprocedureforRANNotificationAreaupdatesissimilartoupdatesoftheUE Registration Area. When a device enters a cell that is not directly orindirectly (via aRAN/TrackingArea) included in theRANNotificationArea,the device accesses the network and makes an RRC RAN Notification AreaUpdate.TheradionetworkregistersthedevicelocationandupdatesthedeviceRANNotificationArea.AsachangeofTrackingAreaalwaysimpliesachangealsoof thedeviceRANArea, anRRCRANNotificationAreaupdate is doneimplicitlyeverytimeadevicemakesaUERegistrationupdate.Inordertotrackitsmovementwithinthenetwork,thedevicesearchesforand

measuresonSSblockssimilartotheinitialcellsearchasdescribedinChapter16.Once thedevicediscoversanSSblockwithareceivedpower thatexceedsthe received power of its current SS block by a certain threshold it reads thesysteminformation(SIB1)ofthenewcellinordertoacquireinformationabouttheTrackingandRANAreas.

6.5.2.2PagingMessageTransmissionSimilartothedeliveryofsysteminformation,pagingmessagesareprovidedbymeans of ordinary scheduledPDSCH transmissions. In order to allow for lowdevice energy consumption, a device is only supposed towake up at specifictimeinstances,forexample,onceevery100msorevenlessoften,tomonitorforpagingmessages.Pagingmessagesare indicatedbyaspecificPI-RNTIcarriedwithintheDCI.OncedetectingsuchaDCI,thedevicedemodulatesanddecodesthecorrespondingPDSCHtoextractthepagingmessage(s).Notethattherecanbe multiple paging messages, corresponding to different devices, within thesamepagingtransmission.ThePI-RNTIisthusasharedidentity.

6.5.3Connected-StateMobilityInaconnectedstatethedevicehasaconnectionestablishedtothenetwork.Theaim of connected-state mobility is to ensure that this connectivity is retainedwithout any interruption or noticeable degradation as the devicemoveswithinthenetwork.To ensure this, the device continuously searches for new cells both on the

currentcarrierfrequency(intra-frequencymeasurements)andondifferentcarrierfrequencies (inter-frequencymeasurements) that the device has been informedabout.SuchmeasurementscanbedoneonanSSblockinessentially thesamewayasfor initialaccessandcellsearchin idleandinactivemode(seeabove).

However,measurementscanalsobedoneonconfiguredCSI-RS.Inaconnectedstate,thedevicedoesnotmakeanydecisionsofitsownwhen

it comes to handover to a different cell. Rather, based on different triggeringconditions,forexample,therelativepowerofameasuredSSblockcomparedtothecurrentcell,thedevicereportstheresultofthemeasurementstothenetwork.Basedon this reporting thenetworkmakesadecisionas towhetherornot thedeviceistohandovertoanewcell.ItshouldbepointedoutthatthisreportingisdoneusingRRCsignaling,thatis,itisnotcoveredbytheLayer-1measurementandreportingframework(Chapter8)used,forexample,forbeammanagement.Except for very small cells that are tightly synchronized to each other, the

currentuplinktransmissiontimingofadevicewilltypicallynotmatchthenewcelltowhichadeviceisassumedtohandover.Toestablishsynchronizationtoanewcelladevicethushastocarryoutaproceduresimilartotherandom-accessprocedureofChapter16.However, thismay thenbeacontention-free randomaccess using resources specifically assigned to the device with no risk forcollisionbutonlyaimingatestablishingsynchronizationtothenewcell.Thus,onlythe twofirststepsof therandom-accessprocedureareneeded, that is, thepreambletransmissionandcorrespondingrandom-accessresponseprovidingthedevicewithupdatedtransmissiontiming.

1Fig.6.2issimplifiedasitdoesnotmakeadistinctionbetweeneNBconnectedtotheEPCandng-eNBconnectedtothe5GCN.2Actually,twocellgroups,themastercellgroup(MCG)andthesecondarycellgroup(SCG)inthecaseofcarrieraggregationascarrieraggregationimpliesmultiplecellsineachofthetwocellgroups.3DuplicatedetectionispartoftheJune2018releaseandnotpresentintheDecember2017releaseofNR.4Thereasonforthetermcellgroupistocoveralsothecaseofcarrieraggregationwheretherearemultiplecells,oneperaggregatedcarriers,ineachcellgroup.5Thesoftcombiningisdonebeforeoraspartofthechanneldecoding,whichclearlyisaphysical-layerfunctionality.Also,theper-CBGretransmissionhandlingisformallypartofthephysicallayer.6LTEchangedtoanasynchronousuplinkhybrid-ARQprotocolforLAA.

7InLTE,eightprocessesareusedforFDDandupto15processesforTDD,dependingontheuplink–downlinkconfiguration.8Forcoderatesbelow¼,thecodeblocksizeis3840.9InLTErelease13,theRRCsuspend/resumemechanismwasintroducedtoprovidesimilarfunctionalityasRRC_INACTIVEinNR.However,theconnectiontothecorenetworkisnotmaintainedinRRCsuspend/resume.

CHAPTER7

OverallTransmissionStructure

Abstract

The overall transmission structure in the time domain (frame, subframe,slots, OFDM symbols) and in the frequency domain (subcarrier, DChandling, bandwidth parts) is described in this chapter. Antenna ports,quasi-colocation,andduplexschemesarealsodiscussed.

KeywordsSlot;subframe;frame;resourceblock;bandwidthpart(BWP);quasi-colocation(QCL);antennaports;frequencyraster;carrieraggregation;supplementaryuplink;FDD;TDD;slotformatindication(SFI)

PriortodiscussingthedetailedNRdownlinkanduplinktransmissionschemes,adescription of the basic time–frequency transmission resource of NR will beprovided in this chapter, including bandwidth parts, supplementary uplink,carrieraggregation,duplexschemes,antennaports,andquasi-colocation.

7.1TransmissionSchemeOFDMwasfoundtobeasuitablewaveformforNRduetoitsrobustnesstotimedispersion and ease of exploiting both the time and frequency domains whendefiningthestructurefordifferentchannelsandsignals.ItisthereforethebasictransmissionschemeforboththedownlinkanduplinktransmissiondirectionsinNR.However,unlikeLTEwhereDFT-precodedOFDMisthesoletransmissionscheme in the uplink, NR uses OFDM as the baseline uplink transmissionscheme with the possibility for complementary DFT-precoded OFDM. ThereasonsforDFT-precodedOFDMintheuplinkarethesameasinLTE,namelyto reduce the cubicmetric andobtain a higher power-amplifier efficiency, buttheuseofDFT-precodingalsohasseveraldrawbacksincluding:

•Spatialmultiplexing(“MIMO”)receiversbecomemorecomplex.ThiswasnotanissuewhenDFT-precodingwasagreedinthefirstLTEreleaseasitdidnotsupportuplinkspatialmultiplexingbutbecomesimportantwhensupportinguplinkspatialmultiplexing.

•Maintainingsymmetrybetweenuplinkanddownlinktransmissionschemesisinmanycasesbeneficial,somethingwhichislostwithanDFT-precodeduplink.Oneexampleofthebenefitswithsymmetricschemesissidelinktransmission,thatis,directtransmissionsbetweendevices.WhensidelinkswereintroducedinLTE,itwasagreedtokeeptheuplinktransmissionschemewhichrequiresthedevicestoimplementareceiverforDFT-precodedOFDMinadditiontotheOFDMreceiverbeingalreadyequippedfordownlinktransmissions.IntroducingsidelinksupportinNRinthefutureisthussimplerasthedevicealreadyhassupportforOFDMtransmissionandreception.

•DFT-precodedOFDMimpliesschedulingrestrictionsasonlycontiguousallocationsinthefrequencydomainarepossible.Inmanycasesthisisanacceptablerestriction,buttherearealsosituationswhenitisdesirabletouseanoncontiguousallocation,forexample,toobtainfrequencydiversity.

Hence,NRhasadoptedOFDMintheuplinkwithcomplementarysupportforDFT-precoding for data transmission. When DFT-precoding is used, uplinktransmissionsarerestrictedtoasinglelayeronly,whileuplinktransmissionsofup to four layers are possible with OFDM. Support for DFT-precoding ismandatoryinthedeviceandthenetworkcanthereforeconfigureDFT-precodingif/whenneeded.Thewaveformtousefortheuplinkrandom-accessmessagesisconfiguredaspartofthesysteminformation.One important aspect of OFDM is the selection of the numerology, in

particularthesubcarrierspacingandthecyclicprefixlength.Alargesubcarrierspacingisbeneficialfromafrequency-errorperspectiveasitreducestheimpactfrom frequency errors and phase noise. However, for a certain cyclic prefixlengthinmicroseconds,therelativeoverheadincreasesthelargerthesubcarrierspacing and from this perspective a smaller cyclic prefixwould be preferable.The selection of the subcarrier spacing therefore needs to carefully balanceoverhead from the cyclic prefix against sensitivity toDoppler spread/shift andphasenoise.For LTE, a choice of 15 kHz subcarrier spacing and a cyclic prefix of

approximately4.7µswasfoundtoofferagoodbalancebetweenthesedifferentconstraints for scenarios for which LTE was originally designed—outdoorcellulardeploymentsuptoapproximately3GHzcarrierfrequency.NR, on the other hand, is designed to support awide range of deployment

scenarios, from large cells with sub-1GHz carrier frequency up tomm-wavedeploymentswithverywidespectrumallocations.Havingasinglenumerologyforall these scenarios isnot efficientor evenpossible.For the lower rangeofcarrier frequencies, frombelow1GHzup toa fewGHz, thecell sizescanberelativelylargeandacyclicprefixcapableofhandlingthedelayspreadexpectedin these type of deployments, a couple of microseconds, is necessary.Consequently,asubcarrierspacingintheLTErangeorsomewhathigher,intherangeof15–30kHz, isneeded.Forhighercarrier frequenciesapproaching themm-wave range, implementation limitations suchasphasenoisebecomemorecritical, calling for higher subcarrier spacings.At the same time, the expectedcell sizes are smaller at higher frequencies as a consequence of the morechallengingpropagationconditions.Theextensiveuseofbeamformingathighfrequenciesalsohelpsreducetheexpecteddelayspread.Hence,forthesetypesof deployments a higher subcarrier spacing and a shorter cyclic prefix aresuitable.Fromthediscussionaboveitisseenthatascalablenumerologyisneeded.NR

therefore supports a flexible numerology with a range of subcarrier spacings,based on scaling a baseline subcarrier spacing of 15 kHz. The reason for thechoice of 15 kHz is coexistencewithLTE and theLTE-basedNB-IoTon thesame carrier. This is an important requirement, for example, for an operatorwhichhasdeployedNB-IoToreMTCtosupportmachine-typecommunication.Unlikesmartphones,suchMTCdevicescanhavearelativelylongreplacementcycle, 10 years or longer. Without provisioning for coexistence, the operatorwouldnotbeable tomigrate thecarrier toNRuntilall theMTCdeviceshavebeenreplaced.AnotherexampleisgradualmigrationwherethelimitedspectrumavailabilitymayforceanoperatortoshareasinglecarrierbetweenLTEandNRinthetimedomain.LTEcoexistenceisfurtherdiscussedinChapter17.Consequently,15kHzsubcarrierspacingwasselectedasthebaselineforNR.

Fromthebaselinesubcarrierspacing,subcarrierspacingsrangingfrom15kHzupto240kHzwithaproportionalchangeincyclicprefixdurationasshowninTable7.1arederived.Notethat240kHzissupportedfortheSSblockonly(seeSection16.1),andnotforregulardatatransmission.AlthoughtheNRphysical-layerspecificationisband-agnostic,notallsupportednumerologiesarerelevant

for all frequency bands. For each frequency band, radio requirements aretherefore defined for a subset of the supported numerologies as discussed inChapter18,RFCharacteristicsandRequirements.

Table7.1

SubcarrierSpacingsSupportedbyNR

SubcarrierSpacing(kHz) UsefulSymbolTime,Tu(µs) CyclicPrefix,TCP(µs)

15 66.7 4.7

30 33.3 2.3

60 16.7 1.2

120 8.33 0.59

240 4.17 0.29

To provide consistent and exact timing definitions, different time intervalswithin the NR specifications are defined as multiples of a basic time unitTc=1/(480000·4096). The basic time unitTc can thus be seen as the samplingtime of an FFT-based transmitter/receiver implementation for a subcarrierspacing of 480 kHz with an FFT size equal to 4096. This is similar to theapproachtakeninLTE,whichusesabasictimeunitTs=64Tc.Asdiscussedabove,thechoiceof15kHzsubcarrierspacingismotivatedby

coexistencewithLTE.Efficientcoexistencealsorequiresalignmentinthetimedomainand for this reason theNRslot structure for15kHz is identical to theLTEsubframestructure.Thismeansthatthecyclicprefixforthefirstandeighthsymbolsaresomewhat larger thanfor theothersymbols.Theslotstructureforhigher subcarrier spacings in NR is then derived by scaling this baselinestructure by powers of two. In essence, an OFDM symbol is split into twoOFDMsymbolsofthenexthighernumerology(seeFig.7.1).Scalingbypowersoftwoisbeneficialasitmaintainsthesymbolboundariesacrossnumerologies,which simplifies mixing different numerologies on the same carrier. For theOFDM symbolswith a somewhat larger cyclic prefix, the excess samples areallocatedtothefirstofthetwosymbolsobtainedwhensplittingonesymbol.

FIGURE7.1 Symbolalignment.

The useful symbol time Tu depends on the subcarrier spacing as shown inTable 7.1, with the overall OFDM symbol time being the sum of the usefulsymbol time and the cyclic-prefix length TCP. In LTE, two different cyclicprefixes are defined, normal cyclic prefix and extended cyclic prefix. Theextended cyclic prefix, although less efficient from a cyclic-prefix-overheadpoint of view, was intended for specific environments with excessive delayspreadwhereperformancewas limitedby timedispersion.However, extendedcyclic prefix was not used in practical deployments (except for MBSFNtransmission), rendering it an unnecessary feature in LTE for unicasttransmission.Withthisinmind,NRdefinesanormalcyclicprefixonly,withtheexceptionof60kHzsubcarrierspacing,wherebothnormalandextendedcyclicprefixaredefinedforreasonsdiscussedbelow.

7.2Time-DomainStructureInthetimedomain,NRtransmissionsareorganizedintoframesoflength10ms,each of which is divided into 10 equally sized subframes of length 1 ms. Asubframeisinturndividedintoslotsconsistingof14OFDMsymbolseach,thatis,thedurationofaslotinmillisecondsdependsonthenumerologyasillustratedinFig.7.2.Onahigherlevel,eachframeisidentifiedbyasystemframenumber(SFN). The SFN is used to define different transmission cycles that have a

periodlongerthanoneframe,forexample,pagingsleep-modecycles.TheSFNperiodequals1024,thustheSFNrepeatsitselfafter1024framesor10.24s.

FIGURE7.2 Frames,subframes,andslotsinNR.

Forthe15kHzsubcarrierspacing,anNRslotthushasthesamestructureasan LTE subframe with normal cyclic prefix, which is beneficial from acoexistenceperspectiveasdiscussedabove.NotethatasubframeinNRservesasanumerology-independent time reference,which isuseful, especially in thecaseofmultiplenumerologiesbeingmixedon thesamecarrier,whileaslot isthe typicaldynamicschedulingunit. Incontrast,LTEwith itssinglesubcarrierspacingusesthetermsubframeforboththesepurposes.Since a slot is defined as a fixed number of OFDM symbols, a higher

subcarrierspacingleadstoashorterslotduration.Inprinciplethiscanbeusedtosupport lower-latency transmission, but as the cyclic prefix also shrinkswhenincreasingthesubcarrierspacing,itisnotafeasibleapproachinalldeployments.Therefore, to facilitate a fourfold reduction in the slot duration and theassociateddelaywhilemaintainingacyclicprefixsimilartothe15kHzcase,anextended cyclic prefix is defined for 60 kHz subcarrier spacing. However, itcomesat thecostof increasedoverhead in termsof cyclicprefixand is a lessefficient way of providing low latency. The subcarrier spacing is thereforeprimarily selected to meet the deployment scenario in terms of, for example,carrier frequency, expected delay spread in the radio channel, and anycoexistencerequirementswithLTE-basedsystemsonthesamecarrier.

Analternativeandmoreefficientway to support low latency is todecouplethetransmissiondurationfromtheslotduration.Insteadofchangingsubcarrierspacing and/or slot duration, the latency-critical transmission uses whatevernumber of OFDM symbols necessary to deliver the payload. NR thereforesupportsoccupyingonlypartofaslotfor the transmission,sometimesreferredto as “mini-slot transmission.” In other words, the term slot is primarily anumerology-dependent timereferenceandonlylooselycoupledwith theactualtransmissionduration.Therearemultiplereasonswhyitisbeneficialtoallowtransmissiontooccupy

onlyapartofaslotasillustratedinFig.7.3.Onereasonis,asalreadydiscussed,support of very low latency. Such transmissions can also preempt an alreadyongoing, longer transmission to another device as discussed inSection 14.1.2,allowingforimmediatetransmissionofdatarequiringverylowlatency.

FIGURE7.3 Decouplingtransmissionsfromslotboundariestoachievelowlatency(top),moreefficientbeamsweeping(middle),andbettersupportforunlicensedspectra(bottom).

Another reason is support for analogbeamforming as discussed inChapters11 and 12 where at most one beam at a time can be used for transmission.Differentdevicesthereforeneedtobetime-multiplexedandwiththeverylargebandwidths available in the mm-wave range, a few OFDM symbols can besufficientevenforrelativelylargepayloads.Athirdreasonisoperationinunlicensedspectra.Unlicensedoperationisnot

partofrelease15butwillbeintroducedinalaterrelease.Inunlicensedspectra,listen-before-talk is typically used to ensure the radio channel is available fortransmission. Once the listen-before-talk operation has declared the channelavailable, it is beneficial to start transmission immediately to avoid another

deviceoccupyingthechannel.Ifdatatransmissionwouldhavetowaituntilthestartofaslotboundary,someformofdummydataorreservationsignalneedstobe transmitted fromthesuccessful listen-before-talkoperationuntil thestartoftheslot,whichwoulddegradetheefficiencyofthesystem.

7.3Frequency-DomainStructureWhen the first release of LTE was designed, it was decided that all devicesshouldbecapableofthemaximumcarrierbandwidthof20MHz,whichwasareasonable assumption at the time given the relatively modest bandwidth,compared to NR. On the other hand, NR is designed to support very widebandwidths,upto400MHzforasinglecarrier.Mandatingalldevicestohandlesuch wide carriers is not reasonable from a cost perspective. Hence, an NRdevice may see only a part of the carrier and, for efficient utilization of thecarrier,thepartofthecarrierreceivedbythedevicemaynotbecenteredaroundthecarrierfrequency.Thishasimplicationsfor,amongotherthings,thehandlingoftheDCsubcarrier.InLTE,theDCsubcarrierisnotusedasitmaybesubjecttodisproportionally

high interference due to, for example, local-oscillator leakage. Since all LTEdevicescanreceivethefullcarrierbandwidthandarecenteredaroundthecarrierfrequency,thiswasstraightforward.1NRdevices,ontheotherhand,maynotbecentered around the carrier frequency and each NR device may have its DClocated at different locations in the carrier, unlike LTE where all devicestypicallyhavetheDCcoincidingwiththecenterofthecarrier.Therefore,havingspecialhandlingoftheDCsubcarrierwouldbecumbersomeinNRandinsteaditwasdecidedtoexploitalsotheDCsubcarrierfordataasillustratedinFig.7.4,acceptingthatthequalityofthissubcarriermaybedegradedinsomesituations.

FIGURE7.4 HandlingoftheDCsubcarrierinLTEandNR.

Aresourceelement,consistingofonesubcarrierduringoneOFDMsymbol,isthesmallestphysicalresourceinNR.Furthermore,asillustratedinFig.7.5,12consecutivesubcarriersinthefrequencydomainarecalledaresourceblock.

FIGURE7.5 Resourceelementandresourceblock.

Note that the NR definition of a resource block differs from the LTEdefinition. An NR resource block is a one-dimensional measure spanning thefrequencydomainonly,whileLTEusestwo-dimensionalresourceblocksof12subcarriersinthefrequencydomainandoneslotinthetimedomain.Onereasonfor defining resource blocks in the frequency domain only in NR is theflexibilityintimedurationfordifferenttransmissionswhereas, inLTE,at leastintheoriginalrelease,transmissionsoccupiedacompleteslot.2NR supports multiple numerologies on the same carrier and, consequently,

therearemultipleresourcesetsofresourcegrids,oneforeachnumerology(Fig.

7.6).Sincearesourceblockis12subcarriers,thefrequencyspanmeasuredinHzisdifferent.TheresourceblockboundariesarealignedacrossnumerologiessuchthattworesourceblocksatasubcarrierspacingofΔfoccupythesamefrequencyrange as one resource block at a subcarrier spacing of 2Δf. In the NRspecifications, the alignment across numerologies in terms of resource blockboundaries, as well as symbol boundaries, is described through multipleresource grids where there is one resource grid per subcarrier spacing andantennaport(seeSection7.9foradiscussionofantennaports),coveringthefullcarrierbandwidthinthefrequencydomainandonesubframeinthetimedomain.

FIGURE7.6 Resourcegridsfortwodifferentsubcarrierspacings.

The resource gridmodels the transmitted signal as seenby the device for agiven subcarrier spacing. However, the device needs to know where in thecarrier the resource blocks are located. In LTE, where there is a singlenumerology and all devices support the full carrier bandwidth, this isstraightforward.NR,ontheotherhand,supportsmultiplenumerologiesand,asdiscussedfurtherbelowinconjunctionwithbandwidthparts,notalldevicesmaysupportthefullcarrierbandwidth.Therefore,acommonreferencepoint,knownaspointA, togetherwith the notion of two types of resource blocks, commonresource blocks and physical resource blocks, are used.3 Reference point Acoincides with subcarrier 0 of common resource block 0 for all subcarrierspacings. This point serves as a reference fromwhich the frequency structurecan be described and pointAmay be located outside the actual carrier.UpondetectinganSSblockaspartoftheinitialaccess(seeSection16.1),thedeviceissignalled the location of point A as part of the broadcast system information(SIB1).The physical resource blocks, which are used to describe the actual

transmittedsignal,arethenlocatedrelativetothisreferencepoint,asillustratedinFig.7.7.Forexample,physicalresourceblock0forsubcarrierspacingΔf islocated m resource blocks from reference point A or, expressed differently,

correspondstocommonresourceblockm.Similarly,physicalresourceblock0forsubcarrierspacing2Δfcorrespondstocommonresourceblockn.Thestartingpoints for the physical resource blocks are signaled independently for eachnumerology (m and n in the example in Fig. 7.7), a feature that is useful forimplementing the filters necessary to meet the out-of-band emissionrequirements(seeChapter18).TheguardinHzneededbetweentheedgeofthecarrier and the first used subcarrier is larger, the larger the subcarrier spacing,whichcanbeaccountedforbyindependentlysettingtheoffsetbetweenthefirstusedresourceblockandreferencepointA.IntheexampleinFig.7.7, thefirstusedresourceblockforsubcarrierspacing2Δfislocatedfurtherfromthecarrieredge than for subcarrier spacing Δf to avoid excessively steep filteringrequirements for the higher numerology or, expressed differently, to allow alargerfractionofthespectrumtobeusedforthelowersubcarrierspacing.

FIGURE7.7 Commonandphysicalresourceblocks.

Thelocationofthefirstusableresourceblock,whichisthesameasthestartoftheresourcegridinthefrequencydomain,issignaledtothedevice.Notethatthismayormaynotbethesameasthefirstresourceblockofabandwidthpart(bandwidthpartsaredescribedinSection7.4).AnNRcarriershouldatmostbe275resourceblockswide,whichcorresponds

to 275·12=3300 used subcarriers. This also defines the largest possible carrierbandwidthinNRforeachnumerology.However,thereisalsoanagreementtolimit the per-carrier bandwidth to 400MHz, resulting in themaximumcarrierbandwidths of 50/100/200/400 MHz for subcarrier spacings of

15/30/60/120 kHz, respectively, as mentioned in Chapter 5. The smallestpossiblecarrierbandwidthof11resourceblocksisgivenbytheRFrequirementsonspectrumutilization(seeChapter18).However,forthenumerologyusedfortheSSblock(seeChapter16)at least20resourceblocksarerequired inorderforthedevicetobeabletofindandsynchronizetothecarrier.

7.4BandwidthPartsAsdiscussedabove,LTEisdesignedunder theassumptionthatalldevicesarecapable of the maximum carrier bandwidth of 20MHz. This avoided severalcomplications,forexample,aroundthehandlingoftheDCsubcarrierasalreadydiscussed,whilehavinganegligible impacton thedevicecost. Italsoallowedcontrol channels to span the full carrier bandwidth to maximize frequencydiversity.The same assumption—all devices being able to receive the full carrier

bandwidth—is not reasonable for NR, given the very wide carrier bandwidthsupported. Consequently, means for handling different device capabilities interms of bandwidth support must be included in the design. Furthermore,reception of a very wide bandwidth can be costly in terms of device energyconsumption compared to receiving a narrower bandwidth. Using the sameapproachasinLTEwherethedownlinkcontrolchannelswouldoccupythefullcarrierbandwidthwouldthereforesignificantlyincreasethepowerconsumptionof thedevice.Abetter approach is, as done inNR, to use receiver-bandwidthadaptation such that the device can use a narrower bandwidth formonitoringcontrolchannelsandtoreceivesmall-to-medium-sizeddatatransmissionsandtoopenthefullbandwidthwhenalargeamountofdataisscheduled.Tohandlethesetwoaspects—supportfordevicesnotcapableofreceivingthe

full carrier bandwidth and receiver-side bandwidth adaptation—NR definesbandwidthparts (BWPs)(seeFig.7.8).Abandwidthpart ischaracterizedbyanumerology (subcarrier spacing and cyclic prefix) and a set of consecutiveresource blocks in the numerology of the BWP, starting at a certain commonresourceblock.

FIGURE7.8 Exampleofbandwidthadaptationusingbandwidthparts.

WhenadeviceenterstheconnectedstateithasobtainedinformationfromthePBCHabout thecontrolresourceset (CORESET;seeSection10.1.2)where itcanfindthecontrolchannelusedtoscheduletheremainingsysteminformation(see Chapter 16 for details). The CORESET configuration obtained from thePBCHalsodefinesandactivatestheinitialbandwidthpartinthedownlink.Theinitial active uplink bandwidth part is obtained from the system informationscheduledusingthedownlinkPDCCH.Once connected, a device can be configured with up to four downlink

bandwidthpartsanduptofouruplinkbandwidthpartsforeachservingcell.InthecaseofSULoperation(seeSection7.7), therecanbeup tofouradditionaluplinkbandwidthpartsonthesupplementaryuplinkcarrier.Oneachservingcell,atagiventimeinstantoneoftheconfigureddownlink

bandwidth parts is referred to as the active downlink bandwidth part for theservingcell andoneof theconfigureduplinkbandwidthparts is referred toasthe active uplink bandwidth part for the serving cell. For unpaired spectra adevice may assume that the active downlink bandwidth part and the activeuplink bandwidth part of a serving cell have the same center frequency. Thissimplifies the implementation as a single oscillator can be used for bothdirections.ThegNBcanactivateanddeactivatebandwidthpartsusingthesamedownlink control signaling as for scheduling information (see Chapter 10),therebyachievingrapidswitchingbetweendifferentbandwidthparts.Inthedownlink,adeviceisnotassumedtobeabletoreceivedownlinkdata

transmissions, more specifically the PDCCH or PDSCH, outside the activebandwidth part. Furthermore, the numerology of the PDCCH and PDSCH arerestrictedtothenumerologyconfiguredforthebandwidthpart.Thus,inrelease15,adevicecanonlyreceiveonenumerologyata timeasmultiplebandwidth

partscannotbesimultaneouslyactive.Mobilitymeasurementscanstillbedoneoutside an active bandwidth part but require a measurement gap similarly tointercell measurements. Hence, a device is not expected to monitor downlinkcontrolchannelswhiledoingmeasurementsoutsidetheactivebandwidthpart.In the uplink, a device transmits PUSCH and PUCCH in the active uplink

bandwidthpartonly.Given the above discussion, a relevant question is why two mechanisms,

carrieraggregationandbandwidthparts,aredefinedinsteadofusingthecarrier-aggregation framework only. To some extent carrier aggregation could havebeen used to handle devices with different bandwidth capabilities as well asbandwidth adaptation.However, from anRF perspective there is a significantdifference.AcomponentcarrierisassociatedwithvariousRFrequirementssuchas out-of-band emission requirements as discussed in Chapter 18, but for abandwidthpartinsideacarrierthereisnosuchrequirement—itisallhandledbythe requirements set on the carrier as such. Furthermore, from an MACperspective there are also some differences in the handling of, for example,hybridARQretransmissionswhichcannotmovebetweencomponentcarriers.

7.5Frequency-DomainLocationofNRCarriersIn principle, anNR carrier could be positioned anywherewithin the spectrumand, similarly to LTE, the basic NR physical-layer specification does not sayanything about the exact frequency location of an NR carrier, including thefrequencyband.However, inpractice, there isaneedforrestrictionsonwherean NR carrier can be positioned in the frequency domain to simplify RFimplementation and to provide some structure to carrier assignments in afrequency band between different operators. In LTE, a 100 kHz carrier rasterservedthispurposeandasimilarapproachhasbeentakeninNR.However,theNRrasterhasamuchfinergranularityof5kHzupto3GHzcarrierfrequency,15 kHz for 3–24.25 GHz, and 60 kHz above 24.25 GHz. This raster has thebenefitofbeinga factor in thesubcarrierspacings relevant foreach frequencyrange,aswellasbeingcompatiblewiththe100kHzLTErasterinbandswhereLTEisdeployed(below3GHz).In LTE, this carrier raster also determines the frequency locations a device

mustsearchforaspartoftheinitialaccessprocedure.However,giventhemuchwidercarrierspossibleinNRandthelargernumberofbandsinwhichNRcanbedeployed,aswellasthefinerrastergranularity,performinginitialcellsearch

onallpossiblerasterpositionswouldbetootimeconsuming.Instead,toreducetheoverall complexityandnot spendanunreasonable timeoncell search,NRalsodefinesasparsersynchronizationraster,whichiswhatanNRdevicehastosearchupon initial access.A consequenceof having a sparser synchronizationrasterthancarrierrasteristhat,unlikeLTE,thesynchronizationsignalsmaynotbecenteredinthecarrier(seeFig.7.9andChapter16forfurtherdetails).

FIGURE7.9 NRcarrierraster.

7.6CarrierAggregationThe possibility of carrier aggregation is part of NR from the first release.Similar to LTE, multiple NR carriers can be aggregated and transmitted inparallel to/from the same device, thereby allowing for an overall widerbandwidth and correspondingly higher per-link data rates. The carriers do nothavetobecontiguousinthefrequencydomainbutcanbedispersed,bothinthesamefrequencybandaswellasindifferentbands,resultinginthreedifferencescenarios:

•Intrabandaggregationwithfrequency-contiguouscomponentcarriers;•Intrabandaggregationwithnoncontiguouscomponentcarriers;•Interbandaggregationwithnoncontiguouscomponentcarriers.

Although the overall structure is the same for all three cases, the RFcomplexitycanbevastlydifferent.Up to 16 carriers, possibly of different bandwidths and different duplex

schemes,canbeaggregatedallowingforoveralltransmissionbandwidthsofup16·400MHz=6.4GHz,whichisfarbeyondtypicalspectrumallocations.A device capable of carrier aggregation may receive or transmit

simultaneously on multiple component carriers while a device not capable ofcarrier aggregation can access one of the component carriers. Thus, in most

respects and unless otherwisementioned, the physical-layer description in thefollowing chapters applies to each component carrier separately in the case ofcarrier aggregation. It is worth noting that in the case of interband carrieraggregation ofmultiple half-duplex (TDD) carriers, the transmission directionondifferentcarriersdoesnotnecessarilyhavetobethesame.Thisimpliesthatacarrier-aggregation-capable TDD device may need a duplex filter, unlike thetypicalscenarioforanoncarrier-aggregation-capabledevice.Inthespecifications,carrieraggregationisdescribedusingthetermcell,that

is, a carrier-aggregation-capable device is able to receive and transmit from/tomultiplecells.Oneofthesecellsisreferredtoastheprimarycell(PCell).Thisisthecellwhichthedeviceinitiallyfindsandconnectsto,afterwhichoneormoresecondary cells (SCells) can be configured once the device is in connectedmode. The secondary cells can be rapidly activated or deceived to meet thevariations in the traffic pattern. Different devices may have different cells astheir primary cell—that is, the configuration of the primary cell is device-specific.Furthermore, thenumberofcarriers (orcells)doesnothave tobe thesame in uplink and downlink. In fact, a typical case is to havemore carriersaggregatedinthedownlinkthanintheuplink.Thereareseveralreasonsforthis.There is typicallymore traffic in thedownlink that in theuplink.Furthermore,the RF complexity from multiple simultaneously active uplink carriers istypicallylargerthanthecorrespondingcomplexityinthedownlink.Schedulinggrantsandschedulingassignmentscanbetransmittedoneitherthe

samecellasthecorrespondingdata,knownasself-scheduling,oronadifferentcellthanthecorrespondingdata,knownascross-carrierscheduling,asillustratedinFig.7.10.Inmostcases,self-schedulingissufficient.

FIGURE7.10 Self-schedulingandcross-scheduling.

7.6.1ControlSignaling

Carrier aggregation usesL1/L2 control signaling for the same reason aswhenoperating with a single carrier. The use of downlink controls signaling forschedulinginformationwastoucheduponintheprevioussection.Thereisalsoaneedforuplinkcontrolsignaling,forexample,hybrid-ARQacknowledgmentstoinform the gNB about the success or failure of downlink data reception. Asbaseline, all the feedback is transmitted on the primary cell,motivated by theneed to support asymmetric carrier aggregation with the number of downlinkcarrierssupportedbyadeviceunrelatedtothenumberofuplinkcarriers.Foralargenumberofdownlinkcomponentcarriers,asingleuplinkcarriermaycarryalargenumberofacknowledgments.Toavoidoverloadingasinglecarrier,itispossible to configure twoPUCCHgroupswhere feedback relating to the firstgroupistransmittedintheuplinkofthePCellandfeedbackrelatingtotheothergroupofcarriersistransmittedontheprimarysecondcell(PSCell)(Fig.7.11).

FIGURE7.11 MultiplePUCCHgroups.

Ifcarrieraggregationisused,thedevicemayreceiveandtransmitonmultiplecarriers, but reception on multiple carriers is typically only needed for thehighestdatarates.Itisthereforebeneficialtoinactivatereceptionofcarriersnotused while keeping the configuration intact. Activation and inactivation ofcomponent carriers can be done through MAC signaling (more specifically,MACcontrolelements,discussedinSection6.4.4.1)containingabitmapwhereeachbitindicateswhetheraconfiguredSCellshouldbeactivatedordeactivated.

7.7SupplementaryUplinkIn addition to carrier aggregation, NR also supports so-called “supplementaryuplink” (SUL). As illustrated in Fig. 7.12, SUL implies that a conventionaldownlink/uplink(DL/UL)carrierpairhasanassociatedorsupplementaryuplinkcarrierwiththeSULcarriertypicallyoperatinginlower-frequencybands.Asanexample,adownlink/uplinkcarrierpairoperatinginthe3.5GHzbandcouldbe

complemented with a supplementary uplink carrier in the 800 MHz band.AlthoughFig.7.12 seems to indicate that theconventionalDL/ULcarrierpairoperatesonpairedspectrawithfrequencyseparationbetweenthedownlinkanduplinkcarriers, itshouldbeunderstood that theconventionalcarrierpaircouldequally well operate in unpaired spectra with downlink/uplink separation bymeansofTDD.Thiswould,forexample,bethecaseinanSULscenariowheretheconventionalcarrierpairoperatesintheunpaired3.5GHzband.

FIGURE7.12 SupplementaryuplinkcarriercomplementingaconventionalDL/ULcarrierpair.

Whilethemainaimofcarrieraggregationistoenablehigherpeakdataratesby increasing the bandwidth available for transmission to/from a device, thetypicalaimofSUListoextenduplinkcoverage,thatis,toprovidehigheruplinkdata rates inpower-limited situations,byutilizing the lowerpath lossat lowerfrequencies. Furthermore, in an SUL scenario the non-SUL uplink carrier istypicallysignificantlymorewidebandcomparedtotheSULcarrier.Thus,undergood channel conditions such as the device located relatively close to the cellsite, the non-SUL carrier typically allows for substantially higher data ratescompared to theSULcarrier.At thesame time,underbadchannelconditions,forexample,atthecelledge,alower-frequencySULcarriertypicallyallowsforsignificantly higher data rates compared to the non-SUL carrier, due to theassumedlowerpathlossatlowerfrequencies.Hence,onlyinarelativelylimitedarea do the two carriers provide similar data rates. As a consequence,aggregatingthethroughputofthetwocarriershasinmostcaseslimitedbenefits.At the same time, scheduling only a single uplink carrier at a time simplifiestransmission protocols and in particular the RF implementation as variousintermodulationissuesisavoided.Notethatforcarrieraggregationthesituationisdifferent:

•Thetwo(ormore)carriersinacarrier-aggregationscenarioareoftenofsimilarbandwidthandoperatingatsimilarcarrierfrequencies,making

aggregationofthethroughputofthetwocarriersmorebeneficial;•Eachuplinkcarrierinacarrieraggregationscenarioisoperatingwithitsowndownlinkcarrier,simplifyingthesupportforsimultaneousschedulingofmultipleuplinktransmissionsinparallel.

Hence,onlyoneofSULandnon-SUListransmittingandsimultaneousSULandnon-SULtransmissionfromadeviceisnotpossible.One SUL scenario iswhen the SUL carrier is located in the uplink part of

pairedspectrumalreadyusedbyLTE(seeFig.7.13). Inotherwords, theSULcarrierexistsinanLTE/NRuplinkcoexistencescenario(seealsoChapter17).Inmany LTE deployments, the uplink traffic is significantly less than thecorresponding downlink traffic. As a consequence, in many deployments, theuplink part of paired spectra is not fully utilized. Deploying an NRsupplementaryuplinkcarrierontopoftheLTEuplinkcarrierinsuchaspectrumis a way to enhance theNR user experiencewith limited impact on the LTEnetwork.

FIGURE7.13 SULcarriercoexistingwithLTEuplinkcarrier.

Finally,asupplementaryuplinkcanalsobeusedtoreducelatency.InthecaseofTDD,theseparationofuplinkanddownlinkinthetimedomainmayimposerestrictions on when uplink data can be transmitted. By combining the TDDcarrierwithasupplementarycarrierinpairedspectra,latency-criticaldatacanbetransmittedonthesupplementaryuplinkimmediatelywithoutbeingrestrictedbytheuplink–downlinkpartitioningonthenormalcarrier.

7.7.1RelationtoCarrierAggregationAlthoughSULmayappearsimilartouplinkcarrieraggregationtherearesome

fundamentaldifferences.In thecaseofcarrieraggregation,eachuplinkcarrierhas itsownassociated

downlinkcarrier.Formally,eachsuchdownlinkcarriercorrespondstoacellofits own and thus different uplink carriers in a carrier-aggregation scenariocorrespondtodifferentcells(seeleftpartofFig.7.14).

FIGURE7.14 Carrieraggregationvssupplementaryuplink.

Incontrast,inthecaseofSULthesupplementaryuplinkcarrierdoesnothaveanassociateddownlinkcarrierofitsown.Ratherthesupplementarycarrierandthe conventional uplink carrier share the same downlink carrier. As aconsequence, thesupplementaryuplinkcarrierdoesnotcorrespond toacellofits own. Instead, in theSUL scenario there is a single cellwith one downlinkcarrierandtwouplinkcarriers(rightpartofFig.7.14).Itshouldbenotedthatinprinciplenothingpreventsthecombinationofcarrier

aggregation,forexample,asituationwithcarrieraggregationbetweentwocells(two DL/UL carrier pairs) where one of the cells is an SUL cell with anadditional supplementary uplink carrier.However, there are currently no bandcombinationsdefinedforsuchcarrier-aggregation/SULcombinations.Arelevantquestionis,ifthereisasupplementaryuplink,istheresuchathing

asasupplementarydownlink?Theanswerisyes—sincethecarrieraggregationframework allows for the number of downlink carriers to be larger than thenumber of uplink carriers, some of the downlink carriers can be seen assupplementary downlinks. One common scenario is to deploy an additionaldownlink carrier in unpaired spectra and aggregate it with a carrier in pairedspectra to increase capacity and data rates.No additionalmechanisms beyondcarrier aggregation are needed and hence the term supplementary downlink ismainlyusedfromaspectrumpointofviewasdiscussedinChapter3.

7.7.2ControlSignalingInthecaseofsupplementary-uplinkoperation,adeviceisexplicitlyconfigured

(bymeansofRRCsignaling)totransmitPUCCHoneithertheSULcarrierorontheconventional(non-SUL)carrier.In terms of PUSCH transmission, the device can be configured to transmit

PUSCHon thesamecarrierasPUCCH.Alternatively,adeviceconfigured forSULoperationcanbeconfiguredfordynamicselectionbetweentheSULcarrieror the non-SUL carrier. In the latter case, the uplink scheduling grant willincludeanSUL/non-SULindicator that indicatesonwhatcarrier thescheduledPUSCH transmission should be carried. Thus, in the case of supplementaryuplink, a devicewill never transmit PUSCH simultaneously on both the SULcarrierandonthenon-SULcarrier.AsdescribedinSection10.2,ifadeviceistotransmitUCIonPUCCHduring

atimeintervalthatoverlapswithascheduledPUSCHtransmissiononthesamecarrier, thedevice insteadmultiplexes theUCIontoPUSCH.Thesame rule istruefortheSULscenario,thatis,thereisnotsimultaneousPUSCHandPUCCHtransmissionevenondifferentcarriers.Rather,ifadeviceistotransmitUCIonPUCCHonecarrier(SULornon-SUL)duringatimeintervalthatoverlapswithascheduledPUSCHtransmissiononeithercarrier(SULornonSUL),thedeviceinsteadmultiplexestheUCIontothePUSCH.Analternativetosupplementaryuplinkwouldbetorelyondualconnectivity

withLTEonthelowerfrequencyandNRonthehigherfrequency.UplinkdatatransmissionwouldinthiscasebehandledbytheLTEcarrierwith,fromadatarate perspective, the benefits would be similar to supplementary uplink.However, in this case, the uplink control signaling related to NR downlinktransmissionshastobehandledbythehigh-frequencyNRuplinkcarrieraseachcarrierpairhastobeself-containedintermsofL1/L2controlsignaling.UsingasupplementaryuplinkavoidsthisdrawbackandallowsL1/L2controlsignalingto exploit the lower-frequency uplink. Another possibility would be to usecarrieraggregation,butinthiscasealow-frequencydownlinkcarrierhastobeconfigured as well, somethingwhichmay be problematic in the case of LTEcoexistence.

7.8DuplexSchemesSpectrum flexibility is one of the key features of NR. In addition to theflexibility in transmission bandwidth, the basic NR structure also supportsseparationofuplinkanddownlinkintimeand/orfrequencysubjecttoeitherhalfduplexorfullduplexoperation,allusingthesamesingleframestructure.This

providesalargedegreeofflexibility(Fig.7.15):

•TDD—uplinkanddownlinktransmissionsusethesamecarrierfrequencyandareseparatedintimeonly;

•FDD—uplinkanddownlinktransmissionsusedifferentfrequenciesbutcanoccursimultaneously;

•Half-duplexFDD—uplinkanddownlinktransmissionsareseparatedinfrequencyandtime,suitableforsimplerdevicesoperatinginpairedspectra.

FIGURE7.15 Duplexschemes.

In principle, the same basic NR structure would also allow full duplexoperationwithuplinkanddownlinkseparatedneitherintime,norinfrequency,although this would result in a significant transmitter-to-receiver interferenceproblemwhosesolutionisstillintheresearchstageandleftforthefuture.LTEalsosupportedbothTDDandFDD,butunlikethesingleframestructure

used inNR,LTEused twodifferent frame structure typesused.4Furthermore,unlikeLTEwhere theuplink–downlinkallocationdoesnotchangeover time,5theTDDoperationforNRisdesignedwithdynamicTDDasakeytechnologycomponent.

7.8.1Time-DivisionDuplex(TDD)InthecaseofTDDoperation,thereisasinglecarrierfrequencyanduplinkanddownlinktransmissionsareseparatedinthetimedomainonacellbasis.Uplinkanddownlinktransmissionsarenonoverlappingintime,bothfromacellandadeviceperspective.TDDcanthereforebeclassifiedashalf-duplexoperation.InLTE,thesplitbetweenuplinkanddownlinkresourcesinthetimedomain

wassemistaticallydeterminedandessentiallyremainedconstantovertime.NR,ontheotherhand,usesdynamicTDDasthebasiswhere(partsof)aslotcanbedynamically allocated to either uplink or downlink as part of the schedulerdecision. This enables following rapid traffic variationswhich are particularlypronounced in dense deploymentswith a relatively small number of users perbase station.Dynamic TDD is particularly useful in small-cell and/or isolatedcell deployments where the transmission power of the device and the basestationisofthesameorderandtheintersiteinterferenceisreasonable.Ifneeded,the scheduling decisions between the different sites can be coordinated. It ismuch simpler to restrict the dynamics in the uplink–downlink allocationwhenneededandtherebyhaveamorestaticoperationthantryingtoadddynamicstoafundamentallystaticscheme,whichwasdonewhenintroducingeIMTAforLTEinrelease12.One example when intersite coordination is useful is a traditional

macrodeployment. In such scenarios, a (more or less) static uplink–downlinkallocationisagoodchoiceasitavoidstroublesomeinterferencesituations.Staticor semistatic TDD operation is also necessary for handling coexistence withLTE, forexample,whenanLTEcarrierandanNRcarrierareusing the samesites and the same frequency band. Such restrictions in the uplink–downlinkallocation can easily be achieved as part of the scheduling implementation byusing a fixed pattern in each base station. There is also a possibility tosemistaticallyconfigurethetransmissiondirectionofsomeoralloftheslotsasdiscussed in Section 7.8.3, a feature that can allow for reduced device energyconsumptionas it isnotnecessary tomonitorfordownlinkcontrolchannels inslotsthatareaprioriknowntobereservedforuplinkusage.AnessentialaspectofanyTDDsystem,orhalf-duplexsystemingeneral, is

thepossibilitytoprovideasufficientlylargeguardperiod(orguardtime),whereneitherdownlinknoruplinktransmissionsoccur.Thisguardperiodisnecessaryfor switching from downlink to uplink transmission and vice versa and isobtainedbyusingslotformatswherethedownlinkendssufficientlyearlypriorto the start of the uplink.The required length of the guard period depends onseveralfactors.First,itshouldbesufficientlylargetoprovidethenecessarytimefor the circuitry in base stations and the devices to switch from downlink touplink.Switchingistypicallyrelativelyfast,oftheorderof20µsorless,andinmostdeploymentsdoesnotsignificantlycontributetotherequiredguardtime.Second, the guard time should also ensure that uplink and downlink

transmissionsdonotinterfereat thebasestation.Thisishandledbyadvancing

the uplink timing at the devices such that, at the base station, the last uplinksubframebeforetheuplink-to-downlinkswitchendsbeforethestartofthefirstdownlinksubframe.Theuplink timingofeachdevicecanbecontrolledby thebasestationbyusingthetimingadvancemechanism,aswillbeelaborateduponinChapter15.Obviously, theguardperiodmustbe large enough to allow thedevice to receive the downlink transmission and switch from reception totransmissionbeforeitstartsthe(timing-advanced)uplinktransmission(seeFig.7.16).Asthetimingadvanceisproportionaltothedistancetothebasestation,alargerguardperiodisrequiredwhenoperatinginlargecellscomparedtosmallcells.

FIGURE7.16 CreationofguardtimeforTDDoperation.

Finally, the selection of the guard period also needs to take interferencebetweenbasestationsintoaccount.Inamulticellnetwork,intercellinterferencefromdownlink transmissions in neighboring cellsmust decay to a sufficientlylowlevelbeforethebasestationcanstarttoreceiveuplinktransmissions.Hence,alargerguardperiodthanismotivatedbythecellsizeitselfmayberequiredasthelastpartofthedownlinktransmissionsfromdistantbasestations,otherwiseitmayinterferewithuplinkreception.Theamountofguardperioddependsonthepropagationenvironments,butinsomemacrocelldeploymentstheinterbase-stationinterferenceisanonnegligiblefactorwhendeterminingtheguardperiod.Depending on the guard period, some residual interferencemay remain at thebeginning of the uplink period. Hence, it is beneficial to avoid placinginterference-sensitivesignalsatthestartofanuplinkburst.

7.8.2Frequency-DivisionDuplex(FDD)

In the case of FDD operation, uplink and downlink are carried on differentcarrierfrequencies,denotedfULandfDLinFig.7.15.Duringeachframe,thereisthus a full set of slots in both uplink and downlink, and uplink and downlinktransmissioncanoccursimultaneouslywithinacell.Isolationbetweendownlinkanduplinktransmissionsisachievedbytransmission/receptionfilters,knownasduplex filters, and a sufficiently large duplex separation in the frequencydomain.Evenifuplinkanddownlinktransmissioncanoccursimultaneouslywithina

cell in the case of FDD operation, a device may be capable of full-duplexoperationoronlyhalf-duplexoperationforacertainfrequencyband,dependingonwhether or not it is capable of simultaneous transmission/reception. In thecase of full-duplex capability, transmission and reception may also occursimultaneously at a device, whereas a device capable of only half-duplexoperation cannot transmit and receive simultaneously. Half-duplex operationallowsforsimplifieddeviceimplementationduetorelaxedornoduplex-filters.Thiscanbeusedtoreducedevicecost,forexample,forlow-enddevicesincost-sensitiveapplciations.Anotherexampleisoperationincertainfrequencybandswithaverynarrowduplexgapwithcorrespondinglychallengingdesignof theduplexfilters.Inthiscase,fullduplexsupportcanbefrequency-band-dependentsuchthatadevicemaysupportonlyhalf-duplexoperationincertainfrequencybandswhilebeingcapableof full-duplexoperation in the remaining supportedbands. It should be noted that full/half-duplex capability is a property of thedevice; the base station can operate in full duplex irrespective of the devicecapabilities. For example, the base station can transmit to one device whilesimultaneouslyreceivingfromanotherdevice.From a network perspective, half-duplex operation has an impact on the

sustained data rates that can be provided to/from a single device as it cannottransmitinalluplinksubframes.Thecellcapacityishardlyaffectedastypicallyit is possible to schedule different devices in uplink and downlink in a givensubframe. No provisioning for guard periods is required from a networkperspective as the network is still operating in full duplex and therefore iscapable of simultaneous transmission and reception.The relevant transmissionstructuresandtimingrelationsareidenticalbetweenfull-duplexandhalf-duplexFDDanda singlecellmay thereforesimultaneouslysupportamixtureof full-duplexandhalf-duplexFDDdevices.Sinceahalf-duplexdeviceisnotcapableofsimultaneoustransmissionandreception, theschedulingdecisionsmust takethis into account and half-duplex operation can be seen as a scheduling

restriction.

7.8.3SlotFormatandSlot-FormatIndicationReturning to theslotstructurediscussed inSection7.2, it is important topointout that there is one set of slots in the uplink and another set of slots in thedownlink, the reason being the time offset between the two as a function oftiming advance. If both uplink and downlink transmissionwould be describedusingthesameslot,whichisoftenseeninvariousillustrationsintheliterature,itwould not be possible to specify the necessary timing difference between thetwo.Dependingonthewhetherthedeviceiscapableoffullduplex,asisthecase

forFDD,orhalf duplexonly, as is the case forTDD, a slotmaynot be fullyused for uplink or downlink transmission. As an example, the downlinktransmissioninFig.7.16hadtostoppriortotheendoftheslotinordertoallowfor sufficient time to switch to downlink reception. Since the necessary timebetween downlink and uplink depends on several factors, NR defines a widerange of slot formats defining which parts of a slot are used for uplink ordownlink.EachslotformatrepresentsacombinationofOFDMsymbolsdenoteddownlink,flexible,anduplink,respectively.Thereasonforhavingathirdstate,flexible, will be discussed further below, but one usage is to handle thenecessary guard period in half-duplex schemes. A subset of the slot formatssupported by NR are illustrated in Fig. 7.17. As seen in the figure, there aredownlink-only and uplink-only slot formats which are useful for full-duplexoperation(FDD),aswellaspartiallyfilleduplinkanddownlinkslotstohandlethecaseofhalf-duplexoperation(TDD).

FIGURE7.17 AsubsetofthepossibleslotformatsinNR(“D”isdownlink,“U”isuplink,and“–”isflexible).

Thenameslot format is somewhatmisleadingas thereareseparateslots foruplinkanddownlinktransmissions,eachfilledwithdatainsuchawaythatthereis no simultaneous transmission and reception in the caseofTDD.Hence, theslotformatforadownlinkslotshouldbeunderstoodasdownlinktransmissionscanonlyoccurin“downlink”or“flexible”symbols,andinanuplinkslot,uplinktransmissions can only occur in “uplink” or “flexible” symbols. Any guardperiodnecessaryforTDDoperationistakenfromtheflexiblesymbols.One of the key features of NR is, as already mentioned, the support for

dynamic TDD where the scheduler dynamically determines the transmissiondirection. Since a half-duplex device cannot transmit and receivesimultaneously,thereisaneedtosplittheresourcesbetweenthetwodirections.InNR, threedifferent signalingmechanismsprovide information to thedeviceonwhethertheresourcesareusedforuplinkordownlinktransmission:

•Dynamicsignalingforthescheduleddevice;•SemistaticsignalingusingRRC;and•Dynamicslot-formatindicationsharedbyagroupofdevices.

Some or all of thesemechanisms are used in combination to determine the

instantaneous transmission direction aswill be discussed below.Although thedescriptionbelowuses the termdynamicTDD,theframeworkcaninprinciplebeappliedtohalf-duplexoperationingeneral,includinghalf-duplexFDD.The firstmechanismand thebasicprinciple is for thedevice tomonitor for

controlsignalinginthedownlinkandtransmit/receiveaccordingtothereceivedscheduling grants/assignments. In essence, a half-duplex device would vieweach OFDM symbol as a downlink symbol unless it has been instructed totransmitintheuplink.Itisuptotheschedulertoensurethatahalf-duplexdeviceis not requested to simultaneously receive and transmit. For a full-duplex-capabledevice (FDD), there isobviouslynosuch restrictionand theschedulercanindependentlyscheduleuplinkanddownlink.The general principle above is simple and provides a flexible framework.

However, if the network knows a priori that it will follow a certain uplink–downlink allocation, for example, in order to provide coexistence with someotherTDDtechnologyortofulfillsomespectrumregulatoryrequirement,itcanbeadvantageoustoprovidethis informationtothedevice.Forexample, if it isknown to a device that a certain set of OFDM symbols is assigned to uplinktransmissions,thereisnoneedforthedevicetomonitoringfordownlinkcontrolsignalinginthepartofthedownlinkslotsoverlappingwiththesesymbols.Thiscan help reducing the device power consumption. NR therefore provides thepossibility to optionally signal the uplink–downlink allocation through RRCsignaling.The RRC-signaled pattern classifies OFDM symbols as “downlink,”

“flexible,” or “uplink.” For a half-duplex device, a symbol classified as“downlink” can only be used for downlink transmission with no uplinktransmission in the same period of time. Similarly, a symbol classified as“uplink” means that the device should not expect any overlapping downlinktransmission.“Flexible”meansthatthedevicecannotmakeanyassumptionsonthetransmissiondirection.Downlinkcontrolsignalingshouldbemonitoredandifaschedulingmessageisfound,thedeviceshouldtransmit/receiveaccordingly.Thus, the fully dynamic scheme outlined above is equivalent to semistaticallydeclaringallsymbolsas“flexible.”The RRC-signaled pattern is expressed as a concatenation of up to two

sequencesofdownlink–flexible–uplink,togetherspanningaconfigurableperiodfrom0.5msupto10ms.Furthermore,twopatternscanbeconfigured,onecell-specific provided as part of system information and one signaled in a device-specificmanner.Theresultingpatternisobtainedbycombiningthesetwowhere

thededicatedpatterncanfurtherrestricttheflexiblesymbolssignaledinthecell-specific pattern to be either downlink or uplink.Only if both the cell-specificpattern and the device-specific pattern indicate flexible should the symbols beforflexibleuse(Fig.7.18).

FIGURE7.18 Exampleofcell-specificanddevice-specificuplink–downlinkpatterns.

The third mechanism is to dynamically signal the current uplink–downlinkallocationtoagroupofdevicesmonitoringaspecialdownlinkcontrolmessageknownastheslot-formatindicator(SFI).Similartothepreviousmechanism,theslot format can indicate the number of OFDM symbols that are downlink,flexible,oruplink,andthemessageisvalidforoneormoreslots.The SFI message will be received by a configured group of one or more

devicesandcanbeviewedasapointerintoanRRC-configuredtablewhereeachrowinthetableisconstructedfromasetofpredefineddownlink/flexible/uplinkpatterns one slot in duration.Upon receiving the SFI, the value is used as anindex into theSFI table toobtain theuplink–downlinkpatternforoneormoreslots as illustrated inFig. 7.19.The set of predefineddownlink/flexible/uplinkpatterns is listed in the NR specifications and covers a wide range ofpossibilities,someexamplesofwhichcanbeseeninFig.7.17andintheleftpartofFig.7.19.TheSFIcanalsoindicatetheuplink–downlinksituationsforothercells(cross-carrierindication).

FIGURE7.19 ExampleofconfiguringtheSFItable.

Since a dynamically scheduled device will know whether the carrier iscurrently used for uplink transmission or downlink transmission from itsscheduling assignment/grant, the group-common SFI signaling is primarilyintendedfornon-scheduleddevices.Inparticular,itoffersthepossibilityforthenetworktooverruleperiodictransmissionsofuplinksoundingsignals(SRS)ordownlink measurements on channel-state information reference signals (CSI-RS).TheSRS transmissions andCSI-RSmeasurements areused for assessingthe channel quality as discussed in Chapter 8, and can be semistaticallyconfigured. Overriding the periodic configuration can be useful in a networkrunningwithdynamicTDD(seeFig.7.20foranexampleillustration).

FIGURE7.20 ControllingperiodicCSI-RSmeasurementsandSRStransmissionsbyusingtheSFI.

The SFI cannot override a semistatically configured uplink or downlinkperiod, neither can it override a dynamically scheduled uplink or downlinktransmission which takes place regardless of the SFI. However, the SFI can

overrideasymbolperiodsemistatically indicatedasflexiblebyrestricting it tobedownlinkoruplink.Itcanalsobeusedtoprovideareservedresource;ifboththeSFIandthesemistaticsignalingindicateacertainsymboltobeflexible,thenthesymbolshouldbe treatedasreservedandnotbeusedfor transmission,norshouldthedevicemakeanyassumptionsonthedownlinktransmission.Thiscanbeusefulasa tool to reserveresourceonanNRcarrier, forexample,usedforotherradio-accesstechnologiesorforfeaturesaddedtofuturereleasesoftheNRstandard.Thedescriptionabovehasfocusedonhalf-duplexdevicesingeneralandTDD

inparticular.However, theSFIcanbeusefulalsoforfull-duplexsystemssuchas FDD, for example, to override periodic SRS transmissions. Since there aretwoindependent“carriers”inthiscase,oneforuplinkandonefordownlink,twoSFIs are needed, one for each carrier. This is solved by using the multislotsupportintheSFI;oneslotisinterpretedasthecurrentSFIforthedownlinkandtheotherasthecurrentSFIfortheuplink.

7.9AntennaPortsDownlink multiantenna transmission is a key technology of NR. Signalstransmitted from different antennas or signals subject to different, and for thereceiver unknown, multiantenna precoders (see Chapter 9), will experiencedifferent “radio channels” even if the set of antennas are located at the samesite.6In general, it is important for a device to understandwhat it can assume in

terms of the relationship between the radio channels experienced by differentdownlinktransmissions.This is, forexample, important inorderfor thedeviceto be able to understand what reference signal(s) should be used for channelestimationforacertaindownlinktransmission.It isalsoimportant inorderforthe device to be able to determine relevant channel-state information, forexample,forschedulingandlink-adaptationpurposes.Forthisreason,theconceptofantennaportisusedintheNR,followingthe

sameprinciplesasinLTE.Anantennaportisdefinedsuchthatthechanneloverwhich a symbol on the antenna port is conveyed can be inferred from thechannel over which another symbol on the same antenna port is conveyed.Expresseddifferently,eachindividualdownlinktransmissioniscarriedoutfroma specific antenna port, the identity of which is known to the device.Furthermore, the device can assume that two transmitted signals have

experiencedthesameradiochannelifandonlyif theyaretransmittedfromthesameantennaport.7In practice, each antenna port can, at least for the downlink, be seen as

correspondingtoaspecificreferencesignal.Adevicereceivercanthenassumethat this reference signal canbeused toestimate thechannel corresponding tothespecificantennaport.Thereferencesignalscanalsobeusedbythedevicetoderivedetailedchannel-stateinformationrelatedtotheantennaport.ThesetofantennaportsdefinedinNRisoutlinedinTable7.2.Asseeninthe

table,thereisacertainstructureintheantennaportnumberingsuchthatantennaports for different purposes have numbers in different ranges. For example,downlink antenna ports starting with 1000 are used for PDSCH. DifferenttransmissionlayersforPDSCHcanuseantennaportsinthisseries,forexample,1000and1001foratwo-layerPDSCHtransmission.Thedifferentantennaportsand their usage will be discussed in more detail in conjunction with therespectivefeature.

Table7.2

AntennaPortsinNR

AntennaPort Uplink Downlink0-series PUSCHandassociatedDM-RS –

1000-series SRS,precodedPUSCH PDSCH

2000-series PUCCH PDCCH

3000-series – CSI-RS

4000-series PRACH SSblock

Itshouldbeunderstood thatanantennaport isanabstractconcept thatdoesnotnecessarilycorrespondtoaspecificphysicalantenna:

•Twodifferentsignalsmaybetransmittedinthesamewayfrommultiplephysicalantennas.Adevicereceiverwillthenseethetwosignalsaspropagatingoverasinglechannelcorrespondingtothe“sum”ofthechannelsofthedifferentantennasandtheoveralltransmissioncouldbeseenasatransmissionfromasingleantennaportbeingthesameforthetwosignals.

•Twosignalsmaybetransmittedfromthesamesetofantennasbutwithdifferent,forthereceiverunknown,antennatransmitter-sideprecoders.Areceiverwillhavetoseetheunknownantennaprecodersaspartoftheoverallchannelimplyingthatthetwosignalswillappearashavingbeentransmittedfromtwodifferentantennaports.Itshouldbenotedthatiftheantennaprecodersofthetwotransmissionswouldhavebeenknowntobethesame,thetransmissionscouldhavebeenseenasoriginatingfromthesameantennaport.Thesamewouldhavebeentrueiftheprecoderswouldhavebeenknowntothereceiveras,inthatcase,theprecoderswouldnotneedtobeseenaspartoftheradiochannel.

ThelastofthesetwoaspectsmotivatestheintroductionofQCLframeworkasdiscussedinthenextsection.

7.10Quasi-ColocationEven if two signals have been transmitted from two different antennas, thechannels experienced by the two signals may still have many large-scalepropertiesincommon.Asanexample,thechannelsexperiencedbytwosignalstransmittedfromtwodifferentantennaportscorrespondingtodifferentphysicalantennas at the same site will, even if being different in the details, typicallyhavethesameoratleastsimilarlarge-scaleproperties,forexample,intermsofDoppler spread/shift, average delay spread, and average gain. It can also beexpected that the channelswill introduce similar average delay.Knowing thatthe radio channels corresponding to two different antenna ports have similarlarge-scale properties can be used by the device receiver, for example, in thesettingofparametersforchannelestimation.Incaseofsingle-antennatransmission, this isstraightforward.However,one

integral part of NR is the extensive support for multiantenna transmission,beamforming, and simultaneous transmission from multiple geographicallyseparatessites.Inthesecases,thechannelsofdifferentantennaportsrelevantforadevicemaydifferevenintermsoflarge-scaleproperties.Forthisreason,theconceptofquasi-colocationwithrespecttoantennaports

is part of NR. A device receiver can assume that the radio channelscorresponding to two different antenna ports have the same large-scalepropertiesintermsofspecificparameterssuchasaveragedelayspread,Dopplerspread/shift,averagedelay,averagegain,andspatialRxparametersifandonly

if theantennaportsarespecifiedasbeingquasi-colocated.Whetherornot twospecific antenna ports can be assumed to be quasi-colocatedwith respect to acertainchannelpropertyisinsomecasesgivenbytheNRspecification.Inothercases, the device may be explicitly informed by the network by means ofsignalingiftwospecificantennaportscanbeassumedtobequasi-colocatedornot.The general principle of quasi-colocation is present already in the later

releasesofLTEwhen it comes to the temporalparameters.However,with theextensive support for beamforming in NR, the QCL framework has beenextended to the spatial domain. Spatial quasi-colocation or, more formally,quasi-colocation with respect to RX parameters is a key part of beammanagement. Although somewhat vague in its formal definition, in practicespatialQCLbetweentwodifferentsignalsimpliesthattheyaretransmittedfromthesameplaceandinthesamebeam.Asaconsequence,ifadeviceknowsthatacertainreceiverbeamdirectionisgoodforoneofthesignals,itcanassumethatthesamebeamdirectionissuitablealsoforreceptionoftheothersignal.Inatypicalsituation,theNRspecificationstatesthatcertaintransmissions,for

example,PDSCHandPDCCHtransmissions,arespatiallyquasi-colocatedwithspecific reference signals, for example, CSI-RS or SS block. The devicemayhavedecidedona specific receiverbeamdirectionbasedonmeasurementsonthe reference signal in question and the device can then assume that the samebeamdirectionisagoodchoicealsoforthePDSCH/PDCCHreception.

1Inthecaseofcarrieraggregation,multiplecarriersmayusethesamepoweramplifier,inwhichcasetheDCsubcarrierofthetransmissiondoesnotnecessarilycoincidewiththeunusedDCsubcarrierintheLTEgrid.2TherearesomesituationsinLTE,forexample,theDwPTSinLTE/TDD,whereatransmissiondoesnotoccupyafullslot.3Thereisathirdtypeofresourceblock,virtualresourceblocks,whicharemappedtophysicalresourceblockswhendescribingthemappingofthePDSCH/PUSCH(seeChapter9:Transport-ChannelProcessing).4Originally,LTEsupportedframestructuretype1forFDDandframestructuretype2forTDD,butinlaterreleasesframestructuretype3wasaddedtohandleoperationinunlicensedspectra.5InLTERel-12theeIMTAfeatureprovidessomesupportfortime-varying

uplink–downlinkallocation.6Anunknowntransmitter-sideprecoderneedstobeseenaspartoftheoverallradiochannel.7Forcertainantennaports,morespecificallythosethatcorrespondtoso-calleddemodulationreferencesignals,theassumptionofthesameradiochannelisonlyvalidwithinagivenschedulingoccasion.

CHAPTER8

ChannelSounding

Abstract

ThischapterdescribestheNRsupportforchannelsounding.Especially,itdescribesthespecificreferencesignals,downlinkchannel-state-informationreferencesignals(CSI-RS)anduplinksoundingreferencesignals(SRS),onwhichchannelsoundingistypicallybased.Italsoprovidesanoverviewofthe NR framework for downlink physical-layer measurements andcorrespondingdevicereportingtothenetwork.

KeywordsChannelsounding;CSI-RS;CSI-IM;sounding-referencesignals;SRS;measurements;reporting

Many transmission features inmodern radio-access technologies are based ontheavailabilityofmoreorlessdetailedknowledgeaboutdifferentcharacteristicsof the radio channel overwhich a signal is to be transmitted.Thismay rangefrom rough knowledge of the radio-channel path loss for transmit-poweradjustmenttodetailedknowledgeaboutthechannelamplitudeandphaseinthetime, frequency, and/or spatial domain. Many transmission features will alsobenefitfromknowledgeabouttheinterferencelevelexperiencedatthereceiverside.Such knowledge about different channel characteristics can be acquired in

differentways and bymeasurements on either the transmitter side or receiverside of a radio link. As an example, knowledge about downlink channelcharacteristicscanbeacquiredbymeansofdevicemeasurements.Theacquiredinformation could then be reported to the network for the setting of differenttransmissionparametersforsubsequentdownlinktransmissions.Alternatively,ifit can be assumed that the channel is reciprocal, that is, the channel

characteristicsofinterestarethesameinthedownlinkanduplinktransmissiondirections, the network can, by itself, acquire knowledge about relevantdownlink channel characteristics by estimating the same characteristics in theuplinkdirection.The same alternatives exist when it comes to acquiring knowledge about

uplinkchannelcharacteristics:

•Thenetworkmaydeterminetheuplinkcharacteristicsofinterestandeitherprovidetheinformationtothedeviceordirectlycontrolsubsequentuplinktransmissionsbasedontheacquiredchannelknowledge;

•Assumingchannelreciprocity,thedevicemay,byitself,acquireknowledgeabouttherelevantuplinkchannelcharacteristicsbymeansofdownlinkmeasurements.

Regardless of the exact approach to acquire channel knowledge, there istypically a need for specific signals onwhich a receiver canmeasure/estimatechannelcharacteristicsofinterest.Thisisoftenexpressedaschannelsounding.This chapter will describe the NR support for such channel sounding.

Especially, wewill describe the specific reference signals, downlink channel-state-information reference signals (CSI-RS) and uplink sounding referencesignals (SRS), on which channel sounding is typically based. We will alsoprovide an overview of the NR framework for downlink physical-layermeasurementsandcorrespondingdevicereportingtothenetwork.

8.1DownlinkChannelSounding—CSI-RSIn the first release of LTE (release 8), channel knowledge for the downlinktransmissiondirectionwassolelyacquiredbymeansofdevicemeasurementsonthe so-called cell-specific reference signals (CRS). The LTE CRS aretransmitted over the entire carrier bandwidth within every LTE subframe oflength1ms,andcanbeassumedtobetransmittedovertheentirecellarea.Thus,adeviceaccessinganLTEnetworkcanassumethatCRSarealwayspresentandcanbemeasuredon.In LTE release 10 the CRS were complemented by so-called CSI-RS. In

contrast toCRS,theLTECSI-RSarenotnecessarily transmittedcontinuously.Rather, anLTEdevice is explicitly configured tomeasureon a set ofCSI-RS

anddoesnotmakeanyassumptionsregardingthepresenceofaCSI-RSunlessitisexplicitlyconfiguredforthedevice.TheoriginfortheintroductionofCSI-RSwastheextensionofLTEtosupport

spatial multiplexing with more than four layers, something which was notpossiblewiththerelease-8CRS.However,theuseofCSI-RSwassoonfoundtobe an, in general, more flexible and efficient tool for channel sounding,compared to CRS. In later releases of LTE, the CSI-RS concept was furtherextended to also support, for example, interference estimation andmulti-pointtransmission.Asalreadydescribed,akeydesignprinciple for thedevelopmentofNRhas

beentoasmuchaspossibleavoid“alwayson”signals.Forthisreason,therearenoCRS-like signals inNR.Rather, theonly “always-on”NRsignal is the so-calledSSblock(seeChapter16)whichistransmittedoveralimitedbandwidthandwithamuchlargerperiodicitycomparedtotheLTECRS.TheSSblockcanbeusedforpowermeasurementstoestimate,forexample,pathlossandaveragechannelquality.However,duetothelimitedbandwidthandlowdutycycle,theSSblock is not suitable formore detailed channel sounding aimed at trackingchannelpropertiesthatvaryrapidlyintimeand/orfrequency.Instead the concept ofCSI-RS is reused inNRand further extended to, for

example,providesupportforbeammanagementandmobilityasacomplementtoSSblock.

8.1.1BasicCSI-RSStructureAconfiguredCSI-RSmaycorrespondtoupto32differentantennaports,eachcorrespondingtoachanneltobesounded.InNR,aCSI-RSisalwaysconfiguredonaper-devicebasis.Itisimportantto

understandthoughthatconfigurationonaper-devicebasisdoesnotnecessarilymean that a transmittedCSI-RScanonlybeusedby a single device.Nothingprevents identical CSI-RS using the same set of resource elements to beseparatelyconfiguredformultipledevices,inpracticeimplyingthatasingleCS-RSissharedbetweenthedevicesAs illustrated in Fig. 8.1, a single-port CSI-RS occupies a single resource

element within a block corresponding to one resource block in the frequencydomain and one slot in the time domain. In principle, the CSI-RS can beconfigured to occur anywherewithin this block although in practice there aresomerestrictionstoavoidcollisionswithotherdownlinkphysicalchannelsand

signals.Especially,adevicecanassumethattransmissionofaconfiguredCSI-RSwillnotcollidewith:

•AnyCORESETconfiguredforthedevice;•DemodulationreferencesignalsassociatedwithPDSCHtransmissionsscheduledforthedevice;

•TransmittedSSblocks.

FIGURE8.1 Single-portCSI-RSstructureconsistingofasingleresourceelementwithinanRB/slotblock.

Amulti-port CSI-RS can be seen asmultiple orthogonally transmitted per-antenna-portCSI-RS sharing the overall set of resource elements assigned fortheconfiguredmulti-portCSI-RS.Inthegeneralcase,thissharingisbasedonacombinationof:

•Code-domainsharing(CDM),implyingthatdifferentper-antenna-portCSI-RSaretransmittedonthesamesetofresourceelementswithseparationachievedbymodulatingtheCSI-RSwithdifferentorthogonalpatterns;

•Frequency-domainsharing(FDM),implyingthatdifferentper-antenna-portCSI-RSaretransmittedondifferentsubcarrierswithinanOFDMsymbol;

•Time-domainsharing(TDM),implyingthatdifferentper-antenna-portCSI-RSaretransmittedindifferentOFDMsymbolswithinaslot.

Furthermore, as illustrated in Fig. 8.2,CDMbetween different per-antenna-portCSI-RScanbe:

•InthefrequencydomainwithCDMovertwoadjacentsubcarriers(2×CDM),allowingforcode-domainsharingbetweentwoper-antenna-portCSI-RS;

•InthefrequencyandtimedomainwithCDMovertwoadjacentsubcarriersandtwoadjacentOFDMsymbols(4×CDM),allowingforcode-domainsharingbetweenuptofourper-antenna-portCSI-RS;

•InthefrequencyandtimedomainwithCDMovertwoadjacentsubcarriersandfouradjacentOFDMsymbols(8×CDM),allowingforcode-domainsharingbetweenuptoeightper-antenna-portCSI-RS.

FIGURE8.2 DifferentCDMstructuresformultiplexingper-antenna-portCSI-RS.

ThedifferentCDMalternativesofFig.8.2,incombinationwithFDMand/orTDM, can then be used to configure different multi-port CSI-RS structureswhere, in general, anN-port CSI-RS occupies a total ofN resource elementswithinanRB/slotblock.1Asafirstexample,Fig.8.3illustrateshowatwo-portCSI-RSconsistsoftwo

adjacent resource elements in the frequencydomainwith sharingbymeansofCDM.Inotherwords,thetwo-portCSI-RShasastructureidenticaltothebasic2×CDMstructureinFig.8.2.

FIGURE8.3 Structureoftwo-portCSI-RSbasedon2×CDM.Thefigurealsoillustratestheorthogonalpatternsofeachport.

InthecaseofCSI-RScorrespondingtomorethantwoantennaportsthereissomeflexibilityinthesensethat,foragivennumberofports,therearemultipleCSI-RSstructuresbasedondifferentcombinationsofCDM,TDM,andFDM.Asan example, there are threedifferent structures for an eight-portCSI-RS

(seeFig.8.4).

•Frequency-domainCDMovertworesourceelements(2×CDM)incombinationwithfourtimesfrequencymultiplexing(leftpartofFig.8.4).TheoverallCSI-RSresourcethusconsistsofeightsubcarrierswithinthesameOFDMsymbol.

•Frequency-domainCDMovertworesourceelements(2×CDM)incombinationwithfrequencyandtimemultiplexing(middlepartFig.8.4).TheoverallCSI-RSresourcethusconsistsoffoursubcarrierswithintwoOFDMsymbols.

•Time/frequency-domainCDMoverfourresourceelements(4×CDM)incombinationwithtwotimesfrequencymultiplexing.TheoverallCSI-RSresourcethusonceagainconsistsoffoursubcarrierswithintwoOFDMsymbols.

FIGURE8.4 Threedifferentstructuresforeight-portCSI-RS.

Finally,Fig. 8.5 illustratesoneoutof threepossible structures for a32-portCSI-RS based on a combination of 8×CDM and four times frequencymultiplexing.ThisexamplealsoillustratesthatCSI-RSantennaportsseparatedin the frequency domain do not necessarily have to occupy consecutivesubcarriers. Likewise, CSI-RS ports separated in the time domain do notnecessarilyhavetooccupyconsecutiveOFDMsymbols.

FIGURE8.5 Onestructure(outofthreesupportedstructures)fora32-portCSI-RS.

Inthecaseofamulti-portCSI-RS,theassociationbetweenper-portCSI-RSandportnumberisdonefirstintheCDMdomain,theninthefrequencydomain,and finally in the timedomain.Thiscan, forexample,beseen from theeight-port example of Fig. 8.4where per-portCSI-RS separated bymeans ofCDMcorrespondtoconsecutiveportnumbers.Furthermore,fortheFDM+TDMcase(centerpartofFig.8.4),portnumberzerotoportnumberthreearetransmittedwithin the sameOFDMsymbol,whileportnumber four toportnumber sevenarejointlytransmittedwithinanotherOFDMsymbol.PortnumberzerotothreeandportnumberfourtosevenarethusseparatedbymeansofTDM.

8.1.2Frequency-DomainStructureofCSI-RSConfigurationsA CSI-RS is configured for a given downlink bandwidth part and is thenassumed tobeconfinedwithin thatbandwidthpart anduse thenumerologyofthebandwidthpart.TheCSI-RScanbeconfiguredtocoverthefullbandwidthof thebandwidth

partorjustafractionofthebandwidth.Inthelattercase,theCSI-RSbandwidthand frequency-domain starting position are provided as part of the CSI-RSconfiguration.Within the configuredCSI-RS bandwidth, aCSI-RSmay be configured for

transmissionineveryresourceblock,referredtoasCSI-RSdensityequaltoone.However, a CSI-RS may also be configured for transmission only in every

secondresourceblock, referred toasCSI-RSdensityequal to1/2. In the lattercase, theCSI-RS configuration includes information about the set of resourceblocks(oddresourceblocksorevenresourceblocks)withinwhichtheCSI-RSwill be transmitted. CSI-RS density equal to 1/2 is not supported for CSI-RSwith4,8,and12antennaports.Thereisalsoapossibilitytoconfigureasingle-portCSI-RSwithadensityof

3 in which case the CSI-RS occupies three subcarriers within each resourceblock.ThisCSI-RSstructure isusedaspartofa so-calledTrackingReferencesignal(TRS)(seefurtherdetailsinSection8.1.7).

8.1.3Time-DomainPropertyofCSI-RSConfigurationsTheper-resource-blockCSI-RSstructureoutlinedabovedescribesthestructureof a CSI-RS transmission, assuming the CSI-RS is actually transmitted in agivenslot.Ingeneral,aCSI-RScanbeconfiguredforperiodic,semi-persistent,oraperiodictransmission.In the case of periodic CSI-RS transmission, a device can assume that a

configuredCSI-RStransmissionoccurseveryNthslot,whereN rangesfromaslowas four, that is,CSI-RS transmissionsevery fourthslot, toashighas640,thatis,CSI-RStransmissiononlyevery640thslot.Inadditiontotheperiodicity,the device is also configured with a specific slot offset for the CSI-RStransmission(seeFig.8.6).

FIGURE8.6 ExamplesofCSI-RSperiodicityandslotoffset.

In the case of semi-persistent CSI-RS transmission, a certain CSI-RSperiodicityandcorrespondingslotoffsetareconfiguredinthesamewayasforperiodic CSI-RS transmission. However, actual CSI-RS transmission can beactivated/deactivated based onMACcontrol elements (MACCE) (see Section6.4.4).OncetheCSI-RStransmissionhasbeenactivated,thedevicecanassume

that the CSI-RS transmission will continue according to the configuredperiodicity until it is explicitly deactivated. Similarly, once the CSI-RStransmissionhasbeendeactivated, thedevicecanassume that therewillbenoCSI-RS transmissions according to the configuration until it is explicitly re-activated.InthecaseofaperiodicCSI-RS,noperiodicityisconfigured.Rather,adevice

is explicitly informed (“triggered”) about eachCSI-RS transmission instant bymeansofsignalingintheDCI.It should be mentioned that the property of periodic, semi-persistent, or

aperiodic is strictly speakingnotapropertyof theCSI-RS itselfbut rather theproperty of a CSI-RS resource set (see Section 8.1.6). As a consequence,activation/deactivation and triggeringof semi-persistent and aperiodicCSI-RS,respectively,isnotdoneforaspecificCSI-RSbutforthesetofCSI-RSwithinaresourceset.

8.1.4CSI-IM—ResourcesforInterferenceMeasurementsAconfiguredCSI-RScanbeusedtoderiveinformationaboutthepropertiesofthechanneloverwhichtheCSI-RSistransmitted.ACSI-RScanalsobeusedtoestimatetheinterferencelevelbysubtractingtheexpectedreceivedsignalfromwhatisactuallyreceivedontheCSI-RSresource.However,theinterferencelevelcanalsobeestimatedfrommeasurementson

so-calledCSI-IM(InterferenceMeasurement)resources.Fig.8.7 illustrates thestructureofaCSI-IMresource.Ascanbeseen, there

are twodifferentCSI-IM structures, each consistingof four resource elementsbut with different time/frequency structures. Similar to CSI-RS, the exactlocationoftheCSI-IMresourcewithintheRB/slotblockisflexibleandpartoftheCSI-IMconfiguration.

FIGURE8.7 AlternativestructuresforaCSI-IMresource.

Thetime-domainpropertyofCSI-IMresourcesisthesameasthatofCSI-RS,that is, a CSI-IM resource could be periodic, semi-persistent(activation/deactivationbymeansofMACCE),oraperiodic(triggeredbyDCI).Furthermore, for periodic and semi-persistent CSI-IM, the set of supportedperiodicitiesisthesameasforCSI-RS.Ina typicalcase,aCSI-IMresourcewouldcorrespond to resourceelements

wherenothingistransmittedwithinthecurrentcellwhiletheactivitywithintheCSI-IMresourceinneighborcellsshouldcorrespondtonormalactivityofthosecells.Thus,bymeasuringthereceiverpowerwithinaCSI-IMresource,adevicewould get an estimate on the typical interference due to transmissions withinothercells.As there should be no transmissions on CSI-IM resources within the cell,

devicesshouldbeconfiguredwiththecorrespondingresourcesasso-calledZP-CSI-RSresources(seebelow).

8.1.5Zero-PowerCSI-RSTheCSI-RSdescribedaboveshouldmorecorrectlybereferredtoasnon-zero-power(NZP)CSI-RStodistinguishthemfromso-calledzero-power(ZP)CSI-RSthatcanalsobeconfiguredforadevice.If a device is scheduled for PDSCH reception on a resource that includes

resourceelementsonwhichaconfiguredCSI-RSistobetransmitted,thedevicecan assume that the PDSCH ratematching and resourcemapping avoid thoseresource elements. However, a device may also be scheduled for PDSCHreceptiononaresourcethatincludesresourceelementscorrespondingtoaCSI-RSconfiguredforadifferentdevice.ThePDSCHmustalsointhiscaseberatematchedaroundtheresourceelementsusedforCSI-RS.TheconfigurationofaZP-CSI-RS is away to inform the device forwhich the PDSCH is scheduledaboutsuchratematching.AconfiguredZP-CSI-RScorresponds toasetof resourceelementswith the

samestructureasanNZP-CSI-RS.However,whileadevicecanassumethatanNZP-CI-RSisactuallytransmittedandissomethingonwhichadevicecancarryout measurements, a configured ZP-CSI-RS only indicates a set of resourceblockstowhichthedeviceshouldassumethatPDSCHisnotmapped.Itshouldbeemphasized that,despite thename,adevicecannotassumethat

there are no transmissions (zero power) within the resource elementscorrespondingtoaconfiguredZP-CSI-RS.Asalreadymentioned,theresources

corresponding to a ZP-CSI-RSmay, for example, be used for transmission ofNZP-CSI-RS configured for other devices.What the NR specification says isthatadevicecannotmakeanyassumptionsregardingtransmissionsonresourcescorrespondingtoaconfiguredZP-CSI-RSandthatPDSCHtransmissionforthedevice is notmapped to resource elements corresponding to a configuredZP-CSI-RS.

8.1.6CSI-RSResourceSetsIn addition tobeingconfiguredwithCSI-RS, adevice canbe configuredwithone or several CSI-RS resource sets, officially referred to as NZP-CSI-RS-ResourceSets. Each such resource set includes one or several configuredCSI-RS.2 The resource set can then be used as part of report configurationsdescribingmeasurements and corresponding reporting to be done by a device(seefurtherdetailsinSection8.2).Alternatively,anddespitethename,anNZP-CSI-RS-ResourceSetmayincludepointerstoasetofSSblocks(seeChapter16).This reflects the fact thatsomedevicemeasurements,especiallymeasurementsrelatedtobeammanagementandmobility,maybecarriedoutoneitherCSI-RSorSSblock.AboveitwasdescribedhowaCSIRScouldbeconfiguredforperiodic,semi-

persistent,oraperiodic transmission.Asmentioned there, thiscategorization isstrictlyspeakingnotapropertyoftheCSI-RSitselfbutapropertyofaresourceset. Furthermore, all CSI-RS within a semi-persistent resource set are jointlyactivated/deactivatedbymeansofaMACCEcommand.Likewise,transmissionofallCSI-RSwithinanaperiodic resourceset is jointly triggeredbymeansofDCI.Similarly, a device may be configured with CSI-IM resource sets, each

including a number of configured CSI-IM that can be jointlyactivated/deactivated (semi-persistent CSI-IM resource set) or triggered(aperiodicCSI-IMresourceset).

8.1.7TrackingReferenceSignal(TRS)Due to oscillator imperfections, the device must track and compensate forvariationsintimeandfrequencytosuccessfullyreceivedownlinktransmissions.To assist the device in this task, a tracking reference signal (TRS) can beconfigured.TheTRSisnotaCSI-RS.RatheraTRSisaresourcesetconsisting

ofmultipleperiodicNZP-CSI-RS.MorespecificallyaTRSconsistsoffourone-port,density-3CSI-RSlocatedwithintwoconsecutiveslots(seeFig.8.8).TheCRS-RS within the resource set, and thus also the TRS in itself, can beconfiguredwithaperiodicityof10,20,40,or80ms.Notethattheexactsetofresource elements (subcarriers andOFDMsymbols)used for theTRSCSI-RSmayvary.There is always a four-symbol time-domain separation between thetwoCSI-RSwithinaslotthough.Thistimedomainseparationsetsthelimitforthe frequency error that can be tracked. Likewise, the frequency-domainseparation (four subcarriers) sets the limit for the timing error that can betracked.

FIGURE8.8 TRSconsistingoffourone-port,density-3CSI-RSlocatedwithintwoconsecutiveslots.

ThereisalsoanalternativeTRSstructurewiththesameper-slotstructureastheTRSstructureofFig.8.8butonlyconsistingoftwoCSI-RSwithinasingleslot,comparedtotwoconsecutiveslotsfortheTRSstructureinFig.8.8ForLTE,theCRSservedthesamepurposeastheTRS.However,compared

totheLTECRS,theTRSimpliesmuchlessoverhead,onlyhavingoneantennaportandonlybeingpresentintwoslotseveryTRSperiod.

8.1.8MappingtoPhysicalAntennasInChapter7, theconceptofantennaportsandtherelation toreferencesignalswerediscussed.Amulti-portCSI-RScorrespondstoasetofantennaportsandthe CSI-RS can be used for sounding of the channels corresponding to thoseantennaports.However,aCSI-RSportisoftennotmappeddirectlytoaphysicalantenna,implyingthatthechannelbeingsoundedbasedonaCSI-RSisoftennotthe actual physical radio channel. Rather, more or less any kind of (linear)transformationorspatial filtering, labeledF inFig.8.9,maybeapplied to theCSI-RS beforemapping to the physical antennas. Furthermore, the number ofphysicalantennas(NinFig.8.9)towhichtheCSI-RSismappedmayverywellbe larger than the number of CSI-RS ports.3 When a device does channel

sounding based on the CSI-RS, neither the spatial filter F nor theN physicalantennas will be explicitly visible. What the device will see is just the M“channels”correspondingtotheMCSI-RSports.

FIGURE8.9 CSI-RSappliedtospatialfilter(F)beforemappingtophysicalantennas.

The spatial filter F may very well be different for different CSI-RS. Thenetworkcould,forexample,maptwodifferentconfiguredCSI-RSsuchthattheyarebeam-formed indifferentdirections (seeFig.8.10).To thedevice thiswillappearas twoCSI-RS transmittedover twodifferentchannels,despite the factthat they are transmitted from the same set of physical antennas and arepropagatingviathesamesetofphysicalchannels.

FIGURE8.10 DifferentspatialfiltersappliedtodifferentCSI-RS.

AlthoughthespatialfilterFisnotexplicitlyvisibletothedevice,thedevicestill has to make certain assumptions regarding F. Especially, F has a strongrelation to theconceptofantennaportsdiscussed inChapter7. Inessenceonecansaythat twosignalsare transmittedfromthesameantennaport if theyaremapped to the same set of physical antennas by means of the sametransformationF.As an example, in the case of downlink multiantenna transmission (see

Chapter 11), a device may measure on a CSI-RS and report a recommendedprecodermatrix to the network. The networkmay then use the recommendedprecodermatrixwhenmapping so-called transmission layers to antenna ports.When selecting a suitable precoder matrix the device will assume that thenetwork,ifusingtherecommendedmatrix,willmaptheoutputoftheprecodingto the antenna ports of the CSI-RS on which the corresponding devicemeasurementswerecarriedout.Inotherwords,thedevicewillassumethattheprecodedsignalwillbemappedtothephysicalantennasbymeansofthesamespatialfilterFasappliedtotheCSI-RS.

8.2DownlinkMeasurementsandReportingAnNRdevicecanbeconfigured tocarryoutdifferentmeasurements, inmostcases with corresponding reporting to the network. In general, such aconfigurationofameasurementandcorrespondingreportingaredonebymeansofareportconfiguration, in the3GPPspecifications[15]referred toasaCSI-ReportConfig.4Eachresourceconfigurationdescribes/indicates:

•Thespecificquantityorsetofquantitiestobereported;•Thedownlinkresource(s)onwhichmeasurementsshouldbecarriedoutinordertoderivethequantityorquantitiestobereported;

•Howtheactualreportingistobecarriedout,forexample,whenthereportingistobedoneandwhatuplinkphysicalchanneltouseforthereporting.

8.2.1ReportQuantityAreportconfigurationindicatesaquantityorsetofquantitiesthatthedeviceis

supposed to report. The report could, for example, include differentcombinations of channel-quality indicator (CQI), rank indicator (RI), andprecoder-matrixindicator(PMI),jointlyreferredtoaschannel-stateinformation(CSI).Alternatively,thereportconfigurationmayindicatereportingofreceivedsignal strength, more formally referred to as reference-signal received power(RSRP).RSRPhashistoricallybeenakeyquantitytomeasureandreportaspartof higher-layer radio-resource management (RRM) and is so also for NR.However,NRalsosupports layer-1reportingofRSRP,forexample,aspartofthe support for beammanagement (see Chapter 12).What is then reported ismore specifically referred to asL1-RSRP, reflecting the fact that the reportingdoesnotincludethemorelong-term(“layer-3”)filteringappliedforthehigher-layerRSRPreporting.

8.2.2MeasurementResourceIn addition to describing what quantity to report, a report configuration alsodescribes the set of downlink signals or, more generally, the set of downlinkresources onwhichmeasurements shouldbe carriedout in order to derive thequantity or quantities to be reported. This is done by associating the reportconfigurationwithoneorseveralresourcesetsasdescribedinSection8.1.6.A resource configuration is associated with at least one NZP-CSI-RS-

ResourceSet to be used formeasuring channel characteristics.As described inSection8.1.6,aNZP-CSI-RS-ResourceSetmayeithercontainasetofconfiguredCSI-RS or a set of SS blocks.Reporting of, for example, L1-RSRP for beammanagementcanthusbebasedonmeasurementsoneitherasetofSSblocksorasetofCSI-RS.Note that the resource configuration is associated with a resource set.

Measurementsandcorrespondingreportingare thus in thegeneralcasecarriedoutonasetofCSI-RSorasetofSSblocks.Insomecases,thesetwillonlyincludeasinglereferencesignal.Anexample

ofthisisconventionalfeedbackforlinkadaptationandmultiantennaprecoding.In this case, the device would typically be configured with a resource setconsisting of a single multi-port CSI-RS on which the device will carry outmeasurementstodetermineandreportacombinationofCQI,RI,andPMI.On the other hand, in the case of beam management the resource set will

typicallyconsistofmultipleCSI-RS,alternativelymultipleSSblocks,whereinpracticeeachCSI-RSorSSblockisassociatedwithaspecificbeam.Thedevice

measuresonthesetofsignalswithintheresourcesetandreportstheresulttothenetworkasinputtothebeam-managementfunctionality.There are also situations when a device needs to carry out measurements

without any corresponding reporting to the network.One such case iswhen adeviceshouldcarryoutmeasurementsforreceiver-sidedownlinkbeamforming.Aswill be described inChapter 12, in such a case a devicemaymeasure ondownlink reference signalsusingdifferent receiverbeams.However, the resultofthemeasurementisnotreportedtothenetworkbutonlyusedinternallywithinthedevicetoselectasuitablereceiverbeam.Atthesametimethedeviceneedstobeconfiguredwiththereferencesignalstomeasureon.Suchaconfigurationisalsocoveredbyreportconfigurationsforwhich,inthiscase,thequantitytobereportedisdefinedas“None.”

8.2.3ReportTypesInadditiontothequantitytoreportandthesetofresourcestomeasureon,thereportconfigurationalsodescribeswhenandhowthereportingshouldbecarriedout.Similar to CSI-RS transmission, device reporting can be periodic, semi-

persistent,oraperiodic.As the name suggests, periodic reporting is done with a certain configured

periodicity.PeriodicreportingisalwaysdoneonthePUCCHphysicalchannel.Thus,inthecaseofperiodicreporting,theresourceconfigurationalsoincludesinformation about aperiodically availablePUCCH resource tobeused for thereporting.In the case of semi-persistent reporting, a device is configured with

periodically occurring reporting instances in the same way as for periodicreporting.However,actualreportingcanbeactivatedanddeactivatedbymeansofMACsignaling(MACCE).Similar to periodic reporting, semi-persistent reporting can be done on a

periodicallyassignedPUCCHresource.Alternatively,semi-persistent reportingcanbedoneonasemi-persistentlyallocatedPUSCH.Thelatteristypicallyusedforlargerreportingpayloads.Aperiodic reporting is explicitly triggeredbymeansofDCI signaling,more

specificallywithinaCSI-request fieldwithin theuplinkschedulinggrant (DCIformat 0-1). The DCI fieldmay consist of up to 6 bits with each configuredaperiodic report associated with a specific bit combination. Thus, up to 63

differentaperiodicreportscanbetriggered.5Aperiodic reporting is always done on the scheduled PUSCH and thus

requires an uplink scheduling grant. This is the reason why the triggering ofaperiodic reporting is only included in the uplink scheduling grant and not inotherDCIformats.It should be noted that, in the case of aperiodic reporting, the report

configuration could actually include multiple resource sets for channelmeasurements,eachwithitsownsetofreferencesignals(CSI-RSorSSblock).EachresourcesetisassociatedwithaspecificvalueoftheCSI-requestfieldintheDCI.BymeansoftheCSIrequestthenetworkcan,inthisway,triggerthesametypeofreportingbutbasedondifferentmeasurementresources.Notethatthe same could, in principle, have been done by configuring the device withmultiple report configurations, where the different resource configurationswould specify the same reporting configuration and report type but differentmeasurementresources.Periodic,semi-persistent,andaperiodicreportingshouldnotbemixedupwith

periodic,semi-persistent,andaperiodicCSI-RSasdescribedinSection8.1.3.Asanexample,aperiodicreportingandsemi-persistentreportingcouldverywellbebased on measurements on periodic CSI-RS. On the other hand, periodicreporting can only be based onmeasurements on periodic CSI-RS but not onaperiodic and semi-persistent CSI-RS. Table 8.1 summarizes the allowedcombinations of reporting type (periodic, semi-persistent, and aperiodic) andresourcetype(periodic,semi-persistent,andaperiodic).

Table8.1

8.3UplinkChannelSounding—SRSToenableuplinkchannelsoundingadevicecanbeconfiguredfortransmissionofsoundingreferencesignals (SRS). InmanyrespectsSRScanbeseenas theuplinkequivalence to thedownlinkCSI-RS in thesense thatbothCSI-RSandSRS are intended for channel sounding, albeit in different transmission

directions.BothCSI-RSandSRScanalsoserveasQCLreferencesinthesensethatotherphysicalchannelscanbeconfiguredtobetransmittedquasi-colocatedwith CSI-RS and SRS, respectively. Thus, given knowledge of a suitablereceiver beam for theCSI-RS/SRS, the receiver knows that the same receiverbeamshouldbesuitablealsoforthephysicalchannelinquestion.However,onamoredetailedlevel,thestructureofSRSisquitedifferentfrom

CSI-RS.

•SRSislimitedtoamaximumoffourantennaports,whileCSI-RSsupportsupto32antennaports.

•Beinganuplinksignal,SRSisdesignedtohavelowcubic-metric[60]enablinghighdevicepower-amplifierefficiency.

Thebasic time/frequencystructureofanSRSisexemplified inFig.8.11. Inthe general case, anSRS spans one, two, or four consecutiveOFDMsymbolsandislocatedsomewherewithinthelastsixsymbolsofaslot.Inthefrequencydomain, an SRS has a so-called “comb” structure, implying that an SRS istransmitted on everyNth subcarrier whereN can take the values two or four(“comb-2”and“comb-4,”respectively).

FIGURE8.11 ExamplesofSRStime/frequencystructures.

SRS transmissions from different devices can be frequency multiplexedwithin the same frequency range by being assigned different combscorresponding todifferent frequencyoffsets.Forcomb-2, that is,whenSRS istransmittedoneverysecondsubcarrier,twoSRScanbefrequencymultiplexed.Inthecaseofcomb-4,uptofourSRScanbefrequencymultiplexed.Fig.8.12illustrates an example of SRSmultiplexing assuming a comb-2 SRS spanningtwoOFDMsymbols.

(8.1)

FIGURE8.12 Comb-basedfrequencymultiplexingofSRSfromtwodifferentdevicesassumingcomb-2.

8.3.1SRSSequencesandZadoff–ChuSequencesThesequencesapplied to thesetofSRSresourceelementsarepartlybasedonso-calledZadoff–Chu sequences [25].Due to their specificproperties,Zadoff–ChusequencesareusedatseveralplaceswithintheNRspecifications,especiallyintheuplinktransmissiondirection.Zadoff–ChusequencesarealsoextensivelyusedinLTE[28].AZadoff–ChusequenceoflengthMisgivenbythefollowingexpression:

Ascanbe seen fromEq. (8.1), aZadoff–Chusequencehasa characterizingparameter u, referred to as the root index of the Zadoff–Chu sequence. For agivensequencelengthM,thenumberofrootindicesgeneratinguniqueZadoff–Chusequencesequals thenumberof integers thatare relativeprime toM.Forthisreason,Zadoff–Chusequencesofprimelengthareofspecialinterestastheymaximize the number of available Zadoff–Chu sequences. More specifically,assuming the sequence lengthM being a prime number there areM–1 uniqueZadoff–Chusequences.

AkeypropertyofZadoff–ChusequencesisthatthediscreteFouriertransformofaZadoff–ChusequenceisalsoaZadoff–Chusequence.6FromEq.(8.1)itisobvious that a Zadoff–Chu sequence has constant time-domain amplitude,makingitgoodfromapower-amplifier-efficiencypointofview.AstheFouriertransformofaZadoff–ChusequenceisalsoaZadoff–Chusequence,therewouldthen also be constant power in the frequency domain, that is, in addition toconstant time-domain amplitude,Zadoff–Chu sequences alsohave flat spectra.Asaflatspectrumisequivalenttozerocyclicautocorrelationforanynon-zerocyclicshift,thisimpliesthattwodifferenttime-domaincyclicshiftsofthesameZadoff–Chusequenceareorthogonaltoeachother.Notethatacyclicshiftinthetime domain corresponds to applying a continuous phase rotation in thefrequencydomain.Although Zadoff–Chu sequences of prime length are preferred in order to

maximize thenumberof available sequences,SRSsequences arenotofprimelength.TheSRSsequencesarethereforeextendedZadoff–Chusequencesbasedon the longest prime-length Zadoff–Chu sequencewith a lengthM smaller orequal to the desired SRS-sequence length. The sequence is then cyclicallyextendedinthefrequencydomainuptothedesiredSRSsequencelength.Astheextension is done in the frequency domain, the extended sequence still has aconstant spectrum, and thus “perfect” cyclic autocorrelation, but the time-domainamplitudewillvarysomewhat.ExtendedZadoff–ChusequenceswillbeusedasSRSsequencesforsequence

lengths of 36 or larger, corresponding to an SRS extending over 6 and 12resource blocks in the cases of comb-2 and comb-4, respectively. For shortersequence lengths, special flat-spectrum sequences with good time-domainenvelopepropertieshavebeenfoundfromcomputersearch.Thereasonisthat,for shorter sequences, there would not be a sufficient number of Zadoff–Chusequencesavailable.The same principle will be used also for other cases where Zadoff–Chu

sequencesareusedwithintheNRspecifications,forexample,foruplinkDMRS(seeSection9.11.1).

8.3.2MultiportSRSInthecaseofanSRSsupportingmorethanoneantennaport,thedifferentportsshare the same set of resource elements and the same basic SRS sequence.Different phase rotations are then applied to separate the different ports as

illustratedinFig.8.13.

FIGURE8.13 SeparationofdifferentSRSantennaportsbyapplyingdifferentphaseshiftstothebasicfrequency-domainSRSsequence

.Thefigureassumesacomb-4SRS.

As described above, applying a phase rotation in the frequency domain isequivalenttoapplyingacyclicshiftinthetimedomain.IntheNRspecificationthe above operation is actually referred to as “cyclic shift,” although it ismathematicallydescribedasafrequency-domainphaseshift.

8.3.3Time-DomainStructureofSRSSimilar toCSI-RS, anSRS canbe configured forperiodic, semi-persistent, oraperiodictransmission:

•AperiodicSRSistransmittedwithacertainconfiguredperiodicityandacertainconfiguredslotoffsetwithinthatperiodicity;

•Asemi-persistentSRShasaconfiguredperiodicityandslotoffsetinthesamewayasaperiodicSRS.However,actualSRStransmissionaccordingtotheconfiguredperiodicityandslotoffsetisactivatedanddeactivatedbymeansofMACCEsignaling;

•AnaperiodicSRSisonlytransmittedwhenexplicitlytriggeredbymeansofDCI.

Itshouldbepointedout that,similar toCSI-RSI,activation/deactivationandtriggering for semi-persistent and aperiodic SRS, respectively, is actually notdoneforaspecificSRSbutratherdoneforaso-calledSRSresourcesetwhich,inthegeneralcase,includedmultipleSRS(seebelow).

8.3.4SRSResourceSets

SimilartoCSI-RS,adevicecanbeconfiguredwithoneorseveralSRSresourcesets, where each resource set includes one or several configured SRS. Asdescribed above, a SRS can be configured for periodic, semi-persistent, oraperiodictransmission.AllSRSincludedwithinaconfiguredSRSresourcesethave to be of the same type. In other words, periodic, semi-persistent, oraperiodictransmissioncanalsobeseenasapropertyofanSRSresourceset.AdevicecanbeconfiguredwithmultipleSRSresourcesetsthatcanbeused

for different purposes, including both downlink and uplink multiantennaprecodinganddownlinkanduplinkbeammanagement.ThetransmissionofaperiodicSRS,ormoreaccurately,transmissionoftheset

of configured SRS included in an aperiodic SRS resource set, is triggered byDCI. More specifically, DCI format 0-1 (uplink scheduling grant) and DCIformat 1-1 (downlink scheduling assignment) include a 2-bitSRS-request thatcantriggerthetransmissionofoneoutofthreedifferentaperiodicSRSresourcesets configured for the device (the fourth bit combination corresponds to “notriggering”).

8.3.5MappingtoPhysicalAntennasSimilar to CSI-RS, SRS ports are often not mapped directly to the devicephysical antennas but via some spatial filter F that mapsM SRS ports toNphysicalchannels(seeFig.8.14).

FIGURE8.14 SRSappliedtospatialfilter(F)beforemappingtophysicalantennas.

In order to provide connectivity regardless of the rotational direction of thedevice,NR devices supporting high-frequency operationwill typically include

multipleantennapanelspointingindifferentdirections.ThemappingofSRStoonesuchpanel isanexampleofatransformationFfromSRSantennaports tothe set of physical antennas. Transmission from different panels will thencorrespondtodifferentspatialfiltersFasillustratedinFig.8.15.

FIGURE8.15 DifferentspatialfiltersappliedtodifferentSRS.

Similar to the downlink, the spatial filteringFhas a real impact despite thefactthatitisneverexplicitlyvisibletothenetworkreceiverbutjustseenasanintegratedpart of theoverall channel.As an example, thenetworkmay soundthe channel basedon a device-transmittedSRSand thendecide on a precodermatrix that the device should use for uplink transmission. The device is thenassumed to use that precoder matrix in combination with the spatial filter FappliedtotheSRS.Inothercases,adevicemaybeexplicitlyscheduledfordatatransmissionusing theantennaportsdefinedbyacertainSRS. Inpractice thisimpliesthatthedeviceisassumedtotransmitusingthesamespatialFthathasbeenusedfortheSRStransmission.Inpractice,thismayimplythatthedeviceshould transmit using the samebeamor panel that has beenused for theSRStransmission.

1The“density-3”CSI-RSusedforTRS(seeSection8.1.7)isanexceptiontothisrule.

2Strictlyspeaking,theresourcesetincludesreferencestoconfiguredCSI-RS.3HavingNsmallerthanMdoesnotmakesense.4Notethatweareheretalkingaboutphysical-layermeasurementsandreporting,tobedistinguishedfromhigher-layerreportingdonebymeansofRRCsignaling.5Theall-zerovalueindicates“notriggering.”6Theinverseobviouslyholdsaswell,thatis,theinverseDFTofaZadoff–ChusequenceisalsoaZadoff–Chusequence.

CHAPTER9

Transport-ChannelProcessing

Abstract

This chapter provides a detailed description of the downlink and uplinkphysical-layer transport-channel processing including coding,modulation,multi-antenna precoding, resource-block mapping, and reference signalstructure.

KeywordsPDSCH;PUSCH;LDPC;ratematching;code-blocksegmentation;coding;hybridARQ;DFTprecoding;DM-RS;demodulationreferencesignal;multi-antennaprecoding;VRB-to-PRBmapping;reservedresources;PT-RS

Thischapterwillprovideamoredetaileddescriptionofthedownlinkanduplinkphysical-layer functionality such as coding, modulation, multi-antennaprecoding,resource-blockmapping,andreferencesignalstructure.

9.1OverviewThephysicallayerprovidesservicestotheMAClayerintheformoftransportchannelsasdescribedinSection6.4.5.Inthedownlink,therearethreedifferenttypesoftransportchannelsdefinedforNR:theDownlinkSharedChannel(DL-SCH),thePagingChannel(PCH),andtheBroadcastChannel(BCH),althoughthelattertwoarenotusedinthenon-standaloneoperation.Intheuplink,thereisonly one uplink transport-channel type carrying transport blocks in NR,1 theUplinkSharedChannel(UL-SCH).TheoveralltransportchannelprocessingforNR follows a similar structure as for LTE (see Fig. 9.1). The processing ismostlysimilarinuplinkanddownlinkandthestructureinFig.9.1isapplicable

for the DL-SCH, BCH, and PCH in the downlink, and the UL-SCH in theuplink. The part of the BCH that ismapped to the PBCH follows a differentstructure,describedinSection16.1,asdoestheRACH.

FIGURE9.1 Generaltransport-channelprocessing.

Within each transmission time interval (TTI), up to two transport blocks ofdynamicsizearedelivered to thephysical layerand transmittedover the radiointerfaceforeachcomponentcarrier.Twotransportblocksareonlyusedinthecaseofspatialmultiplexingwithmorethanfourlayers,whichisonlysupportedinthedownlinkdirectionandmainlyusefulinscenarioswithveryhighsignal-to-noise ratios.Hence, atmost a single transport block per component carrierandTTIisatypicalcaseinpractice.ACRCforerror-detectingpurposesisaddedtoeachtransportblock,followed

by error-correcting coding using LDPC codes. Rate matching, includingphysical-layerhybrid-ARQfunctionality,adaptsthenumberofcodedbitstothescheduled resources.The codebits are scrambled and fed to amodulator, andfinallythemodulationsymbolsaremappedtothephysicalresources,includingthespatialdomain.FortheuplinkthereisalsoapossibilityofaDFT-precoding.The differences between uplink and downlink is, apart from DFT-precodingbeing possible in the uplink only, mainly around antenna mapping andassociatedreferencesignals.Inthefollowing,eachoftheprocessingstepswillbediscussedinmoredetail.

For carrier aggregation, the processing steps are duplicated for each of thecarriers and the description herein is applicable to each of the carriers. Sincemost of the processing steps are identical for uplink and downlink, theprocessing will be described jointly and any differences between uplink anddownlinkexplicitlymentionedwhenrelevant.

9.2ChannelCodingAnoverviewofthechannelcodingstepsisprovidedinFig.9.2anddescribedinmoredetail in the following sections.First, aCRC is attached to the transportblock to facilitate error detection, followed by code block segmentation. Eachcode block isLDPC-encoded and ratematched separately, including physical-layerhybrid-ARQprocessing,andtheresultingbitsareconcatenatedtoformthesequenceofbitsrepresentingthecodedtransportblock.

FIGURE9.2 Channelcoding.

9.2.1CRCAttachmentPerTransportBlockIn the first step of the physical-layer processing, a CRC is calculated for andappendedtoeachtransportblock.TheCRCallowsforreceiver-sidedetectionoferrors in the decoded transport block and can, for example, be used by thehybrid-ARQprotocolasatriggerforrequestingretransmissions.ThesizeoftheCRCdependsonthetransport-blocksize.Fortransportblocks

larger than3824bits,a24-bitCRCisused,otherwisea16-bitCRCisused toreduceoverhead.

9.2.2Code-BlockSegmentationTheLDPCcoderinNRisdefineduptoacertaincode-blocksize(8424bitsforbasegraph1 and3840bits for base graph2).Tohandle transport block sizeslarger than this, code-block segmentation is used where the transport block,includingtheCRC,issplit intomultipleequal-sized2codeblocksasillustratedinFig9.3.

FIGURE9.3 Codeblocksegmentation.

As can be seen in Fig. 9.3, code-block segmentation also implies that anadditionalCRC(alsoof length24bitsbutdifferentcompared to the transport-blockCRCdescribedabove)iscalculatedforandappendedtoeachcodeblock.Inthecaseofasinglecode-blocktransmissionnoadditionalcode-blockCRCis

applied.Onecouldargue that, in thecaseofcode-blocksegmentation, the transport-

blockCRC is redundant and impliesunnecessaryoverhead as the set of code-blockCRCsshould indirectlyprovide informationabout thecorrectnessof thecomplete transport block. However, to handle code-block group (CBG)retransmissions as discussed in Chapter 13, a mechanism to detect errors percode block is necessary. CBG retransmission means that only the erroneouscode-blockgroups are retransmitted insteadof the complete transport block toimprovethespectralefficiency.Theper-CBCRCcanalsobeusedforthedevicetolimitdecodingincaseofaretransmissiononlytothoseCBswhoseCRCsdidnotcheckevenifper-CBGretransmissionisnotconfigured.Thishelpsreducingthedeviceprocessingload.Thetransport-blockCRCalsoaddsanextralevelofprotectionintermsoferrordetection.Notethatcode-blocksegmentationisonlyappliedtolargetransportblocksforwhichtherelativeextraoverheadduetotheadditionaltransport-blockCRCissmall.

9.2.3ChannelCodingChannel coding is based onLDPC codes, a code designwhichwas originallyproposed in the 1960s [34] but forgotten for many years. They were“rediscovered” in the1990s [59] and found tobe anattractive choice fromanimplementationperspective.Froman error-correcting capabilitypoint ofview,turbocodes,asusedinLTE,canachievesimilarperformance,butLDPCcodescanoffer lowercomplexity,especiallyathighercoderates,andwere thereforechosenforNR.The basis for LDPC codes is a sparse (low-density) parity checkmatrixH

where for eachvalidcodewordc the relationHcT=0holds.DesigningagoodLDPCcodetoalargeextentboilsdowntofindingagoodparitycheckmatrixHwhich is sparse (the sparseness implies relatively simple decoding). It iscommon to represent theparity-checkmatrixbyagraphconnectingnvariablenodes at the top with (n–k) constraint nodes at the bottom of the graph, anotation that allows awide range of properties of an (n, k) LDPC code to beanalyzed. This explains why the term base graph is used in the NRspecifications. A detailed description of the theory behind LDPC codes isbeyond the scope of this book, but there is a rich literature in the field (forexample,see[68]).Quasi-cyclicLDPCcodeswithadual-diagonalstructureofthekernelpartof

the parity check matrix are used in NR, which gives a decoding complexitywhich is linear in the number of coded bits and enables a simple encodingoperation. Two base graphs are defined, BG1 and BG2, representing the twobasematrices.The reason for two base graphs instead of one is to handle thewiderangeofpayloadsizesandcoderatesinanefficientway.Supportingaverylargepayloadsizeatamediumtohighcoderate,whichisthecaseforveryhighdatarates,usingacodedesignedtosupportaverylowcoderateisnotefficient.At the same time, the lowest code rates are necessary to provide goodperformance in challenging situations. InNR, BG1 is designed for code ratesfrom 1/3 to 22/24 (approximately 0.33–0.92) and BG 2 from 1/5 to 5/6(approximately 0.2–0.83). Through puncturing, the highest code rate can beincreased somewhat, up to 0.95, beyond which the device is not required todecode.ThechoicebetweenBG1andBG2isbasedonthetransportblocksizeandcoderatetargetedforthefirsttransmission(seeFig.9.4).

FIGURE9.4 SelectionofbasegraphfortheLDPCcode.

The base graphs, and the corresponding base matrices, define the generalstructureof theLDPCcode.To support a rangeof payload sizes, 51differentlifting sizes and sets of shift coefficients are defined and applied to the basematrices. In short, for a given lifting size Z, each “1” in the base matrix isreplacedbytheZ×Zidentifymatrixcircularlyshiftedbythecorrespondingshiftcoefficient and each “0” in the base matrix is replaced by the Z×Z all-zeromatrix. Hence, a relatively large number of parity-check matrices can begenerated to support multiple payload sizes while maintaining the general

structure of the LDPC code. To support payload sizes that are not a nativepayloadsizeofoneofthe51definedparitycheckmatrices,knownfillerbitscanbeappendedtothecodeblockbeforeencoding.SincetheNRLDPCcodesaresystematiccodes,thefillerbitscanberemovedbeforetransmission.

9.3RateMatchingandPhysical-LayerHybrid-ARQFunctionalityThe rate-matching and physical-layer hybrid-ARQ functionality serves twopurposes, namely to extract a suitable number of coded bits to match theresources assigned for transmission and to generate different redundancyversionsneededforthehybrid-ARQprotocol.ThenumberofbitstotransmitonthePDSCHorPUSCHdependsonawiderangeoffactors,notonlythenumberofresourceblocksandthenumberofOFDMsymbolsscheduled,butalsoontheamount of overlapping resource elements used for other purposes and such asreference signals, control channels, or system information. There is also apossibility to, in the downlink, define reserved resources as a tool to providefuture compatibility (see Section 9.10), which affects the number of resourceelementsusableforthePDSCH.Rate matching is performed separately for each code block. First, a fixed

number of the systematic bits are punctured. The fraction of systematic bitspuncturedcanberelativelyhigh,upto1/3ofthesystematicbits,dependingonthecode-blocksize.Theremainingcodedbitsarewrittenintoacircularbuffer,startingwiththenon-puncturedsystematicbitsandcontinuingwithparitybitsasillustratedinFig.9.5.Theselectionofthebitstotransmitisbasedonreadingtherequirednumberof bits from the circular bufferwhere the exact set of bits totransmit depends on the redundancy version (RV) corresponding to differentstartingpositionsinthecircularbuffer.Hence,byselectingdifferentredundancyversions, different sets of coded bits representing the same set of informationbits can be generated, which is used when implementing hybrid-ARQ withincremental redundancy. The starting points in the circular buffer are definedsuchthatbothRV0andRV3areself-decodable,thatis,includesthesystematicbitsunder typical scenarios.This is also the reasonRV3 is locatedafter “nineo’clock”inFig.9.5asthisallowsmoreofthesystematicbitstobeincludedinthetransmission.

FIGURE9.5 Exampleofcircularbufferforincrementalredundancy.

In the receiver, soft combining is an important part of the hybrid-ARQfunctionality as described in Section 13.1. The soft values representing thereceived coded bits are buffered and, if a retransmission occurs, decoding isperformedusingthebufferedbitscombinedwiththeretransmittedcodedbits.Inaddition to a gain in accumulated receivedEb/N0,with different coded bits indifferent transmission attempts, additional parity bits are obtained and theresulting code rate after soft combining is lowerwith a corresponding codinggainobtained.Soft combining requires a buffer in the receiver. Typically, a fairly high

probabilityofsuccessfultransmissiononthefirstattemptistargetedandhencethe soft buffer remains unusedmost of the time. Since the soft buffer size isfairlylargeforthelargesttransportblocksizes,requiringthereceivertobufferallsoftbitsevenforthelargesttransportblocksizesissuboptimalfromacost–performance tradeoff perspective. Hence, limited-buffer rate-matching issupportedasillustratedinFig.9.6.Inprinciple,onlybitsthedevicecanbufferarekeptinthecircularbuffer,thatis,thesizeofthecircularbufferisdeterminedbasedonthereceiver’ssoftbufferingcapability.

FIGURE9.6 Limited-bufferratematching.

For the downlink, the device is not required to buffer more soft bits thancorrespondingtothelargesttransportblocksizecodedatrate2/3.Notethatthisonlylimitsthesoftbuffercapacityforthehighesttransportblocksizes,thatis,thehighestdatarates.Forsmallertransportblocksizes,thedeviceiscapableofbufferingallsoftbitsdowntothemothercoderate.For the uplink, full-buffer rate matching, where all soft bits are buffered

irrespective of the transport block size, is supported given sufficient gNBmemory. Limited-buffer rate matching using the same principles as for thedownlinkcanbeconfiguredusingRRCsignaling.Thefinalstepoftherate-matchingfunctionalityistointerleavethebitsusing

ablock interleaverand tocollect thebits fromeachcodeblock.Thebits fromthecircularbufferarewrittenrow-by-rowintoablockinterleaverandreadoutcolumn-by-column. The number of rows in the interleaver is given by themodulation order and hence the bits in one column correspond to onemodulation symbol3 (see Fig. 9.7). This results in the systematic bits spreadacross the modulation symbols, which improves performance. Bit collectionconcatenatesthebitsforeachcodeblock.

FIGURE9.7 Bitinterleaver(16QAMassumedinthisexample).

9.4Scrambling

Scrambling is applied to theblockof codedbitsdeliveredby thehybrid-ARQfunctionality by multiplying the sequence of coded bits with a bit-levelscrambling sequence.Without scrambling, the channel decoder at the receivercould,atleastinprinciple,beequallymatchedtoaninterferingsignalastothetarget signal, thus being unable to properly suppress the interference. Byapplyingdifferentscramblingsequencesforneighboringcellsinthedownlinkorfordifferentdevicesintheuplink,theinterferingsignal(s)afterdescramblingis(are)randomized,ensuringfullutilizationoftheprocessinggainprovidedbythechannelcode.The scrambling sequence in both downlink (PDSCH) and uplink (PUSCH)

dependsontheidentityofthedevice,thatis,theC-RNTI,andadatascramblingidentityconfiguredineachdevice.Ifnodatascramblingidentityisconfigured,the physical layer cell identity is used as a default value to ensure thatneighboring devices, both in the same cell and between cells, use differentscrambling sequences. Furthermore, in the case of two transport blocks beingtransmitted in the downlink to support more than four layers, differentscramblingsequencesareusedforthetwotransportblocks.

9.5ModulationThemodulationsteptransformstheblockofscrambledbits toacorrespondingblock of complex modulation symbols. The modulation schemes supportedincludeQPSK,16QAM,64QAM,and256QAMinbothuplinkanddownlink.Inaddition,fortheuplinkπ/2-BPSKissupportedinthecasetheDFT-precodingisused, motivated by a reduced cubic metric [60] and hence improved power-amplifierefficiency,inparticularforcoveragelimitedscenarios.Notethatπ/2-BPSK is neither supported nor useful in the absence ofDFT-precoding as thecubicmetricinthiscaseisdominatedbytheOFDMwaveform.

9.6LayerMappingThepurposeof the layer-mappingstep is todistribute themodulationsymbolsacross the different transmission layers. This is done in a similar way as forLTE;everynth symbol ismapped to thenth layer.Onecoded transportblockcanbemappedonuptofourlayers.Inthecaseoffivetoeightlayers,supportedinthedownlinkonly,asecondtransportblockismappedtolayersfivetoeightfollowingthesameprincipleasforthefirsttransportblock.

Multi-layer transmission is only supported in combinationwithOFDM, thebaseline waveform in NR. With DFT-precoding in the uplink, only a singletransmission layer is supported. This is motivated both by the receivercomplexity,whichinthecaseofmulti-layertransmissionwouldbesignificantlyhigherwithaDFT-precoderthanwithout,andtheusecaseoriginallymotivatingtheadditional supportofDFT-precoding,namelyhandlingof coverage-limitedscenarios. In such a scenario, the received signal-to-noise ratio is too low forefficient usage of spatial multiplexing and there is no need to support spatialmultiplexingtoasingledevice.

9.7UplinkDFTPrecodingDFTprecodingcanbeconfiguredintheuplinkonly.Inthedownlink,aswellasthecaseofOFDMintheuplink,thestepistransparent.InthecasethatDFT-precodingisappliedintheuplink,blocksofMsymbols

arefedthroughasize-MDFTasillustratedinFig.9.8,whereMcorrespondstothenumberofsubcarriersassignedforthetransmission.ThereasonfortheDFTprecoding is to reduce the cubic metric for the transmitted signal, therebyenablinghigherpower-amplifierefficiency.Fromanimplementationcomplexitypoint of view theDFT size should preferably be constrained to a power of 2.However,suchaconstraintwouldlimittheschedulerflexibilityintermsoftheamount of resources that can be assigned for an uplink transmission. Rather,from a flexibility point of view all possible DFT sizes should preferably beallowed.ForNR,thesamemiddle-wayasforLTEhasbeenadoptedwheretheDFTsize,andthusalsothesizeoftheresourceallocation,islimitedtoproductsoftheintegers2,3,and5.Thus,forexample,DFTsizesof60,72,and96areallowed but a DFT size of 84 is not allowed.4 In this way, the DFT can beimplemented as a combination of relatively low-complex radix-2, radix-3, andradix-5FFTprocessing.

FIGURE9.8 DFT-precoding.

9.8Multi-AntennaPrecodingThe purpose of multi-antenna precoding is to map the different transmissionlayers to a set of antennaports using aprecodermatrix. InNR, theprecodingand multi-antenna operation differs between downlink and uplink and thecodebook-basedprecodingstep is,exceptforCSIreporting,onlyvisible in theuplinkdirection.Foradetaileddiscussiononhowtheprecodingstepisusedtorealizebeamforming anddifferentmulti-antenna schemes seeChapters 11 and12.

9.8.1DownlinkPrecodingIn the downlink, the demodulation reference signal (DMRS) used for channelestimationissubjecttothesameprecodingasthePDSCH(seeFig.9.9).Thus,theprecoding isnot explicitlyvisible to the receiverbut is seenaspartof theoverall channel. This is similar to the receiver-transparent spatial filteringdiscussedinthecontextofCSI-RSandSRSinChapter8.Inessence,intermsofactualdownlink transmission,anymulti-antennaprecodingcanbeseenaspartofsuch,tothedevice,transparentspatialfiltering.

FIGURE9.9 Downlinkprecoding.

However, for the purpose of CSI reporting, the device may assume that aspecificprecodingmatrixWisappliedat thenetworkside.Thedevice is thenassumingthat theprecodermaps thesignal to theantennaportsof theCSI-RSusedforthemeasurementsonwhichthereportingwasdone.Thenetworkisstillfreetousewhateverprecoderitfindsadvantageousfordatatransmission.To handle receiver-side beamforming, or in general multiple reception

antennaswithdifferentspatialcharacteristics,QCLrelationsbetweenaDM-RSportgroup,which is theantennaportsused forPDSCHtransmission,5 and theantennaportsusedforCSI-RSorSSblocktransmissioncanbeconfigured.TheTransmission Configuration Index (TCI) provided as part of the schedulingassignment indicates the QCL relations to use, or in other words, whichreceptionbeamtouse.ThisisdescribedinmoredetailinChapter12.Demodulationreferencesignalsare,asdiscussedinSection9.11,transmitted

in thescheduledresourceblocksand it is fromthose referencesignals that thedevicecanestimatethechannel,includinganyprecodingWandspatialfilteringF applied for PDSCH. In principle, knowledge about the correlation betweenreference signal transmissions, both in terms of correlation introduced by theradiochannelitselfandcorrelationintheuseofprecoder,isusefultoknowandcanbeexploitedbythedevicetoimprovethechannelestimationaccuracy.Inthetimedomain,thedeviceisnotallowedtomakeanyassumptionsonthe

referencesignalsbeingcorrelatedbetweenPDSCHschedulingoccasions.Thisisnecessary to allow full flexibility in terms of beamforming and spatialprocessingaspartoftheschedulingprocess.In the frequency domain, the device can be given some guidance on the

correlation. This is expressed in the form of physical resource-block groups(PRGs). Over the frequency span of one PRG, the device may assume thedownlink precoder remains the same and may exploit this in the channel-estimation process, while the device may not make any assumptions in thisrespect between PRGs. From this it can be concluded that there is a tradeoffbetween the precoding flexibility and the channel-estimation performance—alarge PRG size can improve the channel-estimation accuracy at the cost ofprecodingflexibilityandviceversa.Hence,thegNBmayindicatethePRGsizeto the device where the possible PRG sizes are two resource blocks, fourresourceblocks,orthescheduledbandwidthasshowninthebottomofFig.9.10.A single value may be configured, in which case this value is used for thePDSCH transmissions. It is also possible to dynamically, through the DCI,indicatethePRGsizeused.Inaddition,thedevicecanbeconfiguredtoassumethatthePRGsizeequalsthescheduledbandwidthinthecasethatthescheduledbandwidthislargerthanhalfthebandwidthpart.

FIGURE9.10 Physicalresource-blockgroups(top)andindicationthereof(bottom).

9.8.2UplinkPrecodingSimilartothedownlink,uplinkdemodulationreferencesignalsusedforchannelestimationare subject to thesameprecodingas theuplinkPUSCH.Thus,alsofor the uplink the precoding is not directly visible froma receiver perspectivebutisseenaspartoftheoverallchannel(seeFig.9.11).

FIGURE9.11 Uplinkprecoding.

However, from a scheduling point of view, the multi-antenna precoding ofFig. 9.1 is visible in theuplink as thenetworkmayprovide thedevicewith aspecificprecodermatrixWthereceivershoulduseforthePUSCHtransmission.This is done through theprecoding information andantennaport fields in theDCI.The precoder is then assumed tomap the different layers to the antennaportsofaconfiguredSRSindicatedbythenetwork.Inpracticethiswillbethesame SRS as the network used for the measurement on which the precoderselection was made. This is known as codebook-based precoding since theprecoder W to use is selected from a codebook of possible matrices andexplicitlysignaled.NotethatthespatialfilterFselectedbythedevicealsocanbe seen as a precoding operation, although not explicitly controlled by thenetwork. The network can however restrict the freedom in the choice of FthroughtheSRSresourceindicator(SRI)providedaspartoftheDCI.Thereisalsoapossibilityforthenetworktooperatewithnon-codebook-based

precoding.InthiscaseWisequaltotheidentitymatrixandprecodingishandledsolelybythespatialfilterFbasedonrecommendationsfromthedevice.

Both codebook-based and non-codebook-based precoding are described indetailinChapter11.

9.9ResourceMappingTheresource-blockmappingtakesthemodulationsymbolstobetransmittedoneachantennaportandmapsthemtothesetofavailableresourceelementsinthesetofresourceblocksassignedbytheMACschedulerforthetransmission.Asdescribed inSection7.3, a resourceblock is12 subcarrierswideand typicallymultiple OFDM symbols, and resource blocks, are used for the transmission.Thesetoftime–frequencyresourcesusedfortransmissionisdeterminedbythescheduler.However,someorallof theresourceelementswithin thescheduledresourceblocksmaynotbe available for the transport-channel transmissionastheyareusedfor:

•Demodulationreferencesignals(potentiallyincludingreferencesignalsforothercoscheduleddevicesinthecaseofmulti-userMIMO)asdescribedinSection9.11;

•OthertypesofreferencesignalssuchasCSI-RSandSRS(seeChapter8);

•DownlinkL1/L2controlsignaling(seeChapter10);•SynchronizationsignalsandsysteminformationasdescribedinChapter16;

•DownlinkreservedresourcesasameanstoprovideforwardcompatibilityasdescribedinSection9.10.

Thetime–frequencyresourcestobeusedfortransmissionaresignaledbythescheduler as a set ofvirtual resource blocks and a set ofOFDMsymbols.Tothese scheduled resources, the modulation symbols are mapped to resourceelements in a frequency-first, time-second manner. The frequency-first, time-secondmappingischosentoachievelowlatencyandallowsboththetransmitterand receiver to process the data “on the fly”. For high data rates, there aremultiple code blocks in eachOFDM symbol and the device can decode thosereceived in one symbol while receiving the next OFDM symbol. Similarly,assembling an OFDM symbol can take place while transmitting the previoussymbols, thereby enabling a pipelined implementation. This would not bepossible in the case of a time-first mapping as the complete slot needs to be

preparedbeforethetransmissioncanstart.Thevirtualresourceblockscontainingthemodulationsymbolsaremappedto

physicalresourceblocksinthebandwidthpartusedfortransmission.Dependingonthebandwidthpartusedfortransmission,thecarrierresourceblockscanbedeterminedandtheexactfrequencylocationonthecarrierdetermined(seeFig.9.12foranillustration).Thereasonforthis,atfirstsightsomewhatcomplicatedmappingprocesswithbothvirtualandphysicalresourceblocksistobeabletohandleawiderangeofscenarios.

FIGURE9.12 Mappingfromvirtualtophysicaltocarrierresourceblocks.

There are two methods for mapping virtual resource blocks to physicalresource blocks, non-interleaved mapping (Fig. 9.12: top) and interleavedmapping(Fig.9.12:bottom).ThemappingschemetousecanbecontrolledonadynamicbasisusingabitintheDCIschedulingthetransmission.Non-interleavedmappingmeansthatavirtualresourceblockinabandwidth

partmaps directly to the physical resource block in the same bandwidth part.This is useful in cases when the network tries to allocate transmissions tophysical resource with instantaneously favorable channel conditions. Forexample,theschedulermighthavedeterminedthatphysicalresourceblockssixto nine in Fig. 9.12 have favorable radio channel properties and are thereforepreferredfortransmissionandanon-interleavedmappingisused.

Interleavedmappingmapsvirtualresourceblockstophysicalresourceblocksusinganinterleaverspanningthewholebandwidthpartandoperatingonpairsorquadrupletsofresourceblocks.Ablockinterleaverwithtworowsisused,withpairs/quadruplets of resource blocks written column-by-column and read outrow-by-row. Whether to use pairs or quadruplets of resource blocks in theinterleavingoperationisconfigurablebyhigher-layersignaling.The reason for interleaved resource-block mapping is to achieve frequency

diversity,thebenefitsofwhichcanbemotivatedseparatelyforsmallandlargeresourceallocations.For small allocations, for example voice services, channel-dependent

scheduling may not be motivated from an overhead perspective due to theamountof feedback signaling required,ormaynotbepossibledue to channelvariations not being possible to track for a rapidlymoving device. Frequencydiversity by distributing the transmission in the frequency domain is in suchcases an alternative way to exploit channel variations. Although frequencydiversity could be obtained by using resource allocation type 0 (see Section10.1.10), this resource allocation scheme implies a relatively large controlsignalingoverheadcomparedtothedatapayloadtransmittedaswellaslimitedpossibilitiestosignalverysmallallocations.Instead,byusingthemorecompactresource allocation type 1, which is only capable of signaling contiguousresource allocations, combinedwith an interleavedvirtual tophysical resourceblock mapping, frequency diversity can be achieved with a small relativeoverhead.ThisisverysimilartothedistributedresourcemappinginLTE.Sinceresourceallocationtype0canprovideahighdegreeofflexibilityintheresourceallocation,interleavedmappingissupportedforresourceallocationtype1only.Forlargerallocations,possiblyspanningthewholebandwidthpart,frequency

diversitycanstillbeadvantageous.Inthecaseofalargetransportblock,thatis,at very high data rates, the coded data are split into multiple code blocks asdiscussedinSection9.2.2.Mappingthecodeddatadirectlytophysicalresourceblocks in a frequency-first manner (remember, frequency-first mapping isbeneficialfromanoverall latencyperspective)wouldresult ineachcodeblockoccupying only a fairly small number of contiguous physical resource blocks.Hence, if the channel quality varies across the frequency range used fortransmission,somecodeblocksmaysufferworsequalitythanothercodeblocks,possiblyresultingintheoveralltransportblockfailingtodecodedespitealmostall code blocks being correctly decoded. The quality variations across thefrequencyrangemayoccureveniftheradiochannelisflatduetoimperfections

inRFcomponents. Ifan interleavedresource-blockmapping isused,onecodeblockoccupyingacontiguoussetofvirtualresourceblockswouldbedistributedin the frequency domain across multiple, widely separated physical resourceblocks, similarly towhat is the case for the small allocations discussed in theprevious paragraph. The result of the interleaved VRB-to-PRB mapping is aquality-averagingeffectacrossthecodeblocks,resultinginahigherlikelihoodofcorrectlydecodingverylargetransportblocks.Thisaspectofresourceblockmappingwas not present in LTE, partially because the data rateswere not ashighasinNR,partlybecausethecodeblocksinLTEareinterleaved.The discussion above holds in general and for the downlink. In the uplink,

release 15 only specifies RF requirements for contiguous allocations andthereforeinterleavedmappingisonlysupportedfordownlinktransmissions.Toobtain frequency diversity also in the uplink, frequency hopping can be usedwherethedatainthefirstsetofOFDMsymbolsintheslotaretransmittedontheresource block as indicated by the scheduling grant. In the remaining OFDMsymbols, data are transmitted on a different set of resource blocks given by aconfigurable offset from the first set. Uplink frequency hopping can bedynamicallycontrolledusingabitintheDCIschedulingthetransmission.

9.10DownlinkReservedResourcesOneofthekeyrequirementsonNRwastoensureforwardcompatibility,thatis,to allow future extensions and technologies to be introduced in a simplewaywithout causing backward-compatibility problems with, at that point in time,already-deployedNRnetworks.SeveralNR technologycomponentscontributetomeeting this requirement,but thepossibility todefinereservedresources inthe downlink is one of the more important tools. Reserved resources aresemistatically configured time–frequency resources around which the PDSCHcanberate-matched.Reservedresourcescanbeconfiguredinthreedifferentways:

•ByreferringtoanLTEcarrierconfiguration,therebyallowingfortransmissionsonanNRcarrierdeployedontopofanLTEcarrier(LTE/NRspectrumcoexistence)toavoidthecell-specificreferencesignalsoftheLTEcarrier(seefurtherdetailsinChapter17);

•ByreferringtoaCORESET;•Byconfiguringresourcesetsusingasetofbitmaps.

There are no reserved resources in the uplink; avoiding transmission oncertainresourcescanbeachievedthroughscheduling.6Configuring reserved resources by referring to a configured CORESET is

used todynamically controlwhether control signaling resources canbe reusedfordataornot(seeSection10.1.2).Inthiscasethereservedresourceisidenticalto the CORESET configured and the gNBmay dynamically indicate whetherthese resources areusable forPDSCHornot.Thus, reserved resourcesdonothavetobeperiodicallyoccurringbutcanbeusedwhenneeded.Thethirdwaytoconfigurereservedresourcesisbasedonbitmaps.Thebasic

building block for a resource-set configuration covers one or two slots in thetimedomainandcanbedescribedbytwobitmapsasillustratedinFig.9.13:

•Afirsttime-domainbitmap,whichintheNRspecificationsisreferredtoas“bitmap-2,”indicatesasetofOFDMsymbolswithintheslot(orwithinapairtwoslots).

•WithinthesetofOFDMsymbolsindicatedbybitmap-2,anarbitrarysetofresourceblocks,thatis,blocksof12resourceelementsinthefrequencydomain,maybereserved.Thesetofresourceblocksisindicatedbyasecondbitmap,intheNRspecificationsreferredtoas“bitmap-1.”

FIGURE9.13 Configuringreservedresources.

If the resource set is defined on a carrier level, bitmap-1 has a lengthcorresponding to the number of resource blocks within the carrier. If theresource set is bandwidth-part specific, the length of bitmap-1 is given by thebandwidthofthebandwidthpart.Thesamebitmap-1isvalidforallOFDMsymbolsindicatedbybitmap-2.In

other words, the same set of resource elements are reserved in all OFDMsymbols indicatedbybitmap-2.Furthermore, the frequency-domaingranularityoftheresource-setconfigurationprovidedbybitmap-1isoneresourceblock.Inotherwords,all resourceelementswithina (frequency-domain) resourceblockareeitherreservedornotreserved.Whether or not the resources configured as reserved resources are actually

reservedorcanbeusedforPDSCHcaneitherbesemistaticallyordynamicallycontrolled.Inthecaseofsemistaticcontrol,athirdbitmap(bitmap-3)determineswhether

ornottheresource-setdefinedbythebitmap-1/bitmap-2pairortheCORSETisvalidforacertainslotornot.Thebitmap-3hasagranularityequaltothelengthofbitmap-2(eitheroneortwoslots)andalengthof40slots.Inotherwords,theoveralltime-domainperiodicityofasemistaticresourcesetdefinedbythetriplet{bitmap-1,bitmap-2,bitmap-3}is40slotsinlength.Inthecaseofdynamicactivation/deactivationofarate-matchingresourceset,

an indicator in the scheduling assignment indicates if the semistaticallyconfigured pattern is valid or not for a certain dynamically scheduledtransmission.Note that,althoughFig.9.14assumesschedulingonaslotbasis,dynamicindicationisequallyapplicabletotransmissiondurationsshorterthanaslot.Theindicator in theDCIshouldnotbeseenascorrespondingtoacertainslot. Rather, it should be seen as corresponding to a certain schedulingassignment.Whattheindicatordoesissimplyindicateif,foragivenschedulingassignmentdefinedbyagivenDCI,aconfiguredresourcesetshouldbeassumedactiveornotduringthetimeoverwhichtheassignmentisvalid.

FIGURE9.14 Dynamicactivation/deactivationofaresourcesetbymeansofaDCIindicator.

In the general case, a device can be configured with up to eight differentresourcesets.EachresourcesetisconfiguredeitherbyreferringtoaCORSESTorbyusingthebitmapapproachdescribedabove.Byconfiguringmorethanoneresource-setconfiguration,moreelaboratepatternsofreservedresourcescanberealized,asillustratedinFig.9.15.

FIGURE9.15 Dynamicactivation/deactivationinthecaseofmultipleconfiguredresourcesets.

Although a device can be configuredwith up to eight different resource-setconfigurations, each of which can be configured for dynamic activation, theconfigurationscannotbe independentlyactivated in theschedulingassignment.Rather,tomaintainareasonableoverhead,theschedulingassignmentincludesatmost two indicators. Each resource set configured for dynamicactivation/deactivationisassignedtoeitheroneorbothoftheseindicationsandjointly activates/deactivates or disables all resource sets assigned to thatindicator. Fig. 9.15 illustrates an examplewith three configured resource sets,where resource set #1 and resource set #2 are assigned to indicator #1 andindicator#2, respectively,while resource set#2 is assigned toboth indicators.NotethatthepatternsinFig.9.15arenotnecessarilyrealistic,butratherchosenforillustrativepurposes.

9.11ReferenceSignalsReference signals are predefined signals occupying specific resource elementswithinthedownlinktime–frequencygrid.TheNRspecificationincludesseveraltypesofreferencesignalstransmittedindifferentwaysandintendedtobeusedfordifferentpurposesbyareceivingdevice.UnlikeLTE,whichreliesheavilyonalways-on,cell-specificreferencesignals

in thedownlink forcoherentdemodulation,channelqualityestimation forCSIreporting, and general time–frequency tracking, NR uses different downlinkreferencesignals fordifferentpurposes.Thisallowsforoptimizingeachof thereference signals for their specific purpose. It is also in line with the overallprinciple of ultralean transmission as the different reference signals can betransmitted only when needed. Later release of LTE took some steps in thisdirection,butNRcanexploitthistoamuchlargerdegreeastherearenolegacyNRdevicestocaterfor.TheNRreferencesignalsinclude:

•Demodulationreferencesignals(DM-RS)forPDSCHareintendedforchannelestimationatthedeviceaspartofcoherentdemodulation.TheyarepresentonlyintheresourceblocksusedforPDSCHtransmission.Similarly,theDM-RSforPUSCHallowsthegNBtocoherentlydemodulatethePUSCH.TheDM-RSforPDSCHandPUSCHisthefocusofthissection;DM-RSforPDCCHandPBCHaredescribedinChapters10and16,respectively.

•Phase-trackingreferencesignals(PT-RS)canbeseenasanextensiontoDM-RSforPDSCH/PUSCHandareintendedforphase-noisecompensation.ThePT-RSisdenserintimebutsparserinfrequencythantheDM-RS,and,ifconfigured,occursonlyincombinationwithDM-RS.Adiscussionofthephase-trackingreferencesignalisfoundlaterinthischapter.

•CSIreferencesignals(CSI-RS)aredownlinkreferencesignalsintendedtobeusedbydevicestoacquiredownlinkchannel-stateinformation(CSI).SpecificinstancesofCSIreferencesignalscanbeconfiguredfortime/frequencytrackingandmobilitymeasurements.CSIreferencesignalsaredescribedinSection8.1.

•Trackingreferencesignals(TRS)aresparsereferencesignalsintendedtoassistthedeviceintimeandfrequencytracking.AspecificCSI-RSconfigurationservesthepurposeofaTRS(seeSection8.1.7).

•Soundingreferencesignals(SRS)areuplinkreferencesignalstransmittedbythedevicesandusedforuplinkchannel-stateestimationatthebasestations.SoundingreferencesignalsaredescribedinSection8.3.

In the following, the demodulation reference signals intended for coherentdemodulationofPDSCHandPUSCHaredescribedinmoredetail,startingwiththe reference signal structure used for OFDM. The same DM-RS structure isused for both downlink and uplink in the case of OFDM. For DFT-spreadOFDM in theuplink, a reference signalbasedonZadoff–Chu sequences as inLTE is used to improve the power-amplifier efficiency but supportingcontiguousallocationsandsingle-layertransmissiononlyasdiscussedinalatersection.Finally,adiscussiononthephase-trackingreferencesignalisprovided.

9.11.1DemodulationReferenceSignalsforOFDM-BasedDownlinkandUplink

OFDM-BasedDownlinkandUplinkThe DM-RS in NR provides quite some flexibility to cater for differentdeploymentscenariosandusecases:afront-loadeddesigntoenablelowlatency,support for up to 12 orthogonal antenna ports for MIMO, transmissionsdurationsfrom2to14symbols,anduptofourreference-signalinstancesperslottosupportveryhigh-speedscenarios.Toachieve low latency, it isbeneficial to locate thedemodulation reference

signals early in the transmission, sometimes known as front-loaded referencesignals.Thisallowsthereceivertoobtainachannelestimateearlyand,oncethechannel estimate is obtained, process the received symbols on the flywithouthavingtobufferacompleteslotpriortodataprocessing.Thisisessentiallythesame motivation as for the frequency-first mapping of data to the resourceelements.Twomain time-domainstructuresaresupported,differencing in the location

ofthefirstDM-RSsymbol:

•MappingtypeA,wherethefirstDM-RSislocatedinsymbol2or3oftheslotandtheDM-RSismappedrelativetothestartoftheslotboundary,regardlessofwhereintheslottheactualdatatransmissionstarts.Thismappingtypeisprimarilyintendedforthecasewherethedataoccupy(mostof)aslot.Thereasonforsymbol2or3inthedownlinkistolocatethefirstDM-RSoccasionafteraCORESETlocatedatthebeginningofaslot.

•MappingtypeB,wherethefirstDM-RSislocatedinthefirstsymbolofthedataallocation,thatis,theDM-RSlocationisnotgivenrelativetotheslotboundarybutratherrelativetowherethedataarelocated.Thismappingisoriginallymotivatedbytransmissionsoverasmallfractionoftheslottosupportverylowlatencyandothertransmissionsthatbenefitfromnotwaitinguntilaslotboundarystartsbutcanbeusedregardlessofthetransmissionduration.

Themapping type for PDSCH transmission can be dynamically signaled aspartoftheDCI(seeSection9.11fordetails),whileforthePUSCHthemappingtypeissemistaticallyconfigured.Although front-loaded reference signals are beneficial from a latency

perspective,theymaynotbesufficientlydenseinthetimedomaininthecaseofrapid channel variations. To support high-speed scenarios, it is possible to

configure up to three additional DM-RS occasions in a slot. The channelestimator in the receiver can use these additional occasions formore accuratechannel estimation, for example, to use interpolation between the occasionswithinaslot.Itisnotpossibletointerpolatebetweenslots,oringeneraldifferenttransmissionoccasions,asdifferentslotsmaybetransmittedtodifferentdevicesand/orindifferentbeamdirections.ThisisadifferencecomparedtoLTE,whereinterslot interpolationof thechannel estimates ispossiblebut also restricts themulti-antennaandbeamformingflexibilityinLTEcomparedtoNR.Thedifferenttime-domainallocationsforDM-RSareillustratedinFig.9.16,

including both single-symbol and double-symbolDM-RS. The purpose of thedouble-symbolDM-RSisprimarilytoprovidealargernumberofantennaportsthanwhat is possiblewith a single-symbol structure as discussed below.Notethat the time-domain location of the DM-RS depends on the scheduled dataduration.Furthermore,not allpatterns illustrated inFig.9.16areapplicable tothePDSCH(forexample,mappingtypeBforPDSCHonlysupportsduration2,4,and7.

FIGURE9.16 Time-domainlocationofDM-RS.

Multiple orthogonal reference signals can be created in each DM-RS

occasion.Thedifferentreferencesignalsareseparatedinthefrequencyandcodedomains,and, in thecaseofadouble-symbolDM-RS,additionally in the timedomain. Two different types of demodulation reference signals can beconfigured,type1andtype2,differinginthemappinginthefrequencydomainand themaximumnumberoforthogonal referencesignals.Type1canprovideup to four orthogonal signals using a single-symbol DM-RS and up to eightorthogonalreferencesignalsusingadouble-symbolDM-RS.Thecorrespondingnumbersfortype2aresixandtwelve.Thereferencesignaltypes(1or2)shouldnotbeconfusedwiththemappingtypes(AorB);differentmappingtypescanbecombinedwithdifferentreferencesignaltypes.Reference signals should preferably have small power variations in the

frequency domain to allow for a similar channel-estimation quality for allfrequencies spanned by the reference signal. Note that this is equivalent to awell-focused time-domain autocorrelation of the transmitted reference signal.For OFDM-based modulation, a pseudo-random sequence is used, morespecificallya length231–1Goldsequence,whichfulfills therequirementsonawell-focusedautocorrelation.Thesequenceisgeneratedacrossall thecommonresource blocks (CRBs) in the frequency domain but transmitted only in theresourceblocksused fordata transmissionas there isno reason forestimatingthechanneloutside the frequencyregionused for transmission.Generating thereference signal sequence across all the resource blocks ensures that the sameunderlying sequence is used for multiple devices scheduled on overlappingtime–frequency resources in the case of multi-user MIMO (see Fig. 9.17)(orthogonalsequencesareusedontopofthepseudo-randomsequencetoobtainmultipleorthogonalreferencesignalsfromthesamepseudo-randomsequenceasdiscussed later). If the underlying pseudo-random sequence would differbetweendifferentcoscheduleddevices,theresultingreferencesignalswouldnotbeorthogonal.Thepseudo-randomsequence isgeneratedusinga configurableidentity,similartothevirtualcellIDinLTE.Ifnoidentityhasbeenconfigured,itdefaultstothephysical-layercellidentity.

FIGURE9.17 GeneratingDM-RSsequencesbasedoncommonresourceblock0.

Returning to the type 1 reference signals, the underlying pseudo-randomsequence ismapped toeverysecondsubcarrier in the frequencydomain in theOFDM symbol used for reference signal transmission, see Fig. 9.18 for anillustrationassumingonlyfront-loadedreferencesignalsarebeingused.Antennaports7 1000 and1001use even-numbered subcarriers in the frequencydomainandareseparatedfromeachotherbymultiplyingtheunderlyingpseudo-randomsequencewithdifferentlength-2orthogonalsequencesinthefrequencydomain,resultingintransmissionoftwoorthogonalreferencesignalsforthetwoantennaports.Aslongastheradiochannelisflatacrossfourconsecutivesubcarriers,thetworeferencesignalswillbeorthogonalalsoatthereceiver.Antennaports1000and1001aresaidtobelongtoCDMgroup0astheyusethesamesubcarriersbutare separated in the code-domain using different orthogonal sequences.Referencesignalsforantennaports1002and1003belongtoCDMgroup1andaregeneratedinthesamewayusingodd-numberedsubcarriers,thatis,separatedinthecodedomainwithintheCDMgroupandinthefrequencydomainbetweenCDM groups. If more than four orthogonal antenna ports are needed, twoconsecutive OFDM symbols are used instead. The structure above is used ineach of the OFDM symbols and a length-2 orthogonal sequence is used toextendthecode-domainseparationtoalsoincludethetimedomain,resultinginuptoeightorthogonalsequencesintotal.

FIGURE9.18 Demodulationreferencesignalstype1.

Demodulationreferencesignalstype2(seeFig.9.19)haveasimilarstructureto type1, but there are somedifferences,mostnotably thenumberof antenna

ports supported. Each CDM group for type 2 consists of two neighboringsubcarriers overwhich a length-2 orthogonal sequence is used to separate thetwo antenna ports sharing the same set of subcarriers. Two such pairs ofsubcarriersareusedineachresourceblockforoneCDMgroup.Sincethereare12subcarriersinaresourceblock,uptothreeCDMgroupswithtwoorthogonalreference signals each can be created using one resource block in oneOFDMsymbol.ByusingasecondOFDMsymbolandatime-domainlength-2sequenceinthesamewasasfortype1,amaximumof12orthogonalreferencesignalscanbecreatedwithtype2.Althoughthebasicstructuresoftype1andtype2havemanysimilarities, therearealsodifferences.Type1 isdenser in the frequencydomain, while type 2 trades the frequency-domain density for a largermultiplexingcapacity, that is, a largernumberoforthogonal reference signals.This is motivated by the support for multi-user MIMO with simultaneoustransmissiontoarelativelylargenumberofdevices.

FIGURE9.19 Demodulationreferencesignalstype2.

Thereferencesignalstructuretouseisdeterminedbasedonacombinationofdynamic scheduling and higher-layer configuration. If a double-symbolreferencesignal isconfigured, theschedulingdecision,conveyed to thedeviceusing thedownlinkcontrol information, indicates to thedevicewhether tousesingle-symbolordouble-symbolreferencesignals.Theschedulingdecisionalsocontains information for thedevicewhich reference signals (more specifically,which CDM groups) that are intended for other devices (see Fig. 9.20). Thescheduleddevicemapsthedataaroundbothitsownreferencesignalsaswellasthe reference signals intended for another device. This allows for a dynamicchangeofthenumberofcoscheduleddevicesinthecaseofmulti-userMIMO.In the case of spatial multiplexing (also known as single-user MIMO) ofmultiple layers for the same device, the same approach is used—each layerleavesresourceelementscorrespondingtoanotherCDMgroupintendedforthesame device unused. This is to avoid interlayer interference for the referencesignals.

FIGURE9.20 RatematchingdataaroundcoscheduledCDMgroups.

The reference signal description above is applicable to both uplink anddownlink.Notethough,thatforprecoder-baseduplinktransmissions,theuplinkreference signal is applied before the precoder (see Fig. 9.11). Hence, thereferencesignaltransmittedisnotthestructureabove,buttheprecodedversionofit.8

9.11.2DemodulationReferenceSignalsforDFT-PrecodedOFDMUplinkDFT-precodedOFDMsupports single-layer transmissiononly and isprimarilydesignedwithcoverage-challengedsituationsinmind.Duetotheimportanceoflowcubicmetricandcorrespondinghighpower-amplifierefficiencyforuplinkDFT-precoded OFDM, the reference signal structure is somewhat differentcompared to the OFDM case. In essence, transmitting reference signalsfrequencymultiplexedwithotheruplinktransmissionsfromthesamedeviceisnot suitable for the uplink as thatwould negatively impact the device power-

amplifier efficiency due to increased cubic metric. Instead, certain OFDMsymbolswithinaslotareusedexclusivelyforDM-RStransmission—thatis,thereferencesignalsare timemultiplexedwith thedata transmittedon thePUSCHfromthesamedevice.Thestructureofthereferencesignalitselfthenensuresalowcubicmetricwithinthesesymbolsasdescribedbelow.In the time domain, the reference signals follow the same mapping as

configuration type 1. As DFT-precoded OFDM is capable of single-layertransmissiononlyandDFT-precodedOFDMisprimarilyintendedforcoverage-challenged situations, there is no need to support configuration type 2 and itscapability of handling a high degree ofmulti-userMIMO. Furthermore, sincemulti-userMIMOisnota targetedscenarioforDFT-precodedOFDM,there isno need to define the reference signal sequence across all common resourceblocks as for the correspondingOFDM case, but it is sufficient to define thesequenceforthetransmittedphysicalresourceblocksonly.Uplinkreferencesignalsshouldpreferablyhavesmallpowervariationsinthe

frequency domain to allow for similar channel-estimation quality for allfrequencies spanned by the reference signal.As already discussed, forOFDMtransmission it is fulfilled by using a pseudo-random sequence with goodautocorrelation properties. However, for the case of DFT-precoded OFDM,limitedpowervariationsasafunctionoftimearealsoimportanttoachievealowcubic metric of the transmitted signal. Furthermore, a sufficient number ofreference-signal sequences of a given length, corresponding to a certainreference-signal bandwidth, should be available in order to avoid restrictionswhenschedulingmultipledevicesindifferentcells.Atypeofsequencefulfillingthese two requirements is the Zadoff–Chu sequence, discussed in Chapter 8.From a Zadoff–Chu sequence with a given group index and sequence index,additional reference-signal sequences can be generated by applying differentlinearphaserotationsinthefrequencydomain,asillustratedinFig.9.21.ThisisthesameprincipleasusedinLTE.

FIGURE9.21 Generationofuplinkreference-signalsequencefromphase-rotatedbasesequence.

9.11.3Phase-TrackingReferenceSignals(PT-RS)Phase-trackingreferencesignals(PT-RS)canbeseenasanextensiontoDM-RS,intended for tracking phase variations across the transmission duration, forexample, one slot. These phase variations can come from phase noise in theoscillators,primarilyathighercarrierfrequencieswherethephasenoisetendstobehigher.ItisanexampleofareferencesignaltypeexistinginNRbutwithnocorresponding signal in LTE. This is partially motivated by the lower carrierfrequenciesusedinLTE,andhencelessproblematicphasenoisesituation,andpartly it ismotivatedby thepresenceof cell-specific reference signals inLTEwhich can be used for tracking purposes. Since the main purpose is to trackphasenoise,thePT-RSneedstobedenseintimebutcanbesparseinfrequency.ThePT-RSonlyoccursincombinationwithDM-RSandonlyifthenetworkhasconfigured the PT-RS to be present.Depending onwhetherOFDMorDFTS-OFDMisused,thestructurediffers.For OFDM, the first reference symbol (prior to applying any orthogonal

sequence) in the PDSCH/PUSCH allocation is repeated every Lth OFDMsymbol, startingwith the firstOFDM symbol in the allocation. The repetitioncounter is reset at each DM-RS occasion as there is no need for PT-RS

immediately after a DM-RS. The density in the time-domain is linked to thescheduledMCSinaconfigurableway.In the frequencydomain,phase-tracking reference signals are transmitted in

every secondor fourth resource block, resulting in a sparse frequencydomainstructure. The density in the frequency domain is linked to the scheduledtransmissionbandwidthsuchthatthehigherthebandwidth,thelowerthePT-RSdensity in the frequency domain. For the smallest bandwidths, no PT-RS istransmitted.To reduce the risk of collisions between phase-tracking reference signals

associated with different devices scheduled on overlapping frequency-domainresources, the subcarrier number and the resource blocks used for PT-RStransmissionaredeterminedbytheC-RNTIofthedevice.Theantennaportusedfor PT-RS transmission is given by the lowest numbered antenna port in theDM-RS antenna port group. Some examples of PT-RSmappings are given inFig.9.22.

FIGURE9.22 ExamplesofPT-RSmappinginoneresourceblockandoneslot.

ForDFT-precodedOFDMintheuplink,thesamplesrepresentingthephase-trackingreferencesignalareinsertedpriortoDFTprecoding.ThetimedomainmappingfollowsthesameprinciplesasthepureOFDMcase.

1Strictlyspeaking,theRandom-AccessChannelisalsodefinedasatransport-channeltype(seeChapter16).However,RACHonlyincludesalayer-1preambleandcarriesnodataintheformoftransportblocks.2Thesetofpossibletransport-blocksizesaresuchthatitisalwayspossibletosplitatoolargetransportblockintosmallerequal-sizedcode-blocks.3Thisstructureimprovestheperformanceforhigher-ordermodulation.4Asuplinkresourceassignmentsarealwaysdoneintermsofresourceblocksof

size12subcarriers,theDFTsizeisalwaysamultipleof12.5ThespecificationispreparedtohandletwoDM-RSportgroupsaspartofmulti-TRPschemesnotpartofrelease15butplannedforlaterreleases.Inthatcase,someofthePDSCHlayersbelongtooneDM-RSportgroupandtheotherlayerstotheotherDM-RSportgroup.6Onereasonisthatonlyfrequency-contiguousallocationsaresupportedintheuplinkinrelease15,resultingin“bitmap-1”beingunabletobeusedasthismayresultinnon-contiguousfrequency-domainallocations.7Thedownlinkantennaportnumberingisassumedinthisexample.Theuplinkstructureissimilarbutwithdifferentantennaportnumbers.8Ingeneral,thereferencesignaltransmittedisinadditionsubjecttoanyimplementation-specificmulti-antennaprocessing,capturedbythespatialfilterFinSection9.8,andtheword“transmitted”shouldbeunderstoodfromaspecificationperspective.

CHAPTER10

Physical-LayerControlSignaling

Abstract

To support the transmission of downlink and uplink transport channels,there is a need for certain associated control signaling. This controlsignalingisoftenreferredtoasL1/L2controlsignaling,indicatingthatthecorrespondinginformationpartlyoriginatesfromthephysicallayer(Layer1)andpartly fromMAC(Layer2). In thischapter, thedownlinkcontrolssignaling, includingschedulinggrantsandassignments,willbedescribed,followed by the uplink control signaling carrying the necessary feedbackfromthedevice.

KeywordsDCI;UCI;PDCCH;PUCCH;CORESET;searchspace;blinddecoding;CCE

Tosupportthetransmissionofdownlinkanduplinktransportchannels,thereisaneed for certain associated control signaling. This control signaling is oftenreferred to as L1/L2 control signaling, indicating that the correspondinginformationpartly originates from thephysical layer (layer 1) andpartly fromMAC(layer2).In this chapter, the downlink control signaling, including scheduling grants

and assignments, will be described, followed by the uplink control signalingcarryingthenecessaryfeedbackfromthedevice.

10.1DownlinkDownlinkL1/L2controlsignalingconsistsofdownlinkschedulingassignments,including information required for the device to be able to properly receive,demodulate, and decode the DL-SCH on a component carrier, and uplink

schedulinggrantsinformingthedeviceabouttheresourcesandtransportformatto use for uplink (UL-SCH) transmission. In addition, the downlink controlsignalingcanalsobeused for specialpurposes suchasconveying informationabout the symbols used for uplink and downlink in a set of slots, preemptionindication,andpowercontrol.InNR, there is only a single control channel, thephysical downlink control

channel(PDCCH).Onahighlevel,theprinciplesofthePDCCHprocessinginNRaresimilartoLTE,namelythatthedevicetriestoblindlydecodecandidatePDCCHs transmitted from the network using one or more search spaces.However, there are some differences compared to LTE based on the differentdesigntargetsforNRaswellasexperiencefromLTEdeployments:

•ThePDCCHinNRdoesnotnecessarilyspanthefullcarrierbandwidth,unliketheLTEPDCCH.ThisisanaturalconsequenceofthefactthatnotallNRdevicesmaybeabletoreceivethefullcarrierbandwidthasdiscussedinChapter5,andledtothedesignofamoregenericcontrolchannelstructureinNR.

•ThePDCCHinNRisdesignedtosupportdevice-specificbeamforming,inlinewiththegeneralbeam-centricdesignofNRandanecessitywhenoperatingatveryhighcarrierfrequencieswithacorrespondingchallenginglinkbudget.

ThesetwoaspectsweretosomeextentaddressedintheLTEEPDCCHdesignin release 11, although in practice EPDCCH has not been used extensivelyexceptasabasisforthecontrolsignalingforeMTC.TwoothercontrolchannelspresentinLTE,thePHICHandthePCFICH,are

notneededinNR.TheformerisusedinLTEtohandleuplinkretransmissionsand is tightly coupled to the use of a synchronous hybrid-ARQ protocol, butsincetheNRhybrid-ARQprotocolisasynchronousinbothuplinkanddownlinkthe PHICH is not needed in NR. The latter channel, the PCFICH, is notnecessaryinNRasthesizeofthecontrolresourcesets(CORESETs)doesnotvarydynamicallyandreuseofcontrolresourcesfordataishandledinadifferentwaythaninLTE,asdiscussedfurtherbelow.Inthefollowingsections,theNRdownlinkcontrolchannel,thePDCCH,will

be described, including the notion of a CORESETs, the time–frequencyresourcesuponwhichthePDCCHis transmitted.First, thePDCCHprocessingincludingcodingandmodulationwillbediscussed,followedbyadiscussionon

theCORESETs structure. There can bemultiple CORESETs on a carrier andpartofthecontrolresourcesetisthemappingfromresourceelementstocontrolchannelelements(CCEs).OneormoreCCEsfromonecontrolresourcesetareaggregated to form the resources used by one PDCCH. Blind detection, theprocesswherethedeviceattemptstodetectifthereareanyPDCCHstransmittedto the device, is based on search spaces. There can bemultiple search spacesusingtheresourcesinasingleCORESET,asillustratedinFig.10.1.Finally,thecontentsofthedownlinkcontrolinformation(DCI)willbedescribed.

FIGURE10.1 OverviewofPDCCHprocessinginNR.

10.1.1PhysicalDownlinkControlChannelThe PDCCHprocessing steps are illustrated in Fig. 10.2.At a high level, thePDCCHprocessing inNR ismore similar to theLTEEPDCCHthan theLTEPDCCHinthesensethateachPDCCHisprocessedindependently.

FIGURE10.2 PDCCHprocessing.

The payload transmitted on a PDCCH is known as Downlink ControlInformation (DCI) to which a 24-bit CRC is attached to detect transmissionerrorsand toaid thedecoder in the receiver.Compared toLTE, theCRCsizehasbeenincreasedtoreducetheriskofincorrectlyreceivedcontrolinformationandtoassistearlyterminationofthedecodingoperationinthereceiver.SimilarlytoLTE,thedeviceidentitymodifiestheCRCtransmittedthrougha

scrambling operation. Upon receipt of the DCI, the device will compute ascrambledCRCon the payload part using the sameprocedure and compare itagainst the received CRC. If the CRC checks, the message is declared to becorrectly receivedand intendedfor thedevice.Thus, the identityof thedevicethat is supposed to receive theDCImessage is implicitlyencoded in theCRCand not explicitly transmitted. This reduces the number of bits necessary totransmitonthePDCCHas,fromadevicepointofview, there isnodifferencebetweenacorruptmessagewhoseCRCwillnotcheck,andamessageintendedfor another device. Note that the RNTI does not necessarily have to be the

identityof thedevice, theC-RNTI,butcanalsobedifferent typesofgrouporcommonRNTIs,forexample,toindicatepagingorarandom-accessresponse.ChannelcodingofthePDCCHisbasedonPolarcodes,arelativelynewform

of channel coding. The basic idea behind Polar codes is to transform severalinstancesof theradiochannelintoasetofchannelsthatareeithernoiselessorcompletely noisy and then transmit the information bits on the noiselesschannels.Decodingcanbedoneinseveralways,butatypicalapproachistousesuccessivecancellationandlistdecoding.ListdecodingusestheCRCaspartofthe decoding process, which means that the error-detecting capabilities arereduced.For example, listdecodingof size eight results in a lossof threebitsfrom an error-detecting perspective, resulting in the 24-bits CRC providingerror-detecting capabilities corresponding to a 21-bit CRC. This is part of thereasonforthelargerCRCsizecomparedtoLTE.UnlikethetailbitingconvolutionalcodesusedinLTE,whichcanhandleany

numberof informationbits,Polar codesneed tobedesignedwithamaximumnumberofbitsinmind.InNR,thePolarcodehasbeendesignedtosupport512codedbits(priortoratematching)inthedownlink.Upto140informationbitscanbehandled,whichprovidesasufficientmarginforfutureextensionsastheDCIpayloadsizeinrelease15issignificantlyless.Toassistearlyterminationinthedecodingprocess,theCRCisnotattachedattheendoftheinformationbits,butinsertedinadistributedmanner,afterwhichthePolarcodeisapplied.Earlyterminationcanalsobeachievedbyexploitingthepathmetricinthedecoder.Rate matching is used to match the number of coded bits to the resources

availableforPDCCHtransmission.This isasomewhat intricateprocessandisbased on one of shortening, puncturing, or repetition of the coded bits aftersubblock interleaving of 32 blocks. The set of rules selecting betweenshortening, puncturing, and repetition, as well as when to use which of theschemes,isdesignedtomaximizeperformance.Finally, the coded and rate-matched bits are scrambled, modulated using

QPSK,andmappedtotheresourceelementsusedforthePDCCH,thedetailsofwhich will be discussed below. Each PDCCH has its own reference signal,which means that the PDCCH can make full use of the antenna setup, forexample, be beamformed in a particular direction. The complete PDCCHprocessingchainisillustratedinFig.10.2.ThemappingofthecodedandmodulatedDCItoresourceelementsissubject

toacertainstructure,basedoncontrol-channelelements (CCEs)andresource-elementgroups(REGs).AlthoughthenamesareborrowedfromLTE,thesizeof

thetwodifferfromtheirLTEcounterparts,asdoestheCCE-to-REGmapping.APDCCH is transmitted using 1, 2, 4, 8, or 16 contiguous control-channel

elementswiththenumberknownastheaggregationlevel.Thecontrol-channelelementistheunituponwhichthesearchspacesforblinddecodingaredefinedaswillbediscussedinSection10.1.3.Acontrol-channelelementconsistsofsixresource-element groups, each ofwhich is equal to one resource block in oneOFDMsymbol.AfteraccountingfortheDM-RSoverhead,thereare54resourceelements (108 bits) available for PDCCH transmission in one control-channelelement.TheCCE-to-REGmappingcanbeeither interleavedornon-interleaved.The

motivationforhavingtwodifferentmappingschemesis,similarlytothecaseofthe LTE EPDCCH, to be able to provide frequency diversity by using aninterleaved mapping or to facilitate interference coordination and frequency-selective transmission of control channels by using non-interleaved mapping.ThedetailsoftheCCE-to-REGmappingwillbediscussedinthenextsectionaspartoftheoverallCORESETstructure.

10.1.2ControlResourceSetCentral to downlink control signaling inNR is the concept ofCORESETs.Acontrol resource set is a time–frequency resource inwhich the device tries todecodecandidatecontrolchannelsusingoneormoresearchspaces.Thesizeandlocation of a CORESET in the time–frequency domain is semistaticallyconfigured by the network and can thus be set to be smaller than the carrierbandwidth.ThisisespeciallyimportantinNRasacarriercanbeverywide,upto400MHz,and it isnot reasonable toassumealldevicescan receive suchawidebandwidth.In LTE, the concept of a CORESET is not explicitly present. Instead,

downlinkcontrolsignalinginLTEusesthefullcarrierbandwidthinthefirst1–3OFDM symbols (four for the most narrowband case). This is known as thecontrol region inLTEand inprinciple thiscontrol regionwouldcorrespond tothe “LTECORESET” if that termwould have been used.Having the controlchannels spanning the full carrierbandwidthwaswellmotivatedby thedesireforfrequencydiversityandthefactthatallLTEdevicessupportthefull20MHzcarrierbandwidth(atleastatthetimeofspecifyingrelease8).However,inlaterLTEreleasesthisleadtocomplicationswhenintroducingsupportfordevicesnotsupportingthefullcarrierbandwidth,forexample,theeMTCdevicesintroduced

inrelease12.AnotherdrawbackoftheLTEapproachistheinabilitytohandlefrequency-domain interference coordination between cells for the downlinkcontrolchannels.Tosomeextent,thesedrawbackswiththeLTEcontrolchanneldesignwereaddressedwiththeintroductionoftheEPDCCHinrelease11,buttheEPDCCHfeaturehassofarnotbeenwidelydeployedinpracticeasanLTEnetwork stillneeds toprovidePDCCHsupport for initial accessand tohandlenon-EPDCCH-capableLTEdevices.Therefore,amoreflexiblestructureisusedinNRfromthestart.A CORESET can occur at any position within a slot and anywhere in the

frequencyrangeofthecarrier(seeFig.10.3).However,adeviceisnotexpectedto handle CORESETs outside its active bandwidth part. The reason forconfiguring CORESETs on the cell level and not per bandwidth part is tofacilitate reuse of CORSETs between bandwidth parts, for example, whenoperatingwithbandwidthadaptationasdiscussedinSection14.1.1.

FIGURE10.3 ExamplesofCORESETconfigurations.

ThefirstCORSET,CORESET0,isprovidedbythemasterinformationblock(MIB) as part of the configuration of the initial bandwidth part to be able toreceive the remaining system information and additional configurationinformation from the network. After connection setup, a device can beconfigured with multiple, potentially overlapping, CORESETs in addition tousingRRCsignaling.In the time domain, a CORESET can be up to three OFDM symbols in

duration and located anywhere within a slot, although a common scenario,suitablefortrafficscenarioswhenaschedulingdecisionistakenonceperslot,istolocatetheCORESETatthebeginningoftheslot.ThisissimilartotheLTEsituation with control channels at the beginning of each LTE subframe.However, configuring a CORESET at other time instances can be useful, for

example to achieve very low latency for transmissions occupying only a fewOFDMsymbolswithoutwaitingforthestartofthenextslot.It is importanttounderstand that a CORESET is defined from a device perspective and onlyindicates where a device may receive PDCCH transmissions. It does not sayanythingonwhetherthegNBactuallytransmitsaPDCCHornot.Dependingonwherethefront-loadedDM-RSforPDSCHarelocated,inthe

third or fourth OFDM symbol of a slot (see Section 9.11.1), the maximumdurationforaCORESETistwoorthreeOFDMsymbols.ThisismotivatedbythetypicalcaseoflocatingtheCORESETbeforethestartofdownlinkreferencesignalsandtheassociateddata.Inthefrequencydomain,aCORESETisdefinedinmultiplesofsixresourceblocksuptothecarrierbandwidth.Unlike LTE, where the control region can vary dynamically in length as

indicatedbyaspecialcontrolchannel(thePCFICH),aCORESETinNRisoffixed size.This is beneficial froman implementationperspective, both for thedeviceandthenetwork.Fromadeviceperspective,apipelinedimplementationissimplerifthedevicecandirectlystarttoprocessthePDCCHwithouthavingtofirstdecodeanotherchannellikethePCFICHinLTE.Havingastreamlinedand implementation-friendly structure of the PDCCH is important in order torealizetheverylowlatencypossibleinNR.However,fromaspectralefficiencypointofview,itisbeneficialifresourcescanbesharedflexiblybetweencontrolanddatainadynamicmanner.Therefore,NRprovidesthepossibilitytostartthePDSCHdatabefore theendofaCORESET. It isalsopossible to, foragivendevice,reuseunusedCORESETresourcesasillustratedinFig.10.4.Tohandlethis, the general mechanism of reserved resources is used (see Section 9.10).Reserved resources that overlap with the CORESET are configured andinformationintheDCIindicatestothedevicewhetherthereservedresourcesareusable by thePDSCHor not. If they are indicated as reserved, thePDSCH israte-matched around the reserved resources overlapping with the CORESET,and if the resources are indicated as available, the PDSCH uses the reservedresourcesfordataexceptfortheresourcesusedbythePDCCHuponwhichthedevicereceivedtheDCIschedulingthePDSCH.

FIGURE10.4 Noreuse(left)andreuse(right)ofCORESETresourcesfor

datatransmission(thedeviceisconfiguredwithtwoCORESETsinthisexample).

ForeachCORESETthereisanassociatedCCE-to-REGmapping,amappingthat isdescribedusing the termREGbundle.AREGbundle is a setofREGsacrosswhichthedevicecanassumetheprecodingisconstant.Thispropertycanbeexploitedtoimprovethechannel-estimationperformanceinasimilarwayasresource-blockbundlingforthePDSCH.Asalreadymentioned,theCCE-to-REGmappingcanbeeitherinterleavedor

non-interleaved,dependingonwhetherfrequency-diverseorfrequency-selectivetransmission is desired. There is only one CCE-to-REGmapping for a givenCORESET, but since the mapping is a property of the CORESET, multipleCORESETscanbeconfiguredwithdifferentmappings,whichcanbeuseful.Forexample,oneormoreCORESETsconfiguredwithnon-interleavedmappingtobenefitfromfrequency-dependentscheduling,andoneormoreconfiguredwithinterleaved mapping to act as a fallback in case the channel-state feedbackbecomesunreliableduetothedevicemovingrapidly.Thenon-interleavedmappingisstraightforward.TheREGbundlesize issix

for thiscase, that is, thedevicemayassumetheprecoding isconstantacrossawholeCCE.ConsecutivebundlesofsixREGsareusedtoformaCCE.Theinterleavedcaseisabitmoreintricate.Inthiscase,theREGbundlesize

isconfigurablebetweentwoalternatives.Onealternativeissix,applicabletoallCORESET durations, and the other alternative is, depending on the CORSETduration,twoorthree.ForadurationofoneortwoOFDMsymbols,thebundlesizecanbe twoorsix,andforadurationof threeOFDMsymbols, thebundlesizecanbethreeorsix.Intheinterleavedcase,theREGbundlesconstitutingaCCE are obtained using a block interleaver to spread out the different REGbundles in frequency, thereby obtaining frequency diversity. The number ofrows in the block interleaver is configurable to handle different deploymentscenarios(Fig.10.5).

FIGURE10.5 ExamplesofCCE-to-REGmapping.

AspartofthePDCCHreceptionprocess,thedeviceneedstoformachannelestimateusingthereferencesignalsassociatedwiththePDCCHcandidatebeingdecoded. A single antenna port is used for the PDCCH, that is, any transmitdiversityormulti-userMIMOschemeishandledinadevice-transparentmanner.ThePDCCHhas itsowndemodulationreferencesignals,basedon thesame

typeofpseudo-randomsequenceasthePDSCH—thepseudo-randomsequenceisgeneratedacrossallthecommonresourceblocksinthefrequencydomainbuttransmittedonlyintheresourceblocksusedforthePDCCH(withoneexceptionasdiscussedbelow).However,duringinitialaccess,thelocationforthecommonresource blocks is not yet known as it is signaled as part of the systeminformation.Hence,forCORESET0configuredbythePBCH,thesequenceisgeneratedstartingfromthefirstresourceblockintheCORESETinstead.Demodulation reference-signals specific for a given PDCCH candidate are

mapped onto every fourth subcarrier in a resource-element group, that is, thereferencesignaloverheadis1/4.ThisisadenserreferencesignalpatternthaninLTE,whichusesareferencesignaloverheadof1/6,butinLTEthedevicecaninterpolate channel estimates in time and frequency as a consequence of LTEusing a cell-specific reference signal common to all devices and presentregardlessofwhetheracontrol-channeltransmissiontakesplaceornot.Theuseofadedicated referencesignalperPDCCHcandidate isbeneficial,despite theslightly higher overhead, as it allows for different types of device-transparentbeamforming. By using a beamformed control channel, the coverage andperformance can be enhanced compared to the non-beamformed controlchannelsinLTE.1Thisisanessentialpartofthebeam-centricdesignofNR.Whenattempting todecodeacertainPDCCHcandidateoccupyingacertain

set of CCEs, the device can compute the REG bundles that constitute the

PDCCHcandidate.Channel estimationmust beperformedperREGbundle asthenetworkmaychangeprecodingacrossREGbundles.Ingeneral,thisresultsin sufficiently accurate channel estimates for good PDCCH performance.However,thereisalsoapossibilitytoconfigurethedevicetoassumethesameprecodingacrosscontiguous resourceblocks inaCORESET, therebyallowingthedevice todo frequency-domain interpolationof thechannelestimates.Thisalsoimpliesthat thedevicemayusereferencesignalsoutsidethePDCCHit istrying todetect, sometimes referred to aswideband reference signals (seeFig.10.6 for an illustration). In some sense this gives the possibility to partiallymimictheLTEcell-specificreferencesignalsinthefrequencydomain,ofcoursewithacorrespondinglimitationintermsofbeamformingpossibilities.

FIGURE10.6 NormalRSstructure(left)andwidebandRSstructure(right).

Related to channel estimation are, ashasbeendiscussed forother channels,the quasi-colocation relations applicable to the reference signals. If the deviceknows that two reference signals are quasi-collocated, this knowledge can beexploited to improve the channel estimation and, more importantly for thePDCCH,tomanagedifferentreceptionbeamsatthedevice(seeChapter12foradetailed discussion on beam management and spatial quasi-colocation). Tohandle this, each CORESET can be configured with a transmissionconfiguration indication (TCI) state, that is, providing information of theantennaportswithwhich thePDCCHantennaportsarequasi-colocated. If thedevice is a certain CORESET spatially colocated with a certain CSI-RS, thedevicecandeterminewhichreceptionisappropriatewhenattemptingtoreceivePDCCHs in this CORESET, as illustrated in Fig. 10.7. In this example, twoCORESETs have been configured in the device, one CORESET with spatial

QCL betweenDM-RS and CSI-RS #1, and one CORESETwith spatial QCLbetweenDM-RS andCSI-RS #2.Based onCSI-RSmeasurements, the devicehas determined the best reception beam for each of the twoCSI-RS:es.WhenmonitoringCORESET#1forpossiblePDCCHtransmissions,thedeviceknowsthespatialQCLrelationandusestheappropriatereceptionbeam(similarlyforCORESET#2).Inthisway,thedevicecanhandlemultiplereceptionbeamsaspartoftheblinddecodingframework.

FIGURE10.7 ExampleofQCLrelationforPDCCHbeammanagement.

Ifnoquasi-colocation isconfiguredforaCORESETthedeviceassumes thePDCCHcandidatestobequasi-colocatedwiththeSSblockwithrespecttodelayspread,Dopplerspread,Dopplershift,averagedelay,andspatialRxparameters.This is a reasonable assumption as the device has been able to receive anddecodethePBCHinordertoaccessthesystem.

10.1.3BlindDecodingandSearchSpacesAs described above, different DCI formats can be used for transmission on aPDCCHand the formatused isaprioriunknown to thedevice.Therefore, thedeviceneeds toblindlydetect theDCI format. InLTE, the formatwas tightlycoupledtotheDCIsizeandmonitoringforacertainDCIformatinmostcasesimpliedmonitoringforanewDCIsize.InNR,thecouplingbetweenDCIformatsandDCIsizesislesspronounced.

DifferentformatscouldstillhavedifferentDCIsizes,butseveralformatssharethe same DCI size. This allows adding more formats in the future withoutincreasingthenumberofblinddecodings.AnNRdeviceneedstomonitorforup

tofourdifferentDCIsizes;onesizeusedforthefallbackDCIformats,onefordownlinkschedulingassignments,and(unlesstheuplinkdownlinknon-fallbackformatsaresize-aligned)oneforuplinkschedulinggrants.Inaddition,adevicemayneedtomonitorforslot-formatindicationandpreemptionindicationusingafourthsize,dependingontheconfiguration.The CCE structure described in the previous section helps in reducing the

numberofblinddecodingattemptsbutisnotsufficient.Hence,itisrequiredtohavemechanisms to limit thenumberofPDCCHcandidates that thedevice issupposedtomonitor.Clearly,fromaschedulingpointofview,restrictionsintheallowed aggregations are undesirable as they may reduce the schedulingflexibilityandrequireadditionalprocessingatthetransmitterside.Atthesametime, requiring the device to monitor all possible CCE aggregations in allconfiguredCORESETsisnotattractivefromadevice-complexitypointofview.Toimposeasfewrestrictionsaspossibleontheschedulerwhileatthesametimelimiting the maximum number of blind decoding attempts in the device, NRdefines so-called search spaces. A search space is a set of candidate controlchannels formed by CCEs at a given aggregation level, which the device issupposedtoattempttodecode.Astherearemultipleaggregationlevelsadevicecanhavemultiplesearchspaces.TherecanbemultiplesearchspacesusingthesameCORESET and, as already described, there can bemultipleCORESETsconfiguredforadevice.AdeviceisnotsupposedforPDCCHoutsideitsactivebandwidth part, which follows from the overall purpose of a bandwidth part.Furthermore, the monitoring instance of a search space is configurable asillustratedinFig.10.8.

FIGURE10.8 ExampleofPDCCHmonitoringconfiguration.

At a configured monitoring occasion for a search space, the devices willattempt to decode the candidate PDCCHs for that search space. Five differentaggregationlevelscorrespondingto1,2,4,8,and16CCEs,respectively,canbeconfigured.Thehighestaggregationlevel,16,isnotsupportedinLTEandwasaddedtoNRincaseofextremecoveragerequirements.ThenumberofPDCCH

candidates can be configured per search space (and thus also per aggregationlevel). Hence NR has a more flexible way of spending the blind decodingattemptsacrossaggregationlevelsthanLTE,wherethenumberofblinddecodesat each aggregation level was fixed. This is motivated by the wider range ofdeploymentsexpectedforNR.Forexample,inasmall-cellscenariothehighestaggregationlevelsmaynotbeused,anditisbettertospendthelimitednumberof blind decoding attempts the device is dimensioned for on the loweraggregationlevelsthanonblinddecodingonanaggregationlevelthatisneverused.Upon attempting to decode a candidate PDCCH, the content of the control

channel is declared as valid for this device if theCRC checks and the deviceprocessestheinformation(schedulingassignment,schedulinggrants,etc.).IftheCRC does not check, the information is either subject to uncorrectabletransmissionerrorsorintendedforanotherdeviceandineithercasethedeviceignoresthatPDCCHtransmission.The network can only address a device if the control information is

transmitted on a PDCCH formed by the CCEs in one of the device’s searchspaces.For example, deviceA inFig. 10.9 cannot be addressedon aPDCCHstartingatCCEnumber20,whereasdeviceBcan.Furthermore, ifdeviceAisusingCCEs16–23,deviceBcannotbeaddressedonaggregationlevel4asallCCEsinitslevel-4searchspaceareblockedbyuseforotherdevices.Fromthisitcanbe intuitivelyunderstood that forefficientutilizationof theCCEs in thesystem, the search spaces should differ between devices. Each device in thesystemcanthereforehaveoneormoredevice-specificsearchspacesconfigured.As a device-specific search space is typically smaller than the number ofPDCCHs the network could transmit at the corresponding aggregation level,there must be a mechanism determining the set of CCEs in a device-specificsearchspace.

FIGURE10.9 Exampleofsearchspacesfortwodifferentdevices.

One possibility would be to let the network configure the device-specificsearchspaceineachdevice,similartothewaytheCORESETsareconfigured.However, this would require explicit signaling to each of the devices andpossiblyreconfigurationathandover.Instead, thedevice-specificsearchspacesfor PDCCH are defined without explicit signaling through a function of thedevice identityunique in the cell, that is, theC-RNTI.Furthermore, the set ofCCEsthedeviceshouldmonitorforacertainaggregationlevelalsovariesasafunction of time to avoid two devices constantly blocking each other. If theycollideatonetimeinstant,theyarenotlikelytocollideatthenexttimeinstant.Ineachofthesesearchspaces,thedeviceisattemptingtodecodethePDCCHsusingthedevice-specificC-RNTIidentity.2Ifvalidcontrolinformationisfound,forexample,aschedulinggrant,thedeviceactsaccordingly.However, there is also information intended for a group of devices.

Furthermore,aspartoftherandom-accessprocedure,itisnecessarytotransmitinformation to a device before it has been assigned a unique identity. Thesemessages are scheduledwith different predefinedRNTIs, for example, theSI-RNTIforschedulingsysteminformation, theP-RNTItransmissionofapagingmessage, theRA-RNTIfor transmissionof the random-access,andTPC-RNTIforuplinkpowercontrol response.Otherexamplesare the INT-RNTIusedforpreemption indication and the SFI-RNTI used for conveying slot-relatedinformation.Thesetypesofinformationcannotrelyonadevice-specificsearchspace as different devices wouldmonitor different CCEs despite themessagebeingintendedforallofthem.Hence,NRalsodefinescommonsearchspaces.3Acommonsearchspaceissimilarinstructuretoadevice-specificsearchspacewith thedifference that the set ofCCEs is predefined andhenceknown to alldevices,regardlessoftheirownidentity.The number of blind decoding attempts depends on the subcarrier spacing

(and hence the slot duration). For 15/30/60/120 kHz subcarrier spacing, up to44/36/22/20 blind decoding attempts per slot can be supported across allDCIpayload sizes—a number selected to offer a good tradeoff between devicecomplexityandschedulingflexibility.However,thenumberofblinddecodedisnot the only measure of complexity but also channel estimation needs to beaccounted for. The number of channel estimates for subcarrier spacings of15/30/60/120kHzhasbeenlimitedto56/56/48/32CCEsacrossallCORESETsinaslot.Dependingontheconfiguration,thenumberofPDCCHcandidatemaybelimitedeitherbythenumberofblinddecodes,orbythenumberofchannel

estimates. Finally, to limit the device complexity, a “3+1”DCI size budget isdefined,meaningthatadeviceatmostmonitorsthreedifferentDCIsizesusingtheC-RNTI(andhencebeingtime-critical)andoneDCIsizeusingotherRNTIs(andhencelesstimecritical).In the case of carrier aggregation, the general blind decoding operation

describedaboveisappliedpercomponentcarrier.Thetotalnumberofchannelestimatesandblinddecodingattemptsisincreasedcomparedtothesinglecarriercase,butnotindirectproportiontothenumberofaggregatedcarriers.

10.1.4DownlinkSchedulingAssignments—DCIFormats1–0and1–1Havingdescribedthe transmissionofDCIonPDCCH,thedetailedcontentsofthecontrolinformationcanbediscussed,startingwiththedownlinkschedulingassignments. Downlink scheduling assignments useDCI format 1–1, the non-fallbackformat,orDCIformat1–0,alsoknownasthefallbackformat.The non-fallback format 1–1 supports all NR features. Depending on the

featuresthatareconfiguredinthesystem,someinformationfieldsmayormaynotbepresent.Forexample,ifcarrieraggregationisnotconfigured,thereisnoneed to include carrier-aggregation-related information in the DCI. Hence theDCIsizeforformat1–1dependsontheoverallconfiguration,butaslongasthedevice knows which features are configured, it also knows the DCI size andblinddetectioncanbeperformed.The fallback format 1–0 is smaller in size, supports a limited set of NR

functionality, and the set of information fields is in general not configurable,resultingina(moreorless)fixedDCIsize.Oneusecaseofthefallbackformatistohandleperiodsofuncertaintyintheconfigurationofadeviceastheexacttimeinstantwhenadeviceappliestheconfigurationinformationisnotknowntothenetwork, for exampledue to transmission errors.Another reason for usingthe fallback DCI is to reduce control signaling overhead. In many cases thefallback format provides sufficient flexibility for scheduling smaller datapackets.Parts of the contents are the same for the differentDCI formats, as seen in

Table10.1, but there are alsodifferencesdue to thedifferent capabilities.TheinformationintheDCIformatsusedfordownlinkschedulingcanbeorganizedintodifferentgroups,with thefieldspresentvaryingbetweentheDCIformats.The content ofDCI formats for downlink scheduling assignments is described

below:

•IdentifierofDCIformat(1bit).ThisisaheadertoindicatewhethertheDCIisadownlinkassignmentoranuplinkgrant,whichisimportantincasethepayloadsizesofmultipleDCIformatsarealignedandthesizecannotbeusedtodifferentiatetheDCIformats(oneexamplehereofisthefallbackformats0–0and1–0whichareofequalsize).

•Resourceinformation,consistingof:•Carrierindicator(0or3bit).Thisfieldispresentifcross-carrierschedulingisconfiguredandisusedtoindicatethecomponentcarriertheDCIrelatesto.ThecarrierindicatorisnotpresentinthefallbackDCIforexampleusedforcommonsignalingtomultipledevices,asnotalldevicesmaybeconfiguredwith(orcapableof)carrieraggregation.

•Bandwidth-partindicator(0–2bit),usedtoactivateoneofuptofourbandwidthpartsconfiguredbyhigher-layersignaling.NotpresentinthefallbackDCI.

•Frequency-domainresourceallocation.ThisfieldindicatestheresourceblocksononecomponentcarrieruponwhichthedeviceshouldreceivethePDSCH.Thesizeofthefielddependsonthesizeofthebandwidthandontheresourceallocationtype,type0only,type1only,ordynamicswitchingbetweenthetwoasdiscussedinSection10.1.10.Format1–0supportsresourceallocationtype1onlyasthefullflexibilityinresourceallocationisnotneededinthiscase.

•Time-domainresourceallocation(1–4bit).ThisfieldindicatestheresourceallocationinthetimedomainasdescribedinSection10.1.11

•VRB-to-PRBmapping(0or1bit)toindicatewhetherinterleavedornon-interleavedVRB-to-PRBmappingshouldbeusedasdescribedinSection9.9.Onlypresentforresourceallocationtype1.

•PRBsizeindicator(0or1bit),usedtoindicatethePDSCHbundlingsizeasdescribedinSection9.9.

•Reservedresources(0–2bit),usedtoindicatetothedeviceifthereservedresourcescanbeusedforPDSCHornotasdescribedinSection9.10.

•Zero-powerCSI-RStrigger(0–2bit),seeSection8.1foradiscussiononCSIreferencesignals.

•Transport-block-relatedinformation:•Modulation-and-codingscheme(5bit),usedtoprovidethedevicewithinformationaboutthemodulationscheme,thecoderate,andthetransport-blocksize,asdescribedfurtherbelow.

•New-dataindicator(1bit),usedtoclearthesoftbufferforinitialtransmissionsasdiscussedinSection13.1.

•Redundancyversion(2bit)(seeSection13.1).•Ifasecondtransportblockispresent(onlyifmorethanfourlayersofspatialmultiplexingaresupportedinDCIformat1–1),thethreefieldsabovearerepeatedforthesecondtransportblock.

•Hybrid-ARQ-relatedinformation:•Hybrid-ARQprocessnumber(4bit),informingthedeviceaboutthehybrid-ARQprocesstouseforsoftcombining.

•Downlinkassignmentindex(DAI,0,2,or4bit),onlypresentinthecaseofadynamichybrid-ARQcodebookasdescribedinSection13.1.5.DCIformat1–1supports0,2,or4bits,whileDCIformat1–0uses2bits.

•HARQfeedbacktiming(3bit),providinginformationonwhenthehybrid-ARQacknowledgmentshouldbetransmittedrelativetothereceptionofthePDSCH.

•CBGtransmissionindicator(CBGTI,0,2,4,6,or8bit),indicatingthecodeblockgroupsretransmittedasdescribedinSection13.1.2.OnlypresentinDCIformat1–1andonlyifCBGretransmissionsareconfigured.

•CBGflushinformation(CBGFI,0–1bit),indicatingsoftbufferflushingasdescribedinSection13.1.2.OnlypresentinDCIformat1–1andonlyifCBGretransmissionsareconfigured.

•Multi-antenna-relatedinformation(presentinDCIformat1–1only):•Antennaports(4–6bit),indicatingtheantennaportsuponwhichthedataaretransmittedaswellasantennaportsscheduledforotherusersasdiscussedinChapters9and11.

•Transmissionconfigurationindication(TCI,0or3bit),usedtoindicatetheQCLrelationsfordownlinktransmissionsas

describedinChapter12.•SRSrequest(2bit),usedtorequesttransmissionofasoundingreferencesignalasdescribedinSection8.3.

•DM-RSsequenceinitialization(0or1bit),usedtoselectbetweentwopreconfiguredinitializationvaluesfortheDM-RSsequence.

•PUCCH-relatedinformation:•PUCCHpowercontrol(2bit),usedtoadjustthePUCCHtransmissionpower.

•PUCCHresourceindicator(3bit),usedtoselectthePUCCHresourcefromasetofconfiguredresources(seeSection10.2.7).

Table10.1

10.1.5UplinkSchedulingGrants—DCIFormats0–0and0–1UplinkschedulinggrantsuseoneofDCIformats0–1,thenon-fallbackformat,orDCI format 0–0, also known as the fallback format.The reason for havingboth a fallback and a non-fallback format is the same as for the downlink,namely to handle uncertainties during RRC reconfiguration and to provide alow-overheadformatfortransmissionsnotexploitingalluplinkfeatures.Asfortheuplink, the informationfieldspresent in thenon-fallbackformatdependonthefeaturesthatareconfigured.TheDCIsizesfortheuplinkDCIformat0–1anddownlinkDCIformat1–1

arealignedwithpaddingaddedtothesmallerofthetwoinordertoreducethenumberofblinddecodes.Parts of the contents are the same for the differentDCI formats, as seen in

Table10.2, but there are alsodifferencesdue to thedifferent capabilities.TheinformationintheDCIformatsusedforuplinkschedulingcanbeorganizedintodifferentgroups,with thefieldspresentvaryingbetweentheDCIformats.ThecontentofDCIformats0–1and0–0isdescribedbelow:

•IdentifierofDCIformat(1bit),aheadertoindicatewhethertheDCIisadownlinkassignmentoranuplinkgrant.

•Resourceinformation,consistingof:•Carrierindicator(0or3bit).Thisfieldispresentifcross-carrierschedulingisconfiguredandisusedtoindicatethecomponentcarriertheDCIrelatesto.ThecarrierindicatorisnotpresentinDCIformat0–0.

•UL/SULindicator(0or1bit),usedtoindicatewhetherthegrantrelatestothesupplementaryuplinkortheordinaryuplink(seeSection7.7).Onlypresentifasupplementaryuplinkisconfiguredaspartofthesysteminformation.

•Bandwidth-partindicator(0–2bit),usedtoactivateoneofuptofourbandwidthpartsconfiguresbyhigher-layersignaling.NotpresentinDCIformat0–0.

•Frequency-domainresourceallocation.ThisfieldindicatestheresourceblocksononecomponentcarrieruponwhichthedeviceshouldtransmitthePUSCH.Thesizeofthefielddependsonthesizeofthebandwidthandontheresource

allocationtype,type0only,type1only,ordynamicswitchingbetweenthetwoasdiscussedinSection10.1.10.Format0–0supportsresourceallocationtype1only.

•Time-domainresourceallocation(0–4bit).ThisfieldindicatestheresourceallocationinthetimedomainasdescribedinSection10.1.11.

•Frequency-hoppingflag(0or1bit),usedtohandlefrequencyhoppingforresourceallocationtype1.

•Transport-block-relatedinformation:•Modulation-and-codingscheme(5bit),usedtoprovidethedevicewithinformationaboutthemodulationscheme,thecoderate,andthetransport-blocksize,asdescribedfurtherbelow.

•New-dataindicator(1bit),usedtoindicatewhetherthegrantrelatestoretransmissionofatransportblockortransmissionofanewtransportblock.

•Redundancyversion(2bit).•Hybrid-ARQ-relatedinformation:

•HybridARQprocessnumber(4bit),informingthedeviceaboutthehybrid-ARQprocessto(re)transmit.

•Downlinkassignmentindex(DAI),usedforhandlingofhybrid-ARQcodebooksincaseofUCItransmittedonPUSCH.NotpresentinDCIformat0–0.

•CBGtransmissionindicator(CBGTI,0,2,4,or6bit),indicatingthecodeblockgroupstoretransmitasdescribedinSection13.1.OnlypresentinDCIformat0–1andonlyifCBGretransmissionsareconfigured.

•Multi-antenna-relatedinformation(presentinDCIformat1–1only):•DMRSsequenceinitialization(1bit),usedtoselectbetweentwopreconfiguredinitializationvaluesfortheDM-RSsequence.

•Antennaports(2–5bit),indicatingtheantennaportsuponwhichthedataaretransmittedaswellasantennaportsscheduledforotherusersasdiscussedinChapters9and11.

•SRSresourceindicator(SRI),usedtodeterminetheantennaportsanduplinktransmissionbeamtouseforPUSCHtransmissionasdescribedinSection11.3.Thenumberofbits

dependsonthenumberofSRSgroupsconfiguredandwhethercodebook-basedornon-codebook-basedprecodingisused.

•Precodinginformation(0–6bit),usedtoselecttheprecodingmatrixWandthenumberoflayersforcodebook-basedprecodingasdescribedinSection11.3.Thenumberofbitsdependsonthenumberofantennaportsandthemaximumranksupportedbythedevice.

•PTRS-DMRSassociation(0or2bit),usedtoindicatetheassociationbetweentheDM-RSandPT-RSports.

•SRSrequest(2bit),usedtorequesttransmissionofasoundingreferencesignalasdescribedinSection8.3.

•CSIrequest(0–6bit),usedtorequesttransmissionofaCSIreportasdescribedinSection8.1.

•Power-control-relatedinformation:•PUSCHpowercontrol(2bit),usedtoadjustthePUSCHtransmissionpower.

•Betaoffset(0or2bit),usedtocontroltheamountofresourcesusedbyUCIonPUSCHincasedynamicbetaoffsetsignalingisconfiguredforDCIformat0–1asdiscussedinSection10.2.8.

Table10.2

10.1.6SlotFormatIndication—DCIFormat2–0DCIformat2–0,ifused,isusedtosignaltheslotformatinformation(SFI)tothedevice as discussed in Section 7.8.3. The SFI is transmitted using the regularPDCCH structure and using the SFI-RNTI, common to multiple devices. Toassist the device in the blind decoding process, the device is configured withinformationabouttheuptotwoPDCCHcandidatesuponwhichtheSFIcanbetransmitted.

10.1.7PreemptionIndication—DCIFormat2–1DCI format 2–1 is used to signal the preemption indicator to the device. It istransmittedusingtheregularPDCCHstructure,usingtheINT-RNTIwhichcanbe common to multiple devices. The details and usage of the preemptionindicatorarediscussedinSection14.1.2.

10.1.8UplinkPowerControlCommands—DCIFormat2–2

Format2–2As a complement to the power-control commands provided as part of thedownlinkschedulingassignmentsandtheuplinkschedulinggrants, thereis thepotentialtotransmitapower-controlcommandusingDCIformat2–2.Themainmotivation for DCI format 2–2 is to support power control for semipersistentscheduling.Inthiscasethereisnodynamicschedulingassignmentorschedulinggrant which can include the power control information for the PUCCH andPUSCH, respectively. Consequently, another mechanism is needed and DCIformat2–2fulfills thisneed.Thepower-controlmessageisdirectedtoagroupofdevicesusinganRNTIspecificforthatgroupandeachdeviceisconfiguredwiththepowercontrolbitsinthejoinmessageitshouldfollow.DCIformat2–2is alignedwith the size ofDCI formats 0–0/1–0 to reduce the blind decodingcomplexity.

10.1.9SRSControlCommands—DCIFormat2–3DCIformat2–3isusedforpowercontrolofuplinksoundingreferencesignalsfordeviceswhichhavenotcoupledtheSRSpowercontroltothePUSCHpowercontrol,eitherbecauseindependentcontrolisdesirableorbecausethedeviceisconfiguredwithoutPUCCHandPUSCH.ThestructureissimilartoDCIformat2–2, but with the possibility to, for each device, configure two bits for SRSrequestinadditiontothetwopowercontrolbits.DCIformat2–2isalignedwiththesizeofDCIformats0–0/1–0toreducetheblinddecodingcomplexity.

10.1.10SignalingofFrequency-DomainResourcesTodetermine the frequency-domain resources to transmitor receiveupon, twofieldsareofinterest:theresource-blockallocationfieldandthebandwidthpartindicator.The resources allocation fields determine the resources blocks in the active

bandwidth part upon which data are transmitted. There are two differentpossibilitiesforsignalingtheresources-blockallocation,type0andtype1,bothinheritedfromLTEwheretheyareknownasdownlinkresourceallocationtype0andtype2.InLTE,theresource-blockallocationsignaledtheallocationacrossthecarrier.However,inNRtheindicationisfortheactivebandwidthpart.Type 0 is a bitmap-based allocation scheme. The most flexible way of

indicating the set of resource blocks the device is supposed to receive thedownlinktransmissionuponistoincludeabitmapwithsizeequaltothenumberof resource blocks in the bandwidth part. This would allow for an arbitrarycombinationof resourceblocks to be scheduled for transmission to the devicebut would, unfortunately, also result in a very large bitmap for the largerbandwidths.Forexample,inthecaseofabandwidthpartof100resourceblocks,thedownlinkPDCCHwouldrequire100bitsforthebitmapalone,towhichtheother pieces of information need to be added.Not onlywould this result in alarge control-signalingoverhead, but it could also result indownlink coverageproblemsasmorethan100bitsinoneOFDMsymbolcorrespondtoadatarateexceeding 1.4 Mbit/s for 15 kHz subcarrier spacing and even higher for thehigher subcarrier spacings.Consequently, there is aneed to reduce thebitmapsize while keeping sufficient allocation flexibility. This can be achieved bypointingnottoindividualresourceblocksinthefrequencydomain,buttogroupsofcontiguousresourceblocks,asshowninatthetopofFig.10.10.Thesizeofsuch a resource-block group is determined by the size of the bandwidth part.Twodifferentconfigurationsarepossible foreachsizeof thebandwidthparts,possiblyresultingindifferentresource-block-groupsizesforagivensizeofthebandwidthpart.

FIGURE10.10 Illustrationofresource-blockallocationtypes(abandwidthpartof25resourceblocksisusedinthisexample).

Resourceallocationtype1doesnotrelyonabitmap.Instead,itencodestheresourceallocationasastartpositionandlengthoftheresource-blockallocation.Thus, it does not support arbitrary allocations of resource blocks but onlyfrequency-contiguousallocations, therebyreducing thenumberofbits requiredforsignalingtheresource-blockallocation.The resource allocation scheme to use is configured according to three

alternatives:type0,type1,ordynamicselectionbetweenthetwousingabitinthe DCI. For the fallback DCIs, only resource block allocation type 1 issupportedasasmalloverheadismoreimportantthantheflexibilitytoconfigure

non-contiguousresources.Bothresource-allocationtypesrefertovirtualresourceblocks(seeSection7.3

foradiscussionofresource-blocktypes).Forresource-allocationtypes0,anon-interleavedmapping fromvirtual tophysical resourceblocks is used,meaningthat the virtual resource blocks are directly mapped to the correspondingphysicalresourceblocks.Forresource-allocationtype1,ontheotherhand,bothinterleaved and non-interleaved mapping is supported. The VRB-to-PRBmapping bit (if present, downlink only) indicates whether the allocationsignalingusesinterleavedornon-interleavedmapping.Returning to the bandwidth part indicator, this field is used to switch the

activebandwidthpart.Itcaneitherpointtothecurrentactivebandwidthpart,orto another bandwidth part to activate. If the field points to the current activebandwidth part, the interpretation of the DCI content is straightforward—theresource allocation applies to the current active bandwidth part as describedabove.However,ifthebandwidthpartindicatorpointstoadifferentbandwidthpart

than the active bandwidth part, the handling becomes more intricate. Manytransmissionparametersingeneralareconfiguredperbandwidthpart.TheDCIpayload size therefore may differ between different bandwidth parts. Thefrequency-domainresourceallocationfieldisanobviousexample;thelargerthebandwidth part, the larger the number of bits for frequency-domain resourceallocation. At the same time, the DCI sizes assumed when performing blinddetection were determined by the currently active bandwidth part, not thebandwidthpart towhichthebandwidthpart indexpoints.Requiringthedeviceto performing blind detection of multiple DCI sizes matching all possiblebandwidth part configurations would be too complex. Hence, the DCIinformationobtainedundertheassumptionoftheDCIformatbeinggivenbythecurrentlyactivebandwidthpartmustbetransformedtothenewbandwidthpart,whichmayhavenotonlyadifferentsizeingeneral,butalsobeconfiguredwitha different set of transmission parameters, for example TCI states which areconfigured per bandwidth part. The transformation is done usingpadding/truncationforeachDCIfieldtomatchtherequirementsofthetargetedbandwidth part. Once this is done, the bandwidth part pointed to by thebandwidth part indicator becomes the new active bandwidth part and thescheduling grant is applied to this bandwidth part. Similar transformation issometimesrequiredforDCIformats0–0and1–0insituationswherethe“3+1”DCIsizebudgetotherwisewouldbeviolated.

10.1.11SignalingofTime-DomainResourcesThe time-domain allocation for the data to be received or transmitted isdynamicallysignaledintheDCI,whichisusefulas thepartofaslotavailablefor downlink reception or uplink transmissionmay vary from slot to slot as aresult of the use of dynamicTDDor the amount of resources used for uplinkcontrols signaling.Furthermore, the slot inwhich the transmissionoccurs alsoneeds to be signaled as part of the time-domain allocation. Although thedownlink data in many cases are transmitted in the same slot as thecorresponding assignment, this is frequently not the case for uplinktransmissions.One approach would be to separately signal the slot number, the starting

OFDM symbol, and the number of OFDM symbols used for transmission orreception. However, as this would result in an unnecessarily large number ofbits, NR has adopted an approach based on configurable tables. The time-domainallocationfieldintheDCIisusedasanindexintoanRRC-configuredtable fromwhich the time-domain allocation is obtained, as illustrated in Fig.10.11.

FIGURE10.11 Signalingoftime-domainallocation(downlink).

There is one table for uplink scheduling grants and one table for downlinkscheduling assignments. Up to 16 rows can be configured where each rowcontains:

•Aslotoffset,thatis,theslotrelativetotheonewheretheDCIwasobtained.Forthedownlink,slotoffsetsfrom0to3arepossible,whilefortheuplinkslotoffsetsfrom0to7canbeused.Thelargeruplinkrangecanbemotivatedbytheneedforschedulinguplinktransmissionsfurtherintothefutureforcoexistencewith(primarily)LTETDD.

•ThefirstOFDMsymbolintheslotwherethedataaretransmitted.•ThedurationofthetransmissioninnumberofOFDMsymbolsintheslot.Notallcombinationsofstartandlengthfitwithinoneslot,forexample,startingatOFDMsymbol12andtransmitduringfiveOFDMsymbolsobviouslyresultsincrossingtheslotboundaryandrepresentsaninvalidcombination.Therefore,thestartandlengtharejointlyencodedtocoveronlythevalidcombinations(althoughinFig.10.11theyareshownastwoseparatecolumnsforillustrativereasons).

•Forthedownlink,thePDSCHmappingtype,thatis,theDM-RSlocationasdescribedinSection9.11,isalsopartofthetable.Thisprovidesmoreflexibilitycomparedtoseparatelyindicatingthemappingtype.

Itisalsopossibletoconfigureslotaggregation,thatis,atransmissionwherethesametransportblockisrepeatedacrossuptoeightslots.However,thisisnotpartofthedynamicsignalingusingatablebutisaseparateRRCconfiguration.Slotaggregationisprimarilyatooltohandlecoverage-challengeddeploymentsandthusthereislessneedforafullydynamicscheme.

10.1.12SignalingofTransport-BlockSizesProper reception of a downlink transmission requires, in addition to the set ofresource blocks, knowledge about the modulation scheme and the transport-blocksize,information(indirectly)providedbythe5-bitMCSfield.Inprinciple,a similar approach as inLTE,namely to tabulate the transport block size as afunctionoftheMCSfieldandtheresource-blockallocationwouldbepossible.However, thesignificantly largerbandwidthssupported inNR, togetherwithawiderangeof transmissiondurationsandvariations in theoverheaddependingonotherfeaturesconfiguredsuchasCSI-RS,wouldresultinalargenumberoftables required to handle the large dynamic range in terms of transport blocksizes. Such a schememay also requiremodificationswhenever some of theseparameters change. Therefore, NR opted for a formula-based approachcombinedwithatableforthesmallesttransport-blocksizesinsteadofapurely

table-basedschemetoachievethenecessaryflexibility.Thefirststep is todetermine themodulationschemeandcoderatefromthe

MCS field.This is doneusingoneof two tables, one table if 256QAM is notconfiguredandanothertableif256QAMisconfigured.Ofthe32combinationsof the 5-bit MCS fields, 29 are used to signal the modulation-and-codingscheme, whereas three are reserved, the purpose of which is described later.Each of the 29 modulation-and-coding scheme entries represents a particularcombinationofmodulation schemeandchannel-coding rateor, equivalently, acertain spectral efficiency measured in the number of information bits permodulation symbol, ranging from approximately 0.2–5.5 bit/s/Hz. For devicesconfiguredwithsupportfor256QAM,fourofthe32combinationsarereservedand the remaining 28 combinations indicate a spectral efficiency in the range0.2–7.4bit/s/Hz.Uptothispoint,theNRschemeissimilartotheoneusedforLTE.However,

toobtainamoreflexiblescheme,thefollowingstepsdiffercomparedtoLTE.Giventhemodulationorder,thenumberofresourceblocksscheduled,andthe

scheduledtransmissionduration,thenumberofavailableresourceelementscanbe computed. From this number the resource elements used for DM-RS aresubtracted.Aconstant,configuredbyhigher layersandmodeling theoverheadbyothersignalssuchasCSI-RSorSRSisalsosubtracted.Theresultingestimateof resource elements available for data is then, together with the number oftransmission layers, themodulationorder, and thecode rateobtained from theMCS, used to calculate an intermediate number of information bits. Thisintermediate number is then quantized to obtain the final transport block sizewhileatthesametimeensuringbyte-alignedcodeblocks,andthatnofillerbitsare needed in the LDPC coding. The quantization also results in the sametransport block size being obtained, even if there are small variations in theamount of resources allocated, a property that is useful when schedulingretransmissionsonadifferentsetofresourcesthantheinitialtransmission.Returning to the threeor four reservedcombinations in themodulation-and-

codingfieldmentionedatthebeginningofthissection,thoseentriescanbeusedforretransmissionsonly.Inthecaseofaretransmission,thetransport-blocksizeis, by definition, unchanged and fundamentally there is no need to signal thispiece of information. Instead, the three or four reserved values represent themodulation scheme—QPSK, 16QAM, 64QAM, or (if configured) 256QAM—whichallowstheschedulertousean(almost)arbitrarycombinationofresourceblocksfortheretransmission.Obviously,usinganyofthethreeorfourreserved

combinationsassumesthatthedeviceproperlyreceivedthecontrolsignalingforthe initial transmission; if this is not the case, the retransmission shouldexplicitlyindicatethetransport-blocksize.The derivation of the transport-block size from the modulation-and-coding

schemeandthenumberofscheduledresourceblocksisillustratedinFig.10.12.

FIGURE10.12 Calculatingthetransportblocksize.

10.2UplinkSimilartoLTE,thereisalsoaneedforuplinkL1/L2controlsignalingtosupportdata transmission on downlink and uplink transport channels. Uplink L1/L2controlsignalingconsistsof:

•Hybrid-ARQacknowledgmentsforreceivedDL-SCHtransportblocks;•Channel-stateinformation(CSI)relatedtothedownlinkchannelconditions,usedtoassistdownlinkscheduling,includingmulti-antennaandbeamformingschemes;and

•Schedulingrequests,indicatingthatadeviceneedsuplinkresourcesforUL-SCHtransmission.

There is no UL-SCH transport-format information included in the uplinktransmission.AsmentionedinSection6.4.4,thegNBisincompletecontroloftheuplinkUL-SCHtransmissionsandthedevicealwaysfollowstheschedulinggrants received from the network, including the UL-SCH transport formatspecifiedinthosegrants.Thus,thenetworkknowsthetransportformatusedforthe UL-SCH transmission in advance and there is no need for any explicit

transport-formatsignalingontheuplink.Thephysicaluplinkcontrolchannel(PUCCH)isthebasisfortransmissionof

uplink control. In principle, the UCI could be transmitted on the PUCCHregardlessofwhether thedevice is transmittingdataon thePUSCH.However,especially if the uplink resources for the PUSCH and the PUCCH are on thesamecarrier(or, tobemoreprecise,use thesamepoweramplifier)butwidelyseparatedinthefrequencydomain,thedevicemayneedarelativelylargepowerback-off to fulfill the spectral emission requirements with a correspondingimpacton theuplinkcoverage.Hence, similarly toLTE,NRsupportsUCIonPUSCH as the basic way of handling simultaneous transmission of data andcontrol.Thus,ifthedeviceistransmittingonthePUSCHtheUCIismultiplexedwithdataonthegrantedresourcesinsteadofbeingtransmittedonthePUCCH.Simultaneous PUSCH and PUCCH is not part of release 15 but may beintroducedinalaterrelease.Beamformingcanbeapplied to thePUCCH.This is realizedbyconfiguring

oneormorespatialrelationsbetweenthePUCCHanddownlinksignalssuchasCSI-RSorSSblock.Inessence,suchaspatialrelationmeansthatthedevicecantransmit the uplink PUCCH using the same beam as it used for receiving thecorresponding downlink signal. For example, if the spatial relation betweenPUCCHandSSblockisconfigured,thedevicewilltransmitPUCCHusingthesamebeamasitusedforreceivingtheSSblock.MultiplespatialrelationscanbeconfiguredandMACcontrolelementsusedtoindicatewhichonetouse.Inthecaseofcarrieraggregation,theuplinkcontrolinformationistransmitted

on the primary cell as a baseline. This is motivated by the need to supportasymmetriccarrieraggregationwiththenumberofdownlinkcarrierssupportedbyadevicethatisunrelatedtothenumberofuplinkcarriers.Foralargenumberof downlink component carriers, a single uplink carrier may carry a largenumberofacknowledgments.Toavoidoverloadingasinglecarrier,itispossibleto configure twoPUCCHgroupswhere feedback relating to the first group istransmittedintheuplinkofthePCellandfeedbackrelatingtotheothergroupofcarriersistransmittedontheprimarysecondcell(PSCell),asillustratedinFig.10.13.

FIGURE10.13 MultiplePUCCHgroups.

In the following section, the basic PUCCH structure and the principles forPUCCH control signaling are described, followed by control signaling onPUSCH.

10.2.1BasicPUCCHStructureUplinkcontrolinformationcanbetransmittedonPUCCHusingseveraldifferentformats.Two of the formats, 0 and 2, are sometimes referred to as short PUCCH

formats,astheyoccupyatmosttwoOFDMsymbols.InmanycasesthelastoneortwoOFDMsymbolsinaslotareusedforPUCCHtransmission,forexample,to transmit ahybrid-ARQacknowledgmentof thedownlinkdata transmission.TheshortPUCCHformatsinclude:

•PUCCHformat0,capableoftransmittingatmosttwobitsandspanningoneortwoOFDMsymbols.Thisformatcan,forexample,beusedtotransmitahybrid-ARQacknowledgmentofadownlinkdatatransmission,ortoissueaschedulingrequest.

•PUCCHformat2,capableoftransmittingmorethantwobitsandspanningoneortwoOFDMsymbols.Thisformatcan,forexample,beusedforCSIreportsorformulti-bithybrid-ARQacknowledgmentsinthecaseofcarrieraggregationorper-CBGretransmission.

Threeoftheformats,1,3,and4,aresometimesreferredtoas longPUCCHformatsas theyoccupyfrom4to14OFDMsymbols.Thereasonforhavingalongertimedurationthantheprevioustwoformatsiscoverage.Ifadurationofone or two OFDM symbols does not provide sufficient received energy forreliable reception, a longer time duration is necessary and one of the longPUCCHformatscanbeused.ThelongPUCCHformatsinclude:

•PUCCHformat1,capableoftransmittingatmosttwobits.•PUCCHformats3and4,bothcapableoftransmittingmorethantwobitsbutdifferinginthemultiplexingcapacity,thatis,howmanydevicesthatcanusethesametime–frequencyresourcesimultaneously.

SincethePUSHuplinkcanbeconfiguredtouseeitherOFDMorDFT-spreadOFDM, one natural thought would be to adopt a similar approach for thePUCCH.However, to reduce the number of options to specify, this is not thecase.Instead,thePUCCHformatsareingeneraldesignedforlowcubicmetric,PUCCH format 2 being the exception and using pure OFDM only. Anotherchoicemade to simplify the overall designwas to only support specification-transparent transmit diversity schemes. In other words, there is only a singleantenna port specified for the PUCCH and if the device is equipped withmultipletransmitantennasit isuptothedeviceimplementationhowtoexploitthese antennas, for example by using some form of delay diversity. In thefollowing, the detailed structure of each of these PUCCH formats will bedescribed.

10.2.2PUCCHFormat0PUCCHformat0,illustratedinFig.10.14,isoneoftheshortPUCCHformatsand is capable of transmitting up to two bits. It is used for hybrid-ARQacknowledgmentsandschedulingrequests.

FIGURE10.14 ExampleofPUCCHformat0.

SequenceselectionisthebasisforPUCCHformat0.Forthesmallnumberofinformation bits supported by PUCCH format 0, the gain from coherentreceptionisnotthatlarge.Furthermore,multiplexinginformationandreferencesignals in one OFDM symbol while maintaining a low cubic metric is notpossible.Therefore,adifferentstructurewheretheinformationbit(s)selectsthesequencetotransmitisused.Thetransmittedsequenceisgeneratedbydifferentphaserotationsofthesameunderlyinglength-12basesequence,wherethebasesequences are the same base sequences defined for generating the referencesignalinthecaseofDFT-precededOFDMasdescribedinSection9.11.2.Thus,thephaserotationappliedtothebasesequencecarriestheinformation.Inotherwords,theinformationselectsoneofseveralphase-rotatedsequences.Twelve different phase rotations are defined for the same base sequence,

providingup to12differentorthogonalsequencesfromeachbasesequence.Alinearphaserotationinthefrequencydomainisequivalenttoapplyingacyclicshiftinthetimedomain,hence,theterm“cyclicshift”issometimesusedwithanimplicitreferencetothetimedomain.Tomaximize theperformance, thephase rotations representing thedifferent

information bits are separated with 2π·6/12 and 2π·3/12 for one and two bitsacknowledgments, respectively. In the case of a simultaneous schedulingrequest,thephaserotationisincreasedby3π/12foroneacknowledgmentbitandby2π/12fortwobits,asillustratedinFig.10.15.

FIGURE10.15 Examplesofphaserotationsasafunctionofhybrid-ARQacknowledgmentsandschedulingrequest.

The phase rotation applied to a certain OFDM symbol carrying PUCCHformat 0 depends not only on the information to be transmitted as alreadymentioned, but also on a reference rotation provided as part of the PUCCHresourceallocationmechanismasdiscussedinSection10.2.7.Theintentionwith

the reference rotation is to multiplex multiple devices on the same time–frequencyresource.Forexample,twodevicestransmittingasinglehybrid-ARQacknowledgmentcanbegivendifferentreferencephaserotationssuchthatonedevice uses 0 and 2π·6/12, while the other device uses 2π·3/12 and 2π·9/12.Finally,thereisalsoamechanismforcyclicshifthoppingwhereaphaseoffsetvaryingbetweendifferentslotsisadded.Theoffsetisgivenbyapseudo-randomsequence.Theunderlyingreasonistorandomizeinterferencebetweendifferentdevices.Thebasesequencetousecanbeconfiguredpercellusinganidentityprovided

as part of the system information. Furthermore, sequence hopping, where thebasesequenceusedvariesonaslot-by-slotbasis,canbeusedtorandomizetheinterference between different cells. As seen from this description manyquantitiesarerandomizedinordertomitigateinterference.PUCCHformat0istypicallytransmittedattheendofaslotasillustratedin

Fig. 10.14.However, it is possible to transmit PUCCH format 0 also in otherpositionswithinaslot.Oneexampleisfrequentlyoccurringschedulingrequests(as frequent as every second OFDM symbol can be configured). Anotherexamplewhenthiscanbeusefulistoacknowledgeadownlinktransmissiononadownlink carrier at a high carrier frequency and, consequently, acorrespondingly higher subcarrier spacing and shorter downlink slot duration.Thisisarelevantscenariointhecaseofcarrieraggregationandsupplementaryuplink,asdiscussed inChapter7. If lowlatency is important, thehybrid-ARQacknowledgmentneedstobefedbackquicklyaftertheendofthedownlinkslot,which is not necessarily at the end of the uplink slot if the subcarrier spacingdiffersbetweenuplinkanddownlink.In the case of two OFDM symbols used for PUCCH format 0, the same

information is transmitted in both OFDM symbols. However, the referencephaserotationaswellasthefrequency-domainresourcesmayvarybetweenthesymbols,essentiallyresultinginafrequency-hoppingmechanism.

10.2.3PUCCHFormat1PUCCHformat1istosomeextentthelongPUCCHcounterpartofformat0.Itis capable of transmittingup to twobits, using from4 to 14OFDMsymbols,eachoneresourceblockwideinfrequency.TheOFDMsymbolsusedaresplitbetween symbols for control information and symbols for reference signals toenablecoherentreception.Thenumberofsymbolsusedforcontrolinformation

and reference signal, respectively, is a tradeoff between channel-estimationaccuracyandenergyintheinformationpart.Approximatelyhalfthesymbolsforreference symbols were found to be a good compromise for the payloadssupportedbyPUCCHformat2.The one or two information bits to be transmitted are BPSK or QPSK

modulated, respectively, and multiplied by the same type of length-12 low-PAPRsequenceasusedforPUCCHformat0.Similartoformat0,sequenceandcyclic shift hopping can be used to randomize interference. The resultingmodulated length-12 sequence is block-wise spread with an orthogonal DFTcode of the same length as the number of symbols used for the controlinformation. The use of the orthogonal code in the time domain increases themultiplexing capacity asmultiple devices having the same base sequence andphaserotationstillcanbeseparatedusingdifferentorthogonalcodes.The reference signals are inserted using the same structure, that is, an

unmodulated length-12 sequence is block-spreadwith an orthogonal sequenceand mapped to the OFDM symbols used for PUCCH reference-signaltransmission.Thus,thelengthoftheorthogonalcode,togetherwiththenumberof cyclic shifts, determines the number of devices that can transmit PUCCHformat1onthesameresource.AnexampleisshowninFig.10.16wherenineOFDM symbols are used for PUCCH transmission, four carrying theinformation and five used for reference signals. Hence, up to four devices,determined by the shorter of the codes for the information part, can share thesame cyclic shift of the base sequence, and a set of resources for PUCCHtransmissioninthisparticularexample.Assumingacell-specificbasesequenceandsixoutofthe12cyclicshiftsbeingusefulfromadelay-spreadperspective,thisresults inamultiplexingcapacityofatmost24devicesonthesametime–frequencyresources.

FIGURE10.16 ExampleofPUCCHformat1withoutfrequencyhopping(top)andwithfrequencyhopping(bottom).

ThelongertransmissiondurationofthelongPUCCHformatscomparedtoashort single-symbol format opens the possibility for frequency hopping as amean to achieve frequency diversity in a similar way as in LTE. However,unlikeLTE,wherehoppingisalwaysdoneattheslotboundarybetweenthetwoslots used for PUCCH, additional flexibility is needed in NR as the PUCCHduration can vary depending on the scheduling decisions and overall systemconfiguration.Furthermore,asthedevicesaresupposedtotransmitwithintheiractive bandwidth part only, hopping is typically not between the edges of theoverall carrier bandwidth as in LTE. Therefore, whether to hop or not isconfigurableanddeterminedaspartofthePUCCHresourceconfiguration.Theposition of the hop is obtained from the length of the PUCCH. If frequencyhopping is enabled, one orthogonal block-spreading sequence is used per hop.Anexample isprovided inFig.10.16where, insteadof a single setof length-4/length-5 orthogonal sequences, two sets of sequences length-2/length-2 andlength-2/length-3,areusedforthefirstandsecondhops,respectively.

10.2.4PUCCHFormat2

PUCCH format 2 is a short PUCCH format based on OFDM and used fortransmissionofmore than twobits, forexample,simultaneousCSIreportsandhybrid-ARQ acknowledgments, or a larger number of hybrid-ARQacknowledgments.Aschedulingrequestcanalsobeincludedinthebitsjointlyencoded. If the bits to be encoded are too large, theCSI report is dropped topreservethehybrid-ARQacknowledgmentswhicharemoreimportant.Theoveralltransmissionstructureisstraightforward.Forlargerpayloadsizes,

a CRC is added. The control information (after CRC attachment) to betransmittediscoded,usingReed–Mullercodesforpayloadsuptoandincluding11bitsandPolar4codingforlargerpayloads,followedbyscramblingandQPSKmodulation. The scrambling sequence is based on the device identity (the C-RNTI) together with the physical-layer cell identity (or a configurable virtualcellidentity),ensuringinterferencerandomizationacrosscellsanddevicesusingthesamesetoftime–frequencyresources.TheQPSKsymbolsarethenmappedtosubcarriersacrossmultipleresourceblocksusingoneortwoOFDMsymbols.A pseudo-random QPSK sequence, mapped to every third subcarrier in eachOFDMsymbol,isusedasademodulationreferencesignaltofacilitatecoherentreceptionatthebasestation.ThenumberofresourceblocksusedbyPUCCHformat2isdeterminedbythe

payload size and a configurablemaximum code rate. The number of resourceblocksis thussmaller if thepayloadsizeissmaller,keepingtheeffectivecoderateroughlyconstant.Thenumberofresourceblocksusedisupperboundedbyaconfigurablelimit.PUCCHformat2istypicallytransmittedattheendofaslotasillustratedin

Fig. 10.17. However, similarly to format 0 and for the same reasons, it ispossibletotransmitPUCCHformat2alsoinotherpositionswithinaslot.

FIGURE10.17 ExampleofPUCCHformat2(theCRCispresentonlyforlargerpayloads).

10.2.5PUCCHFormat3PUCCHformat3canbeseenasthelongPUCCHcounterparttoPUCCHformat2.MorethantwobitscanbetransmittedusingPUCCHformat3usingfrom4to14symbols,eachofwhichcanbemultipleresourceblockswide.Thus,itisthePUCCHformatwiththelargestpayloadcapacity.SimilartoPUCCHformat1,theOFDMsymbolsusedaresplitbetweensymbolsforcontrolinformationandsymbols for reference signals to allow for a low cubicmetric of the resultingwaveform.Thecontrol informationtobetransmittediscodedusingReed–Mullercodes

for11bitsor less andPolar codes for largepayloads, followedby scramblingandmodulation.Thescramblingsequenceisbasedonthedeviceidentity(theC-RNTI) together with the physical-layer cell identity (or a configurable virtualcellidentity),ensuringinterferencerandomizationacrosscellsanddevicesusingthesamesetof time–frequencyresources.Following theprinciplesofPUCCHformat2,aCRCisattached to thecontrol informationfor the largerpayloads.

ThemodulationschemeusedisQPSKbutitispossibletooptionallyconfigureπ/2-BPSKtolowerthecubicmetricatalossinlinkperformance.The resultingmodulation symbols aredividedbetween theOFDMsymbols.

DFT precoding is applied to reduce the cubic metric and improve the poweramplifierefficiency.Thereferencesignalsequenceisgeneratedinthesamewayas for DFT-precoded PUSCH transmissions (see Section 9.11.2) for the samereason,namelytomaintainalowcubicmetric.Frequencyhoppingcanbe configured forPUCCH format3 as illustrated in

Fig.10.18,forexample,toexploitfrequencydiversity,butitisalsopossibletooperate without frequency hopping. The placements of the reference signalsymbolsdependonwhetherthefrequencyhoppingisusedornotandthelengthof thePUCCH transmission, as theremustbeat leastone reference signalperhop.ThereisalsoapossibilitytoconfigureadditionalreferencesignallocationsforthelongerPUCCHdurationstogettworeferencesignalinstancesperhop.

FIGURE10.18 ExampleofPUCCHformat3(theCRCispresentforlargepayloadsizesonly).

Themapping of theUCI is such that themore critical bits, that is, hybrid-ARQ acknowledgments, scheduling request, andCSI part 1, are jointly codedandmappedclosetotheDM-RSlocations,whilethelesscriticalbitsaremappedintheremainingpositions.

10.2.6PUCCHFormat4PUCCHformat4(seeFig.10.19)isinessencethesameasPUCCHformat3butwiththepossibilitytocode-multiplexmultipledevicesinthesameresourceandusing at most one resource block in the frequency domain. Each control-information-carryingOFDMsymbolcarries12/NSFuniquemodulationsymbols.Prior to DFT-precoding, each modulation symbol is block-spread with an

orthogonal sequence of length NSF. Spreading factors two and four aresupported,implyingamultiplexingcapacityoftwoorfourdevicesonthesamesetofresourceblocks.

FIGURE10.19 ExampleofPUCCHformat4.

10.2.7ResourcesandParametersforPUCCHTransmissionInthediscussionofthedifferentPUCCHformatsabove,anumberofparameterswere assumed to be known. For example, the resource blocks to map thetransmittedsignalto,theinitialphaserotationforPUCCHformat0,whethertousefrequencyhoppingornot,andthelengthinOFDMsymbolsforthePUCCH

transmission.Furthermore,thedevicealsoneedstoknowwhichofthePUCCHformatstouse,andwhichtime–frequencyresourcestouse.InLTE,especiallyinthefirstreleases,thereisafairlyfixedlinkagebetween

the uplink control information, the PUCCH format, and the transmissionparameters. For example, LTE PUCCH format 1a/1b is used for hybrid-ARQacknowledgmentsandthetime–frequency-coderesourcestousearegivenbyafixedtimeoffsetfromthereceptionofthedownlinkschedulingassignmentandtheresourcesusedforthedownlinkassignment.Thisisalow-overheadsolution,but has the drawback of being inflexible and was extended to provide moreflexibilityinlaterreleasesofLTEsupportingcarrieraggregationandothermoreadvancedfeatures.NR has adopted a more flexible scheme from the beginning, which is

necessary given the very flexible framework with a wide range of servicerequirements in terms of latency and spectral efficiency, support of nopredefined uplink–downlink allocation in TDD, different devices supportingaggregation of different number of carriers, and different antenna schemesrequiringdifferentamountsoffeedbackjusttonamesomemotivations.Centralin this scheme is thenotionofPUCCHresource sets.APUCCH resource setcontains at least four PUCCH resource configurations, where each resourceconfiguration contains the PUCCH format to use and all the parametersnecessary for that format.Up to fourPUCCHresourcesetscanbeconfigured,each of them corresponding to a certain range of UCI feedback to transmit.PUCCHresourceset0canhandleUCIpayloadsuptotwobitsandhenceonlycontainPUCCHformats0and1,whiletheremainingPUCCHresourcesetsmaycontainanyPUCCHformatexceptformat0and1.When the device is about to transmitUCI, theUCI payload determines the

PUCCHresourcesetand theARI in theDCIdetermines thePUCCHresourceconfiguration within the PUCCH resource set (see Fig. 10.20). Thus, theschedulerhascontrolofwheretheuplinkcontrolinformationistransmitted.Forperiodic CSI reports and scheduling request opportunities, which both aresemistaticallyconfigured,thePUCCHresourcesareprovidedaspartoftheCSIorSRconfiguration.

FIGURE10.20 ExampleofPUCCHresourcesets.

10.2.8UplinkControlSignalingonPUSCHIf the device is transmitting data on PUSCH—that is, has a valid schedulinggrant—simultaneouscontrolsignalingcouldinprincipleremainonthePUCCH.However, as already discussed, this is not the case as in many cases it ispreferable tomultiplex data and control on PUSCH and avoid a simultaneousPUCCH.OnereasonistheincreasedcubicmetriccomparedtoUCIonPUSCHwhenusingDFT-precodedOFDM.Anotherreasonis themorechallengingRFimplementation if out-of-band emission requirements should bemet at highertransmission powers and with PUSCH and PUCCH widely separate in thefrequency domain. Hence, similar to LTE, UCI on PUSCH is the mainmechanism for simultaneous transmission of UCI and uplink data. The sameprinciplesareusedforbothOFDMandDFT-precodedOFDMintheuplink.Only hybrid-ARQ acknowledgments and CSI reports are rerouted to the

PUSCH. There is no need to request a scheduling grant when the device isalreadyscheduled;instead,in-bandbuffer-statusreportscanbesentasdescribedinSection14.2.3.In principle, the base station knows when to expect a hybrid-ARQ

acknowledgment from the device and can therefore perform the appropriatedemultiplexing of the acknowledgment and the data part. However, there is acertainprobabilitythatthedevicehasmissedtheschedulingassignmentonthedownlinkcontrol channel. In this case thebase stationwouldexpect ahybrid-ARQ acknowledgment while the device will not transmit one. If the rate-matchingpatternwoulddependonwhetheranacknowledgmentistransmittedornot,allthecodedbitstransmittedinthedatapartcouldbeaffectedbyamissedassignmentandarelikelytocausetheUL-SCHdecodingtofail.One possibility to avoid this error is to puncture hybrid-ARQ

acknowledgments onto the coded UL-SCH stream in which case the non-punctured bits are unaffected by the presence/absence of hybrid-ARQacknowledgments.ThisisalsothesolutionadoptedinLTE.However,giventhepotentially large number of acknowledgment bits due to, for example, carrieraggregation of the use of codeblock group retransmissions, puncturing is lesssuitableasageneralsolution.Instead,NRhasadoptedaschemewhereuptotwohybrid-ARQacknowledgmentbits arepunctured,while for a largernumberofbitsratematchingoftheuplinkdataisused.Toavoidtheaforementionederrorcases,theuplinkDAIfieldintheDCIindicatestheamountofresourcesreservedfor uplink hybrid ARQ. Thus, regardless of whether the device missed anypreviousschedulingassignmentsornot, theamountofresources tousefor theuplinkhybrid-ARQfeedbackisknown.Themapping of theUCI is such that themore critical bits, that is, hybrid-

ARQ acknowledgments, aremapped to the first OFDM symbol after the firstdemodulationreferencesignal.Lesscriticalbits,thatisCSIreports,aremappedtosubsequentsymbols.Unlikethedatapart,whichreliesonrateadaptationtohandledifferentradio

conditions, this cannot be used for the L1/L2 control-signaling part. Powercontrolcould,inprinciple,beusedasanalternative,butthiswouldimplyrapidpowervariationsinthetimedomain,whichnegativelyimpacttheRFproperties.Therefore,thetransmissionpoweriskeptconstantoverthePUSCHdurationandthe amountof resource elements allocated toL1/L2control signaling—that is,thecoderateofthecontrolsignaling—isvaried.InadditiontoasemistaticvaluecontrollingtheamountofPUSCHresourcesusedforUCI,itisalsopossibletosignalthisfractionaspartoftheDCIshouldatightcontrolbeneeded.

1TheLTEEPDCCHintroduceddevice-specificreferencesignalsinordertoallowbeamforming.2Thereissometimesanadditionaldevice-specificidentity,theCS-RNTI,usedforsemipersistentscheduling,asdiscussedinChapter14.3TheNRspecificationsdefineddifferenttypesofcommonsearchspacesdependingontheRNTImonitored,butthisisnotimportantforunderstandingthegeneralprincipleofsearchspaces.4PolarcodingisusedfortheDCIaswell,butthedetailsofthePolarcodingforUCIaredifferent.

CHAPTER11

MultiAntennaTransmission

Abstract

This chapter gives a background tomultiantenna transmission in general,followedbyadetaileddescriptiononNRmultiantennaprecodingforboththedownlinkanduplinktransmissiondirections.

KeywordsMultiantennatransmission;multiantennaprecoding;codebook-basedtransmission;non-codebook-basedtransmission

Multiantenna transmission is a key component of NR, especially at higherfrequencies. This chapter gives a background to multiantenna transmission ingeneral,followedbyadetaileddescriptiononNRmultiantennaprecoding.

11.1IntroductionThe use of multiple antennas for transmission and/or reception can providesubstantialbenefitsinamobile-communicationsystem.Multiple antennas at the transmitter and/or receiver side can be used to

provide diversity against fading by utilizing the fact that the channelsexperiencedbydifferentantennasmaybeatleastpartlyuncorrelated,eitherduetosufficient inter-antennadistanceordue todifferentpolarizationbetween theantennas.Furthermore, by carefully adjusting the phase, and possibly also the

amplitude,ofeachantennaelement,multipleantennasatthetransmittersidecanbeusedtoprovidedirectivity,thatis,tofocustheoveralltransmittedpowerinacertain direction (beam forming) or, in the more general case, to specificlocations in space. Such directivity can increase the achievable data rates andrange due to higher power reaching the target receiver. Directivity will also

reduce the interference to other links, thereby improving the overall spectrumefficiency.Similarly, multiple receive antennas can be used to provide receiver-side

directivity, focusing the reception in the direction of a target signal, whilesuppressinginterferencearrivingfromotherdirections.Finally, the presence of multiple antennas at both the transmitter and the

receiversidescanbeusedtoenablespatialmultiplexing,thatis,transmissionofmultiple“layers”inparallelusingthesametime/frequencyresources.In LTE, multiantenna transmission/reception for diversity, directivity, and

spatial multiplexing is a key tool to enable high data rates and high systemefficiency. However, multiantenna transmission/reception is an even morecriticalcomponentforNRduetothepossibilityfordeploymentatmuchhigherfrequenciescomparedtoLTE.Thereisawell-establishedandtoalargeextentcorrectassumptionthatradio

communicationathigherfrequenciesisassociatedwithhigherpropagationlossand correspondingly reduced communication range. However, at least part ofthis is due to an assumption that the dimensions of the receiver antenna scalewith the wavelength, that is, with the inverse of the carrier frequency. As anexample,atenfoldincreaseinthecarrierfrequency,correspondingtoatenfoldreduction in the wave length, is assumed to imply a corresponding tenfoldreduction in thephysicaldimensionsof thereceiverantennaora factorof100reductioninthephysicalantennaarea.Thiscorrespondstoa20dBreductionintheenergycapturedbytheantenna.If the receiver antenna sizewould instead be kept unchanged as the carrier

frequency increases, the reduction in captured energy could be avoided.However, thiswould imply that theantennasizewould increaserelative to thewavelength,somethingthatinherentlyincreasesthedirectivityoftheantenna.1The gainwith the larger antenna size can thus only be realized if the receiveantennaiswelldirectedtowardsthetargetsignal.Byalsokeepingthesizeofthetransmitter-sideantennaunchanged,inpractice

increasingthetransmit-antennadirectivity,thelinkbudgetathigherfrequenciescanbefurtherimproved.Assumingline-of-sightpropagationandignoringotherlosses, the overall link budget would then actually improve for higherfrequencies. Inpractice therearemanyother factors thatnegatively impact theoverall propagation losses at higher frequencies such as higher atmosphericattenuation and less diffraction leading to degraded non-line-of-sightpropagation.Still,thegainfromhigherantennadirectivityathigherfrequencies

iswidelyutilizedinpoint-to-pointradiolinkswheretheuseofhighlydirectionalantennasatboththetransmitterandreceiversides,incombinationwithline-of-sightlinks,allowsforrelativelylong-rangecommunicationdespiteoperationatveryhighfrequencies.In a mobile-communication system with devices located in many different

directions relative to the base station and the devices themselves having anessentially random rotational direction, the use of fixed highly directionalantennas is obviously not applicable. However, a similar effect, that is, anextension of the overall receive antenna area enabling higher-directivitytransmission,canalsobeachievedbymeansofanantennapanelconsistingofmany small antenna elements. In this case, the dimension of each antennaelement,aswellasthedistancebetweenantennaelements,isproportionaltothewave length.As the frequency increases, the size of each antenna element, aswell as theirmutual distances, is thus reduced.However, assuming a constantsize of the overall antenna configuration, this can be compensated for byincreasingthenumberofantennaelements.Fig.11.1showsanexampleofsuchanantennapanelconsistingof64dual-polarizedantennaelementsandtargetingthe28GHzband.TheAAAbatteryisincludedinthepictureasanindicationoftheoverallsizeoftheantennapanel.

FIGURE11.1 Rectangularantennapanelwith64dual-polarizedantennaelements.

Thebenefit of such an antennapanelwith a large number of small antennaelements, compared to a single large antenna, is that the direction of thetransmitterbeamcanbeadjustedbyseparatelyadjustingthephaseofthesignalsapplied to each antenna element. The same effect can be achieved when amultiantenna panel, such as the one illustrated in Fig. 11.1, is used on thereceiverside, that is, thereceiverbeamdirectioncanbeadjustedbyseparatelyadjustingthephasesofthesignalsreceivedateachantennaelement.In general, any linear multiantenna transmission scheme can be modeled

accordingtoFig.11.2with layers,capturedbythevector ,beingmappedtotransmitantennas(thevector )bymeansofmultiplicationwithamatrix

ofsize .

FIGURE11.2 GeneralmodelofmultiantennatransmissionmappingNLlayerstoNTantennas.

The general model of Fig. 11.2 applies to most cases of multiantennatransmission. However, depending on implementation there will be variousdegreesofconstraintsthatwillimpacttheactualcapabilitiesofthemultiantennatransmission.Onesuchimplementationaspectrelatestowhere,withintheoverallphysical

transmitterchain,themultiantennaprocessing,thatis,thematrix ofFig.11.3,isapplied.Onahighlevelonecandistinguishbetweentwocases:

•Themultiantennaprocessingisappliedwithintheanalogpartofthetransmitterchain,thatis,afterdigital-to-analogconversion(leftpartofFig.11.3);

•Themultiantennaprocessingisappliedwithinthedigitalpartofthetransmitterchain,thatis,beforedigital-to-analogconversion(rightpartofFig.11.3).

FIGURE11.3 Analogvsdigitalmultiantennaprocessing.

Themain drawback of digital processing according to the right part of Fig.11.3 is the implementation complexity, especially the need for one digital-to-analog converter per antenna element. In the case of operation at higher

frequencies with a large number of closely spaced antenna elements, analogmultiantennaprocessingaccordingtotheleftpartofFig.11.3willthereforebethemost commoncase, at least in the short-andmedium-termperspectives. Inthiscase,themultiantennatransmissionwilltypicallybelimitedtoper-antennaphaseshiftsprovidingbeamforming(seeFig.11.4).

FIGURE11.4 Analogmultiantennaprocessingprovidingbeamforming.

It should be noted that this may not be a severe limitation as operation athigherfrequenciesistypicallymoreoftenpower-limitedthanbandwidth-limited,making beam forming more important than, for example, high-order spatialmultiplexing. The opposite is often true for lower frequency bandswhere thespectrum is amore sparse resourcewith less possibility forwide transmissionbandwidths.Analogprocessingtypicallyalsoimpliesthatanybeamformingiscarriedout

onaper-carrierbasis.Forthedownlinktransmissiondirection,thisimpliesthatit isnotpossible to frequencymultiplexbeam-formed transmissions todeviceslocatedindifferentdirectionsrelativetothebasestation.Inotherwords,beam-formedtransmissionstodifferentdeviceslocatedindifferentdirectionsmustbeseparatedintimeasillustratedinFig.11.5.

FIGURE11.5 Time-domain(non-simultaneous)beamforminginmultipledirections.

Inothercases,especiallyinthecaseofasmallernumberofantennaelementsat lower frequencies, multiantenna processing can be applied in the digitaldomain according to the right part of Fig. 11.3. This enables much higherflexibilityinthemultiantennaprocessingwithapossibilityforhigh-orderspatialmultiplexingandwiththetransmissionmatrix beingageneral matrixwhere each elementmay includeboth a phase shift and a scale factor.Digitalprocessing also allows for independent multiantenna processing for differentsignals within the same carrier, enabling simultaneous beam-formedtransmission to multiple devices located in different directions relative to thebasestationalsobymeansoffrequencymultiplexingasillustratedFig.11.6.

FIGURE11.6 Simultaneous(frequency-multiplexed)beamforminginmultipledirections.

In the case of digital processing, or more generally in the case where theantennaweights canbe flexibly controlled, the transmissionmatrix is oftenrefererred to as a precoder matrix and the multiantenna processing is oftenreferredtoasmultiantennaprecoding.The difference in capabilities between analog and digital multiantenna

processingalsoappliestothereceiverside.Inthecaseofanalogprocessing,themultiantennaprocessingisappliedintheanalogdomainbeforeanalog-to-digitalconversion.Inpractice, themultiantennaprocessingisthenlimitedtoreceiver-sidebeamformingwherethereceiverbeamcanonlybedirectedinonedirectionat a time. Reception from two different directions must then take place atdifferenttimeinstances.Digitalimplementation,ontheotherhand,providesfullflexibility,supporting

receptionofmultiplelayersinparallelandenablingsimultaneousbeam-formedreceptionofmultiplesignalsarrivingfromdifferentdirections.Similar to the transmitter side, the drawback of digital multiantenna

processingonthereceiversideisintermsofcomplexity,especiallytheneedforoneanalog-to-digitalconverterperantennaelement.

For the remainder of this chapterwewill focus onmultiantenna precoding,thatis,multiantennatransmissionwithfullcontrolovertheprecodermatrix.ThelimitationsofanalogprocessingandhowthoseimitationsareimpactingtheNRdesignarediscussedinChapter12.One important aspect of multiantenna precoding is whether or not the

precodingisalsoappliedtothedemodulationreferencesignals(DMRSs)usedtosupportcoherentdemodulationoftheprecodedsignal.IftheDMRSsarenotprecoded,thereceiverneedstobeinformedaboutwhat

precoderisusedat thetransmittersidetoenablepropercoherentdemodulationoftheprecodeddatatransmission.Ontheotherhand,ifthereferencesignalsareprecodedtogetherwiththedata,

theprecodingcan,fromareceiverpointofview,beseenaspartoftheoverallmultidimensionalchannel(seeFig.11.7).Simplyspeaking,insteadofthe“true”

channelmatrix ,thereceiverwillseeachannel ofsize thatistheconcatenationofthechannel withwhateverprecoding isappliedatthetransmitterside.Theprecodingisthustransparenttothereceiverimplyingthatthetransmittercan,atleastinprinciple,selectanarbitraryprecodermatrixanddoesnotneedtoinformthereceiverabouttheselectedprecoder.

FIGURE11.7 DMRSprecodedjointlywithdataimplyingthatanymultiantennaprecodingistransparenttothereceiver.

11.2DownlinkMultiAntennaPrecodingAllNRdownlinkphysicalchannelsrelyonchannel-specificDMRSstosupportcoherentdemodulation.Furthermore,adevicecanassume that theDMRSsarejointly precoded with the data in line with Fig. 11.7. Consequently, anydownlinkmultiantenna precoding is transparent to the device and the networkcan, in principle, apply any transmitter-side precodingwith no need to informthedevicewhatprecodingisapplied.2

The specification impact of downlink multiantenna precoding is thereforemainlyrelatedtothemeasurementsandreportingdonebythedevicetosupportnetwork selection of precoder for downlink PDSCH transmission. Theseprecoder-relatedmeasurements and reporting are part of themoregeneralCSIreportingframeworkbasedonreportconfigurationsasdescribedinSection8.2.Asdescribedthere,aCSIreportmayconsistofoneorseveralofthefollowingquantities:

•ARankIndicator(RI),indicatingwhatthedevicebelievesisasuitabletransmissionrank,thatis,asuitablenumberoftransmissionlayersforthedownlinktransmission;

•APrecoder-MatrixIndicator(PMI),indicatingwhatthedevicebelievesisasuitableprecodermatrix,giventheselectedrank;

•AChannel-QualityIndicator(CQI),inpracticeindicatingwhatthedevicebelievesisasuitablechannel-codingrateandmodulationscheme,giventheselectedprecodermatrix.

Asmentionedabove,thePMIreportedbyadeviceindicateswhatthedevicebelieves is a suitable precodermatrix to use for downlink transmission to thedevice.EachpossiblevalueofthePMIthuscorrespondstoonespecificprecodermatrix. The set of possible PMI values thus corresponds to a set of differentprecoder matrices, referred to as the precoder codebook, that the device canselectbetweenwhenreportingPMI.NotethatthedeviceselectsPMIbasedonacertainnumberofantennaports ,givenbythenumberofantennaportsoftheconfigured CSI-RS associated with the report configuration, and the selectedrank .There is thusat leastonecodebook foreachvalidcombinationofand .It is important to understand that the precoder codebooks for downlink

multiantennaprecodingareonlyusedinthecontextofPMIreportinganddonotimpose any restrictions on what precoder matrix is eventually used by thenetworkfordownlinktransmissiontothereportingdevice.Thenetworkcanusewhateverprecoder itwants and theprecoder selectedby thenetworkdoesnothavetobepartofanydefinedcodebook.Inmanycases itobviouslymakessense for thenetwork touse theprecoder

indicatedby the reportedPMI.However, inothercases thenetworkmayhaveadditional input that speaks in favor of a different precoder. As an example,multiantenna precoding can be used to enable simultaneous downlink

transmission tomultiple devices using the same time/frequency resources, so-calledmulti-userMIMO(MU-MIMO).ThebasicprincipleofMU-MIMObasedonmultiantennaprecoding is tochooseprecodingmatrices thatnotonly focusthe energy towards the target device but also limit interference to othersimultaneously scheduled devices. In this case, the selection of precoding fortransmission to a specific device should not only take into account the PMIreported by that device (which only reflects the channel experienced by thatdevice).Rather,theselectionofprecodingfortransmissiontoaspecificdeviceshould, in the general case, take into account the PMI reported by allsimultaneouslyscheduleddevices.Toconcludeonsuitableprecoding in theMU-MIMOscenario typicallyalso

requiresmore detailed knowledge of the channel experienced by each device,compared to precoding in the caseof transmission to a single device.For thisreason,NRdefinestwotypesofCSIthatdiffer in thestructureandsizeof theprecodercodebooks,TypeICSIandTypeIICSI.

•TypeICSIprimarilytargetsscenarioswhereasingleuserisscheduledwithinagiventime/frequencyresource(noMU-MIMO),potentiallywithtransmissionofarelativelylargenumberoflayersinparallel(high-orderspatialmultiplexing);

•TypeIICSIprimarilytargetsMU-MIMOscenarioswithmultipledevicesbeingscheduledsimultaneouslywithinthesametime/frequencyresourcebutwithonlyalimitednumberofspatiallayers(maximumoftwolayers)perscheduleddevice.

The codebooks for Type I CSI are relatively simple and primarily aim atfocusing the transmittedenergyat the target receiver. Interferencebetween thepotentiallylargenumberofparallellayersisassumedtobehandledprimarilybymeansofreceiverprocessingutilizingmultiplereceiveantennas.ThecodebooksforTypeIICSIaresignificantlymoreextensive,allowingfor

the PMI to provide channel informationwithmuch higher spatial granularity.Themoreextensivechannelinformationallowsthenetworktoselectadownlinkprecoderthatnotonlyfocusesthetransmittedenergyatthetargetdevicebutalsolimits the interference to other devices scheduled in parallel on the sametime/frequency resource. The higher spatial granularity of the PMI feedbackcomesatthecostofsignificantlyhighersignalingoverhead.WhileaPMIreportforTypeICSIwillconsistofatmostafewtensofbits,aPMIreportforTypeII

CSI may consist of several hundred bits. Type II CSI is therefore primarilyapplicableforlow-mobilityscenarioswherethefeedbackperiodicityintimecanbereduced.Belowwewill give an overview of the different types of CSI. For amore

detaileddescription,forexample,see[64].

11.2.1TypeICSITherearetwosubtypesofTypeICSI,referredtoasTypeIsingle-panelCSIandType I multi-panel CSI, corresponding to different codebooks. As the namessuggest, the codebooks have been designed assuming different antennaconfigurationsonthenetwork/transmitterside.Notethatanassumptionofaspecificantennaconfigurationwhendesigninga

codebookdoesnotmeanthatthecodebookcannotbeusedindeploymentsbasedon a different antenna configuration. When a device, based on downlinkmeasurements,selectsaprecodermatrixfromacodebook,itdoesnotmakeanyassumptionsregardingtheantennaconfigurationatthenetworksidebutsimplyselectswhatitbelievesisthemostsuitableprecoderinthecodebook,giventheestimatedchannelconditions.

11.2.1.1Single-PanelCSIAsthenamesuggests, thecodebooksforTypeIsingle-panelCSIaredesignedassumingasingleantennapanelwith cross-polarizedantennaelements.AnexampleisillustratedinFig.11.8forthecaseof( , )=(4,2),thatis,a16-portantenna.3

FIGURE11.8 ExampleofassumedantennastructureforTypeIsingle-

panelCSIwith .

Ingeneral,theprecodermatrices inthecodebooksforTypeIsingle-panelCSIcanbeexpressedastheproductoftwomatrices and withinformationabouttheselected and reportedseparatelyasdifferentpartsoftheoverallPMI.The matrix is assumed to capture long-term frequency-independent

characteristicsofthechannel.Asingle isthereforeselectedandreportedfortheentirereportingbandwidth(widebandfeedback).In contrast, the matrix is assumed to capture more short-term and

potentially frequency-dependent characteristics of the channel. Thematrix canthereforebeselectedandreportedonasubbandbasiswhereasubbandcoversafraction of the overall reporting bandwidth. Alternatively, the devicemay notreport at all, in which case the device, when subsequently selecting CQI,should assume that the network randomly selects on a per PRG (PhysicalResourceBlockGroup, seeSection 9.8) basis.Note that this does not imposeanyrestrictionson theactualprecodingappliedat thenetworksidebut isonlyaboutassumptionsmadebythedevicewhenselectingCQI.On a high level, thematrix can be seen as defining a beamor, in some

casesasetofneighborbeams,pointinginaspecificdirection.Morespecifically,thematrix canbewrittenasW1=[B00B]

where each column of the matrix B defines a beam and the blockstructureisduetothetwopolarizations.Notethat,asthematrix isassumedto only capture long-term frequency-independent channel characteristics, thesamebeamdirectioncanbeassumedtofitbothpolarizationdirections.Selecting thematrix or, equivalently, can thus be seen as selecting a

specificbeamdirectionfromalargesetofpossiblebeamdirectionsdefinedbythefullsetof matriceswithinthecodebook.4In the case of rank 1 or rank 2 transmission, either a single beam or four

neighbor beams are defined by thematrix . In the case of four neighboringbeams,correspondingtofourcolumnsin ,thematrix thenselectstheexactbeamtobeusedforthetransmission.As canbereportedonasubbandbasis,it is thuspossible to fine-tune thebeamdirectionper subband. Inaddition,

provides cophasing between the two polarizations. In the case when onlydefines a single beam, corresponding to being a single-coumn matrix, thematrix onlyprovidescophasingbetweenthetwopolarizations.For transmission ranks larger than 2, thematrix defines orthogonal

beams where . The beams, together with the two polarizationdirectionsineachbeamarethenusedfortransmissionoftheRlayers,withthematrix onlyprovidingcophasingbetweenthetwopolarizations.Uptoeightlayerscanbetransmittedtothesamedevice.

11.2.1.2MultipanelCSIIn contrast to single-panel CSI, codebooks for Type I multi-panel CSI aredesignedassumingthe jointuseofmultipleantennapanelsat thenetworksideand takes into account that it may be difficult to ensure coherence betweentransmissions fromdifferentpanels.Morespecifically, thedesignof themulti-panel codebooks assumes an antenna configuration with two or four two-dimensional panels, each with cross-polarized antenna elements. Anexampleof such amulti-panel antenna configuration is illustrated inFig. 11.9forthecaseoffourantennapanelsand ,thatis,a32-portantenna.

FIGURE11.9 ExampleofassumedantennastructureforTypeImulti-panelCSI;32-portantennawithfourantennapanelsand .

ThebasicprincipleofType1multi-panelCSIis thesameas thatofType1single-panelCSI, except that thematrix defines one beamper polarizationand panel. The matrix then provides per-subband cophasing betweenpolarizationsaswellaspanels.The Type I multi-panel CSI supports spatial multiplexing with up to four

layers.

11.2.2TypeIICSIAs already mentioned, Type II CSI provides channel information withsignificantlyhigherspatialgranularitycomparedtoTypeICSI.SimilartoTypeICSI,TypeIICSIisbasedonwidebandselectionandreportingofbeamsfromalargesetofbeams.However,whileTypeICSIintheendselectsandreportsasinglebeam,TypeIICSImayselectandreportuptofourorthogonalbeams.Foreach selected beam and each of the two polarizations, the reported PMI thenprovidesanamplitudevalue(partlywidebandandpartlysubband)andaphasevalue (subband). In the end, this provides amuchmore detailedmodel of thechannel,capturingthemainraysandtheirrespectiveamplitudeandphase.Atthenetworkside,thePMIdeliveredfrommultipledevicescanthenbeused

toidentifyasetofdevicestowhichtransmissioncanbedonesimultaneouslyona set of time/frequency resources (MU-MIMO) and what precoder to use foreachtransmission.As the Type II CSI is targeting the MU-MIMO scenario, transmission is

limitedtouptotwolayersperdevice.

11.3NRUplinkMultiantennaPrecodingNR support uplink (PUSCH) multiantenna precoding with up to four layers.However, asmentioned earlier, in the case ofDFT-based transformprecoding(seeChapter9),onlysingle-layertransmissionissupported.The device can be configured in two different modes for PUSCH

multiantenna precoding, referred to as codebook-based transmission and non-codebook-based transmission, respectively. The selection between these twotransmissionmodesisatleastpartlydependentonwhatcanbeassumedintermsofuplink/downlinkchannelreciprocity,thatis,towhatextentitcanbeassumedthatthedetaileduplinkchannelconditionscanbeestimatedbythedevicebasedondownlinkmeasurements.Like the downlink, any uplink (PUSCH) multiantenna precoding is also

assumed to be applied to the DMRS used for the PUSCH coherentdemodulation.Similartothedownlinktransmissiondirection,uplinkprecodingis thus transparent to the receiver in the sense that receiver-side demodulationcan be carried out without knowledge of the exact precoding applied at thetransmitter (device) side.Note though that thisdoesnotnecessarily imply thatthe device can freely choose the PUSCH precoder. In the case of codebook-

based precoding, the scheduling grant includes information about a precoder,similartothedeviceprovidingthenetworkwithPMIfordownlinkmultiantennaprecoding.However,incontrasttothedownlink,wherethenetworkmayormaynot use the precodermatrix indicated by the PMI, in the uplink direction thedeviceisassumedtousetheprecoderprovidedbythenetwork.AswewillseeinSection 11.3.2, also in the case of non-codebook-based transmission will thenetworkhaveaninfluenceonthefinalchoiceofuplinkprecoder.Anotheraspect thatmayputconstraintsonuplinkmultiantennatransmission

istowhatextentonecanassumecoherencebetweendifferentdeviceantennas,thatis,towhatextenttherelativephasebetweenthesignalstransmittedontwoantennas can be well controlled. Coherence is needed in the case of generalmultiantenna precoding where antenna-port-specific weight factors, includingspecific phase shifts, are applied to the signals transmitted on the differentantenna ports. Without coherence between the antenna ports the use of suchantenna-port-specific weight factors is obviouslymeaningless as each antennaportwouldanywayintroduceamoreorlessrandomrelativephase.TheNRspecificationallows fordifferentdevicecapabilitieswith regards to

such inter-antenna-port coherence, referred to as full coherence, partialcoherence,andnocoherence,respectively.Inthecaseoffullcoherence,itcanbeassumedthatthedevicecancontrolthe

relative phase between any of the up to four ports that are to be used fortransmission.Inthecaseofpartialcoherence,thedeviceiscapableofpairwisecoherence,

thatis,thedevicecancontroltherelativephasewithinpairsofports.However,there is no guarantee of coherence, that is, a controllable phase, between thepairs.Finally, in the case of no coherence there is no guarantee of coherence

betweenanypairofthedeviceantennaports.

11.3.1Codebook-BasedTransmissionThebasicprincipleofcodebook-basedtransmissionisthatthenetworkdecidesonanuplink transmissionrank, that is, thenumberof layers tobe transmitted,and a correspondingprecodermatrix to use for the transmission.Thenetworkinformsthedeviceabouttheselectedtransmissionrankandprecodermatrixaspart of the uplink scheduling grant.At the device side, the precodermatrix isthen applied for the scheduled PUSCH transmission, mapping the indicated

numberoflayerstotheantennaports.To select a suitable rank and a corresponding precodermatrix, the network

needs estimates of the channels between the device antenna ports and thecorrespondingnetworkreceiveantennas.Toenablethis,adeviceconfiguredforcodebook-based PUSCHwould typically be configured for transmission of atleast onemulti-port SRS.Based onmeasurements on the configuredSRS, thenetwork can sound the channel and determine a suitable rank and precodermatrix.The network cannot select an arbitrary precoder. Rather, for a given

combinationofnumberofantennaports ( or )and transmissionrank ( ),thenetworkselectstheprecodermatrixfromalimitedsetofavailableprecoders(the“uplinkcodebook”).Asanexample,Fig.11.10illustratestheavailableprecodermatrices, that is,

thecodebooksforthecaseoftwoantennaports.

FIGURE11.10 Uplinkcodebooksforthecaseoftwoantennaports.

Whenselectingtheprecodermatrix,thenetworkneedstoconsiderthedevicecapability in terms of antenna-port coherence (see above). For devices notsupportingcoherence,onlythefirsttwoprecodermatrixescanthereforebeusedinthecaseofsingle-ranktransmission.Itcanbenotedthatrestrictingthecodebookselectiontothesetwomatricesis

equivalenttoselectingeitherthefirstorsecondantennaportfortransmission.Inthe case of suchantenna selection, awell-controlled phase, that is, coherencebetween the antenna ports, is not required. On the other hand, the remainingprecodervectorsimplylinearcombinationofthesignalsonthedifferentantennaports,whichrequirescoherencebetweentheantennaports.In the caseof rank-2 transmission ( ) only the firstmatrix,whichdoes

not implyanycouplingbetween theantennaports, canbe selected fordevicesthatdonotsupportcoherence.To further illustrate the impactofno,partial, and full coherence,Fig.11.11

illustrates the full set of rank-1precodermatrices for the caseof four antenna

ports. Once again, the matrices corresponding to no coherence are limited toantenna-port selection. The extended set of matrices corresponding to partialcoherence allows for linear combination within pairs of antenna ports withselection between the pairs. Finally, full coherence allows for a linearcombinationoverallfourantennaports.

FIGURE11.11 Single-layeruplinkcodebooksforthecaseoffourantennaports.

The above-described NR codebook-based transmission for PUSCH isessentiallythesameasthecorrespondingcodebook-basedtransmissionforLTEexcept that NR supports somewhat more extensive codebooks. Another morefundamentalextensionofNRcodebook-basedPUSCHtransmission,comparedtoLTE,isthatadevicecanbeconfiguredtotransmitmultiplemulti-portSRS.5In the case of suchmulti-SRS transmission, the network feedback is extendedwith a one-bit SRS resource indicator (SRI) indicating one of the configuredSRSs.Thedeviceshouldthenusetheprecoderprovidedintheschedulinggrantandmap theoutputof theprecoding to theantennaportscorresponding to theSRSindicatedintheSRI.IntermsofthespatialfilterFdiscussedinChapter8,thedifferentSRSswould typicallybe transmittedusingdifferentspatial filters.ThedeviceshouldthentransmittheprecodersignalusingthesamespatialfilterasusedfortheSRSindicatedbytheSRI.Oneway to visualize the use ofmultiple SRS for codebook-based PUSCH

transmission is to assume that the device transmits themulti-port SRSwithin

separate,relativelylargebeams(seeFig.11.12).Thesebeamsmay,forexample,correspond to different device antenna panels with different directions, whereeachpanelincludesasetofantennaelements,correspondingtotheantennaportsof each multi-port SRS. The SRI received from the network then determineswhatbeamtousefor the transmissionwhile theprecoder information(numberoflayersandprecoder)determineshowthetransmissionistobedonewithintheselectedbeam.Asanexample, in thecaseof full-rank transmission thedevicewill do full-rank transmission within the beam corresponding to the SRSselectedbythenetworkandsignaledbymeansofSRI(upperpartofFig.11.12).Attheotherextreme,inthecaseofsingle-ranktransmissiontheprecodingwillinpracticecreateadditionalbeamformingwithin thewiderbeamindicatedbytheSRI(lowerpartofFig.11.12).

FIGURE11.12 Codebook-basedtransmissionbasedonmultipleSRS.Full-ranktransmission(upperpart)andsingle-ranktransmission(lowerpart).

Codebook-based precoding is typically used when uplink/downlinkreciprocitydoesnothold,thatis,whenuplinkmeasurementsareneededinordertodetermineasuitableuplinkprecoding.

11.3.2Non-codebook-BasedPrecodingIn contrast to codebook-based precoding, which is based on networkmeasurementsandselectionofuplinkprecoder,non-codebook-basedprecodingisbasedondevicemeasurementsandprecoder indications to thenetwork.The

basic principle of uplink non-codebook-based precoding is illustrated in Fig.11.13,withfurtherexplanationbelow.

FIGURE11.13 Non-codebook-basedprecoding.

Basedondownlinkmeasurements,inpracticemeasurementsonaconfiguredCSI-RS, the device selects what it believes is a suitable uplink multi-layerprecoder. Non-codebook-based precoding is thus based on an assumption ofchannelreciprocity,thatis,thatthedevicecanacquiredetailedknowledgeoftheuplink channel based on downlink measurements. Note that there are norestrictions on the device selection of precoder, thus the term “non-codebook-based.”Eachcolumnofaprecodermatrix canbeseenasdefiningadigital“beam”

for thecorrespondinglayer.Thedeviceselectionofprecoderfor layerscanthusbeseenas theselectionof differentbeamdirectionswhereeachbeamcorrespondstoonepossiblelayer.Inprinciple,PUSCHtransmissioncouldbedonedirectlyas transmissionoflayersbasedonthedevice-selectedprecoding.However,deviceselectionofa

precoder based on downlink measurements may not necessarily be the bestprecoder from a network point of view. Thus, the NR non-codebook-basedprecodingincludesanadditionalstepwherethenetworkcanmodifythedevice-selected precoder, in practice remove some “beams,” or equivalently somecolumns,fromtheselectedprecoder.Toenablethis,thedeviceappliestheselectedprecodertoasetofconfigured

SRSs, with one SRS transmitted on each layer or “beam” defined by theprecoder(step2inFig.11.13).BasedonmeasurementsonthereceivedSRS,the

network can then decide to modify the device-selected precoder for eachscheduled PUSCH transmission. This is done by indicating a subset of theconfigured SRSs within the SRS resource indicator (SRI) included in thescheduling grant (step 3).6 The device then carries out the scheduled PUSCHtransmission (step4)usinga reducedprecodermatrixwhereonly thecolumnscorrespondingtotheSRSsindicatedwithintheSRIareincluded.Notethat theSRIthenalsoimplicitlydefinesthenumberoflayerstobetransmitted.It shouldbenoted that thedevice indicationofprecoderselection (step2 in

Fig. 11.13) is not done for each scheduled transmission. The uplink SRStransmission indicating device precoder selection can take place periodically(periodicorsemipersistentSRS)orondemand(aperiodicSRS).Incontrast,thenetworkindicationofprecoder,thatisinpracticethenetworkindicationofthesubsetofbeamsofthedeviceprecoder,isthendoneforeachscheduledPUSCHtransmission.

1Thedirectivityofanantennaisroughlyproportionaltothephysicalantennaareanormalizedwiththesquareofthewavelength.2Thedevicemuststillknowthenumberoftransmissionlayers,thatis,thenumberofcolumnsintheprecodermatrixappliedatthenetworkside.3Notethattherearetwoantennaportspercross-polarizedantennaelement.4Notethatevenifthematrix definesasetofbeams,thesebeamsareneighborbeamsthatpointinessentiallythesamedirection.5Release15islimitedtotwoSRSs.6Foradeviceconfiguredfornon-codebook-basedprecodingtheSRImaythusindicatemultipleSRSs,ratherthanasingleSRSwhichisthecaseforcodebook-basedprecoding(seeSection11.3.1).

CHAPTER12

BeamManagement

Abstract

This chapter describes the NR mechanisms for beam management. Thisincludes mechanisms for initial beam adjustment, beam adjustment, andrecoveryfrombeamfailures.

KeywordsBeammanagement;beampair;beamadjustment;beamrecovery;beamfailure

Chapter11discussedmulti-antennatransmissioningeneralandthenfocusedonmulti-antenna precoding. A general assumption for the discussion on multi-antennaprecodingwasthepossibilityfordetailedcontrol,includingbothphaseadjustmentandamplitudescaling,ofthedifferentantennaelements.Inpracticethisrequiresthatmulti-antennaprocessingatthetransmittersideiscarriedoutinthe digital domain before digital-to-analog conversion. Likewise, the receivermulti-antennaprocessingmustbecarriedoutafteranalog-to-digitalconversion.However,inthecaseofoperationathigherfrequencieswithalargenumberof

closelyspaceantennaelements,theantennaprocessingwillratherbecarriedoutintheanalogdomainwithfocusonbeam-forming.Asanalogantennaprocessingwill be carried out on a carrier basis, this also implies that beam-formedtransmission can only be done in one direction at a time. Downlinktransmissions to different devices located in different directions relative to thebasestationmustthereforebeseparatedintime.Likewise,inthecaseofanalog-based receiver-side beam-forming, the receive beam can only focus in onedirectionatatime.Theultimatetaskofbeammanagementis,undertheseconditions,toestablish

andretainasuitablebeampair, that is,a transmitter-sidebeamdirectionanda

corresponding receiver-side beam direction that jointly provide goodconnectivity.AsillustratedinFig.12.1,thebestbeampairmaynotnecessarilycorrespond

to transmitter and receiver beams that are physically pointing directly towardseach other. Due to obstacles in the surrounding environment, such a “direct”pathbetween the transmitterand receivermaybeblockedanda reflectedpathmayprovidebetterconnectivity,asillustratedintheright-handpartofFig.12.1.This is especially true for operation in higher-frequency bands with less“around-the-corner” dispersion. The beam-management functionality must beabletohandlesuchasituationandestablishandretainasuitablebeampairalsointhiscase.

FIGURE12.1 Illustrationofbeampairsinthedownlinkdirection.Direct(left)andviareflection(right).

Fig.12.1illustratesthecaseofbeamforminginthedownlinkdirection,withbeam-based transmission at the network side and beam-based reception at thedevice side. However, beam forming is at least as relevant for the uplinktransmission direction with beam-based transmission at the device side andcorrespondingbeam-basedreceptionatthenetworkside.In many cases, a suitable transmitter/receiver beam pair for the downlink

transmission direction will also be a suitable beam pair for the uplinktransmission direction and vice versa. In 3GPP this is referred to as(downlink/uplink)beamcorrespondence.Inthecaseofbeamcorrespondence,itis sufficient to explicitly determine a suitable beam pair in one of thetransmission directions. The same pair can then be used also in the oppositetransmissiondirection.As beammanagement is not intended to track fast and frequency-selective

channel variations, beam correspondence does not require that downlink anduplink transmission take place on the same carrier frequency. The concept ofbeam correspondence is thus applicable also for FDD operation in paired

spectrum.Ingeneral,beammanagementcanbedividedintodifferentparts:

•Initialbeamestablishment;•Beamadjustment,primarilytocompensateformovementsandrotationsofthemobiledevice,butalsoforgradualchangesintheenvironment;

•Beamrecoverytohandlethesituationwhenrapidchangesintheenvironmentdisruptthecurrentbeampair.

12.1InitialBeamEstablishmentInitial beam establishment includes the procedures and functions by which abeam pair is initially established in the downlink and uplink transmissiondirections,forexample,whenaconnectionisestablished.AswillbedescribedinmoredetailinChapter16,duringinitialcellsearchadevicewillacquireaso-called SS block transmitted from a cell, with the possibility for multiple SSblocks being transmitted in sequence within different downlink beams. ByassociatingeachsuchSSblock,inpracticethedifferentdownlinkbeams,withacorrespondingrandom-accessoccasionandpreamble(seeSection16.2.1.5),thesubsequent uplink random-access transmission can be used by the network toidentify the downlink beam acquired by the device, thereby establishing aninitialbeampair.Whencommunicationcontinuesafterconnectionsetupthedevicecanassume

that network transmissions to the device will be done using the same spatialfilter,inpracticethesametransmitterbeam,asusedfortheacquiredSSblock.Consequently,thedevicecanassumethatthereceiverbeamusedtoacquiretheSSblockwillbeasuitablebeamalsoforthereceptionofsubsequentdownlinktransmissions.Likewise,subsequentuplinktransmissionsshouldbedoneusingthe same spatial filter (the same beam) as used for the random-accesstransmission, implying that the network can assume that the uplink receiverbeamestablishedatinitialaccesswillremainvalid.

12.2BeamAdjustmentOnce an initial beam pair has been established, there is a need to regularlyreevaluatetheselectionoftransmitter-sideandreceiver-sidebeamdirectionsdueto movements and rotations of the mobile device. Furthermore, even for

stationarydevices,movementsofotherobjectsintheenvironmentmayblockorunblockdifferentbeampairs,implyingapossibleneedtoreevaluatetheselectedbeam directions. This beam adjustment may also include refining the beamshape, for example making the beam more narrow compared to a relativelywiderbeamusedforinitialbeamestablishment.In the general case, beam-forming is about beam pairs consisting of

transmitter-side beam-forming and receiver-side beam-forming. Hence, beamadjustmentcanbedividedintotwoseparateprocedures:

•Reevaluationandpossibleadjustmentofthetransmitter-sidebeamdirectiongiventhecurrentreceiver-sidebeamdirection;

•Reevaluationandpossibleadjustmentofthereceiver-sidebeamdirectiongiventhecurrenttransmitter-sidebeamdirection.

As described above, in the general case beam forming, including beamadjustment, needs to be carried out for both the downlink and uplinktransmission directions. However, as also discussed, if downlink/uplink beamcorrespondencecanbeassumed,explicitbeamadjustmentonlyhastobecarriedoutinoneofthedirections,forexample,inthedownlinkdirection.Itcanthenbe assumed that the adjusted downlink beam pair is appropriate also for theoppositetransmissiondirection.

12.2.1DownlinkTransmitter-SideBeamAdjustmentDownlink transmitter-side beam adjustment aims at refining the networktransmit beam, given the receiver beam currently used at the device side. Toenablethis,thedevicecanmeasureonasetofreferencesignals,correspondingto different downlink beams (see Fig. 12.2). Assuming analog beam forming,transmissionswithin the different downlink beamsmust be done in sequence,thatis,bymeansofabeamsweep.

FIGURE12.2 Downlinktransmitter-sidebeamadjustment.

Theresultofthemeasurementsisthenreportedtothenetworkwhich,basedon the reporting, may decide to adjust the current beam. Note that thisadjustmentmaynotnecessarilyimplytheselectionofoneofthebeamsthatthedevice has measured on. The network could, for example, decide to transmitusingabeamdirectionbetweentwoofthereportedbeams.Also note that, during measurements done for transmitter-side beam

adjustment, the device receiver beam should be kept fixed in order for themeasurementstocapturethequalityofthedifferenttransmitterbeamsgiventhecurrentreceivebeam.Toenablemeasurementsandreportingonasetofbeamsasoutlined inFig.

12.2, the reporting frameworkbasedon reportconfigurations (seeSection8.2)canbeused.Morespecifically, themeasurement/reportingshouldbedescribedbyareportconfigurationhavingL1-RSRPasthequantitytobereported.Thesetofreferencesignalstomeasureon,correspondingtothesetofbeams,

shouldbe included in theNZP-CSI-RS resource set associatedwith the reportconfiguration. As described in Section 8.1.6, such a resource set may eitherinclude a set of configured CSI-RS or a set of SS blocks.Measurements forbeammanagementcanthusbecarriedoutoneitherCSI-RSorSSblock.InthecaseofL1-RSRPmeasurementsbasedonCSI-RS,theCSI-RSshouldbelimitedto single-port or dual-port CSI-RS. In the latter case, the reported L1-RSRPshouldbealinearaverageoftheL1-RSRPmeasuredoneachport.The device can report measurements corresponding to up to four reference

signals(CSI-RSorSSblocks),inpracticeuptofourbeams,inasinglereportinginstance.Eachsuchreportwouldinclude:

•Indicationsoftheuptofourreferencesignals,inpracticebeams,thatthis

specificreportrelatesto;•ThemeasuredL1-RSRPforthestrongestbeam;•Fortheremaininguptothreebeams:ThedifferencebetweenthemeasuredL1-RSRPandthemeasuredL1-RSRPofthebestbeam.

12.2.2DownlinkReceiver-SideBeamAdjustmentReceiver-sidebeamadjustmentaimsatfindingthebestreceivebeam,giventhecurrent transmit beam. To enable this, the device should once again beconfigured with a set of downlink reference signals that, in this case, aretransmittedwithin the samenetwork-sidebeam(thecurrent servingbeam).Asoutlined in Fig. 12.3, the device can then do a receiver-side beam sweep tomeasureon theconfigured referencesignals in sequenceovera setof receiverbeams.Basedonthesemeasurementsthedevicemayadjustitscurrentreceiverbeam.

FIGURE12.3 Downlinkreceiver-sidebeamadjustment.

Downlink receiver-side beam adjustment can be based on similar reportconfigurations as for transmitter-side beam adjustment. However, as thereceiver-sidebeamadjustment isdone internallywithin thedevice, there isnoreport quantity associated with receiver-side beam adjustment. According toSection8.2,thereportquantityshouldthusbesetto“None.”Toallowforanalogbeam-formingatthereceiverside,thedifferentreference

signals within the resource set should be transmitted in different symbols,allowingforthereceiver-sidebeamtosweepoverthesetofreferencesignals.Atthe same time, the device should be allowed to assume that the different

referencesignalsintheresourcesetaretransmittedusingthesamespatialfilter,in practice the same transmit beam. In general, a configured resource setincludes a“repetition” flag that indicateswhetherornot adevicecanassumethatallreferencesignalswithintheresourcesetaretransmittedusingthesamespatial filter. For a resource set to be used for downlink receiver side beamadjustment,therepetitionflagshouldthusbeset.

12.2.3UplinkBeamAdjustmentUplinkbeamadjustmentservesthesamepurposeasdownlinkbeamadjustment,that is, to retain a suitable beam pair which, in the case of uplink beamadjustment, implies a suitable transmitter beam at the device side and acorrespondingsuitablereceiverbeamatthenetworkside.Asdiscussedabove,ifbeamcorrespondencecanbeassumedandifasuitable

downlink beam pair has been established and retained, explicit uplink beammanagement is not needed. Rather, a suitable beam pair for the downlinktransmission direction can be assumed to be suitable also for the uplinkdirection.Note that theoppositewouldalsobe true, that is, if a suitablebeampair is established and retained for the uplink direction, the same beam paircould also be used in the downlink direction without the need for explicitdownlinkbeammanagement.Ifexplicituplinkbeamadjustmentisneededitcanbedoneinessentiallythe

samewayasfordownlinkbeamadjustmentwiththemaindifferencebeingthatmeasurements are done by the network based on configured SRS, rather thanCSI-RSorSSblock.

12.2.4BeamIndicationandTCIDownlink beam-forming can be done transparent to the device, that is, thedevicedoesnotneedtoknowwhatbeamisusedatthetransmitter.However, NR also supports beam indication. In practice this implies

informingthedevicethatacertainPDSCHand/orPDCCHtransmissionusesthesametransmissionbeamasaconfiguredreferencesignal(CSI-RSorSSblock).More formally, it implies informing the device that a certain PDSCH and/orPDCCHis transmittedusing thesamespatial filteras theconfigured referencesignal.Inmore detail, beam indication is based on the configuration and downlink

signalingofso-calledTransmissionConfigurationIndication(TCI)states.EachTCIstate includes,amongother things, informationaboutareferencesignal(aCSI-RS or an SS block). By associating a certain downlink transmission(PDCCHorPDSCH)withacertainTCI,thenetworkinformsthedevicethatitcanassumethatthedownlinktransmissionisdoneusingthesamespatialfilterasthereferencesignalassociatedwiththatTCI.A device can be configured with up to 64 candidate TCI states. For beam

indicationforPDCCH,asubsetoftheMconfiguredcandidatestatesisassignedbyRRCsignalingtoeachconfiguredCORESET.BymeansofMACsignaling,thenetworkcanthenmoredynamicallyindicateaspecificTCIstate,withintheper-CORESET-configured subset, to be valid. When monitoring for PDCCHwithinacertainCORESET,thedevicecanassumethatthePDCCHtransmissionuses the same spatial filter as the reference signal associated with theMAC-indicated TCI. In other words, if the device has earlier determined a suitablereceiver-sidebeamdirectionforreceptionofthereferencesignal,thedevicecanassumethatthesamebeamdirectionissuitableforreceptionofthePDCCH.For PDSCH beam indication, there are two alternatives depending on the

schedulingoffset, that is,dependingon the transmission timingof thePDSCHrelative to the corresponding PDCCH carrying scheduling information for thePDSCH.If thisschedulingoffset is larger thanNsymbols, theDCIof thescheduling

assignmentmayexplicitlyindicatetheTCIstatefor thePDSCHtransmission.1Toenablethis,thedeviceisfirstconfiguredwithasetofuptoeightTCIstatesfromtheoriginallyconfiguredsetofcandidateTCIstates.Athree-bitindicatorwithintheDCIthenindicatestheexactTCIstatevalidforthescheduledPDSCHtransmission.If the schedulingoffset is smaller or equal toN symbols, thedevice should

instead assume that the PDSCH transmission is QCL with the correspondingPDCCH transmission. In other words, the TCI state for the PDCCH stateindicated by MAC signaling should be assumed to be valid also for thecorrespondingscheduledPDSCHtransmission.The reason for limiting the fully dynamic TCI selection based on DCI

signalingtosituationswhentheschedulingoffsetislargerthanacertainvalueissimply that, for shorter schedulingoffsets, therewillnotbe sufficient time forthedevicetodecodetheTCIinformationwithintheDCIandadjustthereceiverbeamaccordinglybeforethePDSCHistobereceived.

12.3BeamRecovery

12.3BeamRecoveryIn some cases,movements in the environment or other events,may lead to acurrentlyestablishedbeampairbeingrapidlyblockedwithoutsufficienttimeforthe regular beam adjustment to adapt. The NR specification includes specificprocedurestohandlesuchbeam-failureevents,alsoreferredtoasbeam(failure)recovery.Inmanyrespects,beamfailure issimilar to theconceptofradio-link failure

(RLF) already defined for current radio-access technologies such as LTE andone could in principle utilize already-established RLF-recovery procedures torecoveralsofrombeam-failureevents.However, therearereasonstointroduceadditionalproceduresspecificallytargetingbeamfailure.

•Especiallyinthecaseofnarrowbeams,beamfailure,thatis,lossofconnectivityduetoarapiddegradationofestablishedbeampairs,canbeexpectedtooccurmorefrequentlycomparedtoRLF,whichtypicallycorrespondstoadevicemovingoutofcoveragefromthecurrentlyservingcell;

•RLFtypicallyimplieslossofcoveragetothecurrentlyservingcellinwhichcaseconnectivitymustbereestablishedtoanewcell,perhapsevenonanewcarrier.Afterbeamfailure,connectivitycanoftenbereestablishedbymeansofanewbeampairwithinthecurrentcell.Asaconsequence,recoveryfrombeamfailurecanoftenbeachievedbymeansoflower-layerfunctionality,allowingforfasterrecoverycomparedtothehigher-layermechanismsusedtorecoverfromRLF.

Ingeneral,beamfailure/recoveryconsistsofthefollowingsteps:

•Beam-failuredetection,thatis,thedevicedetectingthatabeamfailurehasoccurred;

•Candidate-beamidentification,thatis,thedevicetryingtoidentifyanewbeamor,moreexactly,anewbeampairbymeansofwhichconnectivitymayberestored;

•Recovery-requesttransmission,thatis,thedevicetransmittingabeam-recoveryrequesttothenetwork;

•Networkresponsetothebeam-recoveryrequest.

12.3.1Beam-FailureDetection

12.3.1Beam-FailureDetectionFundamentally, a beam failure is assumed to have happened when the errorprobabilityforthedownlinkcontrolchannel(PDCCH)exceedsacertainvalue.However, similar to radio-link failure, rather than actually measuring thePDCCH error probability, the device declares a beam failure based onmeasurementsofthequalityofsomereferencesignal.Thisisoftenexpressedasmeasuringahypotheticalerrorrate.Morespecifically,thedeviceshoulddeclarebeamfailurebasedonmeasuredL1-RSRPofaperiodicCSI-RSoranSSblockthat is spatiallyQCLwith the PDCCH.By default, the device should declarebeam failure based on measurement on the reference signal (CSI-RS or SSblock)associatedwiththePDCCHTCIstate.However,thereisalsoapossibilitytoexplicitlyconfigureadifferentCSI-RSonwhichtomeasureforbeam-failuredetection.Each time instant the measured L1-RSRP is below a configured value is

defined as abeam-failure instance. If the number of consecutive beam-failureinstances exceeds a configured value, the device declares a beam failure andinitiatesthebeam-failure-recoveryprocedure.

12.3.2New-Candidate-BeamIdentificationAs a first step of the beam-recovery procedure, the device tries to find a newbeampaironwhichconnectivitycanbe restored.Toenable this, thedevice isconfiguredwitha resourcesetconsistingofasetofCSI-RS,oralternativelyasetofSSblocks.Inpractice,eachofthesereferencesignalsistransmittedwithina specific downlink beam. The resource set thus corresponds to a set ofcandidatebeams.Similartonormalbeamestablishment, thedevicemeasurestheL1-RSRPon

the reference signals corresponding to the set of candidate beams. If the L1-RSRP exceeds a certain configured target, the reference signal is assumed tocorrespondtoabeambymeansofwhichconnectivitymayberestored.Itshouldbenotedthat,whendoingthis,thedevicehastoconsiderdifferentreceiver-sidebeam directions when applicable, that is, what the device determines is, inpractice,acandidatebeampair.

12.3.3DeviceRecoveryRequestandNetworkResponse

If a beam failure has been declared and a new candidate beam pair has beenidentified, the device carries out a beam-recovery request. The aim of therecovery request is to inform thenetwork that thedevicehasdetected a beamfailure.Therecoveryrequestmayalso includeinformationabout thecandidatebeamidentifiedbythedevice.Thebeam-recovery request is inessencea two-stepcontention-free random-

access request consisting of preamble transmission and random-accessresponse.2Eachreferencesignalcorrespondingtothedifferentcandidatebeamsis associated with a specific preamble configuration (RACH occasion andpreamble sequence, seeChapter 16).Given the identified beam, the preambletransmissionshouldbecarriedoutusingtheassociatedpreambleconfiguration.Furthermore, the preamble should be transmitted within the uplink beam thatcoincideswiththeidentifieddownlinkbeam.Itshouldbenotedthateachcandidatebeammaynotnecessarilybeassociated

withauniquepreambleconfiguration.Therearedifferentalternatives:

•Eachcandidatebeamisassociatedwithauniquepreambleconfiguration.Inthiscase,thenetworkcandirectlyidentifytheidentifieddownlinkbeamfromthereceivedpreamble;

•Thecandidatebeamsaredividedintogroupswhereallbeamswithinthesamegroupcorrespondtothesamepreambleconfiguration,whilebeamsofdifferentgroupscorrespondtodifferentpreambleconfigurations.Inthiscase,thereceivedpreambleonlyindicatesthegrouptowhichtheidentifieddownlinkbeambelongs;

•Allcandidatebeamsareassociatedwiththesamepreambleconfiguration.Inthiscase,thepreamblereceptiononlyindicatesthatbeamfailurehasoccurredandthatthedevicerequestsabeam-failurerecovery.

Undertheassumptionthatthecandidatebeamsareoriginatingfromthesamesite it can also be assumed that the random-access transmission is well time-aligned when arriving at the receiver. However, there may be substantialdifferences in the overall path loss for different candidate beam pairs. Theconfiguration of the beam-recovery-request transmission thus includesparametersforpowerramping(seeSection16.2).Onceadevicehascarriedoutabeam-recoveryrequest itmonitorsdownlink

foranetworkresponse.Whendoingso,thedevicemayassumethatthenetwork,

when responding to the request, is transmitting PDCCH QCL with the RSassociatedwiththecandidatebeamincludedintherequest.The monitoring for the recovery-request response starts four slots after the

transmissionoftherecoverrequest.Ifnoresponseisreceivedwithinawindowofaconfigurablesize,thedeviceretransmitstherecoveryresponseaccordingtotheconfiguredpower-rampingparameters.

1TheexactvalueofNisstillunderdiscussionin3GPP.2SeeSection16.2formoredetailsontheNRrandom-accessprocedureincludingpreamblestructure.

CHAPTER13

RetransmissionProtocols

Abstract

Retransmission functionality in three different protocol layers—hybridARQ,RLC,andPDCP—arediscussedinthischapter.ThetimingofhybridARQacknowledgmentsandtheprocessingtimeisalsodescribed.

KeywordsHARQ;hybridARQ;RLC;PDCP;CBG;softcombining;statusreports;HARQcodebook;DAI;segmentation;in-sequencedelivery

Transmissionsoverwirelesschannelsaresubject toerrors,forexample,duetovariationsinthereceivedsignalquality.Tosomedegree,suchvariationscanbecounteracted through link adaptation as will be discussed in Chapter 14.However, receiver noise and unpredictable interference variations cannot becounteracted. Therefore, virtually all wireless communication systems employsome form of Forward Error Correction (FEC), adding redundancy to thetransmittedsignalallowingthereceivertocorrecterrorsandtracingitsrootstothe pioneering work of Shannon [69]. In NR, LDPC coding is used for errorcorrectionasdiscussedinSection9.2.Despitetheerror-correctingcode,therewillbedataunitsreceivedinerror,for

example,duetoatoohighnoiseorinterferencelevel.HybridAutomaticRepeatRequest(HARQ),firstproposedbyWozencraftandHorstein[72]andrelyingona combination of error-correcting coding and retransmission of erroneous dataunits, is therefore commonly used in many modern communication systems.Data units in error despite the error correcting coding are detected by thereceiver,whichrequestsaretransmissionfromthetransmitter.InNR,threedifferentprotocollayersallofferretransmissionfunctionality—

MAC,RLC,andPDCP—asalreadymentionedintheintroductoryoverviewin

Chapter6.The reasons forhavingamultilevel retransmission structurecanbefound in the trade-off between fast and reliable feedbackof the status reports.Thehybrid-ARQmechanismintheMAClayertargetsveryfastretransmissionsand,consequently,feedbackonsuccessorfailureofthedownlinktransmissionis provided to the gNB after each received transport block (for uplinktransmission no explicit feedback needs to be transmitted as the receiver andschedulerare in thesamenode).Althoughit is inprinciplepossible toattainavery low error probability of the hybrid-ARQ feedback, it comes at a cost intransmission resources such as power. Inmany cases, a feedback error rate of0.1–1% is reasonable, which results in a hybrid-ARQ residual error rate of asimilarorder.Inmanycasesthisresidualerrorrateissufficientlylow,buttherearecaseswhenthisisnotthecase.Oneobviouscaseisservicesrequiringultra-reliable delivery of data combinedwith low latency. In such cases, either thefeedback error rate needs to be decreased and the increased cost in feedbacksignaling has to be accepted, or additional retransmissions can be performedwithout relying on feedback signaling, which comes at a decreased spectralefficiency.AlowerrorrateisnotonlyofinterestforURLLCtypeofservices,butisalso

important from a data-rate perspective.High data rateswithTCPmay requirevirtually error-free delivery of packets to the TCP layer. As an example, forsustainabledata ratesexceeding100Mbit/s,apacket-lossprobability less than10−5isrequired[65].ThereasonisthatTCPassumespacketerrorstobeduetocongestion in the network. Any packet error therefore triggers the TCPcongestion-avoidancemechanismwithacorrespondingdecreaseindatarate.Compared to the hybrid-ARQacknowledgments, theRLC status reports are

transmittedrelativelyinfrequentlyandthusthecostofobtainingareliabilityof10−5 or lower is relatively small.Hence, the combination of hybrid-ARQ andRLCattainsagoodcombinationofsmallroundtriptimeandamodestfeedbackoverhead where the two components complement each other—fastretransmissionsduetothehybrid-ARQmechanismandreliablepacketdeliveryduetotheRLC.The PDCP protocol is also capable of handling retransmissions, as well as

ensuring in-sequence delivery. PDCP-level retransmissions aremainly used inthecaseof inter-gNBhandoveras the lowerprotocols in thiscaseare flushed.Not-yet-acknowledged PDCP PDUs can be forwarded to the new gNB andtransmittedtothedevice.Inthecasethatsomeofthesewerealreadyreceivedbythedevice,thePDCPduplicatedetectionmechanismwilldiscardtheduplicates.

ThePDCPprotocolcanalsobeusedtoobtainselectiondiversitybytransmittingthesamePDUsonmultiplecarriers.ThePDCPinthereceivingendwillinthiscase remove any duplicates in case the same information was receivedsuccessfullyonmultiplecarriers.In the following sections, the principles behind the hybrid-ARQ, RLC, and

PDCPprotocolswillbediscussed inmoredetail.Note that theseprotocolsarepresentalsoinLTEwheretheytoalargeextentprovidethesamefunctionality.However,theNRversionsareenhancedtosignificantlyreducethedelays.

13.1Hybrid-ARQWithSoftCombiningThehybrid-ARQprotocolistheprimarywayofhandlingretransmissionsinNR.In case of an erroneously received packet, a retransmission is requested.However,despiteitnotbeingpossibletodecodethepacket,thereceivedsignalstill contains information, which is lost by discarding erroneously receivedpackets.Thisshortcoming isaddressedbyhybrid-ARQwithsoftcombining. Inhybrid-ARQwithsoftcombining,theerroneouslyreceivedpacketisstoredinabuffermemory and later combinedwith the retransmission to obtain a single,combined packet that is more reliable than its constituents. Decoding of theerror-correctioncodeoperatesonthecombinedsignal.AlthoughtheprotocolitselfprimarilyresidesintheMAClayer,thereisalso

physical layer functionality involved in the form of soft combining.Retransmissions of codeblock groups, that is, retransmission of a part of thetransport block, are handled by the physical layer from a specificationperspective, although it could equallywell have been described as part of theMAClayer.ThebasisfortheNRhybrid-ARQmechanismis,similarlytoLTE,astructure

with multiple stop-and-wait protocols, each operating on a single transportblock. In a stop-and-wait protocol, the transmitter stops and waits for anacknowledgmentaftereachtransmittedtransportblock.Thisisasimplescheme;the only feedback required is a single bit indicating positive or negativeacknowledgment of the transport block. However, since the transmitter stopsaftereachtransmission,thethroughputisalsolow.Therefore,multiplestop-and-wait processes operating in parallel are used such that, while waiting foracknowledgmentfromoneprocess, thetransmittercantransmitdatatoanotherhybrid-ARQprocess.This is illustrated inFig.13.1;whileprocessing thedatareceived in the first hybrid-ARQ process the receiver can continue to receive

using the second process, etc. This structure, multiple hybrid-ARQ processesoperatinginparalleltoformonehybrid-ARQentity,combinesthesimplicityofa stop-and-wait protocol while still allowing continuous transmission of data,andisusedinLTEaswellasNR.

FIGURE13.1 Multiplehybrid-ARQprocesses.

There is one hybrid-ARQ entity per carrier the receiver is connected to.Spatialmultiplexingofmorethanfourlayerstoasingledeviceinthedownlink,wheretwotransportblockscanbetransmittedinparallelonthesametransportchannel as described in Section 9.1, is supported by one hybrid-ARQ entityhaving two sets of hybrid-ARQ processes with independent hybrid-ARQacknowledgments.NRusesanasynchronoushybrid-ARQprotocolinbothdownlinkanduplink,

that is, the hybrid-ARQ process which the downlink or uplink transmissionrelates to is explicitly signaled as part of the downlink control information(DCI).LTEusesthesameschemeforthedownlinkbutnotfortheuplink,whereLTEusesasynchronousprotocol(althoughlaterLTEreleasesaddedsupportforanasynchronousprotocolaswell).Thereare several reasonswhyNRadoptedan asynchronous protocol in both directions. One reason is that synchronoushybrid-ARQ operation does not allow dynamic TDD. Another reason is thatoperation in unlicensed spectra, to be introduced in laterNR releases, ismoreefficientwithasynchronousoperationas it cannotbeguaranteed that the radioresourcesareavailableat the timeforasynchronousretransmission.Thus,NRsettledforanasynchronousschemeinbothuplinkanddownlinkwithupto16processes.Havingalargermaximumnumberofhybrid-ARQprocessesthaninLTE1 is motivated by the possibility for remote radio heads, which incurs a

certain front-haul delay, together with the shorter slot durations at highfrequencies.Itisimportantthough,thatthelargernumberofmaximumhybrid-ARQprocessesdoesnotimplyalongerroundtriptimeasnotallprocessesneedto be used, it is only an upper limit of the number of processes possible toaddress.Large transport block sizes are segmented intomultiple codeblocks prior to

coding,eachwithitsown24-bitCRC(inadditiontotheoveralltransport-blockCRC). This was discussed already in Section 9.2 and the reason is primarilycomplexity; the sizeof a codeblock is large enough togivegoodperformancewhilestillhavingareasonabledecodingcomplexity.SinceeachcodeblockhasitsownCRC,errorscanbedetectedonindividualcodeblocksaswellasontheoveralltransportblock.Arelevantquestionisifretransmissionshouldbelimitedto transport blocks or whether there are benefits of retransmitting only thecodeblocksthatareerroneouslyreceived.Fortheverylargetransportblocksizesusedtosupportdataratesofseveralgigabitspersecond,therecanbehundredsof codeblocks in a transport block. If only one or a few of them are in error,retransmitting the whole transport block results in a low spectral efficiencycomparedtoretransmittingonlytheerroneouscodeblocks.Oneexamplewhereonlysomecodeblocksare inerror isasituationwithbursty interferencewheresomeOFDM symbols are hitmore severely than others, as illustrated in Fig.13.2, for example, due to one downlink transmission preempting another asdiscussedinSection14.1.2.

FIGURE13.2 Codeblock-groupretransmission.

Tocorrectlyreceivethetransportblockfortheexampleabove,itissufficienttoretransmit theerroneouscodeblocks.At thesametime, thecontrolsignalingoverheadwouldbe too large if individual codeblocks canbe addressedby thehybrid-ARQ mechanism. Therefore, so-called codeblock groups (CBGs) aredefined.Ifper-CBGretransmissionisconfigured,feedbackisprovidedperCBG

instead of per transport block and only the erroneously received codeblockgroupsareretransmitted,whichconsumeslessresourcesthanretransmittingthewhole transport block. Two, four, six, or eight codeblock groups can beconfigured with the number of codeblocks per codeblock group varying as afunctionofthetotalnumberofcodeblocksintheinitialtransmission.Notethatthe codeblock group a codeblock belongs to is determined from the initialtransmissionanddoesnotchangebetweenthetransmissionattempts.Thisistoavoiderrorcaseswhichcouldariseifthecodeblockswererepartitionedbetweentworetransmissions.The CBG retransmissions are handled as part of the physical layer from a

specificationperspective.There isnofundamental technical reasonfor thisbutratherawaytoreducethespecificationimpactfromCBG-levelretransmissions.Aconsequenceofthisisthatitisnotpossible,inthesamehybrid-ARQprocess,to mix transmission of new CBGs belonging to another transport block withretransmissionsofCBGsbelongingtotheincorrectlyreceivedtransportblock.

13.1.1SoftCombiningAnimportantpartof thehybrid-ARQmechanismis theuseofsoftcombining,which implies that the receiver combines the received signal from multipletransmission attempts. By definition, a hybrid-ARQ retransmission mustrepresentthesamesetofinformationbitsastheoriginaltransmission.However,the set of coded bits transmitted in each retransmission may be selecteddifferentlyaslongastheyrepresentthesamesetofinformationbits.Dependingon whether the retransmitted bits are required to be identical to the originaltransmission or not, the soft combining scheme is often referred to asChasecombining, firstproposedinRef.[22],orIncrementalRedundancy (IR),whichisusedinNR.Withincrementalredundancy,eachretransmissiondoesnothavetobe identical to theoriginal transmission. Instead,multiple sets of codedbitsare generated, each representing the same set of information bits [67,71]. Therate-matchingfunctionalityofNR,describedinSection9.3,isusedtogeneratedifferentsetsofcodedbitsasafunctionoftheredundancyversionasillustratedinFig.13.3.

FIGURE13.3 Exampleofincrementalredundancy.

InadditiontoagaininaccumulatedreceivedEb/N0, incrementalredundancyalsoresultsinacodinggainforeachretransmission(untilthemothercoderateisreached). The gain with incremental redundancy compared to pure energyaccumulation (Chase combining) is larger for high initial code rates [24].Furthermore, as shown in Ref. [33], the performance gain of incrementalredundancycomparedtoChasecombiningcanalsodependontherelativepowerdifferencebetweenthetransmissionattempts.Inthediscussionsofar,ithasbeenassumedthatthereceiverhasreceivedall

the previously transmitted redundancy versions. If all redundancy versionsprovidethesameamountofinformationaboutthedatapacket,theorderoftheredundancyversions isnot critical.However, for somecode structures,not allredundancy versions are of equal importance. This is the case for the LDPCcodesused inNR; the systematicbitsareofhigher importance than theparitybits.Hence,theinitialtransmissionshouldatleastincludeallthesystematicbitsand some parity bits. In the retransmission(s), parity bits not in the initialtransmissioncanbeincluded.Thisisthebackgroundtowhysystematicbitsareinserted first in the circular buffer in Section 9.3. The starting points in thecircularbufferaredefinedsuchthatbothRV0andRV3areself-decodable,thatis, includes the systematic bits under typical scenarios.This is also the reason

RV3 is located after nine o’clock in Fig. 13.3, as this allows more of thesystematicbitstobeincludedinthetransmission.Withthedefaultorderoftheredundancy versions 0, 2, 3, 1, every second retransmission is typically self-decodable.HybridARQwithsoftcombining,regardlessofwhetherChaseorincremental

redundancyisused, leads toan implicit reductionof thedataratebymeansofretransmissions and can thus be seen as implicit link adaptation. However, incontrast to link adaptation based on explicit estimates of the instantaneouschannel conditions, hybrid-ARQ with soft combining implicitly adjusts thecodingratebasedon theresultof thedecoding. In termsofoverall throughputthiskindofimplicitlinkadaptationcanbesuperiortoexplicitlinkadaptation,asadditional redundancy is only added when needed—that is, when previoushigher-ratetransmissionswerenotpossibletodecodecorrectly.Furthermore,asitdoesnottrytopredictanychannelvariations,itworksequallywell,regardlessofthespeedatwhichtheterminalismoving.Sinceimplicitlinkadaptationcanprovide a gain in system throughput, a valid question is why explicit linkadaptation is necessary at all. One major reason for having explicit linkadaptation is the reduced delay. Although relying on implicit link adaptationalone is sufficient from a system throughput perspective, the end-user servicequalitymaynotbeacceptablefromadelayperspective.Forproperoperationofsoftcombining, the receiverneeds toknowwhen to

performsoftcombiningpriortodecodingandwhentoclearthesoftbuffer—thatis, the receiver needs to differentiate between the reception of an initialtransmission(priortowhichthesoftbuffershouldbecleared)andthereceptionofaretransmission.Similarly, thetransmittermustknowwhether toretransmiterroneouslyreceiveddataor to transmitnewdata.This ishandledby thenew-dataindicatorasdiscussedfurtherbelowfordownlinkanduplinkhybrid-ARQ,respectively.

13.1.2DownlinkHybrid-ARQInthedownlink,retransmissionsarescheduledin thesamewayasnewdata—thatis,theymayoccuratanytimeandatanarbitraryfrequencylocationwithinthedownlinkcellbandwidth.Theschedulingassignmentcontainsthenecessaryhybrid-ARQ-related control signaling—hybrid-ARQprocess number, new-dataindicator,CBGTI,andCBGFIincaseper-CBGretransmissionisconfigured,aswell as information to handle the transmission of the acknowledgment in the

uplinksuchastimingandresourceindicationinformation.Upon receiving a scheduling assignment in the DCI, the receiver tries to

decodethetransportblock,possiblyaftersoftcombiningwithpreviousattemptsasdescribedabove.Sincetransmissionsandretransmissionsarescheduledusingthe same framework in general, the device needs to know whether thetransmission is a new transmission, in which case the soft buffer should beflushed,oraretransmission,inwhichcasesoftcombiningshouldbeperformed.Therefore,anexplicitnew-dataindicatorisincludedforthescheduledtransportblock as part of the scheduling information transmitted in the downlink. Thenew-dataindicatoristoggledforanewtransportblock—thatis,itisessentiallyasingle-bit sequence number. Upon reception of a downlink schedulingassignment, thedevicechecks thenew-data indicator todeterminewhether thecurrenttransmissionshouldbesoftcombinedwiththereceiveddatacurrentlyinthe soft buffer for the hybrid-ARQ process in question, or if the soft buffershouldbecleared.Thenew-dataindicatoroperatesonthetransport-blocklevel.However,ifper-

CBGretransmissionsareconfigured,thedeviceneedstoknowwhichCBGsareretransmittedandwhetherthecorrespondingsoftbuffershouldbeflushedornot.ThisishandledthroughtwoadditionalinformationfieldspresentintheDCIincase per-CBG retransmission is configured, the CBG Transmit Indicator(CBGTI) and the CBG Flush Indicator (CBGFI). The CBGTI is a bitmapindicatingwhetheracertainCBGispresentinthedownlinktransmissionornot(see Fig. 13.4). The CBGFI is a single bit, indicating whether the CBGsindicatedbytheCBGTIshouldbeflushedorwhethersoftcombiningshouldbeperformed.

FIGURE13.4 Illustrationofper-CBGretransmission.

Theresultofthedecodingoperation—apositiveacknowledgmentinthecaseof a successful decoding and a negative acknowledgment in the case ofunsuccessful decoding—is fed back to the gNB as part of the uplink controlinformation. IfCBGretransmissionsare configured, abitmapwithonebitperCBGisfedbackinsteadofasinglebitrepresentingthewholetransportblock.

13.1.3UplinkHybrid-ARQTheuplinkuses thesameasynchronoushybrid-ARQprotocolas thedownlink.Thenecessaryhybrid-ARQ-related information—hybrid-ARQprocess number,new-dataindicator,and,ifper-CBGretransmissionisconfigured,theCBGTI—isincludedintheschedulinggrant.To differentiate between new transmissions and retransmissions of data, the

new-data indicator is used. Toggling the new-data indicator requeststransmissionofanewtransportblock,otherwisetheprevioustransportblockforthis hybrid-ARQ process should be retransmitted (inwhich case the gNB canperform soft combining). The CBGTI is used in a similar way as in thedownlink,namely to indicate thecodeblockgroups toretransmit in thecaseofper-CBGretransmission.NotethatnoCBGFIisneededintheuplinkasthesoftbufferislocatedinthegNBwhichcandecidewhethertoflushthebufferornotbasedontheschedulingdecisions.

13.1.4TimingofUplinkAcknowledgmentsIn LTE, the time from downlink data reception to transmission of theacknowledgment is fixed in the specifications. This is possible for full-duplextransmission, for example, FDD, in which case the acknowledgment istransmitted almost 3 ms after the end of data reception in LTE.2 A similarapproach can be used if the uplink–downlink allocation is semistaticallyconfiguredinthecaseofhalf-duplexoperation,forexample,semistaticTDDasinLTE.Unfortunately, this typeofschemewithpredefined timing instants forthe acknowledgments does not blend well with dynamic TDD, one of thecornerstonesofNR,asanuplinkopportunitycannotbeguaranteedafixedtimeafter the downlink transmission due to the uplink–downlink direction beingdynamically controlled by the scheduler. Coexistence with other TDDdeploymentsinthesamefrequencybandmayalsoimposerestrictionswhenitisdesirable,orpossible,totransmitintheuplink.Furthermore,evenifitwouldbe

possible, it may not be desirable to change the transmission direction fromdownlink touplink ineachslotas thiswould increase theswitchingoverhead.Consequently,amoreflexibleschemecapableofdynamicallycontrollingwhentheacknowledgmentistransmittedisadoptedinNR.The hybrid-ARQ timing field in the downlink DCI is used to control the

transmissiontimingoftheacknowledgmentintheuplink.Thisthree-bitfieldisusedasan indexintoanRRC-configuredtableprovidinginformationonwhenthehybrid-ARQacknowledgmentshouldbetransmittedrelativetothereceptionof the PDSCH (see Fig. 13.5). In this particular example, three slots arescheduled in the downlink before an acknowledgment is transmitted in theuplink. Ineachdownlinkassignment,differentacknowledgment timing indiceshavebeenused,which incombinationwith theRRC-configured tableresult inall three slots being acknowledged at the same time (multiplexing of theseacknowledgmentsinthesameslotisdiscussedbelow).

FIGURE13.5 Determiningtheacknowledgmenttiming.

Furthermore,NRisdesignedwithvery lowlatency inmindandis thereforecapable of transmitting the acknowledgmentmuch sooner after the end of thedownlinkdatareceptionthanthecorrespondingLTEtimingrelation.Alldevicessupport the baseline processing times listed in Table 13.1, with even fasterprocessingoptionallysupportedbysomedevices.Thecapabilityisreportedpersubcarrierspacing.Onepartoftheprocessingtimeisconstantinsymbolsacrossdifferent subcarrier spacing, that is, the time in microseconds scales with thesubcarrier spacing, but there is also a part of the processing time fixed inmicrosecondsandindependentofthesubcarrierspacing.Hence,theprocessing

times listed in the table are not directly proportional to the subcarrier spacingalthough there is a dependency. There is also a dependency on the referencesignalconfiguration;ifthedeviceisconfiguredwithadditionalreferencesignaloccasions later in the slot, the device cannot start the processing until at leastsomeof these reference signals havebeen received and theoverall processingtime is longer. Nevertheless, the processing is much faster than thecorrespondingLTEcaseasaresultofstressingtheimportanceoflowlatencyintheNRdesign.

Table13.1

For proper transmission of the acknowledgment it is not sufficient for thedevice to know when to transmit, which is obtained from the timing fielddiscussed above, but alsowhere in the resource domain (frequency resourcesand, for somePUCCHformats, thecodedomain). In theoriginalLTEdesign,this is primarily obtained from the location of the PDCCH scheduling thetransmission. For NR with its flexibility in the transmission timing of theacknowledgment, sucha scheme isnot sufficient. In the case that twodevicesare instructed to transmit their acknowledgment at the same time even if theywerescheduledatdifferent timeinstants, it isnecessary toprovide thedeviceswithseparateresources.ThisishandledthroughthePUCCHresourceindicator,whichisathree-bitindexselectingoneofeightRRC-configuredresourcesetsasdescribedinSection10.2.7.

13.1.5MultiplexingofHybrid-ARQAcknowledgmentsIn theprevioussection, the timingof thehybrid-ARQacknowledgments in theexamplewassuchthatmultipletransportblocksneedtobeacknowledgedatthesame time. Other examples where multiple acknowledgments need to betransmittedintheuplinkatthesametimearecarrieraggregationandper-CBGretransmissions. NR therefore supports multiplexing of acknowledgments for

multiple transport blocks received by a device into one multi-bitacknowledgmentmessage.Themultiplebits canbemultiplexedusing either asemistatic codebookor adynamiccodebookwithRRCconfiguration selectingbetweenthetwo.The semistatic codebook can be viewed as a matrix consisting of a time-

domain dimension and a component-carrier (or CBG or MIMO layer)dimension, both of which are semistatically configured. The size in the timedomain isgivenby themaximumandminimumhybrid-ARQacknowledgmenttimingsconfiguredinTable13.1,andthesizeinthecarrierdomainisgivenbythe number of simultaneous transport blocks (or CBGs) across all componentcarriers. An example is provided in Fig. 13.6, where the acknowledgmenttimingsareone, two, three,and four, respectively,and threecarriers,onewithtwotransportblocks,onewithonetransportblock,andonewithfourCBGs,areconfigured.Sincethecodebooksizeisfixed,thenumberofbitstotransmitinahybrid-ARQreport isknown(4·7=28bits in theexample inFig.13.6)and theappropriateformatfortheuplinkcontrolsignalingcanbeselected.Eachentryinthe matrix represents the decoding outcome, positive or negativeacknowledgment, of the corresponding transmission. Not all transmissionopportunitiespossiblewiththecodebookareusedinthisexampleandforentriesinthematrixwithoutacorrespondingtransmission,anegativeacknowledgmentis transmitted. This provides robustness; in the case of missed downlinkassignment a negative acknowledgment is provided to the gNB, which canretransmitthemissingtransportblock(orCBG).

FIGURE13.6 Exampleofsemistatichybrid-ARQacknowledgmentcodebook.

Onedrawbackwiththesemistaticcodebookis thepotentiallylargesizeofahybrid-ARQ report. For a small number of component carriers and no CBGretransmissions, this is lessofaproblem,but ifa largenumberofcarriersandcodeblock groups are configured out of which only a small number issimultaneouslyused,thismaybecomemoreofanissue.To address the drawback of a potentially large semistatic codebook size in

somescenarios,NRalsosupportsadynamiccodebook.Infact,thisisthedefaultcodebook used unless the system is configured otherwise. With a dynamiccodebook, only the acknowledgment information for the scheduled carriers3 isincludedinthereport,insteadofallcarriers,scheduledornot,asisthecasewithasemistaticcodebook.Hence,thesizeofthecodebook(thematrixinFig.13.6)is dynamically varying as a function of the number of scheduled carriers. Inessence,onlytheboldentriesintheexampleinFig.13.6wouldbeincludedinthehybrid-ARQreportandthenon-boldentrieswithagraybackground(whichcorrespondtonon-scheduledcarriers)wouldbeomitted.Thisreducesthesizeoftheacknowledgmentmessage.Adynamiccodebookwouldbestraightforwardiftherewerenoerrorsinthe

downlinkcontrolsignaling.However,inthepresenceofanerrorinthedownlinkcontrolsignaling,thedeviceandgNBmayhavedifferentunderstandingonthenumberof scheduledcarriers,whichwould lead to an incorrect codebook sizeand possibly corrupt the feedback report for all carriers, and not only for theones for which the downlink controls signaling was missed. Assume, as anexample, that the device was scheduled for downlink transmission in twosubsequent slots butmissed the PDCCHand hence scheduling assignment forthe first slot. In response the devicewill transmit an acknowledgment for thesecondslotonly,whilethegNBtriestoreceiveacknowledgmentsfortwoslots,leadingtoamismatch.To handle these error cases,NRuses thedownlink assignment index (DAI)

included in the DCI containing the downlink assignment. The DAI field isfurther split into two parts, a counter DAI (cDAI) and, in the case of carrieraggregation,atotalDAI(tDAI).ThecounterDAIincludedintheDCIindicatesthe number of scheduled downlink transmissions up to the point theDCIwasreceived in a carrier first, time secondmanner. The totalDAI included in theDCIindicatesthetotalnumberofdownlinktransmissionsacrossallcarriersupto thispoint in time, that is, thehighestcDAIat thecurrentpoint in time (seeFig.13.7foranexample).ThecounterDAIandtotalDAIarerepresentedwithdecimalnumberswithno limitation; inpractice twobitsareusedforeachandthenumberingwillwraparound,thatis,whatissignaledisthenumbersinthefiguremodulo four.As seen in this example, the dynamic codebook needs toaccountfor17acknowledgments(numbered0–16).Thiscanbecomparedwiththesemistaticcodebookwhichwouldrequire28entriesregardlessofthenumberoftransmissions.

FIGURE13.7 Exampleofdynamichybrid-ARQacknowledgmentcodebook.

Furthermore, in this example,one transmissiononcomponentcarrier five islost.Without the DAImechanism, this would result inmisaligned codebooksbetweenthedeviceandthegNB.However,aslongasthedevicereceivesatleastonecomponentcarrier,itknowsthevalueofthetotalDAIandhencethesizeofthecodebookatthispointintime.Furthermore,bycheckingthevaluesreceivedfor thecounterDAI, itcanconcludewhichcomponentcarrierwasmissedandthat a negative acknowledgment should be assumed in the codebook for thisposition.InthecasethatCBGretransmissionisconfiguredforsomeofthecarriers,the

dynamiccodebookissplitintotwoparts,oneforthenon-CBGcarriersandonefor the CBG carriers. Each codebook is handled according to the principlesoutlinedabove.ThereasonforthesplitisthatfortheCBGcarriers,thedeviceneeds to generate feedback for each of these carriers according to the largestCBGconfiguration.

13.2RLC

Theradio-linkcontrol(RLC)protocoltakesdataintheformofRLCSDUsfromPDCP and delivers them to the corresponding RLC entity in the receiver byusingfunctionalityinMACandphysicallayers.TherelationbetweenRLCandMAC,includingmultiplexingofmultiplelogicalchannelsintoasingletransportchannel,isillustratedinFig.13.8.

FIGURE13.8 MACandRLC.

ThereisoneRLCentityperlogicalchannelconfiguredforadevicewiththeRLCentitybeingresponsibleforoneormoreof:

•SegmentationofRLCSDUs;•Duplicateremoval;and•RLCretransmission.

UnlikeLTE,thereisnosupportforconcatenationorin-sequencedeliveryintheRLCprotocol.Thisisadeliberatechoicedonetoreducetheoveralllatencyas discussed further in the following sections. It has also impacted the headerdesign.Also,notethatthefactthatthereisoneRLCentityperlogicalchanneland one hybrid-ARQ entity per cell (component carrier) implies that RLCretransmissions can occur on a different cell (component carrier) than theoriginal transmission.This is not the case for the hybrid-ARQprotocolwhereretransmissions are bound to the same component carrier as the originaltransmission.Different services have different requirements; for some services (for

example, transfer of a large file), error-free delivery of data is important,whereasforotherapplications(forexample,streamingservices),asmallamount

ofmissing packets is not a problem. The RLC can therefore operate in threedifferentmodes,dependingontherequirementsfromtheapplication:

•Transparentmode(TM),wheretheRLCiscompletelytransparentandisessentiallybypassed.Noretransmissions,noduplicatedetection,andnosegmentation/reassemblytakeplace.Thisconfigurationisusedforcontrol-planebroadcastchannelssuchasBCCH,CCCH,andPCCH,wheretheinformationshouldreachmultipleusers.Thesizeofthesemessagesisselectedsuchthatallintendeddevicesarereachedwithahighprobabilityandhencethereisneitherneedforsegmentationtohandlevaryingchannelconditions,norretransmissionstoprovideerror-freedatatransmission.Furthermore,retransmissionsarenotfeasibleforthesechannelsasthereisnopossibilityforthedevicetofeedbackstatusreportsasnouplinkhasbeenestablished.

•Unacknowledgedmode(UM)supportssegmentationbutnotretransmissions.Thismodeisusedwhenerror-freedeliveryisnotrequired,forexample,voice-over-IP.

•Acknowledgedmode(AM)isthemainmodeofoperationfortheDL-SCHandUL-SCH.Segmentation,duplicateremoval,andretransmissionsoferroneousdataareallsupported.

In the following sections, the operation of the RLC protocol is described,focusingonacknowledgedmode.

13.2.1SequenceNumberingandSegmentationInunacknowledgedandacknowledgedmodes,asequencenumberisattachedtoeachincomingSDUusing6or12bitsforunacknowledgedmodeand12or18bitsforacknowledgedmode.ThesequencenumberisincludedintheRLCPDUheader in Fig. 13.9. In the case of a non-segmented SDU, the operation isstraightforward;theRLCPDUissimplytheRLCSDUwithaheaderattached.NotethatthisallowstheRLCPDUstobegeneratedinadvanceastheheader,intheabsenceofsegmentation,doesnotdependonthescheduledtransportblocksize. This is beneficial from a latency perspective and the reason the headerstructureischangedcomparedtotheoneusedinLTE.

FIGURE13.9 GenerationofRLCPDUsfromRLCSDUs(acknowledgedmodeassumedfortheheaderstructure).

However,dependingonthetransport-blocksizeafterMACmultiplexing,thesizeof(thelast)oftheRLCPDUsinatransportblockmaynotmatchtheRLCSDUsize.Tohandlethis,anSDUcanbesegmentedintomultiplesegments.Ifnosegmentationtakesplace,paddingwouldneedtobeusedinstead,leadingtodegraded spectral efficiency.Hence, dynamically varying the number of RLCPDUsused to fill the transportblock, togetherwith segmentation toadjust thesizeofthelastRLCPDU,ensuresthetransportblockisefficientlyutilized.Segmentationissimple;thelastpreprocessedRLCSDUcanbesplitintotwo

segments,theheaderofthefirstsegmentisupdated,andtothesecondsegmentanewheaderisadded(whichisnottimecriticalasit isnotbeingtransmittedinthe current transport block). Each SDU segment carries the same sequencenumberas theoriginalunsegmentedSDUand this sequencenumber ispartoftheRLCheader.TodistinguishwhetherthePDUcontainsacompleteSDUorasegment,asegmentation information (SI) field isalsopartof theRLCheader,indicatingwhether thePDUisacompleteSDU, the first segmentof theSDU,thelastsegmentoftheSDU,orasegmentbetweenthefirstandlastsegmentsoftheSDU.Furthermore, in thecaseofasegmentedSDU,a16-bitsegmentationoffset(SO)isincludedinallsegmentsexceptthefirstoneandusedtoindicatewhichbyteoftheSDUthesegmentrepresents.Thereisalsoapollbit(P)intheheader used to request a status report for acknowledged mode as describedfurther below, and adata/control indicator, indicatingwhether theRLCPDUcontainsdatato/fromalogicalchannelorcontrolinformationrequiredforRLCoperation.The header structure above holds for acknowledged mode. The header for

unacknowledgedmode issimilarbutdoesnot includeeither thepollbitor thedata/controlindicator.Furthermore,thesequencenumberisincludedinthecaseofsegmentationonly.InLTE,theRLCcanalsoperformconcatenationofRLCSDUsintoasingle

PDU.However,thisfunctionalityisnotpresentinNRinordertoreducelatency.

If concatenationwould be supported, anRLCPDUcannot be assembleduntiltheuplinkgrantisreceivedasthescheduledtransport-blocksizeisnotknowninadvance. Consequently, the uplink grant must be received well in advance toallowsufficientprocessingtimeinthedevice.Withoutconcatenation,theRLCPDUs can be assembled in advance, prior to receiving the uplink grant, andtherebyreducingtheprocessingtimerequiredbetweenreceivinganuplinkgrantandtheactualuplinktransmission.

13.2.2AcknowledgedModeandRLCRetransmissionsRetransmissionofmissingPDUsisoneofthemainfunctionalitiesoftheRLCinacknowledgedmode.Althoughmostoftheerrorscanbehandledbythehybrid-ARQprotocol,thereare,asdiscussedatthebeginningofthechapter,benefitsofhaving a second-level retransmission mechanism as a complement. Byinspecting the sequencenumbers of the receivedPDUs,missingPDUs canbedetectedandaretransmissionrequestedfromthetransmittingside.RLCacknowledgedmodeinNRissimilartoitscounterpartinLTEwithone

exception—reordering to ensure in-sequence delivery is not supported in NR.Removing in-sequence delivery from the RLC also helps reduce the overalllatency as later packets do not have to wait for retransmission of an earliermissing packet before being delivered to higher layers, but can be forwardedimmediately. This also leads to reduced buffering requirements positivelyimpactingtheamountofmemoryusedforRLCbuffering.InLTE,whichdoessupport in-sequence delivery from theRLCprotocol, anRLCSDU cannot beforwarded to higher layers unless all previous SDUs have been correctlyreceived.AsinglemissingSDU,forexample,duetoamomentaryinterferenceburst,canthusblockdeliveryofsubsequentSDUsforquitesometime,evenifthoseSDUswouldbeuseful to theapplication,apropertywhich isclearlynotdesirableinasystemtargetingverylowlatency.In acknowledged mode, the RLC entity is bidirectional—that is, data may

flow in bothdirectionsbetween the twopeer entities.This is necessary as thereceptionofPDUsneedstobeacknowledgedbacktotheentitythattransmittedthosePDUs.InformationaboutmissingPDUsisprovidedbythereceivingendtothetransmittingendintheformofso-calledstatusreports.Statusreportscaneither be transmitted autonomously by the receiver or requested by thetransmitter. To keep track of the PDUs in transit, the sequence number in the

headerisused.Both RLC entities maintain two windows in acknowledged mode, the

transmission and reception windows, respectively. Only PDUs in thetransmissionwindowareeligiblefortransmission;PDUswithsequencenumberbelowthestartofthewindowhavealreadybeenacknowledgedbythereceivingRLC.Similarly,thereceiveronlyacceptsPDUswithsequencenumberswithinthe receptionwindow. The receiver also discards any duplicate PDUs as onlyonecopyofeachSDUshouldbedeliveredtohigherlayers.The operation of the RLC with respect to retransmissions is perhaps best

understood by the simple example in Fig. 13.10, where two RLC entities areillustrated, one in the transmitting node and one in the receiving node.Whenoperatinginacknowledgedmode,asassumedbelow,eachRLCentityhasbothtransmitter and receiver functionality, but in this example only one of thedirectionsisdiscussedastheotherdirectionisidentical.Intheexample,PDUsnumberedfromnton+4areawaitingtransmissioninthetransmissionbuffer.Attime t0, PDUs with sequence number up to and including n have beentransmittedandcorrectlyreceived,butonlyPDUsuptoandincludingn−1havebeen acknowledged by the receiver. As seen in the figure, the transmissionwindowstartsfromn, thefirstnot-yet-acknowledgedPDU,whilethereceptionwindowstartsfromn+1,thenextPDUexpectedtobereceived.UponreceptionofaPDUn,theSDUisreassembledanddeliveredtohigherlayers,thatis,thePDCP. For a PDU containing a complete SDU, reassembly is simply headerremoval,butinthecaseofasegmentedSDU,theSDUcannotbedelivereduntilPDUscarryingallthesegmentshavebeenreceived.

FIGURE13.10 SDUdeliveryinacknowledgedmode.

ThetransmissionofPDUscontinuesand,attimet1,PDUsn+1andn+2havebeentransmittedbut,atthereceivingend,onlyPDUn+2hasarrived.AssoonasacompleteSDUisreceived,itisdeliveredtohigherlayers,hencePDUn+2is

forwarded to the PDCP layerwithout waiting for themissing PDU n+1.Onereason PDU n+1 is missing could be that it is under retransmission by thehybrid-ARQprotocolandthereforehasnotyetbeendeliveredfromthehybrid-ARQtotheRLC.Thetransmissionwindowremainsunchangedcomparedtothepreviousfigure,asnoneofthePDUsnandhigherhavebeenacknowledgedbythe receiver. Hence, any of these PDUs may need to be retransmitted as thetransmitterisnotawareofwhethertheyhavebeenreceivedcorrectlyornot.ThereceptionwindowisnotupdatedwhenPDUn+2arrives,thereasonbeing

themissingPDUn+1.Insteadthereceiverstartsatimer,thet-Reassemblytimer.IfthemissingPDUn+1isnotreceivedbeforethetimerexpires,aretransmissionis requested. Fortunately, in this example, the missing PDU arrives from thehybrid-ARQprotocolattimet2,beforethetimerexpires.ThereceptionwindowisadvancedandthereassemblytimerisstoppedasthemissingPDUhasarrived.PDUn+1isdeliveredforreassemblyintoSDUn+1.Duplicate detection is also the responsibility of the RLC, using the same

sequencenumberasusedforretransmissionhandling.IfPDUn+2arrivesagain(andiswithinthereceptionwindow),despiteithavingalreadybeenreceived,itisdiscarded.The transmissioncontinueswithPDUsn+3,n+4,andn+5,asshowninFig.

13.11.At time t3,PDUsup ton+5havebeen transmitted.OnlyPDUn+5hasarrived and PDUs n+3 and n+4 are missing. Similar to the case above, thiscausesthereassemblytimertostart.However, in thisexamplenoPDUsarrivepriortotheexpirationofthetimer.Theexpirationofthetimerattimet4triggersthe receiver to send a control PDU containing a status report, indicating themissingPDUs, to its peer entity.ControlPDUshavehigher priority thandataPDUs to avoid the status reports being unnecessarily delayed and negativelyimpactingtheretransmissiondelay.Uponreceiptofthestatusreportattimet5,thetransmitterknowsthatPDUsupton+2havebeenreceivedcorrectlyandthetransmission window is advanced. The missing PDUs n+3 and n+4 areretransmittedand,thistime,correctlyreceived.

FIGURE13.11 RetransmissionofmissingPDUs.

Finally, at time t6, all PDUs, including the retransmissions, have beendeliveredbythetransmitterandsuccessfullyreceived.Asn+5wasthelastPDUin the transmission buffer, the transmitter requests a status report from thereceiver by setting a flag in the header of the last RLC data PDU. UponreceptionofthePDUwiththeflagset,thereceiverwillrespondbytransmittingthe requested status report, acknowledging all PDUs up to and includingn+5.Reception of the status report by the transmitter causes all the PDUs to bedeclaredascorrectlyreceivedandthetransmissionwindowisadvanced.Status reports can, as mentioned earlier, be triggered for multiple reasons.

However,tocontroltheamountofstatusreportsandtoavoidfloodingthereturnlink with an excessive number of status reports, it is possible to use a statusprohibittimer.Withsuchatimer,statusreportscannotbetransmittedmoreoftenthanoncepertimeintervalasdeterminedbythetimer.The example above basically assumed eachPDU carrying a non-segmented

SDU. Segmented SDUs are handled the same way, but an SDU cannot bedeliveredtothePDCPprotocoluntilallthesegmentshavebeenreceived.Statusreports and retransmissions operate on individual segments; only the missingsegmentofaPDUneedstoberetransmitted.In thecaseofaretransmission,allRLCPDUsmaynotfit into the transport

blocksizescheduledfortheRLCretransmission.Resegmentationfollowingthesameprincipleastheoriginalsegmentationisusedinthiscase.

13.3PDCPThePacketDataConvergenceProtocol(PDCP)isresponsiblefor:

•Headercompression;•Cipheringandintegrityprotection;•Routingandduplicationforsplitbearers;and

•Retransmission,reordering,andSDUdiscard.

Header compression,with the corresponding decompression functionality atthe receiver side, can be configured and serves the purpose of reducing thenumber of bits transmitted over the radio interface. Especially for smallpayloads, such as voice-over-IP and TCP acknowledgments, the size of anuncompressedIPheaderisinthesamerangeasthepayloaditself,40bytesforIP v4 and 60 bytes for IP v6, and can account for around 60% of the totalnumberofbitssent.Compressingthisheadertoacoupleofbytescanthereforeincrease the spectral efficiency by a large amount. The header compressionscheme in NR is based on Robust Header Compression (ROHC) [38], astandardizedheader-compressionframeworkalsousedforseveralothermobile-communication technologies, for example, LTE. Multiple compressionalgorithms,denotedprofiles,aredefined,eachspecifictotheparticularnetworklayerandtransportlayerprotocolcombinationsuchasTCP/IPandRTP/UDP/IP.HeadercompressionisdevelopedtocompressIPpackets.HenceitisappliedtothedatapartonlyandnottheSDAPheader(ifpresent).Integrityprotectionensuresthatthedataoriginatefromthecorrectsourceand

ciphering protects against eavesdropping. PDCP is responsible for both thesefunctions,ifconfigured.Integrityprotectionandcipheringareusedforboththedata plane and the control plane and applied to the payload only and not thePDCPcontrolPDUsorSDAPheaders.For dual connectivity and split bearers (see Chapter 6, for amore in-depth

discussion on dual connectivity), PDCP can provide routing and duplicationfunctionality.Withdualconnectivity,someof theradiobearersarehandledbythe master cell group, while others are handled by the secondary cell group.Thereisalsoapossibilitytosplitabeareracrossbothcellgroups.Theroutingfunctionality of the PDCP is responsible for routing the data flows for thedifferent bearers to the correct cell groups, as well as handling flow controlbetweenthecentralunit(gNB-CU)anddistributedunit(gNB-DU)inthecaseofasplitgNB.Duplication implies that the same data can be transmitted on two separate

logical channelswhere configuration ensures that the two logical channels aremapped to different carriers. This can be used in combination with carrieraggregation or dual connectivity to provide additional diversity. If multiplecarriersareused to transmit thesamedata, the likelihood that receptionof thedataon at least one carrier is correct increases. Ifmultiple copiesof the same

SDUarereceived,thereceiving-sidePDCHdiscardstheduplicates.Thisresultsinselectiondiversitywhichcanbeessentialtoprovidingveryhighreliability.Retransmissionfunctionality,includingthepossibilityforreorderingtoensure

in-sequencedelivery, isalsopartof thePDCP.Arelevantquestion iswhy thePDCP is capable of retransmissions when there are two other retransmissionfunctions in lower layers, theRLCARQand theMAChybrid-ARQfunctions.Onereason is inter-gNBhandover.Uponhandover,undelivereddownlinkdatapacketswillbeforwardedby thePDCPfromtheoldgNBto thenewgNB.Inthiscase,anewRLCentity (andhybrid-ARQentity) isestablished in thenewgNBandtheRLCstatusislost.ThePDCPretransmissionfunctionalityensuresthat no packets are lost as a result of this handover. In the uplink, the PDCPentity in the device will handle retransmission of all uplink packets not yetdeliveredtothegNBasthehybrid-ARQbuffersareflusheduponhandover.In-sequencedeliveryisnotensuredbytheRLCtoreducetheoveralllatency.

Inmanycases,rapiddeliveryofthepacketsismoreimportantthanguaranteedin-sequencedelivery.However, if in-sequencedelivery is important, thePDCPcanbeconfiguredtoprovidethis.Retransmissionand in-sequencedelivery, ifconfigured, is jointlyhandled in

the same protocol,which operates similarly to theRLCARQprotocol exceptthat no segmentation is supported. A so-called count value is associated witheachSDU,wherethecountisacombinationofthePDCPsequencenumberandthe hyper-frame number. The count value is used to identify lost SDUs andrequest retransmission, as well as reorder received SDUs before delivery toupperlayersisreorderingisconfigured.ReorderingbasicallybuffersareceivedSDU and does not forward it to higher layers until all lower-numbered SDUshave been delivered. Referring to Fig. 13.10, this would be similar to notdelivering SDU n+2 until n+1 has been successfully received and delivered.There is also a possibility to configure a discard timer for each PDCP SDU;whenthetimerexpiresthecorrespondingSDUisdiscardedandnottransmitted.

1InLTE,eightprocessesareusedforFDDandupto15processesforTDD,dependingontheuplink–downlinkconfiguration.2Thetimedependsonthetimingadvancevalue.Forthelargestpossibletimingadvance,thetimeis2.3msinLTE.3Thedescriptionhereusestheterm"carrier"butthesameprincipleisequallyapplicabletoper-CBGretransmissionormultipletransportblocksinthecaseof

MIMOand"transmissioninstant"isamoregenericterm,albeitthedescriptionwouldbehardertoread.

CHAPTER14

Scheduling

Abstract

NR is essentially a scheduled system where the scheduler in the gNBcontrols downlink and uplink transmissions. This chapter describes thedetails around dynamic as well as semipersistent scheduling, includingassociated functionality such as buffer-status report and power-headroomreports.

KeywordsScheduling;dynamicscheduling;semipersistentscheduling;preemption;buffer-statusreport;power-headroomreport;schedulingrequest;discontinuousreception;DRX

NR is essentially a scheduled system, implying that the scheduler determineswhenandtowhichdevicesthetime,frequency,andspatialresourcesshouldbeassigned and what transmission parameters, including data rate, to use.Schedulingcanbeeitherdynamicorsemistatic.Dynamicschedulingisthebasicmode-of-operation where the scheduler for each time interval, for example, aslot, determines which devices are to transmit and receive. Since schedulingdecisions are taken frequently, it is possible to follow rapid variations in thetraffic demand and radio-channel quality, thereby efficiently exploiting theavailable resources. Semistatic scheduling implies that the transmissionparametersareprovidedtothedevicesinadvanceandnotonadynamicbasis.Inthefollowing,dynamicdownlinkanduplinkschedulingwillbediscussed,

including bandwidth adaptation, followed by a discussion on non-dynamicscheduling and finally a discussion on discontinuous reception as a way toreducedevicepowerconsumption.

14.1DynamicDownlinkScheduling

14.1DynamicDownlinkSchedulingFluctuations in the received signal quality due to small-scale aswell as large-scale variations in the environment are an inherent part in any wirelesscommunicationsystem.Historically,suchvariationswereseenasaproblem,butthe development of channel-dependent scheduling, where transmissions to anindividual device take place when the radio-channel conditions are favorable,allowsthesevariationstobeexploited.Givenasufficientnumberofdevicesinthecellhavingdatatotransfer,thereisahighlikelihoodofatleastsomedeviceshaving favorable channel conditions at each point in time and able to use acorrespondinglyhighdatarate.Thegainobtainedbytransmittingtouserswithfavorableradio-linkconditionsiscommonlyknownasmultiuserdiversity.Thelarger the channel variations and the larger the number of users in a cell, thelarger the multiuser diversity gain. Channel-dependent scheduling wasintroducedinthelaterversionsofthe3GstandardknownasHSPA[21]andisalsousedinLTEaswellasNR.There is a rich literature in the field of scheduling and how to exploit

variationsinthetimeandfrequencydomains(see,forexample,Ref.[28]andthereferencestherein).Lately,therehasalsobeenalargeinterestinvariousmassivemultiuserMIMO schemes [55]where a large number of antenna elements areused to create very narrow “beams,” or, expressed differently, isolate thedifferent users in the spatial domain. It can be shown that, under certainconditions, theuseofalargenumberofantennasresultsinaneffectknownas“channel hardening.” In essence, the rapid fluctuations of the radio-channelqualitydisappear,simplifyingthetime–frequencypartoftheschedulingproblematthecostofamorecomplicatedhandlingofthespatialdomain.InNR,thedownlinkscheduler isresponsiblefordynamicallycontrollingthe

device(s) to transmit to. Each of the scheduled devices is provided with ascheduling assignment including information on the set of time–frequencyresourcesuponwhichthedevice’sDL-SCH1istransmitted,themodulation-and-codingscheme,hybrid-ARQ-related information,andmulti-antennaparametersasoutlinedinChapter10.InmostcasestheschedulingassignmentistransmittedjustbeforethedataonthePDSCH,butthetiminginformationintheschedulingassignmentcanalsoscheduleinOFDMsymbolslaterintheslotorinlaterslots.One use for this is bandwidth adaptation as discussed below. Changing thebandwidthpartmaytakesometimeandhencedatatransmissionmaynotoccurinthesameslotasthecontrolsignalingwasreceivedin.It is important to understand that NR does not standardize the scheduling

behavior.Onlyasetofsupportingmechanismsarestandardizedontopofwhicha vendor-specific scheduling strategy is implemented. The information neededby the schedulerdependson the specific schedulingstrategy implemented,butmostschedulersneedinformationaboutatleast:

•Channelconditionsatthedevice,includingspatial-domainproperties;•Bufferstatusofthedifferentdataflows;and•Prioritiesofthedifferentdataflows,includingtheamountofdatapendingretransmission.

Additionally, the interference situation in neighboring cells can be useful ifsomeformofinterferencecoordinationisimplemented.Information about the channel conditions at the device can be obtained in

several ways. In principle, the gNB can use any information available, buttypically theCSI reports from thedeviceareusedasdiscussed inSection8.1.There is awide range ofCSI reports that can be configuredwhere the devicereports the channel quality in the time, frequency, and spatial domains. Theamountofcorrelationbetweenthespatialchannelstodifferentdevicesisalsoofinterest to estimate the degree of spatial isolation between two devices in thecase they are candidates for being scheduled on the same time–frequencyresourcesusingmultiuserMIMO.UplinksoundingusingSRStransmissioncan,together with assumptions on channel reciprocity, also be used to assess thedownlink channel quality. Various other quantities can be used as well, forexample,signal-strengthmeasurementsfordifferentbeamcandidates.Thebufferstatusandtrafficprioritiesareeasilyobtainedinthedownlinkcase

as the scheduler and the transmission buffers reside in the same node.Prioritization of different traffic flows is purely implementation-specific, butretransmissionsare typicallyprioritizedover transmissionofnewdata, at leastfordataflowsofthesamepriority.GiventhatNRisdesignedtohandleamuchwiderrangeoftraffictypesandapplicationsthanprevioustechnologies,suchasLTE, priority handling in the scheduler can in many cases be even moreemphasized than in the past. In addition to selecting data from different dataflows, thescheduleralsohasthepossibilitytoselect thetransmissionduration.For example, for a latency-critical service with its data mapped to a certainlogical channel, it may be advantageous to select a transmission durationcorrespondingtoafractionofaslot,whileforanotherserviceonanotherlogicalchannel, a more traditional approach of using the full slot duration for

transmissionmightbeabetterchoice. Itmayalsobe thecase that, for latencyreasonsandshortageofresources,anurgenttransmissionusingasmallnumberoftransmissionsneedstopreemptanalreadyongoingtransmissionusingthefullslot.Inthiscase,thepreemptedtransmissionislikelytobecorruptedandrequirearetransmission,butthismaybeacceptablegiventheveryhighpriorityofthelow-latencytransmission.TherearealsosomemechanismsinNRwhichcanbeusedtomitigatethis,asdiscussedinSection14.1.2.Differentdownlinkschedulersmaycoordinate theirdecisions to increasethe

overall performance, for example, by avoiding transmission on a certainfrequencyrangeinonecelltoreducetheinterferencetowardsanothercell.Inthecaseof(dynamic)TDD,thedifferentcellscanalsocoordinatethetransmissiondirection,uplinkordownlink,betweenthecellstoavoiddetrimentalinterferencesituations.Suchcoordinationcantakeplaceondifferenttimescales.Typically,thecoordination isdoneat a slower rate than the schedulingdecisions ineachcell as the requirements on the backhaul connecting different gNBs otherwisewouldbetoohigh.In the case of carrier aggregation, the scheduling decisions are taken per

carrier and the scheduling assignments are transmitted separately for eachcarrier, that is, a device scheduled to receive data from multiple carrierssimultaneouslyreceivesmultiplePDCCHs.APDCCHreceivedcaneitherpointto thesamecarrier,knownas self-scheduling,or toanothercarrier, commonlyreferredtoascross-carrierscheduling(seeFig.14.1).Inthecaseofcross-carrierschedulingofacarrierwithadifferentnumerologythantheoneuponwhichthePDCCH was transmitted, timing offsets in the scheduling assignment, forexample, which slot the assignment relates to, are interpreted in the PDSCHnumerology(andnotthePDCCHnumerology).

FIGURE14.1 Self-schedulingandcross-carrierscheduling.

Theschedulingdecisions for thedifferentcarriersarenot taken in isolation.

Rather, the scheduling of the different carriers for a given device needs to becoordinated.Forexample,ifacertainpieceofdataisscheduledfortransmissionon one carrier, the same piece of data should normally not be scheduled onanothercarrieraswell.However,itisinprinciplepossibletoschedulethesamedataonmultiplecarriers.Thiscanbeusedtoincreasereliability;withmultiplecarriers transmitting the samedata the likelihoodof successful receptionon atleastonecarrier is increased.At the receiver theRLC(orPDCP) layercanbeconfiguredtoremoveduplicatesincasethesamedataaresuccessfullyreceivedonmultiplecarriers.Thisresultsinselectiondiversity.

14.1.1BandwidthAdaptationNRsupport averywide transmissionbandwidth, up to several 100MHzon asingle carrier. This is useful for rapid delivery of large payloads but is notneeded for smaller payload sizes or for monitoring the downlink controlchannels when not scheduled. Hence, as mentioned already in Chapter 5 NRsupports receiver-bandwidth adaptation such that the device can use a narrowbandwidth for monitoring control channels and only open the full bandwidthwhen a large amount of data is scheduled. This can be seen as discontinuousreceptioninthefrequencydomain.Opening the wideband receiver can be done by using the bandwidth part

indicator field in theDCI. If thebandwidthpart indicatorpoints to adifferentbandwidth part than the currently active one, the active bandwidth part ischanged (seeFig.14.2).The time it takes tochange theactivebandwidthpartdependsonseveralfactors,forexample,ifthecenterfrequencychangesandthereceiverneedstoretuneornot,butcanbeintheorderofaslot.Onceactivated,thedeviceusesthenew,andwider,bandwidthpartforitsoperation.

FIGURE14.2 Illustrationofbandwidthadaptationprinciple.

Uponcompletionofthedatatransferrequiringthewiderbandwidth,thesamemechanismcanbeused to revertback to theoriginalbandwidthpart.There isalso a possibility to configure a timer to handle the bandwidth-part switchinginstead of explicit signaling. In this case, one of the bandwidth parts isconfigured as the default bandwidth part. If no default bandwidth part isexplicitlyconfigured,theinitialbandwidthpartobtainedfromtherandom-accessprocedureisusedasthedefaultbandwidthpart.UponreceivingaDCIindicatingabandwidthpartotherthanthedefaultone,thetimerisstarted.Whenthetimerexpires, thedevice switchesback to thedefault bandwidthpart.Typically, thedefault bandwidth part is narrower and can hence help reducing the devicepowerconsumption.The introduction of bandwidth adaptation in NR raised several design

questions not present in LTE, in particular related to the handling of controlssignaling as many transmission parameters are configured per bandwidth partand the DCI payload size therefore may differ between different bandwidthparts.Thefrequency-domainresourceallocationfieldisanobviousexample;thelarger the bandwidth part, the larger the number of bits for frequency-domainresource allocation. This is not an issue as long as the downlink datatransmission uses the same bandwidth part as the DCI control signaling.2However, in thecaseofbandwidthadaptationthis isnot trueas thebandwidthpart indicator in theDCI received in one bandwidth part can point toanotherdifferentlysizedbandwidthpartfordatareception.ThisraisestheissueonhowtointerprettheDCIifthebandwidthpartindexpointstoanotherbandwidthpartthanthecurrentone,astheDCIfieldsinthedetectedDCImaynotmatchwhatisneededinthebandwidthpartpointedtobytheindexfield.OnepossibilitytoaddressthiswouldbetoblindlymonitorformultipleDCI

payload sizes, one for each configured bandwidth parts, but unfortunately thiswouldimplyalargeburdenonthedevice.Instead,anapproachwheretheDCIfieldsdetectedarereinterpretedtobeusefulinthebandwidthpartpointedtobythe index isused.Asimpleapproachhasbeenselectedwhere thebitfieldsarepaddedortruncatedtomatchwhatisassumedbythebandwidthpartscheduled.Naturally,thisimposessomelimitationonthepossibleschedulingdecisions,butas soon as the new bandwidth part is activated the devicemonitors downlinkcontrol signaling using the newDCI size and data can be scheduledwith fullflexibilityagain.Althoughthehandlingofdifferentbandwidthpartshasbeendescribedfroma

downlink perspective above, the same approach of reinterpreting the DCI isappliedtotheuplink.

14.1.2DownlinkPreemptionHandlingDynamic scheduling implies, as discussed above, that a schedulingdecision istakenforeach time interval. Inmanycases the time interval isequal toaslot,that is, theschedulingdecisionsare takenonceperslot.Thedurationofaslotdependsonthesubcarrierspacing;ahighersubcarrierspacingleadstoashorterslot duration. In principle this could be used to support lower-latencytransmission,butasthecyclicprefixalsoshrinkswhenincreasingthesubcarrierspacing,itisnotafeasibleapproachinalldeployments.Therefore,asdiscussedinSection7.2,NRsupportsamoreefficientapproachtolowlatencybyallowingfor transmission over a fraction of a slot, starting at anyOFDMsymbol.Thisallowsforverylowlatencywithoutsacrificingrobustnesstotimedispersion.InFig.14.3, an exampleof this is illustrated.DeviceAhasbeen scheduled

with a downlink transmission spanning one slot. During the transmission todevice A, latency-critical data for device B arrives to the gNB, whichimmediately scheduled a transmission to device B. Typically, if there arefrequency resources available, the transmission todeviceB is scheduledusingresourcesnotoverlappingwiththeongoingtransmissiontodeviceA.However,inthecaseofahighloadinthenetwork,thismaynotbepossibleandthereisnochoicebuttouse(someof)theresourcesoriginallyintendedfordeviceAforthelatency-critical transmission to device B. This is sometimes referred to as thetransmission to device B preempting the transmission to device A, whichobviously will suffer an impact as a consequence of some of the resourcesdeviceAassumescontainsdataforitsuddenlycontainingdatafordeviceB.

FIGURE14.3 Downlinkpreemptionindication.

ThereareseveralpossibilitiestohandlethisinNR.Oneapproachistorelyonhybrid-ARQretransmissions.DeviceAwillnotbeabletodecodethedataduetothe resources being preempted and will consequently report a negativeacknowledgment to the gNB, which can retransmit the data at a later timeinstant. Either the complete transport block is retransmitted, or CBG-basedretransmission is used to retransmit only the impacted codeblock groups asdiscussedinSection13.1.There is alsoapossibility to indicate todeviceA that someof its resources

havebeenpreemptedandusedforotherpurposes.Thisisdonebytransmittingapreemption indicator to device A in a slot after the slot containing the datatransmission.ThepreemptionindicatorusesDCIformat2-1(seeChapter10fordetailsondifferentDCIformats)andcontainsabitmapof14bits.InterpretationofthebitmapisconfigurablesuchthateachbitrepresentsoneOFDMsymbolinthetimedomainandthefullbandwidthpart,ortwoOFDMsymbolsinthetimedomain and one half of the bandwidth part. Furthermore, the monitoringperiodicityofthepreemptionindicatorisconfiguredinthedevice,forexample,everynthslot.The behavior of the device when receiving the preemption indicator is not

specified,butareasonablebehaviorcouldbetoflushthepartofthesoftbufferwhichcorrespondstothepreemptedtime–frequencyregiontoavoidsoft-buffercorruptionforfutureretransmissions.Fromasoft-bufferhandlingperspectiveinthe device, themore frequent themonitoring of the preemption indicator, thebetter(ideally,itshouldcomeimmediatelyafterthepreemptionoccurred).

14.2DynamicUplinkScheduling

Thebasicfunctionoftheuplinkschedulerinthecaseofdynamicschedulingissimilar to its downlink counterpart, namely to dynamically control whichdevicesaretotransmit,onwhichuplinkresources,andwithwhat transmissionparameters.The general downlink scheduling discussion is applicable to the uplink as

well. However, there are some fundamental differences between the two. Forexample, theuplinkpower resource isdistributed among thedevices,while inthe downlink the power resource is centralized within the base station.Furthermore,themaximumuplinktransmissionpowerofasingledeviceisoftensignificantlylowerthantheoutputpowerofabasestation.Thishasasignificantimpactontheschedulingstrategy.Eveninthecaseofalargeamountofuplinkdata to transmit there might not be sufficient power available—the uplink isbasically power limited and not bandwidth limited,while in the downlink thesituationcantypicallybetheopposite.Hence,uplinkschedulingtypicallyresultsin a larger degree of frequency multiplexing of different devices than in thedownlink.Eachscheduleddeviceisprovidedwithaschedulinggrant indicatingtheset

of time/frequency/spatial resources to use for the UL-SCH as well as theassociatedtransportformat.Uplinkdatatransmissionsonlytakeplaceinthecasethatthedevicehasavalidgrant.Withoutagrant,nodatacanbetransmitted.Theuplinkschedulerisincompletecontrolofthetransportformatthedevice

shall use, that is, the device has to follow the scheduling grant. The onlyexceptionisthatthedevicewillnottransmitanything,regardlessofthegrant,iftherearenodatainthetransmissionbuffer.Thisreducestheoverallinterferencebyavoidingunnecessarytransmissionsinthecasethatthenetworkscheduledadevicewithnodatapendingtransmission.Logicalchannelmultiplexingiscontrolledbythedeviceaccordingtoasetof

rules (see Section 14.2.1). Thus, the scheduling grant does not explicitlyschedule a certain logical channel but rather the device as such—uplinkschedulingisprimarilyperdeviceandnotperradiobearer(althoughthepriorityhandlingmechanism discussed below in principle can be configured to obtainschedulingperradiobearer).UplinkschedulingisillustratedintherightpartofFig. 14.4, where the scheduler controls the transport format and the devicecontrols the logical channelmultiplexing. This allows the scheduler to tightlycontroltheuplinkactivitytomaximizetheresourceusagecomparedtoschemeswhere the device autonomously selects the data rate, as autonomous schemestypicallyrequiresomemarginintheschedulingdecisions.Aconsequenceofthe

schedulerbeingresponsibleforselectionofthetransportformatisthataccurateanddetailedknowledgeabout thedevicesituationwith respect tobuffer statusand power availability is accentuated compared to schemes where the deviceautonomouslycontrolsthetransmissionparameters.

FIGURE14.4 DownlinkanduplinkschedulinginNR.

Thetimeduringwhichthedeviceshouldtransmitintheuplinkisindicatedaspart of theDCI as described in Section 10.1.11.Unlike in the downlink case,where the scheduling assignment typically is transmitted close in time to thedata,thisisnotnecessarilythecaseintheuplink.Sincethegrantistransmittedusing downlink control signaling, a half-duplex device needs to change thetransmissiondirectionbeforetransmittingintheuplink.Furthermore,dependingon the uplink–downlink allocation, multiple uplink slots may need to bescheduled using multiple grants transmitted at the same downlink occasion.Hence,thetimingfieldintheuplinkgrantisimportant.Thedevicealsoneedsacertainamountoftimetoprepareforthetransmission

asoutlinedinFig.14.5.Fromanoverallperformanceperspective,theshorterthetime thebetter.However, fromadevicecomplexityperspective theprocessingtimecannotbemadearbitrarilyshort.InLTE,morethan3mswasprovidedforthedevice toprepare theuplink transmission.ForNR,amore latency-focuseddesign, for example, the updatedMAC and RLC header structure, as well astechnology development in general has considerably reduced this time. Thedelay from the reception of a grant to the transmission of uplink data is

summarized in Fig. 14.1. As seen from these numbers, the processing timedependsonthesubcarrierspacing,althoughitisnotpurelyscaledinproportiontothesubcarrierspacing.Itisalsoseenthattwodevicecapabilitiesarespecified.All devices need to fulfill the baseline requirements, but a device may alsodeclarewhether it is capable of amore aggressive processing time linewhichcanbeusefulinlatency-criticalapplications(Table14.1).

FIGURE14.5 Exampleofuplinkschedulingintofutureslots.

Table14.1

Similar to the downlink case, the uplink scheduler can benefit frominformation on channel conditions, buffer status, and power availability.However, the transmission buffers reside in the device, as does the poweramplifier. This calls for the reportingmechanisms described below to providetheinformationtothescheduler,unlikethedownlinkcasewherethescheduler,power amplifier, and transmission buffers all are in the same node. Uplinkpriority handling is, as already touched upon, another area where uplink anddownlinkschedulingdiffer.

14.2.1UplinkPriorityHandlingMultiplelogicalchannelsofdifferentprioritiescanbemultiplexedintothesametransport blockusing theMACmultiplexing functionality.Except for the casewhen the uplink scheduling grant provides resources sufficient to transmit alldata on all logical channels, the multiplexing needs to prioritize between thelogicalchannels.However,unlikethedownlinkcase,wheretheprioritizationisuptotheschedulerimplementation,theuplinkmultiplexingisdoneaccordingto

asetofwell-definedrulesinthedevicewithparameterssetbythenetwork.Thereasonforthisisthataschedulinggrantappliestoaspecificuplinkcarrierofadevice,notexplicitlytoaspecificlogicalchannelwithinthecarrier.A simple approachwould be to serve the logical channels in strict priority

order. However, this could result in starvation of lower-priority channels—allresources would go to the high-priority channel until the buffer is empty.Typically,anoperatorwouldinsteadliketoprovideatleastsomethroughputforlow-priorityservicesaswell.Furthermore,asNRisdesignedtohandleamixofawiderangeoftraffictypes,amoreelaborateschemeisneeded.Forexample,trafficduetoafileuploadshouldnotnecessarilyexploitagrantintendedforalatency-criticalservice.The starvation problem is present already in LTE where it is handled by

assigningaguaranteeddataratetoeachchannel.Thelogicalchannelsarethenservedindecreasingpriorityorderuptotheirguaranteeddatarate,whichavoidsstarvationaslongasthescheduleddatarateisatleastaslargeasthesumoftheguaranteeddatarates.Beyondtheguaranteeddatarates,channelsareservedinstrictpriorityorderuntilthegrantisfullyexploited,orthebufferisempty.NRappliesasimilarapproach.However,giventhelargeflexibilityofNRin

terms of different transmission durations and a wider range of traffic typessupported, a more advanced scheme is needed. One possibility would be todefine different profiles, each outlining an allowed combination of logicalchannels,andexplicitlysignaltheprofiletouseinthegrant.However,inNRtheprofiletouseisimplicitlyderivedfromotherinformationavailableinthegrantratherthanexplicitlysignaled.Uponreceptionofanuplinkgrant,twostepsareperformed.First,thedevice

determineswhichlogicalchannelsareeligibleformultiplexingusingthisgrant.Second,thedevicedeterminesthefractionoftheresourcesthatshouldbegiventoeachofthelogicalchannels.The first step determines the logical channels from which data can be

transmitted with the given grant. This can be seen as an implicitly derivedprofile.Foreachlogicalchannel,thedevicecanbeconfiguredwith:

•Thesetofallowedsubcarrierspacingsthislogicalchannelisallowedtouse;

•ThemaximumPUSCHdurationwhichispossibletoscheduleforthislogicalchannel;and

•Thesetofservingcell,thatis,thesetofuplinkcomponentcarriersthe

logicalchannelisallowedtobetransmittedupon.

Onlythelogicalchannelsforwhichtheschedulinggrantmeetstherestrictionsconfiguredareallowedtobetransmittedusingthisgrant,thatis,areeligibleformultiplexing at this particular time instant. In addition, the logical channelmultiplexingcanalsoberestrictedfortransmissionwithoutadynamicgrant.CouplingthemultiplexingruletothePUSCHdurationisin3GPPmotivated

by thepossibility to controlwhether latency-critical data shouldbe allowed toexploitagrantintendedforlesstime-criticaldata.Asanexample,assume thereare twodata flows,eachonadifferent logical

channel. One logical channel carries latency-critical data and is given a highpriority,while theother logical channel carriesnon-latency-criticaldata and isgivenalowpriority.ThegNBtakesschedulingdecisionsbasedon,amongotheraspects,informationaboutthebufferstatusinthedeviceprovidedbythedevice.Assume that the gNB scheduled a relatively long PUSCH duration based oninformationthatthereisonlynontime-criticalinformationinthebuffers.Duringthe reception of the scheduling grant, time-critical information arrives to thedevice.Without the restriction on the maximum PUSCH duration, the devicewould transmit the latency-critical data, possiblymultiplexed with other data,over a relatively long transmission duration and potentially not meeting thelatencyrequirementssetupfortheparticularservice.Instead,abetterapproachwouldbetoseparatelyrequestatransmissionduringashortPUSCHdurationforthe latency critical data, something which is possible by configuring themaximumPUSCHdurationappropriately.Sincethelogicalchannelcarryingthelatency-critical traffic has been configured with a higher priority than thechannelcarryingthenon-latency-criticalservice,thenoncriticalservicewillnotblocktransmissionofthelatency-criticaldataduringtheshortPUSCHduration.Thereasontoalsoincludethesubcarrierspacingissimilartotheduration.In

thecaseofmultiplesubcarrierspacingsconfiguredforasingledevice,alowersubcarrier spacing implies a longer slot duration and the reasoning above canalsobeappliedinthiscase.Restricting the uplink carriers allowed for a certain logical channel is

motivatedbythepossiblydifferentpropagationconditionsfordifferentcarriersand by dual connectivity. Two uplink carriers at vastly different carrierfrequencies can have different reliability. Data which are critical to receivemightbebettertotransmitonalowercarrierfrequencytoensuregoodcoverage,while less-sensitive data can be transmitted on a carrier with a higher carrier

frequencyandpossiblyspottiercoverage.Anothermotivationisduplication,thatis,thesamedatatransmittedonmultiplelogicalchannels,toobtaindiversityasmentionedinSection6.4.2.Ifbothlogicalchannelswouldbetransmittedonthesame uplink carrier, the original motivation for duplication—to obtain adiversityeffect—wouldbegone.At thispoint in theprocess, the setof logical channels fromwhichdataare

allowed to be transmitted given the current grant is established, based on themapping-related parameters configured. Multiplexing of the different logicalchannels also needs to answer the question of how to distribute resourcesbetween the logical channels having data to transmit and eligible fortransmission. This is done based on a set of priority-related parametersconfiguredforeachlocalchannel:

•Priority;•Prioritizedbitrate(PBR);and•Bucketsizeduration(BSD).

Theprioritizedbit rate and thebucket sizeduration together servea similarpurpose as the guaranteed bit rate in LTE but can account for the differenttransmissiondurationspossibleinNR.Theproductoftheprioritizedbitrateandthebucketsizedurationisinessenceabucketofbitsthatataminimumshouldbe transmitted for the given logical channel during a certain time. At eachtransmissioninstant,thelogicalchannelsareservedindecreasingpriorityorder,while trying to fulfill the requirement on the minimum number of bits totransmit. Excess capacity when all the logical channels are served up to thebucketsizeisdistributedinstrictpriorityorder.PriorityhandlingandlogicalchannelmultiplexingareillustratedinFig.14.6.

FIGURE14.6 ExampleoflogicalchannelprioritizationforfourdifferentscheduleddataratesandtwodifferentPUSCHdurations.

14.2.2SchedulingRequest

14.2.2SchedulingRequestTheuplinkschedulerneedsknowledgeofdeviceswithdatatotransmitandthatthereforeneedtobescheduled.Thereisnoneedtoprovideuplinkresourcestoadevicewithnodata to transmit.Hence, as aminimum, the schedulerneeds toknowwhetherthedevicehasdatatotransmitandshouldbegivenagrant.Thisisknownasaschedulingrequest.Schedulingrequestsareusedfordevicesnothaving a valid scheduling grant; devices that have a valid grant providemoredetailedschedulinginformationtothegNBasdiscussedinthenextsection.Aschedulingrequestisaflag,raisedbythedevicetorequestuplinkresources

from the uplink scheduler. Since the device requesting resources by definitionhas noPUSCH resource, the scheduling request is transmitted on thePUCCHusingpreconfiguredandperiodicallyreoccurringPUCCHresourcesdedicatedtothedevice.Withadedicatedscheduling-requestmechanism,thereisnoneedtoprovide the identityof thedevice requesting tobe scheduled as the identity isimplicitly known from the resources upon which the request is transmitted.When data with higher priority than already existing in the transmit buffersarriveat thedeviceand thedevicehasnograntandhencecannot transmit thedata, thedevice transmits a scheduling request at thenextpossible instant andthegNBcanassignagranttothedeviceuponreceptionoftherequest(seeFig.14.7).

FIGURE14.7 Exampleofschedulingrequestoperation.

This is similar to the approach taken by LTE; however, NR supportsconfiguration ofmultiple scheduling requests from a single device. A logicalchannelcanbemappedtozeroormoreschedulingrequestconfigurations.Thisprovides the gNB not only with information that there are data awaitingtransmissioninthedevice,butalsowhattypeofdataareawaitingtransmission.ThisisusefulinformationforthegNBgiventhewiderrangeoftraffictypestheNRisdesignedtohandle.Forexample,thegNBmaywanttoscheduleadevicefor transmission of latency-critical information but not for non-latency-criticalinformation.

EachdevicecanbeassigneddedicatedPUCCHschedulingrequestresourceswith a periodicity ranging from every second OFDM symbol to support verylatency-critical services up to every 80 ms for low overhead. Only onescheduling request can be transmitted at a given time, that is, in the case ofmultiple logical channels having data to transmit a reasonable behavior is totrigger the scheduling request corresponding to the highest-priority logicalchannel. A scheduling request is repeated in subsequent resources, up to aconfigurablelimit,untilagrantisreceivedfromthegNB.It isalsopossibletoconfigure a prohibit timer, controlling how often a scheduling request can betransmitted. In the case of multiple scheduling-request resources in a device,bothoftheseconfigurationsaredoneasperschedulingrequestresource.A device which has not been configured with scheduling request resources

reliesontherandom-accessmechanismtorequestresources.Thiscanbeusedtocreate a contention-based mechanism for requesting resources. Basically,contention-based designs are suitable for situations where there is a largenumberofdevicesinthecellandthetrafficintensity,andhencetheschedulingintensity,islow.Inthecaseofhighertrafficintensities,itisbeneficialtosetupatleastoneschedulingrequestresourceforthedevice.

14.2.3BufferStatusReportsDevicesthatalreadyhaveavalidgrantdonotneedtorequestuplinkresources.However,toallowtheschedulertodeterminetheamountofresourcestogranttoeachdeviceinthefuture,informationaboutthebuffersituation,discussedinthissection,andthepoweravailability,discussedinthenextsection,isuseful.Thisinformation is provided to the scheduler as part of the uplink transmissionthroughMAC control elements (see Section 6.4.4.1 for a discussion onMACcontrolelementsandthegeneralstructureofaMACheader).TheLCIDfieldinoneoftheMACsubheadersissettoareservedvalueindicatingthepresenceofabufferstatusreport,asillustratedinFig.14.8.

FIGURE14.8 MACcontrolelementsforbufferstatusreportingandpowerheadroomreports.

Fromaschedulingperspective,bufferinformationforeachlogicalchannelisbeneficial,althoughthiscouldresultinasignificantoverhead.Logicalchannelsarethereforegroupedintouptoeightlogical-channelgroupsandthereportingisdone per group. The buffer-size field in a buffer-status report indicates theamount of data awaiting transmission across all logical channels in a logical-channel group. Four different formats for buffer status reports are defined,differinginhowmanylogical-channelgroupsareincludedinonereportandtheresolutionofthebufferstatusreport.Abuffer-statusreportcanbetriggeredforthefollowingreasons:

•Arrivalofdatawithhigherprioritythancurrentlyinthetransmissionbuffer—thatis,datainalogical-channelgroupwithhigherprioritythantheonecurrentlybeingtransmitted—asthismayimpacttheschedulingdecision.

•Periodicallyascontrolledbyatimer.•Insteadofpadding.Iftheamountofpaddingrequiredtomatchthescheduledtransportblocksizeislargerthanabuffer-statusreport,abuffer-statusreportisinsertedasitisbettertoexploittheavailablepayloadforusefulschedulinginformationinsteadofpaddingifpossible.

14.2.4PowerHeadroomReportsInadditiontobufferstatus,theamountoftransmissionpoweravailableineachdeviceisalsorelevantfortheuplinkscheduler.Thereislittlereasontoschedulea higher data rate than the available transmission power can support. In thedownlink, the available power is immediately known to the scheduler as the

poweramplifierisinthesamenodeasthescheduler.Fortheuplink,thepoweravailability, or power headroom, needs to be provided to the gNB. Powerheadroom reports are therefore transmitted from the device to the gNB in asimilar way as the buffer-status reports—that is, only when the device isscheduled to transmit on the UL-SCH. A power headroom report can betriggeredforthefollowingreasons:

•Periodicallyascontrolledbyatimer;•Changeinpathloss(thedifferencebetweenthecurrentpowerheadroomandthelastreportislargerthanaconfigurablethreshold);

•Insteadofpadding(forthesamereasonasbuffer-statusreports).

It isalsopossible toconfigureaprohibit timer tocontrol theminimumtimebetween two power-headroom reports and thereby the signaling load on theuplink.There are three different types of power-headroom reports defined in NR,

Type 1, Type 2, and Type 3. In the case of carrier aggregation or dualconnectivity, multiple power headroom reports can be contained in a singlemessage(MACcontrolelement).Type 1 power headroom reporting reflects the power headroom assuming

PUSCH-only transmission on the carrier. It is valid for a certain componentcarrier,assumingthatthedevicewasscheduledforPUSCHtransmissionduringa certain duration, and includes the power headroom and the correspondingvalue of themaximum per-carrier transmit power for component carrier c isdenoted, PCMAX,c. The value of PCMAX,c is explicitly configured and shouldhence be known to the gNB, but since it can be separately configured for anormaluplinkcarrierandasupplementaryuplinkcarrier,bothbelongingtothesamecell(thatis,havingthesameassociateddownlinkcomponentcarrier),thegNBneeds to knowwhich value the device used and hencewhich carrier thereportbelongsto.It can be noted that the power headroom is not ameasure of the difference

betweenthemaximumper-carriertransmitpowerandtheactualcarriertransmitpower. Rather, the power headroom is a measure of the difference betweenPCMAX,candthetransmitpowerthatwouldhavebeenusedassumingthattherewouldhavebeennoupperlimitonthetransmitpower(seeFig.14.9).Thus,thepower headroom can very well be negative, indicating that the per-carriertransmit power was limited by PCMAX,c at the time of the power headroom

reporting—thatis,thenetworkhasscheduledahigherdataratethanthedevicecansupportgiventheavailabletransmissionpower.Asthenetworkknowswhatmodulation-and-coding scheme and resource size the device used fortransmission in the time duration to which the power-headroom reportcorresponds,itcandeterminethevalidcombinationsofmodulation-and-codingscheme and resource size allocation, assuming that the downlink path loss isconstant.

FIGURE14.9 Illustrationofpowerheadroomreports.

Type1powerheadroomcanalsobereportedwhenthereisnoactualPUSCHtransmission. This can be seen as the power headroom assuming a defaulttransmission configuration corresponding to the minimum possible resourceassignment.

Type 2 power headroom reporting is similar to type 1, but assumessimultaneousPUSCHandPUCCHreporting,afeaturethatisnotfullysupportedin thefirst releaseof theNRspecificationsbutplannedforfinalizationin laterreleases.Type3 power headroom reporting is used to handleSRS switching, that is,

SRS transmissions on an uplink carrier where the device is not configured totransmit PUSCH. The intention with this report is to be able to evaluate theuplink quality of alternative uplink carries and, if deemed advantageous,(re)configurethedevicetousethiscarrierforuplinktransmissioninstead.Compared to power control, which can operate different power-control

processes for different beam-pair links (see Chapter 15), the power-headroomreport is per carrier and does not explicitly take beam-based operation intoaccount. One reason is that the network is in control of the beams used fortransmissionandhencecandeterminethebeamarrangementcorrespondingtoacertainpower-headroomreport.

14.3SchedulingandDynamicTDDOne of the key features of NR is the support for dynamic TDD, where thescheduler dynamically determines the transmission direction. Although thedescription uses the term dynamic TDD, the framework can in principle beappliedtohalf-duplexoperationingeneral,includinghalf-duplexFDD.Sinceahalf-duplexdevicecannottransmitandreceivesimultaneously,thereisaneedtosplittheresourcesbetweenthetwodirections.AsmentionedinChapter7threedifferent signaling mechanisms can provide information to the device onwhethertheresourcesareusedforuplinkordownlinktransmission:

•Dynamicsignalingforthescheduleddevice;•SemistaticsignalingusingRRC;and•Dynamicslot-formatindicationsharedbyagroupofdevices,primarilyintendedfornonscheduleddevices.

The scheduler is responsible for the dynamic signaling for the scheduleddevice,thatis,thefirstofthethreebulletsabove.In the case of a device capable of full-duplex operation, the scheduler can

scheduleuplinkanddownlinkindependentlyofeachotherandthereislimited,ifany,needfortheuplinkanddownlinkschedulertocoordinatetheirdecisions.

Inthecaseofahalf-duplexdevice,ontheotherhand,itisuptotheschedulertoensurethatahalf-duplexdeviceisnotrequestedtosimultaneouslyreceiveandtransmit. If a semistatic uplink–downlink pattern has been configured, theschedulersobviouslyneedtoobeythispatternaswellasitcannot,forexample,scheduleanuplinktransmissioninaslotconfiguredfordownlinkusageonly.

14.4TransmissionWithoutaDynamicGrantDynamicscheduling,asdescribedabove,isthemainmodeofoperationinNR.For each transmission interval, for example, a slot, the scheduler uses controlsignalingtoinstructthedevicetotransmitorreceive.Itisflexibleandcanadoptto rapid variations in the traffic behavior, but obviously requires associatedcontrol signaling; control signaling that in some situations it is desirable toavoid.NRthereforealsosupportstransmissionschemesnotrelyingondynamicgrants.In the downlink, semipersistent scheduling is supportedwhere the device is

configured with a periodicity of the data transmissions using RRC signaling.Activation of semipersistent scheduling is done using the PDCCH as fordynamicschedulingbutwiththeCS-RNTIinsteadofthenormalC-RNTI.3ThePDCCH also carries the necessary information in terms of time–frequencyresourcesandotherparametersneededinasimilarwayasdynamicscheduling.Thehybrid-ARQprocessnumber is derived from the timewhen thedownlinkdata transmission starts according to a formula. Upon activation ofsemipersistent scheduling, the device receives downlink data transmissionperiodicallyaccordingtotheRRC-configuredperiodicityusingthetransmissionparametersindicatedonthePDCCHactivatingthetransmission.Hence,controlsignaling is only used once and the overhead is reduced. After enablingsemipersistent scheduling, thedevicecontinues tomonitor the setofcandidatePDCCHs foruplinkanddownlink schedulingcommands.This isuseful in thecasethat thereareoccasional transmissionsof largeamountsofdataforwhichthe semipersistent allocation is not sufficient. It is also used to handle hybrid-ARQretransmissionswhicharedynamicallyscheduled.In the uplink, two schemes for transmission without a dynamic grant are

supported,differinginthewaystheyareactivated(seeFig.14.10):

•Configuredgranttype1,whereanuplinkgrantisprovidedbyRRC,includingactivationofthegrant;and

•Configuredgranttype2,wherethetransmissionperiodicityisprovidedbyRRCandL1/L2controlsignalingisusedtoactivate/deactivatethetransmissioninasimilarwayasinthedownlinkcase.

FIGURE14.10 Uplinktransmissionwithoutadynamicgrant.

The benefits for the two schemes are similar, namely to reduce controlsignalingoverheadand,tosomeextent,toreducethelatencybeforeuplinkdatatransmission as no scheduling request–grant cycle is needed prior to datatransmission.Type1setsallthetransmissionparameters,includingperiodicity,timeoffset,

and frequency resources aswell asmodulation-and-coding schemeof possibleuplink transmissions, using RRC signaling. Upon receiving the RRCconfiguration,thedevicecanstarttousetheconfiguredgrantfortransmissioninthetimeinstantgivenbytheperiodicityandoffset.Thereasonfortheoffsetistocontrolatwhattimeinstantsthedeviceisallowedtotransmit.Thereisnonotionof activation time in the RRC signaling in general; RRC configurations takeeffectassoonas theyare receivedcorrectly.Thispoint in timemayvaryas itdepends on whether RLC retransmissions were needed to deliver the RRCcommandornot.Toavoid this ambiguity, a timeoffset relative to theSFN isincludedintheconfiguration.Type 2 is similar to downlink semipersistent scheduling. RRC signaling is

usedtoconfiguretheperiodicity,whilethetransmissionparametersareprovidedas part of the activation using the PDCCH. Upon receiving the activationcommand, the device transmits according to the preconfigured periodicity ifthere are data in the buffer. If there are no data to transmit, the device will,

similarlytotype1,nottransmitanything.Notethatnotimeoffsetisneededinthis case as the activation time is well defined by the PDCCH transmissioninstant.Thedeviceacknowledges theactivation/deactivationof theconfiguredgrant

type 2 by sending aMAC control element in the uplink. If there are no dataawaiting transmissionwhen the activation is received, the networkwould notknowiftheabsenceoftransmissionisduetotheactivationcommandnotbeingreceived by the device or if it is due to an empty transmission buffer. Theacknowledgmenthelpsinresolvingthisambiguity.In both these schemes it is possible to configure multiple devices with

overlapping time–frequency resources in theuplink. In thiscase it isup to thenetworktodifferentiatebetweentransmissionsfromthedifferentdevices.

14.5DiscontinuousReceptionPacket-datatrafficisoftenhighlybursty,withoccasionalperiodsoftransmissionactivity followed by longer periods of silence. From a delay perspective, it isbeneficialtomonitorthedownlinkcontrolsignalingineachslot(orevenmorefrequently) to receive uplink grants or downlink data transmissions andinstantaneously reacton changes in the trafficbehavior.At the same time thiscomes at a cost in terms of power consumption at the device; the receivercircuitry in a typical device represents a non-negligible amount of powerconsumption. To reduce the device power consumption, NR includesmechanismsfordiscontinuousreception(DRX),followingthesameframeworkas in LTE with enhancements to handle multiple numerologies. Bandwidthadaptation and carrier activation are two other examples of power-savingmechanisms.The basicmechanism forDRX is a configurableDRX cycle in the device.

With a DRX cycle configured, the device monitors the downlink controlsignalingonlywhenactive,sleepingwiththereceivercircuitryswitchedofftheremaining time.This allows for a significant reduction inpower consumption:the longer the cycle, the lower thepower consumption.Naturally, this impliesrestrictions to the scheduler as the device can be addressed only when activeaccordingtotheDRXcycle.In many situations, if the device has been scheduled and is active with

receivingortransmittingdata,itishighlylikelyitwillbescheduledagaininthenearfuture.Onereasoncouldbethatitwasnotpossibletotransmitallthedata

inthetransmissionbufferinusingoneschedulingoccasionandhenceadditionaloccasions are needed. Waiting until the next activity period according to theDRX cycle, although possible, would result in additional delays. Hence, toreduce the delays, the device remains in the active state for a certainconfigurable time after being scheduled. This is implemented by the device(re)startingan inactivity timerevery time it is scheduledand remainingawakeuntilthetimeexpires,asillustratedatthetopofFig.14.11.DuetothefactthatNR can handle multiple numerologies, the DRX timers are specified inmillisecondsinordernottotietheDRXperiodicitytoacertainnumerology.

FIGURE14.11 DRXoperation.

Hybrid-ARQretransmissionsareasynchronousinbothuplinkanddownlink.If the device has been scheduled a transmission in the downlink it could notdecode, a typical situation is that the gNB retransmits the data at a later timeinstant, often as soon as possible. Therefore, the DRX functionality has aconfigurabletimerwhichisstartedafteranerroneouslyreceivedtransportblockandusedtowakeupthedevicereceiverwhenitislikelyforthegNBtoschedulearetransmission.Thevalueofthetimerispreferablysettomatchtheroundtriptime in the hybrid-ARQ protocol; a roundtrip time that depends on theimplementation.The abovemechanism, a (long)DRX cycle in combinationwith the device

remaining awake for some period after being scheduled, is sufficient formostscenarios. However, some services, most notably voice-over-IP, arecharacterizedbyperiodsof regular transmission, followedbyperiodsof noorvery little activity. To handle these services, a second short DRX cycle canoptionallybeusedinadditiontothelongcycledescribedabove.Normally,thedevice follows the long DRX cycle, but if it has recently been scheduled, itfollows a shorter DRX cycle for some time. Handling voice-over-IP in thisscenariocanbedonebysettingtheshortDRXcycleto20ms,asthevoicecodectypicallydeliversavoice-over-IPpacketper20ms.ThelongDRXcycleisthenusedtohandlelongerperiodsofsilencebetweentalkspurts.In addition to theRRC configuration of theDRXparameters, the gNB can

terminatean“onduration”andinstructthedevicetofollowthelongDRXcycle.ThiscanbeusedtoreducethedevicepowerconsumptionifthegNBknowsthatnoadditionaldataareawaitingtransmissioninthedownlinkandhencethereisnoneedforthedevicetobeactive.

1InthecaseofcarrieraggregationthereisoneDL-SCH(orUL-SCH)percomponentcarrier.2Strictlyspeaking,itissufficientifthesizeandconfigurationofthebandwidthpartusedforPDCCHandPDSCHarethesame.3Eachdevicehastwoidentities,the“normal”C-RNTIfordynamicschedulingandtheCS-RNTIforactivation/deactivationofsemipersistentscheduling.

CHAPTER15

UplinkPowerandTimingControl

Abstract

The chapter describes the NR uplink power control and uplink timingcontrol. The similarities and differences to the corresponding LTEmechanisms are high-lighted. Especially, the NR mechanisms for beam-basedpowercontrolarehigh-lighted.

KeywordsPowercontrol;open-looppowercontrol;closed-looppowercontrol;timingcontrol

Uplink power control and uplink timing control are the topics of this chapter.Powercontrolservesthepurposeofcontrollingtheinterference,mainlytowardsothercellsastransmissionswithinthesamecelltypicallyareorthogonal.Timingcontrol ensures that different device are received with the same timing, aprerequisitetomaintainorthogonalitybetweendifferenttransmissions.

15.1UplinkPowerControlNRuplinkpowercontrolisthesetofalgorithmsandtoolsbywhichthetransmitpowerfordifferentuplinkphysicalchannelsandsignalsiscontrolledtoensurethat they, to theextentpossible, are receivedby thenetworkat anappropriatepowerlevel.Inthecaseofanuplinkphysicalchannel,theappropriatepowerissimplythereceivedpowerneededforproperdecodingoftheinformationcarriedby the physical channel. At the same time, the transmit power should not beunnecessarilyhighas thatwouldcauseunnecessarilyhighinterferencetootheruplinktransmissions.The appropriate transmit power will depend on the channel properties,

including the channel attenuation and the noise and interference level at the

receiverside.Itshouldalsobenotedthattherequiredreceivedpowerisdirectlydependentonthedatarate.Ifthereceivedpoweristoolowonecanthuseitherincreasethetransmitpowerorreducethedatarate.Inotherwords,atleastinthecase of PUSCH transmission, there is an intimate relationship between powercontrolandlinkadaptation(ratecontrol).Similar toLTE power control [28],NRuplink power control is based on a

combinationof:

•Open-looppowercontrol,includingsupportforfractionalpath-losscompensation,wherethedeviceestimatestheuplinkpathlossbasedondownlinkmeasurementsandsetsthetransmitpoweraccordingly.

•Closed-looppowercontrolbasedonexplicitpower-controlcommandsprovidedbythenetwork.Inpractice,thesepower-controlcommandsaredeterminedbasedonpriornetworkmeasurementsofthereceiveduplinkpower,thustheterm“closedloop.”

Themaindifference,or rather extension,ofNRuplinkpowercontrol is thepossibilityforbeam-basedpowercontrol(seeSection15.1.2).

15.1.1BaselinePowerControlPower-controlforPUSCHtransmissionscan,somewhatsimplified,bedescribedbythefollowingexpression:where

• isthePUSCHtransmitpower;• isthemaximumallowedtransmitpowerpercarrier;• isanetwork-configurableparameterthatcan,somewhatsimplified,bedescribedasatargetreceivedpower;

• isanestimateoftheuplinkpathloss;• isanetwork-configurableparameterrelatedtofractionalpath-losscompensation;

• relatestothesubcarrierspacing usedforthePUSCHtransmission.Morespecifically, ;

• isthenumberofresourceblocksassignedforthePUSCHtransmission;

• relatestothemodulationschemeandchannel-codingrateusedfor

thePUSCHtransmission;1• isthepoweradjustmentduetotheclosed-looppowercontrol.

Theaboveexpressiondescribesuplinkpowercontrolpercarrier.Ifadeviceisconfigured with multiple uplink carriers (carrier aggregation and/orsupplementaryuplink),powercontrol according to expression (15.1) is carriedout separately for each carrier. The part of the power-controlexpressionthenensuresthatthepowerpercarrierdoesnotexceedthemaximumallowed transmit powerper carrier.However, therewill alsobe a limit on thetotaldevicetransmitpoweroverallconfigureduplinkcarriers. Inorder tostaybelowthislimittherewill,intheend,beaneedtocoordinatethepowersettingbetween the different uplink carriers (see further Section 15.1.4). SuchcoordinationisneededalsointhecaseofLTE/NRdual-connectivity.We will now consider the different parts of the above power control

expressioninsomewhatmoredetail.Whendoingthiswewillinitiallyignoretheparameters j, q, and l. The impact of these parameters will be discussed inSection15.1.2.Theexpression representsbasicopen-looppowercontrolsupporting

fractional path-loss compensation. In the case of full path-loss compensation,correspondingto ,andundertheassumptionthatthepath-lossestimateis an accurate estimate of the uplink path loss, the open-loop power controladjusts the PUSCH transmit power so that the received power alignswith the“target received power” . The quantity is provided as part of the power-controlconfigurationandwouldtypicallydependonthetargetdataratebutalsoonthenoiseandinterferencelevelexperiencedatthereceiver.Thedeviceisassumedtoestimatetheuplinkpathlossbasedonmeasurements

on some downlink signal. The accuracy of the path-loss estimate thus partlydependsonwhatextentdownlink/uplinkreciprocityholdsto.Especially,inthecaseofFDDoperationinpairedspectra,thepath-lossestimatewillnotbeabletocaptureanyfrequency-dependentcharacteristicsofthepathloss.In the case of fractional path-loss compensation, corresponding to , the

pathlosswillnotbefullycompensatedforandthereceivedpowerwillevenonaveragevarydependingonthelocationofthedevicewithinthecell,withlowerreceivedpowerfordeviceswithhigherpathloss,inpracticefordevicesatlargerdistancefromthecellsite.Thismustthenbecompensatedforbyadjustingtheuplinkdatarateaccordingly.The benefit of fractional path-loss compensation is reduced interference to

(15.2)(15.2)

neighborcells.Thiscomesatthepriceoflargervariationsintheservicequality,withreduceddata-rateavailabilityfordevicesclosertothecellborder.Theterm reflectsthefactthat,everythingelseunchanged,the

receivedpower,andthusalsothetransmitpower,shouldbeproportionaltothebandwidth assigned for the transmission. Thus, assuming full path-losscompensation( ), canmoreaccuratelybedescribedasanormalizedtargetreceived power. Especially, assuming full path-loss compensation, is thetarget receivedpowerassuming transmissionoverasingle resourceblockwith15kHznumerology.Theterm triestomodelhowtherequiredreceivedpowervarieswhenthe

number of information bits per resource element varies due to differentmodulation schemes and channel-coding rates. More preciselyΔTF=10·log((21.25·γ−1)·β)where is the number of information bits in the PUSCH transmission,

normalized by the number of resource elements used for the transmission notincludingresourceelementsusedfordemodulationreferencesymbols.Thefactor equals1inthecaseofdatatransmissiononPUSCHbutcanbe

set to a different value in the case that the PUSCH carries layer-1 controlsignaling(UCI).2Itcanbenotedthat,ignoringthefactor ,theexpressionfor isessentially

arewriteoftheShannonchannelcapacity withanadditionalfactor1.25.Inotherwords, canbeseenasmodelinglinkcapacityas80%ofShannoncapacity.Theterm isnotalways includedwhendeterminingthePUSCHtransmit

power.

•Theterm isonlyusedforsingle-layertransmission,thatis, inthecaseofuplinkmulti-layertransmission;

•Theterm can,ingeneral,bedisabled. should,forexample,notbeusedincombinationwithfractionalpowercontrol.Adjustingthetransmitpowertocompensatefordifferentdatarateswouldcounteractanyadjustmentofthedataratetocompensateforthevariationsinreceivedpowerduetofractionalpowercontrolasdescribedabove.

Finally, the term is the power adjustment related to closed-loop powercontrol.Thenetworkcanadjust byacertainstepgivenbyapower-controlcommandprovidedbythenetwork, therebyadjustingthetransmitpowerbased

onnetworkmeasurementsofthereceivedpower.Thepowercontrolcommandsarecarried in theTPC fieldwithinuplink schedulinggrants (DCI formats0–0and 0–1). Power control commands can also be carried jointly to multipledevicesbymeansofDCIformat2–2.Eachpowercontrolcommandconsistsof2bitscorrespondingtofourdifferentupdatesteps(–1dB,0dB,+1dB,+3dB).The reason for including 0 dB as an update step is that a power-controlcommandisincludedineveryschedulinggrantanditisdesirablenottohavetoadjustthePUSCHtransmitpowerforeachgrant.

15.1.2Beam-BasedPowerControlInthediscussionaboveweignoredtheparameterjfortheopen-loopparameters

and ,theparameter inthepathlossestimate ,andtheparameterintheclosed-looppoweradjustment .Theprimaryaimoftheseparametersistotakebeam-formingintoaccountfortheuplinkpowercontrol.

15.1.2.1MultiplePath-Loss-EstimationProcessesInthecaseofuplinkbeam-forming, theuplink-path-lossestimate usedtodetermine the transmitpoweraccording to expression (15.1) should reflect thepathloss,includingthebeam-forminggains,oftheuplinkbeampairtobeusedfor the PUSCH transmission. Assuming beam correspondence, this can beachieved by estimating the path loss based on measurements on a downlinkreferencesignaltransmittedoverthecorrespondingdownlinkbeampair.Astheuplink beam used for the transmission pair may change between PUSCHtransmissions, thedevicemay thushave to retainmultiple path-loss estimates,correspondingtodifferentcandidatebeampairs,inpractice,path-lossestimatesbased onmeasurements on different downlink reference signals.When actualPUSCH transmission is to take place over a specific beam pair, the path-lossestimate corresponding to that beam pair is then used when determining thePUSCHtransmitpoweraccordingtothepower-controlexpression(15.1).This is enabled by the parameter in the path-loss estimate of Eq.

(15.1). The network configures the device with a set of downlink referencesignals (CSI-RSorSSblock)onwhichpath loss is tobeestimated,witheachreferencesignalbeingassociatedwithaspecificvalueof .Inordernottoputtoohighrequirementsonthedevice,therecanbeatmostfourparallelpath-loss-estimationprocesses,eachcorrespondingtoaspecificvalueof .Thenetworkalso configures a mapping from the possible SRI values provided in the

schedulinggranttotheuptofourdifferentvaluesof .IntheendthereisthusamappingfromeachofthepossibleSRIvaluesprovidedintheschedulinggranttooneofuptofourconfigureddownlinkreferencesignalsandthus,indirectly,amapping from each of the possible SRI values to one of up to four path-lossestimates reflecting the path loss of a specific beam pair. When a PUSCHtransmission is scheduled by a scheduling grant including SRI, the path-lossestimateassociatedwiththatSRIisusedwhendeterminingthetransmitpowerforthescheduledPUSCHtransmission.Theprocedure is illustrated inFig.15.1for thecaseof twobeampairs.The

deviceisconfiguredwithtwodownlinkreferencesignals(CSI-RSorSSblock)thatinpracticewillbetransmittedonthedownlinkoverafirstandsecondbeampair, respectively.The device is running two path-loss-estimation processes inparallel, estimating the path loss for the first beam pair based onmeasurements on reference signalRS-1 and the path loss for the secondbeam pair based onmeasurements on reference-signal RS-2. The parameterqassociatesSRI=1withRS-1and thus indirectlywith .Likewise,SRI=2 isassociated with RS-2 and thus indirectly with . When the device isscheduledforPUSCHtransmissionwiththeSRIoftheschedulinggrantsetto1,the transmit powerof the scheduledPUSCH transmission is determinedbasedon the path-loss estimate that is, the path-loss estimate based onmeasurements on RS-1. Thus, assuming beam correspondence the path-lossestimate reflects the path loss of the beam pair over which the PUSCH istransmitted. If the device is instead scheduled for PUSCH transmission withSRI=2, the path-loss estimate , reflecting the path loss of the beam paircorresponding to SRI=2, is used to determine the transmit power for thescheduledPUSCHtransmission.

FIGURE15.1 Useofmultiplepower-estimationprocessestoenableuplinkpowercontrolinthecaseofdynamicbeammanagement.

15.1.2.2MultipleOpen-Loop-ParameterSetsInthePUSCHpower-controlexpression(15.1),theopen-loopparameters and

are associated with a parameter j. This simply reflects that there may bemultipleopen-loop-parameterpairs .Partly,differentopen-loopparameterswill be used for different types of PUSCH transmission (random-access“message 3” transmission, see Section 16.2, grant-free PUSCH transmissions,and scheduled PUSCH transmissions). However, there is also a possibility tohavemultiplepairsofopen-loopparameterforscheduledPUSCHtransmission,wherethepairtouseforacertainPUSCHtransmissioncanbeselectedbasedonthe SRI similar to the selection of path-loss estimates as described above. Inpracticethisimpliesthattheopen-loopparameters and willdependontheuplinkbeam.For the power setting of random-message 3, which in theNR specification

correspondsto , alwaysequals1.Inotherwords,fractionalpowercontrolisnotusedformessage-3 transmission.Furthermore, theparameter can, formessage 3, be calculated based on information in the random-accessconfiguration.ForotherPUSCH transmissions thedevice canbe configuredwithdifferent

open-loop-parameterpairs , corresponding todifferentvalues for theparameterj.Parameterpair shouldbeusedinthecaseofgrant-freePUSCH transmission,while the remaining parameter pairs are associatedwithscheduled PUSCH transmission. Each possible value of the SRI that can beprovided as part of the uplink scheduling grant is associated with one of theconfigured open-loop-parameter pairs. When a PUSCH transmission isscheduled with a certain SRI included in the scheduling grant, the open-loopparameters associated with that SRI are used when determining the transmitpowerforthescheduledPUSCHtransmission.

15.1.2.3MultipleClosed-LoopProcessesThe final parameter is the parameter l for the closed-loop process. PUSCHpower control allows for the configuration of two independent closed-loopprocesses,associatedwith and ,respectively.Similar to thepossibilityfor multiple path-loss estimates and multiple open-loop-parameter sets, theselectionofl,thatis,theselectionofclosed-loopprocess,canbetiedtotheSRIincludedintheschedulinggrantbyassociatingeachpossiblevalueoftheSRItooneoftheclosed-loopprocesses.

15.1.3PowerControlforPUCCH

Power control for PUCCH follows essentially the same principles as powercontrolforPUSCH,withsomeminordifferences.First, for PUCCH power control, there is no fractional path-loss

compensation,thatis,theparameter alwaysequalsone.Furthermore, for PUCCH power control, the closed-loop power control

commandsarecarriedwithinDCIformats1–0and1–1,thatis,withindownlinkschedulingassignmentsratherthanwithinuplinkschedulinggrants,whichisthecaseforPUSCHpowercontrol.OnereasonforuplinkPUCCHtransmissionsisthe transmission of hybrid-ARQ acknowledgments as a response to downlinktransmissions. Such downlink transmissions are typically associated withdownlink scheduling assignments on PDCCH and the corresponding power-controlcommandscouldthusbeusedtoadjustthePUCCHtransmitpowerpriorto the transmission of the hybrid-ARQ acknowledgments. Similar to PUSCH,power-control commands can also be carried jointly to multiple devices bymeansofDCIformat2–2.

15.1.4PowerControlintheCaseofMultipleUplinkCarriersThe above procedures describe how to set the transmit power for a givenphysicalchannelinthecaseofasingleuplinkcarrier.Foreachsuchcarrierthereis amaximumallowed transmit power and the part of thepower-controlexpressionensuresthattheper-carriertransmitpowerofacarrierdoesnotexceedpower .3Inmanycases,adeviceisconfiguredwithmultipleuplinkcarriers:

•Multipleuplinkcarriersinacarrieraggregationscenario;•AnadditionalsupplementaryuplinkcarrierinthecaseofSUL.

Inadditiontothemaximumper-carriertransmitpower ,thereisalimitonthetotaltransmittedpoweroverallcarriers.Foradeviceconfiguredfor

NRtransmissiononmultipleuplinkcarriers, shouldobviouslynotexceed.However, thesumof overallconfigureduplinkcarriersmayvery

well, and often will, exceed . The reason is that a device will often nottransmit simultaneously on all its configured uplink carriers and the deviceshouldthenpreferablystillbeabletotransmitwiththemaximumallowedpower

.Thus,theremaybesituationswhenthesumofthetransmitpowerofeach

carriergivenbythepower-controlexpression(15.1)exceeds .Inthatcase,the power of each carrier needs to be scaled down to ensure that the eventualtransmitpowerofthedevicedoesnotexceedthemaximumallowedvalue.Another situation that needs to be taken care of is the simultaneous uplink

transmission of LTE and NR in the case of a device operating in dual-connectivitybetweenLTEandNR.Notethat,atleastinaninitialphaseofNRdeploymentthiswillbethenormalmode-of-operationasthefirstreleaseoftheNRspecificationsonlysupportnon-standaloneNRdeployments.Inthiscase,thetransmission on LTEmay limit the power available for NR transmission andviceversa.ThebasicprincipleisthattheLTEtransmissionhaspriority,thatistheLTEcarrier is transmittedwith thepowergivenby theLTEuplinkpowercontrol[28].TheNRtransmissioncanthenusewhateverpowerisleftuptothepowergivenbythepower-controlexpression(15.1).ThereasonforprioritizingLTEoverNRismultifold:

•InthespecificationofNR,includingthesupportforNR/LTEdualconnectivity,therehasbeenanaimtoasmuchaspossibleavoidanyimpactontheLTEspecifications.ImposingrestrictionsontheLTEpowercontrol,duetothesimultaneoustransmissiononNR,wouldhaveimpliedsuchanimpact.

•Atleastinitially,LTE/NRdual-connectivitywillhaveLTEprovidingthecontrol-planesignaling,thatis,LTEisusedforthemastercellgroup(MCG).TheLTElinkisthusmorecriticalintermsofretainingconnectivityanditmakessensetoprioritiesthatlinkoverthe“secondary”NRlink.

15.2UplinkTimingControlTheNR uplink allows for uplink intracell orthogonality, implying that uplinktransmissions received from different devices within a cell do not causeinterferencetoeachother.Arequirementforthisuplinkorthogonalitytoholdisthattheuplinkslotboundariesforagivennumerologyare(approximately)timealignedatthebasestation.Morespecifically,anytimingmisalignmentbetweenreceivedsignalsshouldfallwithinthecyclicprefix.Toensuresuchreceiver-sidetime alignment, NR includes a mechanism for transmit-timing advance. Themechanism is similar to the corresponding mechanism in LTE, the maindifference being the use of different timing advance step sizes for different

numerologies.Inessence,timingadvanceisanegativeoffset,atthedevice,betweenthestart

ofadownlinkslotasobservedbythedeviceandthestartofaslotintheuplink.Bycontrollingtheoffsetappropriatelyforeachdevice,thenetworkcancontrolthetimingof thesignalsreceivedat thebasestationfromthedevices.Devicesfarfromthebasestationencounteralargerpropagationdelayandthereforeneedto start their uplink transmissions somewhat in advance, compared to devicesclosertothebasestation,asillustratedinFig.15.2.Inthisspecificexample,thefirst device is located close to the base station and experiences a smallpropagation delay, TP,1. Thus, for this device, a small value of the timingadvanceoffsetTA,1issufficienttocompensateforthepropagationdelayandtoensure the correct timing at the base station. However, a larger value of thetiming advance is required for the second device,which is located at a largerdistancefromthebasestationandthusexperiencesalargerpropagationdelay.

FIGURE15.2 Uplinktimingadvance.

Thetiming-advancevalueforeachdeviceisdeterminedbythenetworkbasedonmeasurements on the respective uplink transmissions. Hence, as long as adevice carries out uplink data transmission, this can be used by the receivingbase station to estimate theuplink receive timingand thusbe a source for thetiming-advancecommands.Soundingreferencesignalscanbeusedasaregularsignal to measure upon, but in principle the base station can use any signaltransmittedfromthedevices.Based on the uplink measurements, the network determines the required

timing correction for each device. If the timing of a specific device needscorrection, the network issues a timing-advance command for this specificdevice,instructingittoretardoradvanceitstimingrelativetothecurrentuplinktiming. The user-specific timing-advance command is transmitted as a MACcontrol element on the DL-SCH. Typically, timing-advance commands to adevicearetransmittedrelativelyinfrequently—forexample,oneorafewtimespersecond—butobviouslythisdependsonhowfastthedeviceismoving.TheproceduredescribedsofarisinessenceidenticaltotheoneusedforLTE.

As discussed above, the target of timing advance is to keep the timingmisalignmentwithinthesizeofthecyclicprefixandthestepsizeofthetimingadvance is thereforechosenasa fractionof thecyclicprefix.However,asNRsupportsmultiple numerologieswith the cyclic prefix being shorter the higherthesubcarrierspacing,thetimingadvancestepsizeisscaledinproportiontothecyclic prefix length and given by the subcarrier spacing of the active uplinkbandwidthpart.If the device has not received a timing-advance command during a

(configurable)period,thedeviceassumesithaslosttheuplinksynchronization.Inthiscase, thedevicemustreestablishuplinktimingusingtherandom-accessprocedurepriortoanyPUSCHorPUCCHtransmissionintheuplink.Forcarrieraggregation,theremaybemultiplecomponentcarrierstransmitted

fromasingledevice.Astraightforwardwayofhandlingthiswouldbetoapplythe same timing-advance value for all uplink component carriers.However, ifdifferent uplink carriers are received at different geographical locations, forexample,byusingremoteradioheadsforsomecarriersbutnotothers,differentcarriers would need different timing advance values. Dual connectivity withdifferentuplinkcarriersterminatedatdifferentsitesisanexamplewhenthisisrelevant. To handle such scenarios, a similar approach as in LTE is taken,

namelytogroupuplinkcarriersinso-calledtimingadvancedgroups(TAGs)andallow for different timing advance commands for different TAGs. Allcomponent carriers in the same group are subject to the same timing-advancecommand.Thetimingadvancestepsizeisdeterminedbythehighestsubcarrierspacingamongthecarriersinatimingadvancegroup.

1TheabbreviationTF=transportformat,atermusedinearlier3GPPtechnologiesbutnotusedexplicitlyforNR.2NotethatonecouldequallywellhavedescribedthisasaseparatetermappliedwhenPUSCHcarriesUCI.3Notethat,incontrasttoLTE,atleastforNRrelease15thereisnotsimultaneousPUCCHandPUSCHtransmissiononacarrier,andthusthereisatmostonephysicalchanneltransmittedonanuplinkcarrieratagiventimeinstant.

CHAPTER16

InitialAccess

Abstract

This chapter provides a detailed description of NR cell search, system-information delivery, and random access. Especially, the NR-specificfeatures related to beam forming and ultra-lean transmission arehighlighted.

KeywordsCellsearch;SSblock;PSS;SSS;PBCH;systeminformation;randomaccess;preambletransmission

WithinNR,theinitial-accessfunctionalityincludes:

•Thefunctionsandproceduresbywhichadeviceinitiallyfindsacellwhenenteringthecoverageareaofasystem;

•Thefunctionsandproceduresbywhichadeviceinidle/inactivestateaccessesthenetwork,typicallytorequesttheset-upofaconnectionandcommonlyreferredtoasrandomaccess.

Toquitea largeextent, similar functionality is alsoused inother situations.Thebasicnetworksignalsusedtoinitiallyfindacellcan,forexample,alsobeusedtofindnewcellswhenadeviceismovingwithinthecoverageareaofthenetwork.Furthermore,whenaccessinganewcell,thesamebasicrandom-accessprocedureasforinitialaccessmaybeused.Therandom-accessproceduremayalsobeusedbyadeviceinconnectedstate,forexample,torequestresourcesforuplinktransmissionortoreestablishuplinksynchronization.In this chapter, the details of cell search, system-information delivery, and

randomaccessaredescribed.

16.1CellSearchCell search covers the functions and procedures bywhich a device finds newcells.Cellsearchiscarriedoutwhenadeviceis initiallyenteringthecoverageareaofasystem.Toenablemobility,cellsearchisalsocontinuouslycarriedoutbydevicesmovingwithinthesystem,bothwhenthedeviceisconnectedtothenetworkandwheninidle/inactivestate.Herewewilldescribecellsearchbasedon so-called SS blocks, which are used for initial cell search as well asidle/inactive-statemobility.CellsearchbasedonSSblockscanalsobeusedforconnected-statemobility,althoughinthatcasecellsearchcanalsobebasedonCSI-RSexplicitlyconfiguredforthedevice.

16.1.1TheSSBlockToenabledevicestofindacellwhenenteringasystem,aswellastofindnewcellswhenmovingwithinthesystem,asynchronizationsignalconsistingoftwoparts, the Primary Synchronization Signal (PSS) and the SecondarySynchronizationSignal(SSS),isperiodicallytransmittedonthedownlinkfromeach NR cell. The PSS/SSS, together with the Physical Broadcast Channel(PBCH),isjointlyreferredtoasaSynchronizationSignalBlockorSSblock.1TheSS block serves a similar purpose and, inmany respects, has a similar

structure as the PSS/SSS/PBCH of LTE [28].2 However, there are someimportantdifferencesbetween theLTEPSS/SSS/PBCHand theNRSSblock.At leastpartly, theoriginof thesedifferencescanbe tracedback tosomeNR-specificrequirementsandcharacteristicsincludingtheaimtoreducetheamountof“always-on”signals,asdiscussedinSection5.2,andthepossibilityforbeam-formingduringinitialaccess.Aswith all NR downlink transmissions, SS-block transmission is based on

OFDM.Inotherwords, theSSblock is transmittedonasetof time/frequencyresources(resourceelements)withinthebasicOFDMgriddiscussedinSection7.3. Fig. 16.1 illustrates the time/frequency structure of a single SS blocktransmission.As can be seen, the SS block spans fourOFDM symbols in thetimedomainand240subcarriersinthefrequencydomain.

•ThePSSistransmittedinthefirstOFDMsymboloftheSSblockandoccupies127subcarriersinthefrequencydomain.Theremainingsubcarriersareempty.

•TheSSSistransmittedinthethirdOFDMsymboloftheSSblockandoccupiesthesamesetofsubcarriersasthePSS.ThereareeightandnineemptysubcarriersoneachsideoftheSSS.

•ThePBCHistransmittedwithinthesecondandfourthOFDMsymbolsoftheSSblock.Inaddition,PBCHtransmissionalsouses48subcarriersoneachsideoftheSSS.

FIGURE16.1 Time/frequencystructureofasingleSSblockconsistingofPSS,SSS,andPBCH.

The total number of resource elements used for PBCH transmission per SSblock thusequals576.Note that this includesresourceelementsfor thePBCHitselfbutalsoresourceelementsforthedemodulationreferencesignals(DMRS)neededforcoherentdemodulationofthePBCH.Different numerologies canbeused forSSblock transmission.However, to

limit the need for devices to simultaneously search for SS blocks of differentnumerologies,thereisinmanycasesonlyasingleSS-blocknumerologydefinedforagivenfrequencyband.Table 16.1 lists the different numerologies applicable for SS-block

transmission together with the corresponding SS-block bandwidth and timeduration,andthefrequencyrangeforwhicheachspecificnumerologyapplies.3

Note that 60 kHz numerology cannot be used for SS-block transmissionregardlessoffrequencyrange.Incontrast,240kHznumerologycanbeusedforSS-blocktransmissionalthoughitiscurrentlynotsupportedforotherdownlinktransmissions.Thereasontosupport240kHzSS-blocknumerologyistoenablea very short time duration for each SS block. This is relevant in the case ofbeam-sweeping overmany beamswith a corresponding large number of timemultiplexedSSblocks(seefurtherdetailsinSection16.1.4).

Table16.1

aTheSS-blockbandwidthissimplythenumberofsubcarriersusedforSSblock(240)multipliedbytheSS-blocksubcarrierspacing.

16.1.2Frequency-DomainPositionofSSBlockInLTE, thePSSandSSSarealways locatedat thecenterof thecarrier.Thus,onceanLTEdevicehasfoundaPSS/SSS,thatis,foundacarrier,itinherentlyknows the center frequency of the found carrier. The drawback with thisapproach,thatis,alwayslocatingthePSS/SSSatthecenterofthecarrier,isthata devicewith no a priori knowledge of the frequency-domain carrier positionmustsearchforPSS/SSSatallpossiblecarrierpositions(the“carrierraster”).Toallowforfastercellsearch,adifferentapproachhasbeenadoptedforNR.

Rather thanalwaysbeing locatedat thecenterof thecarrier, implying that thepossible SS-block locations coincide with the carrier raster, there are, withineach frequency band, a more limited set of possible locations of SS block,referredtoasthe“synchronizationraster”.InsteadofsearchingforanSSblockateachpositionofthecarrierraster,adevicethusonlyneedstosearchforanSSblockonthesparsersynchronizationraster.As carriers can still be located at an arbitrary position on the more dense

carrier raster, the SS blockmay not end up at the center of a carrier. The SSblockmaynotevenendupalignedwiththeresource-blockgrid.Hence,oncetheSSblockhasbeenfound,thedevicemustbeexplicitlyinformedabouttheexactSS-blockfrequency-domainpositionwithinthecarrier.Thisisdonebymeansof

information partly within the SS block itself, more specifically informationcarried by the PBCH (Section 16.1.5.3), and partly within the remainingbroadcastsysteminformation(seefurtherSection6.1.6).

16.1.3SSBlockPeriodicityTheSSblockistransmittedperiodicallywithaperiodthatmayvaryfrom5msup to160ms.However,devicesdoing initialcellsearch,aswellasdevices ininactive/idlestatedoingcellsearchformobility,canassumethattheSSblockisrepeatedatleastonceevery20ms.ThisallowsforadevicethatsearchesforanSS block in the frequency domain to know how long it must stay on eachfrequencybeforeconcludingthatthereisnoPSS/SSSpresentandthatitshouldmoveontothenextfrequencywithinthesynchronizationraster.The20msSS-blockperiodicity is four times longer than the corresponding

5msperiodicityofLTEPSS/SSStransmission.ThelongerSS-blockperiodwasselectedtoallowforenhancedNRnetworkenergyperformanceandingeneraltofollowtheultra-leandesignparadigmdescribedinSection5.2.ThedrawbackwithalongerSS-blockperiodisthatadevicemuststayoneachfrequencyforalonger time in order to conclude that there is no PSS/SSS on the frequency.However,thisiscompensatedforbythesparsesynchronizationrasterdiscussedabove, which reduces the number of frequency-domain locations on which adevicemustsearchforanSSblock.EventhoughdevicesdoinginitialcellsearchcanassumethattheSSblockis

repeated at least once every 20 ms, there are situations when there may bereasonstouseeitherashorterorlongerSS-blockperiodicity:

•AshorterSS-blockperiodicitymaybeusedtoenablefastercellsearchfordevicesinconnectedmode.

•AlongerSS-blockperiodicitymaybeusedtofurtherenhancenetworkenergyperformance.AcarrierwithanSS-blockperiodicitylargerthan20msmaynotbefoundbydevicesdoinginitialaccess.However,suchacarriercouldstillbeusedbydevicesinconnectedmode,forexample,asasecondarycarrierinacarrier-aggregationscenario.

It should be noted that there is even the possibility to deploy secondarycarrierswithoutanySSblock.

16.1.4SSBurstSet:MultipleSSBlocksinthe

16.1.4SSBurstSet:MultipleSSBlocksintheTimeDomainOnekeydifferencebetweentheSSblockandthecorrespondingsignalsforLTEisthepossibilitytoapplybeam-sweepingforSS-blocktransmission,thatis,thepossibility to transmit SS blocks in different beams in a time-multiplexedfashion(seeFig.16.2).ThesetofSSblockswithinabeam-sweepisreferredtoas an SS burst set.4 Note that the SS-block period discussed in the previoussectionisthetimebetweenSS-blocktransmissionswithinaspecificbeam,thatis,itisactuallytheperiodicityoftheSSburstset.Thismakessenseasadevicelocated in a certain downlink beammay only “see” a single SS block and beunawareofanyotherSSblockstransmittedfromthecell.

FIGURE16.2 Multipletime-multiplexedSSblockswithinanSS-burst-setperiod.

By applying beam-forming for the SS block, the coverage of a single SS-blocktransmissionisincreased.Beam-sweepingforSS-blocktransmissionalsoenablesreceiver-sidebeam-sweepingforthereceptionofuplinkrandom-accesstransmissions as well as downlink beam-forming for the random-accessresponse(seefurtherdetailsinSection16.2.1.5).AlthoughtheperiodicityoftheSSburstsetisflexiblewithaminimumperiod

of5msandamaximumperiodof160ms,eachSSburstsetisalwaysconfinedtoa5mstimeinterval,eitherinthefirstorsecondhalfofa10msframe.Themaximum number of SS blockswithin an SS burst set is different for

differentfrequencybands.

•Forfrequencybandsbelow3GHz,therecanbeuptofourSSblockswithinanSSburstset,enablingSS-blockbeam-sweepingoveruptofourbeams;

•Forfrequencybandsbetween3GHzand6GHz,therecanbeuptoeightSSblockswithinanSSburstset,enablingbeam-sweepingoverupto

eightbeams;•Forhigher-frequencybands(FR2)therecanbeupto64SSblockswithinanSSburstset,enablingbeam-sweepingoverupto64beams.

TherearetworeasonswhythemaximumnumberofSSblockswithinanSSburstset,andthusalsothemaximumnumberofbeamsoverwhichtheSSblockcanbeswept,islargerforhigher-frequencybands.

•Theuseofalargenumberofbeamswithmorenarrowbeam-widthistypicallymorerelevantforhigherfrequencies;

•AsthedurationoftheSSblockdependsontheSS-blocknumerology(seeTable16.1),alargenumberofSSblockswithinanSSburstsetwouldimplyaverylargeSS-blockoverheadforlowerfrequenciesforwhichlowerSS-blocknumerology(15or30kHz)mustbeused.

The set of possible SS-block locations in the time domain differ somewhatbetween different SS-block numerologies.As an example, Fig. 16.3 illustratesthe possible SS-block locations within an SS-burst-set period for the case of15kHznumerology.Ascanbeseen,theremaybeSS-blocktransmissioninanyof the first four slots.5 Furthermore, there can be up to two SS-blocktransmissions in each of these slot, with the first possible SS-block locationcorresponding tosymbol two tosymbol fiveand thesecondpossibleSS-blocklocationcorrespondingtosymboleight tosymboleleven.Finally,notethat thefirst and last two OFDM symbols of a slot are unoccupied by SS-blocktransmission.ThisallowsfortheseOFDMsymbolstobeusedfordownlinkanduplink control signaling, respectively, for devices already connected to thenetwork.ThesameistrueforallSS-blocknumerologies.

FIGURE16.3 Possibletime-domainlocationsofanSSblockwithinanSSburstsetfor15kHznumerology.

It should be noted that the SS-block locations outlined in Fig. 16.3 arepossibleSS-blocklocations,thatis,anSSblockisnotnecessarilytransmittedinall the locationsoutlined inFig.16.3.Theremaybeanything fromone singleSS-block transmissionup to themaximumnumber ofSSblockswithin anSSburst setdependingon thenumberofbeamsoverwhich theSSblock is tobebeam-swept.Furthermore, if less than themaximumnumber ofSSblocks is transmitted,

thetransmittedSSblocksdonothavetobetransmittedinconsecutiveSS-blocklocations.Rather,anysubsetofthepossiblesetofSS-blocklocationsoutlinedinFig.16.3canbeused for actualSS-block transmission. In thecaseof fourSSblocks within an SS burst set these may, for example, be located as two SSblockswithineachofthetwofirstslotsorasoneSSblockineachofthefourslotsofFig.16.3.ThePSS andSSSof anSS block only depend on the physical cell identity

(seebelow).Thus,thePSSandSSofallSSblockswithinacellareidenticalandcannotbeusedbythedevicetodeterminetherelativelocationofanacquiredSSblock within the set of possible SS-block locations. For this reason, each SSblock, more specifically, the PBCH, includes a “time index” that explicitlyprovides the relative location of the SS blockwithin the sequence of possibleSS-blocklocations(seefurtherdetailsinSection16.1.5.3).KnowingtherelativelocationoftheSSblockisimportantforseveralreasons:

•Itmakesitpossibleforthedevicetodetermineframetiming(seeSection16.1.5.3).

•ItmakesitpossibletoassociatedifferentSSblocks,inpracticedifferentbeams,withdifferentRACHoccasions.This,inturn,isaprerequisite

fortheuseofnetwork-sidebeamformingduringrandom-accessreception(seefurtherdetailsinSection16.2).

16.1.5DetailsofPSS,SSS,andPBCHAbove we have described the overall structure of an SS block and how itconsists of three parts: PSS, SSS, and PBCH. We have also described howmultipleSSblocksinthetimedomainconstituteanSSburstsetandhowanSSblockismappedtocertainOFDMsymbols.InthissectionwewilldescribethedetailedstructureofthedifferentSS-blockcomponents.

16.1.5.1ThePrimarySynchronizationSequence(PSS)ThePSSisthefirstsignalthatadeviceenteringthesystemwillsearchfor.Atthatstage,thedevicehasnoknowledgeofthesystemtiming.Furthermore,eventhoughthedevicesearchesforacellatagivencarrierfrequency,theremay,dueto inaccuracy of the device internal frequency reference, be a relatively largedeviationbetweenthedeviceandnetworkcarrierfrequency.ThePSShasbeendesignedtobedetectabledespitetheseuncertainties.Once the device has found the PSS, it has found synchronization up to the

periodicityofthePSS.Itcanthenalsousetransmissionsfromthenetworkasareference for its internal frequency generation, thereby to a large extenteliminatinganyfrequencydeviationbetweenthedeviceandthenetwork.Asdescribedabove,thePSSextendsover127resourceelementsontowhicha

PSSsequence ismapped(seeFig.16.4).

FIGURE16.4 PSSstructure.

TherearethreedifferentPSSsequences , ,and ,derivedasdifferentcyclic shifts of a basic length-127 M-sequence [70]generatedaccordingtotherecursiveformula(seealsoFig.16.5):

FIGURE16.5 GenerationofbasicM-sequencefromwhichthreedifferentPSSsequencesarederived.

ByapplyingdifferentcyclicshiftstothebasicM-sequence ,threedifferentPSSsequences , ,and canbegeneratedaccordingto:

WhichofthethreePSSsequencestouseinacertaincellisdeterminedbythephysicalcell identity (PCI)of thecell.Whensearchingfornewcells,adevicethusmustsearchforallthreePSSs.

16.1.5.2TheSecondarySynchronizationSequence(SSS)OnceadevicehasdetectedaPSSitknowsthetransmissiontimingoftheSSS.By detecting the SSS, the device can determine the PCI of the detected cell.There are 1008 different PCIs. However, already from the PSS detection thedevicehasreduced thesetofcandidatePCIsbyafactor3.Thereare thus336differentSSSs,thattogetherwiththealready-detectedPSSprovidesthefullPCI.Notethat,sincethetimingoftheSSSisknowntothedevice,theper-sequencesearchcomplexityisreducedcomparedtothePSS,enablingthelargernumberofSSSsequences.ThebasicstructureoftheSSSisthesameasthatofthePSS(Fig.16.4),that

is,theSSSconsistsof127subcarrierstowhichanSSSsequenceisapplied.On an even more detailed level, each SSS is derived from two basicM-

sequencesgeneratedaccordingtotherecursiveformulas

The actual SSS sequence is then derived by adding the two M sequencestogether,withdifferentshiftsbeingappliedtothetwosequences.

16.1.5.3PBCHWhilethePSSandSSSarephysicalsignalswithspecificstructures,thePBCHis a more conventional physical channel on which explicit channel-codedinformation is transmitted. The PBCH carries the master information block(MIB),whichcontainsa small amountof information that thedeviceneeds in

order tobe able to acquire the remaining system informationbroadcast by thenetwork.6Table 16.2 lists the information carried within the PBCH. Note that the

information differs slightly depending on if the carrier is operating in lower-frequencybands(FR1)orhigher-frequencybands(FR2).

Table16.2

InformationCarriedWithinthePBCH

Information NumberofBitsSS-blocktimeindex 0(FR1)/3(FR2)

CellBarredflag 2

1stPDSCHDMRSposition 1

SIB1numerology 1

SIB1configuration 8

CRBgridoffset 5(FR1)/4(FR2)

Half-framebit 1

Systemframenumber(SFN) 10

Cyclicredundancycheck(CRC) 24

As already mentioned, the SS-block time index identifies the SS-blocklocationwithinanSSburst set.Asdescribed inSection16.1.4,eachSSblockhas awell-definedpositionwithin anSSburst setwhich, in turn, is containedwithinthefirstorsecondhalfofa5msframe.FromtheSS-blocktimeindex,incombinationwiththehalf-framebit (seebelow), thedevicecanthusdeterminetheframeboundary.TheSS-blocktimeindexisprovidedtothedeviceastwoparts:

•AnimplicitpartencodedinthescramblingappliedtothePBCH;•AnexplicitpartincludedinthePBCHpayload.

Eight different scrambling patterns can be used for thePBCH, allowing forthe implicit indication of up to eight different SS-block time indices. This issufficientforoperationbelow6GHz(FR1)wheretherecanbeatmosteightSS

blockswithinanSSburstset.7ForoperationinthehigherNRfrequencyrange(FR2)therecanbeupto64

SSblockswithinanSSburstset, implyingtheneedforthreeadditionalbitstoindicate theSS-block time index.These threebits,whichare thusonlyneededfor operation above 10 GHz, are included as explicit information within thePBCHpayload.TheCellBarredflagconsistoftwobits:

•Thefirstbit,whichcanbeseenastheactualCellBarredflag,indicateswhetherornotdevicesareallowedtoaccessthecell;

•Assumingdevicesarenotallowedtoaccessthecell,thesecondbit,alsoreferredtoastheIntra-frequency-reselectionflag,indicateswhetherornotaccessispermittedtoothercellsonthesamefrequency.

If detecting that a cell is barred and that access to other cells on the samefrequencyisnotpermitted,adevicecanandshouldimmediatelyre-initiatecellsearchonadifferentcarrierfrequency.Itmayseemstrangetodeployacellandthenpreventdevicesfromaccessing

it.Historically this kind of functionality has been used to temporarily preventaccess to a certain cell during maintenance. However, the functionality hasadditional usage within NR due to the possibility for non-standalone NRdeploymentsforwhichdevicesshouldaccessthenetworkviathecorrespondingLTE carrier. By setting the CellBarred flag for the NR carrier in an NSAdeployment,thenetworkpreventsNRdevicesfromtryingtoaccessthesystemviatheNRcarrier.The1stPDSCHDMRSpositionindicatesthetime-domainpositionofthefirst

DMRSsymbolassumingDMRSMappingTypeA(seeSection9.11).TheSIB1numerologyprovidesinformationaboutthesubcarrierspacingused

for the transmission of the so-called SIB1, which is part of the systeminformation (see Section 16.1.6). The same numerology is also used for thedownlinkMessage2andMessage4thatarepartoftherandom-accessprocedure(seeSection16.2).AlthoughNRsupportsfourdifferentnumerologies(15kHz,30kHz,60kHz,and120kHz)fordatatransmission,foragivenfrequencybandthereareonlytwopossiblenumerologies.Thus,onebitissufficienttosignaltheSIB1numerology.The SIB1 configuration provides information about the search space,

corresponding CORESET, and other PDCCH-related parameters that a device

needsinordertomonitorforschedulingofSIB1.TheCRBgridoffsetprovidesinformationaboutthefrequencyoffsetbetween

the SS block and the common resource block grid. As discussed in Section16.1.2, the frequency-domainpositionof theSSblock relative to thecarrier isflexible and does not even have to be aligned with the carrier CRB grid.However, for SIB1 reception, the device needs to know the CRB grid. Thus,informationaboutthefrequencyoffsetbetweentheSSblockandtheCRBgridmustbeprovidedwithin thePBCHinorder tobeavailable todevicesprior toSIB1reception.NotethattheCRBgridoffsetonlyprovidestheoffsetbetweentheSSblock

andtheCRBgrid.InformationabouttheabsolutepositionoftheSSblockwithintheoverallcarrieristhenprovidedwithinSIB1.Thehalf-framebit indicates if theSSblock is located in the first or second

5mspart of a 10ms frame.Asmentioned above, the half-framebit, togetherwith the SS-block time index, allows for a device to determine the cell frameboundary.Allinformationabove,includingtheCRC,isjointlychannelcodedandrate-

matchedtofitthePBCHpayloadofanSSblock.AlthoughalltheinformationaboveiscarriedwithinthePBCHandisjointly

channelcodedandCRC-protected,someoftheinformationisstrictlyspeakingnotpart of theMIB.TheMIB is assumed tobe the sameover an80ms timeinterval (eight subframes) aswell as for all SS blockswithin an SS burst set.Thus, the SS-block time index, which is inherently different for different SSblockswithinanSSburstset,thehalf-framebitandthefourleastsignificantbitsoftheSFNarePBCHinformationcarriedoutsideoftheMIB.8

16.1.6ProvidingRemainingSystemInformationSystem information is a joint name for all the common (non-device-specific)informationthatadeviceneedsinordertoproperlyoperatewithinthenetwork.Ingeneral,thesysteminformationiscarriedwithindifferentSystemInformationBlocks(SIBs),eachconsistingofdifferenttypesofsysteminformation.InLTE, all system information is periodicallybroadcast over the entire cell

areamaking it alwaysavailablebutalso implying that it is transmittedeven ifthereisnodevicewithinthecell.ForNR,adifferentapproachhasbeenadoptedwherethesysteminformation,

beyond thevery limited information carriedwithin theMIB,hasbeendivided

intotwoparts.SIB1, sometimes also referred to as the remaining minimum system

information (RMSI) consists of the system information that a device needs toknowbeforeitcanaccessthesystem.SIB1isalwaysperiodicallybroadcastovertheentirecellarea.OneimportanttaskofSIB1istoprovidetheinformationthedeviceneedsinordertocarryoutaninitialrandomaccess(seeSection16.2).SIB1isprovidedbymeansofordinaryscheduledPDSCHtransmissionswith

a periodicity of 160 ms. As described above, the PBCH/MIB providesinformation about the numerology used for SIB1 transmission as well as thesearchspaceandcorrespondingCORESETusedforschedulingofSIB1.Withinthat CORESET, the device then monitors for scheduling of SIB1 which isindicatedbyaspecialSystemInformationRNTI(SI-RNTI).The remaining SIBs, not including SIB1, consist of the system information

thatadevicedoesnotneedtoknowbeforeaccessingthesystem.TheseSIBscanalsobeperiodicallybroadcastsimilartoSIB1.Alternatively,theseSIBscanbetransmittedondemand, thatis,onlytransmittedwhenexplicitlyrequestedbyaconnecteddevice.ThisimpliesthatthenetworkcanavoidperiodicbroadcastoftheseSIBs incellswherenodevice iscurrentlycamping, therebyallowingforenhancednetworkenergyperformance.

16.2RandomAccessOnceadevicehasfoundacellitmayaccessthecell.Thisisdoneaspartoftherandom-accessprocedure.Similar toLTE,NRuses a four-step random-accessprocedure consistingof

thefollowingsteps(seealsoFig.16.6):

•Step1:DevicetransmissionofapreamblealsoreferredtoasthePhysicalRandom-AccessChannel(PRACH);

•Step2:NetworktransmissionofaRandom-AccessResponse(RAR)indicatingreceptionofthepreambleandprovidingatime-alignmentcommandadjustingthetransmissiontimingofthedevicebasedonthetimingofthereceivedpreamble;

•Steps3/4:Deviceandnetworkexchangeofmessages(uplink“Message3”andsubsequentdownlink“Message4”)withtheaimofresolvingpotentialcollisionsduetosimultaneoustransmissionsofthesamepreamblefrommultipledeviceswithinthecell.Ifsuccessful,Message4

alsotransfersthedevicetoconnectedstate.

FIGURE16.6 Four-steprandom-accessprocedure.

Once the random-access procedure is completed, the device is in connectedstate and network-device communication can continue using normal dedicatedtransmission.Thebasicrandom-accessprocedureisalsousedinothercontextswithinNR,

forexample:

•Forhandover,whensynchronizationneedstobeestablishedtoanewcell;

•Toreestablishuplinksynchronizationtothecurrentcellifsynchronizationhasbeenlostduetoatoolongperiodwithoutanyuplinktransmissionfromthedevice;

•Torequestuplinkschedulingifnodedicatedscheduling-requestresourcehasbeenconfiguredforthedevice.

Parts of the basic random-access procedure are also usedwithin the beam-recoveryprocedure(seeSection12.3).

16.2.1PreambleTransmission

16.2.1PreambleTransmissionAs mentioned above, the random-access preamble is also referred to as thePhysical RandomAccessChannel (PRACH) indicating that, in contrast to theother random-access-related transmissions, the preamble corresponds to aspecialphysicalchannel.

16.2.1.1CharacteristicsofPreambleTransmissionSeveralfactorsimpactthestructureofthepreambletransmission.As described in Section 15.2, the transmission timing of NR uplink

transmissions is typically controlled by the network by means of regularlyprovidedtime-adjustmentcommands(“closed-looptimingcontrol”).Priortopreambletransmission,thereisnosuchclosed-looptimingcontrolin

operation.Rather,thedevicemustbasethepreambletransmissiontimingonthereceivedtimingofsomedownlinksignal,inpracticethereceivedtimingoftheacquiredSSblock.Consequently, therewill be anuncertainty in thepreamblereceptiontimingofatleasttwotimesthemaximumpropagationdelaywithinthecell.Forcellsizesintheorderofafewhundredmeters,thisuncertaintywillbeintheorderoffewmicroseconds.However,forlargecellstheuncertaintycouldbeintheorderof100μsorevenmore.In general, it is up to the base-station scheduler to ensure that there are no

othertransmissionsintheuplinkresourcesinwhichpreambletransmissionsmaytake place.When doing this, the network needs to take the uncertainty in thepreamble reception timing into account. In practice the scheduler needs toprovideanextraguardtimethatcapturesthisuncertainty(seeFig.16.7).

FIGURE16.7 Guard-timeforpreambletransmission.

Notethat thepresenceof theguardtimeisnotpartof theNRspecificationsbut just a resultof scheduling restrictions.Consequently,differentguard timescaneasilybeprovidedtomatchdifferentuncertaintyinthepreamblereceptiontiming,forexample,duetodifferentcellsizes.

Inadditiontothelackofclosed-looptimingcontrol, thereisalsonoclosed-looppowercontrol inoperationprior topreambletransmission.Rather,similarto the transmission timing, thedevicemustdetermine its transmitpowerbasedonthereceivedpowerofsomedownlinksignal,inpracticethereceivedpowerof theacquiredSSblock.Thelackofclosed-looppowercontrolmayleadtoarelativelylargeuncertaintyinthereceivedpreamblepowerforseveralreasons:

•Estimatesoftheabsolutereceivedpowerareinherentlyuncertain;•EspeciallyinthecaseofFDDoperationwithdownlinkanduplinkindifferentfrequencybands,therecanbesignificantdifferencesintheinstantaneousuplinkanddownlinkpathloss.

Finally, while normal uplink transmissions are typically based on explicitschedulinggrants,therebyenablingcontention-freeaccess,initialrandomaccessis inherently contention-based, implying that multiple devices may initiatepreamble transmissionsimultaneously.Thepreambleshouldpreferablybeabletohandle such a situation and asmuch aspossible allow for correct preamblereceptionwhensuch“collisions”occur.

16.2.1.2RACHResourcesWithinacell,preambletransmissioncantakeplacewithinaconfigurablesubsetof slots (theRACH slots) that repeats itself everyRACH configuration period(seeFig.16.8).9

FIGURE16.8 OverallRACHresourceconsistingofasetofconsecutiveresourceblockswithinasetofRACHslotsandwheretheslotpatternrepeatseveryRACH-resourceperiod.

Furthermore, within these “RACH slots”, there may bemultiple frequency-

domain RACH occasions jointly covering K·M consecutive resource blockswhereMisthepreamblebandwidthmeasuredinnumberofresourceblocksandKisthenumberoffrequency-domainRACHoccasions.For a given preamble type, corresponding to a certain preamble bandwidth,

the overall available time/frequencyRACH resourcewithin a cell can thus bedescribedby:

•AconfigurableRACHperiodicitythatcanrangefrom10msupto160ms;

•AconfigurablesetofRACHslotswithintheRACHperiod;•Aconfigurablefrequency-domainRACHresourcegivenbytheindexofthefirstresourceblockintheresourceandthenumberoffrequency-domainRACHoccasions.

16.2.1.3BasicPreambleStructureFig. 16.9 illustrates the basic structure for generating NR random-accesspreambles. A preamble is generated based on a length-L preamble sequence

which is DFT precoded before being applied to a conventionalOFDMmodulator.ThepreamblecanthusbeseenasaDFTS-OFDMsignal.Itshould be noted though that one could equally well see the preamble as aconventionalOFDMsignalbasedonafrequency-domainsequencebeingthediscreteFouriertransformofthesequence .

FIGURE16.9 BasicstructureforgenerationofNRrandom-accesspreamble.

Theoutputof theOFDMmodulator is then repeatedN times, afterwhichacyclicprefixisinserted.Forthepreamble,thecyclicprefixisthusnotinsertedperOFDMsymbolbutonlyoncefortheblockofNrepeatedsymbols.Differentpreamblesequencescanbeusedfor theNRpreambles.Similar to,

for example, uplink SRS, the preamble sequences are based on Zadoff–Chu

sequences [25].Asdescribed inSection8.3.1, forprime-lengthZCsequences,which is the case for the sequences used as a basis for the NR preamblesequences,thereareL–1differentsequences,witheachsequencecorrespondingtoauniquerootindex.Different preamble sequences can be generated from different Zadoff–Chu

sequencescorrespondingtodifferentroot indices.However,differentpreamblesequences can also be generated from different cyclic shifts of the same rootsequence. As described in Section 8.3.1, such sequences are inherentlyorthogonaltoeachother.However,thisorthogonalityisretainedatthereceiverside only if the relative cyclic shift between two sequences is larger than anydifference in their respective receive timing.Thus, inpracticeonlya subsetofthecyclicshiftscanbeusedtogeneratedifferentpreambles,wherethenumberofavailableshiftsdependson themaximumtiminguncertaintywhich, in turn,depends on, for example, the cell size. For small cell sizes a relatively largenumberofcyclicshiftscanoftenbeused.Forlargercells,asmallernumberofcyclicshiftswilltypicallybeavailable.Thesetofcyclicshiftsthatcanbeusedwithinacellisgivenbytheso-called

zero-correlation zone parameter which is part of the cell random-accessconfiguration provided within SIB1. In practice, the zero-correlation zoneparameterpointstoatablethatindicatesthesetofcyclicshiftsavailableinthecell. The name “zero-correlation zone” comes from the fact that the differenttablesindicatedbythezero-correlation-zoneparameterhavedifferentdistancesbetween the cyclic shifts, thusproviding larger or smaller “zones” in termsoftimingerrorforwhichorthogonality(=zerocorrelation)isretained.

16.2.1.4LongvsShortPreamblesNR defines two types of preambles, referred to as long preambles and shortpreambles,respectively.Asthenamesuggests,thetwopreambletypesdifferintermsofthelengthofthepreamblesequence.Theyalsodifferinthenumerology(subcarrierspacing)usedforthepreambletransmission.Thetypeofpreambleispartofthecellrandom-accessconfiguration,thatis,withinacellonlyonetypeofpreamblecanbeusedforinitialaccess.Long preambles are based on a sequence length L=839 and a subcarrier

spacingofeither1.25kHzor5kHz.ThelongpreamblesthususeanumerologydifferentfromanyotherNRtransmissions.ThelongpreamblespartlyoriginatefromthepreamblesusedforLTErandom-access[28].Longpreamblescanonlybeusedforfrequencybandsbelow6GHz(FR1).

As illustrated in Table 16.3 there are four different formats for the longpreamblewhereeachformatcorrespondstoaspecificnumerology(1.25kHzor5 kHz), a specific number of repetitions (the parameterN inFig. 16.9), and aspecificlengthofthecyclicprefix.Thepreambleformatisalsopartofthecellrandom-access configuration, that is, each cell is limited to a single preambleformat.ItcouldbenotedthatthetwofirstformatsofTable16.3areidenticaltotheLTEpreambleformats0and2[14].

Table16.3

In the previous section it was described how the overall RACH resourceconsistsofasetofslotsandresourceblocksinthetime-domainandfrequency-domain, respectively. For long preambles, which use a numerology that isdifferent from other NR transmissions, the slot and resource block should beseenfroma15kHznumerologypointofview.Inthecontextoflongpreambles,a slot thus has a length of 1 ms, while a resource-block has a bandwidth of180kHz.Alongpreamblewith1.25kHznumerologythusoccupiessixresourceblocks in the frequency domain, while a preamble with 5 kHz numerologyoccupies24resourceblocks.It can be observed that preamble format 1 and preamble format 2 in Table

16.3 correspond to a preamble length that exceeds a slot. Thismay appear tocontradicttheassumptionofpreambletransmissionstakingplaceinRACHslotsoflength1msasdiscussedinSection16.2.1.2.However,theRACHslotsonlyindicatethepossiblestartingpositionsforpreambletransmission.Ifapreambletransmissionextendsintoasubsequentslot,thisonlyimpliesthattheschedulerneedstoensurethatnoothertransmissionstakeplacewithinthecorrespondingfrequency-domainresourceswithinthatslot.Shortpreamblesarebasedonasequence lengthL=139andusea subcarrier

spacingalignedwiththenormalNRsubcarrierspacing.Morespecifically,shortpreamblesuseasubcarrierspacingof:

•15kHzor30kHzinthecaseofoperationbelow6GHz(FR1);•60kHzor120kHzinthecaseofoperationinthehigherNRfrequency

bands(FR2).

In the case of short preambles, the RACH resource described in Section16.2.1.2 is based on the same numerology as the preamble.A short preamblethusalwaysoccupies12resourceblocksinthefrequencydomainregardlessofthepreamblenumerology.Table16.4liststhepreambleformatsavailableforshortpreambles.Thelabels

for the different preamble formats originate from the 3GPP standardizationdiscussionsduringwhichanevenlargersetofpreambleformatswerediscussed.The table assumes a preamble subcarrier spacing of 15 kHz. For othernumerologies, the length of the preamble as well as the length of the cyclicprefixscalecorrespondingly,thatis,withtheinverseofthesubcarrierspacing.

Table16.4

Theshortpreamblesare,ingeneral,shorterthanthelongpreamblesandoftenspanonlyafewOFDMsymbols. Inmostcases it is thereforepossible tohavemultiplepreambletransmissionsmultiplexedintimewithinasingleRACHslot.In other words, for short preambles there may not only be multiple RACHoccasions in the frequencydomainbutalso in the timedomainwithinasingleRACHslot(seeTable16.5).

Table16.5

ItcanbenotedthatTable16.5includesadditionalformatsA1/B1,A2/B2,andA3/B3.Theseformatscorrespondtotheuseofamixofthe“A”and“B”formatsofTable16.4,wheretheAformatisusedforallexceptthelastRACHoccasion

within a RACH slot. Note that the A and B preamble formats are identicalexceptforasomewhatshortercyclicprefixfortheBformats.ForthesamereasontherearenoexplicitformatsB2andB3inTable16.5as

theseformatsarealwaysusedincombinationwiththecorrespondingAformats(A2andA3)accordingtotheabove.

16.2.1.5BeamEstablishmentDuringInitialAccessAkey feature of theNR initial access is the possibility to establish a suitablebeam pair already during the initial-access phase and to apply receiver-sideanalogbeam-sweepingforthepreamblereception.This is enabled by the possibility of associating different SS-block time

indiceswithdifferentRACHtime/frequencyoccasionsand/ordifferentpreamblesequences.AsdifferentSS-blocktimeindicesinpracticecorrespondtoSS-blocktransmissions in different downlinkbeams, thismeans that thenetwork, basedonthereceivedpreamble,willbeabletodeterminethedownlinkbeaminwhichthe corresponding device is located. This beam can then be used as an initialbeamforsubsequentdownlinktransmissionstothedevice.Furthermore, if the association between SS-block time index and RACH

occasion is such thatagiven time-domainRACHoccasioncorresponds toonespecific SS-block time index, the networkwill knowwhen, in time, preambletransmission from devices within a specific downlink beam will take place.Assumingbeamcorrespondence,thenetworkcanthenfocustheuplinkreceiverbeam in the corresponding direction for beam-formed preamble reception. Inpracticethisimpliesthatthereceiverbeamwillbesweptoverthecoverageareasynchronized with the corresponding downlink beam sweep for the SS-blocktransmission.Note that beam-sweeping for preamble transmission is only relevant when

analogbeam-forming isappliedat the receiverside. Ifdigitalbeam-forming isapplied,beam-formedpreamblereceptioncanbedonefrommultipledirectionssimultaneously.To associate a certain SS-block time index with a specific random-access

occasionandaspecificsetofpreambles,therandom-accessconfigurationofthecell specifies the number of SS-block time indices per RACH time/frequencyoccasion.Thisnumbercanbelargerthanone,indicatingthatmultipleSS-blocktimeindicescorrespondtoasingleRACHtime/frequencyoccasion.However,itcan also be smaller than one, indicating that one single SS-block time indexcorrespondstomultipleRACHtime/frequencyoccasions.

SS-block time indices are then associated with RACH occasions in thefollowingorder:

•Firstinthefrequencydomain;•Theninthetimedomainwithinaslot,assumingthepreambleformatconfiguredforthecellallowsformultipletime-domainRACHoccasionswithinaslot(onlyrelevantforshortpreambles);

•FinallyinthetimedomainbetweenRACHslots.

Fig. 16.10 exemplifies the association between SS-block time indices andRACHoccasionsunderthefollowingassumptions:

•TwoRACHfrequencyoccasions;•ThreeRACHtimeoccasionsperRACHslot;•EachSS-blocktimeindexassociatedwithfourRACHoccasions.

FIGURE16.10 AssociationbetweenSS-blocktimeindicesandRACHoccasions(example).

16.2.1.6PreamblePowerControlandPowerRampingAsdiscussedabove,preambletransmissionwilltakeplacewitharelativelylargeuncertainty in the required preamble transmit power. Preamble transmissiontherefore includes a power-ramping mechanism where the preamble may berepeatedly transmitted with a transmit power that is increased between eachtransmission.Thedeviceselects the initialpreamble transmitpowerbasedonestimatesof

the downlink path loss in combinationwith a target received preamble powerconfigured by the network. The path loss should be estimated based on thereceivedpowerof theSSblockthat thedevicehasacquiredandfromwhichit

hasdeterminedtheRACHresourcetouseforthepreambletransmission.Thisisaligned with an assumption that if the preamble transmission is received bymeans of beam-forming the corresponding SS block is transmitted with acorresponding beam-shaper. If no random-access response (see below) isreceived within a predetermined window, the device can assume that thepreamblewasnotcorrectlyreceivedbythenetwork,mostlikelyduetothefactthat the preamble was transmitted with too low power. If this happens, thedevice repeats the preamble transmission with the preamble transmit powerincreasedbyacertainconfigurableoffset.Thispowerrampingcontinuesuntilarandom-access response has been received or until a configurable maximumnumber of retransmissions has been carried out, alternatively a configurablemaximumpreambletransmitpowerhasbeenreached.Inthetwolattercases,therandom-accessattemptisdeclaredasafailure.

16.2.2Random-AccessResponseOnceadevicehastransmittedarandom-accesspreamble,itwaitsforarandom-access response, that is, a response from the network that it has properlyreceived the preamble. The random-access response is transmitted as aconventional downlink PDCCH/PDSCH transmission with the correspondingPDCCHtransmittedwithinthecommonsearchspace.Therandom-accessresponseincludesthefollowing:

•Informationabouttherandom-accesspreamblesequencethenetworkdetectedandforwhichtheresponseisvalid;

•Atimingcorrectioncalculatedbythenetworkbasedonthepreamblereceivetiming;

•Aschedulinggrant,indicatingresourcesthedevicewilluseforthetransmissionofthesubsequentMessage3(seebelow);

•Atemporaryidentity,theTC-RNTI,usedforfurthercommunicationbetweenthedeviceandthenetwork.

If the network detects multiple random-access attempts (from differentdevices), the individual response messages can be combined in a singletransmission.Therefore,theresponsemessageisscheduledontheDL-SCHandindicatedonaPDCCHusingan identity reserved for random-access response,theRA-RNTI.Theuseof theRA-RNTI isalsonecessaryasadevicemaynot

haveauniqueidentityintheformofaC-RNTIallocated.Alldevicesthathavetransmitted a preamblemonitor theL1/L2 control channels for random-accessresponse within a configurable time window. The timing of the responsemessageisnotfixedinthespecificationinordertobeabletorespondtomanysimultaneous accesses. It also provides some flexibility in the base-stationimplementation. If thedevicedoesnotdetecta random-access responsewithinthe time window, the preamble will be retransmitted with higher poweraccordingtothepreamblepowerrampingdescribedabove.As long as the devices that performed random access in the same resource

used different preambles, no collision will occur and from the downlinksignalingitiscleartowhichdevice(s)theinformationisrelated.However,thereis a certainprobabilityof contention—that is,multipledevicesusing the samerandom-access preamble at the same time. In this case, multiple devices willreact upon the same downlink response message and a collision occurs.Resolvingthesecollisionsispartofthesubsequentsteps,asdiscussedbelow.Upon reception of the random-access response, the device will adjust its

uplink transmission timing and continue to the third step. If contention-freerandomaccessusingadedicatedpreambleisused,thenthisisthelaststepoftherandom-accessprocedureas there isnoneed tohandlecontention in thiscase.Furthermore,thedevicealreadyhasauniqueidentityallocatedintheformofaC-RNTI.In the case of downlink beam-forming, the random-access response should

followthebeam-formingusedfor theSSblockwhichwasacquiredduringtheinitial cell search.This is important as the devicemayuse receive-side beam-forminganditneedstoknowhowtodirect thereceiverbeam.Bytransmittingthe random-access response using the same beam as the SS block, the deviceknowsthatitcanusethesamereceiverbeamasidentifiedduringthecellsearch.

16.2.3Message3:ContentionResolutionAfter thesecondstep, theuplinkof thedevice is timesynchronized.However,beforeuserdatacanbetransmittedto/fromthedevice,auniqueidentitywithinthecell,theC-RNTI,mustbeassignedtothedevice(unlessthedevicealreadyhas a C-RNTI assigned). Depending on the device state, theremay also be aneedforadditionalmessageexchangeforsettinguptheconnection.Inthethirdstep,thedevicetransmitsthenecessarymessagestothegNBusing

the UL-SCH resources assigned in the random-access response in the second

step.Animportantpartoftheuplinkmessageistheinclusionofadeviceidentity,

as this identity is used as part of the contention-resolution mechanism in thefourthstep.Ifthedeviceisalreadyknownbytheradio-accessnetwork,thatis,inRRC_CONNECTEDorRRC_INACTIVEstate,thealready-assignedC-RNTIis used as the device identity.10Otherwise, a core-network device identifier isusedandthegNBneedstoinvolvethecorenetworkpriortorespondingtotheuplinkmessageinstep4(seebelow).

16.2.4Message4:ContentionResolutionandConnectionSetUpThelaststepintherandom-accessprocedureconsistsofadownlinkmessageforcontention resolution. Note that, from the second step, multiple devicesperforming simultaneous random-access attempts using the same preamblesequenceinthefirststeplistentothesameresponsemessageinthesecondstepand thereforehave thesametemporary identifier.Hence, thefourthstep in therandom-accessprocedureisacontention-resolutionsteptoensurethatadevicedoes not incorrectly use another device’s identity. The contention resolutionmechanism differs somewhat depending on whether the device already has avalididentityintheformofaC-RNTIornot.Notethatthenetworkknowsfromtheuplinkmessagereceivedinstep3whetherthedevicehasavalidC-RNTIornot.IfthedevicealreadyhadaC-RNTIassigned,contentionresolutionishandled

byaddressing thedeviceon thePDCCHusing theC-RNTI.UpondetectionofitsC-RNTI on the PDCCH the devicewill declare the random-access attemptsuccessfulandthereisnoneedforcontention-resolution-relatedinformationontheDL-SCH.SincetheC-RNTIisuniquetoonedevice,unintendeddeviceswillignorethisPDCCHtransmission.IfthedevicedoesnothaveavalidC-RNTI,thecontentionresolutionmessage

is addressed using the TC-RNTI and the associated DL-SCH contains thecontention-resolution message. The device will compare the identity in themessage with the identity transmitted in the third step. Only a device whichobservesamatchbetweentheidentityreceivedinthefourthstepandtheidentitytransmitted as part of the third stepwill declare the random-access proceduresuccessfulandpromotetheTC-RNTIfromthesecondsteptotheC-RNTI.Sinceuplinksynchronizationhasalreadybeenestablished,hybridARQ isapplied to

thedownlinksignalinginthisstepanddeviceswithamatchbetweentheidentitytheytransmittedinthethirdstepandthemessagereceivedinthefourthstepwilltransmitahybrid-ARQacknowledgmentintheuplink.DevicesthatdonotdetectPDCCHtransmissionwiththeirC-RNTIordonot

findamatchbetweentheidentityreceivedinthefourthstepandtherespectiveidentity transmitted as part of the third step are considered to have failed therandom-accessprocedure andneed to restart theprocedure from the first step.No hybrid-ARQ feedback is transmitted from these devices. Furthermore, adevicethathasnotreceivedthedownlinkmessageinstep4withinacertaintimefromthetransmissionoftheuplinkmessageinstep3willdeclaretherandom-accessprocedureasfailedandneedtorestartfromthefirststep.

16.2.5RandomAccessforSupplementaryUplinkSection7.7discussedtheconceptofsupplementaryuplink(SUL),thatis,thatadownlink carrier may be associated with two uplink carriers (the non-SULcarrierandtheSULcarrier),wheretheSULcarrieristypicallylocatedinlower-frequencybandstherebyprovidingenhanceduplinkcoverage.Thatacell isanSULcell, that is, includesacomplementarySULcarrier, is

indicated as part of SIB1. Before initially accessing a cell, a devicewill thusknowifthecelltobeaccessedisanSULcellornot.IfthecellisanSULcelland thedevice supportsSULoperation for thegivenbandcombination, initialrandomaccessmaybecarriedoutusingeithertheSULcarrierorthenon-SULuplink carrier. The cell system information provides separate RACHconfigurationsfortheSULcarrierandthenon-SULcarrierandadevicecapableofSULdetermineswhatcarriertousefortherandomaccessbycomparingthemeasuredRSRPoftheselectedSSblockwithacarrier-selectionthresholdalsoprovidedaspartofthecellsysteminformation.

•IftheRSRPisabovethethreshold,randomaccessiscarriedoutonthenon-SULcarrier.

•IftheRSRPisbelowthethreshold,randomaccessiscarriedoutontheSULcarrier.

In practice the SUL carrier is thus selected by devices with a (downlink)pathlosstothecellthatislargerthanacertainvalue.

The device carrying out a random-access transmission will transmit therandom-access message 3 on the same carrier as used for the preambletransmission.Forotherscenarioswhenadevicemaydoarandomaccess,thatis,fordevices

inconnectedmode,thedevicecanbeexplicitlyconfiguredtouseeithertheSULcarrierorthenon-SULcarrierfortheuplinkrandom-accesstransmissions.

1SometimesonlyPSSandSSSareincludedintheterm“SSblock.”HerewewillrefertothetripletPSS,SSS,andPBCHasanSSblockthough.2EventhoughthetermsPSS,SSS,andPBCHareusedalsoinLTE,thetermSSblockisnotusedwithinthecontextofLTE.3Notethat,althoughthefrequencyrangefor30kHzSS-blocknumerologyfullyoverlapswiththefrequencyrangefor15kHznumerology,foragivenfrequencybandwithinthelowerfrequencyrangethereisinmanycasesonlyasinglenumerologysupported.4ThetermSSburstsetoriginatesfromearly3GPPdiscussionswhenSSblockswereassumedtobegroupedintoSSburstsandtheSSburststhengroupedintoSSburstsets.TheintermediateSS-burstgroupingwaseventuallynotusedbutthetermSSburstsetforthefullsetofSSblockswasretained.5Foroperationbelow3GHz,theSSblockcanonlybelocatedwithinthefirsttwoslots.6SomeoftheinformationonthePBCHisstrictlyspeakingnotpartoftheMIB(seealsobelow).7OnlyuptofourSSblocksforoperationbelow3GHz.8AstheSFNisupdatedevery10msitwouldhavebeensufficienttoplacethethreeleastsignificantbitsoftheSFNoutsideoftheMIB.9AswillbeseeninSection16.1.5,apreambletransmissionmayactuallyextendoutsideRACHslots,thatis,strictlyspeakingtheRACHslotsdefinepossiblestartingpointsofpreambletransmissions.10ThedeviceidentityisincludedasaMACcontrolelementontheUL-SCH.

CHAPTER17

LTE/NRInterworkingandCoexistence

Abstract

This chapter describe LTE/NR interworking including the differentarchitecture options for LTE/NR dual-connectivity. Key deploymentscenariosforLTE/NRdual-connectivityisalsodescribed.ThechapteralsodescribesthemechanismsforLTE/NRspectrumcoexistence.

KeywordsLTE/NRinterworking;LTE/NRdual-connectivity;LTE/NRcoexistence;single-TXoperation

The initial deployment of a new generation of mobile-communicationtechnologytypicallytakesplaceinareaswithhightrafficdensityandwithhighdemandsfornewservicecapabilities.Thisisthenfollowedbyagradualfurtherbuild-out that can be more or less rapid depending on the operator strategy.Duringthissubsequentgradualdeployment,ubiquitouscoveragetotheoperatornetworkwillbeprovidedbyamixofnewandlegacytechnology,withdevicescontinuously moving in and out of areas covered by the new technology.Seamlesshandoverbetweennewandlegacytechnologyhasthereforebeenakeyrequirementatleastsincetheintroductionofthefirst3Gnetworks.Furthermore,eveninareaswhereanewtechnologyhasbeendeployed,earlier

generationsmust typically be retained and operated in parallel for a relativelylongtimeinordertoensurecontinuedserviceforlegacydevicesnotsupportingthe new technology. The majority of users will migrate to new devicessupportingthelatesttechnologywithinafewyears.However,alimitedamountoflegacydevicesmayremainforalongtime.Thisbecomesevenmorethecasewithanincreasingnumberofmobiledevicesnotbeingdirectlyusedbypersons

but ratherbeingan integratedpartofotherequipment,suchasparkingmeters,cardreaders,surveillancecameras,etc.Suchequipmentmayhavealifetimeofmorethan10yearsandwillbeexpectedtoremainconnectableduringthis lifetime.This isactuallyone important reasonwhymanysecond-generationGSMnetworks are still in operation even though both 3G and 4G networks havesubsequentlybeendeployed.However, the interworking between NR and LTE goes further than just

enablingsmoothhandoverbetweenthetwotechnologiesandallowingfortheirparalleldeployment.

•NRallowsfordual-connectivitywithLTE,implyingthatdevicesmayhavesimultaneousconnectivitytobothLTEandNR.AsalreadymentionedinChapter5,thefirstreleaseofNRactuallyreliesonsuchdual-connectivity,withLTEprovidingthecontrolplaneandNRonlyprovidingadditionaluser-planecapacity;

•NRcanbedeployedinthesamespectrumasLTEinsuchawaythattheoverallspectrumcapacitycanbedynamicallysharedbetweenthetwotechnologies.SuchspectrumcoexistenceallowsforamoresmoothintroductionofNRinspectraalreadyoccupiedbyLTE.

17.1LTE/NRDual-ConnectivityThe basic principle of LTE/NR dual-connectivity is the same as LTE dual-connectivity[28],seealsoFig.17.1):

•Adevicehassimultaneousconnectivitytomultiplenodeswithintheradio-accessnetwork(eNBinthecaseofLTE,gNBinthecaseofNR);

•Thereisonemasternode(inthegeneralcaseeitheraneNBoragNB)responsiblefortheradio-accesscontrolplane.Inotherwords,onthenetworksidethesignalingradiobearerterminatesatthemasternodewhichthenalsohandlesallRRC-basedconfigurationofthedevice;

•Thereisone,orinthegeneralcasemultiple,secondarynode(s)(eNBorgNB)thatprovidesadditionaluser-planelinksforthedevice.

FIGURE17.1 Basicprincipleofdual-connectivity.

17.1.1DeploymentScenariosInthecaseofLTEdual-connectivity,themultiplenodestowhichadevicehassimultaneous connectivity are typically geographically separated. The devicemay, for example, have simultaneous connectivity to a small-cell layer and anoverlaidmacrolayer.Thesamescenario,thatis,simultaneousconnectivitytoasmall-celllayerand

an overlaid macrolayer, is a highly relevant scenario also for LTE/NR dual-connectivity. Especially,NR in higher-frequency bandsmay be deployed as asmall-celllayerunderanexistingmacrolayerbasedonLTE(seeFig.17.2).TheLTEmacrolayerwouldthenprovidethemasternodes,ensuringthatthecontrolplaneisretainedeveniftheconnectivitytothehigh-frequencysmall-celllayeristemporarilylost.Inthiscase,theNRlayerprovidesveryhighcapacityandveryhigh data rates, while dual-connectivity to the lower-frequency LTE-basedmacro layer provides additional robustness to the inherently less robust high-frequencysmall-celllayer.NotethatthisisessentiallythesamescenarioastheLTE dual-connectivity scenario described above, except for the use of NRinsteadofLTEinthesmall-celllayer.

FIGURE17.2 LTE/NRdual-connectivityinamulti-layerscenario.

However, LTE/NRdual-connectivity is also relevant in the case of co-sitedLTE and NR network nodes (Fig. 17.3).1 As an example, for initial NRdeployment an operatormaywant to reuse an already deployedLTE site gridalsoforNRtoavoidthecostofdeployingadditionalsites.Inthisscenario,dual-connectivity enables higher end-user data rates by allowing for aggregation ofthethroughputoftheNRandLTEcarriers.Inthecaseofasingleradio-accesstechnology, suchaggregationbetweencarriers transmitted from the samenodewouldbemoreefficientlyrealizedbymeansofcarrieraggregation(seeSection7.6).However,NRdoesnotsupportcarrieraggregationwithLTEandthusdual-connectivityisneededtosupportaggregationoftheLTEandNRthroughput.

FIGURE17.3 LTE/NRdual-connectivity,co-siteddeployment.

Co-siteddeploymentsareespeciallyrelevantwhenNRisoperatinginlower-frequencyspectrum,thatis, inthesameorsimilarspectrumasLTE.However,co-siteddeploymentscanalsobeusedwhenthetwotechnologiesareoperatinginverydifferentspectra,includingthecasewhenNRisoperatinginmm-wave

bands(Fig.17.4).Inthiscase,NRmaynotbeabletoprovidecoverageovertheentirecellarea.However,theNRpartofthenetworkcouldstillcapturealargepart of the overall traffic, thereby allowing for the LTE part to focus onprovidingservicetodevicesinpoor-coveragelocations.

FIGURE17.4 LTE/NRdual-connectivity,co-siteddeploymentindifferentspectrum.

InthescenarioinFig.17.4,theNRcarrierwouldtypicallyhavemuchwiderbandwidthcomparedtoLTE.Aslongasthereiscoverage,theNRcarrierwouldtherefore, in most cases, provide significantly higher data rates compared toLTE,makingthroughputaggregationlessimportant.Rather,themainbenefitofdual-connectivityinthisscenariowould,onceagain,beenhancedrobustnessforthehigher-frequencydeployment.

17.1.2ArchitectureOptionsDuetothepresenceoftwodifferentradio-accesstechnologies(LTEandNR)aswellasthefutureavailabilityofanew5Gcorenetworkasanalternativetothelegacy 4G core network (EPC), there are several different alternatives, oroptions, for the architecture ofLTE/NRdual-connectivity (seeFig. 17.5).Thelabeling of the different options in Fig. 17.5 originates from early 3GPPdiscussions on possible NR architecture options where a number of differentalternativeswere on the table, a subset ofwhichwas eventually agreed to besupported(seeChapter6forsomeadditional,non-dual-connectivity,options).

FIGURE17.5 DifferentarchitectureoptionsforLTE/NRdual-connectivity.

ItcanbenotedthatLTE/NRdual-connectivityusingEPCwithNRprovidingthemasternodeisnotincludedamongtheoptionsoutlinedinFig.17.5.Atthetimeof thewritingof thisbook, thepossiblesupportfor thisalternativeisstillunderdiscussion.

17.1.3Single-TXOperationIn the case of dual-connectivity between LTE and NR there will be multipleuplink carriers (at least one LTE uplink carrier and one NR uplink carrier)transmitted from the same device. Due to non-linearities in the RF circuitry,simultaneous transmissionon twocarrierswill create intermodulationproductsat the transmitter output. Depending on the specific carrier frequencies of thetransmittedsignals,someof these intermodulationproductsmayendupwithinthe device receiver band causing “self-interference,” also referred to asintermodulationdistortion (IMD).The IMDwill add to the receivernoise andlead to adegradationof the receiver sensitivity.The impact from IMDcanbereducedbyimposingtighterlinearityrequirementsonthedevice.However,thiswill have a corresponding negative impact on device cost and energyconsumption.ToreducetheimpactofIMDwithoutimposingverytightRFrequirementson

all devices, NR allows for single-TX dual-connectivity for “difficult bandcombinations.” In this context, difficult band combinations correspond tospecificallyidentifiedcombinationsofLTEandNRfrequencybandsforwhichlower-order intermodulationproducts between simultaneously transmittedLTEandNRuplinkcarriersmayfallintoacorrespondingdownlinkband.Single-TXoperation implies that therewill not be simultaneous transmissionon theLTEandNRuplinkcarrierswithin adevice even though thedevice isoperating inLTE/NRdual-connectivity.It is the taskof theLTEandNRschedulers to jointlyprevent simultaneous

transmission on the LTE and NR uplink carriers in the case of single-TX

operation.Thisrequirescoordinationbetweentheschedulers,thatisbetweenaneNB and a gNB. The 3GPP specifications include explicit support for theinterchangeofstandardizedinter-nodemessagesforthispurpose.SingleTXoperation inherently leads to timemultiplexingbetween theLTE

andNR uplink transmissionswithin a device, with none of the uplinks beingcontinuously available. However, it is still desirable to be able to retain fullutilizationofthecorrespondingdownlinkcarriers.ForNR,with itshighdegreeof schedulingandhybrid-ARQflexibility, this

caneasilybeachievedwithnoadditionalimpactontheNRspecifications.FortheLTEpartoftheconnectionthesituationissomewhatdifferentthough.LTEFDD is based on synchronousHARQ,where uplinkHARQ feedback is to betransmitted a specified number of subframes after the reception of thecorresponding downlink transmission. With a single-TX constraint, not alluplink subframes will be available for transmission of HARQ feedback,potentially restricting the subframes in which downlink transmission can takeplace.However, the same situation may already occur within LTE itself, more

specifically in thecaseofFDD/TDDcarrier aggregationwith theTDDcarrierbeingtheprimarycell[28].Inthiscase,theTDDcarrier,whichisinherentlynotcontinuously available for uplink transmission, carries uplinkHARQ feedbackcorresponding to downlink transmissions on the FDD carrier. To handle thissituation,LTErelease13 introducedso-calledDL/ULreferenceconfigurations[28]allowingforaTDD-like timingrelation, forexampleforuplinkfeedback,for anFDDcarrier.The same functionality canbeused to support continuousLTE downlink transmission in the case of LTE/NR dual-connectivityconstrainedbysingle-TXoperation.IntheLTEFDD/TDDcarrier-aggregationsscenario,theuplinkconstraintsare

duetocell-leveldownlink/uplinkconfigurations.Ontheotherhand,inthecaseof single-TX dual-connectivity the constraints are due to the need to avoidsimultaneoustransmissionontheLTEandNRuplinkcarriers,butwithoutanytight interdependencybetweendifferent devices.The set of unavailable uplinksubframesmaythusnotneedtobe thesamefordifferentdevices.Toenableamoreeven loadon theLTEuplink, theDL/ULreferenceconfigurations in thecase of single-TX operation can therefore be shifted in time on a per-devicebasis.

17.2LTE/NRCoexistence

17.2LTE/NRCoexistenceThe introduction of earlier generations of mobile communication has alwaysbeen associated with the introduction of a new spectrum in which the newtechnologycanbedeployed.ThisisthecasealsoforNR,forwhichthesupportforoperationinmm-wavebandsopensupfortheuseofaspectrumrangeneverbeforeappliedtomobilecommunication.Even taking into account the use of antenna configurations with a large

numberofantennaelementsenablingextensivebeamforming,operationinsuchhigh-frequency spectrum is inherently disadvantageous in terms of coverage.Rather,toprovidetrulywide-areaNRcoverage,lower-frequencyspectrummustbeused.However, most lower-frequency spectrum is already occupied by current

technologies, primarily LTE. Furthermore, an additional low-frequencyspectrum is planned to be deployedwithLTE in the relatively near future. InmanycasesNRdeploymentsinlower-frequencyspectrumwillthereforeneedtotakeplaceinspectrumalreadyusedbyLTE.Themost straightforwardway to deployNR in a spectrum already used by

LTE is static frequency-domain sharing, where part of the LTE spectrum ismigratedtoNR(seeFig.17.6).

FIGURE17.6 MigrationofLTEspectrumtoNR.

Therearetwodrawbackswiththisapproachthough.Atleastataninitialstage,themainpartofthetrafficwillstillbeviaLTE.At

the same time, the static frequency-domain sharing reduces the spectrumavailableforLTE,makingitmoredifficulttosatisfythetrafficdemands.Furthermore, static frequency-domain sharing will lead to less bandwidth

being available for each technology, leading to a reduced peak data rate percarrier.ThepossibleuseofLTE/NRdual-connectivitymaycompensateforthisfor new devices capable of such operation. However, at least for legacy LTEdevicestherewillbeadirectimpactontheachievabledatarates.AmoreattractivesolutionistohaveNRandLTEdynamicallysharethesame

spectrum as illustrated in Fig. 17.7. Such spectrum coexistencewill retain thefull bandwidth and corresponding peak data rates for each technology.Furthermore, the overall spectrum capacity could be dynamically assigned tomatchthetrafficconditionsoneachtechnology.

FIGURE17.7 LTE/NRspectrumcoexistence.

The fundamental tool to enable such LTE/NR spectrum coexistence is thedynamicschedulingofbothLTEandNR.However,thereareseveralotherNRfeaturesthatplayaroleintheoverallsupportforLTE/NRspectrumcoexistence:

•TheavailabilityoftheLTE-compatible15kHzNRnumerologythatallowsforLTEandNRtooperateonacommontime/frequencygrid;

•ThegeneralNRforward-compatibilitydesignprincipleslistedinSection5.1.3.ThisalsoincludesthepossibilitytodefinereservedresourcesbasedonbitmapsasdescribedinSection9.10;

•ApossibilityforNRPDSCHmappingtoavoidresourceelementscorrespondingtoLTEcell-specificreferencesignals(seefurtherdetailsbelow).

As already mentioned in Section 5.1.11 there are two main scenarios forLTE/NRcoexistence(seealsoFig.17.8):

•Coexistenceinbothdownlinkanduplink;•Uplink-onlycoexistence.

FIGURE17.8 Downlink/uplinkcoexistencevsuplink-onlycoexistence.

A typical use case for uplink-only coexistence is the deployment of asupplementaryuplinkcarrier(seeSection7.7).In general, coexistence in the uplink direction is more straightforward

comparedtothedownlinkdirectionandcan,toalargeextent,besupportedbymeans of scheduling coordination/constraints. NR and LTE uplink schedulingshould be coordinated to avoid collision between LTE and NR PUSCHtransmissions. Furthermore, the NR scheduler should be constrained to avoidresources used for LTE uplink layer 1 control signaling (PUCCH) and viceversa.Depending on the level of interaction between the eNB and gNB, suchcoordinationandconstraintscanbemoreorlessdynamic.Also for the downlink, scheduling coordination should be used to avoid

collision between scheduled LTE and NR transmissions. However, the LTEdownlink also includes several non-scheduled “always-on” signals that cannotbereadilyscheduledaround.Thisincludes(see[28]fordetails):

•TheLTEPSSandSSS,whicharetransmittedovertwoOFDMsymbolsandsixresourceblocksinthefrequencydomainonceeveryfifthsubframe;

•TheLTEPBCH,whichistransmittedoverfourOFDMsymbolsandsixresourceblocksinthefrequencydomainonceeveryframe(10subframes);

•TheLTECRS,whichistransmittedregularlyinthefrequencydomainandinfourorsixsymbolsineverysubframedependingonthenumberofCRSantennaports.2

Rather than being avoided bymeans of scheduling, the concept or reservedresources(seeSection9.10)canbeused to ratematch theNRPDSCHaroundthesesignals.Ratematching around the LTE PSS/SSS can be done by defining reserved

resourcesaccordingtobitmapsasdescribedinSection9.10.Morespecificallya

singlereservedresourcegivenbya{bitmap-1,bitmap-2,bitmap-3}tripletcouldbedefinedasfollows(seealsoFig.17.9):

•Abitmap-1ofalengthequaltothenumberofNRresourceblocksinthefrequencydomain,indicatingthesixresourceblockswithinwhichLTEPSSandSSSaretransmitted;

•Abitmap-2oflength14(oneslot),indicatingthetwoOFDMsymbolswithinwhichthePSSandSSSaretransmittedwithinanLTEsubframe;

•Abitmap-3oflength10indicatingthetwosubframeswithinwhichthePSSandSSSaretransmittedwithina10msframe.

FIGURE17.9 ConfigurationofreservedresourcetoenablePDSCHratematchingaroundLTEPSS/SS.Notethatthefigureassumes15kHzNRnumerology.

Thisassumesa15kHzNRnumerology.Notethoughthattheuseofreservedresources based on bitmaps is not limited to 15 kHz numerology and, inprinciple,asimilarapproach to ratematcharoundLTEPSSandSSScouldbeusedalsowith,forexample,a30kHzNRnumerology.ThesameapproachcanbeusedtoratematcharoundtheLTEPBCHwiththe

only difference that bitmap-2 would, in this case, indicate the four symbolswithin which PBCH is transmitted, while bitmap-3 would indicate a singlesubframe.Regarding the LTECRS, theNR specification includes explicit support for

PDSCH rate matching around resource elements corresponding to CRS of anoverlaidLTEcarrier.Inordertobeabletoproperlyreceivesucharate-matchedPDSCH,thedeviceisconfiguredwiththefollowinginformation:

•TheLTEcarrierbandwidthandfrequencydomainlocation,toallowforLTE/NRcoexistenceeventhoughtheLTEcarriermayhaveadifferentbandwidthandadifferentcenter-carrierlocation,comparedtotheNR

carrier;•TheLTEMBSFNsubframeconfiguration,asthiswillinfluencethesetofOFDMsymbolsinwhichCRStransmissiontakesplacewithinagivenLTEsubframe;

•ThenumberofLTECRSantennaportsasthiswillimpactthesetofOFDMsymbolsonwhichCRStransmissiontakesplaceaswellasthenumberofCRSresourceelementsperresourceblockinthefrequencydomain;

•TheLTECRSshift,thatis,theexactfrequency-domainpositionoftheLTECRS.

Rate matching around LTE CRS is only possible for the 15 kHz NRnumerology.

1Notethatthreewouldinthiscasestillbetwodifferentlogicalnodes(aneNBandagNB)althoughthesecouldverywellbeimplementedinthesamephysicalhardware.2Onlyoneortwosymbolsincaseofso-calledMBSFNsubframes.

CHAPTER18

RFCharacteristics

Abstract

ThischapterpresentstheRFrequirementsthatdefinetheRFcharacteristicsofbothbasestationsanddevices.Bothanoverviewandfurtherdetailsoftransmitter and receiver requirements are given, including how they aresubdividedintoconductedandradiatedrequirements.

KeywordsRFcharacteristics;RFrequirements;soectrumflexibility;frequencyrange;radiated;conducted;BStypes;BSclasses;multistandard;multiband;non-contiguous

TheRFcharacteristicsofNRarestronglytiedtothespectrumavailablefor5Gas described in Chapter 3 and the spectrum flexibility required to operate inthosespectrumallocations.Whilespectrumflexibilityhasbeenacornerstoneforpreviousgenerationsofmobilesystems,thisbecomesevenmoreaccentuatedforNR. It consistsof severalcomponents, includingdeployment indifferent-sizedspectrumallocationsanddiversefrequencyranges,bothinpairedandunpairedfrequencybandsandwithaggregationofdifferentfrequencyallocationswithinand between different bands.NRwill also have the capability to operatewithmixed numerology on the same RF carrier and will have an even higherflexibilitythanLTEintermsoffrequencydomainschedulingandmultiplexingofdeviceswithin abase stationRFcarrier. It is theuseofOFDM inNR thatgivesflexibilitybothintermsofthesizeofthespectrumallocationneededandin the instantaneous transmission bandwidth used, and that enables frequency-domainscheduling.ImplementationofActiveAntennaSystems (AAS)andmultipleantennas in

deviceshasbeeninuseforLTE,butistakenonestepfurtherinNR,whichwill

supportmassiveMIMOandbeam-forming applications both in existing bandsand in the newmm-wave bands. Beyond the physical layer implications, thisimpacts theRF implementation in termsof filters,amplifiers,andallotherRFcomponentsthatareusedtotransmitandreceivethesignalandmustbedefinedtakingalsothespectrumflexibilityintoaccount.ThesearefurtherdiscussedinChapter19.Note that for the purpose of defining RF characteristics, the physical

representationofthegNBiscalledabasestation(BS).Abasestationisdefinedwith interfaces where the RF requirements are defined, either as conductedrequirementsatanantennaportorasradiatedrequirementsover-the-air(OTA).

18.1SpectrumFlexibilityImplicationsSpectrum flexibilitywas a fundamental requirement forLTEand it hadmajorimplications for how LTEwas specified. The need for spectrum flexibility isevenhigherforNR,becauseofthediversespectrumwhereNRneedstooperateand the way the physical layer is designed to meet the key characteristicsrequired for5G.The followingare some important aspects impactinghow theRFcharacteristicsaredefined:

•Diversespectrumallocations:Thespectrumusedfor3Gand4Gisalreadyverydiverseintermsofthesizesofthefrequencyofoperation,bandwidthallocations,howtheyarearranged(pairedandunpaired),andwhattherelatedregulationis.ForNRitwillbeevenmorediverse,withthefundamentalfrequencyvaryingfrombelow1GHzupto40–50GHzandabove;themaximumfrequencypresentlyunderstudyinITU-Ris86GHz.ThesizeofallocatedbandswhereNRistobedeployedvariesfrom5MHzto3GHz,withbothpairedandunpairedallocations,wheretheintentionistousesomeallocationsassupplementarydownlinksoruplinkstogetherwithotherpairedbands.Thespectrumthatisplannedandunderinvestigationtobeusedfor5GandtherelatedoperatingbandsdefinedforNRaredescribedinChapter3.

•Variousspectrumblockdefinitions:Withinthediversespectrumallocations,spectrumblockswillbeassignedforNRdeployment,usuallythroughoperatorlicenses.Theexactfrequencyboundariesoftheblockscanvarybetweencountriesandregionsanditmustbe

possibletoplacetheRFcarriersinpositionswheretheblocksareusedefficientlywithoutwastingspectrum.Thisputsspecificrequirementsonthechannelrastertouseforplacingcarriers.

•LTE-NRcoexistence:TheLTE/NRcoexistenceinthesamespectrummakesitpossibletodeployNRwithin-carriercoexistenceinbothuplinkanddownlinkofexistingLTEdeployments.ThisisfurtherdescribedinChapter17.SincethecoexistingNRandLTEcarriersneedtobesubcarrier-aligned,thisposesrestrictionsontheNRchannelrasterinordertoaligntheplacingofNRandLTEcarriers.

•Multipleandmixednumerologies:AsdescribedinSection7.1,thetransmissionschemeforNRhasahighflexibilityandsupportsmultiplenumerologieswithsubcarrierspacingsrangingfrom15to240kHz,withdirectimplicationsforthetimeandfrequencydomainstructure.ThesubcarrierspacinghasimplicationsfortheRFintermsoftheroll-offofthetransmittedspectrum,whichimpacttheresultingguardbandsthatareneededbetweenthetransmittedresourceblocksandtheRFcarrieredgedefinedforthepurposeofdefiningRFrequirements(seeSection18.3).NRalsosupportsmixednumerologiesonthesamecarrier,whichhasfurtherRFimpactssincetheguardbandsmayneedtobedifferentatthetwoedgesoftheRFcarrier.

•Independentchannelbandwidthdefinitions:NRdevicesingeneraldonotreceiveortransmitusingthefullchannelbandwidthoftheBSbutcanbeassignedwhatiscalledabandwidthpart(seeSection7.4).WhiletheconceptdoesnothaveanydirectRFimplications,itisimportanttonotethatBSanddevicechannelbandwidtharedefinedindependentlyandthatthedevicebandwidthcapabilitydoesnothavetomatchtheBSchannelbandwidth.

•Variationofduplexschemes:AsshowninSection7.2,asingleframestructureisdefinedinNRthatsupportsTDD,FDD,andhalf-duplexFDD.TheduplexmethodisspecificallydefinedforeachoperatingbanddefinedforNRasshowninChapter3.Somebandsarealsodefinedassupplementarydownlink(SDL)orsupplementaryuplink(SUL)tobeusedinFDDoperation.ThisisfurtherdescribedinSection7.7.

Many of the frequency bands identified for deployment of NR are existingbands identified for IMT (seeChapter 3) and theymay already have 2G, 3G,

and/or4Gsystemsdeployed.Manybandsarealsoinsomeregionsdefinedandregulated in a “technology-neutral” manner, which means that coexistencebetweendifferenttechnologiesisarequirement.Thecapabilitytooperateinthiswide range of bands for any mobile system, including NR, has directimplications for the RF requirements and how those are defined, in order tosupportthefollowing:

•Coexistencebetweenoperatorsinthesamegeographicalareaintheband:OperatorsinthesamebandmaydeployNRorotherIMTtechnologies,suchasLTE,UTRA,orGSM/EDGE.Theremayinsomecasesalsobenon-IMTtechnologies.Suchcoexistencerequirementsaretoalargeextentdevelopedwithin3GPP,buttheremayalsoberegionalrequirementsdefinedbyregulatorybodiesincertaincases.

•Co-locationofbase-stationequipmentbetweenoperators:Thereareinmanycaseslimitationstowherebase-stationequipmentcanbedeployed.Often,sitesmustbesharedbetweenoperatorsoranoperatorwilldeploymultipletechnologiesinonesite.Thisputsadditionalrequirementsonbothbase-stationreceiversandtransmitterstooperateincloseproximitytootherbasestations.

•Coexistencewithservicesinadjacentfrequencybandsandacrosscountryborders:TheuseoftheRFspectrumisregulatedthroughcomplexinternationalagreements,involvingmanyinterests.Therewillthereforeberequirementsforcoordinationbetweenoperatorsindifferentcountriesandforcoexistencewithservicesinadjacentfrequencybands.Mostofthesearedefinedindifferentregulatorybodies.Insomecases,theregulatorsrequestthat3GPPincludessuchcoexistencelimitsinthe3GPPspecifications.

•CoexistencebetweenoperatorsofTDDsystemsinthesamebandisingeneralprovidedbyinter-operatorsynchronization,inordertoavoidinterferencebetweendownlinkanduplinktransmissionsofdifferentoperators.Thismeansthatalloperatorsneedtohavethesamedownlink/uplinkconfigurationsandframesynchronization,whichisnotinitselfanRFrequirement,butitisimplicitlyassumedinthe3GPPspecifications.RFrequirementsforunsynchronizedsystemsbecomemuchstricter.

•Release-independentfrequency-bandprinciples:Frequencybandsaredefinedregionally,andnewbandsareaddedcontinuouslyforeach

generationofmobilesystems.Thismeansthateverynewreleaseof3GPPspecificationswillhavenewbandsadded.Throughthe“releaseindependence”principle,itispossibletodesigndevicesbasedonanearlyreleaseof3GPPspecificationsthatsupportafrequencybandaddedinalaterrelease.ThefirstsetofNRbands(seeChapter3)isdefinedinrelease15andadditionalbandswillbeaddedinarelease-independentway.

•Aggregationofspectrumallocations:Operatorsofmobilesystemshavequitediversespectrumallocations,whichinmanycasesdonotconsistofablockthateasilyfitsexactlywithinonecarrier.Theallocationmayevenbenon-contiguous,consistingofmultipleblocksspreadoutinabandorinmultiplebands.Forthesescenarios,theNRspecificationssupportscarrieraggregation,wheremultiplecarrierswithinaband,orinmultiplebands,canbecombinedtocreatelargertransmissionbandwidths.

18.2RFRequirementsinDifferentFrequencyRangesAsdiscussedaboveandinChapter3,therewillbeaverywiderangeofdiversespectrumallocationswhereNRcanoperate.Theallocationsvaryinblocksize,channelbandwidthandduplexspacingsupported,butwhatreallydifferentiatesNR from previous generations is the wide frequency range over whichrequirementsneedtobedefined,wherenotonlytherequirementlimitsbutalsothe definitions and conformance testing aspects may be quite different atdifferent frequencies. Measurement equipment, such as spectrum analyzers,becomesmorecomplexandexpensiveathigherfrequenciesandforthehighestfrequencies considered, including theharmonicsof thehighestpossible carrierfrequencies,requirementsmaynotevenbepossibletotestinareasonableway.For this reason, theRF requirements for both devices and base stations are

dividedintofrequencyranges(FRs),wherepresentlytwoaredefined(FR1andFR2)in3GPPrelease15asshowninTable18.1.Thefrequencyrangeconceptisnot intended to be static. If new NR band(s) are added that are outside theexisting frequency ranges, one of them could be extended to cover the newband(s) if the requirements will align well with that range. If there are largedifferencescomparedtoexistingFR,anewfrequencyrangecouldbedefinedforthenewband.

Table18.1

FrequencyRangesDefinedin3GPPRelease15

FrequencyRangeDesignation CorrespondingFrequencyRangeFrequencyrange1(FR1) 450–6,000MHz

Frequencyrange2(FR2) 24,250–52,600MHz

ThefrequencyrangesarealsoillustratedinFig.18.1ona logarithmicscale,where the relatedbands identified for IMT (in at least one region) are shown.FR1startsat450MHzatthefirstIMTallocationandendsat6GHz.FR2coversasubsetofthebandsthatarepresentlyunderstudyforIMTidentificationintheITU-R (see Section 3.1). The subset ends at 52.6 GHz, which is the highestfrequencywithinthescopeofthespecificationworkin3GPPrelease15.

FIGURE18.1 FrequencyrangesFR1andFR2andcorrespondingIMTidentifications.Notethatthefrequencyscalesarelogarithmic.

All existingLTEbands arewithinFR1 andNR is expected to coexistwithLTEandpreviousgenerationsofsystemsinmanyoftheFR1bands.Itisonlyinwhat isoften referred toas the“midbands”around3.5GHz(in fact spanning3.3–5GHz)thatNRwilltoalargerextentbedeployedina“new”spectrum,thatisaspectrumpreviouslynotexploitedformobileservices.FR2coversapartofwhat is often referred to as the mm-wave band (strictly, mm-wave starts at30GHzwith10mmwavelength).Atsuchhighfrequenciescompared toFR1,propagation properties are different, with less diffraction, higher penetrationlosses,andingeneralhigherpathlosses.Thiscanbecompensatedforbyhaving

more antenna elements both at the transmitter and receiver, to be used fornarrower antenna beamswith higher gain and formassiveMIMO. This givesoverall different coexistence properties and therefore leads to different RFrequirementsforcoexistence.mm-waveRFimplementationforFR2bandswillalso have different complexity and performance compared to FR1 bands,impactingallcomponentsincludingA/DandD/Aconverters,LOgeneration,PAefficiently,filtering,etc.ThisisfurtherdiscussedinChapter19.

18.3ChannelBandwidthandSpectrumUtilizationTheoperatingbandsdefinedforNRhaveaverylargevariationinbandwidth,asshown inChapter 3.The spectrumavailable for uplinkor downlink canbe assmall as5MHz in someLTE re-farmingbands,while it isup to900MHz in“new”bandsforNRinfrequencyrange1,anduptoseveralGHzinfrequencyrange2.Thespectrumblocksavailableforasingleoperatorwilloftenbesmallerthanthis.Furthermore,themigrationtoNRinoperatingbandscurrentlyusedforotherradio-accesstechnologiessuchasLTE,mustoftentakeplacegraduallytoensurethatasufficientamountofspectrumremainstosupporttheexistingusers.Thus, the amount of spectrum that can initially be migrated to NR can berelatively small but may then gradually increase. The variation of the size ofspectrumblocksandpossiblespectrumscenariosimpliesarequirementforveryhigh spectrum flexibility for NR in terms of the transmission bandwidthssupported.ThefundamentalbandwidthofanNRcarrieriscalledthechannelbandwidth

(BWChannel) and is a fundamental parameter for defining most of the NR RFrequirements.ThespectrumflexibilityrequirementpointsouttheneedforNRtobe scalable in the frequency domain over a large range. In order to limitimplementation complexity, only a limited set of bandwidths is defined in theRF specifications. A range of channel bandwidths from 5 to 400 MHz issupported.Thebandwidthofacarrierisrelatedtothespectrumutilization,whichisthe

fraction of a channel bandwidth occupied by the physical resource blocks. InLTE,themaximumspectrumutilizationwas90%,butahighernumberhasbeentargetedforNRtoachieveahigherspectrumefficiency.Considerationshowevermust be taken for the numerology (subcarrier spacing), which impacts theOFDM waveform roll-off, and for the implementation of filtering and

(18.1)

windowing solutions. In addition, spectrum utilization is related to theachievableerrorvectormagnitude (EVM)and transmitterunwantedemissions,andalso to receiverperformance includingadjacentchannel selectivity (ACS).Thespectrumutilizationisspecifiedasamaximumnumberofphysicalresourceblocks, NRB, which will be the maximum possible transmission bandwidthconfiguration,definedseparatelyforeachpossiblechannelbandwidth.Whatthespectrumutilizationultimatelydefinesisaguardbandateachedge

oftheRFcarrier,asshowninFig.18.2.Outsideoftheguardbandandtherebyoutside the RF channel bandwidth, the “external” RF requirements such asunwanted emissions are defined, while only requirements on the actual RFcarriersuchasEVMaredefinedinside.ForachannelbandwidthBWChannel,theguardbandwillbe

FIGURE18.2 ThechannelbandwidthforoneRFcarrierandthecorrespondingtransmissionbandwidthconfiguration.

whereNRBisthemaximumnumberofresourceblockspossibleandΔfisthesubcarrierspacing.TheextraΔf/2guardappliedoneachsideofthecarrierisdue

totherelationtotheRFchannelraster,whichhasasubcarrier-basedgranularityandisdefinedindependentlyoftheactualspectrumblocks.Itmaythereforenotbepossible toplacea carrier exactly in thecenterof a spectrumblockandanextraguardwillberequiredtomakesureRFrequirementscanbemet.AsshowninEq. (18.1), theguardbandandtherebythespectrumutilization

will dependon the numerology applied.As described inSection 7.3, differentbandwidths will be possible depending on the subcarrier spacing of thenumerology, since the maximum value for NRB is 275. In order to havereasonablespectrumutilization,valuesofNRBbelow11arenotusedeither.Theresult is a range of possible channel bandwidths and corresponding spectrumutilization numbers defined for NR, as shown in Table 18.2. Note that thesubcarrierspacinguseddiffersbetweenfrequencyranges1and2.Thespectrumutilization expressed as a fraction is up to 98% for the widest channelbandwidthsanditisabove90%forallcases,exceptforthesmallerbandwidths,whereNRB≤25.

Table18.2

Since the channel bandwidth is defined independently for base stations anddevices (see aboveand inSection7.4), the actual channelbandwidths that aresupportedbythebasestationanddevicespecificationswillalsobedifferent.Foraspecificbandwidth,thesupportedspectrumutilizationishoweverthesameforbasestationanddevice,ifthecombinationofbandwidthandsubcarrierspacingissupportedbyboth.

18.4OverallStructureofDeviceRFRequirementsThedifferencesincoexistencepropertiesandimplementationbetweenFR1andFR2meansthatdeviceRFrequirementsforNRaredefinedseparatelyforFR1andFR2.Foramoredetaileddiscussionof the implementationaspects inFR2

usingmm-wavetechnologyfordevicesandbasestations,seeChapter19.For LTE and previous generations, RF requirements have in general been

specifiedasconductedrequirementsthataredefinedandmeasuredatanantennaconnector.Sinceantennasarenormallynotdetachableonadevice,thisisdoneatanantennatestport.DevicerequirementsinFR1aredefinedinthisway.WiththehighernumberofantennaelementsforoperationinFR2andthehigh

level of integration expected when using mm-wave technology, conductedrequirementsarenolongerseenasfeasible.FR2willthereforebespecifiedwithradiated requirements and testingwill have to be doneOTA.While this is anextrachallengewhendefiningrequirements,inparticularfortesting,itisseenasanecessityforFR2.Therewill also be a set of device requirements for interworkingwith other

radios within the same device. This concerns primarily interworking with E-UTRAfornon-standalone(NSA)operationandinterworkingbetweenFR1andFR2radiosforcarrieraggregation.Finally, there is a set of device performance requirements, which set the

basebanddemodulationperformanceofphysicalchannelsofthedevicereceiveracrossarangeofconditions,includingpropagationindifferentenvironments.Because of the differences between the different types of requirements, the

specificationfordeviceRFcharacteristicsisseparatedintofourdifferentparts,wherethedeviceiscalleduserequipment(UE)in3GPPspecifications:

•TS38.101-1[5]:UEradiotransmissionandreception,FR1;•TS38.101-2[6]:UEradiotransmissionandreception,FR2;•TS38.101-3[7]:UEradiotransmissionandreception,interworkingwithotherradios;

•TS38.101-4[8]:UEradiotransmissionandreception,performancerequirements.

TheconductedRFrequirementsforFR1aredescribedinSections18.6–18.11.

18.5OverallStructureofBase-StationRFRequirements18.5.1ConductedandRadiatedRFRequirementsforNRBS

For the continuing evolution of mobile systems, AAS have an increasingimportance. While there were several attempts to develop and deploy basestationswithpassiveantennaarraysofdifferentkindsformanyyears,therehavebeen no specificRF requirements associatedwith such antenna systems.WithRFrequirementsingeneraldefinedatthebasestationRFantennaconnector,theantennashavealsonotbeenseenaspartofthebasestation,atleastnotfromastandardizationpointofview.Requirementsspecifiedatanantennaconnectorare referred toasconducted

requirements,usuallydefinedasapowerlevel(absoluteorrelative)measuredatthe antenna connector. Most emission limits in regulation are defined asconductedrequirements.Analternativewayistodefinearadiatedrequirement,whichisassessedincludingtheantenna,oftenaccountingfortheantennagainina specific direction. Radiated requirements demand more complex OTA testprocedures, using for example an anechoic chamber. With OTA testing, thespatial characteristics of the whole BS, including the antenna system, can beassessed.For base stations with AAS, where the active parts of the transmitter and

receivermaybeanintegralpartoftheantennasystem,itisnotalwayssuitabletomaintain the traditional definitionof requirements at the antenna connector.Forthispurpose,3GPPdevelopedRFrequirementsinrelease13forAASbasestationsinasetofseparateRFspecificationsthatareapplicabletobothLTEandUTRAequipment.For NR, radiated RF requirements and OTA testing will be a part of the

specificationsfromthestart,bothinFR1andFR2.MuchoftheworkfromAAShas thereforebeen takendirectly into theNRspecifications.The termAASassuch is not used within the NR base-station RF specification [4], howeverrequirementsareinsteaddefinedfordifferentBStypes.The AAS BS requirements are based on a generalized AAS BS radio

architecture, as shown in Fig. 18.3. The architecture consists of a transceiverunit array that is connected to a composite antenna that contains a radiodistributionnetwork andanantennaarray.The transceiverunit arraycontainsmultiple transmitter and receiver units. These are connected to the compositeantenna through a number of connectors on the transceiver array boundary(TAB).TheseTABconnectorscorrespondtotheantennaconnectorsonanon-AASbasestationandserveasareferencepointforconductedrequirements.Theradiodistributionnetworkispassiveanddistributesthetransmitteroutputstothecorrespondingantennaelementsandviceversaforthereceiverinputs.Notethat

theactualimplementationofanAASBSmaylookdifferentintermsofphysicallocation of the different parts, array geometry, type of antenna elements used,etc.

FIGURE18.3 Generalizedradioarchitectureofanactiveantennasystem(AAS),usedalsofortheNR-radiatedrequirements.

BasedonthearchitectureinFig.18.3,therearetwotypesofrequirements:

•ConductedrequirementsaredefinedforeachRFcharacteristicatanindividualoragroupofTABconnectors.Theconductedrequirementsaredefinedinsuchawaythattheyareinasense“equivalent”tothecorrespondingconductedrequirementforanon-AASbasestation,thatis,theperformanceofthesystemortheimpactonothersystemsisexpectedtobethesame.

•RadiatedrequirementsaredefinedOTAinthefarfieldoftheantennasystem.Sincethespatialdirectionbecomesrelevantinthiscase,itisdetailedforeachrequirementhowitapplies.Radiatedrequirementsaredefinedwithreferencetoaradiatedinterfaceboundary(RIB),somewhereinthefar-fieldregion.

18.5.2BSTypesinDifferentFrequencyRangesforNRAnumberofdifferentbase-stationdesignpossibilitieshavetobeconsideredfortheRFrequirements.FirstinFR1,therearebasestationsbuiltinawaysimilar

to“classical”3Gand4Gbase stationswithantennaconnectors throughwhichexternal antennas are connected. Then we have base stations with AAS, butwhereantennaconnectorscanstillbeaccessedfordefinitionandtestingofsomeRFrequirements.Finally,wehavebasestationswithhighlyintegratedantennasystems where all requirements must be assessed OTA, since there are noantenna connectors. It is assumed that in FR2wheremm-wave technology isused for implementation of the antenna systems, only the latter type of basestationneedstobespecified.3GPP has defined four base-station types based on the above assumptions,

withreferencetothearchitecturedefinedaboveinFig.18.3:

•BStype1-C:NRbasestationoperatinginFR1,specifiedonlywithconductedrequirementsdefinedatindividualantennaconnectors.

•BStype1-O:NRbasestationoperatinginFR1,specifiedonlywithconducted(OTA)requirementsdefinedattheRIB.

•BStype1-H:NRbasestationoperatingatFR1,specifiedwitha“hybrid”setofrequirementsconsistingofbothconductedrequirementsdefinedatindividualTABconnectorsandsomeOTArequirementsdefinedattheRIB.

•BStype2-O:NRbasestationoperatinginFR2,specifiedonlywithconducted(OTA)requirementsdefinedattheRIB.

BStype1-ChasrequirementsdefinedinthesamewayasforUTRAorLTEconductedrequirements.ThesearedescribedinSections18.6–18.11.BStype1-Hcorrespondstothefirst typeofAASbasestationsspecifiedfor

LTE/UTRAin3GPPRelease13,where two radiated requirementsaredefined(radiated transmit power andOTA sensitivity),while all others are defined asconducted requirements, asdescribed inSections18.6–18.11.Manyconductedrequirements,suchasunwantedemissionlimits,areforBStype1-Hdefinedintwosteps.Firstabasiclimitisdefined,whichisidenticaltotheconductedlimitatanindividualantennaconnectorforBStype1-Candtherebyequivalenttothelimit at aTABconnector forBS type1-H. Ina second step, thebasic limit isconverted to a radiated limit at theRIB through a scaling factor based on thenumber of active transmitter units. The scaling is capped at amaximum of 8(9 dB), which is the maximum number of antenna elements used in definingcertainregulatorylimits.Notethatthemaximumscalingmayvarydependingonregionalregulation.

BStype1-OandBStype2-Ohaveallrequirementsdefinedasradiated.BStype 1-O has many requirements defined with reference to the correspondingFR1 conducted requirements, where unwanted emission limits also have ascaling applied as for BS type 1-H. The overall differences in coexistencepropertiesandimplementationbetweenFR1andFR2meanthatBStype2-OhasseparateFR2requirementsdefinedthatinmanycasesaredifferentfromtheFR1requirementsforBStype1-O.AnoverviewoftheradiatedrequirementsusedforBStypes1-Oand2-O,and

tosomeextentforBStype1-H,isgiveninSection18.12.

18.6OverviewofConductedRFRequirementsforNRTheRFrequirementsdefinethereceiverandtransmitterRFcharacteristicsofabasestationordevice.Thebasestation is thephysicalnode that transmitsandreceivesRFsignalsononeormoreantennaconnectors.Note thatanNRbasestationisnotthesamethingasagNB,whichisthecorrespondinglogicalnodeintheradio-accessnetwork(seeChapter6).ThedeviceisdenotedUEinallRFspecifications. Conducted RF requirements are defined for operating bands inFR1,whileonlyradiated(OTA)requirementsaredefinedforFR2(seeSection18.12).The set of conductedRF requirements defined forNR is fundamentally the

sameasthosedefinedforLTEoranyotherradiosystem.Somerequirementsarealso based on regulatory requirements and are more concerned with thefrequencybandofoperationand/ortheplacewherethesystemisdeployed,thanwiththetypeofsystem.What is particular to NR is the flexible channel bandwidths and multiple

numerologiesofthesystem,whichmakessomerequirementsmorecomplextodefine. These properties have special implications for the transmitterrequirements on unwanted emissions, where the definition of the limits ininternationalregulationdependonthechannelbandwidth.Suchlimitsarehardertodefineforasystemwherethebasestationmayoperatewithmultiplechannelbandwidthsandwherethedevicemayvaryitschannelbandwidthofoperation.ThepropertiesoftheflexibleOFDM-basedphysicallayeralsohaveimplicationsforspecifyingthetransmittermodulationqualityandhowtodefinethereceiverselectivity and blocking requirements. Note that the channel bandwidth ingeneralisdifferentfortheBSandthedeviceasdiscussedinSection18.3.

Thetypeoftransmitterrequirementsdefinedforthedeviceisverysimilartowhatisdefinedforthebasestation,andthedefinitionsoftherequirementsareoften similar. The output power levels are, however, considerably lower for adevice, while the restrictions on the device implementation are much higher.There is tight pressure on cost and complexity for all telecommunicationsequipment,butthisismuchmorepronouncedfordevices,duetothescaleofthetotalmarket,beingclosetotwobilliondevicesperyear.Incaseswheretherearedifferences in how requirements are defined between device and base station,theyaretreatedseparatelyinthischapter.The detailed background of the conducted RF requirements for NR is

described inRefs. [74] and [75].The conductedRF requirements for the basestation are specified in Ref. [4] and for the device in Ref. [5]. The RFrequirementsaredividedintotransmitterandreceivercharacteristics.Therearealso performance characteristics for base stations and devices that define thereceiver baseband performance for all physical channels under differentpropagation conditions. These are not strictly RF requirements, though theperformancewillalsodependontheRFtosomeextent.Each RF requirement has a corresponding test defined in the NR test

specificationsforthebasestationandthedevice.Thesespecificationsdefinethetest setup, test procedure, test signals, test tolerances, etc. needed to showcompliancewiththeRFandperformancerequirements.

18.6.1ConductedTransmitterCharacteristicsThe transmitter characteristics define RF requirements for the wanted signaltransmitted from the device and the base station, but also for the unavoidableunwanted emissions outside the transmitted carrier(s). The requirements arefundamentallyspecifiedinthreeparts:

•Outputpowerlevelrequirementssetlimitsforthemaximumallowedtransmittedpower,forthedynamicvariationofthepowerlevel,andinsomecasesforthetransmitterOFFstate;

•Transmittedsignalqualityrequirementsdefinethe“purity”ofthetransmittedsignalandalsotherelationbetweenmultipletransmitterbranches;

•Unwantedemissionsrequirementssetlimitstoallemissionsoutsidethetransmittedcarrier(s)andaretightlycoupledtoregulatoryrequirements

andcoexistencewithothersystems.

A list of the device and base-station transmitter characteristics arrangedaccording to the three parts defined above is shown in Table 18.3. A moredetailed description of the specific requirements can be found later in thischapter.

Table18.3

18.6.2ConductedReceiverCharacteristicsThesetof receiver requirements forNR isquite similar towhat isdefined forother systems such as LTE and UTRA. The receiver characteristics arefundamentallyspecifiedinthreeparts:

•Sensitivityanddynamicrangerequirementsforreceivingthewantedsignal;

•Receiversusceptibilitytointerferingsignalsdefinesreceivers’susceptibilitytodifferenttypesofinterferingsignalsatdifferentfrequencyoffsets;

•Unwantedemissionslimitsarealsodefinedforthereceiver.

A list of the device and base-station receiver characteristics arrangedaccording to the three parts defined above is shown in Table 18.4. A moredetaileddescriptionofeachrequirementcanbefoundlaterinthischapter.

Table18.4

Table18.4

18.6.3RegionalRequirementsThere are a number of regional variations to the RF requirements and theirapplication.Thevariationsoriginateindifferentregionalandlocalregulationsofthe spectrum and its use. Themost obvious regional variation is the differentfrequency bands and their use, as discussed above.Many of the regional RFrequirementsarealsotiedtospecificfrequencybands.Whenthereisaregionalrequirementon,forexample,spuriousemissions,this

requirementshouldbereflectedinthe3GPPspecifications.Forthebasestationit is entered as an optional requirement and is marked as “regional.” For thedevice, the same procedure is not possible, since a devicemay roam betweendifferentregionsandwillthereforehavetofulfillallregionalrequirementsthataretiedtoanoperatingbandintheregionswherethebandisused.ForNR(andalso for LTE), this becomesmore complex than for UTRA, since there is anadditional variation in the transmitter (and receiver) bandwidth used, makingsome regional requirements difficult tomeet as amandatory requirement.TheconceptofnetworksignalingofRFrequirementsisthereforeintroducedforNR,where a device can be informed at call setup of whether some specific RFrequirementsapplywhenthedeviceisconnectedtoanetwork.

18.6.4Band-SpecificDeviceRequirementsThroughNetworkSignalingFor the device, the channel bandwidths supported are a function of the NRoperating band, and also have a relation to the transmitter and receiver RFrequirements.ThereasonisthatsomeRFrequirementsmaybedifficulttomeet

under conditionswith a combination ofmaximum power and high number oftransmittedand/orreceivedresourceblocks.InbothNRandLTE,someadditionalRFrequirementsapply for thedevice

whenaspecificnetworksignalingvalue(NS_x)issignaledtothedeviceaspartof the cell handover or broadcastmessage. For implementation reasons, theserequirements are associated with restrictions and variations to RF parameterssuch as device output power, maximum channel bandwidth, and number oftransmitted resource blocks. The variations of the requirements are definedtogether with the NS_x in the device RF specification, where each valuecorresponds to a specific condition.The default value for all bands isNS_01.NS_x values are connected to an allowed power reduction called additionalmaximum power reduction (A-MPR) and may apply for transmission using acertain minimum number of resource blocks, depending also on the channelbandwidth.

18.6.5Base-StationClassesInordertoaccommodatedifferentdeploymentscenariosforbasestations,therearemultiplesetsofRFrequirementsforNRbasestations,eachapplicabletoabasestationclass.WhentheRFrequirementswerederivedforNR,base-stationclasseswereintroducedthatwereintendedformacrocell,microcell,andpicocellscenarios.The termsmacro,micro, andpico relate to thedeployment scenarioand are not used in 3GPP to identify the base-station classes, instead thefollowingterminologyisused:

•Wideareabasestations:Thistypeofbasestationisintendedformacrocellscenarios,withaBS-to-deviceminimumdistancealongthegroundequalto35m.Thisisthetypicallargecelldeploymentwithhigh-towerorabove-rooftopinstallations,givingwideareaoutdoorcoverage,butalsoindoorcoverage.

•Mediumrangebasestations:Thistypeofbasestationisintendedformicrocellscenarios,withaBS-to-deviceminimumdistancealongthegroundequalto5m.Typicaldeploymentsareoutdoorbelow-rooftopinstallations,givingbothoutdoorhotspotcoverageandoutdoor-to-indoorcoveragethroughwalls.

•Localareabasestations:Thistypeofbasestationisintendedforpicocellscenarios,definedwithaBS-to-deviceminimumdistance

alongthegroundequalto2m.Typicaldeploymentsareindoorofficesandindoor/outdoorhotspots,withtheBSmountedonwallsorceilings.

Thelocalareaandmediumrangebasestationclasseshavemodificationstoanumberofrequirementscomparedtowideareabasestations,mainlyduetotheassumptionofalowerminimumbasestationtodevicedistance,givingalowerminimumcouplingloss:

•Maximumbasestationpowerislimitedto38dBmoutputpowerformediumrangebasestationsand24dBmoutputpowerforlocalareabasestations.Thispowerisdefinedperantennaandcarrier.Thereisnomaximumbasestationpowerdefinedforwideareabasestations.

•Thespectrummask(operatingbandunwantedemissions)haslowerlimitsformediumrangeandlocalarea,inlinewiththelowermaximumpowerlevels.

•Receiverreferencesensitivitylimitsarehigher(morerelaxed)formediumrangeandlocalarea.Receiverdynamicrangeandin-channelselectivity(ICS)arealsoadjustedaccordingly.

•Limitsforco-locationformediumrangeandlocalareaarerelaxedcomparedtowideareaBS,correspondingtotherelaxedreferencesensitivityforthebasestation.

•Allmediumrangeandlocalarealimitsforreceiversusceptibilitytointerferingsignalsareadjustedtotakethehigherreceiversensitivitylimitandthelowerassumedminimumcouplingloss(basestation-to-device)intoaccount.

18.7ConductedOutputPowerLevelRequirements18.7.1Base-StationOutputPowerandDynamicRangeThere is no generalmaximumoutput power requirement for base stations.Asmentioned in thediscussionofbase-stationclasses above, there is, however, amaximum output power limit of 38 dBm formedium range base stations and24 dBm for local area base stations. In addition to this, there is a tolerancespecified,defininghowmuchtheactualmaximumpowermaydeviatefromthe

powerleveldeclaredbythemanufacturer.Thebase stationalsohas a specificationof the totalpower controldynamic

rangeforaresourceelement,definingthepowerrangeoverwhichitshouldbepossible to configure. There is also a dynamic range requirement for the totalbase-stationpower.For TDD operation, a power mask is defined for the base-station output

power, defining the off power level during the uplink subframes and themaximum time for the transmitter transient period between the transmitter onandoffstates.

18.7.2DeviceOutputPowerandDynamicRangeThedeviceoutputpowerlevelisdefinedinthreesteps:

•UEpowerclassdefinesanominalmaximumoutputpowerforQPSKmodulation.Itmaybedifferentindifferentoperatingbands,butthemaindevicepowerclassistodaysetat23dBmforallbands.

•Maximumpowerreduction(MPR)definesanallowedreductionofmaximumpowerlevelforcertaincombinationsofmodulationusedandresourceblockallocation.

•Additionalmaximumpowerreduction(A-MPR)maybeappliedinsomeregionsandisusuallyconnectedtospecifictransmitterrequirementssuchasregionalemissionlimitsandtocertaincarrierconfigurations.Foreachsuchsetofrequirements,thereisanassociatednetworksignalingvalueNS_xthatidentifiestheallowedA-MPRandtheassociatedconditions,asexplainedinSection18.6.4.

A minimum output power level setting defines the device dynamic range.Thereisadefinitionofthetransmitteroffpowerlevel,applicabletoconditionswhen thedevice isnotallowed to transmit.There isalsoageneralon/off timemask specified, plus specific timemasks for PRACH, PUCCH, SRS, and forPUCCH/PUSCH/SRStransitions.Thedevice transmit power control is specified through requirements for the

absolute power tolerance for the initial power setting, the relative powertolerance between two subframes, and the aggregated power tolerance for asequenceofpower-controlcommands.

18.8TransmittedSignalQuality

18.8TransmittedSignalQualityThe requirements for transmitted signal quality specify how much thetransmitted base station or device signal deviates from an “ideal” modulatedsignal in the signal and frequency domains. Impairments on the transmittedsignalareintroducedbythetransmitterRFparts,withthenon-linearpropertiesofthePAbeingamajorcontributor.Thesignalqualityisassessedforthebasestation and device through requirements on EVM and frequency error. Anadditionaldevicerequirementisdevicein-bandemissions.

18.8.1EVMandFrequencyErrorWhile the theoretical definitions of the signal quality measures are quitestraightforward,theactualassessmentisaveryelaborateprocedure,describedingreat detail in the 3GPP specification. The reason is that it becomes amultidimensionaloptimizationproblem,wherethebestmatchforthetiming,thefrequency,andthesignalconstellationarefound.The EVM is a measure of the error in the modulated signal constellation,

taken as the rootmean square of the error vectors over the active subcarriers,considering all symbols of the modulation scheme. It is expressed as apercentage value in relation to the power of the ideal signal. The EVMfundamentallydefinesthemaximumSINRthatcanbeachievedatthereceiver,if there are no additional impairments to the signal between transmitter andreceiver.Sinceareceivercanremovesomeimpairmentsofthetransmittedsignalsuch

as time dispersion, the EVM is assessed after cyclic prefix removal andequalization.Inthisway,theEVMevaluationincludesastandardizedmodelofthe receiver. The frequency offset resulting from the EVM evaluation isaveragedandusedasameasureofthefrequencyerrorofthetransmittedsignal.

18.8.2DeviceIn-BandEmissionsIn-bandemissionsareemissionswithinthechannelbandwidth.Therequirementlimitshowmuchadevicecantransmitintonon-allocatedresourceblockswithinthe channel bandwidth. Unlike the out-of-band (OOB) emissions, the in-bandemissionsaremeasuredaftercyclicprefixremovalandFFT,sincethisishowadevicetransmitteraffectsarealbase-stationreceiver.

18.8.3Base-StationTimeAlignment

18.8.3Base-StationTimeAlignmentSeveral NR features require the base station to transmit from two or moreantennas, suchas transmitterdiversityandMIMO.Forcarrieraggregation, thecarriersmayalsobetransmittedfromdifferentantennas.Inorderforthedeviceto properly receive the signals from multiple antennas, the timing relationbetweenanytwotransmitterbranchesisspecifiedintermsofamaximumtimealignment error between transmitter branches. The maximum allowed errordependsonthefeatureorcombinationoffeaturesinthetransmitterbranches.

18.9ConductedUnwantedEmissionsRequirementsUnwanted emissions from the transmitter are divided intoOOBemissions andspuriousemissionsinITU-Rrecommendations[42].OOBemissionsaredefinedas emissions on a frequency close to the RF carrier, which results from themodulationprocess.SpuriousemissionsareemissionsoutsidetheRFcarrierthatmaybereducedwithoutaffectingthecorrespondingtransmissionofinformation.Examples of spurious emissions are harmonic emissions, intermodulationproducts,andfrequencyconversionproducts.ThefrequencyrangewhereOOBemissions are normally defined is called theOOB domain, whereas spuriousemissionlimitsarenormallydefinedinthespuriousdomain.ITU-RalsodefinestheboundarybetweentheOOBandspuriousdomainsata

frequency separation from the carrier center of 2.5 times the necessarybandwidth,whichcorrespondsto2.5timesthechannelbandwidthforNR.Thisdivision of the requirements is easily applied for systems that have a fixedchannelbandwidth.Itdoes,however,becomemoredifficultforNR,whichisaflexible bandwidth system, implying that the frequency range whererequirementsapplywouldthenvarywiththechannelbandwidth.Theapproachtakenfordefiningtheboundaryin3GPPisslightlydifferentforbase-stationanddevicerequirements.With the recommended boundary between OOB emissions and spurious

emissions set at 2.5 times the channel bandwidth, third-and fifth-orderintermodulation products from the carrier will fall inside the OOB domain,whichwillcoverafrequencyrangeoftwicethechannelbandwidthoneachsideof thecarrier.For theOOBdomain, twooverlapping requirementsaredefinedforbothbasestationanddevice:spectrumemissionsmask(SEM)andadjacentchannelleakageratio(ACLR).Thedetailsofthesearefurtherexplainedbelow.

18.9.1ImplementationAspectsThe spectrum of an OFDM signal decays rather slowly outside of thetransmission bandwidth configuration. Since the transmitted signal for NRoccupies up to 98% of the channel bandwidth, it is not possible to meet theunwantedemissionlimitsdirectlyoutsidethechannelbandwidthwitha“pure”OFDM signal. The techniques used for achieving the transmitter requirementsare, however, not specified or mandated in NR specifications. Time-domainwindowingisonemethodcommonlyusedinOFDM-basedtransmissionsystemstocontrolspectrumemissions.Filteringisalwaysused,bothtime-domaindigitalfilteringofthebasebandsignalandanalogfilteringoftheRFsignal.Thenon-linearcharacteristicsofthepoweramplifier(PA)usedtoamplifythe

RF signal must also be taken into account, since it is the source ofintermodulationproductsoutsidethechannelbandwidth.Powerback-offtogiveamorelinearoperationofthePAcanbeused,butatthecostofalowerpowerefficiency.Thepowerback-offshouldthereforebekepttoaminimum.Forthisreason,additional linearizationschemescanbeemployed.Theseareespeciallyimportant for the base station, where there are fewer restrictions onimplementation complexity and use of advanced linearization schemes is anessential part of controlling spectrum emissions. Examples of such techniquesarefeed-forward,feedback,predistortion,andpostdistortion.

18.9.2EmissionMaskintheOOBDomainTheemissionmaskdefinesthepermissibleOOBspectrumemissionsoutsidethenecessary bandwidth. As explained above, how to take the flexible channelbandwidth into account when defining the frequency boundary betweenOOBemissionsandspuriousemissionsisdonedifferentlyfortheNRbasestationanddevice.Consequently,theemissionmasksarealsobasedondifferentprinciples.

18.9.2.1Base-StationOperatingBandUnwantedEmissionLimitsFor theNRbase station, theproblemof the implicitvariationof theboundarybetween OOB and spurious domain with the varying channel bandwidth ishandledbynotdefininganexplicitboundary.Thesolutionisaunifiedconceptofoperatingbandunwantedemissions(OBUEs)fortheNRbasestationinsteadof the spectrummaskusuallydefined forOOBemissions.Theoperatingband

unwantedemissionsrequirementappliesoverthewholebase-stationtransmitteroperatingband, plus an additional 10–40MHzon each side, as shown inFig.18.4.All requirements outside of that range are set by the regulatory spuriousemission limits, based on the ITU-R recommendations [42]. As seen in Fig.18.4, a large part of the operating band unwanted emissions is defined over afrequencyrangethatforsmallerchannelbandwidthscanbebothinspuriousandOOBdomains.Thismeansthatthelimitsforthefrequencyrangesthatmaybein the spurious domain also have to alignwith the regulatory limits from theITU-R.Theshapeofthemaskisgenericforallchannelbandwidths,withamaskthatconsequentlyhas toalignwith theITU-Rlimitsstarting10–40MHzfromthe channel edges.Theoperatingbandunwanted emissions aredefinedwith a100 kHz measurement bandwidth and align to a large extent with thecorrespondingmasksforLTE.

FIGURE18.4 FrequencyrangesforoperatingbandunwantedemissionsandspuriousemissionsapplicabletoNRbasestation(FR1).

Inthecaseofcarrieraggregationforabasestation,theOBUErequirement(asotherRF requirements) applies as for anymulticarrier transmission,where theOBUEwillbedefinedrelativetothecarriersontheedgesoftheRFbandwidth.Inthecaseofnon-contiguouscarrieraggregation,theOBUEwithinasub-blockgap ispartlycalculatedas thecumulativesumofcontributions fromeachsub-block.Therearealsospecial limitsdefined tomeeta specific regulationsetby the

FCC (FederalCommunicationsCommission,Title 47) for the operating bandsusedintheUSAandbytheECCforsomeEuropeanbands.Thesearespecified

asseparatelimitsinadditiontotheoperatingbandunwantedemissionlimits.

18.9.2.2DeviceSpectrumEmissionMaskFor implementation reasons, it is not possible to define a generic devicespectrummaskthatdoesnotvarywiththechannelbandwidth,sothefrequencyranges for OOB limits and spurious emissions limits do not follow the sameprinciple as for the base station. The SEM extends out to a separation ΔfOOBfrom the channel edges, as illustrated in Fig. 18.5. For 5 MHz channelbandwidth, this point corresponds to 250% of the necessary bandwidth asrecommendedby the ITU-R,but forhigherchannelbandwidths it is setcloserthan250%.

FIGURE18.5 FrequencyrangesforspectrumemissionmaskandspuriousemissionsapplicabletoanNRdevice.

TheSEMisdefinedasageneralmaskandasetofadditionalmasksthatcanbe applied to reflect different regional requirements. Each additional regionalmaskisassociatedwithaspecificnetworksignalingvalueNS_x.

18.9.3AdjacentChannelLeakageRatioInadditiontoaspectrumemissionsmask,theOOBemissionsaredefinedbyanACLR requirement. The ACLR concept is very useful for analysis ofcoexistence between two systems that operate on adjacent frequencies. TheACLR defines the ratio of the power transmitted within the assigned channel

bandwidth to thepower of the unwanted emissions transmittedon an adjacentchannel. There is a corresponding receiver requirement called ACS, whichdefinesareceiver’sabilitytosuppressasignalonanadjacentchannel.ThedefinitionsofACLRandACSareillustratedinFig.18.6forawantedand

an interfering signal received in adjacent channels. The interfering signal’sleakage of unwanted emissions at the wanted signal receiver is given by theACLR and the ability of the receiver of the wanted signal to suppress theinterfering signal in the adjacent channel is defined by the ACS. The twoparameterswhencombineddefine the total leakagebetween two transmissionsonadjacentchannels.Thatratioiscalledtheadjacentchannelinterferenceratio(ACIR)andisdefinedastheratioofthepowertransmittedononechanneltothetotal interference received by a receiver on the adjacent channel, due to bothtransmitter(ACLR)andreceiver(ACS)imperfections.

FIGURE18.6 IllustrationofACLRandACS,withexamplecharacteristics

(18.2)

foran“aggressor”interfererandareceiverfora“victim”wantedsignal.

Thisrelationbetweentheadjacentchannelparametersis[11]:

ACLRandACScanbedefinedwithdifferentchannelbandwidthsforthetwoadjacentchannels,whichisthecaseforsomerequirementssetforNRduetothebandwidth flexibility. Eq. (18.2) will also apply for different channelbandwidths,butonlyifthesametwochannelbandwidthsareusedfordefiningallthreeparameters,ACIR,ACLR,andACS,usedintheequation.The ACLR limits for NR device and base station are derived based on

extensiveanalysis[11]ofNRcoexistencewithNRorothersystemsonadjacentcarriers.For anNR base station, there areACLR requirements both for an adjacent

channelwithanNRreceiverofthesamechannelbandwidthandforanadjacentLTE receiver. The ACLR requirement for NR BS is set to 45 dB. This isconsiderably more strict than the ACS requirement for the device, whichaccording to Eq. (18.2) implies that in the downlink, the device receiverperformance will be the limiting factor for ACIR and consequently forcoexistencebetweenbasestationsanddevices.Fromasystempointofview,thischoice is cost-efficient since it moves implementation complexity to the BS,insteadofrequiringalldevicestohavehigh-performanceRF.In thecaseofcarrieraggregation forabasestation, theACLR(asotherRF

requirements) apply as for any multicarrier transmission, where the ACLRrequirementwillbedefinedforthecarriersontheedgesoftheRFbandwidth.Inthe case of non-contiguous carrier aggregation where the sub-block gap is sosmall that the ACLR requirements at the edges of the gap will “overlap,” aspecial cumulative ACLR requirement (CACLR) is defined for the gap. ForCACLR, contributions from carriers on both sides of the sub-block gap areaccountedforintheCACLRlimit.TheCACLRlimitisthesameastheACLRforthebasestationat45dB.ACLR limits for the device are set bothwith assumedNR and an assumed

UTRAreceiveron theadjacentchannel. In thecaseofcarrieraggregation, thedeviceACLRrequirementappliestotheaggregatedchannelbandwidthinsteadof per carrier. The ACLR limit for NR devices is set to 30 dB. This is

considerably relaxed compared to the ACS requirement for the BS, whichaccording to Eq. (18.2) implies that in the uplink, the device transmitterperformance will be the limiting factor for ACIR and consequently forcoexistencebetweenbasestationsanddevices.

18.9.4SpuriousEmissionsThe limits for base station spurious emissions are taken from internationalrecommendations[42],butareonlydefinedintheregionoutsidethefrequencyrangeofoperatingbandunwanted emission limits as illustrated inFig. 18.4—that is, at frequencies that are separated from the base-station transmitteroperating band by at least 10–40MHz. There are also additional regional oroptionallimitsforprotectionofothersystemsthatNRmaycoexistwithorevenbe co-located with. Examples of other systems considered in those additionalspuriousemissionsrequirementsareGSM,UTRAFDD/TDD,CDMA2000,andPHS.Devicespuriousemission limitsaredefined forall frequency rangesoutside

the frequency range covered by the SEM. The limits are generally based oninternational regulations [42] but there are also additional requirements forcoexistence with other bands when the device is roaming. The additionalspuriousemissionlimitscanhaveanassociatednetworksignalingvalue.Inaddition, therearebase-stationanddeviceemission limitsdefinedfor the

receiver.Since receiveremissionsaredominatedby the transmitted signal, thereceiverspuriousemissionlimitsareonlyapplicablewhenthetransmitterisnotactive,andalsowhenthetransmitterisactiveforanNRFDDbasestationthathasaseparatereceiverantennaconnector.

18.9.5OccupiedBandwidthOccupiedbandwidthisaregulatoryrequirementthatisspecifiedforequipmentin some regions, such as Japan and theUSA. Itwas originally definedby theITU-Ras amaximumbandwidth, outsideofwhichemissionsdonot exceedacertain percentage of the total emissions. The occupied bandwidth is for NRequal to the channel bandwidth, outside of which a maximum of 1% of theemissionsareallowed(0.5%oneachside).

18.9.6TransmitterIntermodulation

An additional implementation aspect of anRF transmitter is the possibility ofintermodulation between the transmitted signal and another strong signaltransmittedintheproximityofthebasestationordevice.Forthisreason,thereisarequirementfortransmitterintermodulation.Forthebasestation,therequirementisbasedonastationaryscenariowitha

co-locatedotherbase-stationtransmitter,withitstransmittedsignalappearingatthe antenna connector of the base station being specified but attenuated by30 dB. Since it is a stationary scenario, there are no additional unwantedemissions allowed, implying that all unwanted emission limits alsohave tobemetwiththeinterfererpresent.Forthedevice,thereisasimilarrequirementbasedonascenariowithanother

device-transmittedsignalappearingattheantennaconnectorofthedevicebeingspecified but attenuated by 40 dB. The requirement specifies the minimumattenuationoftheresultingintermodulationproductbelowthetransmittedsignal.

18.10ConductedSensitivityandDynamicRangeThe primary purpose of the reference sensitivity requirement is to verify thereceivernoisefigure,whichisameasureofhowmuchthereceiver’sRFsignalchain degrades the SNR of the received signal. For this reason, a low-SNRtransmission scheme using QPSK is chosen as a reference channel for thereferencesensitivitytest.Thereferencesensitivityisdefinedatareceiverinputlevelwherethethroughputis95%ofthemaximumthroughputforthereferencechannel.Forthedevice,referencesensitivityisdefinedforthefullchannelbandwidth

signalsandwithallresourceblocksallocatedforthewantedsignal.Theintentionofthedynamicrangerequirementistoensurethatthereceiver

canalsooperateatreceivedsignallevelsconsiderablyhigherthanthereferencesensitivity.Thescenarioassumedforbase-stationdynamicrangeisthepresenceofincreasedinterferenceandcorrespondinghigherwantedsignallevels,therebytesting the effects of different receiver impairments. In order to stress thereceiver, a higher SNR transmission scheme using 16QAM is applied for thetest.Inordertofurtherstressthereceivertohighersignallevels,aninterferingAWGNsignal at a level 20dBabove the assumednoise floor is added to thereceivedsignal.Thedynamicrangerequirementforthedeviceisspecifiedasamaximumsignallevelatwhichthethroughputrequirementismet.

18.11ReceiverSusceptibilitytoInterfering

18.11ReceiverSusceptibilitytoInterferingSignalsThereisasetofrequirementsforbasestationanddevice,definingthereceiver’sabilitytoreceiveawantedsignalinthepresenceofastrongerinterferingsignal.The reason for the multiple requirements is that, depending on the frequencyoffset of the interferer from the wanted signal, the interference scenario maylook very different and different types of receiver impairmentswill affect theperformance.Theintentionofthedifferentcombinationsofinterferingsignalsisto model as far as possible the range of possible scenarios with interferingsignalsofdifferentbandwidths thatmaybeencountered insideandoutside thebase-stationanddevicereceiveroperatingband.While the types of requirements are very similar between base station and

device, the signal levels are different, since the interference scenarios for thebasestationanddeviceareverydifferent.There isalsonodevice requirementcorrespondingtothebase-stationICSrequirement.The following requirements are defined for NR base station and device,

startingfrominterfererswithlargefrequencyseparationandgoingclosein(seealsoFig.18.7).InallcaseswheretheinterferingsignalisanNRsignal,ithasthesameorsmallerbandwidththanthewantedsignal,butatmost20MHz.

•Blocking:Thiscorrespondstothescenariowithstronginterferingsignalsreceivedoutsidetheoperatingband(out-of-bandblocking)orinsidetheoperatingband(in-bandblocking),butnotadjacenttothewantedsignal.In-bandblockingincludesinterferersinthefirst20–60MHzoutsidetheoperatingbandforthebasestationandthefirst15MHzforthedevice.Thescenariosaremodeledwithacontinuouswave(CW)signalfortheout-of-bandcaseandanNRsignalforthein-bandcase.Thereareadditional(optional)base-stationblockingrequirementsforthescenariowhenthebasestationisco-locatedwithanotherbasestationinadifferentoperatingband.Forthedevice,afixednumberofexceptionsareallowedfromtheout-of-bandblockingrequirement,foreachassignedfrequencychannelandattherespectivespuriousresponsefrequencies.Atthosefrequencies,thedevicemustcomplywiththemorerelaxedspuriousresponserequirement.

•Adjacentchannelselectivity:TheACSscenarioisastrongsignalinthechanneladjacenttothewantedsignalandiscloselyrelatedtothe

correspondingACLRrequirement(seealsothediscussioninSection18.9.3).TheadjacentinterfererisanNRsignal.Forthedevice,theACSisspecifiedfortwocaseswithalowerandahighersignallevel.

•Narrowbandblocking:Thescenarioisanadjacentstrongnarrowbandinterferer,whichintherequirementismodeledasasingleresourceblockNRsignalforthebasestationandaCWsignalforthedevice.

•In-channelselectivity(ICS):Thescenarioismultiplereceivedsignalsofdifferentreceivedpowerlevelsinsidethechannelbandwidth,wheretheperformanceoftheweaker“wanted”signalisverifiedinthepresenceofthestronger“interfering”signal.ICSisonlyspecifiedforthebasestation.

•Receiverintermodulation:Thescenarioistwointerferingsignalsneartothewantedsignal,wheretheinterferersareoneCWandoneNRsignal(notshowninFig.18.7).Thepurposeoftherequirementistotestreceiverlinearity.Theinterferersareplacedinfrequencyinsuchawaythatthemainintermodulationproductfallsinsidethewantedsignal’schannelbandwidth.ThereisalsoanarrowbandintermodulationrequirementforthebasestationwheretheCWsignalisveryclosetothewantedsignalandtheNRinterfererisasingleRBsignal.

FIGURE18.7 Base-stationanddevicerequirementsforreceiversusceptibilitytointerferingsignalsintermsofblocking,ACS,narrowbandblocking,andICS(BSonly).

For all requirements except ICS, thewanted signal uses the same reference

channel as in the corresponding reference sensitivity requirement. With theinterferenceadded,thesame95%relativethroughputismetasforthereferencesensitivity,butata“desensitized”higherwantedsignallevel.

18.12RadiatedRFRequirementsforNRManyoftheradiatedRFrequirementsdefinedfordevicesandbasestationsarederived directly from the corresponding conducted RF requirements. Unlikeconducted requirements, the radiated requirements will account also for theantenna.Whendefiningemission levels suchasbasestationoutputpowerandunwantedemissions,thiscanbedoneeitherbyincorporatingtheantennagainasadirectionalrequirementusinganeffective isotropicradiatedpower (EIRP)orby definition of limits using total radiated power (TRP). Two new radiatedrequirementsaredefinedasdirectionalforthebasestation(seeSection18.12.2),butmostradiateddeviceandbasestationrequirementsforNRaredefinedwithlimitsexpressedasTRP.Thereareseveralreasonsforthischoice[19].TRPandEIRParedirectlyrelated throughthenumberofradiatingantennas

and depend also on specific base station implementation, considering thegeometryof the antenna array and the correlationbetweenunwanted emissionsignalsfromdifferentantennaports.TheimplicationisthatanEIRPlimitcouldresultindifferentlevelsoftotalradiatedunwantedemissionpowerdependingontheimplementation.AnEIRPlimitwillthusnotgivecontrolofthetotalamountofinterferenceinthenetwork,whileaTRPrequirementlimitsthetotalamountof interference injected in the network regardless of the specific BSimplementation.Another relevant element behind the 3GPP choice of defining unwanted

emissionasTRPisthedifferentbehaviorbetweenpassiveandAAS.Inthecaseof passive systems, the antenna gain does not varymuch between thewantedsignalandunwantedemissions.Thus,EIRPisdirectlyproportionaltoTRPandcanbeused as a substitute.For an active system such asNR, theEIRPcouldvary greatly between the wanted signal and unwanted emissions and alsobetweenimplementations,soEIRPisnotproportionaltoTRPandusingEIRPtosubstituteTRPwouldbeincorrect.TheradiatedRFrequirementsfordeviceandbasestationaredescribedbelow.

18.12.1RadiatedDeviceRequirementsinFR2

Asdescribed inSection18.4, theRF requirements inFR2operatingbandsaredescribed in a separate specification [6] for devices, because of the highernumber of antenna elements for operation in FR2 and the high level ofintegrationexpectedwhenusingmm-wavetechnology.Thesetofrequirementsis basically the same as the conducted RF requirements defined for FR1operating bands.The limits formany requirements are however different.Thedifference in coexistence atmm-wave frequencies leads to lower requirementson, for example, ACLR and spectrum mask. This is demonstrated throughcoexistencestudiesperformedin3GPPanddocumentedin[11].Thepossibilityfordifferentlimitshasalsobeendemonstratedinacademia[73].The implementation usingmm-wave technologies is more challenging than

usingthemorematuretechnologiesinthefrequencybandsbelow6GHz(FR1).Themm-waveRFimplementationaspectsarefurtherdiscussedinChapter19.Itshouldalsobenotedthatthechannelbandwidthsandnumerologiesdefined

for FR2 are in general different fromFR1,making it not possible to comparerequirementlevels,especiallyforreceiverrequirements.ThefollowingisabriefoverviewoftheradiatedRFrequirementsinFR2:

•Outputpowerlevelrequirements:MaximumoutputpowerisofthesameorderasinFR1butisexpressedbothasTRPandEIRP.TheminimumoutputpowerandtransmitterOFFpowerlevelsarehigherthaninFR1.Radiatedtransmitpowerisanadditionalradiatedrequirement,whichunlikethemaximumoutputpowerisdirectional.

•Transmittedsignalquality:FrequencyerrorandEVMrequirementsaredefinedsimilartowhatisdoneinFR1andmostlywiththesamelimits.

•Radiatedunwantedemissionsrequirements:Occupiedbandwidth,ACLR,spectrummask,andspuriousemissionsaredefinedinthesamewayasforFR1.ThelattertwoarebasedonTRP.ManylimitsarelessstrictthaninFR1.ACLRisontheorderof10dBrelaxedcomparedtoFR1,duetomorefavorablecoexistence.

•Referencesensitivityanddynamicrange:DefinedinthesamewayasinFR1,butlevelsarenotcomparable.

•Receiversusceptibilitytointerferingsignals:ACS,in-bandandout-of-bandblockingaredefinedasforFR1,butthereisnonarrowbandblockingscenariodefinedsincethereareonlywidebandsystemsinFR2.ACSisontheorderof10dBrelaxedcomparedtoFR1,duetomorefavorablecoexistence.

18.12.2RadiatedBase-StationRequirementsinFR1AsdescribedinSection18.5,theRFrequirementsforBStype1-Oconsistedofonly radiated (OTA) requirements. These are in general based on thecorresponding conducted requirements, either directly or through scaling.Twoadditional radiated requirementsdefinedareradiated transmitpower andOTAsensitivity,describedfurtherbelow.BStype1-Hisdefinedwitha“hybrid”setofrequirementsconsistingmostly

ofconductedrequirementsandinadditiontworadiatedrequirements,whicharethesameasforBStype1-O:

•Radiatedtransmitpowerisdefinedaccountingfortheantennaarraybeam-formingpatterninaspecificdirectionasEIRPforeachbeamthatthebasestationisdeclaredtotransmit.InawaysimilartoBSoutputpower,theactualrequirementisontheaccuracyofthedeclaredEIRPlevel.

•OTAsensitivityisadirectionalrequirementbasedonaquiteelaboratedeclarationbythemanufacturerofoneormoreOTAsensitivitydirectiondeclarations(OSDDs).Thesensitivityisinthiswaydefinedaccountingfortheantennaarraybeam-formingpatterninaspecificdirectionasdeclaredequivalentisotropicsensitivity(EIS)leveltowardsareceivertarget.TheEISlimitistobemetnotonlyinasingledirectionbutwithinarangeofangleofarrival(RoAoA)inthedirectionofthereceivertarget.DependingonthelevelofadaptivityfortheAASBS,twoalternativedeclarationsaremade:

•Ifthereceiverisadaptivetodirection,sothatthereceivertargetcanberedirected,thedeclarationcontainsareceivertargetredirectionrangeinaspecifiedreceivertargetdirection.TheEISlimitshouldbemetwithintheredirectionrange,whichistestedatfivedeclaredsensitivityRoAoAwithinthatrange.

•Ifthereceiverisnotadaptivetodirectionandthuscannotredirectthereceivertarget,thedeclarationconsistsofasinglesensitivityRoAoAinaspecifiedreceivertargetdirection,inwhichtheEISlimitshouldbemet.

NotethattheOTAsensitivityisdefinedinadditiontothereferencesensitivityrequirement,whichexistsbothasconducted(forBStype1-H)andradiated(forBStype1-O).

18.12.3RadiatedBase-StationRequirementsinFR2AsdescribedinSection18.5,theRFrequirementsforBStype2-Oareradiatedrequirements for base stations in FR2 operating bands. These are describedseparately, togetherwith the radiated requirements forBS type1-O,but in thesamespecification[4]astheconductedbase-stationRFrequirements.The setof requirements is identical to the radiatedRF requirementsdefined

forFR1operatingbandsdescribedabove,butthelimitsformanyrequirementsare different. As for the device, the difference in coexistence at mm-wavefrequencies leads to lower requirements on, for example, ACLR, ACS asdemonstratedthrough3GPPcoexistencestudies[11].Theimplementationusingmm-wave technologies is also more challenging than using the more maturetechnologiesinthefrequencybandsbelow6GHz(FR1)asfurtherdiscussedinChapter19.ThefollowingisabriefoverviewoftheradiatedRFrequirementsinFR2:

•Outputpowerlevelrequirements:MaximumoutputpoweristhesameforFR1andFR2,butisscaledfromtheconductedrequirementandexpressedasTRP.Thereisinadditionadirectionalradiatedtransmitpowerrequirement.ThedynamicrangerequirementisdefinedsimilarlytoFR1.

•Transmittedsignalquality:Frequencyerror,EVM,andtime-alignmentrequirementsaredefinedsimilartowhatisdoneinFR1andmostlywiththesamelimits.

•Radiatedunwantedemissionsrequirements:Occupiedbandwidth,spectrummask,ACLR,andspuriousemissionsaredefinedinthesamewayasforFR1.ThethreelatterarebasedonTRPandalsohavelessstrictlimitsthaninFR1.ACLRisontheorderof15dB,relaxedcomparedtoFR1,duetomorefavorablecoexistence.

•Referencesensitivityanddynamicrange:DefinedinthesamewayasinFR1,butlevelsarenotcomparable.ThereisinadditionadirectionalOTAsensitivityrequirement.

•Receiversusceptibilitytointerferingsignals:ACS,in-band,andout-of-bandblockingaredefinedasforFR1,butthereisnonarrowbandblockingscenariodefinedsincethereareonlywidebandsystemsinFR2.ACSisrelaxedcomparedtoFR1,duetomorefavorablecoexistence.

18.13OngoingDevelopmentsofRFRequirementsforNRThefirstsetofNRspecificationsin3GPPrelease15doesnothavefullsupportforsomeRFdeploymentoptionsthatexistforLTE.Multistandardradio(MSR)basestations,multibandbasestations,andnon-contiguousoperationarefeaturesthat are under development in 3GPP and will have full support in the finalrelease15specifications,orinsomecasesinrelease16.Ashortdescriptionofthose features isgivenbelow,whereamoredetaileddescription (applicable toLTE)canbefoundinRef.[28].

18.13.1MultistandardRadioBaseStationsTraditionally the RF specifications have been developed separately for thedifferent 3GPP radio-access technologiesGSM/EDGE,UTRA, LTE, andNR.The rapid evolution ofmobile radio and the need to deploy new technologiesalongside the legacy deployments has, however, led to implementation ofdifferent radio-access technologies (RAT) at the same sites, sharing antennasand other parts of the installation. In addition, operation of multiple RATs isoftendonewithinthesamebase-stationequipment.Theevolutiontomulti-RATbase stations is fosteredby the evolutionof technology.WhilemultipleRATshavetraditionallysharedpartsofthesiteinstallation,suchasantennas,feeders,backhaul,orpower, theadvanceofbothdigitalbasebandandRF technologiesenablesamuchtighterintegration.3GPPdefinesanMSRbasestation,asabasestationwhere thereceiverand

the transmitter are capable of simultaneously processing multiple carriers ofdifferentRATs in commonactiveRF components.The reason for this stricterdefinitionisthatthetruepotentialofmulti-RATbasestations,andthechallengeintermsofimplementationcomplexity,comesfromhavingacommonRF.ThisprincipleisillustratedinFig.18.8withanexamplebasestationcapableofbothNRandLTE.SomeoftheNRandLTEbasebandfunctionalitymaybeseparate

in thebase stationbut ispossibly implemented in the samehardware.TheRFmust,however,beimplementedinthesameactivecomponentsasshowninthefigure.

FIGURE18.8 ExampleofmigrationfromLTEtoNRusinganMSRbasestationforallmigrationphases.

While development of MSR BS specifications including NR is part of theworkin3GPPrelease15,thesetofspecificationsfirstissuedforNSAoperationwillinthefirststepnotcovertheMSRBSspecifications.ItisexpectedthatNRwillbeaddedasanewRATforMSRBSduring2018.ThemainadvantagesofanMSRbasestationimplementationforNRaretwofold:

•MigrationbetweenRATsinadeployment,forexample,frompreviousmobilegenerationstoNR,ispossibleusingthesamebase-stationhardware.TheoperationofNRcanthenbeintroducedgraduallyovertimewhenpartsofthespectrumusedforpreviousgenerationsisfreedupforNR.

•AsinglebasestationdesignedasanMSRbasestationcanbedeployedinvariousenvironmentsforsingle-RAToperationforeachRATsupported,aswellasformulti-RAToperation,wherethatisrequiredbythedeploymentscenario.Thisisalsoinlinewiththerecenttechnologytrendsseeninthemarket,withfewerandmoregenericbase-station

designs.Havingfewervarietiesofbasestationisanadvantagebothforthebase-stationvendorandfortheoperator,sinceasinglesolutioncanbedevelopedandimplementedforavarietyofscenarios.

The MSR concept has a substantial impact for many requirements, whileothers remain completely unchanged. A fundamental concept introduced forMSRbasestationsisRFbandwidth,whichisdefinedasthetotalbandwidthoverthe set of carriers transmitted and received. Many receiver and transmitterrequirements are usually specified relative to the carrier center or the channeledges. For anMSR base station, they are instead specified relative to theRFbandwidthedges,inawaysimilartocarrieraggregation.ByintroducingtheRFbandwidth concept and introducing generic limits, the requirements for MSRshift frombeingcarrier centric towardsbeing frequencyblockcentric, therebyembracingtechnologyneutralitybybeingindependentoftheaccesstechnologyoroperationalmode.For the specification ofMSRbase stations, the operating bands are divided

intobandcategories(BC)dependingonwhatRATsaresupportedintheband.There are presently three band categories, BC1–BC3, covering paired bandswithoutGSMoperation,pairedbandswithGSMoperationandunpairedbands,respectively. It is not yet determined whether new band categories could beneededwhenNRisaddedasadditionalRAT.AnotherimportantconceptforMSRbasestationsisthesupportedcapability

set (CS), which is part of the declaration of base-station capabilities by thevendor. The capability set defines all supported single RATs and multi-RATcombinations.Therearecurrently15capabilitysets,CS1–CS15,definedintheMSRBStestspecification[2].WhenNRisaddedasanewRAT,itisexpectedthat the new CS will be added to cover RAT combinations that include NRoperation.Carrier aggregation is also applicable toMSRbase stations.Since theMSR

specification has most of the concepts and definitions in place for definingmulticarrierRFrequirements,whetheraggregatedornot,thedifferencesfortheMSRrequirementscomparedtonon-aggregatedcarriersareveryminor.MoredetailsontheRFrequirementsforMSRbasestationssupportingLTE,

UTRA,andGSM/EDGEoperationaregiveninSection22.5ofRef.[28].

18.13.2Multiband-CapableBaseStations

The3GPPspecificationshavebeencontinuouslydevelopedtosupportlargerRFbandwidthsfor transmissionandreception throughmulticarrierandmulti-RAToperationandcarrieraggregationwithinabandandwith requirementsdefinedforonebandat a time.Thishasbeenmadepossiblewith theevolutionofRFtechnology supporting larger bandwidths for both transmitters and receivers.From 3GPP release 11, there is support in the LTE and MSR base-stationspecifications for simultaneous transmission and/or reception in two bandsthroughacommonradio.Suchamultibandbasestationcoversmultiplebandsoverafrequencyrangeofafew100MHz.Supportformorethantwobandsisgivenfrom3GPPrelease14.While development of NR specifications for multiband base stations is not

excludedfromtheworkin3GPPrelease15,thesetofspecificationsfirstissuedforNSAoperationdoesnothavefulldescriptionsofmultibandoperationofNRforbandsinfrequencyrange2.One obvious application formultiband base stations is for interband carrier

aggregation. It shouldhoweverbenoted that base stations supportingmultiplebandswereinexistencelongbeforecarrieraggregationwasintroducedinLTEandUTRA.AlreadyforGSM,dual-bandbasestationsweredesignedtoenablemore compact deployments of equipment at base-station sites, but they werereally two separate sets of transmitters and receivers for the bands that wereintegrated in thesameequipmentcabinet.Thedifferencefor“true”multiband-capable base stations is that the signals for the bands are transmitted andreceivedincommonactiveRFinthebasestation.AnexamplebasestationisillustratedinFig.18.9,whichshowsabasestation

with a common RF implementation of both transmitter and receiver for twooperatingbandsXandY.Throughaduplex filter, the transmitterand receiverare connected to a common antenna connector and a common antenna. Theexample is also amulti-RAT-capableMB-MSR base station,with LTE+GSMconfigured inbandXandLTEconfigured inbandY.Note that the figurehasonlyonediagramshowingthefrequencyrangeforthetwobands,whichcouldeitherbethereceiverortransmitterfrequencies.

FIGURE18.9 Exampleofmultibandbasestationwithmultibandtransmitterandreceiverfortwobandswithonecommonantennaconnector.

While having only a single antenna connector and a common feeder thatconnects toacommonantenna isdesirable toreduce theamountofequipmentneeded in a site, it is not always possible. It may also be desirable to haveseparateantennaconnectors,feeders,andantennasforeachband.AnexampleofamultibandbasestationwithseparateconnectorsfortwooperatingbandsXandYisshowninFig.18.10.Notethatwhiletheantennaconnectorsareseparateforthetwobands,theRFimplementationfortransmitterandreceiverisinthiscasecommon for the bands.TheRF for the two bands is separated into individualpathsforbandXandbandYbeforetheantennaconnectorsthroughafilter.Asformultibandbasestationswithacommonantennaconnectorforthebands,itisalso here possible to have either the transmitter or receiver be a single-bandimplementation,whiletheotherismultiband.

FIGURE18.10 Multibandbasestationwithmultibandtransmitterandreceiverfortwobandswithseparateantennaconnectorsforeachband.

Further possibilities are base station implementations with separate antennaconnectorsforreceiverandtransmitter,inordertogivebetterisolationbetweenthe receiver and transmitterpaths.Thismaybedesirable for amultibandbasestation, considering the large total RF bandwidths, which will in fact alsooverlapbetweenreceiverandtransmitter.For a multiband base station, with a possible capability to operate with

multiple RATs and several alternative implementations with common orseparateantennaconnectorsforthebandsand/orforthetransmitterandreceiver,the declaration of the base station capability becomes quite complex. Whatrequirementswillapplytosuchabasestationandhowtheyaretestedwillalsodependonthesedeclaredcapabilities.Moredetails on theRF requirements formultibandbase stations supporting

LTEoperationisgiveninSection22.12of[28].

18.13.3OperationinNon-contiguousSpectrumSomespectrumallocationsconsistoffragmentedpartsofspectrumfordifferentreasons. The spectrum may be a recycled 2G spectrum, where the originallicensedspectrumwas“interleaved”betweenoperators.Thiswasquitecommon

for original GSM deployments, for implementation reasons (the originalcombiner filters used were not easily tuned when spectrum allocations wereexpanded).Insomeregions,operatorshavealsopurchasedspectrumlicensesonauctionsandhavefordifferentreasonsendedupwithmultipleallocationsinthesamebandthatarenotadjacent.For deployment of non-contiguous spectrum allocations there are a few

implications:

•Ifthefullspectrumallocationinabandistobeoperatedwithasinglebasestation,thebasestationhastobecapableofoperationinnon-contiguousspectrum.

•Ifalargertransmissionbandwidthistobeusedthanwhatisavailableineachofthespectrumfragments,boththedeviceandthebasestationhavetobecapableofintrabandnon-contiguouscarrieraggregationinthatband.

Note that the capability for the base station to operate in non-contiguousspectrumisnotdirectlycoupledtocarrieraggregationassuch.FromanRFpointof view,whatwill be required by the base stations is to receive and transmitcarriersoveranRFbandwidththatissplitintwo(ormore)separatesub-blocks,witha sub-blockgap in-between,as shown inFig.18.11.Thespectrum in thesub-blockgapcanbedeployedbyanyotheroperator,whichmeansthattheRFrequirements for the base station in the sub-block gap will be based oncoexistenceforuncoordinatedoperation.ThishasafewimplicationsforsomeofthebasestationRFrequirementswithinanoperatingband.

FIGURE18.11 Exampleofnon-contiguousspectrumoperation,illustratingthedefinitionsofRFbandwidth,sub-block,andsub-blockgap.

WhiledevelopmentofNRBSspecifications fornon-contiguous spectrum isnotexcluded from thework in3GPPRelease15, thesetof specifications firstissued for NSA operation does not have a full description of non-contiguousoperation.MoredetailsontheRFrequirementsfornon-contiguousoperationforLTEaregiveninSection22.4ofRef.[28].

CHAPTER19

RFTechnologiesatmm-WaveFrequencies

Abstract

This chapter describes some of theRF technologies needed atmm-WavefrequenciestoimplementdevicesandBSinfrequencyrange2.Challengesandpossibilitiesarediscussedfordifferentimplementationalternatives.

Keywordsmm-Wavetechnologies;ACD;DAC;LOgeneration;phasenoise;poweramplifier;filtering;noisefigure;dynamicrange;bandwidth

The existing 3GPP specifications for 2G, 3G, and 4Gmobile communicationsare applicable to frequency ranges below 6 GHz and the corresponding RFrequirementsconsiderthetechnologyaspectsrelatedtobelow6GHzoperation.NRalsooperatesinthosefrequencyranges(identifiedasfrequencyrange1)butwillinadditionbedefinedforoperationabove24.25GHz(frequencyrange2orFR2), also referred to as mm-wave frequencies. A fundamental aspect fordefiningtheRFperformanceandsettingRFrequirementsforNRbasestationsanddevicesisthechangeintechnologiesusedforRFimplementationinordertosupport operation in those higher frequencies. In this chapter, some importantandfundamentalaspectsrelatedtomm-wavetechnologiesarepresentedinorderto better understand the performance thatmm-wave technology can offer, butalsowhatthelimitationsare.In this chapter, Analog-to-Digital/Digital-to-Analog converters and power

amplifiersarediscussed,includingaspectssuchastheachievableoutputpowerversusefficiencyandlinearity.Inaddition,somedetailedinsightsareprovidedinto receiveressentialmetricssuchasnoise figure,bandwidth,dynamic range,power dissipation, and the dependencies betweenmetrics. Themechanism for

frequencygenerationandtherelatedphasenoiseaspectsarealsocovered.Filtersformm-wavesareanotherimportantpart,indicatingtheachievableperformancefor various technologies and the feasibility of integrating filters into NRimplementations.The data sets used in this chapter indicate the current state-of-the-art

capability and performance and are either published elsewhere or have beenpresentedaspartof the3GPPstudyfordevelopingNR[11].Note thatneitherthe 3GPP specifications nor the discussion here mandate any restrictions,specific models, or implementations for NR in frequency range 2. ThediscussionhighlightsandanalyzesdifferentpossibilitiesforRFimplementationofmm-wavereceiversandtransmitters.AnadditionalaspectisthatessentiallyalloperationinFrequencyRange2will

bewithActiveAntennaSystembasestationsusinglargeantennaarraysizesanddevices with multi-antenna implementations. While this is enabled by thesmaller scale of antennas at mm-wave frequencies, it also drives complexity.The compact building practice needed for mm-wave systems with manytransceivers and antennas requires careful and often complex considerationregarding the power efficiency and heat dissipation within a small area orvolume. These considerations directly affect the achievable performance andpossibleRFrequirements.ThediscussionhereinmanyaspectsappliesforbothNR base stations and NR devices, noting also that the mm-wave transceiverimplementation between device and base station will have less differencescomparedtofrequencybandsbelow6GHz.

19.1ADCandDACConsiderationsThelargerbandwidthsavailableatmm-wavecommunicationwillchallengethedataconversioninterfacesbetweenanaloganddigitaldomainsinbothreceiversandtransmitters.Thesignal-to-noise-and-distortionratio(SNDR)-basedSchreierFigure-of-Merit (FoM) is a widely accepted metric for Analog-to-DigitalConverters(ADCs)definedby[61]

withSNDRindB,powerconsumptionPinW,andNyquistsamplingfrequencyfs inHz.Fig.19.1showstheSchreierFoMforalargenumberofADCsvsthe

Nyquist sampling frequency fs (=2×BW formost converters), published at thetwomost acknowledgedconferences [62] in this fieldof research.Thedashedline indicates the FoM envelope which is constant at roughly 180 dB forsampling frequencies below some 100 MHz. With constant FoM, the powerconsumption doubles for every doubling of bandwidth or 3 dB increase inSNDR.Above100MHz there is an additional 10dB/decadepenalty, and thismeansthatadoublingofbandwidthwillincreasepowerconsumptionbyafactorof4.

FIGURE19.1 Schreierfigure-of-meritforpublishedADCs[62].

AlthoughtheFoMenvelope isexpected tobeslowlypushedtowardshigherfrequencies by continued development of integrated circuit technology, RFbandwidths in the GHz range inevitably give poor power efficiency in theanalog-to-digitalconversion.ThelargebandwidthsandarraysizesassumedforNR at mm-wave will thus lead to a large ADC power footprint and it isimportant that specificationsdrivingSNDRrequirementsarenotunnecessarilyhigh.Thisappliestodevicesaswellasbasestations.

Digital-to-Analog Converters (DACs) are typically less complex than theirADCcounterparts for thesameresolutionandspeed.Furthermore,whileADCoperationcommonlyinvolvesiterativeprocesses, theDACsdonot.DACsalsoattract substantially less interest in the research community.While structurallyquitedifferentfromtheirADCcounterpartstheycanstillbebenchmarkedusingthesameFoMandrendersimilarnumbersasforADCs.InthesamewayasforADC, a larger bandwidth and unnecessarily high SNDR requirement on thetransmitterwillresultinhigherDACpowerfootprint.

19.2LOGenerationandPhaseNoiseAspectsLocalOscillator (LO) is an essential component in allmodern communicationsystems for shifting carrier frequency up-or downwards in transceivers. AparameterfeaturingtheLOqualityistheso-calledphasenoise(PN)ofthesignalgeneratedbytheLO.Inplainwords,phasenoiseisameasureofhowstablethesignal is in frequency domain. Its value is given in dBc/Hz for an offsetfrequency Δf and it describes the likelihood that the oscillation frequencydeviatesbyΔffromthedesiredfrequency.LO phase noise may significantly impact system performance; this is

illustrated in Fig. 19.2, though somewhat exaggerated for a single-carrierexample,wheretheconstellationdiagramfora16-QAMsignaliscomparedforcaseswithandwithoutphasenoise, including inbothcasesanAdditiveWhiteGaussian Noise (AWGN) signal modeling thermal noise. For a given symbolerrorrate,phasenoiselimitsthehighestmodulationschemethatmaybeutilized,asdemonstratedinFig.19.2.Inotherwords,differentmodulationschemesposedifferentrequirementsontheLOphasenoiselevel.

FIGURE19.2 Constellationdiagramofasingle-carrier16-QAMsignalwithout(left)andwith(right)LOphasenoise.

19.2.1PhaseNoiseCharacteristicsofFree-RunningOscillatorsandPLLsThemostcommoncircuitsolutionforfrequencygenerationistouseaVoltage-ControlledOscillator(VCO).Fig.19.3showsamodelandthecharacteristicPNbehaviorofafree-runningVCOindifferentoffsetfrequencyregions,wheref0isthe oscillation frequency, Δf is the offset frequency from f0, Ps is the signalstrength,Qistheloadedqualityfactoroftheresonator,Fisanempiricalfittingparameter but has physicalmeaning of noise figure, andΔf1/f3 is the 1/f-noisecornerfrequencyoftheactivedeviceinuse[57].

FIGURE19.3 Phasenoisecharacteristicforatypicalfree-runningVCO[57]:phasenoiseindBc/Hz(y-axis)versusoffsetfrequencyinHz(x-axis,logarithmicscale).

ThefollowingcanbeconcludedfromtheLeesonformulainFig.19.3:

1.PNincreasesby6dBpereverydoublingoftheoscillationfrequencyf0;2.PNisinverselyproportionaltosignalstrength,Ps;3.PNisinverselyproportionaltothesquareoftheloadedqualityfactorof

theresonator,Q;4.1/fnoiseup-conversiongivesrisetoclose-to-carrierPNincrease(atsmall

offset).

Thus, thereare severalparameters thatmaybeused fordesign trade-offs inVCO development. To make performance comparison of the VCOs made indifferentsemiconductortechnologiesandcircuitrytopologies,aFigure-of-Merit(FoM) is often used which takes into account power consumption and thusallowsforafaircomparison:

Here isthephasenoiseoftheVCOindBc/Hzand isthepowerconsumptioninwatt.Onenoticeableresultofthisexpressionisthatbothphase

noiseandpowerconsumption in linearpowerareproportional to 2.Thus, tomaintainaphasenoiselevelatacertainoffsetwhileincreasing byafactorNwouldrequirethepowertobeincreasedbyN2(assumingafixedFoMvalue).AcommonwaytosuppressthephasenoiseistoapplyaPhaseLockedLoop

(PLL)[18].BasicPLLbuildingblockscontainaVCO,frequencydivider,phasedetector,loopfilter,andalow-frequencyreferencesourceofhighstability,suchasacrystaloscillator.The totalphasenoiseof thePLLoutput iscomposedofcontributions from the VCO outside the loop bandwidth and the referenceoscillator inside the loop.Asignificantnoisecontribution isalsoaddedby thephasedetectorandthedivider.Asanexampleforthetypicalbehaviorofanmm-waveLO,Fig.19.4shows

themeasuredphasenoisefroma28GHzLOproducedbyapplyingaPLLatalower frequency and thenmultiplying up to 28GHz. There are four differentoffsetrangesthatshowdistinctivecharacteristics:

1.f1,forsmalloffset,<10kHz:~30dB/decaderoll-off,dueto1/fnoiseup-conversion;

2.f2,foroffsetwithinthePLLbandwidth:relativelyflatandcomposedofseveralcontributions;

3.f3,foroffsetlargerthanPLLbandwidth:~20dB/decaderoll-off,dominantbyVCOphasenoise;

4.f4,forevenlargeroffset,>10MHz:flat,duetofinitenoisefloor.

FIGURE19.4 ExampleofmeasuredphasenoisebehaviorforaphaselockedVCOmultipliedto28GHz.EricssonAB,usedwithpermission.

19.2.2ChallengesWithmm-WaveSignalGenerationAs phase noise increases with frequency, increasing the oscillation frequencyfrom3GHzto30GHz,forinstance,willresultinfundamentalPNdegradationof20dBatagivenoffsetfrequency.ThiswillcertainlylimitthehighestorderofPN-sensitivemodulationschemesusableatmm-waveandthusposesalimitationonachievablespectrumefficiencyformm-wavecommunications.Millimeter-waveLOsalsosufferfromthedegradationinqualityfactorQand

thesignalpowerPs.Leeson’sequationtellsusthatinordertoachievelowphasenoise,QandPsneedtobemaximized,whileminimizingthenoisefigureoftheactive device. Unfortunately, these three factors contribute in an unfavorablemanner when oscillation frequency increases. In monolithic VCOimplementation, the Q-value of the on-chip resonator decreases rapidly withfrequency increases duemainly to (1) the increase of parasitic losses such asmetallossand/orsubstratelossand(2)thedecreaseofvaractorQ.Meanwhile,thesignalstrengthoftheoscillatorbecomesincreasinglylimitedwhengoingtohigher frequencies. This is because higher-frequency operation requires moreadvanced semiconductor devices whose breakdown voltage decreases as their

feature size shrinks. This is manifested by the observed reduction in powercapabilityversusfrequencyforpoweramplifiers(–20dBperdecade)asdetailedin Section 19.3. For this reason, a method widely applied in mm-wave LOimplementation is to generate a lower-frequency PLL and then multiply thesignaluptothetargetfrequency.Except for the challenges discussed above, up-conversion of the 1/f noise

createsanaddedslopeclosetothecarrier.The1/fnoiseisstronglytechnology-dependent, where planar devices such as CMOS and HEMT (High ElectronMobilityTransistor)generallyshowhigher1/fnoisethanverticalbipolardevicessuch as bipolar andHBTs.Technologies used in fully integratedMMIC/RFICVCO and PLL solution range from CMOS and BiCMOS to III–V materialswhere InGaP HBT is popular due to its relatively low 1/f noise and highbreakdown.OccasionallyalsopHEMTdevicesareused,evenifsufferingfromsevere1/fnoise.SomedevelopmentshavebeenmadeusingGaNFETstructuresinordertobenefitfromtheveryhighbreakdownvoltage,but1/fisevenhigherthan in GaAs FET devices and therefore seems to offset the gain due to thebreakdownvoltage.Fig.19.5summarizesphasenoiseperformanceat100kHzoffsetvsoscillationfrequencyfordifferentsemiconductortechnologies.

FIGURE19.5 Phasenoiseversusoscillationfrequencyforoscillatorsindifferentsemiconductortechnologies[36].

Lastbutnotleast,recentresearchrevealstheimpactoftheLOnoiseflooronsystemperformance[23].Thisimpactisinsignificantifthesymbolrateislow.When the rate increases, such as in 5G NR, the flat noise floor starts toincreasingly affect the EVM of the modulated signal. Fig. 19.6 shows themeasuredEVMfromatransmitterfordifferentsymbolrateanddifferentnoisefloorlevel.TheimpactfromreceiverLOnoisefloorissimilar.Thisobservationmay imply that it requires extra care when generating mm-wave LOs forwideband systems in terms of choice of technology, VCO topology, andmultiplicationfactor,tomaintainareasonablylowPNfloor.

FIGURE19.6 MeasuredEVMofa64-QAMsignalfroma7.5GHztransmitterfordifferentsymbolrateandLOnoisefloorlevel[23].

19.3PowerAmplifierEfficiencyinRelationtoUnwantedEmissionRadio Frequency (RF) building block performance generally degrades withincreasingfrequency.Thepowercapabilityofpoweramplifiers(PA)foragiven

integratedcircuittechnologyroughlydegradesby20dBperdecade,asshowninFig. 19.7 for a number of various semiconductor technologies. There is afundamental cause for this degradation; increased power capability andincreasedfrequencycapabilityareconflictingrequirementsasobservedfromtheso-called Johnson limit [54]. In short, higher operational frequencies requiresmaller geometries, which subsequently result in lower operational power inorder to prevent dielectric breakdown from the increased field strengths. ToupholdMoore’s law, the gate geometries are constantly shrunk and hence theintrinsicpowercapabilityisreduced.

FIGURE19.7 Poweramplifieroutputpowerversusfrequencyforvarioussemiconductortechnologies.Thedashedlineillustratestheobservedreductioninpowercapabilityversusfrequency(–20dBperdecade).Thedatapointsarefromasurveyofpublishedmicrowaveandmm-wavepoweramplifiercircuits.

Aremedyishoweverfoundinthechoiceofintegratedcircuitmaterial.mm-Wave integrated circuits have traditionally beenmanufactured using so-called

III–Vmaterials,thatisacombinationofelementsfromgroupsIIIandVoftheperiodic table, such as Gallium Arsenide (GaAs) and more recently GalliumNitride (GaN). Integrated circuit technologies based on III–V materials aresubstantiallymore expensive than conventional silicon-based technologies andtheycannothandletheintegrationcomplexityof,forexample,digitalcircuitsorradiomodems for cellular handsets.Nevertheless,GaN-based technologies arenowmaturing rapidly and deliver power levels an order of magnitude highercomparedtoconventionaltechnologies.There are mainly three semiconductor material parameters that affect the

efficiency of an amplifier: maximum operating voltage, maximum operatingcurrent density, and knee-voltage. Due to the knee-voltage, the maximumattainableefficiencyisreducedbyafactorthatisproportionalto:

wherek is the ratio of knee-voltage to themaximumoperatingvoltage.Formosttransistortechnologiestheratiokisintherangeof0.05–0.01,resultinginanefficiencydegradationof10%–20%.Themaximumoperatingvoltageand thecurrentdensity limit themaximum

outputpowerfromasingletransistorcell.Tofurtherincreasetheoutputpower,theoutput frommultiple transistorcellsmustbecombined.Themostcommoncombination techniques are stacking (voltage combining), paralleling (currentcombining), and corporate combiners (power combining). Either choice ofcombinationtechniquewillbeassociatedwithacertaincombiner-efficiency.Alower power density requiresmore combination stages andwill incur a loweroverall combiner-efficiency.Atmm-wave frequencies the voltage-and current-combiningmethods are limited due to thewavelength.The overall size of thetransistor cell must be kept less than about 1/10th of the wavelength. Hence,paralleling and/or stacking are used to some extent and then corporatecombiningisusedtogetthewantedoutputpower.Themaximumpowerdensityof CMOS is about 100 mW/mm compared to 4000mW/mm for GaN. Thus,GaNtechnologywillrequirelessaggressivecombiningstrategiesandhencegivehigherefficiency.Fig.19.8showsthesaturatedpower-addedefficiency(PAE)asafunctionof

frequency. The maximum reported PAE is approximately 40% and 25%, at30GHzand77GHz,respectively.

FIGURE19.8 Saturatedpower-addedefficiencyversusfrequencyforvarioussemiconductortechnologiesfromasurveyofpublishedmicrowaveandmm-wavepoweramplifiercircuits.

PAEisexpressedas

Atmm-wavefrequencies,semiconductortechnologiesfundamentallylimittheavailableoutputpower.Furthermore,theefficiencyisalsodegradedwithhigherfrequency.ConsideringthePAEcharacteristicsinFig.19.8,andthenon-linearbehavior

oftheAM-AM/AM-PMcharacteristicsofthepoweramplifier,significantpowerback-offmaybenecessarytoreachlinearityrequirementsuchasthetransmitterACLRrequirements(seeSection18.9).Consideringtheheatdissipationaspectsand significantly reduced area/volume for mm-wave products, the complexinterrelation between linearity, PAE, and output power in the light of heat

dissipationmustbeconsidered.

19.4FilteringAspectsUsingvarious typesof filters inbase stationanddevice implementations is anessentialcomponentformeetingtheoverallRFrequirements.ThishasbeenthecaseforallgenerationsofmobilesystemsandwillbeessentialalsoforNR,bothbelow6GHzandinthenewmm-wavebands.Thefiltersmitigatetheunwantedemissionsarisingfrom,forexample,non-linearity in the transmittersgenerateddue to intermodulation, noise, harmonics generation, LO leakage, and variousunwantedmixingproducts.Inthereceiverchain,filtersareusedtohandleeitherself-interferencefromowntransmittersignalinpairedbands,ortosuppresstheinterfereratadjacentorotherfrequencies.The RF requirements are differentiated in terms of levels for different

scenarios. For base station spurious emission, there are general requirementsacross a very wide frequency range, coexistence requirements in the samegeographicalareas,andco-locationrequirementsfordensedeployments.Similarrequirementsaredefinedfordevices.Considering the limited size (area/volume) and level of integrations needed

formm-wave frequencies, the filteringcanbechallengingwherediscretemm-wave filtersarequitebulkyand there isachallenge toembedsuch filters intohighlyintegratedstructuresformm-waveproducts.

19.4.1PossibilitiesofFilteringattheAnalogFront-EndDifferent implementations provide different possibilities for filtering. For thepurposeofdiscussion,twomaincasescanbeidentified:

•Low-cost,monolithicintegrationwithoneorafewmulti-chainCMOS/BiCMOScore-chipswithbuilt-inpoweramplifiersandbuiltindown-converters.Thiscasewillgivelimitedpossibilitiestoincludehigh-performancefiltersalongtheRF-chainssincetheQ-valuesforonchipfilterresonatorswillbepoor(5–20).

•High-performance,heterogeneousintegrationwithseveralCMOS/BiCMOScorechips,combinedwithexternalamplifiersandexternalmixers.Thisimplementationallowstheinclusionofexternal

filtersalongtheRF-chains(atahighercomplexity,size,andpowerconsumption).

Thereareat least threeplaceswhereitmakessensetoputfilters,dependingonimplementation,asshowninFig19.9:

•Behindorinsidetheantennaelement(F1orF0),whereloss,size,cost,andwidebandsuppressionareimportant;

•Behindthefirstamplifiers(lookingfromtheantennaside),wherelowlossislesscritical(F2);

•Onthehigh-frequencysideofmixers(F3),wheresignalhavebeencombined(inthecaseofanalogandhybridbeamforming).

FIGURE19.9 Possiblefilterlocations.

ThemainpurposeofF1/F0isnormallytosuppressinterferenceandemissionsfarfromthedesiredchannelacrossawidefrequencyrange(forexample,DCto60GHz).Thereshouldnotbeanyunintentionalresonancesorpassbandsinthiswidefrequencyrange.Thisfilterwillhelprelaxthedesignchallenge(bandwidthto consider, linearity requirements, etc.) of all following blocks. Insertion lossmustbeverylow,andtherearestrictsizeandcostrequirementssincetheremustbeone filter at each subarray (Figs. 19.9 and19.10). In somecases, this filtermustfulfillstrictsuppressionrequirementsclosetothepassband,particularlyforhighoutputpowerclosetosensitivebands.

FIGURE19.10 Filterexampleforthe28GHzband.

ThemainpurposeofF2issuppressionofLO-,image-,spurious-,andnoise-emission,andsuppressionofincominginterferersrelativelyfarfromthedesiredfrequency band. There are still strict size requirements, but more loss can beaccepted (behind the first amplifiers) and even unintentional passbands(assuming F1/F0 will handle that). This allows better discrimination (morepoles),andbetterfrequencyprecision(forexample,usinghalf-waveresonators).Themain purpose of F3 is typically suppression ofLO-, image-, spurious-,

andnoise-emission,andsuppressionofincominginterferersthataccidentallyfallin the IF-band after the mixer, and strong interferers that tend to block themixersorADCs.Foranalog(orhybrid)beam-formingitisenoughtohavejustone (or a few) such filters. This relaxes requirements on size and cost,whichopensthepossibilitytoachievesharpfilterswithmultiplepolesandzeroes,andwithhighQ-valueandgoodfrequencyprecisionintheresonators.Thedeeper into theRF-chain (starting from theantennaelement), thebetter

protectedthecircuitswillget.ForthemonolithicintegrationcaseitisdifficulttoimplementfiltersF2andF3.Onecanexpectperformancepenaltiesforthiscase,andoutputpowerperbranchislower.Furthermore,itischallengingtoachievegood isolation across a wide frequency range, as microwaves tend to bypassfiltersbypropagatingingroundstructuresaroundthem.

19.4.2InsertionLoss(IL)andBandwidth

Sharpfilteringoneachbranch(atpositionsF1/F0)withnarrowbandwidthleadstoexcessive lossatmicrowaveandmm-wavefrequencies.Toget the insertionloss down to a reasonable level the passband can bemade significantly largerthan the signal bandwidth. A drawback of such an approach is that moreunwantedsignalswillpassthefilter.Inchoosingthebestloss-bandwidthtrade-offtherearesomebasicdependenciestobeawareof:

•ILdecreaseswithincreasingBW(forfixedfc);•ILincreaseswithincreasingfc(forfixedBW);•ILdecreaseswithincreasingQ-value;•ILincreaseswithincreasingN.

Toexemplify the trade-off, a three-poleLC-filterwithQ=20,100,500, and5000, for100and800MHz3dB-bandwidth is studied, tuned to15dBreturnloss(withQ=5000)isexamined,asshowninFig.19.11.

FIGURE19.11 Examplethree-poleLCfilterwith800and4×800MHzbandwidth,fordifferentQvalues.

Fromthisstudyitisobservedthat:

•800MHzbandwidthorsmaller,requiresexoticfiltertechnologies,withaQ-valuearound500orbettertogetanILbelow1.5dB.SuchQ-

valuesareverychallengingtoachieveconsideringconstraintsonsize,integrationaspects,andcost;

•Byrelaxingtherequirementonselectivityto4×800MHz,itissufficienttohaveaQ-valuearound100toget2dBIL.Thisshouldbewithinreachwithalow-lossprintedcircuitboard(PCB).TheincreasedbandwidthwillalsohelptorelaxthetolerancerequirementsonthePCB.

19.4.3FilterImplementationExamplesThere aremanyways to implement filters in a5Garray radio.Keyaspects tocompare are:Q-value, discrimination, size, and integration possibilities. Table19.1givesaroughcomparisonbetweendifferenttechnologiesandtwospecificexamplesaregivenbelow.

Table19.1

19.4.3.1PCBIntegratedImplementationExampleAsimpleandattractivewaytoimplementantennafilters(F1)istousestriplineormicrostrip filters, embedded in a PCB close to each antenna element. Thisrequiresalow-lossPCBwithgoodprecision.Productiontolerances(permittivityandpatterningandvia-positioning)willlimittheperformance,mainlythoughashift in the passband and increased mismatch. In most implementations thepassbandmustbesetlargerthantheoperatingfrequencybandwithasignificantmargintoaccountforthis.Typical characteristics of such filters can be illustrated by looking at the

followingdesignexample,withthelayoutshowninFig.19.12:

•Five-pole,coupledline,striplinefilter;•Dielectricpermittivity:3.4;•Dielectricthickness:500μm(groundtoground);

•UnloadedresonatorQ:130(assuminglow-lossmicrowavedielectrics).

FIGURE19.12 LayoutofstriplinefilteronaPCB.

The filter is tuned to give 20 dB suppression at 24 GHz, while passing asmuch as possible of the band 24.25–27.5 GHz (with 17 dB return loss).SignificantmarginsareaddedtomakeroomforvariationsinthemanufacturingprocessesofthePCB.AMonteCarloanalysiswasperformedtostudytheimpactofvariationsinthe

manufacturing process on filter performance, using the following quiteaggressivetoleranceassumptionsforthePCB:

•Permittivitystandarddeviation:0.02;•Linewidthstandarddeviation:8μm;•Thicknessofdielectricstandarddeviation:15μm.

With these distribution assumptions, 1000 instances of the filter weregeneratedandsimulated.Fig.19.13showsthefilterperformance(S21)forthese1000 instances (blue traces), together with the nominal performance (yellowtrace).Redlinesinthegraphindicatepossiblerequirementlevelsthatcouldbemetconsideringthisfilter.

FIGURE19.13 SimulatedimpactofmanufacturingtolerancesonthefiltercharacteristicsofastriplinefilterinPCB.

From this design example, the following rough description of a PCB filterimplementationisfound:

•3–4dBinsertionloss;•20dBsuppression(17dBifILissubtracted);•1.5GHztransitionregionwithmarginsincluded;•Size:25mm2,whichcanbedifficulttofitinthecaseofindividualfeedand/ordualpolarizedelements;

•Ifa3dBIListargeted,therewouldbesignificantyieldlosswiththesuggestedrequirement,inparticularforchannelsclosetothepassbandedges.

19.4.3.2LTCCFilterImplementationExampleAnotherpromisingwaytoimplementfiltersistomakecomponentsforSurfaceMountAssembly(SMT),includingbothfiltersandantennas,forexamplebasedon Low-Temperature Cofired Ceramics (LTCC). One example of a prototypeLTCCcomponentwasoutlinedinRef.[31]andisalsoshowninFig.19.14.

FIGURE19.14 ExampleofprototypeofanLTCC-componentcontainingbothantennaelementsandfilters.TDKCorporation,usedwithpermission.

ThemeasuredperformanceofthecorrespondingfilterisshowninFig.19.15anditshowsthattheLTCC-filteraddsabout2dBofinsertionlossfora2GHzpassband, while providing 22 dB of additional attenuation 1 GHz from thepassbandedge.

FIGURE19.15 Measuredperformanceofthecorrespondingfilterwithoutantenna.TDKCorporation,usedwithpermission.

Additionalmargins relative to thisexampleshouldbeconsidered toaccountformanufacturing tolerancesandfutureadjustmentsofbandwidth, suppressionlevel,guardbandwidth,antennaproperties,integrationaspects,etc.Accountingforsuchmargins,theLTCC-filtershowncouldbeassumedtoaddapproximately3dBofinsertionloss,for17dBsuppression(ILsubtracted)at1.5GHzfromthepassbandedge.Technologydevelopment,particularlyregardingQ-valuesandmanufacturing

tolerances,willlikelyleadtoimprovementsinthesenumbers.

19.5ReceiverNoiseFigure,DynamicRange,andBandwidthDependencies19.5.1ReceiverandNoiseFigureModelAreceivermodelas shown inFig.19.16 is assumedhere.Thedynamic range

(DR) of the receiverwill in general be limited by the front-end insertion loss(IL), the receiver (RX) Low-noise Amplifier (LNA), and the ADC noise andlinearityproperties.

FIGURE19.16 Typicalzero-IFtransceiverschematic.

Typically DRLNA DRADC so the RX use Automatic Gain Control (AGC)and selectivity (distributed) in-between theLNAand theADC tooptimize themappingofthewantedsignalandtheinterferencetotheDRADC.Forsimplicity,afixedgainsettingisconsideredhere.AfurthersimplifiedreceivermodelcanbederivedbylumpingtheFrontEnd

(FE),RX, andADC into three cascaded blocks, as shown in Fig. 19.17. Thismodel cannot replace a more rigorous analysis but will demonstrateinterdependenciesbetweenthemainparameters.

FIGURE19.17 Asimplifiedreceivermodel.

Focusing on the small signal co-channel noise floor, the impact of varioussignalandlinearityimpairmentscanbestudiedtoarriveatasimplenoisefactor,ornoisefigure,expression.

19.5.2NoiseFactorandNoiseFloorAssumingmatchedconditions,Friis’formulacanbeusedtofindthenoisefactoratthereceiverinputas(linearunitsunlessnoted),

TheRXinputreferredsmall-signalco-channelnoisefloorwillthenequal

whereN0=k···T···BWandNADCaretheavailablenoisepowerandtheADCeffective noise floor in the channel bandwidth, respectively (k and T beingBoltzmann’s constant and absolute temperature, respectively). TheADCnoisefloor is typically set by a combination of quantization, thermal, andintermodulationnoise,buthere a flatnoise floor is assumedasdefinedby theADCeffectivenumberofbits.TheeffectivegainGfromLNAinputtoADCinputdependsonsmall-signal

gain, AGC setting, selectivity, and desensitization (saturation), but here it isassumed that the gain is set such that the antenna referred input compressionpoint (CPi) corresponds to theADC clipping level, that is theADC full scaleinputvoltage(VFS).Forweaknon-linearities, there isadirectmathematical relationshipbetween

CP and the third-order intercept point (IP3), such that IP3≈CP+10 dB. Forhigher-ordernon-linearities,thedifferencecanbelargerthan10dB,butthenCPis still agoodestimateof themaximumsignal levelwhile intermodulation forlowersignallevelsmaybeoverestimated.

19.5.3CompressionPointandGainBetweentheantennaandtheRXthereistheFEwithitsassociatedinsertionloss

(IL>1),forexampleduetoaT/Rswitch,apossibleRFfilter,andPCB/substratelosses.Theselosseshavetobeaccountedforinthegainandnoiseexpressions.KnowingIL,theCPicanbefoundthatcorrespondstotheADCclippingas

Theantennareferrednoisefactorandnoisefigurewillthenbecome

and

respectively.Whencomparingtwodesigns,forexample,at2and30GHz,respectively,the

30 GHz IL will be significantly higher than that of the 2 GHz. From the Fiexpressionitcanbeseenthattomaintainthesamenoisefigure(NFi)forthetwocarrierfrequencies,thehigherFElossat30GHzneedstobecompensatedforbyimprovingtheRXnoisefactor.Thiscanbeaccomplishedby(1)usingabetterLNA, (2) relaxing the input compression point, that is increasing G, or (3)increasingtheDRADC.UsuallyagoodLNAisalreadyusedat2GHztoachievea lowNFi, so thisoption is rarelypossible.RelaxingCPi is anoptionbut thiswill reduce IP3 and the linearity performancewill degrade. Finally, increasingDRADC comes at a power consumption penalty (4× per extra bit). EspeciallywidebandADCsmayhaveahighpowerconsumption,thatiswhenBWisbelowsome100MHztheN0···DRADCproduct(thatisBW···DRADC)isproportionalto the ADC power consumption, but for higher bandwidths the ADC powerconsumption isproportional toBW2 ···DRADC, therebypenalizinghigherBW(seeSection19.1).IncreasingDRADCistypicallynotanattractiveoptionanditis inevitable that the30GHzreceiverwillhaveasignificantlyhigherNFi thanthatofthe2GHzreceiver.

19.5.4PowerSpectralDensityandDynamic

19.5.4PowerSpectralDensityandDynamicRangeA signal consisting of many similar subcarriers will have a constant power-spectraldensity(PSD)overitsbandwidthandthetotalsignalpowercanthenbefoundasP=PSD···BW.When signals of different bandwidths but similar power levels are received

simultaneously, their PSDs will be inversely proportional to their BW. Theantenna-referred noise floor will be proportional to BW and Fi, or Ni=Fi···k···T···BW,asderivedabove.SinceCPiwillbefixed,givenbyGandADCclipping,thedynamicrange,ormaximumSNR,willdecreasewithsignalbandwidth,thatisSNRmax∝1/BW.The above signal can be considered as additive white Gaussian noise

(AWGN) with an antenna-referred mean power level (Psig) and a standarddeviation(σ).Basedonthisassumptionthepeak-to-average-powerratiocanbeapproximatedasPAPR=20···log10(k),wherethepeaksignalpowerisdefinedasPsig+k···σ,thatistherearekstandarddeviationsbetweenthemeanpowerlevelandtheclippinglevel.ForOFDManunclippedPAPRof10dBisoftenassumed(thatis3σ)andthismarginmustbesubtractedfromCPitoavoidclippingofthereceivedsignal.AnOFDMsignalwithanaveragepowerlevel,forexample,3σbelowtheclippinglevelwillresultinlessthan0.2%clipping.

19.5.5CarrierFrequencyandmm-WaveTechnologyAspectsDesigning a receiver at, for example, 30GHzwith a 1GHz signal bandwidthleavesmuchlessdesignmarginthanwhatwouldbethecasefora2GHzcarrierfrequency, fcarrier with, for example, 50 MHz signal bandwidth. The ICtechnologyspeedissimilarinbothcasesbutthedesignmarginandperformancedependonthetechnologybeingmuchfasterthantherequiredsignalprocessing,whichmeansthatthe2GHzdesignwillhavebetterperformance.Thegraphshowsexpectedevolutionofsometransistorparametersimportant

formm-waveICdesign,aspredictedbytheInternationalTechnologyRoadmapfor Semiconductors (ITRS). Here ft, fmax, and Vdd/BVceo data from the ITRS2007 targets [39] for CMOS and bipolar RF technologies are plotted vs thecalendaryearwhen the technology is anticipated tobecomeavailable. ft is the

transistortransitfrequency(thatis,wheretheRFdevice'scurrentgainis0dB),andfmaxisthemaximumfrequencyofoscillation(thatis,whentheextrapolatedpowergainis0dB).VddistheRF/high-performanceCMOSsupplyvoltageandBVceo isthebipolartransistor’scollector–emitterbaseopenbreakdownvoltagelimits.Forexample,anRFCMOSdeviceisexpectedtohaveamaximumVddof750mVby2020(othersupplyvoltageswillbeavailableaswell,butatalowerspeed).Thefreespacewavelengthat30GHzisonly1cm,whichisonetenthofwhat

isthecaseforexisting3GPPbandsbelow6GHz.Antennasizeandpathlossarerelated to wavelength and carrier frequency, and to compensate the smallphysicalsizeofasingleantennaelementmultipleantennas,forexample,arrayantennaswillhavetobeused.Whenbeam-formingisusedthespacingbetweenantennaelementswillstillberelatedtothewavelength,constrainingthesizeoftheFEandRX.Someoftheimplicationsofthesefrequencyandsizeconstraintsare:

•Theratiosft/fcarrierandfmax/fcarrierwillbemuchloweratmillimeterwavefrequenciesthanforbelow6GHzapplications.Asreceivergaindropswithoperatingfrequencywhenthisratioislessthansome10−100×,theavailablegainatmillimeterwaveswillbelowerandconsequentlythedevicenoisefactor,Fi,higher(similartowhenFriis’formulawasappliedtoatransistor’sinternalnoisesources).

•Thesemiconductormaterial’selectricalbreakdownvoltage(Ebr)isinverselyproportionaltothechargecarriersaturationvelocity(Vsat)ofthedeviceduetotheJohnsonlimit.ThiscanbeexpressedasVsat···Ebr=constantorfmax···Vdd=constant.Consequently,thesupplyvoltagewillbelowerformillimeter-wavedevicescomparedtodevicesinthelowGHzfrequencyrange.ThiswilllimittheCPiandthemaximumavailabledynamicrange.

•Ahigherleveloftransceiverintegrationisrequiredtosavespace,eitherassystem-on-chip(SoC)orsystem-in-package(SiP).ThiswilllimitthenumberoftechnologiessuitablefortheRFtransceiverandlimitFRX.

•RFfilterswillhavetobeplacedclosetotheantennaelementsandfitintothearrayantenna.Consequently,theyhavetobesmall,resultinginhigherphysicaltolerancerequirements,possiblyatthecostofinsertionlossandstop-bandattenuation.Thatis,ILandselectivitygetworse.

Thefilteringaspectformm-wavefrequenciesisfurtherelaboratedoninSection19.4.

Increasingthecarrierfrequencyfrom2GHzto30GHz(that is>10×)hasasignificant impact on the circuit design and itsRF performance. For example,modern high-speed CMOS devices are velocity saturated and their maximumoperatingfrequencyisinverselyproportionaltotheminimumchannellength,orfeaturesize.Thisdimensionhalves roughlyevery4years,asperMoore’s law(stating that complexity, that is transistor density, doubles every other year).With smaller feature sizes, internal voltages must also be lowered to limitelectrical fields to safe levels. Thus, designing a 30 GHz RF receivercorrespondstodesigninga2GHzreceiverusingabout15-year-oldlow-voltagetechnology (that is today’s breakdown voltage but 15 years old Ft (see Fig.19.18)withITRSdevicetargets).Withsuchamismatchindeviceperformanceanddesignmarginitisnottobeexpectedtomaintainboth2GHzperformanceandpowerconsumptionat30GHz.

FIGURE19.18 Expectedevolutionovertimeofsometransistorparameters:ft,fmax,andVdd/BVceo[39].

The signal bandwidth at mm-wave frequencies will also be significantlyhigherthanat2GHz.Foranactivedevice,orcircuit,thesignalswingislimitedbythesupplyvoltageatoneendandbythermalnoiseattheother.Theavailablethermal noise power of a device is proportional to BW/gm, where gm is theintrinsicdevicegain(trans-conductance).Asgmisproportionaltobiascurrentitcanbeseenthatthedynamicrangebecomestheratio

or

wherePisthepowerdissipation.Receivers formm-wave frequencieswillhave increasedpowerconsumption

due to their higherBW, aggravated by the low-voltage technology needed forspeed, compared to typical 2 GHz receivers. Thus, considering the thermalchallengesgiven the significantly reducedarea/volume formm-waveproducts,thecomplexinterrelationbetweenlinearity,NF,bandwidth,anddynamicrangeinthelightofpowerdissipationshouldbeconsidered.

19.6SummaryThis chapter gave an overview of what mm-wave technologies can offer andhow toderive requirements.Theneed forhighly integratedmm-wave systemswith many transceivers and antennas will require careful and often complexconsideration regarding the power efficiency and heat dissipation in smallarea/volumeaffectingtheachievableperformance.ImportantareaspresentedwereDA/ADconverters,poweramplifiers,andthe

achievable power versus efficiency as well as linearity. Receiver essentialmetricsarenoisefigure,bandwidth,dynamicrange,andpowerdissipationandtheyall have complexdependencies.Themechanism for frequencygeneration

aswellasphasenoiseaspectswerealsocovered.Filteringaspectsformm-wavefrequencieswere shown to have substantial impact in newNR bands and theachievable performance for various technologies and the feasibility ofintegratingsuchfiltersintoNRimplementationsneedstobeaccountedforwhendefining RF requirements. All these aspects are accounted for throughout theprocessofdevelopingtheRFcharacteristicsofNRinFrequencyRange2.

CHAPTER20

BeyondtheFirstReleaseof5G

Abstract

Thischapterprovidesahigh-leveloverviewofsomepossibleareasofNRevolutionbeyondthefirstrelease.Thisincludesintegratedaccessbackhaul(IAB),operationinunlicensedspectrum,device-to-devicecommunication,andnon-orthogonalmultipleaccess.

KeywordsIntegratedaccessbackhaul;IAB;unlicensedoperation;non-orthogonalmultipleaccess;NOMA;fullduplex;device-to-devicecommunication;D2D;sidelink

ThefirstreleaseofNR,release15,hasfocusedonbasicsupportforeMBBand,tosomeextent,URLLC.1Release15asdescribedinthepreviouschaptersisthefoundationuponwhichthefutureevolutionofNRwillbebuiltfor thecomingreleases.TheNRevolutionwillbringadditionalcapabilitiesandfurtherenhancethe performance. Not only will the additional capabilities provide betterperformance in existing applications, they may also open for, or even bemotivatedby,newapplicationareas.In the following, someareas inwhichNR is likely to evolve arediscussed.

Studiesinsomeoftheareasarealreadyongoingin3GPP,whileotherareasaremorerelevantforlaterreleases.

20.1IntegratedAccess-BackhaulTheuseofwirelesstechnologyforbackhaulhasbeenusedextensivelyformanyyears. In some regions of the world, wireless backhaul constitutes more than50%of total backhaul.Currentwireless-backhaul solutions are typicallybased

onproprietary(non-standardized)technologyoperatingaspoint-to-pointline-of-sightlinksusingspecialfrequencybandsabove10GHz.Thewirelessbackhaulisthususingdifferenttechnologyandoperatingindifferentspectra,comparedtotheaccess(base-station/device)links.Relaying,introducedinrelease10ofLTE,isbasicallyawirelessbackhaullink,althoughwithsomerestrictions.However,ithassofarnotbeenusedinpracticetoanysignificantextent.Onereasonisthatwirelesslyconnectedsmall-celldeployments, forwhich relayingwasdesigned,havenotyetbeenextensivelyusedinpractice.Anotherreasonisthatoperatorsprefertousetheirpreciouslow-frequencyspectrafortheaccesslink.Asalreadymentioned,currentwirelessbackhaulingreliesonnon-LTEtechnologiescapableof exploiting significantly higher-frequency bands thanLTE, thereby avoidingwastingvaluableaccessspectraforbackhaulpurposes.However,forNR,aconvergenceofbackhaulandaccesscanbeexpectedfor

severalreasons:

•Theaccesslinkcanexploitmm-wavefrequencies—thatis,thesamefrequencyrangethatiscurrentlyusedforwirelessbackhaul.

•Theexpecteddensificationofthemobilenetworks,withmanybasestationslocatedindoorandoutdooronstreetlevel,willrequirewirelessbackhaulcapableofoperatingundernon-line-of-sightconditionsand,moregenerally,verysimilarpropagationconditionsastheaccesslink.

The requirements and characteristics of the wireless backhaul link and theaccesslinkarethusconverging.Inessence,withreferencetoFig.20.1,thereis,radio-wise, no major difference between the wireless backhaul link and thenormal wireless link. Consequently, there are strong reasons to consider aconvergencealsointermsoftechnologyandspectrumwithasingleradio-accesstechnologythatcanbeusedforbothaccessandwirelessbackhaul.Thereshouldpreferably also be a common spectrum pool for both the access link and thewireless backhaul. It should also be noted that a common spectrum pool foraccessandwirelessbackhauldoesnotnecessarilymeanthattheaccesslinkandthewirelessbackhaullinkshouldoperateonthesamecarrierfrequency(“inbandrelaying”).Insomecases,thiswillbepossible.However,inothercases,havinga frequency separation between the backhaul link and the access link ispreferred.Thekeythingisthattheseparationofspectrumbetweenbackhaulandaccessshould,asmuchaspossible,notbearegulatoryissue.Rather,anoperatorshouldhaveaccesstoasinglespectrumpool.Itisthenanoperatordecisionhow

tousethisspectruminthebestpossiblewayandhowtosplititbetweenaccessandbackhaul.

FIGURE20.1 Wirelessbackhaulvstheaccesslink.

Toaddressbackhaulscenarios,astudyitemonintegratedaccess-backhaul[1]ispartof release15 toassess thepossibilitiesand techniquesforusingNRforbackhaulpurposes.TheNRradioaccessiswellpreparedtosupportthebackhaullinkandmostofthenecessaryworkisonhigher-layerprotocols.

20.2OperationinUnlicensedSpectraSpectrumisfundamentalforwirelesscommunicationandthereisanever-endingquest for more spectra to meet the ever-increasing demands of increasedcapacityandhigherdatarates.This isoneof thereasonsforsupportinghighercarrier frequencies inNR. The first release ofNRwas primarily designed forlicensedspectra.Suchspectraoffermanybenefitssincetheoperatorcanplanthenetworkandcontroltheinterference.Licensedspectrumisthusinstrumentaltoproviding quality-of-service guarantees andwide-area coverage. However, theamountof licensedspectraanoperatorhasaccess tomaynotbesufficientandthereistypicallyacostassociatedwithobtainingaspectrumlicense.Unlicensedspectra,ontheotherhand,areopenforanyonetouseatnocost,

subject to a set of rules, for example onmaximum transmission power. Sinceanyone can use the spectra, the interference situation is typically much moreunpredictable than for licensed spectra. Consequently, quality-of-service andavailability cannot be guaranteed. Furthermore, the maximum transmissionpower is modest, making it unsuitable for wide-area coverage. Wi-Fi andBluetooth are two examples of communication systems exploiting unlicensed

spectra in the lower-frequency range:2.4GHzor5GHz. Inaddition, someofthehigher-frequencybandswhichNRislikelytoaddressareunlicensed.Fromthediscussionabove,itcanbeseenthatthesetwospectrumtypeshave

different benefits and drawbacks. An attractive option is to combine the twosuchthatlicensedspectraareusedtoprovidewide-areacoverageandquality-of-serviceguarantees,withunlicensedspectrausedasa local-areacomplement toincrease user data rates and overall capacitywithout compromising on overallcoverage, availability, and reliability. This has been done as part of the LTEevolution, seeLicense-Assisted Access (LAA) in Chapter 4 and Fig. 20.2. ForNR,astudyonNR-basedAccesstoUnlicensedSpectrum[9]ispartofrelease15withspecificationworktargetingrelease16.

FIGURE20.2 License-assistedaccess.

Although NR release 15 does not support unlicensed spectra, it wasconsideredinthedevelopmentofthebasicNRframework.Oneexamplehereofisthepossibilitytotransmitoverafractionofaslot(seeChapter7).Extending

NRintoanLAA-likeoperationisthereforerelativelysimple,usingtheexistingflexibilityandfollowingtheapproachdevelopedforLTE.One important characteristic of operation in unlicensed spectra, which was

accountedfor in theLTE/LAAwork, isfairsharingofunlicensedspectrawithother operators and other systems, in particular Wi-Fi. There are severalmechanismsthatcanbeusedtoenablethis.Dynamicfrequencyselection(DFS),wherethenetworknodesearchesandfindsapartoftheunlicensedspectrawithlow load, can be used to avoid other systems if possible. Listen-before-talk(LBT) mechanism, where the transmitter ensures there are no ongoingtransmissions on the carrier frequency prior to transmitting, is anothermechanismwell proven at lower-frequency bands that could be added toNR.For higher-frequency bands, where extensive beam-forming is typically used,theLBTmechanismmayneedsomemodifications.Beyondlicense-assistedaccesstounlicensedspectra,acompletesolutionfor

standalone operation in unlicensed spectra can also be envisioned. Thisobviously requires mechanisms for system-information delivery and mobilitycapableofhandlingunlicensedspectra.

20.3Non-orthogonalMultipleAccessNR primarily uses orthogonal multiple-access where different devices areseparated in time and/or frequency. However, non-orthogonal access has thepotential toincreasecapacityinsomescenarios.DuringtheearlystagesofNRdevelopment,non-orthogonalmultipleaccess (NOMA)wasbrieflystudiedbutdown-prioritized.Nevertheless,studiesonNOMAareongoinginrelease15andmaybecomerelevantforNRinlaterreleases.

20.4Machine-TypeCommunicationMachine-typecommunicationisaverywideterm,coveringmanydifferentusecases and scenarios. It is common todividemachine-type communication intomassive machine-type communication and ultra-reliable low-latencycommunication(URLLC),asalreadydiscussedatthebeginningofthisbook.Massive machine-type communication refers to scenarios where a device

typically sendsavery small amountofdata,has relaxed latency requirements,butlowpowerconsumptionandlowcostareatpremium.Thenumberofdevicesisoftenverylarge.SuchscenarioswillbeaddressedbyLTEandNB-IoTforthe

near-to mid-term perspective, in particular for the low-end massive MTCregime. Specific mechanisms such as the reserved resources discussed inChapter 17havebeen introduced to simplify the coexistencebetweenNRandthese access technologies. In the longer time perspective, NR is expected toevolvewith improvednativesupportofmassivemachine-typecommunication,primarily focusing on themid-to-high-endmassiveMTC.Reduced bandwidthsupport,extendedsleep-modesolutions,wake-upsignaling,andnon-orthogonalwaveformsareexamplesofwhatcouldbe relevant tostudyaspartof suchanevolution.Factoryautomation isanexampleofanapplicationarea related tomachine-

typecommunication. Inmanycases, suchapplicationsaredemanding in termsof reliability and latency and the URLLC aspects of NR are therefore highlyrelevant. Examples of possible enhancements to NR relevant for factoryautomationarehigher-layerenhancementstosupportcommonlyusedindustrialprotocols(otherthanTCP/IP)andlocalbreakoutfromthecorenetwork.

20.5Device-To-DeviceCommunicationSupportfordirectdevice-to-device(D2D)connectivity(Fig.20.3),alsoreferredtoassidelinkconnectivity,usingLTEwasintroducedin3GPPrelease12withtwomainusecasesinmind:

•Device-to-devicecommunication,focusingonthepublic-safetyusecase;•Device-to-devicediscovery,targetingpublicsafetybutalsocommercialusecases.

FIGURE20.3 Device-to-deviceconnectivity.

TheD2Dframeworkhasalsoservedas thebasis for theV2V/V2Xwork intheLTEevolutioninlaterreleasesasdiscussedinChapter4.NRrelease15doesnotsupportdirectdevice-to-devicecommunication,butit

is a likely candidate for a future release. Insteadof focusingon a specific usecase,device-to-deviceconnectivityshouldbeseenasageneral tool toenhanceconnectivity within the 5G network. In essence, direct data transfer betweendevicesshouldbeconfiguredifthenetworkconcludesthatthisismoreefficient(requires less resources) or provides better quality (higher data rates and/orlower latency) compared to indirect connectivity via the infrastructure. Thenetworkshouldalsobeabletoconfiguredevice-basedrelaylinkstoenhancetheconnectivityquality,forexampleformassivemachine-typedeviceswithbadornocoverage.ThelowerlatencyofNRcouldalsoprovevaluableforsomeD2Dapplications,forexampleplatooning,asmentionedinChapter4.

20.6SpectrumandDuplexFlexibilityDuplexflexibilityisawidearea,aimingatimprovingtheusageoftheavailablespectrum.ThetoolspartofNRfromthestart—forexamplebandwidthparts,aflexible slot structure, and carrier aggregation also across duplex schemes—providea lotof flexibility andensureNRcanbedeployed in awide rangeofscenarios.Nevertheless,furtherenhancementsinthisareacanbeenvisioned.Currently,theFDDspectrumissplitintoadownlinkpartandanuplinkpart.

However,whatisrelevantfromatechnicalperspectiveisprimarilynotdownlinkvsuplink,butlowpowervshighpower.Thedownlinktypicallyuseshighpowerand relatively high above-rooftop antennas,while the uplinkuses significantlylowertransmissionpowerandantennainstallations.Hence,fromaninterferenceperspective, a low-power downlink transmission in the uplink spectrum is notdifferent from a low-power uplink transmission in the same spectrum.Consequently, there are ideas on allowing downlink transmission also in theuplinkbands.Tosomeextent,thisistheFDDcounterparttodynamicTDDasitallows for adynamicchange to the “transmissiondirection.”Froma technicalperspective,NRiswellprepared tosuchenhancementsbecauseof the flexibleslotstructure.Thepotentialissuesareprimarilyregulatory.Another area related to spectra and possible future enhancements is

interferencemeasurementsanddynamicTDD.TheTDDschemeinNRisbuiltupon a dynamic framework and dynamicTDD is therefore part of release 15.However, inpractice,suchdeploymentsareprimarily limited tosmallcells. In

larger cells, with a correspondingly higher downlink transmission power, theintercell interference typically calls for a more static duplex operation. OnepossibilitytoimprovethenumberofscenarioswheredynamicTDDisfeasiblecouldbetoincludevariousinterferencemeasurementmechanisms.Forexample,iftheschedulerknowstheinterferencesituationforthedifferentdevices,itcanscheduledynamicallyforsomedeviceswhiletakingamorestaticapproachforotherdevices.Differentintercellinterferencecoordinationmechanismscanalsobethoughtof.Therehaverecentlybeendifferentproposalsfor“true”full-duplexoperation

[53].Inthiscontext,full-duplexoperationmeansthattransmissionandreceptionarecarriedoutatthesamefrequencyatthesametime(seealsoFig.20.4).2Full-duplex operation obviously leads to very strong “self” interference from thetransmittertothereceiver,aninterferencethatneedstobesuppressed/canceledbeforetheactualtargetsignalcanbedetected.

FIGURE20.4 Fullduplexonlinklevelvscelllevel.

Inprinciple, such interference suppression/cancellation is straightforward,asthe interfering signal is in principle completely known to the receiver. Inpractice, the suppression/cancellation is far from straightforward due to theenormousdifferencebetween the target signal and the interference in termsofreceivedpower.Tohandlethis,currentdemonstrationsoffull-duplexoperationrelyonacombinationof spatial separation (separateantennas for transmissionand reception), analog suppression, anddigital cancellation.The technology isstilltoalargedegreeattheresearchlevelandnotmatureenoughforlarge-scaledeployments. Implementation on the network-side only (see right part of Fig.20.4)might be less complex than implementation on the device-side due to ahigher degree of spatial separation of receive and transmit antennas on the

networkside.Even if full duplex would be feasible in real implementation, its benefits

should not be overestimated. Full duplex has the potential to double the linkthroughput by allowing for continuous transmission in both directions on thesame frequency.However, therewill then be two simultaneous transmissions,implying increased interference to other transmissions, something which willnegativelyimpacttheoverallsystemgain.Thelargestgainfromfullduplexcanthereforebeexpectedtooccurinscenarioswithrelativelyisolatedradiolinks.

20.7ConcludingRemarksAbove, some examples of technology areas relevant for NR evolution areoutlined.SomeofthesearelikelytobepartoffutureNRreleases,whileothermaynot happen at all.However, as always,when trying to predict the future,there are a lot of uncertainties and new, not-yet-known requirements ortechnologies, which may motivate evolutions into directions not discussedabove.TheemphasisonfuturecompatibilityinthebasicNRdesignensuresthatintroductionof extension inmost cases is relatively straightforward. It is clearthoughthatNRisaveryflexibleplatform,capableofevolvinginawiderangeofdirectionsandanattractivepathtofuturewirelesscommunication.

1ThefirstNRversionprimarilyaddressedthelow-latencypartofURLLC.Meanstoincreasethereliabilityareworkeduponinthelatterpartsofrelease15,targetingthefinalNRrelease15inJune2018.2Nottobeupmixedwithfull-duplexFDDasusedinLTE.

References

1.3GPPRP-172290,NewSIDProposal:StudyonIntegratedAccessandBackhaulforNR.

2.3GPPTS37.141,E-UTRA,UTRAandGSM/EDGE;Multi-StandardRadio(MSR)BaseStation(BS)ConformanceTesting.

3.3GPPR1-163961,FinalReportof3GPPTSGRANWG1#84bis.

4.3GPPTS38.104,NR;BaseStation(BS)RadioTransmissionandReception.

5.3GPPTS38.101-1,NR;UserEquipment(UE)RadioTransmissionandReception.Part1.Range1Standalone.

6.3GPPTS38.101-2,NR;UserEquipment(UE)RadioTransmissionandReception.Part2.Range2Standalone.

7.3GPPTS38.101-3,NR;UserEquipment(UE)RadioTransmissionandReception.Part3.Range1andRange2InterworkingOperationwithOtherRadios.

8.3GPPTS38.101-4,NR;UserEquipment(UE)RadioTransmissionandReception.Part4.PerformanceRequirements.

9.3GPPRP-172021,StudyonNR-BasedAccesstoUnlicensedSpectrum.

10.3GPPTR36.913,RequirementsforFurtherAdvancementsforEvolvedUniversalTerrestrialRadioAccess(E-UTRA)(LTE-Advanced)(Release9).

11.3GPPTR38.803,StudyonNewRadioAccessTechnology:RadioFrequency(RF)andCoexistenceAspects.

12.3GPPTS23.402,ArchitectureEnhancementsforNon-3GPP

12.3GPPTS23.402,ArchitectureEnhancementsforNon-3GPPAccesses.

13.3GPPTS23.501,SystemArchitectureforthe5GSystem.14.3GPPTS36.211,EvolvedUniversalTerrestrialRadioAccess

(E-UTRA);PhysicalChannelsandModulation.15.3GPPTS38.331,NR;RadioResourceControl(RRC)

ProtocolSpecification(Release15).16.3GPPTR36.913,RequirementsforFurtherAdvancements

forEvolvedUniversalTerrestrialRadioAccess(E-UTRA)(LTE-Advanced).

17.ArikanE.Channelpolarization:amethodforconstructingcapacity-achievingcodesforsymmetricbinaryinputmemorylesschannels.IEEETransInformTheory.July2009;55(7):3051–3073.

18.BestRolandE.PhasesLockedLoops:Design,SimulationandApplicationssixthed.McGraw-HillProfessional2007.

19.CEPT/ERCRecommendation74-01onunwantedemissionsinthespuriousdomain,Cardiff2011.

20.CEPT,LSfromtoCEPT/ECCSE21,SE21(17)38,September2017.

21.ChapmanT,LarssonE,vonWryczaP,DahlmanE,ParkvallS,SköldJ.HSPAEvolution:TheFundamentalsforMobileBroadbandAcademicPress2014.

22.ChaseD.Codecombining—amaximum-likelihooddecodingapproachforcombiningandarbitrarynumberofnoisypackets.IEEETransCommun.May1985;33:385–393.

23.ChenJ.DoesLOnoisefloorlimitperformanceinmulti-Gigabitmm-wavecommunication?.IEEEMicrowComponLett.2017;27(8):769–771.

24.ChengJ-F.CodingperformanceofhybridARQschemes.IEEETransCommun.June2006;54:1017–1029.

25.ChuDC.Polyphasecodeswithgoodperiodiccorrelationproperties.IEEETransInformTheory.July1972;18(4):531–532.

26.ChungST,GoldsmithAJ.Degreesoffreedominadaptivemodulation:aunifiedview.IEEETransCommun.September2001;49(9):1561–1571.

27.ColombiD,ThorsB,TörnevikC.ImplicationsofEMFexposurelimitsonoutputpowerlevelsfor5Gdevicesabove6GHz.IEEEAntennasWirelPropagLett.February2015;14:1247–1249.

28.DahlmanE,ParkvallS,SköldJ.4GLTE-AdvancedProandtheRoadto5GElsevier2016.

29.DIGITALEUROPE,5GSpectrumOptionsforEurope,October2017.

30.Ericsson,EricssonMobilityReport,November2017.https://www.ericsson.com/assets/local/mobility-report/documents/2017/ericsson-mobility-report-november-2017.pdf.

31.Ericsson,Onmm-waveFiltersandRequirementImpact,R4-1712718,3GPPTSG-RANWG4Meeting#85,December2017.

32.FederalCommunicationsCommission,Title47oftheCodeofFederalRegulations(CFR).

33.P.Frenger,S.Parkvall,E.Dahlman,PerformancecomparisonofHARQwithchasecombiningandincrementalredundancyforHSDPA.In:ProceedingsoftheIEEEVehicularTechnologyConference,AtlanticCity,NJ,USA,pp.1829–1833.October2001.

34.GallagerRG.LowDensityParityCheckCodesMonograph,M.I.T.Press1963.

35.GlobalmobileSuppliersAssociation(GSA),ThefutureofIMTinthe3300–4200MHzfrequencyrange,June2017.

36.M.Hörberg,LowphasenoiseGaNHEMToscillatordesignbasedonhigh-Qresonators(Ph.D.Thesis),ChalmersUniversityofTechnology,April2017.

37.IEEE,IEEEStandardforLocalandmetropolitanareanetworksPart16:AirInterfaceforBroadbandWireless

networksPart16:AirInterfaceforBroadbandWirelessAccessSystemsAmendment3:AdvancedAirInterface,IEEEStd802.16m-2011(AmendmenttoIEEEStd802.16-2009).

38.IETF,Robustheadercompression(ROHC):frameworkandfourprofiles:RTP,UDP,ESP,andUncompressed,RFC3095.

39.ITRS,RadioFrequencyandAnalog/Mixed-SignalTechnologiesforWirelessCommunications,EditionInternationalTechnologyRoadmapforSemiconductors(ITRS),2007.

40.ITU-R,Workplan,timeline,processanddeliverablesforthefuturedevelopmentofIMT,ITU-RDocument5D/758,Attachment2.12.

41.ITU-R,FrameworkandoverallobjectivesofthefuturedevelopmentofIMT-2000andsystemsbeyondIMT-2000.RecommendationITU-RM.1645,June2003.

42.ITU-R,Unwantedemissionsinthespuriousdomain.RecommendationITU-RSM.329-12,September2012.

43.ITU-R,FuturetechnologytrendsofterrestrialIMTsystems.ReportITU-RM.2320,November2014.

44.ITU-R,TechnicalfeasibilityofIMTinbandsabove6GHz.ReportITU-RM.2376,November2014.

45.ITU-R,DetailedspecificationsoftheterrestrialradiointerfacesofInternationalMobileTelecommunicationsAdvanced(IMT-Advanced).RecommendationITU-RM.2012-2,September2015.

46.ITU-R,FrequencyarrangementsforimplementationoftheterrestrialcomponentofInternationalMobileTelecommunications(IMT)inthebandsidentifiedforIMTintheRadioRegulations.RecommendationITU-RM.1036-5,October2015.

47.ITU-R,IMTVision—FrameworkandoverallobjectivesofthefuturedevelopmentofIMTfor2020andbeyond.RecommendationITU-RM.2083,September2015.

RecommendationITU-RM.2083,September2015.48.ITU-R,Radioregulations,Editionof2016.49.ITU-R,Detailedspecificationsoftheterrestrialradio

interfacesofInternationalMobileTelecommunications-2000(IMT-2000).RecommendationITU-RM.1457-13,February2017.

50.ITU-R,GuidelinesforevaluationofradiointerfacetechnologiesforIMT-2020.ReportITU-RM.2412November2017.

51.ITU-R,MinimumrequirementsrelatedtotechnicalperformanceforIMT-2020radiointerface(s).ReportITU-RM.2410November2017.

52.ITU-R,Requirements,evaluationcriteriaandsubmissiontemplatesforthedevelopmentofIMT-2020.ReportITU-RM.2411November2017.

53.JainM,etal.Practical,Real-Time,Full-duplexWirelessLasVegas,NV,USA:MobiCom’11;September19–23,2011.

54.JohnsonEO.Physicallimitationsonfrequencyandpowerparametersoftransistors.RCARev.June,1965;26:163–177.

55.LarssonEG,EdforsO,TufvessonF,MarzettaTL.MassiveMIMOfornextgenerationwirelesssystems.IEEECommunMag.February2014;52(2):186–195.

56.LeeJ,etal.Spectrumfor5G:globalstatus,challenges,andenablingtechnologies.IEEECommunMag.March2018.

57.LeesonDB.Asimplemodeloffeedbackoscillatornoisespectrum.ProcIEEE.February1966;54.

58.LibergO,SundbergM,WangE,BergmanJ,SachsJ.CellularInternetofThings:Technologies,Standards,andPerformanceAcademicPress2017.

59.MacKayDJC,NealRM.Nearshannonlimitperformanceoflowdensityparitycheckcodes.ElectronLett.July1996;33.

60.Motorola,ComparisonofPARandCubicMetricforPowerDe-rating,R1-040642.

61.MurmannB.Theracefortheextradecibel:abriefreviewof

currentADCperformancetrajectories.IEEESolStateCircMag.Summer2015;7(3):58–66.

62.Murmann,B.,ADCPerformanceSurvey1997–2017[Online].Available:http://web.stanford.edu/Bmurmann/adcsurvey.html.

63.OlssonM,SultanaS,RommerS,FridL,MulliganC.SAEandtheEvolvedPacketCore—DrivingtheMobileBroadbandRevolutionAcademicPress2009.

64.OnggosanusiE,etal.Modularandhigh-resolutionchannelstateinformationandbeammanagementfor5Gnewradio.IEEECommunMag.March2018;56.

65.PadhyeJ,FiroiuV,TowsleyDF,KuroseJF.Modelling,TCPrenoperformance:asimplemodelanditsempiricalvalidation.ACM/IEEETransNetw.2000;8(2):133–145.

66.ParkvallS,DahlmanE,FuruskaärA,FrenneM.NR:thenew5Gradioaccesstechnology.IEEECommunStandMag.December2017;1(4):24–30.

67.PursleyMB,SandbergSD.Incremental-redundancytransmissionformeteorburstcommunications.IEEETransCommun.May1991;39:689–702.

68.RichardsonT,UrbankeR.ModernCodingTheoryCambridgeUniversityPress2008.

69.ShannonCE.Amathematicaltheoryofcommunication.BellSystTechJ.JulyandOctober1948;27(379–423):623–656.

70.SpecialIssueonSpreadSpectrum,IEEETrans.Commun.25,745–869.August1977.

71.WickerSB,BartzM.Type-IhybridARQprotocolsusingpuncturedMDScodes.IEEETransCommun.April1994;42:1431–1440.

72.Wozencraft,J.M.,Horstein,M.,DigitalisedCommunicationOverTwo-wayChannels,FourthLondonSymposiumonInformationTheory,London,UK,September1960.

73.MollenC,LarssonEG,GustavssonU,ErikssonT,HeathRW.

Out-of-bandradiationfromlargeantennaarrays.IEEECommunMag.April2018;56.

74.3GPP,NR;GeneralaspectsforUERFforNR,3GPPTR38.817-01.

75.3GPP,NR;GeneralaspectsforBSRFforNR,3GPPTR38.817-02.

Index

Note:Pagenumbersfollowedby“f”and“t”refertofiguresandtables,respectively.

A

Absolutepowertolerance,366

AccessandMobilityManagementFunction(AMF),74–75

Accessstratum(AS),74–75

ACIR,SeeAdjacentchannelinterferenceratio(ACIR)

Acknowledgedmode(AM),85,266,267,269–273

SDUdelivery,272f

ACLR,SeeAdjacentchannelleakageratio(ACLR)

ACS,SeeAdjacentchannelselectivity(ACS)

ActiveAntennaSystembasestations,389–390

Activeantennasystems(AASs),358,358

BSrequirements,358–359

generalizedradioarchitectureof,358f

Activedownlinkbandwidthpart,113–114

Activeuplinkbandwidthpart,113–114

ADCs,SeeAnalog-to-DigitalConverters(ADCs)

Additionalmaximumpowerreduction(AMPR),363–364,365

AdditivewhiteGaussiannoise(AWGN),391,408

Adjacentchannelinterferenceratio(ACIR),371–372

Adjacentchannelleakageratio(ACLR),367,371–372,371f,378,398

Adjacentchannelparameters,372

Adjacentchannelselectivity(ACS),354,371f,372,375

AdvancedAntennaSystems(A-ASs),349

AdvancedMobilePhoneSystem(AMPS),1

Advancedmultiantennatransmission/reception,59

Aerials,55

AF,SeeApplicationFunction(AF)

AGC,SeeAutomaticGainControl(AGC)

Aggregatedpowertolerance,366

Aggregationlevel,188,195

Aggregationofspectrumallocations,352

Allocations,352

Always-onsignals,59–60

Always-ontransmissions,60

AM,SeeAcknowledgedmode(AM)

AMF,SeeAccessandMobilityManagementFunction(AMF)

AMPR,SeeAdditionalmaximumpowerreduction(AMPR)

AMPS,SeeAdvancedMobilePhoneSystem(AMPS)

Analogantennaprocessing,243

Analogbeamforming,63

Analogfront-end,possibilitiesoffilteringat,399–401

Analogmultiantennaprocessing,231

Analog-to-DigitalConverters(ADCs),390–391

Analysis,21

Antenna,155–156SeealsoMultiantennatransmission

array,358–359

composite,358–359

portfields,167

ports,128–130,129t,165

selection,239

Aperiodic

CSI-RStransmission,140

reporting,147,147

SRS,151,242

ApplicationFunction(AF),75

Architecture

options,342,343f

phase,22

Areatrafficcapacity,18

“Around-the-corner”dispersion,243

ARQ,SeeAutomaticrepeat-request(ARQ)

AS,SeeAccessstratum(AS)

Associatedcontrolsignaling,185

Asynchronoushybrid-ARQprotocol,93

AuthenticationServerFunction(AUSF),75

AutomaticGainControl(AGC),405

Automaticrepeat-request(ARQ),67

AWGN,SeeAdditivewhiteGaussiannoise(AWGN)

B

Backwardscompatibility,42

Bandcategories(BC),382

Band-specificdevicerequirements,363–364

Bandwidth(BW),207–209,401–402

adaptation,62,280–282,281f

ofcarrier,354

dependencies,405–411

Bandwidthparts(BWPs),61–64,62,112–114,113f

Bandwidth-partindicator(0–2bit),204

Basestation(BS),41,349

classes,364–365

colocationofBSequipmentbetweenoperators,351

conductedRFrequirementsforNR,357–359

controlof,47

OBUElimits,368–370

outputpoweranddynamicrange,365

radiatedRFrequirementsforNR,357–359

spuriousemission,399

structureofBSRFrequirements,357–360

conductedandradiatedRFrequirementsforNRBS,357–359

timealignment,367

type1-C,359,359

type1-H,359–360,359,378–379

type1-O,359,360,378

type2-O,359,360

typesindifferentFRsforNR,359–360

Base-stationdynamicrange,374

Baselinepowercontrol,304–306

Basiclimit,359–360

Basicrandom-accessprocedure,325

BC,SeeBandcategories(BC)

BCCH,SeeBroadcastControlChannel(BCCH)

BCH,SeeBroadcastChannel(BCH)

Beamadjustment,245–249

beamindicationandTCI,248–249

downlinkreceiver-side,247,247f

downlinktransmitter-side,245–246,246f

uplink,247–248

Beamcorrespondence,243–244

Beamestablishmentduringinitialaccess,332–333

Beamfailure/recovery,249,250

Beamindication,248–249

Beammanagement,231,243

beamadjustment,245–249

beamrecovery,249–252

Beamrecovery,249–252

beam-failuredetection,250

devicerecoveryrequestandnetworkresponse,251–252

new-candidate-beamidentification,250–251

procedure,325

request,251,251–252

Beam-basedpowercontrol,306–308

multipleclosed-loopprocesses,308

multipleopen-loop-parametersets,307–308

multiplepath-loss-estimationprocesses,306–307

useofmultiplepower-estimationprocesses,307f

Beam-centricdesign,68–69

Beam-failuredetection,250,250

Beam-failureevents,SeeBeamfailure/recovery

Beam-failureinstance,250

Beamforming,55,68f,243,245,409

function,41

forSSblock,317

Beam-sweeping

forpreambletransmission,332

forSS-blocktransmission,317

Bipolardevice,394–395

Bit-levelscramblingsequence,162–163

Bitmap-1,172,347

Bitmap-2,172

Blinddecoding,195–199

Blocking,375

Bluetooth,415

BroadcastChannel(BCH),87,155

BroadcastControlChannel(BCCH),87,266

BS,SeeBasestation(BS)

Bucketsizeduration(BSD),290

Bufferstatusreports,292–294

BW,SeeBandwidth(BW)

BWPs,SeeBandwidthparts(BWPs)

C

C-MTC,SeeCriticalmachinetypecommunication(C-MTC)

C-RNTI,SeeCellRadio-NetworkTemporaryIdentifier(C-RNTI)

CA,SeeCarrieraggregation(CA)

CACLR,SeeCumulativeACLRrequirement(CACLR)

Candidatebeams,250

identification,250

Candidatetechnology,21

Capabilityset(CS),382

Carrieraggregation(CA),27–28,44–45,44f,90,90,90f,91,115–117,118,119f,341,352,382

controlsignaling,116–117

relationto,119–120

Carrierfrequencyandmm-wavetechnologyaspects,408–411

Carrierindicator(0or3bit),204

Carrierraster,70,316

Carrierresourceblocks,168

Carrier-selectionthreshold,336–337

CBG,SeeCode-blockgroup(CBG)

CBGFlushIndicator(CBGFI),259–260

CBGtransmissionindicator(CBGTI),202,204,259–260

CBGTransmitIndicator,SeeCBGtransmissionindicator(CBGTI)

CBGFI,SeeCBGFlushIndicator(CBGFI)

CBGTI,SeeCBGtransmissionindicator(CBGTI)

CCCH,SeeCommoncontrolchannel(CCCH)

CCEs,SeeControlchannelelements(CCEs)

cDAI,SeeCounterDAI(cDAI)

CDM,SeeCode-domainsharing(CDM)

CDMA-basedIS-95technology,1–2

Cell,116,336–337

group,84–85

reselection,99

systeminformation,336–337

CellRadio-NetworkTemporaryIdentifier(C-RNTI),98–99,335,335–336

Cellsearch,313–324

detailsofPSS,SSS,ANDPBCH,319–323

frequency-domainpositionofSSblock,315–316

providingremainingsysteminformation,324

SSblock,313–315

periodicity,316–317

SSburstset,317–319,317f

Cell-specificreferencesignals(CRS),40,134

CellBarredflag,322

Cellularsystems,52–53

Channelbandwidth(BWChannel),353–356,355f,356t

independent,350

Channelcharacteristicsofinterest,133

Channelcoding,157–160,157f,158–160

code-blocksegmentation,157–158

CRCattachmentpertransportblock,157

ofPDCCH,187–188

“Channelhardening”effect,277–278

Channelqualityindicator(CQI),145,233

Channelsounding,133

downlink,134–144

basicCSI-RSstructure,134–137,135f

CSI-IM,140–141

CSI-RSresourcesets,142

frequency-domainstructureofCSI-RSconfigurations,137–139

mappingtophysicalantennas,143–144

time-domainpropertyofCSI-RSconfigurations,139–140

TRS,142–143,143f

zero-powerCSI-RS,141–142

downlinkmeasurementsandreporting,144–147

measurementresource,145–146

reportquantity,145

reporttypes,146–147

uplink,147–153

mappingtophysicalantennas,152–153

multiportSRS,150–151,150f

SRSresourceset,151

SRSsequencesandZadoff–Chusequences,149–150

time-domainstructureofSRS,151

Channel-dependentscheduling,66,91,277

Channel-estimation

accuracy,217

process,166

Channel-stateinformation(CSI),68,92,145,174,213

Channel-state-informationforinterferencemeasurements(CSI-IM),140–141

alternativestructures,141f

resourcesets,142

Channel-state-informationreferencesignals(CSI-RS),127,128f,133,134–144,146,167,174,211,246,248,250SeealsoSoundingreferencesignals(SRS)

basicstructure,134–137,135f

CSI-IM,140–141

densityequaltoone,139

frequency-domainstructureofCSI-RSconfigurations,137–139

mappingtophysicalantennas,143–144

periodicityandslotoffset,140f

resourcesets,142

time-domainpropertyofCSI-RSconfigurations,139–140

TRS,142–143,143f

zero-power,141–142

Chasecombining,257,257–258

Closed-loop

powercontrol,303

spatialmultiplexing,41

timingcontrol,326,326–327

CMOS,394–395,397

CN,SeeCoreNetwork(CN)

Co-siteddeployments,341,342f,342f

Code-blockgroup(CBG),67,94–95,95f,158,257

retransmissions,256f,257

Code-blocksegmentation,157–158,158f

Code-domainsharing(CDM),135,136f

frequency-domain,137

time/frequency-domain,137

Codebook-basedbeamforming,41

Codebook-basedprecoding,167,167,240,241

Codebook-basedtransmission,237,238–240,239f,241f

single-layeruplinkcodebooksforcaseoffourantennaports,240f

CodedUL-SCHstream,225

Coexistencebetweenoperators,351

ofTDDsystems,351

Coexistencewithservices,351

ColocationofBSequipmentbetweenoperators,351

“Comb”structure,148

Commoncontrolchannel(CCCH),87,97,266

Commonresourceblocks(CRBs),110–111,111f,176

gridoffset,323,323

Commonsearchspaces,199

CoMP,SeeCoordinatedmultipoint(CoMP)

ComplementarySULcarrier,336–337

Componentcarriers,44

Compositeantenna,358–359

Compressionpointandgain,407–408

Conductedoutputpowerlevelrequirements

BSoutputpoweranddynamicrange,365

deviceoutputpoweranddynamicrange,365–366

Conductedreceivercharacteristics,362,363t

ConductedRFrequirements

forNR,360–366

band-specificdevicerequirementsthroughnetworksignaling,363–364

BSclasses,364–365

conductedoutputpowerlevelrequirements,365–366

conductedreceivercharacteristics,362

conductedtransmittercharacteristics,361

regionalrequirements,362–363

forNRBS,357–359

Conductedsensitivity,374

Conductedtransmittercharacteristics,361,362t

Conductedunwantedemissionsrequirements,367–374

ACLR,371–372

emissionmaskinOOBdomain,368–370

implementationaspects,367–368

occupiedbandwidth,373

spuriousemissions,373

transmitterintermodulation,373–374

Configurablefrequency-domainRACHresource,327

ConfigurableRACHperiodicity,327

Configuredgranttype1,297

Configuredgranttype2,298

Configuringreservedresources,171,172f

Connected-statemobility,102

Connectiondensity,19

Connectionmanagement,97

Contention

contention-freerandomaccess,334

resolution,335

resolutionandconnectionsetup,335–336

Continuouswavesignal(CWsignal),375

Controlchannelelements(CCEs),186,188,188,192f

Controlchannels,67–68,86–87

structureenhancement,48

Controlindicator,267

Controlresourcesets(CORESETs),67,113,186,189–195,190f,191f,324

exampleofQCLrelationforPDCCHbeammanagement,194f

normalRSstructureandwidebandRSstructure,194f

Controlsignaling,65–66,116–117,120

Control-planefunctions,74–75

Control-planeprotocols,97–102SeealsoUser-planeprotocols

connected-statemobility,102

idle-stateandinactive-statemobility,99–102

RRCstatemachine,97–99,98f

Control-plane/user-planesplit,74

Coordinatedmultipoint(CoMP),43,47,48f

hypotheses,48

CoreNetwork(CN),73

deviceidentifier,335

CORESETs,SeeControlresourcesets(CORESETs)

Corporatecombiners,397

Countvalue,276

CounterDAI(cDAI),264–265

CPi,SeeInputcompressionpoint(CPi)

CQI,SeeChannelqualityindicator(CQI)

CRBs,SeeCommonresourceblocks(CRBs)

CRC,SeeCyclicredundancycheck(CRC)

Criticalmachinetypecommunication(C-MTC),14–15

Cross-carrierscheduling,116,279,280f

Cross-scheduling,116f

CRS,SeeCell-specificreferencesignals(CRS)

CS,SeeCapabilityset(CS)

CSI,SeeChannel-stateinformation(CSI)

CSI-IM,SeeChannel-state-informationforinterferencemeasurements(CSI-IM)

CSI-ReportConfig,144–145

CSI-RS,SeeChannel-state-informationreferencesignals(CSI-RS)

Cubicmetric,61,61–62,163

CumulativeACLRrequirement(CACLR),372

CWsignal,SeeContinuouswavesignal(CWsignal)

Cyclicredundancycheck(CRC),256,323

attachmentpertransportblock,157

forerror-detectingpurposes,155–156

Cyclicshift,151,215,328

D

D-AMPS,SeeDigitalAMPS(D-AMPS)

D2Dcommunication,SeeDevice-to-devicecommunication(D2Dcommunication)

DACs,SeeDigital-to-AnalogConverters(DACs)

DAI,SeeDownlinkassignmentindex(DAI)

Data

allocation,175

indicator,267

radiobearers,79

scramblingidentity,163

transmission,48,66–67,287t

DCCH,SeeDedicatedcontrolchannel(DCCH)

DCI,SeeDownlinkcontrolinformation(DCI)

Decoding,187–188

Dedicatedcontrolchannel(DCCH),87,97

DedicatedTrafficChannel(DTCH),87

Demodulationreferencesignals(DMRSs),129–130,165,165–166,167,174,177f,178f,193,232,315

forDFT-precodedOFDMuplink,181–183

forOFDM-baseddownlinkanduplink,175–181,178f,180f

DenseUrban-eMBB,21

Denserreferencesignalpattern,193

Densification,48–52

Deploymentscenarios,21,340–341

Detailedspecification,23

Device

device-specificsearchspaces,197

enhancements,52

in-bandemissions,366

outputpoweranddynamicrange,365–366

recoveryrequest,251–252

RFrequirements,structureof,356–357

SEM,370

spuriousemissionlimits,373

transmissionofpreamble,324

Device-to-devicecommunication(D2Dcommunication),52–53,53f,417,418f

Device-to-devicediscovery,417

DFS,SeeDynamicfrequencyselection(DFS)

DFT,164,328

DFT-precodedOFDM,40,61,103–104,215

OFDMuplink,181–183

DFT-precoding,155–156,163,164f,221SeealsoMultiantennaprecoding

uplink,164

Difficultbandcombinations,343

DigitalAMPS(D-AMPS),1–2

Digitalbeamforming,332

Digitalmultiantennaprocessing,231

Digitalprocessing,229–230

Digital-to-AnalogConverters(DACs),390–391

DIGITALEUROPE,31

DirectD2Dconnectivity,417

Discontinuousreception(DRX),87–88,98–99,298–302,301f

functionality,300

Discretemm-wavefilters,399

Diversespectrumallocations,350

DL-SCH,SeeDownlinkSharedChannel(DL-SCH)

DMRSs,SeeDemodulationreferencesignals(DMRSs)

Donorcell,49

Double-symbolreferencesignal,179–181

Downlink,155,185–212,418SeealsoUplink

beam,332

blinddecodingandsearchspaces,195–199

channel-dependentscheduling,92

controlchannels,67

controlresourceset,189–195

controlsignaling,185

downlinkschedulingassignments,199–202

hybrid-ARQ,259–260

interferencescenario,50

L1/L2controlsignaling,168,185

measurementsandreporting,144–147

measurementresource,145–146

reportquantity,145

reporttypes,146–147

multiantennatransmission,128

PDCCH,186–189

precoding,165–166,165f

preemptionhandling,282–283

preemptionindication,205

receiver-sidebeamadjustment,247,247f

reservedresources,168,171–173

scheduler,91,278

scheduling,91

assignments,199–202

signaling

offrequency-domainresources,206–209

oftime-domainresources,209–211

oftransport-blocksizes,211–212

slot,216–217

formatindication,205

spatialmultiplexing,46

SRScontrolcommands,206

symbols,126

time–frequencygrid,174

transmissions,243,308–309

direction,230–231

suitabletransmitter/receiverbeampairfor,243–244

transmitter-sidebeamadjustment,245–246,246f

uplinkpowercontrolcommands,206

uplinkschedulinggrants,202–205

Downlinkassignmentindex(DAI),202,204,264–265

Downlinkchannelsounding,134–144SeealsoUplinkchannelsounding

basicCSI-RSstructure,134–137,135f

CSI-IM,140–141

CSI-RSresourcesets,142

frequency-domainstructureofCSI-RSconfigurations,137–139

mappingtophysicalantennas,143–144

time-domainpropertyofCSI-RSconfigurations,139–140

TRS,142–143,143f

zero-powerCSI-RS,141–142

Downlinkcontrolinformation(DCI),96,186,200,255–256

format0–0,202–203,202–205,203t

format0–1,202–205,203t

format2–0,205

format2–1,205

format2–2,206

format2–3,206

formats1–0and1–1,199–202,201t

schedulingassignmentin,259

Downlinkmultiantennaprecoding,232–237SeealsoNRuplinkmultiantennaprecoding

typeICSI,234–236

typeIICSI,236–237

DownlinkSharedChannel(DL-SCH),88,155

Downlink/uplink(DL/UL)

carrierpair,117

referenceconfigurations,344

DR,SeeDynamicrange(DR)

DRX,SeeDiscontinuousreception(DRX)

DTCH,SeeDedicatedTrafficChannel(DTCH)

Dualconnectivity,50,51f,78,78,78,84,90,91

withsplitbearer,84f

Dual-bandbasestations,383

Duplexfilters,123

Duplexflexibility,418–419

fullduplexonlinklevelvs.celllevel,419f

Duplexschemes,64–65,64f,121–128,122f

FDD,123–124

slotformatandslot-formatindication,124–128,125f

TDD,121–123

variationof,351

Duplicationfunctionality,275

Dynamicactivation/deactivation,173f

incaseofmultipleconfiguredresourcesets,173f

ofrate-matchingresourceset,172–173

Dynamicdownlinkscheduling,277–283SeealsoDynamicuplinkscheduling

bandwidthadaptation,280–282,281f

downlinkpreemptionhandling,282–283

Dynamicfrequencyselection(DFS),415–416

DynamicPointSelection,47–48

Dynamicrange(DR),374,405,408

BSoutputpowerand,365

deviceoutputpowerand,365–366

referencesensitivityand,378

requirements,362

Dynamicscheduling,67,91,92,277,282,297

DynamicTDD,50–51,64–65,121–122,125,296–297,418

Dynamicuplinkscheduling,283–296SeealsoDynamicdownlinkscheduling

bufferstatusreports,292–294

downlinkpreemptionindication,284f

powerheadroomreports,294–296,296f

schedulingrequest,290–292,293f

uplinkpriorityhandling,288–290

E

Effectiveisotropicradiatedpower(EIRP),377

Efficientmobilityhandling,99

Eight-portCSI-RS,137,138f

eIMTA,SeeEnhancedInterferenceMitigationandTrafficAdaptation(eIMTA)

EIRP,SeeEffectiveisotropicradiatedpower(EIRP)

EIS,SeeEquivalentisotropicsensitivity(EIS)

Electricalbreakdownvoltage(Ebr),409

Electromagneticfields(EMFs),36

eMBB,SeeEnhancedMobileBroadband(eMBB)

EMFs,SeeElectromagneticfields(EMFs)

Emission

maskinOOBdomain,368–370

BSOBUElimits,368–370

deviceSEM,370

unwantedemission

limits,362

requirements,361

EnhancedInterferenceMitigationandTrafficAdaptation(eIMTA),51

EnhancedMobileBroadband(eMBB),4,11–12,14,57

EPC,SeeEvolvedPacketCore(EPC)

Equivalentisotropicsensitivity(EIS),378–379

Errorvectormagnitude(EVM),354,366,366

EuropeanTelecommunicationsStandardsInstitute(ETSI),3

Evaluationconfigurations,21

Evaluationguideline,13

EVM,SeeErrorvectormagnitude(EVM)

EvolvedPacketCore(EPC),39,57,73

Explicitmapping,79

Extendedmultiantennatransmission,46–47

ExtendedZadoff–Chusequence,150

F

Factoryautomation,417

Fallbackformat,SeeDownlinkcontrolinformation(DCI)—format0–0

FasthybridARQwithsoftcombining,41

FCC,SeeFederalCommunicationsCommission(FCC)

FDD,SeeFrequency-divisionduplex(FDD),Full-duplex-capabledevice(FDD)

FDD–TDDaggregation,45

FDM,SeeFrequencydomainsharing(FDM)

FE,SeeFrontEnd(FE)

FEC,SeeForwardErrorCorrection(FEC)

FederalCommunicationsCommission(FCC),36

Fifth-generation(5G),3

firstrelease

D2Dcommunication,417,418f

integratedaccess-backhaul,413–414

machine-typecommunication,416–417

nonorthogonalaccess,416

operationinunlicensedspectra,415–416

spectrumandduplexflexibility,418–419

3GPPandstandardizationofmobilecommunication,2–3

5GAmericas,8

5G/NR,3–6,5–6,395

5Gusecases,4,4f

5GCN,6

evolutionofLTEandNR,6f

evolvingLTEto5Gcapability,5

radio-accesstechnology,5–6

standardization,7

3GPPstandardization,22–26

5GandIMT-2020,14–21

ITU-Ractivitiesfrom3Gto5G,9–14

andregulation,7–8

Figure-of-Merit(FoM),390–391

Filtering,367–368,398–404

filterimplementationexamples,402–404

LTCCfilterimplementationexample,404

PCBintegratedimplementationexample,402–404

ILandbandwidth,401–402

possibilitiesoffilteringatanalogfront-end,399–401

filterexamplefor28GHzband,400f

possiblefilterlocations,400f

Firstgeneration

ofmobilecommunication,1

NMTtechnology,3

1stPDSCHDMRSposition,323

5Gcorenetwork(5GCN),6,73,74–76

FlexibleOFDM-basedphysicallayer,360–361

“Flexible”symbols,126

FoM,SeeFigure-of-Merit(FoM)

Forwardcompatibility,60–61

ForwardErrorCorrection(FEC),253

Four-steprandom-accessprocedure,324–325,325f

Fourth-generation(4G),2SeealsoLong-TermEvolution(LTE)

mobilecommunication,389

FPLMTS,SeeFuturePublicLandMobileSystems(FPLMTS)

Fractionalpath-losscompensation,303,305

Fragmentedspectra,44

Frames,106–107,107f

structure,61–64

Free-runningoscillators,PNcharacteristicsof,392–393

Frequency

error,366,366

hopping,221

multiplexbeamformedtransmissions,230–231

offset,366

Frequencybands,27

frequency-band-dependent,123–124

forNR,32–36

release-independentfrequency-bandprinciples,351–352

Frequencydomainsharing(FDM),135

Frequencyranges(FRs),32–33,352,352–353,352t,353f,367,369f,370f

FR1,33,62

radiatedbase-stationrequirementsin,378–379

FR2,33,62,389

radiatedbase-stationrequirementsin,379–380

radiateddevicerequirementsin,377–378

forNRBStypesin,359–360

RFrequirementsin,352–353

Frequency-divisionduplex(FDD),1–2,27–28,39,64,121,123–124,260–261,418

Frequency-domain,166,193

CDM,137

locationofNRcarriers,114–115

positionofSSblock,315–316

resource

allocation,204

resource-blockallocationtypes,208f

signaling,206–209

structure,109–112

ofCSI-RSconfigurations,137–139

Frequency-hoppingflag(0or1bit),204

Friis’formula,406

FrontEnd(FE),405

Front-loadedreferencesignals,65–66,175–176

FRs,SeeFrequencyranges(FRs)

Fullcoherence,238

Fullduplex,419

onlinklevelvs.celllevel,419f

Full-dimensionMIMO,46

Full-duplexoperation,123–124,124–125

Full-duplex-capabledevice(FDD),126

FundamentalbandwidthofNRcarrier,354

FuturePublicLandMobileSystems(FPLMTS),10

G

5G,SeeFifth-generation(5G)

5GCN,See5Gcorenetwork(5GCN)

Gain,compressionpointand,407–408

Galliumarsenide(GaAs),397

Galliumnitride(GaN),397

FETstructures,394–395

technology,397

GlobalmobileSuppliersAssociation(GSA),31

Globalspectrumsituationfor5G,31–32

GlobalSystemforMobilecommunication(GSM),1–2,383

gNB,76,76–77,263–264,283

distributedunits(gNB-DU),77

entralunit(gNB-CU),77

gNB-DU,SeegNBdistributedunits(gNB-DU)

Goldsequence,176

3GPP,SeeThird-GenerationPartnershipProject(3GPP)

Groupindex,182–183

GSA,SeeGlobalmobileSuppliersAssociation(GSA)

GSM,SeeGlobalSystemforMobilecommunication(GSM)

GSMAssociation(GSMA),8

Guardperiod,SeeGuardtime

Guardtime,122,122–123,123f,326,326,326f

H

Half-duplex

FDD,121

operation,123–124

Half-framebit,321,323

Harmonizedstandards,8

HARQ,SeeHybridAutomaticRepeatRequest(HARQ)

HBTs,394–395

Headercompression,273–275

Heterogeneousdeployments,48–52,49,50f

HighElectronMobilityTransistor(HEMT),394–395

HighSpeedPacketAccess(HSPA),1–2,277

HigherSNRtransmissionscheme,374

Higher-frequency

bands,32,318,321,415–416

operation,59

Higher-layerprotocols,66

HSPA,SeeHighSpeedPacketAccess(HSPA)

HybridAutomaticRepeatRequest(HARQ),67,253,336

acknowledgments,212,216f,262–265,308–309

hybrid-ARQ-relatedinformation,202,204

mechanism,257,260,297

protocol,254

retransmission,257,300

withsoftcombining,93–95,254–265

downlink,259–260

dynamichybrid-ARQacknowledgmentcodebook,265f

multiplexingofhybrid-ARQacknowledgments,262–265

semistatichybrid-ARQacknowledgmentcodebook,263f

softcombining,257–259

timingofuplinkacknowledgments,260–262,261f

uplink,260

Hybrid-ARQ,SeeHybridAutomaticRepeatRequest(HARQ)

“Hybrid”set,359,378–379

Hypotheticalerrorrate,250

I

ICIC,SeeInter-CellInterferenceCoordination(ICIC)

ICNIRP,SeeInternationalCommissiononNon-IonizingRadiation(ICNIRP)

ICS,SeeIn-channelselectivity(ICS)

Identityoflogicalchannel(LCID),89

Idle-statemobility,99–102

pagingmessagetransmission,101–102

trackingdevice,100–101

III–Vmaterials,397

IL,SeeInsertionloss(IL)

IMD,SeeIntermodulationdistortion(IMD)

IMTsystem,SeeInternationalMobileTelecommunicationssystem(IMTsystem)

In-channelselectivity(ICS),364,375

Inactive-statemobility,99–102

pagingmessagetransmission,101–102

trackingdevice,100–101

Inbandrelaying,414

Incrementalredundancy(IR),257,258f

Independentchannelbandwidthdefinitions,350

IndoorHotspot-eMBB,21

Industryforums,8

Initialaccess,70–71,313

associationbetweenSS-blocktimeindicesandRACHoccasionsassuming,333f

beamestablishmentduring,332–333

cellsearch,313–324

randomaccess,324–337

Initialbeamestablishment,244–245

Inputcompressionpoint(CPi),407

Insertionloss(IL),401–402,405,407

Integratedaccess-backhaul,413–414

wirelessbackhaulvs.accesslink,414f

Integratedcircuittechnology,391,395–397,397

Intelligenttransportationsystems(ITSs),54

Inter-CellInterferenceCoordination(ICIC),47

Interbandaggregation,115

Interference

avoidancebyspatialseparation,68

interference-mitigationtechniques,55

suppression/cancellation,419

Interferingsignals

leakage,371–372

receiversusceptibilityto,362,374–376

Interleavedcase,191

Interleavedmapping,168

InterleavedVRB-to-PRBmapping,170

Intermodulationdistortion(IMD),342–343

InternationalCommissiononNon-IonizingRadiation(ICNIRP),36

InternationalMobileTelecommunicationssystem(IMTsystem),9–10

IMT-2000,10–11,11f

coreband,28

IMT-2020,14–21

capabilities,16–19

minimumtechnicalperformancerequirementsfor,20t

performancerequirementsandevaluation,19–21

processinITU-RWP5D,11–14,13f

usagescenariosfor,14–16

usecasesandmappingtousagescenarios,15f

IMT-Advanced,10–11,11f,12f

spectrumdefinedfor,28–31

technologies,351

InternationalRFEMFexposurelimits,36

InternationalTechnologyRoadmapforSemiconductors(ITRS),408–409

InternationalTelecommunicationsUnion(ITU),8SeealsoITURadioRegulations(ITU-R)

Interworking,71–72

Intra-frequency-reselectionflag,322

Intraband

aggregation,115,115

noncontiguouscarrieraggregation,386

IP3,SeeThird-orderinterceptpoint(IP3)

IR,SeeIncrementalredundancy(IR)

ITRS,SeeInternationalTechnologyRoadmapforSemiconductors(ITRS)

ITSs,SeeIntelligenttransportationsystems(ITSs)

ITU,SeeInternationalTelecommunicationsUnion(ITU)

ITURadioRegulations(ITU-R),16,28,30,367

activitiesfrom3Gto5G,9–14

IMT-2000,10–11

IMT-2020processinITU-RWP5D,11–14

IMT-ADVANCED,10–11

roleofITU-R,9–10

relationbetweenkeycapabilitiesandthreeusagescenarios,17f

spectrumdefinedforIMTsystemsby,28–31

J

Johnsonlimit,395–397,409

JointTransmission,47–48

K

Keycapabilities,19

ofIMT-2020,16,16f

relationbetweenkeycapabilitiesandusagescenariosofITU-R,17f

Keyperformanceindicator(KPI),17

Knee-voltage,397

L

L1-RSRP,145,246,250,250–251

L1/L2control

channels,334

signaling,185

LAA,SeeLicense-assistedaccess(LAA)

Latency,18

latency-wiseLTE,41

reduction,54

Layermapping,163

LBT,SeeListen-before-talk(LBT)

LCID,SeeIdentityoflogicalchannel(LCID)

LDPC,SeeLow-densityparity-check(LDPC)

Leesonformula,392–393,392f

License-assistedaccess(LAA),43,45–46,46f,415,416,416f

Licensedspectra,415

Licensedspectrum,45–46

Limited-bufferratematching,161,162f

Linearmultiantennatransmission,229

Listen-before-talk(LBT),415–416

procedure,63

LNA,SeeLow-noiseAmplifier(LNA)

LO,SeeLocalOscillator(LO)

LocalareaBS,364

LocalOscillator(LO),391

generation,391–395

Logicalchannel(s),82,86–91

groups,292–294

multiple,288

multiplexing,285

Logicalnode,76–77

Longpreambles,328–332

numberofRACHtime-domainoccasions,331t

preambleformatsfor,330t

shortpreambles,331t

LongPUCCHformats,214–215

Long-TermEvolution(LTE),39,73,109,227,260–261,279,315–316,317,324,324–325,354,416–417SeealsoLTE/NR,NewRadio(NR)

bands,353

coexistence,71–72

CRS,134,346

densification,48–52

design,60

deviceenhancements,52

dualconnectivity,50,51f

dynamicTDD,50–51

andevolution,40f,42–43,42f

heterogeneousdeployments,48–52,49

LTE-basedtechnologies,57

multiantennaenhancements,46–48

newscenarios,52–55

aerials,55

device-to-devicecommunication,52–53,53f

latencyreduction,54

MTC,53–54

V2V,54–55,55f

V2X,54–55,55f

PBCH,346

PSSandSSS,346,347

re-farmingbands,33

release8,39–41,42

release-8/9devices,49

release9,42

release10,42,44–45

release11,43,45

release12,43,45

release13,43,45

release14,43

release15,43

smallcells,48–52

spectrumflexibility,43–46

technology,2

WLANinterworking,51–52

LongerSS-blockperiodicity,317

Low-densityparity-check(LDPC),66

coderinNR,157

codes,158,159,159f

Low-frequencybands,31

Low-latencysupport,65–66

Low-noiseAmplifier(LNA),405

Low-SNRtransmissionscheme,374

Low-TemperatureCofiredCeramics(LTCC),404

filterimplementationexample,404

Lower-frequencybands,71,321,344

LTCC,SeeLow-TemperatureCofiredCeramics(LTCC)

LTE,SeeLong-TermEvolution(LTE)

“LTECORESET”,189

LTE-Advanced,24

LTE-AdvancedPro,24,43

LTE/NRSeealsoLong-TermEvolution(LTE)

coexistence,344–348,345f,350

configurationofreservedresource,347f

downlink/uplinkcoexistencevs.uplink-onlycoexistence,346f

dual-connectivity,340–344,340f

architectureoptions,342,343f

deploymentscenarios,340–341

inmultilayerscenario,341f

single-TXoperation,342–344

interworking,339–340

migrationofLTEspectrumtoNR,345f

spectrumcoexistence,71

M

M-sequence,320,320–321,320f

MAC,SeeMedium-AccessControl(MAC)

MACcontrolelements(MACCE),89,89–90,117,139–140,292

forbufferstatusreportingandpowerheadroomreports,294f

Machine-typecommunication(MTC),53–54,416–417

Macrocell,364

Mappingtophysicalantennas

CSI-RS,143–144

SRS,152–153

MassiveMachine-TypeCommunication(mMTC),4,11–12,15,57,416–417

MassiveMIMO,68

MasterCellGroup(MCG),84,310

MasterInformationBlock(MIB),87,189,321,323

Masternode,340

Maximumpowerreduction(MPR),365

MCG,SeeMasterCellGroup(MCG)

MediumrangeBS,364

Medium-AccessControl(MAC),66,82,86–95,268f

hybridARQwithsoftcombining,93–95

layer,155

logicalchannelsandtransportchannels,86–91

multiplexingfunctionality,288

protocollayers,253–254

scheduling,91–93

Medium-frequencybands,31

MIB,SeeMasterInformationBlock(MIB)

Microcell,364

Millimeter-waveLos,394

MIMO,39–40

distributed,69

full-dimension,46

massiveMIMOimplementation,29

“Mini-slot”transmission,62–63,63,65–66,107–108

Minimumprocessingtime

inOFDMsymbolsfromgrantreceptiontodatatransmission,287t

PDSCHmappingtypeA,feedbackonPUCCH,262t

mm-wavedomain,operationin,63

mm-wavefrequencies,389,397

RFtechnologiesat

ADCandDACconsiderations,390–391

filtering,398–404

LOgenerationandphasenoiseaspects,391–395

PAefficiencyinrelationtounwantedemission,395–398

receivernoisefigure,DR,andbandwidthdependencies,405–411

mm-wavesignalgeneration,challengeswith,393–395

mm-wavetechnology,377,378

mMTC,SeeMassiveMachine-TypeCommunication(mMTC)

MobilecommunicationSeealsoInternationalTelecommunicationsUnion(ITU)

3GPPandstandardization,2–3

generations,2f

first,1

second,1–2

third,1–2

system,227,228

Mobileservices,30

Mobilesystems

operators,352

spectrumfor,27–32

Mobility,18–19

Modernhigh-speedCMOSdevices,409

Modulation,163

symbol,162

MonolithicVCOimplementation,394

MonteCarloanalysis,403

Moore’slaw,395–397,409

MPR,SeeMaximumpowerreduction(MPR)

MSR,SeeMultistandardradio(MSR)

MTC,SeeMachine-typecommunication(MTC)

MU-MIMO,SeeMultiuserMIMO(MU-MIMO)

Multi-RAT-capableMB-MSRbasestation,383

Multi-SRStransmission,239–240

Multiantenna

multiantenna-relatedinformation,202,204–205

processing,229,229

schemes,41

Multiantennaenhancements,46–48

controlchannelstructureenhancement,48

transmission

extendedmultiantenna,46–47

multipointcoordinationand,47–48

Multiantennaprecoding,128,164–167,167,231,243SeealsoDFT-precoding

downlinkprecoding,165–166

uplinkprecoding,167

Multiantennatransmission,68–69,227

analogmultiantennaprocessingprovidingbeamforming,230f

analogvs.digitalmultiantennaprocessing,230f

DMRSprecoded,232f

downlinkmultiantennaprecoding,232–237

generalmodelofmultiantennatransmissionmapping,230f

multiantennatransmission/reception,227

NRuplinkmultiantennaprecoding,237–242

simultaneous(frequency-multiplexed)beamforming,232f

time-domain(nonsimultaneous)beamforming,231f

Multiband-capablebasestations,382–385

Multilayertransmission,163

Multinationalbasis,3

MultipanelCSI,236,237f

Multipleantennas,227

Multipleclosed-loopprocesses,308

Multiplecompressionalgorithms,273–275

Multiplehybrid-ARQprocesses,255,255f

Multipleopen-loop-parametersets,307–308

Multipleorthogonalreferencesignals,176

Multipleparallelhybrid-ARQprocesses,94,94f

Multiplepath-loss-estimationprocesses,306–307

MultipleperiodicNZP-CSI-RS,142

MultipleRATs,380

Multipleuplinkcarriers,powercontrolincaseof,309–310

Multiplexingcapacity,179

Multiplexingofhybrid-ARQacknowledgments,262–265

Multipoint

coordination,47–48

reception,48

transmission,47–48

Multiport

CSI-RS,135

SRS,150–151,150f

Multistandardradio(MSR),380

basestation,380–382

Multiuserdiversity,277

MultiuserMIMO(MU-MIMO),233–234

NNAICS,SeeNetwork-assistedinterferencecancellation(NAICS)

Nameslotformat,125

NarrowbandInternet-of-Things(NB-IoT),54,416–417

Narrowbandblocking,375

Narrowbandintermodulation,375

NAS,SeeNon-AccessStratum(NAS)

NAX1precodervector,41

NB-IoT,SeeNarrowbandInternet-of-Things(NB-IoT)

NEF,SeeNetworkExposureFunction(NEF)

Neighboringsubcarriers,179

Network,197,326

energyefficiency,18

network-sidebeam-sweeping,71

response,251–252

slicing,74

transmissionofRAR,324–325

NetworkExposureFunction(NEF),75

Networksignaling,362–363

band-specificdevicerequirementsthrough,363–364

Network-assistedinterferencecancellation(NAICS),52

Newbands,27–28

NewRadio(NR),5–6,57,58,73,104,104,253–254,255–256,277,296,313,324,324–325,328–330,349,349–351,350,351,413,414SeealsoLong-TermEvolution(LTE)

antennaports,129t

bands,352

beamforming,68f

BStypesindifferentFRs,359–360

carrier,341

frequency-domainlocation,114–115

fundamentalbandwidth,354

raster,115f

conductedRFrequirements,360–366

band-specificdevicerequirementsthroughnetworksignaling,363–364

BSclasses,364–365

conductedoutputpowerlevelrequirements,365–366

conductedreceivercharacteristics,362

conductedtransmittercharacteristics,361

regionalrequirements,362–363

controlchannels,67–68

CSI-RSin,134–135

developmentsofRFrequirements,380–387

device,144–145,350

downlink

physicalchannels,232–233

transmissions,314–315

anduplinkscheduling,286f

duplexschemes,64–65,64f

forwardcompatibility,60–61

frequencybandsfor,32–36

3GPPtimeline,58f

higher-frequencyoperationandspectrumflexibility,59

hybrid-ARQprotocol,186

initialaccess,70–71,332

interworkingandLTEcoexistence,71–72

low-latencysupport,65–66

NRBSconductedRFrequirements,357–359

radiatedRFrequirements,357–359

NR–LTEcoexistence,72f

radiatedRFrequirementsfor,377–380

release15,413

resourceblock,109

specifications,172,199

spectraidentifiedforNRandcorrespondingsubcarrierspacings,62f

subcarrierspacingssupportedby,105t

time-domainstructure,62–63

transmission

beam-centricdesignandmultiantenna,68–69

schedulinganddata,66–67

scheme,bandwidthparts,andframestructure,61–64

timingofNRuplinktransmissions,326

ultraleandesign,59–60

uplinkpowercontrol,303,303,304

New-candidate-beamidentification,250–251

New-dataindicator,259,259

NextGenerationMobileNetworks(NGMN),8

NGcontrol-planepart(NG-c),77

NGinterface,77

NGuser-planepart(NG-u),77

NG-c,SeeNGcontrol-planepart(NG-c)

ng-eNB,76,76–77

NG-RAN,76

NG-u,SeeNGuser-planepart(NG-u)

NGMN,SeeNextGenerationMobileNetworks(NGMN)

NMT,SeeNordicMobileTelephony(NMT)

Nocoherence,238

Noise

factorandnoisefloor,406–407

figure,374

NOMA,SeeNonorthogonalmultipleaccess(NOMA)

Non-AccessStratum(NAS),74–75

control-planefunctionality,97

RegistrationUpdate,101

Non-DFT-precodedOFDM,61

Non-LTEtechnologies,413

Noncodebook-basedprecoding,167,167,241–242,242f

Noncodebook-basedtransmission,237

Noncontiguousspectra,operationin,386–387,386f

Noninterleavedmapping,191

Nonorthogonalaccess,416

Nonorthogonalmultipleaccess(NOMA),416

Nonstandalone(NSA),357

mode,6

operation,75

Nonzero-powerCSI-RS(NZP-CSI-RS),141,141–142

multipleperiodic,142

NordicMobileTelephony(NMT),1

Normalizedtargetreceivedpower,305

NR,SeeNewRadio(NR)

NRRepositoryFunction(NRF),75

NRuplinkmultiantennaprecoding,237–242SeealsoDownlinkmultiantennaprecoding

codebook-basedtransmission,238–240

noncodebook-basedprecoding,241–242

NR-basedAccesstoUnlicensedSpectrum,415

NRF,SeeNRRepositoryFunction(NRF)

NSA,SeeNonstandalone(NSA)

Numerologies,315

multipleandmixed,350

numerology-independenttimereference,107

240kHznumerology,315

Nyquistsamplingfrequency,390–391

NZP-CSI-RS,SeeNonzero-powerCSI-RS(NZP-CSI-RS)

NZP-CSI-RS-ResourceSets,142,145

OOBUEs,SeeOperatingbandunwantedemissions(OBUEs)

Occupiedbandwidth,373

OFDM,SeeOrthogonalfrequency-divisionmultiplexing(OFDM)

OOB,SeeOut-of-band(OOB)

OOBblocking,SeeOutsideoperatingband(OOBblocking)

Open-loopparameters,307–308

pairs,308

Open-looppowercontrol,303

Operatingbandunwantedemissions(OBUEs),368

BSOBUElimits,368–370

Operatingbands,33,34t,34t,35f,35f,36f

Operationallifetime,19

Operators

coexistencebetweenoperators

ingeographicalareainband,351

ofTDDsystems,351

colocationofBSequipmentbetween,351

ofmobilesystems,352

Orthogonalfrequency-divisionmultiplexing(OFDM),39–40,61,103–104,314–315,349,408

modulatoroutput,328

OFDM-baseddownlinkanduplink,175–181

OFDM-basedphysicallayer,flexible,360–361

OFDM-basedtransmission,2

spectrumofOFDMsignal,367–368

symbols,126,283,314–315,318–319

Orthogonalsequences,176

Orthogonality,328

OSDDs,SeeOTAsensitivitydirectiondeclarations(OSDDs)

OTA,SeeOver-the-air(OTA)

OTAsensitivitydirectiondeclarations(OSDDs),378–379

Out-of-band(OOB),32

domain,367

emissionmaskin,368–370

emissions,366,367

Outputpoweranddynamicrange

BS,365

device,365–366

Outputpowerlevelrequirements,361

conducted,365–366

Outsideoperatingband(OOBblocking),375

Over-the-air(OTA),349,378

sensitivity,378–379

testing,357–358

PPA,SeePoweramplifier(PA)

PacketDataConvergenceProtocol(PDCP),81,83–85,273–276

header,82–83

layer,71

protocol,82–83,254

layers,253–254

retransmissionfunctionality,275

PAE,SeePower-addedefficiency(PAE)

PagingChannel(PCH),87–88,155

PagingControlChannel(PCCH),87,266

Pagingmessagetransmission,101–102

Pairedbands,27–28

Pairwisecoherence,238

Parallelingtechnique,397

Partialcoherence,238

Path-lossestimate(PLestimate),304,305,306–307

PayloadtransmittedonPDCCH,186

PBCH,SeePhysicalBroadcastChannel(PBCH)

PBR,SeePrioritizedbitrate(PBR)

PCB,SeePrintedcircuitboard(PCB)

PCCH,SeePagingControlChannel(PCCH)

PCell,SeePrimarycell(PCell)

PCF,SeePolicyControlFunction(PCF)

PCH,SeePagingChannel(PCH)

PCI,SeePhysicalcellidentity(PCI)

PDC,SeePersonalDigitalCellular(PDC)

PDCCH,SeePhysicalDownlinkControlChannel(PDCCH)

PDCP,SeePacketDataConvergenceProtocol(PDCP)

PDSCH,SeePhysicalDownlinkSharedChannel(PDSCH)

PDU,SeeProtocolDataUnit(PDU)

Peakdatarate,17

Peakspectralefficiency,17

Per-CBCRC,158

Per-CBGretransmission,259–260,260f

Per-slotscheduling,91

Performancecharacteristics,361

PeriodicCSI-RStransmission,139

Periodicreporting,146,147

PeriodicSRS,151,242

PersonalDigitalCellular(PDC),1–2

PhaseLockedLoop(PLL),392–393

Phasenoise(PN),391–395

challengeswithmm-wavesignalgeneration,393–395

characteristicsoffree-runningoscillatorsandPLLs,392–393

Phase-trackingreferencesignals(PT-RS),174,183–184,184f

pHEMTdevices,394–395

PHY,SeePhysicalLayer(PHY)

PhysicalBroadcastChannel(PBCH),70,96,313–314,315,315,319,319–323,321–323

informationcarriedwithin,322t

PBCH/MIB,324

Physicalcellidentity(PCI),321

Physicalchannel,96

Physicaldatasharedchannels,SeePhysicalDownlinkSharedChannel(PDSCH)

PhysicalDownlinkControlChannel(PDCCH),41,66,67,96,185,186–189,186f,187f,196f,250,297

transmission,248,249

PhysicalDownlinkSharedChannel(PDSCH),69,96,141

downlink,163

PDSCH/PUSCHallocation,183

transmission,248,249

PhysicalLayer(PHY),82,95–96,155

PhysicalRandom-AccessChannel(PRACH),96,324,325

Physicalresourceblocks,110–111,111–112,111f,168

Physicalresource-blockgroups(PRGs),166,166f,235

PhysicalUplinkControlChannel(PUCCH),41,67–68,96,146,213,214f

format0,215–217,216f

format1,217–219,218f

format2,219–220,220f

format3,220–222,221f

format4,222,222f

groups,116–117

powercontrolfor,308–309

PUCCH-relatedinformation,202

reporting,295

resource

indicator,262

andparametersfortransmission,223

sets,223,224f

structure,214–215

PhysicalUplinkSharedChannel(PUSCH),96,146

reporting,295

transmission,120,303,306–307

power-controlfor,304

uplink,163

controlsignalingon,223–225

Physical-layercontrol

channels,68

signaling

downlink,185–212

uplink,212–225

Physical-layerhybrid-ARQfunctionality,155–156,160–162

bitinterleaver,162f

circularbufferforincrementalredundancy,161f

Picocell,364

PLestimate,SeePath-lossestimate(PLestimate)

Planardevices,394–395

Platooning,54

PLL,SeePhaseLockedLoop(PLL)

PMI,SeePrecodermatrixindicator(PMI)

PN,SeePhasenoise(PN)

PointA(referencepoint),110–111

Polarcode,188

PolicyControlFunction(PCF),75

Power

availability,294–295

back-off,368

consumption,300

headroomreports,294–296,296f

ramping,333

Poweramplifier(PA),368,395–397

efficiencyinrelationtounwantedemission,395–398

outputpowervs.frequency,396f

saturatedpower-addedefficiencyvs.frequency,398f

Powercontrol,295–296,303

power-controlcommands,303,306

power-control-relatedinformation,205

forPUCCH,308–309

forPUSCHtransmissions,304

Power-addedefficiency(PAE),398

Power-spectraldensity(PSD),408

PRACH,SeePhysicalRandom-AccessChannel(PRACH)

Preamble,328

powercontrol,333

sequence,328,328,328

structure,328

generationofNRrandom-accesspreamble,329f

Preambleformat,330

forlongpreambles,330t

forshortpreambles,331t

Preambletransmission,325–333

basicpreamblestructure,328

beamestablishmentduringinitialaccess,332–333

characteristics,326–327

guard-timeneedsfor,326f

longvs.shortpreambles,328–332

preamblepowercontrolandpowerramping,333

RACHresources,327

Precodercodebook,233

Precodermatrix,41,231

Precodermatrixindicator(PMI),145,233

Precoder-baseduplinktransmissions,181

Precodinginformation,167

Preemption,67

indication,205

indicator,283

PRGs,SeePhysicalresource-blockgroups(PRGs)

Primarycell(PCell),116

Primarysecondcell(PSCell),116–117,213

PrimarySynchronizationSequence,SeePrimarySynchronizationSignal(PSS)

PrimarySynchronizationSignal(PSS),70,313–314,314,315–316,319–323,320–321,320f

PSS/SSS,313–314

sequences,320

ofSSblock,319

Prime-lengthZCsequences,328

Printedcircuitboard(PCB),402

Prioritizedbitrate(PBR),290

ProtocolDataUnit(PDU),82

sessions,79,79f

PSCell,SeePrimarysecondcell(PSCell)

PSD,SeePower-spectraldensity(PSD)

Pseudo-randomsequence,176–179,193

PSS,SeePrimarySynchronizationSignal(PSS)

PT-RS,SeePhase-trackingreferencesignals(PT-RS)

PUCCH,SeePhysicalUplinkControlChannel(PUCCH)

PUSCH,SeePhysicalUplinkSharedChannel(PUSCH)

QQCL,147–148,249

QFI,SeeQuality-of-serviceflowidentifier(QFI)

QPSK,365,374

Quality-of-service(QoS),79

flows,79,79f

handling,79

Quality-of-serviceflowidentifier(QFI),79,83

Quasi-colocation,130–131

Quasi-cyclicLDPCcodes,159

R

RA-RNTI,199,334

RACH,SeeRandom-AccessChannel(RACH)

Radiatedbase-station

requirementsinFR1,378–379

requirementsinFR2,379–380

RadiateddevicerequirementsinFR2,377–378

Radiatedinterfaceboundary(RIB),359

RadiatedRFrequirementsforNR,377–380

BS,357–359

radiatedbase-stationrequirements

inFR1,378–379

inFR2,379–380

radiateddevicerequirementsinFR2,377–378

Radiatedtransmitpower,378

Radiatedunwantedemissionsrequirements,378,379–380

Radio

access,39–41

communication,227–228

distributionnetwork,358–359

protocolarchitecture,80

RadioAccessNetwork(RAN),23,73,73,76–78,77f,335

Radiofrequency(RF),8,23,395–397SeealsoReferencesignal(s)

ADCandDACconsiderations,390–391

bandwidth,381–382

channelbandwidthandspectrumutilization,353–356

characteristics,349

conductedRFrequirementsforNR,360–366

band-specificdevicerequirementsthroughnetworksignaling,363–364

BSclasses,364–365

conductedoutputpowerlevelrequirements,365–366

conductedreceivercharacteristics,362

conductedtransmittercharacteristics,361

regionalrequirements,362–363

conductedsensitivityanddynamicrange,374

conductedunwantedemissionsrequirements,367–374

developmentsofRFrequirementsforNR,380–387

MSRbasestation,380–382

multiband-capablebasestations,382–385

operationinnoncontiguousspectra,386–387

exposureabove6GHz,36–37

filtering,398–404

filters,409

LOgenerationandphasenoiseaspects,391–395

PAefficiencyinrelationtounwantedemission,395–398

radiatedRFrequirementsforNR,377–380

receivernoisefigure,DR,andbandwidthdependencies

carrierfrequencyandmm-wavetechnologyaspects,408–411

compressionpointandgain,407–408

noisefactorandnoisefloor,406–407

PSDandDR,408

receiverandnoisefiguremodel,405

receiversusceptibilitytointerferingsignals,374–376

requirements,352–353

indifferentFRs,352–353

spectrumflexibilityimplications,349–352

structure

ofBS,357–360,359–360

ofBSRFrequirements,357–360

conductedandradiatedRFrequirementsforNRBS,357–359

ofdevice,356–357

technologiesatmm-wavefrequencies,389

transmittedsignalquality,366–367

RadioInterfaceSpecifications(RSPCs),10

RadioInterfaceTechnologies(RITs),10

RadioRegulations,9

RadioResourceControl(RRC),97

RRCRANNotificationAreaUpdate,101

RRC-IDLEstate,97–98,98

RRC-signaledpattern,126,126

RRC_ACTIVEstate,97–98

RRC_CONNECTEDstate,98–99

RRC_INACTIVEstate,97–98,99

signaling,298

statemachine,97–99,98f

Radioresourcemanagement(RRM),23,77,145

Radio-accesstechnologies(RAT),342,380

Radio-interfacearchitectureSeealsoNewRadio(NR)

control-planeprotocols,97–102

overallsystemarchitecture,73–78

combinationsofcorenetworksandradio-accesstechnologies,76f

5Gcorenetwork,74–76

high-levelcorenetworkarchitecture,75f

radio-accessnetwork,76–78,77f

QoShandling,79

radioprotocolarchitecture,80

user-planeprotocols,80–96,82f

Radio-LinkControl(RLC),66,82,85–86,85f,266–273,268fSeealsoNewRadio(NR)

acknowledgedmodeandRLCretransmissions,269–273

generationofRLCPDUsfromRLCSDUs,270f

PDUs,83

protocol,83,253–254

retransmissions,269–273

sequencenumberingandsegmentation,267–269

Radio-linkfailure(RLF),249–250

RAN,SeeRadioAccessNetwork(RAN)

RANAreaIdentifier(RAI),100

RANAreas,100,100f,101

RANNotificationArea,101

Randomaccess,313,324–337

channel,155–156

contentionresolution,335

andconnectionsetup,335–336

preamble,70,325

transmission,325–333

procedure,325

random-access-relatedMACcontrolelements,89

response,334–335

forSUL,336–337

Random-AccessChannel(RACH),88

configurationperiod,327

occasions,327

resources,327,327f,330

slots,327,327,330

Random-AccessResponse(RAR),324–325,334–335

Rangeofangleofarrival(RoAoA),378–379

Rankindicator(RI),145,233

RAR,SeeRandom-AccessResponse(RAR)

RAT,SeeRadio-accesstechnologies(RAT)

Ratematching,188

andphysical-layerhybrid-ARQfunctionality,160–162

Re-farming,31

Receiver

characteristics,362,363t

intermodulation,375

multiantennaprocessing,243

noisefigure,405–411

andnoisefiguremodel,405

simplifiedreceivermodel,406f

zero-IFtransceiverschematic,406f

receiver-bandwidthadaptation,62,112–113,280

receiver-sidedirectivity,227

susceptibilitytointerferingsignals,362,374–376,378,380

BSanddevicerequirementsforreceiversusceptibility,376f

Recovery-requesttransmission,250

Redundancyversion(RV),160

Referencesensitivity,374

anddynamicrange,378,380

Referencesignalreceivedpower(RSRP),145,336,336

Referencesignal(s),174–184,176,217–219SeealsoRadiofrequency(RF)

demodulation

forDFT-precodedOFDMuplink,181–183

forOFDM-baseddownlinkanduplink,175–181

occasions,40

PT-RS,183–184

structure,47,179–181

Reflectivemapping,79

Regionalrequirements,362–363

REGs,SeeResource–elementgroups(REGs)

Regulatorybodiesandadministrations,8,9f

Relativepowertolerance,366

Relaynode,49

Relaying,49,49f

Release-independentfrequency-bandprinciples,351–352

Reliability,19

Remainingminimumsysteminformation(RMSI),324

Remainingsysteminformation,324

“Repetition”flag,247

Reportconfigurations,142,144–145,233

Requirementsphase,22

Reservedresources,61,160,171

Resilience,19

Resource

allocation

type0,170

type1,170,207

blocks,91,109

configuration,146

element,109

grids,109–110,110–111,111f

mapping,167–171,169f

Resource–elementgroups(REGs),188

bundle,191

Retransmission,161,259SeealsoTransmission

functionality,253–254,275

protocols

hybrid-ARQwithsoftcombining,254–265

PDCP,273–276

RLCprotocol,266–273,268f

RF,SeeRadiofrequency(RF)

RI,SeeRankindicator(RI)

RIB,SeeRadiatedinterfaceboundary(RIB)

RITs,SeeRadioInterfaceTechnologies(RITs)

RLC,SeeRadio-LinkControl(RLC)

RLF,SeeRadio-linkfailure(RLF)

RMSI,SeeRemainingminimumsysteminformation(RMSI)

RoAoA,SeeRangeofangleofarrival(RoAoA)

Robustheadercompression(ROHC),83,273–275

ROHC,SeeRobustheadercompression(ROHC)

RootindexofZadoff–Chusequence,149

RRC,SeeRadioResourceControl(RRC)

RRM,SeeRadioresourcemanagement(RRM)

RSPCs,SeeRadioInterfaceSpecifications(RSPCs)

RSRP,SeeReferencesignalreceivedpower(RSRP)

Rural-eMBB,21

RV,SeeRedundancyversion(RV)

SSaturationvelocity(Vsat),409

SCells,SeeSecondarycells(SCells)

SCG,SeeSecondaryCellGroup(SCG)

Scheduledcarriers,264

Scheduler,91

Scheduling,66–67,91–93,296–297

assignments,116,278

decisions,41

discontinuousreception,298–302

dynamicdownlink,277–283

anddynamicTDD,296–297

dynamicuplink,283–296

grants,116,285

request,290–292,293f

scheduling-relatedMACcontrolelements,89

transmissionwithoutdynamicgrant,297–298,299f

Scrambling,162–163

SDAP,SeeServiceDataApplicationProtocol(SDAP)

SDLbands,SeeSupplementaryDownlinkbands(SDLbands)

SDOs,SeeStandardsDevelopingOrganizations(SDOs)

SDPA,SeeServiceDataAdaptationProtocol(SDPA)

SDU,SeeServiceDataUnit(SDU)

Searchspaces,195–199,198f

Secondgeneration(2G)

ofmobilecommunication,1–2,389

technologies,1–2

SecondaryCellGroup(SCG),84

Secondarycells(SCells),116

Secondarynode,340

Secondarysynchronizationsignal(SSS),70,313–314,314,315–316,319–323,321,321

sequence,321

ofSSblock,319

Securityandprivacy,19

Segmentation,85–86,85f,267–269

Segmentationinformation(SI),267

Segmentationoffset(SO),267

Self-containedslots,67–68

Self-interference,342–343

Self-scheduling,116,116f,279,280f

SEM,SeeSpectrumemissionsmask(SEM)

Semipersistent

CSI-RStransmission,139–140

reporting,147

scheduling,297

SRS,151,242

Semistaticcodebook,264

Semistaticscheduling,277

Sensitivityanddynamicrangerequirements,362

Sequence

index,182–183

numbering,267–269

ServiceDataAdaptationProtocol(SDPA),82,83

ServiceDataApplicationProtocol(SDAP),81

ServiceDataUnit(SDU),82

Service-basedarchitecture,74

SessionManagementFunction(SMF),74–75

700MHzband,31

SFI,SeeSlot-formatindication/indicator(SFI)

SFN,SeeSystemframenumber(SFN)

Shannonchannelcapacity,305

Sharpfiltering,401

Shiftcoefficients,160

Shortpreambles,328–332,331t

formatsforlongpreambles,330t

RACHtime-domainoccasionswithinRACHslot,331t

ShortPUCCHformats,214

ShortTTI(sTTI),43,54

ShorterSS-blockperiodicity,316

SI,SeeSegmentationinformation(SI)

SI-RNTI,SeeSystemInformationRNTI(SI-RNTI)

SIBs,SeeSystemInformationBlocks(SIBs)

Sidelink

connectivity,417

transmission,57

Signal-to-noise-and-distortionratio(SNDR),390–391

SNDR-basedSchreierFoM,390–391

Signaling

offrequency-domainresources,206–209

tosupportbeam-managementprocedures,69

oftime-domainresources,209–211

oftransport-blocksizes,211–212

Signalingradiobearers(SRBs),97

Simplifiedreceivermodel,405,406f

Simulation,21

Singleradio-accesstechnology,414

Single-antennatransmission,130

Single-panelCSI,235–236,235f

Single-portCSI-RS,139

Single-TXoperation,342–344

Single-userMIMO,41

SiP,SeeSystem-in-package(SiP)

Sixteen-QAMsignalΨ16Ψ,374,391,392f

Slot,107

aggregation,211

format,124–128,125f

Slot-formatindication/indicator(SFI),124–128,125f,126,126–127,127f,205

Smallcells,48–52

on/off,49–50

SMF,SeeSessionManagementFunction(SMF)

SNDR,SeeSignal-to-noise-and-distortionratio(SNDR)

SO,SeeSegmentationoffset(SO)

SoC,SeeSystem-on-chip(SoC)

Softcombining,161,254–265,257–259

hybrid-ARQwith

downlinkhybrid-ARQ,259–260

dynamichybrid-ARQacknowledgmentcodebook,265f

multiplexingofhybrid-ARQacknowledgments,262–265

semistatichybrid-ARQacknowledgmentcodebook,263f

timingofuplinkacknowledgments,260–262,261f

uplinkhybrid-ARQ,260

Soundingreferencesignals(SRS),92,133,147–153,167,174,310SeealsoChannel-state-informationreferencesignals(CSI-RS)

comb-basedfrequencymultiplexing,149f

controlcommands,206

mappingtophysicalantennas,152–153

multiport,150–151,150f

resourceset,151

sequences,149–150

time-domainstructure,151

time/frequencystructures,148f

Zadoff–Chusequences,149–150

Sparsefrequencyraster,70

SparseSS-blockraster,70

Sparsesynchronizationraster,316

Spatialfiltering,143–144

Spatialmultiplexing,103,179–181,227

Spectrum,415,418–419

for5G

frequencybandsforNR,32–36

globalspectrumsituationfor5G,31–32

newIMTbandsunderstudyinITU-RTG5/1,30f

RFexposureabove6GHz,36–37

spectrumdefinedforIMTsystemsbyITU-R,28–31

spectrumformobilesystems,27–32

allocations

aggregation,352

diverse,350

analyzers,352

andbandwidthflexibility,19

blockdefinitions,350

coexistence,339–340

efficiency,18

flexibility,39,43–46,59,121,349,354

CA,44–45,44f

implications,349–352

LAA,45–46,46f

fullduplexonlinklevelvs.celllevel,419f

mask,364

formobilesystems,27–32

ofOFDMsignal,367–368

regulation,8

utilization,353–356,356t

Spectrumemissionsmask(SEM),367

device,370

Spiderweb”diagrams,16,16f

Splitbearers,84

Spuriousdomain,367

Spuriousemissions,367,373

Spuriousresponsefrequencies,375

SRBs,SeeSignalingradiobearers(SRBs)

SRI,SeeSRSresourceindicator(SRI)

SRS,SeeSoundingreferencesignals(SRS)

SRSresourceindicator(SRI),167,205,239–240,241–242

SSblock,SeeSynchronizationSignalblock(SSblock)

SS-blockperiodicity,316

SSS,SeeSecondarysynchronizationsignal(SSS)

Stackingtechnique,397

StandardsDevelopingOrganizations(SDOs),7

Staticfrequency-domainsharing,344,345

Staticsplit,50–51

sTTI,SeeShortTTI(sTTI)

Subcarrierspacing,107

Subframe(s),106–107,107f

durationof1ms,40

Submissiontemplate,13

Suitablebeampair,243,244,245

adjusteddownlink,245

indownlinkdirection,244f

suitabledownlink,247–248

suitabletransmitter/receiver,243–244

SUL,SeeSupplementaryuplink(SUL)

SupplementaryDownlinkbands(SDLbands),27–28,120,351

Supplementaryuplink(SUL),71–72,117–120,118f,119f,119f,351

bands,27–28

controlsignaling,120

randomaccessfor,336–337

relationtocarrieraggregation,119–120

SUL/non-SULindicator,120

Synchronizationraster,115,316

SynchronizationSignalblock(SSblock),70,134,146,244,244–245,246,248,250,313,313–315

burstset,317–319,317f

time-domainlocationsofSSblockwithin,318f

frequency-domainposition,315–316

numerologiesandfrequencyranges,315t

periodicity,316–317

timeindex,322,323,332

time–frequencystructureofsingleSSblock,314f

Synchronoushybrid-ARQprotocol,186

Systemframenumber(SFN),106–107

SystemInformationBlocks(SIBs),324

SIB1,323,324,324

configuration,323

numerology,323

reception,323

SystemInformationRNTI(SI-RNTI),324

System-in-package(SiP),409

System-levelsimulations,21

System-on-chip(SoC),409

TTAB,SeeTransceiverarrayboundary(TAB)

TACS,SeeTotalAccessCommunicationSystem(TACS)

Tactileinternet,14–15

TAGs,SeeTimingadvancedgroups(TAGs)

TAI,SeeTrackingAreaIdentifier(TAI)

Targetreceivedpower,304

TC-RNTI,334,336

TCI,SeeTransmissionConfigurationIndex(TCI)

TD-SCDMA,2

tDAI,SeeTotalDAI(tDAI)

TDD,SeeTimeDivisionDuplex(TDD)

TDM,SeeTime-domainsharing(TDM)

Technicalrequirements,13

TechnicalSpecifications(TS),25

TechnicalSpecificationsGroups(TSGs),23

Technology,419

“Technology-neutral”manner,351–352

Testenvironments,21

Testingandverificationphase,23

TF,SeeTransportFormat(TF)

TG5/1taskgroup,30

Thirdgeneration(3G),1–2

mobilecommunication,389

Thirdgenerationofmobilecommunication,1–2

Third-GenerationPartnershipProject(3GPP),2–3,7,359,377,380–381,414

organization,24f

process,22–25

radio-accesstechnologies,380

specifications,382,389

of5G,25–26

standardization,22–26

phasesanditerativeprocess,22f

timeline,58f

Third-orderinterceptpoint(IP3),407

32-portCSI-RS,137,139f

3Dgaming,14–15

Timedomain,166,171–172,225

allocation,209,210f

forDM-RS,176

bitmap,172

propertyofCSI-RSconfigurations,139–140

resource

allocation,204

signaling,209–211

structure,106–108

ofSRS,151

windowing,367–368

Timeindex,319

Timemultiplexedreferencesignals,181–182

TimeDivisionDuplex(TDD),1–2,27–28,39,64,121,121–123

carrier,344

coexistencebetweenoperatorsofTDDsystems,351

operation,365

scheme,418

TDD-capabledevice,45

Time-domainsharing(TDM),135

Time–frequency

resource,168,189

time/frequency-domainCDM,137

time–frequency-coderesources,223

Timingadvancedgroups(TAGs),312

Timing-advance,310,311,311f

MACcontrolelements,89

TM,SeeTransparentmode(TM)

TotalAccessCommunicationSystem(TACS),1

TotalDAI(tDAI),264–265

Totalradiatedpower(TRP),377

TrackingAreaIdentifier(TAI),100

TrackingAreas,100,100f,101

Trackingdevice,100–101

Trackingreferencesignal(TRS),142–143,143f,174

Traffic

channel,86–87

situation,51

Transceiverarrayboundary(TAB),358–359

Transceiverunitarray,358–359

Transmission,259

bandwidthconfiguration,354

todeviceAandB,283

withoutdynamicgrant,297–298,299f

parameters,298

rank,41

scheme,61–64,103–106

structure

antennaports,128–130,129t

BWPs,112–114,113f

carrieraggregation,115–117

duplexschemes,121–128

frequency-domainlocationofNRcarriers,114–115

frequency-domainstructure,109–112

quasi-colocation,130–131

subcarrierspacingssupportedbyNR,105t

SUL,117–120,118f,119f,119f

symbolalignment,106f

time-domainstructure,106–108

transmissionscheme,103–106

timingofNRuplinktransmissions,326

TransmissionConfigurationIndex(TCI),165,193–194,248–249,248,249

Transmissionconfigurationindication,SeeTransmissionConfigurationIndex(TCI)

TransmissionReceptionPoint(TRP),18

TransmissionTimeInterval(TTI),87,155

TransmitmultiplemultiportSRS,239–240

Transmit-timingadvance,310

Transmittedsignalquality,366–367,378,379

BStimealignment,367

devicein-bandemissions,366

EVMandfrequencyerror,366

requirements,361

Transmitter

characteristics,361,362t

intermodulation,373–374

requirements,361

Transparentmode(TM),85,266

Transportblock(s),87,157

sizessignaling,211–212,212f

transport-block-relatedinformation,201–202,204

Transportchannels,86–91

processing,156f

channelcoding,157–160

downlinkreservedresources,171–173

layermapping,163

modulation,163

multiantennaprecoding,164–167

ratematchingandphysical-layerhybrid-ARQfunctionality,160–162

referencesignals,174–184

resourcemapping,167–171

scrambling,162–163

uplinkDFTprecoding,164

transmission,167–168

types,87–88

TransportFormat(TF),87,304

Transport-formatselection,87

TRP,SeeTotalradiatedpower(TRP),TransmissionReceptionPoint(TRP)

TRS,SeeTrackingreferencesignal(TRS)

TS,SeeTechnicalSpecifications(TS)

TSGRAN,23

TSGs,SeeTechnicalSpecificationsGroups(TSGs)

TTI,SeeTransmissionTimeInterval(TTI)

26GHzband,32

Two-dimensionalbeamforming,46

Two-portCSI-RS,136,136f

Type0,bitmap-basedallocationscheme,206–207

Type1powerheadroomreporting,295

Type2powerheadroomreporting,295

Type3powerheadroomreporting,295

TypeICSI,234–236

multipanelCSI,234,236

single-panelCSI,234,235–236

TypeIICSI,236–237

UUCI,SeeUplinkcontrolinformation(UCI)

UDM,SeeUnifiedDataManagement(UDM)

UE,SeeUserEquipment(UE)

UEpowerclass,365

UERegistrationArea,101,101

UL-SCH,SeeUplinkSharedChannel(UL-SCH)

UL/SULindicator,204

Ultra-Low-LatencyandReliablecommunication(URLLC),4,11–12,14–15,53,416

Ultraleandesign,59–60

Unacknowledgedmode(UM),85,266,267

UnifiedDataManagement(UDM),75

Unlicensedspectra,operationin,415–416

Unpairedbands,27–28

Unwantedemissions

limits,362

requirements,361,367–374

UPF,SeeUserPlaneFunction(UPF)

Uplink,155–156,212–225,418SeealsoDownlink

acknowledgmenttiming,260–262,261f

beamadjustment,247–248

codebook,239,239f

constraints,344

controlsignalingonPUSCH,223–225

DFTprecoding,164

hybrid-ARQ,260

message,335

orthogonality,310

π/2-BPSK,163

precoding,167,182f

priorityhandling,288–290

PUCCHformat0,215–217

format1,217–219

format2,219–220

format3,220–222

format4,222

structure,214–215

reference

andparametersforPUCCHtransmission,223

signals,182–183,183f

scheduler,91,283

scheduling,91

assignments,308–309

grants,202–205

soundingsignals,127

spatialmultiplexing,47

symbols,126

timingcontrol,310–312

uplink-onlycoexistence,346

uplink-path-lossestimate,306

uplink–downlinkallocation,39,65

Uplinkchannelsounding,147–153SeealsoDownlinkchannelsounding

mappingtophysicalantennas,152–153

multiportSRS,150–151,150f

SRSresourceset,151

SRSsequencesandZadoff–Chusequences,149–150

time-domainstructureofSRS,151

Uplinkcontrolinformation(UCI),67–68,96

Uplinkpowercontrol,303–310SeealsoBeam-basedpowercontrol

baselinepowercontrol,304–306

beam-basedpowercontrol,306–308

incaseofmultipleuplinkcarriers,309–310

commands,206

forPUCCH,308–309

UplinkSharedChannel(UL-SCH),88,155

UrbanMacro-mMTC,21

UrbanMacro-URLLC,21

URLLC,SeeUltra-Low-LatencyandReliablecommunication(URLLC)

Usagescenarios,11–12,29

forIMT-2020,14–16

UserEquipment(UE),74–75,357

Userexperienceddatarate,18

UserPlaneFunction(UPF),74

User-planeprotocols,80–96,82fSeealsoControl-planeprotocols

MAC,86–95

PDCP,83–85

physicallayer,95–96

RLC,85–86,85f

SDAP,83

Uuinterface,77

VVandiagram,10,11f

Vehicle-to-everythingcommunication(V2Xcommunication),43,54–55,55f

Vehicle-to-vehiclecommunication(V2Vcommunication),14–15,43,54–55,55f

Virtualresourceblocks,111–112,168,207

“Vision”recommendation,11–12,12

Voltage-ControlledOscillator(VCO),392

W

WARC,SeeWorldAdministrativeRadioConference(WARC)

Wi-Fi,45–46,415

WidebandCDMA(WCDMA),3

Widebandreferencesignals,193

Wireless

communicationsystems,97–98

technologyforbackhaul,413

wireless-backhaulsolutions,413

WLANinterworking,51–52

WorkingParty5D(WP5D),9–10

WorldAdministrativeRadioConference(WARC),28–29

WARC-92,28

WorldRadio-communicationConference(WRC),9,28–29

WRC-15,12,13,29

WRC-19,12

WP5D,SeeWorkingParty5D(WP5D)

WRC,SeeWorldRadio-communicationConference(WRC)

XXninterface,77

Z

Zadoff–Chusequences(ZCsequences),149–150,182–183,328,328

Zero-correlationzoneparameter,328

Zero-powerCSI-RS(ZP-CSI-RS),141–142