utran radio resource managementutran radio resource · pdf fileutran radio resource...

UTRAN Radio Resource ManagementUTRAN Radio Resource Management

IntroductionHandover Control

Soft/Softer HandoverInter Frequency HandoverBTS 3 Inter Frequency HandoverCell Selection/Re-selection

Power Control

BTS 3

Closed Loop Power ControlOpen Loop Power Control

Load ControlBTS 2UE

Load Control

Call Admission ControlCongestion ControlBTS 1

Packet Data Transmission

Packet Data ControlDynamic SchedulingDynamic Scheduling

References

H Holma A Toskala (Ed ) “WCDMA for UMTS” Wiley 4th edition Wiley 2007H. Holma, A. Toskala (Ed.), WCDMA for UMTS , Wiley, 4th edition, Wiley, 2007Walke, Althoff, Seidenberg: UMTS – Ein Kurs. J. Schlembach Fachverlag, 2001

H. Kaaranen, et.al., “UMTS Networks: Architecture, Mobility and Services”, Wiley, 2001. (see chapter 4)

A. Viterbi: “CDMA: Principles of Spread Spectrum Communications,” Addison Wesley, 1995.y,

J. Laiho, A. Wacker, T. Novosad (ed.): “Radio Network Planning and Optimisation for UMTS,“ Wiley, 2001

T Ojanperä R Prasad “Wideband CDMA for Third Generation MobileT. Ojanperä, R. Prasad, Wideband CDMA for Third Generation Mobile Communication”, Artech House, 1998.

R. Prasad, W. Mohr, W. Konhäuser, “Third Generation Mobile Communications S t ” A t h H M h 2000Systems”, Artech House, March 2000.

3GPP standards:3GPP standards:

TS 25.214: “Physical Layer Procedures“ (esp. power control)TR 25.922: “Radio Resource Management Strategies“

UMTS Networks 2Andreas Mitschele-Thiel, Jens Mueckenheim Oct. 2008

TR 25.942: “RF System Scenarios“

UMTS – Context of Radio Resource Management

System evaluation and standardisation activities on three levels:

• Access Network (Architecture)• Mobility Management

UTRANInterfaces Architecture y gArchitecture

• Cellular Network Aspects (Layer 2&3) • Radio Resource Management

Cellular Network Protocols

R di Li k (Ph i l L )• Radio Link (Physical Layer)


RRM – High-Level Requirements

Efficient use of limited radio resources (spectrum, power, code space)Minimizing interference Flexibility regarding services (Quality of Service, user behaviour)Simple algorithms requiring small signalling overhead onlyStability and overload protectionStability and overload protectionSelf adaptive in varying environmentsAllow interoperability in multi-vendor environments

Radio Resource Management algorithms control thealgorithms control the efficient use of resources with respect to interdependent objectives:interdependent objectives:-- cell coverage-- cell capacity-- quality of service


q y

RRM – Components

R di R M t

non-accessstratum

Radio Resource Management

Packet DataC t l

LoadC t l

HandoverC t l

typically in RNC

ControlControlControl

MediumAccess

PowerC t l typically in Access

ControlControl yp y

NodeB

Physical layer


Example of Coverage and Best Server Map

p ap

rage

ma

erve

r m

a

cove

r

best

se

Application: RF engineering (cell layout) Application: HO decision


Legend: red indicates high signal level, yellow indicates low level

Legend: color indicates cell with best CPICH in area

Interference in CDMA Networks


Interference in CDMA - Uplink


Interference in CDMA - Downlink


Cell Breathing


Cell Breathing

coverage coverage coverage

Coverage depending on load: load causes interference which reduces the area where a SIR sufficient for communication can be provided

coverage low load

coverage medium load

coverage high load


yellow area: connection may drop or blocked

Coverage vs. Capacity

Capacity depends on:QoS of the users (data rate error performance (bit-error-rate))QoS of the users (data rate, error performance (bit error rate))User behaviour (activity)Interference (intra- & inter-/noise)Number of carriers/ sectors

Coverage (service area) depends on:Interference (intra- & inter-cell) + noisePathloss (propagation conditions)QoS of the users (data rate error performance (bit error rate))QoS of the users (data rate, error performance (bit-error-rate))

Thus, trade-off between capacity and coverage


Coverage vs. Capacity

3

3.513kbps circuit switched service capacity versus maximum cell radius

2

2.5

dius

(km

)

1

1.5←Downlink

Max

imum

cel

l rad

0.5

1

Uplink→

M

0 20 40 60 80 100 120 140 160 180 2000

Erlangs (2% GOS)

Downlink limits capacity while uplink limits coverageDownlink depends more on the load (user share total transmit BS power)


Handover Control: Basics

General: mechanism of changing a cell or base station during a call or sessionGeneral: mechanism of changing a cell or base station during a call or session

Handover in UMTS:UE may have active radio links to more than one Node BUE may have active radio links to more than one Node BMobile-assisted & network-based handover in UMTS:

UE reports measurements to UTRAN if reporting criteria (which are set by the UTRAN) are metthe UTRAN) are metUTRAN then decides to dynamically add or delete radio links depending on the measurement results

Types of Handover:Soft/Softer Handover (dedicated channels)H d H d ( h d h l )Hard Handover (shared channels)Inter Frequency (Hard) HandoverInter System Handover (e.g. UMTS-GSM)Cell selection/re-selection (inactive or idle)

All handover types require heavy support from the UMTS network infrastructure!


Macro Diversity & Soft Handover (Wrap-Up)

NodeB 1NodeB 1NodeB 2NodeB 2

UEUE

Downlink: combining in the mobile station Uplink: combining in the base station and/or radio network controller


Uplink: combining in the base station and/or radio network controller

Soft/Softer Handover

In soft/softer handover the UE maintains active radio links to more than one Node BCombination of the signals from multiple active radio links is necessary

Soft HandoverThe mobile is connected to (at least) two cells belonging to different NodeBsI li k th i l bi d i th RNCIn uplink, the signals are combined in the RNC, e.g. by means of selection combining using CRC

Softer HandoverThe mobile is connected to two sectors within one NodeBThe mobile is connected to two sectors within one NodeBMore efficient combining in the uplink is possible like maximum ratio combining (MRC) in the NodeB instead of RNC

Note:In uplink no additional signal is transmitted, while in downlink each new link causes interference to other users, therefore:

Uplink: HO general increase performanceUplink: HO general increase performanceDownlink: Trade-off


Soft and Softer Handover in Practice


Soft Handover Control

NodeB 1NodeB 1 NodeB 2NodeB 2UEUE

soft handoversoft handoverareaarea

• Measurement quantity, e.g. EC/I0 on CPICHR l ti th h ld δ &

MeasurementMeasurement

NodeB 1NodeB 1 UEUE

• Relative thresholds δadd & δdrop for adding & dropping

• Preservation time Tlink to a oid “ping pong” effects

QuantityQuantityCPICH 1CPICH 1

δδdropdrop avoid “ping-pong” effects• Event triggered measurement

reporting to decrease signalling loadCPICH 2CPICH 2

δδdropdrop

δδaddadd TTlinklink

signalling load

Link to 1Link to 1 Link to 1 & 2Link to 1 & 2 Link to 2Link to 2 timetime


Soft Handover – Simulation Results

25%

20%

ility

oppi

ng)

1 link

10%

15%

ge P

roba

bg

and

Dro 1 link

max 2 SHO linksmax 4 SHO links

6 SHO li k

5%

10%

Out

ag(B

lock

in max 6 SHO links

0%5 15 25 35 45 55

Offered Traffic [Erlang per site]


Soft handover significantly improves the performance, but …

Soft Handover – Simulation Results II

2

1,5

ve L

inks

1

er o

f Act

iv

0,5

ean

Num

b

01 2 4 6

Me

1 2 4 6

Max. Active Set Size


… the overhead due to simultaneous connections becomes higher!

( CS) H

Inter-Frequency Handover

Hierarchical cell structure (HCS) Hot-spot

Macro Micro Macrof2

f2

Hot spot

Handover f1 ⇔ f2 always needed

f1

f2

f1

Handover f1 ⇔ f2 needed

f1

2f1 f1

1 2 ybetween layers

1 2sometimes at hot spot

Hard handoverInter-frequency measurements of target cell needed in both scenariosq y gMobile-assisted handover (MAHO)

slotted (compressed) mode for inter-frequency measurements to find suitable target cellgalso supports GSM system measurements

Database assisted handover (DAHO)no measurements performed on other frequencies or systems


no measurements performed on other frequencies or systemsuse cell mapping information stored in data base to identify the target cell

Power Control: Basics

C t l th tti f th t it i d tControls the setting of the transmit power in order to:Keep the QoS within the required limits, e.g. data rate, delay and BERMinimise interference, i.e. the overall power consumption

Power control handles:Path Loss (Near-Far-Problem), Shadowing (Log-Normal-Fading) and Fast Fading (Rayleigh- Ricean-Fading)Fast Fading (Rayleigh , Ricean Fading) Environment (delay spread, UE speed, …) which implies different performance of the de-interleaver and decoder

Uplink: per mobileDownlink: per physical channel

Three types of power control:Inner loop power controlOuter loop power control (SIR-target adjustment)O l t l ( ll ti )Open loop power control (power allocation)

Downlink power overload control to protect amplifierGain Clipping (GC)


Gain Clipping (GC)Aggregated Overload Control (AOC)

Near-Far Problem

Near-Far Problem:Near Far Problem:• Spreading sequences are not orthogonal

(multi-user interference)• Near mobile dominates

UE 1UE 1

• Near mobile dominates• Signal to interference ratio (SIR) is lower for far

mobiles and performance degradesN d BN d B

The problem can be resolved through dynamic power control to equalize all received power levels

NodeBNodeB

AND/OR: by means of joint multi-user detection (MUD)

UE 2UE 2( )


Impact of Power Control


Source: H. Holma, A. Toskala (Ed.), “WCDMA for UMTS”,

Closed Loop Power Control

Closed loop power control is used on channels which are established in both directions, such as DCHThe receiver generates transmit power commands (TPC) based on the

i d i d li h C d b k h i i hestimated received quality; the TPC are send back to the transmitter in the opposite directionThe transmit power is adjusted according to the received TPC

ReceiverReceiver DecoderDecoderPAPAUser dataUser dataUser dataUser data

ReceiverReceiver DecoderDecoder

BLERBLER--estimateestimate

PAPA

OuterOuterlooploop

SIRSIR--estimateestimate

BLERBLER estimateestimateTPCTPC

commandscommands

TPCTPCAdjust SIRAdjust SIRtargettarget

Inner loopInner loop

MUXMUXDeMUXDeMUX

TPCTPCcommandscommands


Inner/Outer Loop Power Control

Inner Loop Power Control (1500 Hz)objective: adjust transmit power to keep the received SIR at a given SIR targetrealisation using the SIR-estimate after RAKE combining:

if SIRest > SIRtarget, then generate TPC “DOWN”if SIR ≤ SIR then generate TPC “UP”if SIRest ≤ SIRtarget, then generate TPC UP

realisation in base station (nodeB)similar mechanisms for up- and downlink

Outer Loop Power Control (100 Hz max)objective: meet the required reception quality, e.g. average BLERj q p q y, g grealisation using the CRC attachment

if CRC ok, then decrease SIRtarget = SIRtarget – Δdown

if CRC fails then increase SIR SIR + Δif CRC fails, then increase SIRtarget = SIRtarget + Δup

Δdown = Δup × BLERtarget / (1–BLERtarget)realisation in radio network controller (RNC)


Power Control Results

7.5

6.5

7IR

[dB]

6

6.5

quire

d UL

S

5.5

Req

PedAVehA

50 20 40 60 80 100 120

Velocity [km/h]

SIR requirement strongly depends on the environment (due to different fast fading conditions – Jakes models)


⇒ outer loop power control needed to adapt SIR

Open Loop Power Control

Open loop power control is used on channels that cannot apply closed loop power control, e.g. RACH, FACHThe transmitter power is determined on the basis of a path loss estimate p pfrom the received power measure of the opposite directionTo avoid excessive interference, probes with incremental power steps until a response is obtained: “power ramping”p p p g

R iR iReceiverReceiver

Measure Rx powerMeasure Rx power

Adjust Tx powerAdjust Tx power

PAPAUser dataUser data


Power Ramping during Random Access

Preamble #1 Message

RACH

Preamble #N

AICH/FACH

AcquisitionIndicator

MessageAcknowledgement

Random access on RACH used forInitial accessShort packets: user data, signalling (e.g. measurement reports)S o t pac ets use data, s g a g (e g easu e e t epo ts)

Power RampingMS starts with relatively low transmit powerR t ith i t l t til i iti i di tRepeats with incremental power steps until acquisition indicator receivedMessage sent with constant power offset


Load Control: Basics

General limitations in 3rd generation mobile systems:

Li it d t ( d )Limited resources: spectrum, power, (code space)

Not all service requests can be granted

Different types of services in the same cellular environmentDifferent types of services in the same cellular environmentService parameters (user behaviour and QoS) and environmental conditions (propagation) vary over time

CDMA-specific impacts:Coverage in CDMA systems depends on the cell loading:Coverage in CDMA systems depends on the cell loading:“cell breathing”CDMA systems may become unstable in highly loaded situationsdue to fast inner-loop power controldue to fast inner loop power control


Load Control

Main objective:Main objective:Avoid overload situations by controlling system loadMonitor and controls radio resources of usersMonitor and controls radio resources of users

Call Admission Control (CAC)Admit or deny new users, new radio access bearers or new radio li klinksAvoid overload situations, e.g. by means of blocking a new callDecisions are based on interference and resource measurementsDecisions are based on interference and resource measurements

Congestion Control (ConC)Monitor, detect and handle overload situations with the already connected usersBring the system back to a stable state, e.g. by means of dropping an existing calldropping an existing call


Basic Resource Equations (CDMA)Rbi: data rate of user i hi: channel coefficient of user iRbi: data rate of user i hi: channel coefficient of user iPpilot: Pilot power Fi: Orthogonality factor of user i (multipath)Ith: thermal noise

DownlinkDownlinkQuality (RAKE):

thinterpilotj jii

iibi

t

bi

IIPPhFPhRW

NE

+++⋅⋅⋅⋅

=∑ )(

Resource of user i:

U li k

Power dTransmitte Total:Poo

ii P

P=α

UplinkQuality (RAKE):

thinterij j

ibi

t

bi

IIPPRW

NE

++⋅

=∑ ≠

ˆˆ

Resource of user i: ceInterferen Total:Iˆ

otbb

tb

o

ii NERW

NEIP

+==α

Resource Consumption strongly depends on: data rate, quality (Eb/Nt), receiver structure (RAKE etc., channel estimation, path tracking, …)N li l ti b t d t t d i d E /N


Non-linear relation between resource, data rate and required Eb/Nt

Resource Consumption

• Service/BLER-dependent resource consumption9

10resource consumption

• Uplink example:– Service I: Voice

R 12 2kb E /N 5dB6789

B]

Rb = 12.2kbps, Eb/Nt = 5dBαI = 0.99%

– Service II: DataR 144kbps E /N 3 1dB

345

Eb/N

t [d

Voice (12.2 kbps)BLER = 1% Rb = 144kbps, Eb/Nt = 3.1dB

αII = 7.11%

with012

10 100 1000

BLER 1%Data (144 kbps)

BLER = 10%

t10 100 1000

Data Rate Rb [kbps]

0,5% 1% 2% 5%α)(

)(

tbb

tb

NERWNE

+=α

where W is the channel bandwidth

and Rb is the data rate

10% 20% 35% 50%α


UL/DL Load Measures

• UL Measurement

∑ ˆ

16

18

20

[dB

] • Load Estimate

thintercelli i IIPI ++=∑∈0

10

12

14

Pow

er R

ise

thr_CAC = 75%

thr_ConC = 90%

th

thcurrent

IINRNRI

II

0

0

0 11

=

−=−

=η

2

4

6

8

Noi

se R

ise/

P

• DL Measurement

th0

Pi PPP +=∑0

0

2

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

N

• Load Estimate

Pcelli i∑∈0

PP 1System Load

p

Pcurrent

PPPRPRP

PP

0

0

0 11

=

−=−

=η


p

Call Admission Control

• Admitting new call alwaysAdmitting new call always increases the cell loading

• CAC avoids overload by limiting this increase16

18

20

[dB

]

thr CAC

• CAC load check:

?CACnewcurrent thr≤α+η10

12

14

Pow

er R

ise

t

_

• Threshold setting thrCACtrade-off between

2

4

6

8

Noi

se R

ise/

P η_current + α_new

– Maximize capacity: prevent excessive blocking

– Avoid overload: thrCAC < thrConC

0

2

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

N

η_current

System Load


Call Admission Control (CAC)

Admitting a new call always increases the cell loadAdmitting a new call always increases the cell loadIn order to avoid overload situations, the admission control will limit this load increaseThe principle is to check the current system load plus the expected resource consumption of the new call against the call admission threshold:

l d ti ≤ th ?load + consumption ≤ thrCAC ?

In case the admission check fails, the basic strategy is to protect ongoing calls by denying the new user access to the system since dropping is assumed to be more annoying than blockingAdmission control is required for uplink and downlinkq pArrivals of high-data-rate users that require a large amount of resources (especially in the downlink) may demand global information


Congestion Control

• Even with efficient CAC overload

16

18

20

[dB

]

η current

Even with efficient CAC overload situations still occur due to

– Mobility (esp. downlink)

– Activity

10

12

14

Pow

er R

ise

η_

NR/PR_max

y

• During overload quality of allusers is deteriorated !

• Triggered by measurement

2

4

6

8

Noi

se R

ise/

P

η_reducedthr_ConC

Triggered by measurement

• Action: reduce offered traffic by

ConCcurrent thr≥η

0

2

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

N • Action: reduce offered traffic by downgrading one user

• Threshold setting thrConC to preserve the coverage:

System Loadp g

– NRmax: limitation of MS single transmit power Pimax

– PRmax: limitation of BTS total t it P


transmit power P0max

Congestion Control (ConC)

Due to the mobility (especially of high data rate users) overload situationsDue to the mobility (especially of high data rate users) overload situations occur even if an efficient admission control algorithm is used!The congestion control is activated once the system load exceeds the congestion threshold:congestion threshold:

load ≥ thrConC

In order to overcome the overload situation the load must be lowered until load < thrConC

This is done by reducing the offered trafficLowering the data rate of one or several services that are insensitive to increased delays – this might be the most preferred methody g pPerforming inter-frequency (inter-system) handoverRemoving one or several connections

Global information may be required to minimize the number of alteredGlobal information may be required to minimize the number of altered connections


T adeoff bet een blocking and d opping

Call Admission Control: Simulation Results I

Tradeoff between blocking and droppingExample: 64k per user, urban

45%

50%thr_CAC = 50%thr_CAC = 75%thr_CAC = 90%

16%

18%

20%thr_CAC = 50%thr_CAC = 75%thr_CAC = 90%

30%

35%

40%

abili

ty

12%

14%

16%

abili

ty

20%

25%

ocki

ng P

roba

8%

10%

oppi

ng P

rob

5%

10%

15%Blo

2%

4%

6%Dro

0%

5%

5 15 25 35 45 55

Offered Traffic [Erlang per s ite ]

0%

%

5 15 25 35 45 55

Offered Traffic [Erlang per s ite ]


Offered Traffic [Erlang per s ite ] Offered Traffic [Erlang per s ite ]

Cell load depending on CAC threshold

Call Admission Control: Simulation Results II

Cell load depending on CAC thresholdExample: 64k per user, urban

80%

90%

50%

60%

70%

adin

g

20%

30%

40%

Cell

Loa

0%

10%

20% thr_CAC = 50%thr_CAC = 75%thr_CAC = 90%

5 15 25 35 45 55

Offered Traffic [Erlang per site]


Packet Data Control: Channel Switching

FlexibilityAsymmetrical data ratesVery low to very high data ratesVery low to very high data ratesControl information/user information

Efficient transmission making good use of CDMA characteristicsDedicated channel (DCH)

Minimise transmission power by closed-loop power controlp y p pIndependence between uplink and downlink capacity

Common channelRandom access in the uplink (RACH)Random access in the uplink (RACH)Dynamic scheduling in the downlink (FACH)

Adaptive channel usage depending on traffic characteristicsInfrequent or short packets ⇒ Common channel (Cell_FACH)Frequent or large packets ⇒ Dedicated channel (Cell_DCH)


Channel Switching – Example

CELL_DCH CELL_FACH CELL_DCH

DCH Active Time

Page Download Time Reading Time

DCH Active Time“Chatty Applications”

Example: Web serviceChatty apps : keep alive message stock tickers etcChatty apps.: keep alive message, stock tickers, etc.(e.g. 100 bytes every 15 sec)Second stage: when no activity in CELL_FACH then switch to URA_PCH


Packet Data Control

Burst Admission Controlk dDecision on starting packet data transmission

Similar principle like call admission control, i.e. check the system load against a threshold that is usually different from thrCAC

Time horizon: 100msec … 10sec

Rate AdaptationRate AdaptationChoose data rate according to transmit power

UE nearby NodeB ⇒ high data rateUE at cell edges ⇒ low data rate

Decrease data rate in case of overload, cf. congestion control


Power Control vs. Rate Adaptation

•• Power Control:Power Control:–– Balances user received quality (BLER, SIR)Balances user received quality (BLER, SIR)–– Users at cell center get less share of BTS Users at cell center get less share of BTS

transmit power assigned than at cell edgetransmit power assigned than at cell edgetransmit power assigned than at cell edgetransmit power assigned than at cell edge–– Occurrence of power overloadOccurrence of power overload

•• Rate Adaptation:Rate Adaptation:

UE 1UE 1

–– Transmit power ~ data rateTransmit power ~ data rate–– Users at cell edge get lower data rate Users at cell edge get lower data rate

assigned than at cell centerassigned than at cell center–– Reduces also power overloadReduces also power overloadReduces also power overloadReduces also power overload

•• In UMTS combination of power control and In UMTS combination of power control and rate adaptation on DCHrate adaptation on DCH

NodeBNodeB

high data ratehigh data rateareaarea

UE 2UE 2low data ratelow data rate

areaarea

areaarea


Rate Adaptation Performance

UMTS_urban, 50 kByte UMTS_urban, 50 kByte

30%

35%

40%384k64kadaptive

6

7

8

384k

20%

25%

30%

Prob

abili

ty

4

5

6

elay

[sec

]

384k64kadaptive

10%

15%

Out

age

P

2

3

Mea

n D

e

0%

5%

200 300 400 500 6000

1

200 300 400 500 600

Rate adaptation significantly improves the RRM performance.

Cell Throughput [kBit/sec] Cell Throughput [kBit/sec]


p g y p p

Dynamic Scheduling

• “Statistical multiplexing” of data packets from different data

Sample Flow

packets from different data flows over one shared medium, e.g. on DSCH or HSDPAS h d li ith ti h i f

Flow #1

• Scheduling with time-horizon of 2msec … 1sec

• Optimised usage of radio

UE 1

Fl #2 resources• Exploitation of the short-term

variations on the radio NodeB

Flow #2

channels (opportunistic scheduling)

• Provides certain degree of

Flow #3 UE 2

o des ce a deg ee oQoS

UE 3


UE 3

Conclusions

Basic RRM algorithms presented here:Basic RRM algorithms presented here:Handover ControlPower ControlLoad ControlPacket Data Control

RRM procedures not discussed here:Spreading code managementRRM f TDD d ti l t tRRM for TDD mode: time-slot management

Related issue: RF engineering

Enhanced RRM algorithms in order to:Support advanced packet data transmission techniques such as HSDPASupport advanced packet data transmission techniques such as HSDPA (High-Speed Downlink Packet Access)Support intelligent transmitter/receiver structures like adaptive antenna or MIMO techniques


or MIMO techniques

utran radio resource managementutran radio resource · pdf fileutran radio resource...

Documents