femtocell by son

1Copyright © 2009 BeFEMTO– Broadband Evolved FEMTO Network. All Rights reserved

Mischa DohlerCTTC, Spain

Manuel Palmowski & Dimitri MarandinmimoOn, Germany

Copyright © 2009 BeFEMTO– Broadband Evolved FEMTO Network. All Rights reserved.

LTE/LTE-A SON (for Femtocells)


System Degrees of Freedom [1/2]

Degrees of Freedom (DOF) / Area: new system DOF = old system DOF x 20‐30 new system density = old system density x 4 new DOF/km2 = old DOF/km2 x 100

1

100

10000

1000000

00000000

1E+10

1E+12

GSM EDGE UMTSHSPA

LTELTE‐A

CELLULAR + WIFI + FEMTO

CELLULAR + WIFI

CELLULAR


System Degrees of Freedom [2/2]

With femtocells, the DOF increases significantly!

[presented by Interdigital: Globecom’11 – IWM2M, Houston]


Likelihood of Failure/Outage

Important High‐Level Insides: DOF/area increases failures/outages increase:

loss in revenue if nothing done, and/or very costly if addressed by human labour

[© http://www.scien

cedirect.com

/scien

ce/article/pii/S0951832004000031]


Importance of SON

Previously Manual Processes: reduce manual intervention for deployment savings automate repetitive processes examples: automatic planning & self‐configuration

Too Fast/Complex Processes: improve run‐time operation based on real‐time data automate optimization of critical network elements example: self‐optimization & self‐healing

SON automation, however, is not new but rather a further evolution of prior automation efforts


Elements of SON

Important Elements: elements of “autonomous” elements of “distributed” elements of “intelligent/cognitive” elements of “optimality”

SON ≠ SON: despite large DOF, SON has to be (close to) optimal since it makes no sense to have a system with amazing components which however underperforms due to poor “management”


SON in Cellular Networks

Optimisation/Maintenance

Self Optimisation

Self Healing

Deployment

Planning

Self Configuration

Automated Planning


Presentation Overview

1. SON in NGMN & 3GPP LTE

2. SON Architectural Elements

3. 3GPP SON Features

4. 3GPP SON Functional Details

5. Conclusions


1. SON in NGMN & 3GPP LTE


NGMN’s Input To SON

Next Generation Mobile Networks (NGMN) Alliance: created in 2006 by group of operators business requirements driven often based on use‐cases of daily networking routines

NGMN and SON: SON input to 3GPP since 2006 10 SON use cases have been defined (see right) which input to 3GPP Rx

QoS SON is likely to appear shortly


3GPP High‐Level Structure

maintenance/developmentof GSM/GPRS/EDGE RAN

maintenance/developmentof UMTS/HSPA/LTE RAN

system architecture, service capabilities, codecs (inc. EPC)

CN interfaces, protocols, interworking, IMS, terminals, SIM

Management!


SON in 3GPP LTE Release 8

3GPP R8 is about eNB self‐configuration: automatic neighbor relation automatic physical cell ID (PCI) assignment automatic inventory automatic software download


SON in 3GPP LTE Release 9

3GPP R9 is about network optimization procedures: mobility robustness and handover optimization RACH optimization load balancing optimization inter‐cell interference coordination (ICIC)


SON in 3GPP LTE‐A Release 10

3GPP R10 is about overlaid networks: coverage & capacity optimization enhanced ICIC cell outage detection and compensation self‐healing functions minimize drive test energy savings


SON in 3GPP LTE‐A Release 11

3GPP R11 will be about heterogeneous networks: automated network management troubleshooting multi‐layer, multi‐RAT heterogeneous networks among other issues


3GPP SON Status

PCI_Sel.

ANR

MLB, MRO, RACH

Energy Saving (ES)

Minimize Drive Test (MDT)

SON MgmtSelf‐Healing

MDT

PCI_Selection

Energy Saving

MLB, MRO,CCO

SA5

RAN

Rel9 Rel10Rel8

ANR = Automatic Neighbor RelationPCI = Physical Cell IDMLB = Mobile Load BalancingMRO = Mobility Robustness Opt.RACH = Random Access ChannelCCO = Capacity & Coverage Opt.


2. SON Architectural Elements


SON Architecture Approaches [1/2]

Levels of SON Execution: localised: autonomous SON execution based on purely local information at (H)eNB & UE

distributed: autonomous SON execution based on information exchanged with neighbouring (H)eNB (eg via X2 interface)

centralized: decision taking based on (fairly complete) system information (eg at NM/DM/EM levels)

hybrid: any mixture of above

X2

NM

DM/EM

SON

SON

DM/EM

SON SON

SON

NM = Network ManagementDM = Device ManagementEM = Element Management


SON Architecture Approaches [2/2]

centralized distributed localised hybrid

KPI = Key Performance Indicator


Key Time Intervals

For Centralised SON, timing is important: Collection Interval: time period during which statistics and data are collected; limited by vendor’s OAM bandwidth; typical 5min (i.e. not at scheduling level!)

Analysis Interval: time period needed to draw decision; typically several collection intervals (filtering effect by considering also prior data history)

Change Interval: time period between executing the changes in the network; typically limited by system’s operational constraints


Important SON Trade‐Offs

Centralised

multiple cellslong term statisticsuses OAM multi‐vendor/multi‐RAT

self‐configurationcoverage optimisation(load balancing)

Distributed

normally ca. 2 cellsuses X2multi‐vendor (X2)

handover opt.load balancingRACH opt

Localised

adv. RRMsmall impact on ncellsshort term statistics

scheduler optLink adapt. optRACH opt

faster & less prone to “single point of failure”

info from number of cells/RATs


Coordinated SON Model

very difficult to standardise: lots of different ways to implement a SON function are possible e.g. load balancing (LB) done at NM level but handover (HO) optimization done at eNB level and both want to adjust the same parameter in eNB

practical way around today: if function at NM level, “configure” is standardised at eNB level not standardised currently primary & secondary targets are defined

[based on S5‐102029]

man

agem

ent

decision

taking

execution

feedback


BeFEMTO HeNB Architecture

SO‐RRIM

MAC

RLC

PDCP

PHYRF

BB

Radio Functions

SON Enabling Functions

Coverage EstimationPositioning …

SO-RRIM Sub-functions

LoadBalancing

Mobility Control

AdmissionControl

EnergySaving

…Scheduler

MAC

IP/IPsec

TCP

PHY

Geo Sublayer

SCTPUDP

RRCHTTPS

SOAP RPC

GTP‐U

S1AP/X2AP

Network /Mgmt Functions

Fault Diag.Probe

Routing

Network Sync

Local NetwElem Agent

Protocol Stacks

S1 / X2 / Type 1C / Rt /Lm interfaces Uu / Un interfaces

Radio SON Coordinator

LocalLocationMgmt.

Perform.Mgmt

IKEv2

EAP‐AKA

NMM

TCP/UDP

Security

ManagementFunctions

NetworkFunctions

ContextLearning

usersnetwork


3. 3GPP SON Features


SON in Cellular Networks

Optimisation/Maintenance

Self Optimisation

Self Healing

Deployment

Planning

Self Configuration

Automated Planning


SON Functions (NGMN & 3GPP)

Self‐Optimisation

Self‐Healing

Self‐Configuration

Automated Planning


Automated Planning

From “trial & error” to “planning automation”:

[© ICT MOMENTUM]


Self‐Configuration

Important Self‐Configuration Functions: authentication address allocation secure OAM tunnel setup SW installation inventory management transport parameters setup radio parameters setup self‐test


Self‐Optimisation

• Automatic Neighbour Relation• Mobility Robustness/Handover Optimisation• (Mobility) Load Balancing• RACH Optimisation• Coverage and Capacity Optimisation• Energy Saving• Interference Coordination• HeNB SON

3GPP

• QoS Optimisation (advanced RRM)• Scheduler• admission control• congestion control• link level/L2/HARQ re‐tx/MIMO parameters

additional from Socrates/NGMN (not discus)


Self‐Healing

Important Self‐Configuration Functions: failure recovery cell outage compensation


4. 3GPP SON Functional Details


eNB Self‐Configuration (Rel. 8)

Benefits: reduction of manual labour faster and cheaper network deployment

Description: detect transport link & establish connection download & upgrade software set up initial configuration parameters perform self‐test go into “operational” mode


Benefits: keeping centralized neighborhood tables is impossible automating this is one of the most important SON features

Description: keep updated neighborcell(ncell) table through UE obtain ECGI & IP via MME to communicate to target eNB

ECGI = E‐UTRAN Cell Global IdentifierOAM = Operating Administration & MaintenanceMME = Mobility Management EntityRRC = Radio Resource ControlerTCI = Target Cell Identifier


PCI Selection (Rel 8/9)

Planned/Centralised PCI direct value is assigned (3 x 168 = 504 in total only!)

SON/Distributed PCI selection a range of PCIs is given

eNB will not select cell IDs that arereported by UEsreported over the X2 interface (not for HeNB as of now)acquired through NW listen mode (will not catch all)

select PCI value randomly from the remaining PCIs

PCI collisions: >1 eNB seeing each other choose same ID

PCI = Physical Cell ID


PCI Selection ‐ Collisions

PCI collision probability depends on the number of PCIs available for such cells

PCI reported belongs to two different ECGI Collision recovery

Number of PCIsavailable for

non-macro eNBs

Percentage of UEs observing collision of

PCIs with random PCI allocation

Percentage of UEs observing collision of PCIs

with random PCI allocation plus downlink receiver (sensitivity at -6

dB)

Percentage of UEs observing collision of PCIs

with random PCI allocation plus downlink receiver

(sensitivity at -8 dB)

10 15.2% 12.2% 11.7%

50 3.3% 2.6% 2.4%

100 1.7% 1.3% 1.2%

200 0.9% 0.7% 0.6%

500 non‐macro eNBs (e.g. HeNBs) dropped randomly in each macro eNB [R3‐091018]


MRO (Rel‐9, ext Rel‐10)

too early intra‐RAT HO: wrong cell intra‐RAT HO:too late handover (HO):

Mobility Robustness Optimization (MRO) because …… Handovers (HOs) between same (intra) and different (inter) radio access technologies (RATs) can go very wrong:



Mobility Robustness Optimization detect radio link connection failures due to being too early, too late

handovers or handover to wrong cells caused by not optimised intra/inter‐RAT handover parameters also used for optimising idle mode cell reselection parameters

UE sends Radio Link Failure (RLF) measurement report to eNBwhere UE did re‐establishment

Optimised for Rel‐10 to use also ConnectionRequest

too early handover


eNB


OAM

MRO

MobilityControl

RRM config examples:Trigger ThresholdTime‐to‐TriggerHysteresis against Ping‐PongSpeed & Neighbour Param.Electric Antenna Tilt

config target:Rate of failures related to HO

monitor KPIs, alarms:Nr of HO eventsNr of HO failuresNr of too early HO


MLB (Rel‐9)

Mobility Load Balancing Distribute cell load evenly among cells or Transfer part of traffic to other cells so that radio resources remain highly

utilized Optimisation of mobility parameters or handover actions Also includes setting of cell reselection mode for idle mode balancing

Load Reporting the radio resource usage (UL / DL PRB usage) HW and TNL load indicator (UL/DL load: low, mid, high, overload) a cell capacity class indicator and capacity value (UL / DL)

PRB = Physical Resource BlockTNL = Transport Network LayerHW = HardwareUL = uplinkDL = DownlinkRAB = Radio Access Bearer (next slide)


MLB – Active Mode LB

target cell is informed proposed source cell

change proposal for target cell

change

Serving eNB

OAM

MLB

MobilityControl

RRM configHO Trigger‐/+ 10dB (incr/reduce time until HO decision)

config target:‐ RAB setup failure rate due to load‐ Rate of failures related to HO

monitor KPIs, alarms:Nr RAB setup failure due to loadTotal nr of RAB setup failure

Target eNB LOAD INFO

MOBILITY CHANGE


RACH optimisation (Rel 9)

Optimises the (P)RACH configuration access to internal measurements and access to feedback from UEs (via RRC) and from other (H)eNBs

The setting of RACH parameters that can be optimized are: RACH configuration (resource unit allocation), preamble groups split,

backoff, power control

UEs which receive UEInformationRequest shall feedback:Number of RACH preambles sent until the successful RACH completion (indication of quality of link)

Contention resolution failure (indication of cell load) X2 allows to exchange RACH config between eNBs Rel 10 OAM policy targets: Access Probability and Access Delay

Probability


Min. Drive Tests (R10 procedure)

Performs logging of available measurements together with location and time for logged measurement configured UEs

Types: Immediate and logged measurements Immediate measurements (UE optional FFS)Reuses normal measurements with addition of location

Enhancement of RLF Report to include location Idle mode logged measurements (UE optional)Input: duration, interval, cell ids and/or tracking areasTime, Location, Cell id, scell/ncell RSRP/Q

Add. eNB meas: eNB RIP and CQI under disc, + PHR and UL SINRMDT activation via OAM per cell/region (not reused at HO) S1 UE specific trace activation (can be cont. on HO)


Coverage and Capacity Opt (Rel‐10)

Coverage definition“coverage hole” (should cover UL also)

unsuccessful RX of DL control channels and signalling radio bearersdifferent control channels have to be optimised jointly

“weak coverage”: cell‐edge users, minimum service rate not reached

Types of coverage holesother RATs/no radio, coverage too small/too large

CCO detection using MDT/MRO Inputs: UE measurements, Performance measurements,

alarms Outputs: power control, antenna tilt, azimuth

Capacity & Coverage improvements


Energy Saving

Optimise energy consumption: a cell providing additional capacity can be switched off when its capacity is no longer needed and to be re‐activated when needed

The decision typically based on cell load information Peer eNBs are about the switch‐off over the X2 interface (eNB

Configuration Update procedure) Informed eNBs keep the cell config when a cell is off and may request re‐

activation if needed (X2: Cell Activation procedure) How to select to correct cell to switch‐on? The operators can configure ES function:

enable/disable ES OAM policies for switch‐off and reactivation

open interRAT, overlaid inter‐eNB savings solutions to capacity limited scenario of inter‐eNB energy saving ES in network sharing


5. Conclusions


SON Conclusions

Benefits of SON: reduction of human intervention real‐time optimization of network significant CAPEX & OPEX savings!

NGMN & 3GPP: important SON features standardised from R8‐R11 focus shifting from auto‐planning & self‐configurationtowards self‐optimization and self‐healingmost important SON features have been discussed in thispresentation whereas concrete embodiments of SON comes later today

femtocell by son

Documents

automated planning planning

ues observing collision

random pci allocation

soninngmn

ngmn

downlink receiver

configuration deployment

optimisation maintenance