smar tech overview

11 issue 11/09 TECHCOM Consulting TECHCOM.DE Copyright © All rights reserved

Radio Performance Optimisation RAN network in process

RAN optimisation

Key Performance

Nominal planning


Radio Performance Optimisation Service trace process for: Mobile Operator Network

NetworkNetworkserviceservice

•• Iub/IuIub/Iu data data capturingcapturingby Agilent DNA/SARTby Agilent DNA/SART

•• Air interface Air interface analysisanalysisby TECHtraceby TECHtrace

Agilent/TECHCOM successfully perform a 3G RAN workshop in South Africa, November 2009. The operator bought in the past already the Agilent DNA and SART application to be able to troubleshoot 3G RAN related problems. Target of the workshop was to present the additional Agilent 3G RAN troubleshooting application – data mining database – to analyze high level KPI and drill down quickly to the rootproblems and RF Optimization application from smartTECH

Following Performance KPI where seen in the network base on the 3 days Call Data Record generation for customers only at one RNC:

Vendor feature activation Auto tuning enabled UL (max. range 20dB) Noise PwR-105 dBm default enabled UL Interference limitation deactivated EcNo Filter coefficient is default 600 msRRM Reports)* TCP PWR is related to Release 99


Radio Performance Optimisation

Monitoring during Call Setup (RRC connection request) Quality Analysis EC/I0 (all calls, RT, NRT) Interference Analysis Little i (all calls, RT, NRT) Propagation Delay (all calls, RT, NRT - call setup versus distance)

Monitoring during SHO (Event 1A – best active cell) SHO Window (all calls, RT, NRT) Cell Matrix

Monitoring of RSCP (all calls, RT, NRT)

Monitoring of load Transmitted Carrier Power Received Total Wideband Power

Conclusions quality Suggestions

Content

RAN optimisation

Key Performance

Nominal planningNode B


Radio Performance Optimisation Service architecture

TracingTracing•• Data filteringData filtering•• analyzinganalyzing•• BenchmarkBenchmark

Service Service processprocess

AgilentRAN Monitoring with DNA/SARTRoute cause trouble shootingCDR/CSSR diagnosticsSignalling analysis UTRAN

RANanalyzing

GISUE analyzing

Monitoring

RPO/KPI


Radio Performance Optimisation SW tools

smartRadiosmartRadioInterference monitoring

smartRANsmartRANRANPAR-libraryOSS/RANPAR

smartCountsmartCountKPI -libraryOSS/counter

Protocol analysis SARTProtocol analysis SART

3G RAN/UTRAN 3G RAN/UTRAN performanceperformance

Cell/GIS MAP Cell/GIS MAP performanceperformance

smartInfo smartInfo GIS/MAP & QoS

We used… SmartTECH solutions DNA/SART

Benchmark with… OSS KPI (tools) RANPAR (vendor)



Quality Analysis EQuality Analysis ECC/I/I00 (all calls, RT, NRT(all calls, RT, NRT Interference Analysis Little i (all calls, RT, NRT)Interference Analysis Little i (all calls, RT, NRT) Propagation Delay (all calls, RT, NRT Propagation Delay (all calls, RT, NRT -- call setup versus distance)call setup versus distance)

[RACH] RRC:RRC Connection Request

ALCAP:ERQ

NBAP: RL Setup Request

Start TX/RX

[FACH] RRC: RRC Connection Setup

NBAP: RL Setup ResponseAC to check to accept or reject RRC

Connection Request ALCAP:ECF

RRC Connection Setup phase

RNC

RANanalyzing

Monitoring during Call Setup - Busy hour - RRC Connection set-up


Radio Performance Optimisation EC/I0 map – from trouble shooting at Iub

Break point Break point to initiate HO eventsto initiate HO eventsEcEc/Io/Io

RSCPRSCP

GISUE analyzing


Radio Performance Optimisation Monitoring during Call Setup - Busy hour - RRC Connection set-up Quality Analysis EQuality Analysis ECC/I/I00 (all calls, RT, NRT)(all calls, RT, NRT) RAN

analyzing

Reference RAN Parameter range for good/acceptable quality: Ec/Io = -3 dB …-8 dB

Busy hoursIn general, during RRC connection request statistics describes acceptable median Ec/Io conditions, which are similar both for RT and NRT calls.

Medium Medium EcEc/Io of cells /Io of cells for RTfor RT

Number of cells versus median Ec/Io, for RT calls


Radio Performance Optimisation Monitoring during Call Setup - Busy hour - RRC Connection set-up Quality Analysis EQuality Analysis ECC/I/I00 (all calls, RT, NRT)(all calls, RT, NRT) RAN

analyzing

Reference RAN Parameter range for good/acceptable quality: Ec/Io = -3 dB …-8 dB

Busy hoursIn general, during RRC connection request statistics describes acceptable median Ec/Io conditions. Some cells show a median Ec/Io below the good quality threshold of -8 dB. The worst cells for all calls are (both cell ID and SC is indicated):

EcEc/Io of cells /Io of cells for NRT less for NRT less good then RTgood then RT

Number of cells versus median Ec/Io, for NRT calls


Radio Performance Optimisation Monitoring during SHO (Event 1A – best active cell)

ConclusionsConclusions

•• PP--RACH RACH -- EcEc/Io/Ioanalyzinganalyzing

PP--RACH:RACH:Out of Synchronization, high UL RTWPOpen loop Power control, initial PwR, EcEc/Io offset /Io offset --20 dB20 dB(required SIR - operator parameter)Timer for P-RACH PwR rampingC1 (UL/DL in-balance - CCCH > PwR max MS)

Initial RACH power: CPICH Tx power – RSCP + RTWP + Required Received C/I DPCCH_Initial PwR: TxCPICH – CPICH_RSCP + RTWP + SIRDPCCH – 10 • log (SF DPCCH)

Initial RACH power = 33 – (-95 dBm) – 105 dBm – 17 = 6 dBm correct with normal UL- interference levelInitial RACH power = 33 – (-100 dBm) – 97 dBm – 17 = 19 dBm For PS NRT = too low coverage - too high PwR for RACH necessary, Pmax UE= 21 dBm.



own

other

powerRxTotalpowerRxTotali

____

ceinterferen cellown ceinterferen cellother

GISUE analyzing

CELL A

CELL B

CELL C

Little i map – from trouble shooting at Iub



Interference Analysis Little i (all calls, RT, NRT)Interference Analysis Little i (all calls, RT, NRT) RANanalyzing


Little i of cells Little i of cells for RTfor RT

Number of cells versus median little i, for RT calls

Number of cells versus median little i, for NRT calls

Little i of cells Little i of cells for NRT less for NRT less good then RTgood then RT

A huge fraction of the cells shows a median little i of 1 and higher (this means, that the adjacent cell interference exceeds the own cell one).

The strongest interfered cells during RT call setup with median little i > 1 are …



Propagation Delay (all calls, RT, NRT Propagation Delay (all calls, RT, NRT -- call setup versus distance)call setup versus distance) RANanalyzing


Number of cells versus median propagation delay, for RT calls

PD distribution PD distribution of cells of cells max 7.5 km/RTmax 7.5 km/RT

For RT calls, the median propagation delay ranges typically from 6-24 chips, corresponding to about 0.5 to 1.9 km. Some cells, however, show a very large median extension of up to 96 chips (about 7.5 km).

Number of cells versus median propagation delay, for NRT calls

PD distribution of cells PD distribution of cells max 5.5 km/NRTmax 5.5 km/NRT

For NRT a higher median propagation delay is indicated than for RT calls, typically ranging from 9-30 chips.

Much more cells now show a median propagation delay of 18 chips and higher. The more distance access explains the higher adjacent cell interference for NRT calls.




•• Little iLittle i•• PDPD

analyzinganalyzingLittle i is too muchLittle i is too much

More Antenna tiltingLess CPICH overshootingChange/downgrade service request on cell edge for NRT

Cell range: Cell range: Some cells show a very large median extension of up to 96 chips (about 7.5 km).

The more distance access explains the higher adjacent cell interference for NRT calls.



SHO Window (all calls, RT, NRT) Cell Matrix RSCP (all calls, RT, NRT)

RANanalyzing

Monitoring during SHO (Event 1A – best active cell)

RNSAP: RL Setup RequestRRC: Measurement Report

RNSAP: RL Setup Response

NBAP: RL Setup

NBAP: RL Setup Response

UP FP: Downlink Synch.

UP FP: Uplink Synch.

RRC: Active Set Update (RL Addition, Deletion, Replacement)

RRC: Active Set Update Complete

NewNode B

Drift Serving

Event 1A,Cevent-triggered

DCH DL RLC AMD rrcMeasurementControl

RNCRNC RNCRNC

Iub Bearer Setup Iub Bearer Setup



SHO addition Window E1A is too late triggedSHO addition Window E1A is too late trigged ::

• UE is often released, due to too late E1A and CDR goes up • Low add. Window offset < 4 dB (serving cell – adjacent cell) reduce SHO

overhead KPI – at least no overlap of cells – coverage gaps are produced.

CELL ANeighbour

CELL BServing

Ec/Io/RSCP

Too late trigger from cell AToo late trigger from cell A-- BB

SHO areaSHO area

• The consequence of the delayed SHO process decrease the coverage area.

+ 4 dB offset

RANanalyzing


Ec/Io/RSCP

SHO Window (all calls, RT, NRT)


Radio Performance Optimisation RAN

analyzingMonitoring during SHO (Event 1A – best active cell)

For RT calls the SHO performance is consistent with an addition window of 4 dBaddition window of 4 dB. In most cells the median EC/I0 difference between server and best neighbor indicated by the first event 1A report is not much lower than this threshold. The SHO process is initiated in time, when the server still is clearly dominating.

Number of cells versus median event 1a window, for RT calls

Too early active set updateToo early active set update–– enlarge SHO overheadenlarge SHO overhead

Too late active updateToo late active update–– too small SHO overheadtoo small SHO overhead



Radio Performance Optimisation RAN

analyzingMonitoring during SHO (Event 1A – best active cell)

Too late active updateToo late active update–– too small SHO overheadtoo small SHO overhead

Number of cells versus median event 1a window, for NRT calls

Too early active set updateToo early active set update–– enlarge SHO overheadenlarge SHO overhead

For NRT calls the first event 1A report indicates a lower median qualitylower median qualitydifference between server and best neighbor in comparison to RT calls. The SHO process is initiated with delay, the dominance of the server is already weak in comparison to the neighbor.

Very late active set update of controllable service comparing wiVery late active set update of controllable service comparing with RTth RT




SHO overhead - trace results at NW/cell layer (example)

HighHighInterferenceInterference

lowlowInterferenceInterferenceE1A in small

SHO areas, byhigh interference

Most significant causes at Cell layer e.g neighbor cell 18dB strMost significant causes at Cell layer e.g neighbor cell 18dB stronger than serving cellonger than serving cell

Release margin

Coverage holes/no cell overlap),averaging of filter coefficient,

caused E1A trigger

If SHO E1A window < 4 dB If SHO E1A window < 4 dB –– than too than too early E1A trigger, small SHO areaearly E1A trigger, small SHO area

Too late E1A trigger, big Too late E1A trigger, big ““SHO/power SHO/power overlapoverlap”” area area -- Release margin will drop UERelease margin will drop UE

RANanalyzing





•• Event 1AEvent 1Aanalyzinganalyzing

Results:In most cells the median Ec/Io difference between server and best neighbor indicated by the first event 1A report is not much lower than recommended threshold 2 dB…3 dB. But please change shift Ec/Io filter coefficient from 600 ms to 400 ms.


Radio Performance Optimisation Cell matrix evaluations and improvements (description)

Cell Matrix report show adjacent cell Cell Matrix report show adjacent cell properties in particular serving cell:properties in particular serving cell:

• N (number) - Serving cell visibility = Number of 1a reports indicating serving cell

• ADJ N (number) - Adjacent cell visibility = Number of 1a reports indicating adjacent cell

• ADJ WIN_E1A - Average adjacent cell window offset (dB)

• ADJ i - Adjacent cell Little i * visibility• INTRA - Adjacent DL SC (cell) is from same

site (Node B)

(CPICH SC)

Visibility =Visibility =Number of reports indicating cell n /Total number of reports

Example to add a cellExample to add a cell –to become active set enhancements,e.g. 153 times of E1a reports indicating in ADJ 1a/SC 25 (adjacent cell update) and the addition window offset is measured only by 1.24 dB, recommended parameter 4 dB – too late access for new cell

RANanalyzing Cell Matrix



Results:As example the cell matrix of cell 14683 is shown - see cell matrix analysis.

We consider too long list of neighbors.CPICH over-shooting

Monitoring during SHO (Event 1A – best active cell) Cell Matrix ConclusionsConclusions

•• Cell matrix Cell matrix analyzinganalyzing


Radio Performance Optimisation Monitoring of RSCP (all calls, RT, NRT) RAN

analyzing

Busy Hours

Even in SHO, for RT calls in many cases an acceptable median coverage is indicated.

Number of cells versus median RSCP, for RT calls

Reference RAN Parameter range for good/acceptable coverage: RSCP = -85 dBm …-95 dBm

RSCP (all calls, RT, NRT)


Radio Performance Optimisation Monitoring of RSCP (all calls, RT, NRT) RAN

analyzing

Busy Hours

Even in SHO, for NRT calls the median coverage is lower than for RT ones, much more cells now are below the acceptable RSCP threshold of -95 dBm. This is a consequence of the delayed SHO process.

Number of cells versus median RSCP, for NRT calls

Reference RAN Parameter range for good/acceptable coverage: RSCP = -85 dBm …-95 dBm

RSCP (all calls, RT, NRT)



PwR PwR ((dBmdBm))

distancedistance

Ec/Io

Ec/Io

RSCPRSCP

RSSIRSSI

GISUE analyzing

RSCP map – from trouble shooting at Iub



Conclusion:Shift filter values for E1A Ec/Io requirements (see also chapter SHO) This will improve RSCP (coverage due to MDC/MRC & SHO gain)

DYLO already impacts too early NRT service activation at cell edge (feature) – down grades service request at cell edge – decrease coverage & improve capacity, but due to high little i values (RT > 1) especially for NRT 1 …2, we consider too much overshooting of adjacencies at cell level power.

Summery:Summery:Small SHO areas (too late E1A activation) and high neighbor cell PwR overshooting (little i for NRT) decrease services, therefore total interference will increase and coverage will decrease but DYLO can not compensate all negative impacts of interference.


•• Event 1A/ Event 1A/ coverage RSCPcoverage RSCP

PtxRL

Ptxave

Triggering of DyLO

Ptxmax

Offset

Time/distance

Ptx

Ptxave

Triggering of DyLO

Ptxmax

Offset

Too lateE1a

Too early entering trigger offset thresholdToo early entering trigger offset threshold



Monitoring of load Transmitted Carrier Power Received Total Wideband Power

RANanalyzing

Monitoring of load

RNSAP: RL Setup RequestRRC: Measurement Report

RNSAP: RL Setup Response

NBAP: RL Setup

NBAP: RL Setup Response

UP FP: Downlink Synch.

UP FP: Uplink Synch.

RRC: Active Set Update (RL Addition, Deletion, Replacement)

RRC: Active Set Update Complete

NewNode B

Drift Serving

Event 1A,Cevent-triggered

DCH DL RLC AMD rrcMeasurementControl

RNCRNC RNCRNC

Iub Bearer Setup Iub Bearer Setup



Ptx Target [dBm]

load ()

total transmitted power Ptx Total area [dBm]

Overloaded area for 20 W

More feasible load and

coverage with more PwR …

but much more interference in received part

DL performance (Tx PwR)

Ptx Target [dBm] + PtxOffset [dB]

More visible capacity and coverage, but more inter cell interferMore visible capacity and coverage, but more inter cell interferenceence

Causes of DL interference:Causes of DL interference:• Too much common Ch PwR• User down link service allocation• TX IMP (3rd order)• CPICH over shouting

Marginal Load Area 1

CCCH

37.5 dBm

41dBm

40 dBm

Fixed PwR.

capacity

43 dBm

Decrease power to perform more capacity by

less neighbour interference in RSSI part

max,_

_

BTStx

totaltxDL P

P

max,_

_

BTStx

totaltxDL P

PMore coverage

RANanalyzing Transmitted Carrier Power

Monitoring of load



Transmitted Carrier PowerMonitoring of load RAN

analyzing

Busy HoursThe median TCP within the network is low. The highest load is indicated for cell 10873, with a median TCP of 27.5 % of the maximum cell power. The median TCP, however, is not very informative, as the load undergoes very strong short term fluctuations. On the other side the raw TCP measurements indicate, that even the cell with highest load seldom is in overload state.

Raw TCP versus time for cell 10873



Transmitted Carrier PowerMonitoring of load

Results/conclusion:No congestions due to low overload situations.The highest load is indicated for cell 10873 (< 30% median max. load)


•• DL DL TxTx PwRPwRanalyzinganalyzing



• High interference level due to IMP and traffic load limits service and throughput in the cell per user

• It limits also coverage area of cells

LRT UnloadedRT (2%) and LNRT UnloadedNRT(1%) Sum (3%)Sum (3%)

PwR rise PwR rise (dBm)(dBm)

loadload

I own cellI own cell

Thermal noise Thermal noise --108 dBm + Io108 dBm + Io

PrxOffset

PrxTarget

60 %60 %3%3%

I other cells I other cells

25 %25 % 30 %30 %

RANanalyzing

Received Total Wideband PowerMonitoring of load


Radio Performance Optimisation Monitoring of load RAN

analyzing

Busy HoursThe median RTWP within the network is very high, ranking from about -97 to -95 dBm for all cells, which is far above the default overload threshold. There is also a considerable fluctuation, as shown here for cell 16822, the cell with highest instability.

Raw RTWP versus time for cell 16822

Received Total Wideband Power


Radio Performance Optimisation Monitoring of load


•• UL RTWP UL RTWP analyzinganalyzing

Received Total Wideband Power

Results/conclusion:Results/conclusion:

Too much interference within UL direction Little i within UL direction is too high and/or out of band IM (e.g. video cameras) RRC performance can suffer. In-balance of UL/DL path loss auto tuning saves capacity within cell, already enabled, but too much little i Auto tuning can perform high UL sensitivity and save need of services within

cells. But it will reduce cell coverage too Use GIS analysis to detect UE faults and interference causes


Radio Performance Optimisation Geo data map & user allocation (GIS)

Mobility and Events

UE specific

Broadcastpilot

Broadcastpilot

Movement of UE

t5

t4

t3

t2

t1

Evaluations by … Drive testing reduction Use probing, make analysis more quick Benchmark with OSS - RANPAR/KPI

GISUE analyzing


Radio Performance Optimisation Call trace – no simulations – from trouble shooting at Iub

Only about 50-100 m offset from correct course of the roads due to prediction error

GISUE analyzing


Radio Performance Optimisation Service trace process

3G RPO3G RPOserviceservice

•• GIS GIS -- profileprofileservice managementservice management

•• Cell Service Cell Service capability capability RNC benchmarkRNC benchmark

Agilent/TECHCOM successfully perform 3G RAN diagnostic systems.The new applications can optimize the radio & signalingperformance of RNC areas.

• GIS service management (PoP/PoS)• OSS counter /KPI improvements• RANPAR tuning and radio monitoring without drive test

Agilent 3G RAN troubleshooting application – using data mining database – to analyze high level KPI and drill down quickly to the root problems and RF optimization application from smartTECH


Radio Performance Optimisation Cell service capability

advanced service:advanced service:

Detecting of user service utilisation within cells Overall NW throughput performance detection Design of service MAP with data rate capability Traffic analysis within networks BLER optimization by RANPAR calibration Drive test substitution and fault detection


RNC analysis by SARTRNC analysis by SART


smartRANsmartRANRANPAR-libraryOSS/RANPAR

smartCountsmartCountKPI -libraryOSS/counter

••UTRAN benchmarkUTRAN benchmarkRANPAR,KPI,GISRANPAR,KPI,GIS




Protocol analysis SARTProtocol analysis SART

Detection of Detection of RANPAR valuesRANPAR values• Little i (PwR overlap)• PDU credit delivery• Cell orthogonality• RSCP, Ec/Io etc.

smartInfo smartInfo GIS/MAP & QoS•• Cell Service capability Cell Service capability

e.g.e.g. BLER/SIR BLER/SIR –– target improvementstarget improvements

Cell service capability

Cell/GIS MAP Cell/GIS MAP performanceperformance



CELL A

CELL B

CELL C

2MB/area 2

Cell Service capabilityCell Service capabilityCell throughput & capacity improvementCell throughput & capacity improvement

Vendor featureVendor feature•• EbEb/No flexible adjustable/No flexible adjustable•• ServiceService--specific, specific, •• QualityQuality--Target Target •• on Rxon Rx--DiversityDiversity•• Streaming Service on DCH or HSPAStreaming Service on DCH or HSPA•• Orthogonality adjustableOrthogonality adjustable

EbEb/No parameters can be configured by operator /No parameters can be configured by operator (parameter object). Final DCH (parameter object). Final DCH EbEb/No must be /No must be determined for determined for throughput & power estimationsthroughput & power estimations..

throughputthroughputperformanceperformance

0.5 MB/area 1

Cell service capability





••UE UE –– service managementservice managementGIS, PoP, PoSGIS, PoP, PoS

Always on by SART/DNA Always on by SART/DNA

Geo information system (GIS)

advanced service:advanced service:

User service providing platform 7/24 Google application without GPS/MS New services (security, Point of Present) New dedicated LBS flat rates Virtual shopping market placement


Radio Performance Optimisation Flexible - experienced - quick

We bring technologies and learning products in time to markets

Please contact:TECHCOM ConsultingD-82041 Oberhaching / Munich

phone: +49-89-638-488-0fax: +49-89-638-488-99

e mail: [email protected]

made by …

smar tech overview

Technology

techcom consulting

nrt propagation

nrtall calls

g ran workshop

distance monitoring

g ran related problems

additional agilent

distancecall setup