Initial Tuning and Optimization

8/13/2019 Initial Tuning and Optimization

1/34

UMTS Initial Tuning


2/34

9-16-2012 | Page 2

UMTS Initial Tuning


3/34

9-16-2012 | Page 3

Are there any differences between 2G and 3G?

2G Initial Tuning 3G RAN-Tuning

New Operators Mature Operators, but not in 3G...

Mature System and UE NEW and sophisticated Systems features and Ue

Coverage focused Coverage & Capacity focus already in the initial network

Stable processes Modified/Enhanced processes

Stable tools New tools

Medium requirements on Project Management High requirements on Project Management

Over shooting sites are causing interference Overshooting and Neighbor sites are causing interference

Neighbor relations are basic Optimized Neighbor relations are key

Co-sited with GSM and Dual band antennas

Inter-working with GSM


4/34

9-16-2012 | Page 4

Typical Tuning Process Flow

Preparation

PreparationRootAnalysis

Analysis

Change

Proposals

ReportingChange

Verification

Implementation& Verification

PostProcessing

Measurement

Data

Collection


5/34

9-16-2012 | Page 5

Initial tuning overview

Main RF System performance indicators:

BLER (connection quality)

Access Failures

Dropped Calls Pilot Channel Ec/Io

Mobile Transmitted and Received Power

Forward/Reverse Link BLER

Call Processing Information

Service Measurements


6/34

9-16-2012 | Page 6

Initial tuning overview

Network tunning parameters

Neighbour list

Antenna height, tilt and orientation

Network parameters:

-Output power

- Cell Selection & Reselection

- Handover

- Power Control

- Other


7/349-16-2012 | Page 7

Test Equipment

Pilot Scanner

Scan pilots measuring Ec/Io and RSCP

Independent of system parameters (soft-HO, etc)

UEs: Mobile phones and PC Cards

GPS

Mobile Monitor software installed in Laptop

Mobile, GPS and Scanner connected to laptop

Collects all data measured and processed by Mobile:

- RSCP and Ec/Io of cells in Active Set+Neighbour Set

-Mobile Tx and Rx power

- L2 and L3 messaging

- Call processing

- BLER

-Service Measurements


8/349-16-2012 | Page 8

USB2USB1

Com 1

scanner

Short call Long Call

USB1

CS64

GPS

Drive Test Equipments for Voice, CS64 & PS call

Drive test tools configuration

USB2

PS


9/349-16-2012 | Page 9

Data Analysis

Post processing data analysis software Generate plots, table, show L3 messaging

TEMS (Ericsson), X-Cal (Couei), etc

GIS Software

To display graphical results from coverage surveys overlaidon street maps

MAPINFO, ARCVIEW, etc

RF Planning tool

Planning and evaluation of modifications

TEMS CPU, Asset, Atoll, etc


10/34

9-16-2012 | Page 10

Initial tuning procedure

1. Spectrum clearance Check external interference

2. Sector Tests

Basic call processing functionality checking

3. Unloaded Cluster optimization

Group of 15-20 NodeBs (3 rings)

Clusters should overlapp

Detection of main problems and basic adjustment

4. Complete system-wide optimization

Optimization under loaded conditions (OCNS) Adjustement of Inter-cluster areas

Fine tuning to reach KPI objectives


11/34

9-16-2012 | Page 11

Initial tuning Steps

First step: check UMTS Spectrum Possible sources of external interference:

Microwave data transmissions

Wideband noise from machinery

Monitor Uplink and Downlink interference in the UMTS band


12/34

9-16-2012 | Page 12


Second Step: Site testing Test of basic call processing functions: origination, termination of

Call, Videocall, PDP context, Data transmissions

Test of softer handover

Check correct power settings

Check correct PSCs of cells

Check antenna orientation and footprint


13/34

9-16-2012 | Page 13


Unloaded Cluster optimization Collect basic site parameters: coordinates, antenna height,

orientations and tilts, PSC.

Check design with RF planning tool

Visit sites Verify SC plan (1 cell cant have 2 neighbours with same PSC)

Check Cell parameters (LAC, power settings, HO, etc..)

Neighbour list definition

Define extensive drive test

Check O&M alarms before drive test


14/34

9-16-2012 | Page 14


Unloaded Cluster optimization First Pass Optimization

Neighbour list correction

Detect low coverage areas

Detect areas with Pilot Polution

Plot and check coverage of each cell

Aerial modifications: antenna height, azimuth, tilt.

Useful plots:

- UE Rx and Tx power

- Dominant PSC, RSCP and Ec/Io

- BLER

- Scanner Top1 PSC, RSCP, Ec/Io

- Call Events (Drop, Fail, etc)


15/34

9-16-2012 | Page 15


Unloaded Cluster optimization Second Pass Optimization

Same as 1st pass:

- Neighbour list correction

-Detect low coverage areas

- Detect areas with Pilot Polution

- Check coverage of each sector

- Aerial modifications: antenna height, azimuth, tilt.

Adjustment of power settings (CPICH power)

Adjustment of HO parameters:

- Soft-HO: addition, deletion, max active set size

- Hard HO GSMUMTS parameters

Exhaustive Call Events analysis


16/34

9-16-2012 | Page 16


Complete system-wide optimization Neighbour list correction in inter-cluster areas

Load simulation in downlink and uplink

Under loaded conditions additional tuning is required

acceptable areas without load may suffer from interference

under load

Fine tuning (aerial modifications, parameters)

Exhaustive Call Event Analysis for KPI calculation


17/34

9-16-2012 | Page 17

What can be achieved with a scanner?

The basic measurements of scanner is CPICH_RSCP (received signal code power)

CPICH_Ec/No (received energy per chip divided by the

power density in the band)

RSSI (received signal strength indicator)

What can you detect with scanner?

Crossed feeder issues (DL)

Coverage verification

Interference problems (cell overlap, pilot pollution)

Missing neighbours

Detect overshooters


18/34

9-16-2012 | Page 18

COVERAGE VERIFICATION - Primary Common Pilot

Channel (Example)

Verify P-CPICH detection to minimize coverage holes

P-CPICH RSCP

P-CPICH Ec/No

Verify coverage predictions

Use threshold events and/or coverage maps to detect coverage holes.

Use Best Server Indication to identify interferer


19/34

9-16-2012 | Page 19

Best Server Signal Strength (RSCP)

Green is good

Yellow can

generate problems

Brown is bad

RSCP in dBm

A

B

D

CProblems occur

in areas with

bad coverage


20/34

9-16-2012 | Page 20

Pilot Ec/No measurements (Example)

Green is good

Yellow can

generate problems

Brown is bad

Ec/No in dB

A

B

D

CProblem areas

-9 to 0

-12 to10

-30 to -13


21/34

9-16-2012 | Page 21

Finding interference (Example)

By correlating low Ec/No

with high RSCP, areas

with high interference

can be detected

-100

-90

-80

-70

-60

-50

-40

-30

-25 -20 -15 -10 -5 0 5

Ec/Io [dB]

RSCP[dBm]

High interference

nenenee

amesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesamesames

B3031

B4009

A4

A3

A4155

A329

Reading StationReading StationReading StationReading StationReading StationReading StationReading StationReading StationReading Station

est Stationest Stationest Stationest Stationest Stationest Stationest Stationest Stationest Station

ReadingReadingReadingReadingReadingReadingReadingReadingReading

Museum of ReadingMuseum of ReadingMuseum of ReadingMuseum of ReadingMuseum of ReadingMuseum of ReadingMuseum of ReadingMuseum of ReadingMuseum of Reading

The University ofThe University ofThe University ofThe University ofThe University ofThe University oThe University oThe University oThe University of

1408

1685

23400 23

23405

25003

6734

6741

RSCP [dBm]

-50 to -41,6 (100)-60 to -50 (357)-70 to -60 (598)-80 to -70 (376)-90 to -80 (70)


22/34

9-16-2012 | Page 22

2nd priority

RN aspects

Initial Tuning experiences

Parameter consistency

Consistent network

configuration

O&M aspects (CM)

Securing coverage

Identifying interference

Tuning neighbor relations

Detecting crossed

feeders

Measurement principles

LA/RA planning

Power settings Correct code plan

Feature tuning

1st priority

3rd priority


23/34

9-16-2012 | Page 23

0%

2%

4%

6%

8%

10%

12%14%

16%

18%

20%

-16 -14 -12 -10 -8 -6 -4

Pilot Ec/No

%binsatgivenE

c/No

0%

10%

20%

30%

40%

50%

60%70%

80%

90%

100%

Cumulativedistribution

Drive test to verify p ilot RSCP and Ec/No

Securing CoverageChecking UL and DL service coverage

Verify RSCP Verify Ec/No

1st priority

Actu al DL service coverage based on Ec/No measurments

RAB

Required

Ec/No

unloaded

Required

Ec/No

With load

Required Ec

unloaded

Required Ec

With load

AMR 12.2 -16 -12 -114 -111

P64/64 -14 -10 -110 -107

P64/128 -12 -8 -110 -107

P64/384 -9 -5 -110 -107

Values depend o n

CPICH sett ing and

antenna conf igurat ion

384 kbps: 80%128 kbp s: 95%


24/34

9-16-2012 | Page 24

Identifying interference

In case of low RSCP:

Up tilt/re-direct antenna

Build new site

A pilot power increase will only help

marginally

In case of low Ec/No (and high

RSCP) down-tilt/redirect antenna

if necessary turn off interfering sector

Reducing pilot power is not a

preferred solution

1st priority

Potentially, high interference can be due to co-code

problems

Make sure to utilize all 512 codes

10090807060504030

25 20 15 10 5 0Ec/N0[dB]

RSCP

dBm

Interference limited Both design criteria

Coverage limitedCoverage and interference

RSCP req

outdoor

Minimum

quality


25/34

9-16-2012 | Page 25

Tuning neighbour definitionsConsequences of missing/too many neighbour definitions

Good neighbour definitions are

essential

A missing neighbour relation

can make call drop

Common reasons for poor non-working neighbours

Neighbour has not been

defined=> missing neighbour

The neighbour list has been

truncated

best server

cell not in active set

Pilot Ec/No UE released

releaseConnOffset [12 dB]

1st priority


26/34

9-16-2012 | Page 26

Tuning neighbour definitionsEffect of truncation

Active set (AS)

max 32 cells

Monitored set is the union of the

Ncell list of the cells in AS.

Note: Duplicated Ncell entities are

removed

Cell A

Cell B

Cell C

Cell A1

Cell B1

Cell C1

Cell A10

Cell B10

Cell X

Cell Y

Cell Z

Monitored set

Unmonitored set not measured by all UE

Neighbours are truncated

Avoiding truncation is important to keep drop rate low

A healthy network has not more than 15-20 relations per cell

1st priority


27/34

9-16-2012 | Page 27

Crossed feeder detectionCommon scenarios

DL

UL

Reference case

DL

UL

No coverage, loss of UL diversity

Low signal level, poor Ec/NoDropped calls

Handover fails

Call set up problems

High UL power

Swapped TX&RX

By monitoring Code, signal level, quality

and UE performance (handover, setup success,

power) crossed feeder can be detected.

RX2

TX/RX1

DL

UL

Swapped TX (or Swapped RX)

Loss of UL diversity High UL power

2nd priority


28/34

9-16-2012 | Page 28

Feature tuningThings to start with

HSDPAPurposeTo ensure high user bitrate and high HS

utilisation

ParametershsHysteresis1d

hsTimetoTrigger1d

hsdschInactivityTimer

other HS parameters

Packet dataPurpose

To secure a good trade off between

available resources and PS performance

Parameters

downswitchtimer (channel switching)

SF8 (number of 384 users per cell)

BLER target

IRATPurpose

To secure a well working mobility/prioritization

between WCDMA and GSM network

Parameters

UsedFreqThres2dEcno

UsedFreqThres2dRscpgsmThresh3a

hoType

idle mode and GSM parameters

3rd priority


29/34

9-16-2012 | Page 29

Trouble shooting: Call Events Analysis

1. Classify the reason of dropped and failed callsbetween

a) Radio related cases

b) Network related cases

c) other related cases

2. Drops and Fails due to Radio related cases to befixed:

Missing Neighbours

Pilot polution

Low coverage (if this can be improved)

3. Network issues are transmitted to the UTRANequipment supplier for system investigation

4. Other issues types are being classified andinvestigated (TMA problems, Test equipmentmalfunction, etc)

RF P f T l L3 ll i


30/34

9-16-2012 | Page 30

RF Performance Tools : L3 call processing

messaging

[ 2004 Apr 14 03:09:55.445 ] DL DCCH

log_chan_type : 3

rb_id : 3

length : 6

value DL-DCCH-Message ::=

message downlinkDirectTransfer : r3 : downlinkDirectTransfer-r3

rrc-TransactionIdentifier 1,

cn-DomainIdentity cs-domain,

nas-Message '052265'H

mobility management messages

===> mobility management messages

===> Connection management messages:CM SERVICE REJECT

Reject cause value : Message not

compatible with the protocol state

D C ll R


31/34

9-16-2012 | Page 31

Drop Call Reasons

Example

Drop calls reasons Type of drop

Missing neighbour Radio Issue

Good RSCP, high Ec/IoPilot polution Radio IssueGood RF but high BLER, release cause

unspecified

Network issue

Test equipment malfunction Other issue

Out of cluster Other issue


32/34

9-16-2012 | Page 32

Call Events Analysis : Drops and Fails

Correct classification is not always obvious: for a correctAnalysis a variety of RF parameters need to be takeninto account:

CPICH RSCP and Ec/Io of detected cells measuredby UE and Scanner

BLER DL (Drive Test Equipment) BLER UL (Traces)

UE Tx power

L3 messaging

UL Interference (Uplink measurements)

Etc


33/34

9-16-2012 | Page 33

Summary RAN Tuning

Ensure that there is coverage

All sites are in place correctly installed with correct power settings

Make sure to minimize missing neighbours

Use techniques such as NCS, counters, drive tests

Keep a consistent radio network

To ensure that same performance in all parts of the network

Selected feature tuning when there is significant amount of traffic

Current recommended settings has proven to give good performance equalto GSM

Eliminate over shooters Well defined cell coverage

Smooth transition to GSM

RND have impact on RAN tuning

Efficient RAN tuning => Shorter TTM


34/34

For more details

[email protected]

Documents

Initial Tuning and Optimization