Part1_Fundamentals of Network Optimization[1]

NSN Internal DocumentFundamentals of Network Optimization/JP/NNPO/ 1st to 3rd July 2010 @ VF RoB1 © Nokia Siemens Networks

Fundamentals of Network OptimizationNPO Refresher CourseJuly, 1 st to 3 rd 2010Vodafone MS – RoB

Jignesh [email protected] Siemens NetworksNational NPO, Ahmedabad, India


Maximize network quality :• as perceived by the subscriber• as perceived by the operator within the given scope

Balance investment versus achieved quality

What is Network Optimization?


Need for Optimization

In most cases, the network will not run optimal from the very beginning due to the following reasons:

• Systematic errors of input data (e.g. wrong estimate of traffic)• Statistical nature of involved processes (e.g. average reflection / diffraction

coefficients instead of building material specific ones) – uncertainty of the statistical values used in the propagation models

• Dynamical nature of involved processes (e.g. increasing number of subscribers and change of their behavior, e.g. SMS traffic increases more than voice traffic)

• Installation errors (e.g. wrong cabling: transmitting into cell A, but receiving from cell B)

• Hardware / software trouble

• Configuration errors


How Network Optimization can be done?

Note! Case 1 and 2, Same quality improvements over the pe riod

Optimization projects=> quality is improved after projects

Quality trend

Quality trend

quality

time

Case1:Optimization projects to improve quality

Case2:Monitoring =>all time optimization (part of network planning)


What is Network Quality?

• Network quality can be measure• Interference level• Coverage• Drops• Drops / Erlang• MOS• Customer complaints

⇒There are many different ways to measure the quality of the network


OPERATOR

CUSTOMER

NETWORK

SERVICES

MOBILE

COST

• Mail Box, Data, Fax• HSCSD, GPRS, WAP• Customer Care

• Faulty H/W or S/W• Mobile Quality• Misuse of Equipment

• HW/SW Failures• Network Configuration• Network Traffic• Spectrum Efficiency

• Coverage yes/no• Service Probability• Quality• Call Set Up Time• Call Setup Success Rate• Call Completion Rate

• H/W Costs• Subscription/Airtime costs• Additional Services Costs

• Network Equipment Costs• Maintenance Costs• Site Leasing Costs• Transmission Link Costs

What is Network Quality?

OPERATOR

CUSTOMER

NETWORK

SERVICES

MOBILE

COST

• Mail Box, Data, Fax• HSCSD, GPRS, WAP• Customer Care

• Faulty H/W or S/W• Mobile Quality• Misuse of Equipment

• HW/SW Failures• Network Configuration• Network Traffic• Spectrum Efficiency

• Coverage yes/no• Service Probability• Quality• Call Set Up Time• Call Setup Success Rate• Call Completion Rate

• H/W Costs• Subscription/Airtime costs• Additional Services Costs

• Network Equipment Costs• Maintenance Costs• Site Leasing Costs• Transmission Link Costs

Main focus on Network


Optimization Targets

• Weight of quality targets changes in time• In green field the coverage is most important• Later capacity / quality / coverage are all most important

⇒Same with optimization, how to optimize to get the biggest benefit

• Which are the biggest bottlenecks in project

• Coverage, capacity or quality

QualityCoverage

Capacity

QualityCapacity

Coverage

Time


Optimisation needed to...

• Maximise/Maintain network Quality

Radio Network has to follow city developments

• Achieve Quality Criteria set by

Operator:

- Key Performance Indicators balance investment versus achieved quality

Subscriber:

- 95% coverage probability everywhere

• Quality targets are NOT set by network optimization engineers !




The targets are always defined by the customer but thefollowing figures can be considered as satisfactory results.

95 %> 98 %Good Quality samples (0..5)

96 %> 98 %Handover Success Rate

4 %< 2 %DroppedCalls

Lowest Acceptable

TargetItem


How Network Quality can be measured?

Customer complaints• Can detect problems in

certain regions that are masked by statistics.

• A bit too late!!

• These are consequences, not reasons

Field Tests•Geographically locate problems

•Typically downlink

• Snapshot

• Expensive and time consuming!

From Network Data•Allows centralized data collection

•A cost efficient way to monitor network quality

•Problems can be found very quickly, before customer complaints

Three main forms to monitor and detect the needs of optimization:


ToolsKPI

Documentation

Analyze

N

Case closed

YSolution worked ?

Solution

More info

needed ?

Y

N

HW change

Parameter change

Performance Analysis

O&MFT CC DB

Optimization Flow Chart


Young Network Case

In a young network the primary target is normally the coverageIn this phase usually there is a massive use of drive test measurement

• Check the signal and

• The performance of the competitors

GPS

NMSX

MMAC

Optimization Process


Mature Network Case

In a mature network the primary targets are quality indicators• Drop call rate, average quality, handover failures

Important use the information from OMC• A general view of the network performance

Drive test measurements are still used• But not in a massive way

• In areas where new sites are on air

• Where interference and similar problems are pointed out by OMC data analysis

Drop Ca ll Ra te (%)

0

0.5

1

1.5

2

2.5

3

3.5

Mon Tue Wed Thu Fri Sat Sun Mon Tue Wed

Call Bids / 10000

A verage

Busy Hour

Optimization Process


Operation & Maintenance

• Statistics• Alarms• Parameters (OMC/BSS)

Field Tests (NEMO Outdoor, TEMS, Agilent, Scanner, power meter, spectrum analyzer, transmitters)

• Drive tests• CW measurement• Customer Complaints

Database concistancy checks (NetAct)• Planning tool �� BSS/OMC

Optimization Inputs and Tools


Content: Performance Measurements

• OSS Statistics– Key performance indicators– Counters

• Drive Tests– Planning and execution– Analysis (example)


OSS Statistics

Statistics obtained by O&M system

• Cover a short or long time interval (e.g. one hour, one day, one week)

• Cover a small or large part of the network (e.g. single TRX, single cell, all cells of a

BSS)

• Are specified (GSM 12.04) or vendor specific

• Each time a certain events occurs (e.g. SDCCH / TCH request / allocation /

rejection), the corresponding counter is updated


Call processingRequirements

Call processingRequirements

UTRAN Formulas

UTRAN Formulas

UTRAN Counters

UTRAN Counters

GERANFormulas

GERANFormulas

GERANCounters

GERANCounters

PM Customer Requirements- Network Operation - Customer Care- Network Planning - Management- Network Optimization - Marketing

PM Customer Requirements- Network Operation - Customer Care- Network Planning - Management- Network Optimization - Marketing

Performance Indicators,QoS Parameters

Performance Indicators,QoS Parameters

CommonCommon SpecificSpecific- Measurements- Message flows- KPIs- Use cases- Traffic Engineering- Network Planning

- Measurements- Message flows- KPIs- Use cases- Traffic Engineering- Network Planning

StandardizationRequirements (GSM 12.04, CWTS, 3GPP

TS32.104)

StandardizationRequirements (GSM 12.04, CWTS, 3GPP

TS32.104)

OSS Statistics – KPI Concepts 1/2


OSS Statistics – KPIs Concept 2/2

KPIs are figures used to evaluate Network performance

• Post processing of NMS data or

• Drive test measurements data

Usually one short term target and one long term target

• Check the network evolution and which targets are achieved

KPIs calculated with NMS data

• Network performance on the operator side.

KPIs from drive test

• Performance on the subscribers side

Usually turn key projects are evaluated according to some predefined KPIs figures like drop call rate


The most reliable KPIs to evaluate the network performance with NMS are:

• SDCCH and TCH congestion• Blocking percentage [%]

• Drop call rate [%]

• Handover failure and/or success rate• Call setup success rate

• Average quality DL and UL

OSS Statistics – Performance Evaluation 1/2

With NMS


With Drive Tests

Evaluate network performance from the subscriber point of viewKPIs information

• DL quality, call success rate, handover success rate, DL signal level• Not statistically as reliable as NMS information

Added value of drive test measurement• Find out the geographical position of problems like bad DL quality to look

for a possible interference source in the area• Compare the performance of different networks (benchmarking)• Display the signal level on the digital maps to individuate areas with lack of

coverage eventually improve the propagation model • Verify the neighbour list parameter plan

OSS Statistics – Performance Evaluation 2/2


Percentage of refused Mobile Originated CallsMOC refused call percentage

Percentage of refused Mobile Terminated CallsMTC refused call percentage

Percentage of refused callsRefused call percentage

Percentage of dropped callsDrop call percentage

Percentage of blocked callsBlocking call percentage

Percentage of drops connections during a SDCCH connection

SDCCH drop rate

Rate of unsuccessful handovers without a loss of the MS

connection because of reversion to the old cell

Handover failure rate

Percentage of TCH drops related to all TCH connections

TCH drop rate

Percentage of successful Setup CallsCall Setup Successful Rate

Performance MeasurementKPI

OSS Statistics – KPIs

• Among others, some often considered KPIs:


CALL SUCCESS FACTORS (breakdown I)====================

• SDCCH access probability(before FCS) .......................................csf_1 99.61 %(after FCS) ........................................./csf_1a 99.61 %

• SDCCH success ratio(SDCCH fail based, incl.LU) ............./csf_2k 216.40 %(SDCCH to TCH based) .................../csf_2o 95.37 %

• TCH access probability(before DR and queuing) ................./csf_3m 118.70 %(before DR) ......................................./csf_3i 116.11 %(real) ................................................./csf_3p 99.83 %

• TCH success ratio (after seizure)(before re-est.) ................................/csf_41a 97.43 %(after re-est) ..................................../csf_42a 97.56 %

Note: See the formula descriptions for the accuracy limitations.Note: For a cell level list, run report 250.CALL SUCCESS FACTORS (breakdown II)====================

• TCH assignment probability .................…/100-blck_29 97.88 %• TCH success ratio after assignment ......../100-dcr_8h 99.04 %

OSS Statistics – KPIs – Examples


OSS Statistics – Counters – Examples

• SDCCH requests ................................./c1000 7967688

• HO in..................................................../c1006 28324 ( 0.36 %)

• blocked ............................../blck_15 (blck_5a) 30707 ( 0.39 %)

• To FACCH call setup .........................../c1099 0 ( 0.00 %)

• LU ......................................................../c3019 2918916 ( 36.63 %)

• MTC (incl. SMS) ................................../c3012 1996039 ( 25.05 %)

• MOC (incl. SMS,SS) ............................/c3013 2209808 ( 27.73 %)

• supplementary service request (S9) … /c3044 104573 ( 1.31 %)

• IMSI detach (S7) ................................../c3033 66587 ( 0.84 %)

• call re-establishment............................./c3020 3568 ( 0.04 %)

• emergency call ...................................../c3021 1521 ( 0.02 %)

• other (fails, ghosts) .............................../sd_1b 607623 ( 10.25 %)

Counters Formulas KPIs

lsNumTermCal

TCHDropteCallDropRa =e.g.


Alarms - Introduction

Alarm analysis is priority in worst cell troubleshootingBad performance may caused by HW problem would be listed in Alarm analysis.Check alarms on bad cells and its neighboursAlarms are transferred to NetAct database.MML command (describe in next slides), Nokia NetAct Top Level User Interface

Top-level User Interface in network monitoring is to show the alarm situation in all managed objects


Alarms – Groups

Alarm number in: Notices (NOT ICE)

Disturbance printouts (DIST UR)

Failure printouts

(ALARMS)

Diagnosis reports

(DIAGN)

Base station alarms

Transmission equipment

alarms

Numbers reserved for

possible external alarms

switching equipment 0–799 1000–17992000–2799 3000–3799

4000–4799

O&M equipment 800 - 899 1800–18992800–2899 3800–3899

4800–4899

transmission equipment

900 - 999 1900–19992900–2999 3900–3999

4900–4999

diagnosis report number

3700–3999

base station/ transmission equipment alarms

7000–7999 8000–8999

power equipment 5000–5499

external equipment 5500–5999

Alarms – Printout Fields• Alarm Number• BCF number, BTS number, TRX

number, Alarm Object, Unit, Date, Time, Alarm Number

• Urgency level(* Low Priority, ** Med Priority, *** High Priority)

• This is output in all alarm printouts except notices (NOTICE). The terminated alarms are indicated by dots (.) instead of asterisks (*).

• Printout type– ALARM fault situation – CANCEL fault terminated – DISTUR disturbance – NOTICE notice

• Event typeCOMM communication failure QUAL quality of service PROCES processing failure – EQUIPM equipment failure – ENVIR environmental failure


Alarms - Alarms in MMLAlarms in BSC Level:

– ZAHO: PRINT ALARMS CURRENTLY ON– ZAHP: PRINT ALARM HISTORY

Alarms in BTS Level: – ZEOL: LIST ALARMS CURRENTLY ON– ZEOH: LIST ALARM HISTORY

BCF, SEG or BTS configuration and status– ZEEI: OUTPUT RADIO NETWORK CONFIGURATION

TRX and RTSL configuration and status– ZERO: OUTPUT TRANSCEIVER PARAMETERS

HW Tests:• ZUBK: HANDLE ABIS LOOP TEST

(Parameters: BTS, TRX, RTSL, Fixed or Dynamic Abis connection and Abis TSL and Sub-TSL, looping time)

• ZUBS: START TRANSCEIVER TEST(Parameters: BTS, TRX, RTSL, test mode, RTSL,test selection, diversity path selection,test connection, RF test signal attenuation,BTS RX level,STM antenna attenuation,BS TX power attenuation, loop duration)

Verify if BCF, SEG, BTS, TRX and RTSL are LOCKED or UNLOCKED; WO (working), BL- USR (blocked by user) or Restarting.


Alarms - Examples

• 2993: BTS AND TC UNSYNCHRONIZATION CLEAR CALLS ON ABIS INTERFACE

– Transcoder and transmission alarm. Abis test is needed.

• 7045: TRX/FU DATA TRANSFER ERROR BETWEEN FU AND CU16– Base Station alarm. Observed when TCH failure rate is very high. A possible

solution can be to change the TRX.

• Example of a BTS alarm printout:


• Different KPI group can be seen here

• Example of final KPI report

Monitoring – Reporting Suite


Drive Test (1/3)

• Statistics (counters) help to find bad cells but the informationoffered are insufficient to localize problematic areas inside of one cell. The counters can offer information about the distance dispersion of the MSs with a best case resolution of 550 m.

• Field measurements give information with a resolution of less than 10 m about fading in the area, that can not be captured with the statistics.

• Field measurements offer valuable conclusions about the interaction between the cells.

• Field measurements are used in the optimization process but even in the planning phase e.g. to tune the propagation models.

• An often used type of the field measurements is the drive test.


Drive Test (2/3)

• Drive tests are performed by the network operator for various reasons :– To check the coverage in a certain area– To check the quality of service in a certain area– To find the answer for customer complaints– To realise that the network is not properly running– To verify that the network is properly running– To verify that certain optimisation steps have been successful– ...

• Drive tests must be well prepared. Before, during and after the drive test the following steps should be performed:


- Make back-up files of captured data- Replay captured data and analyse them- Find out problem areas and problem events- Use further post-processing tools to display the captured data more - clearly and to display further values graphically- Perform statistics and summarise the results

• After drive test

- Monitor test equipment- Reconnect dropped calls- Insert notes in recording file- Note interesting events separately

• During drive test

- Plan the route where to drive- Plan the time when to drive- Determine MS mode (idle mode/ connected mode) and also call strategy

(long / short calls) - Decide which values to focus on (e.g. RXQUAL, RXLEV, SQI...)- Select an appropriate test equipment and check it- Think of notes to be inserted later on in recording file

• Before drive test

Drive Tests (3/3)


Drive Tests – Example 1






• Parameter changes- Frequency changes- Neighbour changes etc...

• Hardware changes- Antenna tilt- Antenna orientation- Antenna changes- HW check- Adding/deleting cell- Adding/deleting site- Adding/deleting TRX

Optimization Solutionse.g. :


Parameter Optimization

• Cell Re-selection– C1– C2

• Power Control Triggered by– Receive level– Receive quality

• Handover Triggered by– Power budget– Graduate drop of receive level– Receive quality– Interference– Rapid drop of receive level

• Adjacency Plan– Number of adjacencies– Selection of best adjacency in case of HO


Physical Optimization

• Antenna– Mechanical and electrical tilt– Height above ground / roof– Orientation– Type

• Other Equipment– Booster (DL) and mast head / tower mounted amplifier (UL)– Feeder cable– Diversity antennas– Combiner, duplexer and diplexer


The antenna configuration is a key parameter of the optimization.

Has an impact on:

• coverage of the cell,

• co-channel interference and

• coverage holes (which can also be covered using repeaters).

It’s possible to change:

• the antenna tilt,

• azimuth

• height

• type

Antenna (1/2)


Altering antenna tilt:

• to reduce interference

• to limit coverage area

• to improve coverage (e.g. coverage weakness below main lobe)

• to improve in-building penetration

Altering the Antenna tilt must be done very carefully to really improve the situation.

Typical down-tilts are between 0° and 10°, however even higher values (up to 25°) have already been used.

Altering antenna azimuth:

• to overcome coverage weakness between different sectors

• to reduce interference in certain directions

Increasing or decreasing antenna height:

• to reduce or improve coverage

• to reduce interference

Change of antenna type

• to achieve desired ration characteristics

Antenna (2/2)


• If the link budget is not balanced (which leads to different coverage areas in uplink and downlink), there will be a bad overall performance of the cell.

• This situation can be overcome using the TMAs, which will improve the link balance, noise figure and sensitivity thus leading to a better performance.

Other Equipment


Other Equipment – TMA

Amplifiers help to balance the link UL & DL

Different coverage areas for UL and DL - too low RXLEV and/or RXQUAL at cell edge- many call drops and handover failures

• DL stronger than UL (usual case) � Add tower mounted amplifier to improve UL (see chap.2)

• UL stronger than DL � Add booster or reduce downtilt to improve DL


Other Equipment – Repeater

Repeater to improve coverage:

• Full duplex RF amplifier

• Receives, amplifies and retransmits downlink signal from donor base station and uplink signal from close mobile stations without any changes

• Overcomes local coverage problems (tunnels, valleys, in buildings etc.)

• Extends BTS coverage area


Let’s optimize!!!


Before starting the optimization …

Actual Network Status:• Status of the network, actual performance

Target Performance:• Defining the parameters that will serve to determinate the grade

of success of the optimisation• Can be based on Key Performance Indicators (KPIs) from

statistics and, optionally, on Drive Tests results.


Examples from real networkAutomated network performance optimization

Our challenge• To ensure good performing GSM network• Not to have penalty from CSP

Our solution• Automated measurement based optimization process :� Analyze the current performance in network

(mobility and interference)� Identify actions necessary to improve performance� Optimize accurately by using knowledge on current network configuration and network

performance and traffic measurements

Customer benefits• Accurate and repeatable optimization (automation)• Optimized usage of network resources • Minimized usage of available frequency spectrum


Optimization Example

Quality Definitions• The operator should define what is “quality” according to

its focus

To offer much coverage as possible

To maintain the best speech quality

Etc.

Quality?

Operator A wants to offer the best speech quality



Quality Target• Based on the quality definition the targets to achieve

should be defined accordingly, e.g:

DCR < x% (e.g 2%)

(% samples having RxQual > 4) < x

Call success setup ratio > x%

(% samples having FER > y) < x

Operator A sets:• DCR < 2%

• (% samples having FER >5%) < 5%



Quality Monitoring• The operator define the necessary KPIs to analyze the

performance of the network and assess the quality targets

Operator A has defined several KPIs:-DCR

-% cells having blocking >2%

-% samples having FER > 5%

The operator has several ways to obtain the reports:-Sometimes MML commands to the BSC are used (e.gservice terminal)

-NDW daily reports are also obtained

-Reporting Suit reports



Alarm Analysis• Faulty behaviors should be detected since they may

prevent the network from accomplishing the quality targets

• The basic monitoring tasks may include:

Alarm monitoringReal time traffic monitoring and troubleshooting

Operator A monitors the alarms and no critical alarms are found.

New parameters or functionalities have not been activated lately



Configuration Analysis

Operator A checks the parameters consistency and all the parameters are according to planned



Troubleshooting

Operator A uses several measurements and obtains:

•DCR close to the limit or even slightly above depending on the day

•DCR causes are mainly “tch_radio_fail” and “tch_rf_old_ho”. Thus radio reasons are causing the majority of the drops.

•RXLEV distribution shows:•No coverage problems (majority of samples have an acceptable value of RXLEV)

•RXLEVEL values in DL are higher than in UL. In UL still RXLEVEL values are quite good.

•RXQUAL distribution shows that the percentage of samples having bad RxQUAL is on the limit and significantly higher in DL than in UL (DL is the limited link due to interference)



Change RequestObjective: Reduce DCR

Reduce interference

DCR due to radio reasons (interference)

In UL the quality is much better than in DL and RXLEV values are good �There is still room to reduce power in UL

More aggressive UL PC settings

Enable “MS Power Optimization” feature



Quality Improvement

• After the changes have been made the next phase is to start monitoring the network to check if there is some negative impact and continue the “Network Optimization Cycle” again.

The second option is chosen to ensured that the transmission at maximum power is reduced as

much as possible.


Lastly …

• Parameters set during the planning phase must be reviewed according to network statistics

• As number of users increase, network expansions must be considered as well as new strategies

• Frequency plan may have to be altered in order to avoid interference and network quality degradation during network growth

Why optimize a network? Hasn’t everything been

done during planning phase?

NO!


Thank You

Documents

Part1_Fundamentals of Network Optimization[1]