Testing of GSM network
[email protected]
Omnitele Ltd. 2003 1
Contents η
GSM network quality audit
η
Network quality benchmarking
η
Optimisation
η
Statistics / Operation & Maintenance measurements
η
Field measurements
η
GPRS measurements
η
Testing of special features
Omnitele Ltd. 2003 2
GSM Radio Network Quality Audit and Optimisation
Omnitele Ltd. 2003 3
Background • On what basis does the customer rate GSM operatorÅs quality ? _ _ _ _ _ _
Prices Coverage area Call blocking/dropping Speech quality Customer service Else?
Radio Radio network network quality quality related! related!
• The user experienced service quality in GSM links directly to the performance of the radio network • Differentiation from competitors
Omnitele Ltd. 2003 4
Optimisation outcome ? • Network optimisation is a tradeoff between quality, traffic/revenues and investments. • Without fine-tuned network the customer complaints are increas work load is increased and marketing becomes inefficient GSM network indicators of European operator customer
Do you know how your network is performing ?
Direct impact on customer satisfaction ! Omnitele Ltd. 2003 5
Network quality benchmarking How does your networkÅs quality match against the competitors ? Åand against the world standard ? Quality Survey / Example Urban area benchmarking statistics
MTC Competitor Operator 1 (+1m) Operator 2 (+1m) Operator 3 (< MTC) Operator 4 (< MTC) Operator 5 (< MTC)
Omnitele Ltd. 2003 6
Call success rate 95,70 % 96,80 % 97,90 % 96,90 % 95,40 % 95,20 % 95,20 %
Handover DL success signal rate quality 97,90 % 99,00 % 99,20 % 95,50 % 96,10 % 95,00 % 99,10 %
0,4 0,4 0,2 0,5 0,6 0,3 0,5
MS TX Handovers/ SQI DL signal MS TX power level power class power class call 900 1800 -67.1 dBm -61.6 dBm -
9,6 5,5 8,7 7,2 7,8 8,7 7,3
5,0 3,5 1,8 1,5 N/A N/A 1,4
3,6 3,5 2,5 1,4 3,1 1,7 2,4
28,1 28,8 29,1 28,2 27,3 28,9 28,8
Field measurements • Benchmarking • Troubleshooting _ statistics, customer complaints
• Base station start-up • Testing equipment _ _ _ _
Testing software e.g. TEMS Ericsson in a laptop Test mobile phones (one or more) Indoor/outdoor antenna Cables + battery chargers
Omnitele Ltd. 2003 7
Field measurements • Testing route _ Roads, train _ Hot spot, pedestrian _ Urban, suburban, rural
• Test setup _ Idle mode _ Continuous call _ Call sequence (90s calls / 15s idle)
• Tested frequencies: 900/900E/1800
Omnitele Ltd. 2003 8
Example _ Call Success Rate 99,0 %
GSM1800 large European GSM900/1800 large European GSM900/1800 large European
98,0 %
97,0 %
GSM900 small Middle East GSM900 small Middle East
96,0 %
GSM900/1800 large European 95,0 %
GSM900/1800 large European GSM900 very small European
94,0 %
GSM900 small European
93,0 % Call success rate
Omnitele Ltd. 2003 9
GSM900/1800 very small Middle East
Example _ SQI 35
GSM1800 large European
30
GSM900/1800 large European GSM900/1800 large European
25
GSM900 small Middle East 20
GSM900 small Middle East
15
GSM900/1800 large European
10
GSM900/1800 large European GSM900 very small European
5
GSM900 small European
0 SQI (max 30)
Omnitele Ltd. 2003 10
GSM900/1800 very small Middle East
Network quality benchmarking • Could there be an indicator that shows your network is underperforming ?
Independent benchmark needed • What kind of action should be taken to improve the situation ?
Correctly targeted network optimisation
Quality Survey / Example Urban area
benchmarking statistics MTC
Competitor Operator 1 (+1m) Operator 2 (+1m) Operator 3 (< MTC) Operator 4 (< MTC) Operator 5 (< MTC)
Omnitele Ltd. 2003 11
Call success rate
95,70 % 96,80 % 97,90 % 96,90 % 95,40 % 95,20 % 95,20 %
Handover DL success signal rate quality 97,90 % 99,00 % 99,20 % 95,50 % 96,10 % 95,00 % 99,10 %
0,4 0,4 0,2 0,5 0,6 0,3 0,5
DL signal MS TX MS TX Handovers/ SQI power level power class power class call 900 1800 -67.1 dBm -61.6 dBm -
9,6 5,5 8,7 7,2 7,8 8,7 7,3
5,0 3,5 1,8 1,5 N/A N/A 1,4
3,6 3,5 2,5 1,4 3,1 1,7 2,4
28,1 28,8 29,1 28,2 27,3 28,9 28,8
Network optimisation process Customer Feedback
INPUTS
MANAGEMENT MARKETING
Benchmarking
NETWORK PLANNING Field measurements and performance evaluations Operation & Maintenance Measurements
Omnitele Ltd. 2003 12
Postprocessing
CUSTOMER SERVICE
Continuous process!
Optimisation targets
Satisfied customers
Maximum utilisation of hardware
Omnitele Ltd. 2003 13
Effective and economic use of network !
Troubleshooting OmniteleÅs engineers are experienced to encounte the major issues effecting the quality of GSM netwo • • • • • • • • • •
Handover problems Quality problems Coverage problems & New site location identification Network layers fine-tuning Capacity problematics LAC optimisation New feature implementation (software and hardware) Micro cells introduction Indoor coverage fine-tuning BSS parameter fine-tuning
Omnitele Ltd. 2003 14
Dual Band considerations Dual Band Strategy ¥ Spectrum requirements ¥ Network evolution from traffic hot spots to seamless coverage ¥ 900/1800 spectrum for indoor / outdoor, interference reduction, data ¥ Capacity gain calculations for different scenarios
Utilisation • Traffic steering between 900/1800 layers • Parameter optimisation • Coverage/capacity • GPRS strategy
Omnitele Ltd. 2003 15
Planning and Implementation ¥ ¥ ¥ ¥ ¥
Planning methods BSS parameters Site suitability criteria Sharing of RF components Planning / measurement tool requirements
Frequency planning & interference analysis • Interference analysis
• Frequency planning objectives
_
Used by turns with the
QoS objectives allocation in order to validate (and optimize, if _ Overall C/Ic and C/Ia requirements necessary) the frequency _ Cell-to-cell C/I requirements (reuse) plan • Interference check per _
channel 5,13 13,1 6,13
13,8
_ Evaluation of the significance of the residual interference _ HO/Cell selection margin should be included in the analysis _ C /Ic (and C/Ia ) statistics per channel
• Interference per network
Omnitele Ltd. 2003 16
_ On the dominance area _ C /Ic (and C/Ia ) statistics _ Shows the overall interference picture
Interference reduction means ¥ Optimising is done frequency by frequency ¥ Antenna redirection
Ð the most predictable Ð coverage is in danger
¥ Antenna tilting Ð ideally very useful: steepens the slope Ð in practice difficult to predict è should be always measured Ð useful for large clearence angles
¥ Power reduction Ð UL/DL interference power balance is lost Ð coverage is lost Ð not recommended
Omnitele Ltd. 2003 17
• All interference reduction means may generate problems in other frequencies
Reducing interference Improves network quality!
Optimising adjacencies & LAC • Missing neighbours very often result in unnecessary dropped calls and bad quality • Correct neighbour relations can be determined by analysing measurement results Unnecessary LAC updates can easily increase the singalling load significantly Increased signalling reduces the room for payload
Are your adjacencies and LACs optimised ? Omnitele Ltd. 2003 18
Optimising parameters study theoretically the impact of the intended change select the set of performance parameters and test cells change the cell parameters under optimisation measure with a test MS happy with the results ? + measure the performance and compare to the ref. performance happy with the results ? + apply the changes permanently in the network/service area Omnitele Ltd. 2003 19
new conclusions
O&M based performance parameters • QOS as well as RF and capacity planning parameters assessed per cell, per BSC and per network • RF planning parameters • QOS parameters _ Dropped call rate _ Dropped call rate due to radio _ Cumulative UL/DL quality statistics _ TCH BH blocking rate _ SDCCH BH blocking rate _ (Call setup success rate) _ (Handover success rate)
_ Number of calls _ TCH RF loss rate _ SDCCH RF loss rate _ TCH mean holding time _ Handovers per call _ Cumulative UL/DL level statistics _ Idle channel UL interference _ Power balance
• Capacity planning parameters _ TCH BH traffic (e.g. weekly and average of daily BH traffics) _ BH activity per subscriber (segment) _ Total TCH time (per subscriber) _ SDCCH weekly BH traffic _ BH paging load Omnitele Ltd. 2003 20
Example _ BH traffic
20 June: 290 000 subs ⇒ 17 mErl/subscriber Mean holding time ? -> BHCA/subscriber
Omnitele Ltd. 2003 21
Conclusion • The GSM network optimisation process consists of complex technical measurements and analysis • Every network is individually planned and thus have individual problems which decrease the quality • Professionally performed optimisation will increase customer satisfaction How to increase the quality and capacity GSM network with minimum investments ? indicators of European operator customer
Omnitele Ltd. 2003 22
How can Omnitele assist you? Radio network quality audit and optimisation
Omnitele Ltd. 2003 23
Performing the quality audit
Where?
To observe and follow the quality of the network Predefined areas
How often?
Regularly, e.g. once a year
How?
With measurements By comparing Independent from the operator and vendor
Why?
Who?
Omnitele Ltd. 2003 24
Network quality audit -- Objectives -Complete analysis of the quality perceived by subscribers Location of interference, inadequate signal strength and quality, congestion Proposal for corrective actions in both short and long term Comparison to the competitors and world class standard On-the-job training for OperatorÅs own staff Omnitele Ltd. 2003 25
Network quality audit -- Area definition -The most important cities, roads, suburbs Areas representing different kind of coverage, different subscriber density etc. Same basic area every time, expanded with possible new coverage areas
Omnitele Ltd. 2003 26
Measurement route definition ¥ ca. 1/3 of the # of cells in the ¥ ¥ ¥ ¥ ¥ ¥ ¥
area 30 min. of driving for each covering the whole sub-service area distinction of areas with different required coverage probabilities following the foreseen traffic distribution all HO pairs tested closing loops & major highways marketing should be consulted
Omnitele Ltd. 2003 27
¥ emphasis on Ð Ð Ð Ð
the city the most important roads the most important suburbs the areas with foreseen difficulties
Network quality audit -- Regularity -Once a year if the network is quite stable More often if the network is still under construction In addition to complete quality audit weekly measurements are recommended
Omnitele Ltd. 2003 28
Network quality audit --The auditor -Should be independent from the OperatorÅs Technical Department and the Vendor Should have experience from other networks Because, then ... The comparison will be objective The Operator gets the information what is th performance level of the network compared the other networks Omnitele Ltd. 2003 29
Quality audit -- Comparison -Predefined criteria the network should fulfill Quality Index (measured indicators) Comparison with competitors Comparison with previous quality audits By Quality Index Omnitele Ltd. 2003 30
Why Omnitele ? •
•
Long and extensive experience on radio network planning _ Dimensioning and nominal planning _ Site planning _ Parameter planning _ Network tuning and optimisation Experience from several equipment suppliers
Projects: • GSM planning in Finland _ Radiolinja project (worldÅs first GSM network) _ GSM1800 city networks _ Ongoing Finnish 2G project (3rd Finnish operator) • GSM planning abroad _ Italy Czech republic _ Netherlands Brazil Omnitele Ltd. 2003 31
Why Omnitele ? •
•
GSM network dimensioning/nominal planning for start-up operators _ Tanzania _ Jordan _ Pakistan Network audits and optimisation projects _ Finland • Radiolinja • City networks • Finnish 2G _ Abroad: • Greece Ivory Coast Kuwait • Togo Jersey Czech Republic • Guernsey Faroe Islands Cyprus • Andorra Bahrain
Omnitele Ltd. 2003 32
GPRS Optimisation in general
Omnitele Ltd. 2003 33
Optimisation Process Input Input The Thebusiness businessstrategy strategyfor forGPRS GPRS
Output Output ¥Optimal ¥Optimalparameter parametersettings: settings:
¥Existing ¥Existingand anddesired desiredquality qualityofofthe the network (field measurements & OMC network (field measurements & OMC studies): studies): ¥Congestion ¥Congestion ¥QoS ¥QoSofofthe theexisting existingnetwork network
¥Interference ¥Interference ¥Different ¥Differentvendor vendorsolutions solutions
¥Required ¥Requiredservices servicesand andexpected expected usage usage
¥Dedicated/dynamic ¥Dedicated/dynamicradio radio capacity allocation capacity allocation
¥Traffic ¥Trafficsteering steeringbetween betweenthe the radio layers radio layers ¥Coding ¥Codingscheme schemeadaptation adaptation ¥Different ¥Differenttimers timers
¥Optimal ¥Optimalhardware hardwareconfiguration configuration ¥Protocol ¥Protocoloptimisation optimisation ¥Key ¥Keyperformance performanceindicator indicator definition and measurements definition and measurements
¥Future ¥Futureplan planÐÐwinner winnerstrategy strategy
Best quality with minimum investment
Omnitele Ltd. 2003 34
Measurement System Set-up End-to-End data performance
Data at application layer
Data at application layer
GPRS layer data performance
Data at RLC/LLC layer
Data at RLC/LLC layer
RF performance
Air-interface
GPRS Mobile Omnitele Ltd. 2003 35
Air-interface
Server
Monitored Key Performance Indicators (KPI) Ð Ð Ð Ð Ð Ð Ð Ð
C/I of the measured channels Average RLC throughput (kbit/s) Average LLC throughput (kbit/s) Average number of timeslots used by a single user CS usage (% of time in the CS1/CS2 *) Average cell reselect time Network response time (RTT) RLC bit error rate
*) Available only in networks with dynamic CS adaptation
Omnitele Ltd. 2003 36
GPRS Parameter Planning (1) ¥ Goals: Ð To handle GPRS in cells with high CS (speech) traffic Ð To steer the GPRS traffic to wanted cells Ð Adequate definition of GPRS timers Ð Radio interface capacity division between CS/PS Ð Optimise the performance of specific service
Optimised end-user experience ! Omnitele Ltd. 2003 37
GPRS Parameter Planning (2) ¥
Omnitele has profound hands-on experience with
Ð Alcatel GPRS parameters Ð Ericsson GPRS parameters Ð Nokia GPRS parameters
TroubleshootingÅ Daily planning & OptimisingÅ Drive testsÅ Omnitele Ltd. 2003 38
Measuring GPRS network performance
RLC and LLC Throughput
Omnitele Ltd. 2003 39
• Number of time slots used per mobile • Coding scheme usage
• Data flow behaviour • Retransmissions • Cell reselections, RA updates etc.
GPRS Network performance benchmarking
Omnitele Ltd. 2003 40
Why benchmark the network performance ? ¥ Performance of the GPRS networks vary significantly between the operators ¥ The QoS status of operatorÕs GPRS network is best determined by benchmarking the network performance against competitorsÕ performance and/or world class operators operating in different markets
Omnitele Ltd. 2003 41
Benchmarking Results Examples ¥
Example country: Ð Three nation-wide GPRS operators Ð Same drive routes in the benchmarking Ð Significant differences in KPIÕs of the three operators Network response time (s)
Throughput DL [kbit/s] 35.0
33.3 31.3
33.2 31.3
34.7 32.4
1.8
1.8 1.6
1.5
1.4
30.0
1.1
1.2
25.0
RLC Throughput [kbit/s]
1
LLC Throughput [kbit/s]
0.8
Network response time (s)
0.6
20.0
0.4 0.2
15.0
X
Y
Omnitele Ltd. 2003 42
Z
0
X
Y
Z
Improving the Performance of Your Network ¥ After benchmarking the GPRS performance, the problem areas will be identified ¥ Recommendations for improvements are given ¥ Service optimisation plan will be made
Omnitele Ltd. 2003 43
BSC Measurements ¥
In addition to end-to-end field measurements, BSC measurements can be performed from the BSC of the benchmarked operator Ð Better overall picture of the problem cells: Suggestions to improvements
¥
Some of the key issues that can be identified from BSC are: Ð Overall GPRS traffic Ð Blocking of GPRS calls Ð Retransmissions Ð PCU loading
¥
Note: The BSC measurements can only be benchmarked against Omnitele archive since there is no access to competitorsÕ BSC
Omnitele Ltd. 2003 44
How can Omnitele assist you ?
Omnitele Ltd. 2003 45
Example of Omnitele Ås GPRS Projects Network acceptance and optimisation 1 † 2 † 2.1 2.1.1 2.1.2 2.1.3 2.2 2.2.1 2.2.2 2.2.3 2.2.4 2.2.5 2.2.6 † 3 3.1 3.1.1 3.1.2 3.2 3.2.1 3.2.2 † 4 4.1 4.2 4.3 4.3.1 4.3.2 4.4 4.5 4.6 4.6.1 4.6.2 †
Int roduc tion Descript ion of field measurement s GPRS measurement model Applicat ion layer performance measurements GPRS layer performance measurement s RF performance measurement s 5 Monit ored Key Performance Indicat ors 5.1 Carrier t o Int erference rat io for t he measured channels 5.2 (dB) 5.2.1 Coding scheme usage (%) 5.2.2 RLC/LLC throughput (kbit/s) 5.2.3 Cell reselec tion t ime (s) 5.3 Rout ing area updat e t ime (s) 5.4 Ne twork response t ime (s) 5.5 5.5.1 Findings during the Installat ion 5.5.2 Alcat el acceptance tests † MFS accep tance t est s 6 Core accep tance t est s Detect ed problems 6.1 MFS 6.2 Core 7 7.1 Int egrat ion t est ing 7.2 Carrier t o Int erference rat io 7.3 Coding scheme usage 7.4 Throughput 8 Normal cell load t hroughput measurement s 8.1 High cell load t hroughput measurement s 8.2 Cell reselec tion 8.3 rou ting area updat e time 9 Ne twork response t ime Pinging wit h Windows Command prompt Pinging wit h dif ferent int ervals be tween t he pings
Omnitele Ltd. 2003 46
GPRS Parameter fine-tuning Further routing area update testing Optimisation of the coding scheme usage CS_Lev parameter CS_Qual_1_2 / CS_Qual_ 2 _ 1 TBF_CS_Period1 / TBF_CS_Period2 Congestion control network response time optimisation Other investigated issues Frequency scans in the east coast Missing adjacencies Comparison of finnish GPRS operatorsÅ network Performance indicators Throughput Cell Reselection time (s) comparison of different vendorsÅ GPRS features Automatic Link adaptation (ALA) Congestion control Multiplexing users into the Timeslots TRX preference suggestions for GPRS network development Alcatel GPRS core rel2 and BSS software B7 Traffic monitoring Air-Interface compression software Conclusion
How Can Omnitele Assist You? ¥
¥
¥
¥
Contract negotiations: Ð Assistance in definition of the technical framework to the system specifications: How should the GPRS system perform? Ð Hands-on experience: Alcatel, Ericsson and Nokia GPRS systems Implementation: Ð Verification of the system performance: Specification of the operator acceptance tests and system integration tests Ð Parameter planning, parameter fine-tuning and optimisation of the GPRS network Service optimisation: Ð WAP&Web: Parameters&timers Ð Content optimisation: Performance Enhancement Proxy selection process Ð MMS: Radio capacity optimisation GPRS performance benchmarking Ð Comparison of your GPRS networkÕs performance to your competitors
Omnitele Ltd. 2003 47
Please visit our web site www.omnitele.fi for more information
Omnitele Ltd. 2003 48