Introduction to UMTS RAN KPI GSM-to-UMTS Training Series_V1.0 Change History Date Revision Version Description Author 2008-12-13 1.0 Draft completed. Zang Liang 2009-01-08 1.1 The remarks of basic concept are added in pages 4, 13, 19, 28, and 43. The remarks in page 26 are modified. Guo Hao Page 2 Objectives • • • Understand the definition of the RAN KPI Familiar with basic KPI test method Know how to analyze the network performance through the KPI After learning this course, we can know: Basic items for network acceptance test DT and CQT test methods of the KPI and traffic statistics analysis Simple signaling analysis Page 3 Foreword: UMTS&GSM KPI List UMTS KPI GSM KPI G&U KPI Difference Coverage rate Coverage rate CPICH Ec/Io and RSCP thresholds are set in the WCDMA network RXLEV threshold is set in the GSM network. SHO rate None SHO rate is a specific KPI for the CDMA service, which is closely related to the resource usage. RRC connection setup success rate Success rate of immediate assignments When the RRC signaling is established in the WCDMA network, it cannot differentiate the PS service and the CS service. RAB connection setup success rate Success rate of assignments The RAB connection setup success rate involves the Voice, VP and PS services, and is similar to the success rate of the TCH assignment and TBF setup success rate. Call drop rate Call drop rate The service type of the WCDMA network is different from that of the GSM network. Delay Delay It is tested by the drive test (DT) and call quality test (CQT), The KPIs related to the delay in the WCDMA network are the same as the KPIs related to the GPRS service in the GSM network. Handover success rate Handover success rate The classification of the KPIs in the WCDMA network and the GSM network is different. Page 4 Contents Introduction to KPI Test Method of KPI RAN KPI Counters Page 5 Introduction to KPI • Definition of KPI Key Performance Indicator • Management of KPI RNC-level, cell-level, and customized level • Function of KPI Define the network performance counters Monitor the network performance Locate and rectify the network faults Analyze the network trend • Purpose of KPI Test The verification and enhancement of the network optimization. That is, evaluate the network performance through the network KPI test, solve the network problem, continue to optimize the network to achieve the objective of building a refined network. Page 6 Introduction to KPI End-to-End Network Performance Evaluation System Other PLMN Core Network Service Network PSTN Internet Radio/Transmissi User Service Core Network on Equipmen Network Performance Network t Performance Performance Performance End-to-End Service Performance External Network Performance End-to-End Quality of Experience UMTS network performance evaluation system is used to map the service quality onto the end-to-end counters, which comprehensively indicate the UE performance, radio network performance, transport network performance, core network performance, and service network performance, and external network performance. Page 7 Introduction to KPI Relation of QoE, KQI, and KPI Quality of Experience (QoE) Availability Key Quality Indicator (KQI) Key Performance Indicator (KPI) Accuracy Speed Voice, Video Phone, MMS, WAP, Email, FTP, HTTP… AMR, CSD64k, CS Streaming, PS Steaming, BE64k/128k/384k, HSDPA Accessibility Retain ability Integrity Mobility Raw Counters, Drive Test, … Page 8 …. Coverage Availability Introduction to KPI Customer: As a customer, I do not care about KPIs, only in the hope of obtaining good and rich services. Operator: Our ultimate purpose is to meet increasing requirements of network subscribers and provide wireless networks with perfect user experience. As a channel between subscribers and the network, subscribers experience the network quality through services "Focus on User Experience " is a prerequisite and feature of a perfect network KPI system. Page 9 Introduction to KPI Huawei RAN KPIs Coverage AccessabilityRetainability Service Integrity Mobility • Coverage rate • RRC connection setup success rate • SHO rate • RAB connection setup success rate • Call setup success rate • Pilot pollution ratio • PDP activation success rate • Call completion rate of the calling party (voice, video, and SMS) • Call completion rate of the called party (voice, video, and SMS) • Call drop rate (voice, video, and PS services) • Mean opinion score (MOS) • Soft handover success rate • Video call (synchronization performance of voice and image • Intra-frequency hard handover success rate • Uplink/downlink BLER (Voice, Video Phone) • Average throughput of PS services (64K/128K/384 K/HSPA) • Inter-frequency hard handover success rate • Inter-RAT handover success rate Delay Availability • Call setup delay • Worst cell rate • PDP activation delay • Admission rejection rate • Point-to-point user plane delay • Paging congestion rate • Ping delay • Downlink transmission resource usage rate • Service disruption delay • Flow delay Page 10 • Congested cell rate Contents Introduction to KPI Test Method of KPI RAN KPI Counters Page 11 Test Method of KPI • Test Method • Prerequisite • The system works normally and stably. The test route must meet the following conditions: ① Test requirements of coverage rate: PCPICH RSCP >= -95dBm & Ec/Io >= -12dB ② Test requirements of other counters: PCPICH RSCP >= -90dBm & Ec/Io >= -11dB Basic operation of the network optimization is complete. The downlink analog load cannot exceed the planned load (optional). Test Tool and Software Page 12 Traffic statistics, drive test (DT), and call quality test (CQT) Mobile phone, frequency scanner, GPS, and laptop computer LMT, M2000, Probe, Assistant, and Nastar DT——Drive Test Classification of KPI Test Methods Evaluate the overall network performance after network optimization. − For example, mileage call drop ratio, Voice/VP call access rate/call drop rate of the calling party or called party, soft handover success rate, and Inter-RAT CS/PS HO success rate ST——Static/Stationary Test Evaluate the processing capability and speed of each network element. − Such as voice/VP call completion rate/call drop rate of the calling party or called party, PDP activation success rate/delay, uplink/downlink throughput, MOS, UE2UE/PSTN Voice/VP call setup delay/user plane delay, and Ping delay Stat.——Statistics Evaluate the overall network quality. − Such as SHO overhead, uplink interference cell rate, RRC/RAB connection setup success rate, Voice/VP/PS call drop rate, SHO success rate, Inter-RAT CS/PS HO success rate, worst cell rate, admission rejection rate, and congested cell rate Page 13 Test Method of KPI • Advantages and Disadvantages of DT/ST and stat. – Complementary to each other, DT/ST and Stat. must be flexibly used at different network phases and under different network conditions. Pro Con 1. Close to actual user experience. DT/ST 2. Unrestricted to the traffic. 1. Restricted to network conditions. 3. Easy to locate the fault. 2. Unsuitable for long-time observation. 4. Easy to test the specific network performance 3. Often affected by subjective and human factors. 1. Evaluating overall network performance Stat. 2. Suitable for long-time observation 3. Less affected by subjective factors and human factors – Page 14 1. Unnecessarily close to actual user experience 2. Restricted to the traffic volume 3. Difficult to locate the fault 4. Difficult to adjust test items to test network performance Another prerequisite of the KPI of the high-quality network is that all the KPIs of DT/ST/Stat must be covered and all kinds of counters must be complementary to each other. Test Method of KPI • Life Cycle of DT/ST/Stat. Network lifecycle Contract Requirements for KPIs Planning Implementation Optimization Cluster Optimization by •Field Test •Field Test KPIs KPIs (Drive Test/ (Drive Test/ Stationary Test) Stationary Test) for PAC •Statistic KPIs for FAC Page 15 Ongoing Optimization Commercial Launch Network Expansion FAC1 •Stat.: •RRC/RAB SSR •CDR •SHO SSR •... PAC •ST: •Call Setup Delay •PDP Context Activation Delay •Ping Delay •... PAC •DT: •CDR •CSSR •MOS •... Ongoing Optimization by •Field Test KPIs (Drive Test/Stationary Test) FAC2 •Stat.: •RRC/RAB SSR •CDR •SHO SSR •... Contents Introduction to KPI Test Method of KPI RAN KPI Counters Page 16 RAN KPI List KPI Type KPI Name KPI Name Mileage call drop ratio Soft handover success rate Coverage rate Intra-frequency handover success rate SHO rate Coverage KPI Type Mobility Inter-frequency handover success rate SHO overhead Inter-RAT handover success rate Pilot pollution ratio ...... Uplink Interference cell rate Call setup delay ...... PDP activation delay ...... UE2UE user plane delay RRC connection setup success rate UE2PSTN user plane delay Delay Accessibility Retainability RAB connection setup success rate Ping delay Call completion rate of the calling party Service disruption delay Call completion rate of the called party Flow delay PDP activation success ...... ...... Worst cell rate Call drop rate Admission rejection rate ...... Rate of relatively high uplink BLERs Availability Congested Cell rate ...... Rate of relatively high downlink BLERs Average downlink throughput Service integrity Average uplink throughput Mean Opinion Score (MOS) ...... Page 17 Remarks: 1. All the KPIs have two modes: Stat. mode and DT/ST mode. 2. According to the services, some KPIs have sub-item. 3. According to the calling part or the called party, some KPIs have sub-item. RAN KPI Counters Coverage Accessibility Retainability Delay Mobility Page 18 Coverage Analysis • KPI (DT/CQT) Coverage Rate At the early phase of the network deployment, the coverage rate is an important factor in the network planning and optimization. • KPI (stat.) SHO Rate and SHO Overhead The impact of the overlapped coverage on the GSM network is not obvious. For the WCDMA network, however, it may decrease the network capacity and deteriorate the coverage quality (pilot pollution). Page 19 Coverage Coverage rate • • • • Page 20 Test Method: DT Coverage rate=A1/A2*100% A1: length of valid test route (km) (CPICH Ec/Io≥-12dB & RSCP≥-95dBm) A2: total length of test route The coverage rate is calculated after the distance binning of 5 m is performed. Main Influencing Factors NodeB CPICH transmit power NodeB distribution condition Environmental factor Total interference level Test Conditions – Perform the test on the customer-specified route. – The area without coverage should not be included in the customer-specified route. Reference value of class A network: coverage rate >=96% Coverage SHO rate • Test Method: DT or stat. (evaluate the proportion of valid overlapped area) SHO rate = (Number of samples that 2 radio links in active set + Number of samples that 3 radio links in active set)/Total number of samples x 100% SHO_FACTOR_UE = (A2 + A3) / (A1 + A2 + A3) • • Page 21 A1: 1 radio links in active set A2: 2 radio links in active set A3: 3 radio links in active set Main Influencing Factors Soft handover parameter Site density and transmit power Average antenna height, radio propagation environment, antenna azimuth, and antenna tilt angle of the BTS Traffic load of the sector (mainly specifies the load of the Um interface, and is not related to the actual capacity of the network) Traffic distribution and MS distribution Reference value of class A network: SHO rate <=40% Coverage-SHO Overhead SHO overhead SHO Overhead={(ACTIVESET_1_UE_MEAN + ACTIVESET_2_2SOFTER_UE_MEAN*2 + ACTIVESET_2_2SOFT_UE_MEAN*2 + ACTIVESET_3_3SOFTER_UE_MEAN*3 + ACTIVESET_3_3SOFT_UE_MEAN*3 + ACTIVESET_3_2SOFTER_UE_MEAN*3) / (ACTIVESET_1_UE_MEAN + ACTIVESET_2_2SOFTER_UE_MEAN + ACTIVESET_2_2SOFT_UE_MEAN + ACTIVESET_3_3SOFTER_UE_MEAN + ACTIVESET_3_3SOFT_UE_MEAN + ACTIVESET_3_2SOFTER_UE_MEAN)} -1 Note : ACTIVESET_1_UE_MEAN: average number of UEs configured with one radio links. ACTIVESET_2_2SOFTER_UE_MEAN: average number of UEs configured with two radio links ACTIVESET_2_2SOFT_UE_MEAN: average number of UEs configured with two radio links that are not combined ACTIVESET_3_3SOFTER_UE_MEAN: average number of UEs configured with three radio links that are not combined. ACTIVESET_3_3SOFT_UE_MEAN: average number of UEs configured with three radio links that are not combined. ACTIVESET_3_2SOFTER_UE_MEAN: average number of UEs configured with three radio links that two links are combined. Reference value of class A network: SHO overhead <=45% Formula for calculating SHO overhead: SHO_Ratio = 100*{[VS.SHO.AS.1 + (VS.SHO.AS.2Softer + VS.SHO.AS.2Soft)*2+ (VS.SHO.AS.3Softer+ VS.SHO.AS.3Soft+VS.SHO.AS.3Soft2Softer)*3)]/ ( VS.SHO.AS.1 + VS.SHO.AS.2Softer + VS.SHO.AS.2Soft + VS.SHO.AS.3Softer + VS.SHO.AS.3Soft + VS.SHO.AS.3Soft2Softer )-1} SHO overhead In the Nastar software, the counter defined by the SHO overhead is as follows: SHO Overhead=VS.SHO.Factor.RL Page 22 RAN KPI Counter Coverage Accessibility Retainability Delay Mobility Page 23 • KPIs (stat.): Accessibility Analysis RRC connection setup success rate and RAB connection setup success rate These counters are mapped onto the success rate of immediate assignments and success rate of assignments in the GSM network. Similar to the analysis of the GSM accessibility, you can focus on these counters in the daily KPI monitor and network optimization. There is no KPI about the congestion in the WCDMA network. Therefore, you need to analyze the network congestion through these counters. • KPIs (DT/CQT) Call completion rate of the calling/called party and PDP activation success rate The test requirements of KPIs in the WCDMA network are basically the same as the test requirements of KPIs in the GSM network. The service type in the WCDMA test is increased, so the classification of the tested KPIs is also increased. Like the KPIs of GPRS/EDGE services, the KPIs of PS services are classified according to different rates. For the CS services in the WCDMA network, the VP service test is added. Page 24 RRC Connection Setup Success Rate • Test Method: stat. Stat. Number of RRC Connection Request in Point A1 of Figure1 Stat. Number of RRC Connection Setup in Point B of Figure1 Stat. Number of RRC Connection Setup complete in Point C of Figure1 Stat. Number of RRC Connection Reject in point A of Figure2 UE RNC RRC CONNECTION REQUEST A1 UE RNC A2 RRC CONNECTION REQUEST RRC CONNECTION SETUP B RRC CONNECTION REJECT RRC CONNECTION SETUP COMPLETE A C F1:successful RRC connection setup procedure F2: RRC connection reject procedure Page 25 • RRC Connection Setup Success Rate RRC Setup KPI (reference value of class A network: RRC connection setup success rate >=98.5%) RRC Connection Setup Success Rate = RRC connection setup success time / RRC connection request time received by RNC × 100% – • Note:here the RRC connection setup request time is the request time of initiating the service Calculation Formula of RRC Connection Setup Success Rate: – RRC Connection Setup Success Rate =100 *∑(VS.RRC.SuccConnEstab.OrgConv + VS.RRC.SuccConnEstab.OrgStr + VS.RRC.SuccConnEstab.OrgInt + VS.RRC.SuccConnEstab.OrgBkg + VS.RRC.SuccConnEstab.OrgSubs + VS.RRC.SuccConnEstab.TrmConv + VS.RRC.SuccConnEstab.TrmStr + VS.RRC.SuccConnEstab.TrmInt + VS.RRC.SuccConnEstab.TrmBkg + VS.RRC.SuccConnEstab.OrgHPSi + VS.RRC.SuccConnEstab.OgLPSig + VS.RRC.SuccConnEstab.TmHPSi + VS.RRC.SuccConnEstab.TmLPSi + VS.RRC.SuccConnEstab.Rest) / ∑(VS.RRC.AttConnEstab.OrgConv + VS.RRC.AttConnEstab.OrgStr + VS.RRC.AttConnEstab.OrgInt + VS.RRC.AttConnEstab.OrgBkg + VS.RRC.AttConnEstab.OrgSubs + VS.RRC.AttConnEstab.TrmConv + VS.RRC.AttConnEstab.TrmStr + VS.RRC.AttConnEstab.TrmInt + VS.RRC.AttConnEstab.TrmBkg + VS.RRC.AttConnEstab.OrgHPSig + VS.RRC.AttConnEstab.OrgLPSig + VS.RRC.AttConnEstab.TmHPSig + VS.RRC.AttConnEstab.TmLPSig + VS.RRC.AttConnEstab.Rest) • Main Influencing Factors Page 26 RRC Reject( Power Congestion、CE Congestion、NodeB RL Fail、ALCAP AAL2 Fail、FP Fail、Code Congestion、Other) RRC setup no response Exclude the success rate of the WCDMA networks in covering the edge cell of the network. The failed RRC connection setup due to the network congestion is excluded. Exclude the RRC connection setup failure caused by network congestion • RAB Connection Setup Success Rate Test Method: stat. Stat. Number of RAB Setup Request in Point A of Figure1 Stat. Number of RAB Setup Success in Point B of Figure1 Stat. Number of RAB Setup Failure in Point B of Figure2 UE RNC CN UE RNC RAB ASSIGNMENT REQUEST RAB ASSIGNMENT REQUEST A A RADIO BEARER SETUP RADIO BEARER SETUP RADIO BEARER SETUP COMPLETE RADIO BEARER SETUP FAILURE RAB ASSIGNMENT RESPONSE B RAB ASSIGNMENT RESPONSE F1: Successful RAB setup procedure CN B F2: RAB setup failure procedure Page 27 CS RAB Connection Setup Success Rate RAB connection setup success rate = Number of received CS RB setup complete messages reported by the UE / Number of CS RAB assignment requests sent by the CN x 100% Note: The CS RAB connection setup success rate consists of the Voice RAB connection setup success rate and the VP RAB connection setup Reference value of class A network success rate. Voice RAB connection setup success rate >= 99.5%, VP RAB connection setup success rate >= 99.5% Formula for calculating the CS RAB connection setup success rate 1. Voice RAB connection setup success rate = 100 x <VS.RAB.SucEstCSConv0.32.RNC> / <VS.RAB.AttEstCSConv0.32.RNC> 2. VP RAB connection setup success rate = 100 x <VS.RAB.SucEstCSConv32.64.RNC> / <VS.RAB.AttEstCSConv32.64.RNC> In the Nastar software Counter of the Voice RAB connection setup success rate: RAB.SuccEstabCSConv0.32.RNC.Rate Counter of the VP RAB connection setup success rate: RAB.SuccEstabCSConv32.64.RNC.Rate Page 28 PS RAB Connection Setup Success Rate PS RAB Connection Setup Success Rate PS RAB success rate = Number of received RAB setup complete messages of the PS domain reported by the UE / Number of RAB assignment requests of the PS domain sent by the CN x 100% Reference value of class A network PS RAB connection setup success rate >= 98.5% Formula for calculating the PS RAB connection setup success rate PS RAB connection setup success rate = 100 * (RAB.SuccEstabPSNoQueuing.Conv + RAB.SuccEstabPSNoQueuing.Strm + RAB.SuccEstabPSNoQueuing.Intact + RAB.SuccEstabPSNoQueuing.Bgrd + RAB.SuccEstabPSQueuing.Conv + RAB.SuccEstabPSQueuing.Strm + RAB.SuccEstabPSQueuing.Intact + RAB.SuccEstabPSQueuing.Bgrd) / (RAB.AttEstabPS.Conv + RAB.AttEstabPS.Strm + RAB.AttEstabPS.Intact + RAB.AttEstabPS.Bgrd) In the Nastar software Counter of the PS RAB connection setup success rate: Page 29 Call Completion Rate of the Calling Party Call completion rate of the calling party (DT test) Call completion rate of the calling party = Number of call completions of the calling party/Number of call requests of the calling party x 100% Note: The number of call requests of the calling party is the number of RRC Connection Requests initiated by the calling UE. If the RRC Connection Request is sent for multiple times during a call, it is recorded once only. The number of call completions of the calling party is the number of Alerting messages received by the calling UE. Reference value of class A network Voice >= 99%, VP >= 98% Service Voice/VP Test conditions ① ② ③ Carry out the test in the area of Ec/Io>-11dB and RSCP>-90dBm. Carry out the test in the case of low traffic, that is, the uplink load and the downlink load cannot exceed the planned load. Exclude the access failure caused by the admission failure. Exclude the access failure caused by problems not relating to access networks or by the UE. Test methods ① ② ③ ④ ⑤ ⑥ ⑦ Collect statistics on the Voice or the VP separately. Adopt WCDMA ONLY of SCANNER + Voice to consecutively make a call (short-time call). The called party uses a fixed-line phone or the UE with strong signals. The calling party makes a call. Then, the number of calling party requests is increased by 1. If the calling party hears the ring back tone, the number of successful call setups is increased by 1. Otherwise, the call fails to be set up. After the call is set up, maintain the conversation for 15 seconds. Then, the calling party hangs up. After 10 seconds, the calling party makes a call again. In the case of a call failure, makes a call after 30 seconds. At least carry out the test 200 times. Page 30 Call Completion Rate of the Called Party Callcompletion completion rate of of the the called party = Number call completions of the called Call rate called party (DToftest) party/ Number of call requests of the called party x 100% Note: The number of call requests of the called party is the number of CM Service Requests initiated by the calling UE. The number of call connections of the called party is the number of Alerting messages received by the calling UE. Reference value of class A network Voice >= 98%, VP >= 98% Service Voice/VP Test conditions ① ② ③ ④ ⑤ ⑥ Carry out the test in the area of Ec/Io>-11dB and RSCP>-90dBm. Carry out the test in the case of low traffic, that is, the uplink load and the downlink load cannot exceed the specified load. Exclude the access failure caused by the admission failure. Exclude the access failure caused by problems not related to the access network or by the UE. Exclude the paging failure caused by power-off of the UE. Exclude the paging failure for the reason that the called party is busy. Exclude the paging failure for the reason that the called UE is out of service. Test methods ① ② ③ Collect statistics on the voice call and VP call separately. Adopt WCDMA ONLY of SCANNER + Voice/VP to consecutively make a call (short-time call). The calling party uses a fixed-line phone or the UE with strong signals. The calling party cannot uses a fixed-line phone during a VP call. 31called party hears the The calling party makes a call. Then, the number of call requests of the called party increases byPage 1. If the ring back tone, the number of call completions of the called party increases by 1. Otherwise, the call fails to be set up. PDP Activation Success Rate PDP activation success rate (DT test) PDP activation success rate = Number of successful PDP activations / Number of PDP activation requests x 100% Note: There are two ways to define the number of PDP activation requests. a) Number of Activate PDP Context Requests initiated by the UE. b) Number of RRC Connection Requests initiated by the UE. The cause value contained in the request is the interactive service and background service of the calling party and called party. If the RRC Connection Request is sent for multiple times during a call, it is recorded once only. The number of successful PDP activations is the number of Activate PDP Context Accept messages received by the UE. Reference value of class A network >= 99% Service For the PDP activation success rate of R99, the PS service is established on the DCH. For the HSDPA, the PS service is established on the HSDPA. Test conditions ① ② ③ ④ Carry out the test in the area of Ec/Io > -11dB and RSCP > -90dBm. Carry out the test in the case of low traffic, that is, the uplink load and the downlink load cannot exceed the specified load. Exclude the access failure caused by the admission failure. Exclude the access failure caused by problems not related to the access networks or by the UE. The activation success rate of the service varies slightly with the rate. The service with a higher rate has a relatively lower activation rate. Test methods ① ② ③ ④ ⑤ ⑥ If the PS service is established on the DCH, adopt WCDMA ONLY of SCANNER + PS128K to consecutively perform the PDP activation test, that is, make short-time calls. If the PS service is established on the HSDPA, adopt WCDMA ONLY of Scanner + HSDPA to consecutively perform the PDP activation test. Carry out the test or collect statistics based on the typical subscription rate. Carry out the PDP activation test. If the PDP activation is successful, deactivate the PDP after 10 seconds. If the PDP activation fails, activate the PDP again after 10 seconds. Calculate the number of successful PDP activations and the number of PDP activation requests. Page 32 RAN KPI Counters Coverage Accessibility Retainability Delay Mobility Page 33 Analysis of Call Drop Rate • Call drop rate is a KPI that telecom operators are required to be tested, regardless of the WCDMA network or the GSM network. It is an important counter for the analysis of the traffic measurement and DT/CQT. • On the basis of the GSM network, the video call service is added to the WCDMA network. Compared with the analysis of the call drop rate in the GSM network, the analysis of the call drop rate in the WCDMA network is very difficult. Therefore, you should make more efforts to analyze it during the network optimization. Page 34 CS Call Drop Rate (DT Test) CS call drop rate (DT test) CS call drop rate = Number of CS call drops / Number of CS call connections x 100% Note: The number of CS call connections is the number of Alerting messages received by the UE after the service request is initiated Number of CS call drops: From the viewpoint of the Um interface signaling recorded on the UE side, the number of CS call drops is increased by 1 when one of the following messages of the Um interface is received during the conversation (in the call connection state). (1) Any BCH message (system message) (2) RRC Release message in which the release cause value is Not Normal (3) CC Disconnect message, CC Release Complete message, or CC Release message in which the release cause value is Not Normal Clearing, Not Normal, or Unspecified. Reference value of class A network Voice <= 1%, VP <= 1.5% Test conditions ① ② ③ Carry out the test in the area of Ec/Io>-11dB and RSCP>-90dBm. Carry out the test in the case of low traffic, that is, the uplink load and the downlink load cannot exceed the specified load. Exclude call drops caused by network congestion. Exclude call drops caused by problems not related to the access networks or by the UE. Test methods ① ② ③ ④ ⑤ ⑥ Adopt WCDMA ONLY of SCANNER + Voice/VP to consecutively make a call (long-time call). During a voice/VP call, the called party is in the strong-signal area of RSCP > -80dBm and Ec/Io > -8dB Set the call duration of the UE to 90 seconds and the waiting time after the call is terminated to 15 seconds. If the call drop occurs, initiate a call after 30 seconds. Calculate the number of call connections and the number of call drops. At least, carry out the test 200 times. Page 35 PS Call Drop Rate (DT Test) PS call drop rate (DT test) PS call drop rate = Number of PS call drops / Number of successful PDP activations x 100% Note: The number of successful PDP activations is the number of Activate PDP Context Accept messages received by the UE Number of PS call drops: From the viewpoint of the RRC signaling recorded on the UE side, the number of PS call drops is increased by 1 when the air interface messages meet one of the following conditions in the data service connection state. (1) The RRC Release message is not received, but the UE changes from the connection state (CELL_DCH, CELL_FACH, CELL_PCH, URA_PCH) to the idle state (IDLE). (2) The RRC Release message is received and the release cause value is not Normal or User Inactivity. Reference value of class A network <= 2% Service For the PS call drop rate of R99, the service adopts the planned PS full-coverage rate; for the PS call drop rate of HSDPA, the service is carried on HSDPA and adopts the planned HSDPA cell margin rate. Test conditions ① ② ③ Carry out the test in the area of Ec/Io>-11dB and RSCP>-90dBm. Carry out the test in the case of low traffic, that is, the uplink load and the downlink load cannot exceed the specified load. Exclude call drops caused by network congestion. Exclude call drops caused by problems not relating to access networks or by the UE. Test methods ① ② ③ ④ ⑤ For the PS call drop rate of R99, adopt WCDMA ONLY period of SCANNER + PS full-coverage rate to download the test (short-time call). For HSDPA PS call drop rate, adopt WCDMA ONLY period of SCANNER + HSDPA to download the test (short-time call). Carry out the test or collect statistics according to the full-coverage rate. Perform the PDP activation. If the PDP activation is successful, the number of successful PDP activations is increased by 1. Start the FTP server to download. After 90s, perform the PDP deactivation. If the call drop occurs during the download, the number of PS call drops is increased by 1. At least, carry out the test 200 times. Page 36 Call Drop Rate (Stat.) • Test Method: stat. Stat. Number of Call Drop in Point A of Figure 1 and in Point A of Figure 2 Definition of Call Drop: Abnormal RAB release triggered by RNC RNC CN RNC CN IU RELEASE REQUEST IU RELEASE COMMAND RAB RELEASE REQUEST IU RELEASE COMPLETE A A F1: RAB release request procedure F2: Iu connection release request procedure Page 37 CS Call Drop Rate (Stat.) CS call drop rate (stat.) CS call drop rate = Number of abnormal CS releases/(Number of abnormal CS releases + Number of normal CS releases) Reference value of class A network: Voice call drop rate <=1%; VP call drop rate <=1.5% Formula for calculating the CS call droop rate: CS AMR call drop rate =100* (VS.RAB.Loss.CS.AMR) / (VS.RAB.Loss.CS.AMR + VS.RAB.Loss.CS.Norm.AMR) VP call drop rate = 100 * (VS.RAB.Loss.CS.Conv64K.RNC) / (VS.Norm.Rel.CS.Conv.RB.64 + VS.RAB.Loss.CS.Conv64K) In the Nastar software: Counter of the CS AMR call drop rate: VS.AMR.Call.Drop.Rate.RNC Page 38 PS Call Drop Rate (Stat.) PS call drop rate (stat.) PS call drop rate = Number of abnormal PS releases initiated by the RNC/(Number of abnormal PS releases initiated by the RNC + Number of normal PS releases initiated by the RNC) Reference value of class A network: PS call drop rate <=3% Formula for calculating the PS call drop rate: PS call drop rate =100 * ( <VS.RAB.Loss.PS.RF.RNC> + <VS.RAB.Loss.PS.Abnorm.RNC> )/(VS.RAB.Loss.PS.RF + VS.RAB.Loss.PS.Abnorm + VS.RAB.Loss.PS.Norm) In the Nastar software: Counter of the PS call drop rate: VS.PS.Call.Drop.Rate.RNC Influencing factors and deviation suggestions 1. Exclude the case that the equipment is faulty. 2. Exclude the case of overload. Page 39 • • • • Page 40 KPI Counters Related to Call Drop Rate (Stat.) Overall counter CS call drop CS call drop rate PS call drop PS call drop rate Traffic class based counter Conversational Streaming Interactive Background Typical bearer based counter AMR speech CS 64 kbit/s PS 64 kbit/s PS 128 kbit/s PS 384 kbit/s Main Influencing Factors RF loss RLC reset AAL2 failure Other cause RAN KPI Counters Coverage Accessibility Retainability Delay Mobility Page 41 Delay • Delay is mainly tested by the call quality test (CQT). There is no great difference in the GSM network and the WCDMA network. For the CS services, the VP service is added to the WCDMA network. For the PS services, more channel occupation modes and multiple rates are applied to the WCDMA network. • Delay is applicable to the delivery or when customers perform the network CQT survey. Daily KPI monitor and network optimization are mainly performed on the basis of the exported data of the KPIs. Page 42 • AMR/VP Voice Delay AMR/VP delay test method: CQT • Test conditions • Connection delay: Refers to the interval from the time when the UE sends the RRC Connection Request for the first time to the time when the UE receives an Alerting message. 1. UEs of the calling party and called party are in the strong-signal area of RSCP > 90dBm and Ec/Io > -11dB. 2. Both the calling party and called party are in the local office. 3. When the static test is performed, the location of UE remains unchanged. Test methods 1. If you use a stopwatch to calculate the delay, the timing begins when you press the dial key and ends when a ring back tone is heard. 2. If the ring back tone from the peer end is not received, the call is defined to be unsuccessful and this delay is not recorded. 3. Record call-related messages (the interval from the time when the UE sends the RRC Connection Request to the time when the UE receives an Alerting message). 4. At least, carry out the test 20 times. Then, calculates the average delay. 5. Change a cell to perform the test at five locations. Page 43 • VP Image Delay CQT VP image delay test method: CQT • Connection delay: Refers to the interval from the time when the UE sends the RRC Connection Request for the first time to the time when the first frame image appears. Test conditions 1. UEs of the calling party and called party are in the strong-signal area of RSCP > -90dBm and Ec/Io > -11dB. 2. Both the calling party and called party are in the local office. 3. When the static test is performed, the location of UE remains unchanged. • Test methods 1. Record the time when you see the first frame image after you press the talk key. 2. If the Alerting message from the peer end is not received, the call is defined to be unsuccessful and this delay is not recorded. 3. At least, carry out the test 20 times. Then, calculates the average delay. 4. Change a cell to perform the test at five locations. 5. Exclude the excessive access delay caused by non-network reasons. Page 44 AMR/VP Call Delay UE->UE Means:4.5s 95%<5.5s UE->PSTN Means:3.3s 95%<3.54s UE->UE Means:4.5s 95%<5.5s Voice VP-Alerting VP-display UE->UE Means:9s (The VP delay difference is great between different UEs) Page 45 Ping Delay • Ping delay test method: CQT • • Ping delay: Refers to the time when the response message is received when the Ping request is sent. Test conditions – RSCP>-90dBm, Ec/Io>-11dB ; When the static test is performed, the UE should be placed in the area with good coverage, and then is connected to the test equipment. RSCP>-90dBm, Ec/Io>11dB – When the uplink or downlink TCH is mapped onto the DCH, the DCCC algorithm must be disabled to perform the Ping delay test. If the DCCC algorithm is enabled, the speed may decrease during the Ping test. In this case, the delay may increase. – The server adopts the Windows operating system, and the Ping server is directly connected to the Gi interface. Test methods 1. According to the requirements, the size of the Ping packets should be 0 bytes and 1460 bytes. 2. Exclude the excessive delay caused by the non-network reason. 3. If the timeout occurs, the average value of the delay should be calculated manually. Do not use the average value of the delay displayed in the CMD command. 4. At least, carry out 100 times. The average value is used as the value of the delay. 5. Change a cell to perform the test at five locations. 6. Exclude the excessive access delay caused by non-network reasons. Page 46 Reference KPIs Related to Ping Delay Counter Name Definition Service Reference Value of Delay DCH Ping It specifies the ping delay when the uplink or downlink TCH is mapped onto the DCH. UL 128/DL 384 0 bytes: 170 ms 1460 bytes: 380 ms FACH Ping It specifies the ping delay in the FACH state. RACH/FACH 0 bytes: 200 ms HSDPA Ping It specifies the ping delay in the HSDPA channel. UL 128k/DL HSDPA 0 bytes: 90ms 1460 bytes: 150ms FACH to DCH Ping It specifies the Ping delay when the state of the TCH transits from the FACH state to the DCH state after the Ping is performed. Ping 1460 Bytes 1300 ms FACH to HSDPA Ping It specifies the Ping delay when the state of the TCH transits from the FACH state to the HSDPA state after the Ping is performed. Ping 1460Bytes 1000 ms Idle to DCH Ping It specifies the Ping delay when the state of the TCH transits from the idle state to the DCH state after the Ping is performed. Ping 1460 Bytes 2800ms Idle to HSDPA Ping It specifies the Ping delay when the state of the TCH transits from the idle state to the HSDPA state after the Ping is performed. Ping 1460 Bytes 2650 ms Page 47 RAN KPI Counters Coverage Accessibility Retainability Delay Mobility Page 48 Analysis of Handover Success Rate • Like the call drop rate, the data of the handover success rate needs to be also exported from the test software and the drive test (DT). Call drop rate and handover success rate are important KPIs for engineering delivery. • According to the attributes of the target cell, the KPIs of the WCDMA network are categorized into the following: soft handover success rate, intra-frequency/inter-frequency hard handover, inter-RAT handover. During the analysis process, the soft handover success rate (similar to the G-BSC internal handover) is mainly analyzed. The intra-frequency handover occurs in the edge of the Radio Access Network (RAN) (refer to the GSM outgoing BSC handover). The inter-frequency handover occurs in the multi-band network (refer to the GSM dual-band handover). The inter-RAT handover indicates the WCDMA-to-GSM handover. • Compared with the GSM network, the handover counters related to the PS services is added to the WCDMA network. The procedure for the PS service handover in the WCDMA network can refer to the procedure for the CS handover in the GSM network. Page 49 Soft Handover Success Rate Soft handover success rate = Number of successful soft handovers / Number of Soft handover success rate (DT & Stat.) soft handover attempts x 100% Note: The number of soft handover attempts is the number of Active Set Update messages sent by the RNC The number of successful soft handovers is the number of Active Set Update Reference value of class A network Complete messages received by the RNC. >= 99.5% Service CS & PS Test conditions ① ② ③ ④ Perform the test in the area of Ec/Io>-11 and RSCP>-90dBm. Carry out the test in the case of low traffic, that is, the uplink load and the downlink load do exceed the specified load. Exclude the failure to add handover links caused by network congestion. Exclude the handover failure caused by the failure to establish the user plane. Exclude the handover failure caused by the UE. Test methods ① ② ③ ④ ⑤ Perform a long-conversation test by adopting different services, and place the UE of the called party in the strong-signal area. The RNC starts the single user signaling tracing of the tested UE. The UE records the RRC signaling process. Count the number of activity sets to be updated sent by the RNC and the number of updated activity sets received by the RNC. If the conversation is interrupted during the test, re-establish the conversation. Counters for Traffic Measurement ① ② Soft handover counter: VS.SHO.Succ/VS.SHO.Att.RNC Counter of the soft handover success rate in the Nastar software: VS.SHO.Succ.Rate Page 50 Soft Handover Success Rate • Measurement Point of Soft Handover Figure 1: procedure for successful soft handover Figure 2: procedure for failed soft handover Figure 3: procedure for soft handover timeout Main Influencing Factors Handover Delay parameter factor Movement speed Interference Signal UE RNC MEASUREMENT REPORT UE RNC change UE RNC MEASUREMENT REPORT MEASUREMENT REPORT A ACTIVE SET UPDATE A A ACTIVE SET UPDATE ACTIVE SET UPDATE B ACTIVE SET UPDATE COMPLETE B B WAITING ACTIVE SET UPDATE COMPLETE TIME OUT ACTIVE SET UPDATE FAILURE C factor C C F1: Successful soft handover procedure F3: Soft handover timeout p F2: Failed soft handover procedure Page 51 Inter-Frequency Hard Handover Success Rate Inter-frequency hard handover success rate (DT & stat.) Inter-frequency hard handover success rate = Number of successful inter-frequency hard handovers / Number of inter-frequency hard handover attempts x 100% Note: The number of successful inter-frequency hard handovers is the number of physical channels (used for inter-frequency hard handover) reconfiguration completion messages received by the RNC. The number of inter-frequency hard handover attempts is the number of physical channel (used for inter-frequency hard handover) reconfiguration requests sent by the RNC Reference value of class A network Voice >= 95%, VP >= 93%, PS64K > 93%, PS144K >= 90%, PS384K >= 88% Service CS & PS Test conditions ① ② ③ ④ Perform the test in the area of Ec/Io > -11dB and RSCP > -90dBm. Carry out the test in the case of low traffic, that is, the uplink load and the downlink load cannot exceed the specified load. Exclude the failure to add handover links caused by network congestion. Exclude the hard handover failure caused by the failure to establish the user plane. Exclude the hard handover failure caused by the UE. Test methods ① ② ③ ④ ⑤ Make a long-time CS call or download the PS service. The RNC starts the single-UE signaling tracing of the UE. The UE records the RRC signaling process. Calculate the number of physical channels (used for inter-frequency hard handover) reconfiguration requests sent by the RNC and the number of physical channel (used for inter-frequency hard handover) reconfiguration completion messages received by the RNC . If the conversation is interrupted during the test, re-establish the conversation. At least, perform the test 20 times in each handover scenario. Page 52 Inter-Frequency Handover Success Rate (Outgoing) • Measurement Point of Hard Handover UE Figure 1: procedure for successful hard handover Figure 2: procedure for failed hard handover Figure 3: procedure for hard handover timeout NodeB RNC MEASUREMENT REPORT A RADIO LINK SETUP REQUEST RADIO LINK SETUP RESPONSE PHYSICAL CHANNEL RECONFIGURATION PHYSICAL CHANNEL RECONFIGURATION COMPLETE B C F1: Successful hard handover procedure UE RNC UE RNC PHYSICAL CHANNEL RECONFIGURATION PHYSICAL CHANNEL RECONFIGURATION PHYSICAL CHANNEL RECONFIGURATION FAILURE WAITING PHYSICAL CHANNEL RECONFIGURATION TIME OUT A A F2: Failed hard handover procedure F3: Hard handover timeout procedure Page 53 AMR Inter-RAT Handover Success Rate (DT & stat.) AMR inter-RAT handover success rate (DT & stat.) Collect statistics on the UE side AMR inter-RAT handover success rate (outgoing) = Number of successful inter-RAT handovers / Number of inter-RAT handover attempts x 100% Note: The number of successful inter-RAT handovers is the number of Handover Complete messages sent by the UE. The number of inter-RAT handover attempts is the number of Handover from UTRAN Command messages received by the UE. In the flow chart of inter-RAT handover of the CS: A and B indicate the handover preparation process. C and D indicate the handover execution process. UE SRNC CN RELOCATION REQUIRED A RELOCATION COMMAND HANDOVER FROM UTRAN COMMAND B C IU RELEASE COMMAND D Successful inter-RAT HO outgoing procedure Page 54 AMR Inter-RAT Handover Rate (Outgoing) AMR inter-RAT handover rate (outgoing) Reference value of class A network Voice >= 97% Test conditions ① ② ③ ④ ⑤ ⑥ ⑦ The speed of the WCDMA network signal decreases by less than 30dB within 3 seconds. Exclude the handover scenario in the elevator. When the handover occurs, RxLex of the GSM network signal must be greater than -85dBm. Exclude handover failures caused by GSM network congestion. Exclude handover failures caused by GSM networks. Exclude handover failures caused by the UE. It is not a failure when the UE returns to the WCDMA network during the handover. Test methods ① ② ③ ④ ⑤ Determine three to five scenarios in which the UE is handed over from the WCDMA network to the GSM network. For example, the GSM network is equipped with the indoor distribution system, and the WCDMA network does not have the indoor distribution system. Make a long-time Voice call for a test and move the UE from the WCDMA coverage area to the GSM coverage area. Check whether the UE is handed over from the WCDMA network to the GSM network, that is, check whether the icon on the screen changes from WCDMA to GSM. If the conversation is not interrupted, the number of successful handovers is increased by 1. If the conversation is interrupted, the number of failed handovers is increased by 1. Return to the WCDMA area. Perform the cell reselection forcibly for the UE to camp on a WCDMA cell. Record signaling messages on the UE side and the RNC side. At least, perform the test over 20 times in each handover scenario. Formula for calculating the handover success rate: Number of successful handovers/(Number of successful handovers + Number of failed handovers) Counters for traffic measurement and calculation formula ① ② Success rate of inter-RAT handovers (outgoing) in the CS domain = 100 x (VS.IRATHO.SuccCSOut) / (VS.IRATHO.AttCSOut.RNC) In the Nastar software, the counter of the success rate of inter-RAT handovers (outgoing) in the CS domain is VS.IRATHO.SuccCSOut.RNC.Rate. Page 55 • PS Inter-RAT Handover Success Rate (Outgoing) PS inter-RAT handover success rate (outgoing) DT Formula for calculating PS inter-RAT handover success rate (outgoing) PS inter-RAT handover success rate (outgoing) = Number of successful PS inter-RAT handovers / Number of PS inter-RAT handover attempts x 100% Note: The number of successful PS inter-RAT handovers is the number of Routing Area Update Complete messages sent by the UE. The number of PS inter-RAT handover attempts is the number of CELL CHANGE ORDER FROM UTRAN messages received by the UE. In the flow chart of the PS inter-RAT handover, A and B indicate the handover request process. C and D indicate the successful handover process. UE SRNC CN MEASUREMENT REPORT CELL CHANGE ORDER FROM UTRAN A B IU RELEASE COMMAND C IU RELEASE COMPLETE D UTRAN-controlled cell change outgoing procedure Page 56 PS Inter-RAT Handover Success Rate (Outgoing) PS inter-RAT handover success rate (outgoing) DT Reference value of class A network >= 96% Test conditions ① Exclude the handover scenario in the elevator. ② The WCDMA network signal speed must decrease by less than 30dB within 3 seconds. ③ When the handover occurs, RxLex of the GSM signal must be greater than 85dBm. ④ Exclude handover failures caused by GSM network congestion. ⑤ Exclude handover failures caused by GSM networks. ⑥ Exclude handover failures caused by the UE. ⑦ It is not a failure when the UE returns to the WCDMA network during the handover. Test methods ① Place the UE in the area where the inter-RAT handover can occur. The UE camps on the WCDMA cell. Page 57 PS Inter-RAT Handover Success Rate (Stat.) inter-RAThandover handover success Number of successful PS inter-RAT handovers / PSPSinter-RAT successrate rate= (stat.) Number of PS inter-RAT handover attempts Note: The number of PS inter-RAT handover attempts is the number of CELL CHANGE ORDER FROM UTRAN messages sent from the RNC to the UE. The number of successful PS inter-RAT handovers is the number of IU RELEASE COMMAND messages received by the RNC in which the cause value is Successful or Normal Release. Reference value ofRelocation class A network PS inter-RAT handover success rate: >= 90% Formula for calculating the PS inter-RAT handover success rate PS inter-RAT handover success rate = 100 x (VS.IRATHO.SuccPSOutUTRAN.RNC) / (VS.IRATHO.AttPSOutUTRAN.RNC) In the Nastar software Counter of the PS inter-RAT handover Page 58 Conclusion • • • • • Page 59 Understand the test purpose of the KPI Know the application scenario of the KPI Master the test method of the KPI Compared with G-KPI, understand the definition of the KPI Understand the KPI baseline Thank You www.huawei.com