Masters Thesis Mikko Nieminen Espoo February 14 th

Master’s Thesis, Mikko Nieminen Espoo, February 14 th, 2006 TROUBLESHOOTING IN LIVE WCDMA NETWORKS Supervisor: Professor Heikki Hämmäinen

Background to the Study • The number of live WCDMA networks is growing quickly. • The first commercial Third Generation Partnership Project (3 GPP) compliant network, J-phone, was opened in December 2002. • By October of 2005, there were 80 live commercial WCDMA networks and the amount of subscribers was nearly 40 million. By that time, around 140 licenses had been awarded for WCDMA, the current WCDMA license holders having more than 500 million subscribers in their Second Generation (2 G) networks. • Especially in Europe and Asia, WCDMA network deployment after successful field trials and service launches has entered a new critical stage: the phase of network optimisation and network troubleshooting.

Research Problem • As the amount of WCDMA subscribers quickly increases, operators and equipment vendors are facing big challenges in maintaining and troubleshooting their networks. – We may raise the question of how one can efficiently narrow down the root causes of the problems when there is a huge amount of subscribers and traffic in a live WCDMA network. – What are the principles of examination of the fault scenarios and narrowing down the problem investigation into logical manageable pieces? – Which are the tools and methods that are in practice used in WCDMA network troubleshooting today? • In order tackle these questions and challenges, this Thesis presents a Framework for KPI-triggered troubleshooting in live WCDMA networks. • The applicability of the Framework is demonstrated by applying it to a selection of real troubleshooting cases that have occurred in commercial WCDMA networks.

Scope of the Study • This study concentrates on the KPI-triggered problems in live WCDMA networks. In general, the faults can be classified into three categories • – – – • • Critical, which are emergency problems that require immediate actions, Major (which we refer in this study as KPI-triggered problems) Minor which do not affect the services of the network. The viewpoint of is from the equipment vendor’s side, the main objective being to create guidelines for troubleshooting experts and technical support personnel of WCDMA network manufacturers in order to perform troubleshooting and narrow the problems down following a defined logic. This Thesis mainly concentrates on WCDMA network troubleshooting from a Radio Access Network perspective. The reasoning behind this approach is that the UTRAN covers most of the WCDMA specific functionality and intelligence, and therefore brings the majority of the troubleshooting challenges also.

Research Methods • This Thesis is mainly based on the study of various technical specifications and interviews of WCDMA network troubleshooting experts. • The main literature sources are the 3 GPP specifications of release 99, since the majority of the live WCDMA networks were based on 3 GPP release 99 during the writing of this Thesis. • It can be noted that 3 GPP release 4 networks are currently gaining foothold in the live WCDMA networks. However, there are only minor differences in the Radio Access functionality of the afore-mentioned two 3 GPP specification releases.

Structure of the Thesis • • • Introduction to WCDMA Networks UTRAN Protocols Call Trace Analysis Key Performance Indicators Framework for KPI-Triggered Troubleshooting • Cases from Live WCDMA Networks

WCDMA network architecture PSTN INTERNET GMSC GGSN Au. C CORE NETWORK HLR EIR SGSN MSC/VLR UTRAN Node B cell RNC Node B cell UE ME USIM Node B cell

UTRAN architecture UTRAN Iu-CS Node B 3 G MSC RNC Uu Node B Iub User Equipment (UE) Core Network (CN) Iur Node B SGSN RNC Node B Iu-PS

UMTS Bearer Services Non-Access Stratum Radio Access Bearer Signalling connection RRC connection Iu connection Access Stratum Radio bearer service Iu bearer service : SAP UE RAN Uu CN Iu

Summary of Protocols (CS user plane) Iub Uu Iu CS application and coding RLC MAC WCDMA L 1 Node B Iu-UP protocol FP FP AAL 2 ATM ATM PDH/SDH UE Iu-UP protocol PDH/SDH RNC PDH/SDH MSC

Summary of Protocols (UE control plane) Iub Uu Iu NAS RRC RANAP RLC SCCP MAC MTP 3 b SSCF-NNI SSCOP WCDMA L 1 UE RANAP FP FP AAL 2 AAL 5 ATM ATM PDH/SDH WCDMA L 1 Node B RNC CN

Overview of WCDMA Call Setup MT Call Paging MO Call RRC Connection Establishment Radio Access Bearer Establishment User Plane Data Flow

RRC connection establishment (DCH) UE RRC Node B RNC 1. RRC CONNECTION REQUEST RRC 2. Admission Control C-NBAP 3. RADIO LINK SETUP REQUEST C-NBAP 4. Start RX 5. RADIO LINK SETUP ESPONSE C-NBAP ALCAP FP C-NBAP 6. ESTABLISH REQUEST 7. ESTABLISH CONFIRM 8. UPLINK & DOWNLINK SYNC ALCAP FP 9. Start TX RRC 10. RRC CONNECTION SETUP RRC 11. L 1 SYNCH D-NBAP RRC 13. RRC CONNECTION SETUP COMPLETE 12. RL RESTORE INDICATION D-NBAP RRC

Protocol Analysers Company Product Home Country Nethawk [47] 3 G Analyser Finland Agilent [48] Signaling Analyzer United States K 15 United States Tektronix [49] Radcom [50] Performer Analyser Israel Acterna [51] Telecom Protocol Analyzer United States

RRC Connection Events and KPIs UE RNC RRC CONNECTION REQUEST CN Event 1 Setup phase RRC CONNECTION SETUP Event 2 RRC_CONN_ATT_COMP incremented Access phase RRC CONNECTION SETUP COMPLETE Event 1 RRC_CONN_ATT_EST incremented Event 3 Active phase Event 3 RRC_CONN_ACC_COMP incremented Event 4 RRC_CONN_ACT_COMP incremented Event 4 IU RELEASE COMMAND RRC Setup Complete Rate = Sum of RRC_CONN_STP_COMP Sum of RRC_CONN_STP_ATT x 100 % Sum of RRC_CONN_ACC_COMP RRC Establishment Complete Rate = x 100 % Sum of RRC_CONN_STP_ATT Sum of RRC_CONN_ACT_COMP RRC Retainability Rate = x 100 % Sum of RRC_CONN_ACC_COMP

RRC connection Phases Phase: Setup complete Access Active Complete Success Active Release Active Failures Attempts Access Failures Setup Failures, Blocking RRC Drop

Other WCDMA network KPIs RAB Setup Complete Rate = Sum of RAB_STP_COMP Sum of RAB_STP_ATT RAB Establishment Complete Rate = RAB Retainability Rate = x 100 % Sum of RAB_ACC_COMP Sum of RAB_STP_ATT Sum of RAB_ACT_COMP Sum of RAB_ACC_COMP Sum of CSSR = RAB_ACC_COMP x 100 % Sum of RRC_CONN_STP_ATT Sum of CCSR x 100 % RAB_ACT_COMP = Sum of RRC_CONN_STP_ATT x 100 %

Fault Classification Fault Class Description Examples A-CRITICAL Total or major outages that are not avoidable with a workaround solution. Critical (emergency duty contacted) problems severely affect service, capacity/traffic, billing, and maintenance capabilities and require immediate corrective action, regardless of time of day or day of the week as viewed by the operator. • System restart, all links down • Simultaneous restarts of active computer units • More than 50 per cent of traffic handling capacity out of use • Subscriber related network element functionality is not working B-MAJOR The problem leads to degradation of network performance or the fault affects traffic randomly. Major problems cause conditions that seriously affect system performance, operation, maintenance, and administration and require immediate attention as viewed by the operator. The urgency is less than in critical situations because of a lesser immediate or impending effect on system performance, customers, and the customers operation and revenue. • Capacity/quality related functionality is not working as supposed to • Problems seriously affecting end user service, but avoidable with a workaround solution • Configuration changes (network, HW, and SW) are not working as supposed to • Subscriber related functions are not working completely • Performance measurement, alarm management or activation of a new feature fails • Single restart of computer units C-MINOR Minor fault not affecting operation or service quality Other problems that the operator does not view as critical or major are considered minor. Minor problems do not significantly impair the functioning of the system or affect the service to customers. These problems are tolerable during system use. • Failures not seriously affecting traffic • Errors in operating commands syntax • Cosmetic errors in operational commands or statistics output • Minor errors in documentation

Framework for KPI-Triggered Troubleshooting • Framework is designed for investigating and soelving B-MAJOR level i. e. “KPI-triggered” faults • Before applying the Framework – The general alarm status of the network has been checked. No clear network alarms pointing to the root cause of the fault can be detected. – Traces from external interfaces of RNC have been taken with a protocol analyser in order to record the fault scenario. Also RNC internal trace has been taken when the fault took place. – The basic fault scenario has been analysed and clarified.

A Is the problem new in the operator network? No B Yes Perform simulation of the fault in test bed. Does the fault still occur? C No New SW, HW, parameters, UE model or feature introduced? No Yes D No Yes E Is the fault operator specific? Perform simulation of the fault with reference conditions. Does the fault still occur? Yes F H CN specific Q R No Use RNC Performance Tester to generate load in test bed and perform analysis. I J G Has average network load increased significantly and/or does the problem occur at a specific time of day? Yes No Analyse and investigate the differences between the working and faulty conditions. Analyse the traces. Investigate fault scope. K L RNC specific Node B specific M Transmission specific N Service specific O Country specific P UE specific Analyse network element and interface specific alarms, parameters, capacity, logs and traces. Take specific actions depending on problem scope (refer to detailed Framework notes). In case of MVI environment, check IOT results and contact foreign vendor. Investigate own vendor’s default parameters and compare implementation againts 3 GPP specifications. Compare own default parameters with other default parameters of other vendors. Execute air interface protocol analysis and drive tests.

Case: Increased AMR call drop rate • A decrease in RAB Retainability Rate KPI for AMR telephony service was experienced during the last three months in an operator network. • The decrease was around 2% on each RNC compared to the time when the network was performing well. Actions that had already been taken with no positive effect: – Soft reset for all Node Bs and for all RNCs – Hard reset and re-commissioning of Node Bs – Alarms checked and no major alarms found

Case: Increased AMR call drop rate I. A Is the problem new in the operator network? Yes II. C New SW, HW, parameters, UE model or feature introduced? Yes III. E Perform simulation of the fault in reference conditions. Does the fault still occur? No G IV. Analyse and investigate the differences between the working and faulty conditions.

Case: Increased AMR call drop rate • Solution – The short term solution was that the parameter for planned maximum downlink transmission power of all the Node Bs in the operator network was changed to the default value of 34 d. Bm. In this way, the problem disappeared in the operator network. – The long term solution was to implement a fix of the bug into the next software release of the Node B.

Results • • • As a result of thorough research conducted for this Thesis, a Framework for KPItriggered troubleshooting for live WCDMA networks was developed. The Framework is mainly targeted for WCDMA network equipment vendors, to help them in solving major service affecting faults occurring in the live WCDMA networks of today. Troubleshooting cases from live WCDMA networks were solved using the Framework developed, in order to verify the results and test the applicability and practicality of the Framework.

Assessment of the results • • The applicability and relevance of the troubleshooting Framework was tested against three different fault cases from live WCDMA networks. The results were fairly promising since all the cases were successfully solved by utilising the Framework. The Framework was found to be quite practical and suitable for solving KPI-triggered problems in live WCDMA networks. However, it must be taken into account that the Framework was tested with a limited number of cases, because of time and resource limitations. If more extensive testing and verification with a large number of cases would be applied, there is a possibility that optimisations and improvements to the Framework could be done. Still, the basic logic of the Framework was proven with reasonable relevance. The results presented in this study can be easily tested in the future against a number of cases in order to verify the results with more extensive statistical reliability.

Exploitation of the results • The results of this study will be used as source material in the development of UTRAN troubleshooting competence development and advanced learning solution creation, targeted for troubleshooting experts and customer support engineers of one of the leading WCDMA network equipment vendors. • Also, the results of the Thesis will be used as an input in creation of customer documentation for UTRAN troubleshooting. • There is also an intention to further test the relevance and reliability of the results of this Thesis by applying it in the 24/7 RAN technical support operator service of the equipment vendor in question.

Future Research • The significance of Performance Indicator based troubleshooting is increasing continuously in live WCDMA networks. • Once the PI and KPI specifications become more mature, more extensive study of the most relevant Performance Indicators used in WCDMA network troubleshooting is essential. • Also, there is a need to develop a Framework and logic for solving emergency problems in WCDMA networks. • As the growth of complexity of telecommunication networks increases, effective and efficient troubleshooting procedures are essential in order to manage the diversity of network technologies and the increasing quality requirements of the operators.