GPRS Optimization Handbook

GPRS Network Optimization
Optimization Handbook (Preliminary)
For internal use only
Issued by:
Siemens Shanghai Mobile Communications, Ltd. Network Performance Headquarter 777 ChuanQiao Road, 201206, Shanghai, China
SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook
1/135
2005-09-12
History of Document:
Version: 1.0
Date: 2005.09.12
Changes: First release
2/135
2005-09-12
Table of Contents
1 INTRODUCTION ................................................................................................................................ 5 1.1 1.2 1.3 2 NETWORK OPTIMIZATION CONCEPTS ............................................................................................... 5 NETWORK OPTIMIZATION PROCESS ................................................................................................. 6 AIM OF THIS HANDBOOK.................................................................................................................. 7
GPRS BASIC ........................................................................................................................................ 8 2.1 GENERAL INTRODUCTION ................................................................................................................ 8 2.2 GPRS NETWORK SYSTEM ARCHITECTURE ...................................................................................... 8 2.3 GPRS PROTOCOLS ......................................................................................................................... 11 2.4 GPRS PROCEDURES BASIC ............................................................................................................ 14 2.4.1 Mobility Management States ................................................................................................ 14 2.4.2 Packet Data Protocol states ................................................................................................. 15 2.4.3 GPRS attach ......................................................................................................................... 16 2.4.4 Routing area update ............................................................................................................. 17 2.4.5 Cell update............................................................................................................................ 18 2.4.6 PDP context activation ......................................................................................................... 19 2.5 GPRS RADIO NETWORK ................................................................................................................ 20 2.5.1 GPRS Um interface .............................................................................................................. 20 2.5.2 GPRS radio network features in BR7 ................................................................................... 22 2.6 GB INTERFACE ............................................................................................................................... 24 2.6.1 Gb physical connection......................................................................................................... 25 2.7 GPRS CORE NETWORK .................................................................................................................. 26 2.7.1 SGSN architecture ................................................................................................................ 26 2.7.2 GGSN architecture ............................................................................................................... 29 2.7.3 GSN Hardware components capacity assumptions .............................................................. 32
GPRS PARAMETERS IN BSC DATABASE.................................................................................. 33 3.1 SBS DATABASE STRUCTURE WITH GPRS RELATED OBJECTS ........................................................ 33 3.2 INTRODUCTION OF GPRS RADIO PARAMETERS ............................................................................. 34 3.2.1 GPRS channel configuration parameters ............................................................................. 34 3.2.2 Coding scheme and Link Adaptation parameter .................................................................. 35 3.2.3 GPRS feature parameter....................................................................................................... 36 3.2.4 RLC/MAC parameter............................................................................................................ 38 3.2.5 Other parameters.................................................................................................................. 39 3.2.6 Recommended GPRS radio parameter settings from TAC3 ................................................. 40
INTRODUCTION OF GPRS RADIO PERFORMANCE MEASUREMENTS........................... 40 4.1 4.2 4.3 4.4 GENERAL CONCEPTS...................................................................................................................... 40 STRUCTURE OF SCANNERS ............................................................................................................ 42 SBS OPTIMIZATION WORK FLOW OF PERFORMANCE MEASUREMENT ............................................. 44 MAIN GPRS RADIO NETWORK PERFORMANCE KPIS (BASED ON BR7) ......................................... 45
INTRODUCTION OF GPRS CORE PERFORMANCE MEASUREMENTS ............................ 46 5.1 GENERAL CONCEPTS...................................................................................................................... 46 5.2 PERFORMANCE DATA HANDLING ................................................................................................... 48 5.3 DATA REQUIRED FOR CORE MEASUREMENT EVALUATION (BASED ON PO2.1)............................... 49 5.3.1 SGSN evaluation................................................................................................................... 49 5.3.2 GGSN evaluation.................................................................................................................. 51 5.3.3 Set up the PDC ..................................................................................................................... 51 5.4 GPRS CORE NETWORK KPIS ......................................................................................................... 52
INTRODUCTION OF COMMON GPRS NETWORK OPTIMIZATION TOOLS ................... 54

3/135 2005-09-12
MAIN GPRS OPTIMIZATION SKILLS ........................................................................................ 55 7.1 END-TO-END GPRS PERFORMANCE TESTS .................................................................................... 55 7.1.1 General introduction ............................................................................................................ 55 7.1.2 GPRS end-to-end test tools................................................................................................... 56 7.1.3 Preparations before test ....................................................................................................... 57 7.1.4 Using TEMS investigation .................................................................................................... 59 7.1.5 Test items .............................................................................................................................. 65 7.2 GPRS RADIO MEASUREMENT ANALYSIS WITH BATRANA2004 ...................................................... 72 7.2.1 Introduction .......................................................................................................................... 72 7.2.2 Using MySQL ....................................................................................................................... 72 7.2.3 Getting Data from project .................................................................................................... 75 7.2.4 MySQL connection and configuration.................................................................................. 77 7.2.5 Project Operation ................................................................................................................. 78 7.2.6 Import Data .......................................................................................................................... 79 7.2.7 Network Performance Analysis ............................................................................................ 81 7.3 GPRS CORE MEASUREMENT ANALYSIS WITH NETAN2000 ............................................................ 84 7.3.1 Introduction to NETAN2000................................................................................................. 84 7.3.2 NETAN Basics ...................................................................................................................... 85 7.3.3 NETAN2000 Result Files...................................................................................................... 87 7.3.4 Evaluations of SSNC/GPRS.................................................................................................. 91 7.3.5 Long Term Statistics ........................................................................................................... 104 7.4 GPRS SIGNALING ANALYSIS (CALL ANALYZER, K1205)............................................................ 105 7.4.1 Introduction ........................................................................................................................ 105 7.4.2 K1205 introduction............................................................................................................. 105 7.4.3 Call Analyzer introduction ................................................................................................. 121
GPRS OPTIMIZATION CASE STUDIES .................................................................................... 126 8.1 GPRS RADIO OPTIMIZATION CASES ............................................................................................. 126 8.1.1 Increase throughput by improving C/I................................................................................ 126 8.1.2 Activating GPRS forced intra cell handover ...................................................................... 127 8.1.3 Peak throughput management ............................................................................................ 127 8.1.4 Reduce delay on Gb interface............................................................................................. 128 8.1.5 Cell reselection parameter: CELLRESH ............................................................................ 129 8.1.6 Reconfigure number of Abis subslots under FAAS ............................................................. 129 8.2 GPRS CORE OPTIMIZATION CASES............................................................................................... 130 8.2.1 GMM/SM procedure analysis by call analyzer .................................................................. 130 8.2.2 Investigation on data transfer pause .................................................................................. 131 8.2.3 WAP performance analysis................................................................................................. 132 8.2.4 Mis-Configuration of VLAN ............................................................................................... 134 8.2.5 Detect radio problems from core network side................................................................... 135
4/135
2005-09-12
1 Introduction
1.1 Network optimization concepts
Network optimization is a value-added service product which can be provided to a mobile network operator. A service package may comprise many customer-oriented contents such as: comprehensive audit of the running network quality, in-depth analysis of network problems, solutions to improve network performance, network monitoring, cutover support, etc. The network performance is usually quantified by several number of key performance indicators (KPIs) defined in a network optimization contract. These KPIs will be evaluated before and after the provision of network optimization service in order to quantify the result of contract execution. General contents and evolvement of network optimization service is shown in the following chart:
Opt. of new services
Evolvement
Efficiency & Profitability Network Quality Coverage and capacity for basic services
As a first step we ensure the basic coverage and capacity for basic services such as: voice call, SMS, etc. by necessary expansions. Secondly, the quality of existing coverage and capacity will be optimized in order to ensure subscriber satisfaction and keep the operator competitive in the market. Thirdly, in order to maximize the operators revenue, the network has to be more profitable and efficient. Special optimization tasks will carried out such as signaling load optimization (hence network can serve more subscriber with available equipments), TRX re-allocation (hence investment on expansion is delayed), cell load balancing (traffic is shared among the available cells) , dual-band optimization (hence network capacity is fully utilized), etc. As a last step, new services (e.g. packet services) will be optimized in order to allow the operator to operate them at high quality level (i.e. subscriber satisfaction) and share network resources properly between basic services and new services (i.e. operator benefit), so that return of investment from new services is also high. In general, the network optimization service chain is provided in order to guarantee an mobile network operators long-term success in the competitive market.
SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook 5/135 2005-09-12
1.2 Network optimization process

An overview flow of the network optimization process is presented in the chart below:
Contract Acquisition
Project Preparation
Project Kickoff
Audit Audit Report
Analysis
Solution Change Request Implementation Stage Report Verification
Problem Solved?
No
Yes
Targets Reached?
No
Yes
Monitoring
New problem Found?
Yes
No No
Time Reached?
Yes
Customer approval
Final Report
6/135
2005-09-12
1.3 Aim of this handbook

The aim of this handbook is to provide information of basic 2.5G GPRS network optimization skills and related background knowledge of packet switched data services. Both GPRS radio and core network will be concerned. A few case studies will be provided in chapter 8 in this handbook to give the reader a little feeling on what GPRS optimization can do and how they are implemented in the field. The target readers of this handbook are beginners who intend to learn basic skills of GPRS network optimization or field engineers who would like to use this handbook as an easy reference for daily tasks.
7/135
2005-09-12
2 GPRS basic
2.1 General introduction
The original GSM network is only able to provide circuit-switched data service with limited data rate of 9.6kbps. With more and more mobile data service requirements from the mobile subscribers, the GSM phase2+ standard had introduced enhanced mobile data communication technologies into GSM network to meet the demand. As one of the new technologies of GSM phasae2+, GPRS(General Packet Radio Service) was designed to provide the packet-switched high speed data service based on the traditional GSM network. The main features why GPRS can provide higher speed mobile data service are the new coding schemes and channel bundling of GPRS radio network. Channel bundling means that one GPRS subscriber can use more than one radio timeslot simultaneously for the same data session while the traditional GSM data service can use only one radio timeslot. As for the new coding schemes, GPRS provides 4 coding schemes: CS1, CS2, CS3, CS4. The higher coding scheme, the higher data rate is expected over air interface. On the other hand, one GPRS radio timeslot can be shared not only by multiple GPRS subscribers at the same time, but also shared dynamically by GPRS data service and GSM voice service at different time. All these features improve the radio channel resource efficiency, which is the traditional bottleneck, of the mobile network and finally benefit the mobile operator by higher network resource efficiency and being more attractive to mobile data users. Since the GPRS offers a new packet data service and brings much complexity into the mobile network, more GPRS network details will be introduced in the next four small chapters: network architecture, protocols, procedures and radio network specific information. Please be noted that the information given in this handbook is quite brief, for more detailed and complete technical information, please refer to plenty of other Siemens official documentations. In the last chapter, a few recommended documents and links will be provided.
2.2 GPRS Network system architecture

GPRS network consists of radio and core network: GPRS network shares the common radio network with GSM network but has its own new GPRS core network. The network architecture of GPRS network is shown in the chart below: (The red color marks the GPRS network elements and the yellow color marks the GSM network elements; The two vertical dashed lines separate the radio network, core network and external network)
8/135
2005-09-12
In the radio network part, GPRS doesnt need new network node while PCUs (Packet Control Unit) shall be added as part of the BSC no matter the PCU modules are integrated into BSC racks or installed in an extra new PCU rack (Siemens solution is to integrate PCU module into BSC rack). The basic function of PCU is to take care of GPRS radio channel scheduling/multiplexing, data packet segmentation/assembling and the routing of data packets to GPRS core. In addition to the new PCU module, new BSS software is also necessary in both BSC and BTS to accomplish the GPRS functionalities. In Siemens solution, SBS BR55 is the first BSS software release starting to support GPRS. The GPRS functionality unit in BTS is normally named as CCU(Channel Codec Unit) and please note that CCU stands for only software enhancement in BTS without addition of any new hardware modules. In the core network part, GPRS introduces two new major nodes: SGSN(Serving GPRS Support Node) and GGSN(Gateway GPRS Support Node). SGSN handles mainly the mobility of the subscribers and radio related protocol toward the mobile station. GGSN interworks with the external packet data network. Besides the SGSN and GGSN, the GPRS core network consists of some other network elements, which are also important and necessary to accomplish other GPRS control functionalities, such as DNS server, Border Gateway, etc. GPRS core nodes have to be connected to the radio network and other GSM network nodes in order to convey the user data and control message across the whole network, accordingly a lot of new interfaces have been introduced by GPRS as shown in the chart below:
9/135
2005-09-12
The major GPRS interfaces are following: Gb interface: between SGSN and BSC Gn interface: between SGSN and GGSN Gs interface: between SGSN and MSC Gi interface: between GGSN and external data network Gr interface: between SGSN and HLR Gd interface: between SGSN and SMS-G-MSC Gp interface: between GGSN and Border Gateway Even though all above interfaces are defined in the GPRS standard, not all of them are mandatory to be implemented in live GPRS networks. Take the Gs interface for example: although Gs interface can improve the GPRS network performance by supporting the combined GPRS-GSM procedures and reducing the overall network signaling load, it is not really implemented yet in most of the GPRS networks. On the other hand, we shall note that some interfaces carry only signaling (for example Gs), while the others carry both signaling and user data (for example Gb). GPRS user subscription data is stored in HLR. This is the reason why Gr interface is mandatory so that SGSN can communicate with the HLR to ask for user subscription data. Consequently the HLR has to be upgraded to be higher than Siemens SR8.0 in order to support GPRS functionality. Finally the mobile phone shall be GPRS capable. According to standard, there are three different classes of MS regarding to their GPRS capability: Class A: MS supports simultaneous and independent handling of GPRS and GSM services (operation & signaling).Class A MS consists of two independent Tx & Rx. Class B: MS supports GPRS and GSM services, but cannot handle them simultaneously (signalling for both, GPRS & GSM). GPRS traffic will be suspended in case of pending or established CS connection.
10/135
2005-09-12
Class C: Alternate use of GSM or GPRS services. Class C MS are either GSM or GPRS MS at one time. Moreover, GPRS capable MS can be divided into different multislot classes according to its capability of supporting radio timeslots. The multislot class information is listed in the table below.
Multislot class Maximum number of slots Rx 1 2 3 4 5 6 7 8 9 10 11 12 1 2 2 3 2 3 3 4 3 4 4 4 Tx 1 1 2 1 2 2 3 1 2 2 3 4 Sum 2 3 3 4 4 4 4 5 5 5 5 5 Minimum number of slots Tta 3 3 3 3 3 3 3 3 3 3 3 2 Ttb 2 2 2 1 1 1 1 1 1 1 1 1
For instance, Ericsson mobile phone R520 used widely in GPRS field test is a multislot class 4 and class B GPRS mobile phone, which means that it can support up to 3 downlink radio timeslots at the same time.
2.3 GPRS protocols

A lot of new protocols are introduced into the GPRS network on all interfaces, most of them are designed to fit the character of packet data transferring. One of the major highlights of GPRS network is that it uses IP protocol both internally and externally so that GPRS network could have direct connection to external data networks. The protocol layers of the GPRS network have two types: one for transmission plane (for user data) and one for control plane (for signaling). The protocol layer of transmission plane is shown in chart below:
11/135
2005-09-12
G PRS Transm ission Planes

MS
Relay
Um
BSS
Gb
SG SN
Gn
G G SN
Relay IP/ X.25
Gi
R outer
Application/ Higher level protocols IP/ X.25 Relay SND G TP CP LLC Relay RLC BSS GP M AC G SM RF FR L1 bis
LLC : M AC : RLC :
IP/ X.25
SND CP LLC RLC
G TP
IP/ X.25
IP/ X.25
UDP / TCP IP L2 L1
UDP / TCP IP L2 L1 L2 L1 L2 L1
BSS GP FR L1 bis
L2 L1
BSSG P : FR : G TP : IP :
M AC G SM RF
BSS G PRS Protocol Fram e Relayl G PRS Tunnelling Protocol Internet Protocol
Logical Link Control M edium Access Control Radio Link Control
SNDCP : SubNetwork Dependent Convergence Protocol TCP : Transm ission Control Protocol UDP : User Datagram Protocol
The transmission plane involves Um, Gb, Gn and Gi interfaces, which handles the flow of GPRS user data. For the control plane, all interfaces in GPRS network are involved because signaling message exchange takes places on every interface. The protocol layers of control plane on Um and Gb interface are shown in chart below: (as we can see, the layer3 protocol used between MS and SGSN are GMM and SM, which will be discussed in details in later chapters)
12/135
2005-09-12
S ig n a llin g P la n e s b e tw e e n S G S N a n d M S
MS
IP / X .2 5
G P R S a tta c h / d e ta c h s e c u rity ro u tin g a re a u p d a te , lo c a tio n u p d a te P D P c o n te x t a c tiv a tio n , m o d ific a tio n & d e a c tiv a tio n lo g ic a l c o n n e c tio n
a c k n o w le d g e / u n a c k n o w le d g e d p e e r-to -p e e r o p e ra tio n c ip h e rin g S A P s to h ig h e r la y e r (S N D C P , G M M , SM S)
Um
Gb
SGSN
G M M /S M
G P R S M o b ility M anagem ent and S e s s io n M anagem ent
G M M /S M
G P R S M o b ility M anagem ent and S e s s io n M anagem ent
LLC
L o g ic a l L in k C o n tro l
LLC
L o g ic a l L in k C o n tro l
R e la y RLC
R a d io L in k C o n tro l
BSSGP RLC BSS GP

BSS GPRS P ro to c o l
MAC
M e d iu m A c c e s s C o n tro l
MAC GSM RF
FR L 1 b is
FR
F ra m e R e la y
L 1 b is
P h y s ic a l Layer
GSM RF
The control plane protocols used in Gn and Gi interfaces are similar to the one used on transmission plane of these two interfaces. Comparatively the control plane protocol of Gc,Gr,Gf,Gd and Gs interface are different and they are close to the standard CCS7 signaling protocol, as shown in charts below:
G P R S S ig n a llin g P la n e s
SGSN GGSN
G PRSs p e c ific MAP e x te n sio n
G r,f,d H L R , E IR ,
Gc HLR
S M S -G M S C
SGSN
Gs
M S C /V L R
BSSAP+
BSS A p p lic a tio n P a rt +
MAP
M o b ile A p p lic a tio n P a rt
MAP
M o b ile A p p lic a tio n P a rt
BSSAP+
S u b se t o f BSSAP fu n c tio n a litie s
TCAP
T ra n s a c tio n C a p a b ilitie s A p p lic a tio n P a rt
TCAP SCCP
SCCP
S ig n a llin g C o n n e c tio n C o n tro l P a rt
SCCP M TP L3 M TP L2 L1
SCCP M TP L3 M TP L2 L1
M TP L3 M TP L2 L1
M TP L3 M TP L2 L1
13/135
2005-09-12
2.4 GPRS procedures basic

They layer3 protocols in the GPRS signaling plane contains two parts: GMM (GPRS Mobility Management) and SM (Session Management). The mobility management functions are used to keep track of the current location of an MS within the PLMN. The MM functions are located in MS and SGSN. The session management functions are used for activation, modification and deactivation of PDP (packet data protocol) contexts, i.e. of packet routing and transfer capability. The SM functions are located in MS, SGSN and GGSN. 2.4.1 Mobility Management States In IDLE state, the MS is not attached (i.e. detached) to the PLMN. There is no valid location or routing information for the MS available in SGSN or GGSN. Only GPRS subscription information in the HLR is available. User data transfer and paging of the GPRS MS are not possible. In READY state, the SGSN knows as well the cell selected by the MS within the RAI. If the MS enters a new radio cell, it updates the SGSN by signaling. Therefore paging is not necessary, if the SGSN wants to transmit packet data downlink. The MS may send and receive PDP PDUs (protocol data unit) in this state. However, the ready state does not imply any established physical connection between MS and SGSN. In STANDBY state, the MS is attached to the PLMN. The SGSN knows the routing area identity (RAI) of the roaming MS, but not the specific cell within that RAI. The MS as well as the SGSN may initiate data transfer at that state by PDP context activation (MS) or paging (SGSN). As a consequence, the MS and the SGSN will change to ready state.
Mobility Management State Transitions

GPRS Detach IDLE READY STANDBY timer expired PDU transmission STANDBY READY timer expired, Force to STANDBY, abnormal RLC condition STANDBY timer expired, Cancel Location PDU transmission READY STANDBY READY timer expired, Force to STANDBY, abnormal RLC condition
GPRS Attach
MM State Model of MS
Cancel Location, GPRS Detach IDLE
GPRS Attach
MM State Model of SGSN
The MM procedures are: GPRS Attach, GPRS Detach, Routing Area Update, Authentication, Identity Check, GPRS Paging.
14/135
2005-09-12
2.4.2 Packet Data Protocol states In INACTIVE state PDP context is not activated. There is no related routing context in the MS and SGSN/GGSN. The transition to the ACTIVE state is performed by the activation of a PDP context. The mobility management states STANDBY or READY at MS and SGSN are a precondition for PDP context activation. Otherwise transmission of PDUs from MS side will be rejected. In ACTIVE state the MS and SGSN/GGSN hold a routing context, i.e. a PDP context is activated. PDU transfer is possible. The mobility management states STANDBY or READY at MS and SGSN are a precondition for the state ACTIVE. With PDP context deactivation, the state of a PDP context changes to INACTIVE. In case of change to mobility management state idle (GPRS Detach or STANDBY timer expiry), all activated PDP contexts change to INACTIVE. The SM procedures are: PDP Context Activation, PDP Context Modification, PDP Context Deactivation.
Functional PDP State Model

Activate PDP Context
INACTIVE no routing and mapping of PDP PDUs possible no data transmission
ACTIVE routing and mapping of PDP PDUs possible location update takes place
Deactivate PDP Context MM state change to IDLE
The GMM/SM procedures are crucial to GPRS network performance. For example, if a GPRS end-user wants to have fast GPRS network access and data transfer, fast and successful GPRS attach and PDP context activation are always the prerequisite before any user data packet is transmitted or received. Therefore, performance of GMM/SM procedures shall be under careful monitoring and they are normally regarded to as major GPRS Key Performance Indicators. Some of the important GMM/SM procedures are illustrated in the followings subchapters.
15/135
2005-09-12
2.4.3 GPRS attach The GPRS attach procedure is used to establish a GMM context in the MS and the SGSN. After a successful GPRS attach, the GPRS mobile subscriber data are stored in the SLR (SGSN location register). After switching on, the MS initiates the attach procedure by the transmission of an Attach Request and the following procedures are shown in chart below:
16/135
2005-09-12
2.4.4 Routing area update Routing Area (RA) is defined as a group of cells which is the minimum range of GPRS packet paging. When the mobile phone moves from one RA to another RA, a Routing Area Update (RAU) will be initiated by the MS. One routing area is always a subset of one location area. By initial planning, the size routing areas are set same to the size of location areas. In the GMM STANDBY mode, Periodic RAUs will be initiated by MS under control of the periodic RAU timer. Normal RAUs (when MS moves into a new RA), comprise intra-SGSN RAU and inter-SGSN RAU. The message flow of an intra-SGSN RAU is shown in chart below:
BSS MS SGSN
R o u tin g A r e a U p d a te R e q u e s t (o ld R A I, U p d a te T y p e ) S e c u rity F u n c tio n s
R o u tin g A r e a U p d a te A c c e p t (P -T M S I, P -T M S I R e a llo c a tio n ) R o u tin g A r e a U p d a te C o m p le te (P -T M S I)
17/135
2005-09-12
2.4.5 Cell update In GPRS network, cell reselection instead of handover will be carried out when the MS moves from one cell to another cell. If the MS is in MM idle or standby state, the GPRS cell reselection is just the same as the non-GPRS mobile cell reselection. While in GMM READY state, the GPRS core network shall always be aware of the MS location on cell level and consequently, the cell update procedure will be initiated by MS which will generate some extra signaling on Gb interface as shown in chart below:
The detailed description for the above messages are: (1) A cell update takes place when the MS enters a new cell inside the current RA and the MS is in READY state. The MS performs the cell update procedure by sending an uplink LLC frame of any type containing the MS's identity to the SGSN. Towards the SGSN, the BSS adds the BVCI and Cell Global Identity including RAC and LAC. (2) When the SGSN detects the cell change, it send a FLUSH-LL PDU to the old BVC to initiate the following procedure: - at a cell change within one NSEI (e.g. one PCU represents one NSEI) and within one routing area, LLC-PDU(s) for a given TLLI stored at an "old" BVCI (corresponding to the old cell) are either deleted or transferred to a "new" BVCI (corresponding to the new cell) with which the TLLI is currently associated. (3) In response to a FLUSH-LL PDU the BSS shall send a FLUSH-LL-ACK PDU to the SGSN containing: - the TLLI received in the FLUSH-LL PDU and - an indication of whether the LLC-PDU(s) were "transferred" or "deleted". In case the SDUs were "transferred" the BVCI (new) IE shall be included. (4) Data transfer is continued in the new cell.
18/135
2005-09-12
2.4.6 PDP context activation The PDP context contains necessary information for GPRS data service like PDP address, QoS parameter and so on, accordingly the PDP context has to be set up in both MS and network via PDP context activation procedure before data packet can be transferred. PDP context activation can be initiated by either GGSN or MS depending on whether the first data packet transfer request comes from in downlink or uplink direction. The MS initiated PDP context activation process is shown in chart below:
MS
BSS SGSN Activate PDP Context Request (NSAPI, TI, PDP Type, ... ) GGSN
Security Functions
Create PDP Context Request (PDP Type, PDP Address, TID, ...) Create PDP Context Response (TID, PDP Address, ... ) Activate PDP Context Accept (PDP Type, PDP Address, TI, ...)
19/135
2005-09-12
2.5 GPRS radio network

With the introduction of GPRS into the GSM network, the radio network has introduced a lot of new terms, features and procedures even though there is not much hardware changes in the radio network. These new functionalities are completely different from the traditional ones of circuit switched services because they are adapted to be suitable to the packet data service characteristics. In order to understand better the GPRS data service and GPRS network, it is quite important and useful to get more knowledge of the GPRS radio network. Moreover, the GPRS radio network optimization is a dispensable part of the overall GPRS network optimization. 2.5.1 GPRS Um interface In the GPRS radio network, the involved interfaces and protocols are shown in chart below:
LLC RLC M A C G SM RF G SM RF L 1 b is RLC M A C G SM RF
LLC tr a n s m is s io n to S G S N
M S U m
BTSE A b is
BSC
The main functions of the protocol layers are: GSM RF: MAC: RLC: The GSM RF is the physical radio channel used to transfer packet data. The Medium Access Control provides the access to the physical radio resources. It is responsible for the physical allocation of a packet data channels (PDTCH). The Radio Link Control layer provides a reliable link over the air interface that fits the block structure of the physical channel. Therefore it segments and re-assembles the LLC frames. Additionally, it performs a sub-multiplexing to support more than one MS by one physical channel and the channel combining to provide up to eight physical channels to one MS. The Logical Link Control layer provides a logical connection between MS and SGSN even if no physical connection is established and the logical connection is identified by TLLI(Temporary Logical Link
LLC:
20/135
2005-09-12
Identity). The physical connection is set up by RLC/MAC layer when there is data to transmit. The MAC provides the establishment of a Temporary Block Flow (TBF), which is a physical connection between one specific MS and the network to support the unidirectional transfer of LLC Packet Data Units on packet data physical channels. TBF is an allocated radio resource on one or more PDCHs and comprises a number of RLC/MAC blocks carrying one or more LLC PDUs. A TBF to/from the MS is maintained only for the duration of the data transfer. When the data transfer is finished, the TBF shall be released and the radio resource shall be reserved for further allocation. Each TBF is assigned a Temporary Flow Identity (TFI) by the network. The TFI is assigned in a resource assignment message and will be stored in the first octet of the RLC/ MAC blocks. The importance of the TFI can be understood when considering the multiplexing of more MSs on the same PDCH. On uplink direction, GPRS radio network uses USF(Uplink State Flag) to control the multiplexing of different MS on uplink PDCH to prevent collision on uplink channel: Uplink state flag (USF) sent from the Packet Control Unit (PCU) is included in the header of each RLC/MAC block on a downlink PDCH. The MS monitors the USFs on the allocated PDCHs and transmits Radio Blocks on those, which currently bear the USF value reserved for the usage of the MS. The RLC/MAC has two working mode: Acknowledged mode and Unacknowledged mode. The transfer of RLC data blocks in the RLC acknowledged mode uses retransmissions of RLC data blocks. The transmitting side numbers the RLC data blocks via the block sequence number (BSN). The BSN is used for retransmission and for reassembly. The receiving side sends PACKET Ack./ Nack messages in order to request retransmission of RLC data blocks. As shown by the name GSM RF, the physical layer of GPRS radio network is the same with the physical layer of GSM network on Um interface. But the channel coding of GPRS Um interface is totally different from the channel coding of voice traffic: 4 types of coding schemes (CS1,CS2,CS3,CS4) are used by GPRS radio network. The coding technical details of these 4 coding schemes are listed in table below: Scheme Code rate Radio block Coded bits Punctur ed bits Data rate kb/s 9.05 13,4 15,6 21,4 Redun dancy
CS-1 CS-2 CS-3 CS-4
1/2 2/3 3/4 1
181 268 312 428
456 588 676 456
0 132 220 -
high middle low no
It can be seen that the higher coding scheme can provide higher data rate, whereas the drawback is clear too: higher coding schemes (e.g. CS4) are more vulnerable to packet loss on the Um interface due to poor radio link quality because it has less protection and redundancy.
21/135
2005-09-12
In other words, the GPRS data service can benefit from higher coding schemes only when the radio link quality (e.g. higher radio C/I ) is good enough. If the radio condition becomes worse (for instance lower radio C/I), the higher coding scheme will have high BLER(Block Erase Rate) and retransmission always take place in acknowledged mode. Consequently, the data rate of higher coding scheme might be even lower than the one of lower coding schemes.
2.5.2 GPRS radio network features in BR7 Siemens BSS network starts to support GPRS service from SBS BR5.5, but the GPRS network of BR5.5 works with quite limited GPRS functions. For example, in BR5.5 GPRS radio network supports only coding scheme CS1 and CS2. With the introduction new SBS release, GPRS functions have been enhanced and GPRS performance has been improved as well. In the following contents, some new GPRS radio features brought by BR7 will be introduced briefly. Support of CS3/CS4 and EDGE The reason why CS3/CS4 is not supported in previous release is that: CS3 and CS4 needs more than 16kbps transport capacity on the Abis interface while the previous release network can provide only 16 kbps Abis transport channel. Consequently, CS3 and CS4 implementation is only possible when new transport mechanism on Abis interface is realized. EDGE (Enhanced Data Rates for the GSM Evolution) brings higher data rates (up to 60kbps) by introducing 8PSK modulation to the GSM RF. Apparently the traditional Abis 16Kbps bottleneck must be broken in order to support EDGE. Thanks to the new Abis channel control feature FAAS , Siemens GPRS radio network is able to support GPRS coding scheme CS3, CS4 and EDGE starting from BR7. FAAS The Flexible Abis Allocation Strategy (FAAS) is introduced to allocate the Abis resources in a more flexible way. Without FAAS, one radio timeslot is statically associated to one 16Kbps Abis subslot (terrestrial traffic channel). The Abis subslots are configured by O&M procedures and stored into BSC database. While for voice and GPRS coding scheme CS1 and CS2 each time slot on the air interface is assigned a fixed 16 kb/s Time Slot on the Abis interface. With the introduction of coding scheme CS3 and CS4 (and EDGE coding schemes), characterized by data rates with more than 16 kb/s on the air interface, may require up to 5x16 kbit/s Abis capacity in order to transport an RLC/MAC radio block via the Abis interface. In this case a static Abis allocation would be highly inefficient. Hence Flexible Abis Allocation Strategy is introduced to allocate the appropriate number (one or more than one) of Abis 16kbps timeslots according to the service requirement. As soon as the radio timeslot is released the allocated Abis resources are set free again to Abis resource pool. In order to allocate more than one Abis timeslots for one radio timeslot, a new format of PCU frames, concatenated PCU frame, is introduced.
22/135
2005-09-12
The number of used concatenated PCU frames depends on the coding scheme used for radio block transmission: for GPRS when using the coding scheme CS1, a single PCU frame is needed, while the other coding schemes (CS2, CS3, CS4) require two concatenated PCU frames. In general the FAAS feature can improve the Abis resource efficiency by applying Abis channel dynamic allocation and reuse with the concept of abis pool. Accordingly the network operator can save costs for abis transmission especially when the higher GPRS coding scheme or EGPRS coding scheme is applied. Link Adaptation The Link Adaptation algorithm is performed dynamically in order to select the coding scheme that best suits a certain radio condition of the air interface. The switching points are determined via appropriate BLER measurements. The principles and the advantage for switching to a more robust coding scheme will be illustrated in the following example for the switch between GPRS CS-1 and CS-2. The figure shows, that an optimal data throughput can be achieved, if the system switches on the switching point =. Let us assume that currently the system works with CS-1. In case of improved radio conditions there is a switch to CS-2. In the other versus, if system works with CS-2, in case of worse radio conditions the system switches back to CS-1. Theoretically a switch (up or downgrade) exactly at = would give the best overall performance. However in reality this could result in a unstable systems behavior and so a hysteresis can be configured for the switching points.
As C/I values are difficult to estimate in a real network, the Link Adaptation algorithm is based on BLER measurements. Once the connection is established the block error rate (BLER) is continuously updated by checking if RLC blocks have been transmitted faultlessly. The ratio of repeated blocks versus the number of transmitted blocks in total is calculated. The goal of Link Adaptation is throughput maximization: The data transmission is started with an initial coding scheme set by the O&M. In case of bad conditions (noise, fading, interference) a more robust coding scheme is chosen soon. If conditions improve later on an upgrade of coding scheme (following the recommendations) is performed.
2.6 Gb interface
The Gb interface connects the BSC to the SGSN, transferring signaling information and user data. Several BSCs may be interfaced to one SGSN on the Gb interface. The main characteristics of the Gb interface are: a) resources are given to a user upon activity (when data is sent or received) and they are reallocated immediately thereafter; this is in contrast to the A interface, where a single user has the only use of a dedicated physical resource throughout the lifetime of a call, irrespective of activity; b) GPRS/EGPRS signalling and user data are sent in the same physical channel. No dedicated physical resources are required to be allocated for signalling purposes (like e.g. the A interface where SS7 links are used to transmit signalling between the BSC and the MSC). The protocol stack of the Gb interface is shown in chart below:
SNDCP
LLC Relay
GTP
LLC
BSSGP
RLC BSSGP MAC

New TRAU frame
L2 L1
NS L1
Gb
NS L1 SGSN
BSC/PCU
The various layers of Gb interface realize the following functions: Network Service The Network Service (NS) performs transport of NS Service Data Units (SDU) between the SGSN and BSS. The Gb interface is based on FR(Frame Relay) as specified in GSM 08.16. FR supports high rate data transmission with low delay. Frames of different sizes may be transmitted. FR performs congestion control and error detection, however error correction is not supported. Primary functions of the Base Station Subsystem GPRS protocol (BSSGP) are providing connection-less link between the SGSN and BSS (layer 2 level), providing tools for bi-directional control of the data flow and handling paging requests from the SGSN to the BSS. The Logical Link Control layer (LLC) provides logical links between an MS and a corresponding SGSN. The transport of both data and signalling is supported. SubNetwork Dependent Convergence Protocol (SNDCP) supports a direct peer to peer (i.e. point-to-point) communication between an MS
BSSGP
LLC
SNDCP
24/135
2005-09-12
and a SGSN. User data is transported from a network layer protocol, e.g. IP or X.25.
2.6.1 Gb physical connection In principle four types of configurations are possible in connecting the GPRS entities on the Gb interface: 1. a direct line (e.g. PCM30) between the two entities (static and permanent physical point to point connections); 2. an intermediate frame relay network; 3. Nailed Up Connection (NUC) through the MSC via a frame relay network; 4. NUC through MSC, without using an intermediate frame relay network.
In the early phase of GPRS network implementation, NUC connection had been adapted by operators because NUC is cost effective by taking advantage of existing transmission resource. The negative impact of the NUC is its limited bandwidth and long delay. For commercial use of GPRS the direct connection is recommended which can provide larger bandwidth and improve GPRS network performance to some extent.
25/135
2005-09-12
2.7 GPRS core network

General architecture of GPRS core network will be introduced in this chapter. 2.7.1 SGSN architecture In general, the SGSN architecture must be able to provide single as well as combined SGSN entities both in combination with the Circuit Switched System. The following entities are required for PO2.0/2.1: 2G-SGSN entity only 3G-SGSN entity only combined 2G/3G-SGSN entity All these entities are derived from a common architecture with a minimum set of different configurations. However in chapter only the general architecture of 2GSGSN will be introduced. The 2G SGSN is the Serving GPRS Support Node in GPRS and handles beside others the mobility of the subscribers and radio related protocol toward the mobile station. The system functions of the 2G SGSN are structured in functional areas as shown below.
2G-SGSN
Resource Management
Packet Handling
Accounting
Interception Handling
Configuration Management
Mobility Management
Handover Control
CAMEL Handling
Performance Management
Session Management
SMS Handling
Protocol & Interfaces
Fault & Maintenance Management
Figure 1 2G-SGSN System Functions / Functional Areas The functional area Packet Handling contains the functions: Packet Routing and Transfer Sub-network Dependent Convergence (SNDCP) Logical Link Control (LLC) including ciphering Flow Control Overload Control GPRS Tunnelling Protocol (GTP) Buffer Management on MP Message Handler MP-SP Protocol (MSP) Yet Another Tunnel Protocol (YATP)
The functional area Mobility Management contains the functions: Mobility Management (including security functions and paging) SGSN Location Register (SLR) GTP-MAP conversion The functional area Session Management contains the functions: PDP Context Handling Subscription Check Network Facility Check Context Table The functional area Resource Management contains the functions: management of internal resources for admission control management of packet flow contexts The functional area SMS Handling contains the functions: SMS handling in SGSN The functional area Handover Control contains the functions: SGSN change handling The functional area Accounting contains the functions: Accounting Management Charging Collection Record formatting for accounting and interception Support of a centralised as well as distributed charging gateway The functional area CAMEL Handling contains the functions: PrePaid Service VPN Service for SMS and Change of APN The functional area Interception contains the functions: Interception Management Interception Data Base (IDB) Delivery of interception related data and interception products The functional area Configuration Management contains the functions: Configuration Management Security Management The functional area Performance Management contains the functions: Performance Management The functional area Fault & Maintenance Management contains the functions: Interface to System Maintenance Interface to other maintenance areas, e.g. Trunk/Line Maintenance and SS7 Maintenance Central GPRS maintenance functions, e.g. audit framework, alarming framework error handling The functional area Protocols and Interfaces contains the functions: Gb Interface including BSSGP Gr Interface including MAP Gn Interface including IP and related protocols Gd Interface including MAP Gs Interface CAP Interface SS7, SCCP, TCAP, including MAP service provider The following figure shows the general architecture of the 2G-SGSN. it is derived from the general node GSN.
27/135
2005-09-12
s SGSN
Gb interface
Xp140 XP140
Gn interface
LIC:FR LIC:FR/*
SP:BSSGP SP:BSSGP
MSP YATP
MSP YATP
SP:GTP SP:GTP MCP
LIC:* LIC:*
MP:PD/SH MP:PD/SH
SSNC
MP:MM MP:MM MP:SM

LIC:E1 LIC:E1
(SS7) (SS7)
MP:RANAP MP:RANAP
MP:ACC MP:ACC
(SS7+ +Traffic) Traffic) (SS7
MP:SLT MP:SLT
LIC:ATM LIC:ATM
MP:ACCIO
(OAM dependent)
MP:OAM
Gr/Gd/Gs/Ge interface
Iu interface
Ga/charging interface
O&M interface (Q3, SNMP) Interception interface
Legend
light blue
SSNC parts XP140 parts
light yellow
atomic function running on a specific hardware platform
traffic path
continuous border dotted border
packet oriented system specific software parts software parts which may be shared by the ciruit switched system
light red
signalling path
Figure 2 General 2G-SGSN Architecture (unused parts of the SGSN architecture shadowed) First it consists of the SSNC common part which covers the basic functions like SS7 handling (MTP, SCCP) and accounting management (charging collection) and may be shared with the circuit switched part. Second there is the SGSN specific core part of the SSNC hosting the functions packet routing, session management, mobility management, resource management, interception handling and CAMEL support for PrePaid services. Last but not least the inter-connection toward the external world with respect to the traffic channel is represented by two XP140 each serving the two interfaces Gb and Gn. The XP140 covers also the interface specific protocol handling (BSSGP and GTP). Whereas Figure 2 shows the general architecture of the SGSN, Figure 3 depicts the architectural approach with respect to the Gb side of the SGSN. The main constraint, which influences the architecture, is the distribution of subscribers to MP:PD/SH as well as the flow control function on the Gb interface. To solve this issue the approach of the cluster solution is used where one or more MP:PD/SH serves a certain area which covers one or more routing areas entirely. If a subscriber enters the SGSN area his/her context will be assigned to one of the MP:PD/SH serving the originating routing area. As long as the subscriber roams inside the area which is served by the current cluster the context will be hold on the selected MP:PD/SH which was selected first. Otherwise, if the subscriber leaves this area but still roams within the SGSN area his/her context will be migrated to an MP:PD/SH of the cluster which serves the new area. Because each MP:PD/SH of one cluster must operate all cells of the served area it must be able to handle the flow control inside the cluster in a co-ordinated way between all MP:PD/SH of the cluster. It is assumed that a cluster size of three MP:PD/SH will be the upper limit due to coordination effort. In PO2.0/2.1 a cluster size of one is used because the expected
28/135
2005-09-12
traffic of one routing area is lower than the performance (throughput) of one MP:PD/SH.
MP:MM
MP:PD/SH
MP:PD/SH
MP:PD/SH
area 1
area 2
area n
Figure 3 SGSN Cluster Solution
2.7.2 GGSN architecture The Gateway GPRS Support Node is the node which interworks with the external IP network. The GGSN is structured in functional areas as shown below.
GGSN
Resource Management Configuration Management
Packet Handling
Accounting
Performance Management
Session Management
Protocol & Interfaces
Fault & Maintenance Management
Figure 4 GGSN System Functions / Functional Areas The functional area Packet Handling contains the functions: Packet Routing and Transfer Overload Control GPRS Tunnelling Protocol (GTP)
29/135
2005-09-12
Message Handler MP-SP Protocol (MSP) Yet Another Tunnel Protocol (YATP) The functional area Session Management contains the functions: PDP Context Handling Network Facility Check Context Table The functional area Resource Management contains the functions: mapping of QoS parameters management of internal resources like processor performance The functional area Accounting contains the functions: Accounting Management Charging Collection Record formatting for accounting Support of a centralised and distributed charging gateway The functional area Configuration Management contains the functions: Configuration Management Security Management Software Management The functional area Performance Management contains the functions: Performance Management The functional area Fault & Maintenance Management contains the functions: Interface to System Maintenance Interface to other maintenance areas Central GPRS maintenance functions, including error handling The functional area Protocols and Interfaces contains the functions: Gi Interface including IP and related protocols Gn Interface including IP and related protocols Ga Interface including IP and related protocols Figure 5 below shows the general architecture of the GGSN. First it consists of the XP140 part which covers the functions like session management, remote access server functions, accounting management (charging collection) and the interfaces to the GPRS backbone network as well as the external IP network. Second there is an OEM part which covers the Accounting File Server (AFS). The GGSN provides interworking with external IP networks. It may serve one or more ISPs or corporate IP networks with possible overlapping IP address spaces. One SP:ISP is able to serve different external IP networks and one external IP network may be served by one or more SP:ISPs.
30/135
2005-09-12
s GGSN
Gn/Ga interface
36140 XP140
MSP
LIC:FR LIC:*
SP:GTP SP:GTP
MSP YATP
SP:GTP SP:ISP MCP
LIC:* LIC:*
Gi interface
OEM
AFS AFS
charging interface
O&M interface (SNMP)
Legend
light blue
OEM parts XP140 parts
light yellow
atomic function running on a specific hardware platform
traffic path
light red
signalling path
Figure 5 GGSN Architecture Since GR3.1c, a new IP router based GGSN solution (CPG-3200, CPG-3300) is introduced with Cisco router.
Fig. Core PO release structure Detailed information about Cisco GGSN will be provided in later versions of this handbook.
31/135
2005-09-12
2.7.3 GSN Hardware components capacity assumptions For the GSN hardware configurations the following assumptions are done: the ASNcompact (5Gbit/s) is available including a new shelf the MPU-D is available for PO2.0, while the MPU-E is available for PO2.1 MP-OAMD (MP:ACCIO) can be used for charging data storage and IO processing from PO2.1 on 2 port STM1 LIC available for XP140 The following performance figures are used as base for the hardware configuration. Note, that throughput values are given for an average value of 300 byte packet size according to the PO traffic model (Sysload 1), while larger values given in brackets indicate throughput at 600 byte average packet size (Sysload 2). Component Performance PO2.0 with MPU-D MP:MM (including CAP) MP:PD/SH for UMTS MP:PD/SH for GPRS 150.000 subscriber 100.000 subscriber 110.000 sessions about 3 Mbit/s (6 Mbit/s) 50.000 subscriber 55.000 contexts 150.000 subscriber 1.600 MSU/sec 250.000 subscriber 22 Mbit/s (45 Mbit/s) 100.000 subscriber 30 Mbit/s (60 Mbit/s) 100.000 subscriber 18,5 Mbit/s (37 Mbit/s) Performance PO2.1 with MPU-E 230.000 subscriber 100.000 subscriber 80.000 sessions about 6 Mbit/s (12 Mbit/s) 50.000 subscriber 55.000 contexts* 300.000 subscriber 3.200 MSU/sec 250.000 subscriber 22 Mbit/s (45 Mbit/s) 100.000 subscriber 30 Mbit/s (60 Mbit/s) 100.000 subscriber 25 Mbit/s (50 Mbit/s)
MP:RANAP MP:SLT SP:BSSGP SP:GTP (2G-SGSN & GGSN) SP:GTP (3G-SGSN)

*
39.000 contexts with IP header compression
32/135
2005-09-12
3 GPRS parameters in BSC database

In addition to new added GPRS related hardware modules, BSC database has to be updated with new GPRS related objects in order to enable GPRS function in the network. In BSC database, the new GPRS objects are PCU, PTPPKF, FRL and so on. Under these objects, there are many new GPRS control parameters. Some old database objects, such as TRX has to include new parameters with the introduction of GPRS functionality.
3.1 SBS Database structure with GPRS related objects

Based on Siemens SBS BR7, all the GPRS new and related database objects are shown below in an object-tree structure:
33/135
2005-09-12
3.2 Introduction of GPRS radio parameters

In this chapter, main GPRS radio parameters concerned by optimization will be explained briefly based on their functional groups. Please be noted that the existence and definition SBS database parameters are closely related with the SBS software release versions, i.e. their existence and meaning may vary in different releases. The introduction in this chapter will base on BR7. For complete information about SBS database parameters, please refer to official BSC Database Parameter Description.
3.2.1 GPRS channel configuration parameters For packet switched service, four basic Packet Data Logical Channels have been introduced: Packet Broadcast Control Channel (PBCCH) Packet Common Control Channel (PCCCH) Packet Data Traffic Channel (PDTCH) Packet dedicated control channels (PTCCH and PACCH) Accordingly the GPRS channel configuration can be divided into two parts: control channel and traffic channel (PDTCH). In the early phase of GPRS deployment, there is normally no dedicated GPRS control channel PBCCH and PCCCH, which means that the GPRS control message (like system information) has to be sent within the old GSM control channel (like BCCH). The first reason of no PBCCH/PCCCH is to save radio resource consumed by GPRS when GPRS traffic was still too low in early phase; The second reason is that the old GPRS mobile phones mostly dont support PBCCH and they cant work properly in case of PBCCH existence. Following table lists the main parameters for GPRS control channel configurations. Object/ Package CHANNEL DB Name GDCH Range 1 ... 3 GPRSPBCCH, GPRSPCCCH, Null 0 .. 3 Meaning Is used to control if the channel is configured to carry dedicated control information for GPRS Indicates the number of blocks allocated to the PBCCH in the multiframe
PTPPKF
BSPBBLK
34/135
2005-09-12
Basically the GPRS traffic channel (PDTCH) configuration in radio cells shall base on the data traffic volume and circuit-switched traffic load; Furthermore the PDTCH is divided into reserved GPRS channel or dynamic GPRS channel. Reserved GPRS channel means that the channel is reserved for PS traffic only so that new CS channel request will be rejected if all the channel except reserved PDTCH are occupied. The reserved PDTCH is used to guarantee the radio channel availability for GPRS data service. Comparatively the dynamic PDTCH can be assigned on demand for either by PS or CS traffic. If the PS call ends in dynamic PDTCH, this PDTCH shall be released soon and be ready to be used for either PS or CS. Moreover, in BR7 GPRS channel allocation is based on Automatic Horizontal/Vertical Allocation and its controlling parameters shall be well considered during GPRS channel configuration. Following table lists main parameters of GPRS traffic channel configuration: Object/ Package TRX DB Name GSUP Range 0..1 False/True 1..254 sec. Meaning This attribute indicates if service is supported or not. the GPRS
PCU
TEMPCH
Timer Empty Default: Channel sec. PTPPKF
Specifies the time for releasing PDTCH if no activities are done. This timer should 90 prevent a faster PDTCH channel allocation/ release that can increase TDPC load It indicates the percentage of available channels dynamically dedicated by the system to GPRS traffic. Maximum Number of Packet channel reserved Defines the percentage of idle channel in the cell in order to trigger the switch between horizontal and vertical allocation. Defines the percentage of idle channel in the cell in order to trigger the switch between vertical and horizontal allocation. Threshold Idle Channel EU - indicates the threshold over which the upgrading of radio resources is enabled..
GPDPDTCHA 0...100; Default = 30 GMANPRES GASTRTHHV GASTRTHVH GASTRTHEU 0 190 Default = 0 0...100; Default = 10 0...100; Default = 20 0...100; Default = 20
PTPPKF PTPPKF PTPPKF
PTPPKF
3.2.2 Coding scheme and Link Adaptation parameter The induction of CS3 and CS4 in BR7 makes it possible to reach much higher data rate over air interface than before. On the other hand, CS3 and CS4 require much better radio link quality (radio C/I) than before. As the result, in order to optimize the coding scheme utilization and finally get real higher data rate for GPRS, the
35/135
2005-09-12
parameter of controlling coding schemes and link adaptation must be carefully optimized based on the radio quality of the target live networks. The coding scheme and link adaptation parameters are listed in table below: Object/ Package PTPPKF DB Name INICSCH Range Meaning Indicates the coding scheme to be used when the packet transfer starts It enables/disables the CS3/CS4 support at BSC level It enables/disables the CS3/CS4 support at cell level Enables or disables Link Adaptation algorithm. Defines the initial BLER estimation in a cell to be used in Resource Management to calculate the initial number of radio resources to be assigned to packet services when no 'historical' information about BLER is available. Specifies the filtering period for BLER calculation in downlink (expressed in number of radio block, one radio block = 20ms)
BSC PTPPKF PTPPKF PTPPKF
CS-1, CS-2, CS-3, CS-4 Default = CS-2 CSCH3CSCH4SUP FALSE, TRUE CSCH3CSCH4SUP FALSE, TRUE ELKADPT INIBLER FALSE, TRUE PER10, PER30, PER50, PER70, PER90 UNIT025, UNIT050, UNIT100, UNIT150, UNIT200, UNIT250, UNIT300, UNIT350, UNIT400 (unit= blocks) PER20, PER40, PER60, PER80,
PTPPKF
BLERAVEDL
radio
3.2.3 GPRS feature parameter In order to take full advantage of the new features and finally keep the GPRS network in optimum status, all feature related parameter must be set properly, otherwise the feature implementation might not naturally and surely lead to the network performance improvement. Besides the Automatic Horizontal/Vertical Allocation and Link Adaptation features that have been mentioned already, there are still many more GPRS features in SBS: FAAS (Flexible Abis Allocation Strategy) NCCR (Network Controlled Cell Reselection) Downlink TBF Delay Release
Forced Intracell Handover GPRS Resource Management
Part of the GPRS feature related parameters used in optimization are listed in table below: Object/ Package BSCB DB Name Range Meaning
ENFOIAHO
TRUE, FALSE Default = FALSE
The parameter enables Forced Intracell Handovers which are used to move Circuit Switched Calls to other free timeslots in the same cell in order to get adjacent timeslots for GPRS.
PCU
TIMEDTBFREL
0..49 (100msec Time delay TBF release. This attribute is stepsize) used to delay the release of a DL Default: 15 Temporary Block Flow (TBF). Default = Specifies the BSC channel allocation NO_PREFERENCE policy when different services are required.
BSCB
CPOLICY
BSCB
BTSM
BSCB
Default = Defines the downgrade strategy for NO_DOWNGRADE HSCSD and GPRS calls used during the resource reallocation procedure for TCH requests for single-channel circuit switched calls. GASTRABISTH Default: Control the switch from/to thresholdIdleAbis vertical/horizontal allocation strategy HV = 10 and the stop/restore of Abis upgrading thresholdIdleAbis due to Abis scarcity. VH = 20 It is composed of four fields. thresholdIdleAbis SU = 0 thresholdIdleAbis RU = 10 UPGRFREQ Default: Controls the time to pass between two Uplink = SEC1 consecutive radio resource upgrade Downlink = SEC1 attempts for packet services separately for the uplink and the downlink in steps of 1 second. NCRESELFLAG ENABLE, DISABLE Determines whether GPRS network Default = controlled cell reselection is enabled or DISABLE disabled.
DGRSTRGY
BSCB
37/135
2005-09-12
PTPPKF
CRESELTRHS OUT
50 100 Cell RESELection ThReSHold OUTput (unit= %) defines the % of 'occupation' threshold: Default = 85 when the number of active GPRS mobile is over this value, the network controlled cell reselection procedure can order gprs mobiles to reselect another cell until the number of GPRS active mobile is again under the threshold.
3.2.4 RLC/MAC parameter In the RLC/MAC layer, there are some timers and parameter to control the TBF related procedures. In some cases, these timers and parameter might be tuned to optimize the TBF procedure on RLC/MAC layer. Part of the timer and parameters are listed in table below: Object/ Package PCU DB Name Range Meaning
T3191
1 to 30 sec.
PCU
On network side Default: sec. T3193 1 to 42 Default = 4
This timer defines the waiting time for reuse of the TFI and USF after 5 having sent the last RLC block This timer defines the waiting time for reuse of TFI after reception of the final Packet Downlink Ack./Nack. from the MS. This parameter implements the threshold for not received RLC/MAC control message from the MS after sending a RRBP field on downlink. If this threshold is reached, the communication with the associated MS is broken This parameter represents the value of timer T3198, at MS side, expressed in number of RLC blocks This parameter implements the threshold for not received PACKET CONTROL ACK as answer of the PACKET UPLINK ACK/NACK. In case this threshold is reached, the communication with the associated MS is broken.
PCU
N3105
1..255
Counter on Default: 10 Network side
PTPPKF
BSCDVMA
1..15 Default: 15
PCU
N3103
1..255
Counter on Default: 10 Network side
38/135
2005-09-12
PTPPKF
PKTNDEC
0..7 Default:2
PTPPKF
PKTNINC
0..7 Default:2
This parameter defines the number of decrement for counter N3102 performed if the T3182 expires without reception packet Uplink ACK/NACK. It is used for RLC layer at MS side. This parameter defines the number of increment for counter N3102 performed after receiving packet UL ACK/NACK. It is used for RLC layer at MS side.
3.2.5 Other parameters Besides the parameters mentioned in above chapters, there are still many more parameters directly or indirectly related to GPRS radio performance. Especially some parameters are also involved in GSM voice network optimization because the GPRS data and GSM voice traffic in fact share the same radio network. In order to give brief hints on these parameters, some introduction are given below for reference: Cell reselection GPRS MS uses cell reselection procedure other than handover procedure when it moves around in the radio network. Without NCCR and PBCCH, the GPRS cell reselection is similar with the normal cell reselection procedure (C1,C2 algorithm) as voice MS in the GSM network. The only difference is that the CellReselectionHysterisis needs to taken into account in CellUpdate of GPRS while in GSM voice MS it works only in the location update procedure. In case of PBCCH presence in GPRS network, there will be much more GPRS dedicated control information broadcasted in PBCCH and the new GPRS cell reselection algorithms CS31 and C32 might be applied by lots of other new parameters. Radio coverage tuning Radio coverage is the fundamental basis for everything of the GSM network, accordingly the radio coverage enhancement or tuning is very important to radio network performance. As for GPRS, the aim of radio coverage optimization is to provide sufficient signal level for GPRS MS and furthermore manage the radio coverage in an optimum way to avoid chaotic radio signal overlapping, which will cause unnecessary ping pong cell reselection. The radio coverage tuning is mainly related to the base station power parameters, cell access parameter and other antenna physical parameters. Radio network quality Radio network quality is always one of the main targets of network optimization because radio network quality is important to both voice quality and GPRS data throughput. On the other hand, radio quality optimization is certainly a complicated and systematic work, which might involve a lot of efforts, e.g. internal and external interference reduction, radio coverage control, hardware malfunction removal.
39/135
2005-09-12
Consequently the radio quality optimization is always part of the GPRS optimization work and the traditional optimization methodology for GSM voice optimization shall be carried out.
3.2.6 Recommended GPRS radio parameter settings from TAC3 For each SBS release, TAC3 issues Recommendations for GPRS/EDGE related Parameter settings in BRxx. Updates can be found on IMS folder: GERAN data service. The complete path on IMS: Enterprise | ICM | ICM N | Products & Solutions | Technical Support | Mobile Radio Network | GERAN | GERAN_Feature and product description | Data services Here we will attach the document for recommendations for BR7.0:
G PR S_Par am et er _Se t t i ngs_f or _BR 7. 0.
4 Introduction of GPRS radio performance measurements

4.1 General concepts
Performance measurements are essential for the surveillance of a mobile network. They provide information about: The availability of network resources Traffic behavior and load Quality of Service These information can be used to identify failures and problem zones within the network which are not detected by fault management. This helps also to optimize and extend the existing network. Each measurement taken alone only gives elementary information. However, combined with other measurements and together with additional information (e.g. alarms or configuration data), they provide valuable overall information about what is going on in the network. The Siemens Base Station (SBS) provides a set of performance measurements for network surveillance. These measurements are partly compliant to GSM 12.04, partly Siemens proprietary. The measurements can be grouped into several categories as follows:
40/135
2005-09-12
Fig. Categories of SBS Performance Measurements As part of the SBS performance measurements campaign, the introduction of GPRS has brought new GPRS performance measurements counters. Since BR7.0 the number of GPRS related counters has been greatly enlarged which gives the GPRS radio optimizers much more possibilities to quantify the GPRS radio network performance.
41/135
2005-09-12
4.2 Structure of Scanners

The processes which accumulate data and assemble it for collection and inspection are called measurement jobs or shortly, scanners. Each scanner will collect data at a particular frequency known as the granularity of the measurement. The scanners have following structure: Every measurement job (scanner) consists of one or more measurement types. The measurement type(s) contained in a scanner may apply to one or more network resources of the same type, e.g. a measurement job may be related to one or several BTSs. A measurement type has a fixed number of counters, there are precisely defined trigger events for a counter.
Following picture gives the structure of scanners:

Object Class Measurement type Counter
...
... ...
Measurement type
Counter Counter
...
Counter Object Class Measurement type Counter
...
...
Measurement type
Counter Counter
...
Counter
Fig. Scanner structure According to their nature, different measurement types can be subdivided into different administration groups or we can say, object classes. The administration groups of BR7.0 is given below in the table below:
42/135
2005-09-12
Tab. Overview of administration groups of BR7.0 GPRS related performance measurements counters in BR7.0 are categorized into following groups: packet channel configuration, packet flow establishment, active packet flow, packet flow release, Gb interface and cell reselection. Moreover the processor measurements provide the PCU load statistics.
Fig. GPRS related performance measurements
43/135
2005-09-12
4.3 SBS optimization work flow of performance measurement

The following flow chart gives a general work flow of SBS performance measurement which is commonly followed by SBS optimization engineers. 1
Create scanner Modify scanner if necessary
Uploading measurement data (Automatic or Manual)
Exporting measurement data into ASCII file
5 6 Import ASCII measurement files and BSC databases into BATRANA2004

Upload BSC database and convert it to ASCII file by DBAEM.
Analyze SBS performance with BATRANA2004
To complete procedure 1 to 5, we need to operate the BSC via Radio Commander, there are two basic approaches for operating Radio Commander:
Working under RC GUI Executing CLI scripts
Radio commander GUI offers user-friendly graphical approach for OAM operations, which is suitable for simple and less numbered tasks. CLI (Command Line Interface) offers Unix based interface for executing so called CLI commands, which is suitable for advanced and large numbered tasks, such as a batch task.
44/135
2005-09-12
For detailed information about step 1 to 5, please refer to BSC or RC operation manual, step 6 and 7 will be explained in chapter 7.2: GPRS radio measurement analysis with Batrana2004.
4.4 Main GPRS radio network performance KPIs (Based on BR7)

In BR7 there are much more GPRS performance counters available and therefore GPRS KPIs can be more abundant than earlier releases. For simplicity and completeness, here we will simply insert << PM: SBS Key Performance Indicators (optional issue) A30808-X3247-B912-1-7618 >> issued by Network Engineering. Please refer to chapter 14 for GPRS related KPIs.
KPIs_SBS_BR_7.pd
It shall be aware that all available counters can be used in flexible ways to build up operator or project specific KPIs in order to best match customers requirements and concerns on network quality. Other Reference Documents: For detailed information about SBS counters and message flows, please refer to following documents: PM:SBS Counter (A30308-X3247-L40-2-7618) PM:SBS Message Flow, A30808-X3247-L41-2-7618
45/135
2005-09-12
5 Introduction of GPRS core performance measurements

5.1 General concepts
Within the GPRS core network a main processor (MP) platform and a server processor (SP) platform are distinguished. The MP platform provides for the core functions such as accounting, SS7, mobility management, session management, resource management and mobile related packet handling. It is managed by means of the Q3 interface. The SP platform mainly controls the interconnection towards the external interfaces with respect to the traffic channel, and the required protocols for these interfaces. It is managed by means of SNMP. Performance management handles the count of events, administration of measurement parameters and storage and transfer of measurement data. It is used to monitor the load of the system, to detect problems with regard to erroneous or inadequate dimensioned system components or for statistical reasons. MP platform: On the MP platform we have the STATS functionality as a generic solution to the problem of implementing performance management counters. Every 15 minutes (the granularity period) the SGSN collects performance data from its counters with respect to: The system load The processor load The protocol load Managed object classes and their attributes are used on the object-oriented interface (Q3) between the SGSN and the switch commander (SC) in the operations system (OS). Before measurement data from the counters can be collected, the measurement jobs that generate the required data must be activated in the SGSN. In object-oriented terms, this corresponds to the instantiation of managed objects that model the measurement process. As long as the measurement job is active, the measurement results are generated every 15 minutes. The results of each period are recorded in a report notification that is forwarded to a measurement log file in the SGSN. These measurement reports can then be transferred to the switch commander. Packet-switched performance measurements are grouped in object classes. The object classes are called currentDatas from the STATS point of view. The following figure gives the structure of PM object classes implemented on the MP platform:
46/135
2005-09-12
SP platform: On the SP platform the STATS concept is not supported. All counters are defined in management information bases (MIBs), and are handled in a de-central way by each application. Every 15 minutes the switch commander reads all these variables using SNMP. The main functions of performance measurement: administration, collection and storage are carried out by the switch commander, which offers a graphical user interface (GUI) that allows access to the combined CMIP/SNMP interface. In general, the picture below shows the exact processors from which different types of measurements are generated.
47/135
2005-09-12
Note: MPU load types for PO: SM, SLT: SS7 HANDLING PD/SH: PACKET DISPATCHING AND SESSION HANDLING RANAP: RADIO ACCESS NETWORK APPLICATION PART MM/CAP: MOBILITY MANAGEMENT AND CAMEL ACC: ACCOUNTING OAM, OAMD: O&M FUNCTIONS (ACCOUNTING IN OAMD) MP:OAM is also known as MP:SA
5.2 Performance data handling In the figure below, it can be seen that measurements are started from Switch Commander and collected later via different interface protocols by the SC with PDC (Performance Data Collector) application.
The PDC is responsible for Fetching measurement data from the defined NE Editing the collected data (e.g. insert counter names) Form a unique output design for all types of measurement The measurement files are now ready on the SC to be copied for post-processing. Transfer the files to your PC via LAN, USB hard disk, etc. You can easily check the correct format of the file after transferring to your PC, the file must be readable in a normal text editor and must not contain any unreadable characters. Example of a SSNC/SGSN raw file format: HEADER;vcpuCurrentdata;26;EWSDV13; ;0;SCANNER; ;0;vcpu00mpTotalAverageLoad;vcpu00mpTotalMaximumLoad;vcpu01SystemIntegrit yAverageLoad;vcpu01SystemIntegrityMaximumLoad;vcpu02sltPrhAverageLoad;vcpu 02sltPrhMaximumLoad;vcpu03oamAverageLoad;vcpu03oamMaximumLoad;vcpu04ca
llpAverageLoad;vcpu04callpMaximumLoad;vcpu05SystemUpgradeAverageLoad;vcpu 05SystemUpgradeMaximumLoad;vcpu06ccs7NmAverageLoad;vcpu06ccs7NmMaximu mLoad;vcpu07vcpuSpare07AverageLoad;vcpu07vcpuSpare07MaximumLoad;vcpu08v cpuSpare08AverageLoad;vcpu08vcpuSpare08MaximumLoad;vcpu14backgroundAver ageLoad;vcpu14backgroundMaximumLoad;vcpu15auditAverageLoad;vcpu15auditMa ximumLoad;vcpu16idleAverageLoad;vcpu16idleMaximumLoad;vcpu17SystemBaseLo adAverageLoad;vcpu17SystemBaseLoadMaximumLoad vcpuCurrentdata;EWSDV13;2001/10/11;05:30:00;SGSN4.1;900;FALSE;SCANNER;2 6;997;999;0;0;0;0;15;500;0;0;0;0;0;0;0;0;0;0;0;0;0;3;960;976;22;43 vcpuCurrentdata;EWSDV13;2001/10/11;05:30:00;SGSN4.8;900;FALSE;SCANNER;2 6;997;999;0;0;0;0;1;10HEADER;switchingElementMeasurement;4;EWSDV13; ;0;SCANNER; ;0;numInMSU;numLostMSURoutErr;numOutMSU;numTransitMSU;-1;1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1 switchingElementMeasurement;EWSDV13;2001/10/11;05:30:00;SGSN4;900;FALSE; SCANNER;4;10;0;0;0;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;1;-1 HEADER;sgsn2gCamelCurrentData;3;GR20; ;0;SCANNER; ;0;attCamelDialogues;unsuccCamelDialoguesSsf;unsuccCamelDialoguesScf;-1;-1;1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1 sgsn2gCamelCurrentData;GR20;2001/10/11;05:30:00;SGSN4.1;900;FALSE;SCANN ER;3;0;0;0;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1;-1 NETAN is a Siemens official software tool for core network measurement postprocessing. It is a software installed on PC. At the moment, V7 is highest version which is mandatory to process PS core release PO2/GR3.
5.3 Data required for core measurement evaluation (Based on PO2.1)

5.3.1 SGSN evaluation This chapter will describe the actions on the switch commander that are necessary as preparation for the Performance Data Collector PDC. This summary of Measurement counter commands is an extract from Task Manual TML.. For complete information of these commands, please refer to the original Task Manual. All listed commands are for measurements in the 36190, the measurements in the 36140s are always running; i.e. no actions have to be made (except in the PDC itself of course). 1. MP load evaluation For every entity SGSN and every MP we require for load evaluation - CR MPLDRP / CR MPLDME As the load budget for the MP types is different, we need also the LOG file of - DISPMP
49/135
2005-09-12
(in ASCII format, not in UNICODE option in SC; see page below)
2. C7 Link Load distribution For analysis of the C7-Link Load we need for each SSNC the files from - CR MTPSCHD / CR SIGLME In case you would like SCCP information please start also measurement - CR SPMESCHD / CR SPME Additionally we need for every SSNC the MLOG files for - DISPSIGDL; - DISPSIGLINK:Display names=Yes; Attention: After every reconfiguration of C7 links (database changes) the C7 measurement must be stopped and restarted. Otherwise the counters for modified or new links are not activated! 3. For every SGSN we need the data from commands - CR GMESCHD / CR GMEALL (2G) - CR UMESCHD / CR UMEALL (3G) 4. Create the corresponding Event log for the data collector, which is used to store Event Reports. - CR SGSNMELO In the picture below, you will find a graphic display on which commands for which measurement. The commands with greater font are the commands we need and
request for NETAN evaluation, the commands with smaller font are the rest, i.e. commands which might be used for specific measurements.
5.3.2 GGSN evaluation There is no need to start GGSN measurements on the switch commander, as the counters on the 36140s are always
5.3.3 Set up the PDC When the PDC is installed, first of all the network elements to be monitored have to be set. And the path (directory) has to be defined, where the collected data are to be stored. Then the procedure is as follows (simplified): Start Programs Switch Commander Performance Data Collector PDC Performance data collector window opens Select configuration Q3 scheduling tasks (new window opens) | insert Fill in time schedule etc. Repeat same procedure (last 2 lines) for SNMP, FTP, FTAM Start measurement
51/135
2005-09-12
5.4 GPRS core network KPIs

The Siemens official key performance indicators for GPRS core network performance evaluation are implemented in the NETAN tool. Due to the reason of intellectual property, the complete set of GPRS core KPIs will not be listed here. For complete information about NETAN as well as its KPIs, please contact SAG ICM N OP MS ND6. Here only the major GPRS KPIs from 36190 measurements are given in the following table:
Registration all registered subscribers sgsn2GmmafCurrenData meanNbrHomeSub+meanNbrVisitingNatSub+ meanNbrVisitingForeign
CAMEL subscribers meanNbrCamelSub number of attached subscribers meanNbrOfAttachedSub number of subscribers with activated [sgsn2gSm]meanActivePdpPerSgsn contexts number of activated contexts SMS SMS MO attempts SMS MO success rate SMS MT attempts SMS MT success rate [sgsn2gSm]meanActPDPContextHighPriority+ meanActPDPContextLowPriority sgsn2gSms attSmsMo 100*succSmsMo / attSmsMo attSmsMt 100*succSmsMt / attSmsMt
PDP Contexts: sgsn2gSm activation attempts with dynamic attActPdpContextDyn addresses "dynamic addresses" success rate 100 * succActPdpContextDyn / attActPdpContextDyn activation attempts with static addresses attActPdpContext - attActPdpContextDyn 100 * (succActPDPContextHighPriority+ "static addresses" success rate succActPDPContextLowPriority - succActPdpContextDyn) / (attActPdpContext - attActPdpContextDyn) Throughput 2G LLC frames sgsnLLC DL sent llcFramesSent UL received llcFramesReceived SNDCP N-PDU packets sgsnSNDCP DL ack. DL'NpduAckPkts DL not ack. DL'NpduNackPkts UL ack. UL'NpduAckPkts UL not ack. UL'NpduNackPkts SNDCP N-PDU Octets sgsnSNDCP DL ack. DL'NpduAckOctets DL not ack. DL'NpduNackOctets UL ack. UL'NpduAckOctets UL not ack. Routing area Attach attempts GPRS only GPRS only success rate Detach attempts from MS attempts from SGSN "Detach from SGSN" success rate RAU in SGSN attempts (non combined) UL'NpduNackOctets ra2gGMMAF attGprsAttach 100 * succGprsAttach / attGprsAttach attGprsDetachMs attGprsDetachSgsn 100 * succGprsDetachSgsn / attGprsDetachSGSN attIntraSgsnRaUpdate
52/135
2005-09-12
success rate RAU between SGSN attempts (non combined) success rate PS Paging paging attempts 2g successful paging 2g
100 * succIntraSgsnRaUpdate / attIntraSgsnRaUpdate attInterSgsnRaUpdate 100 * succInterSgsnRaUpdate / attInterSgsnRaUpdate raPsPaging nbrPsPagingMes2G 100 * succPsPaging2g / nbrPsPagingMes2G
53/135
2005-09-12
6 Introduction of common GPRS network optimization tools

Tools used in GPRS network optimization are normally divided into two parts: Data collection tools (the left part in the picture below) Data processing & analysis tools (the right part in the picture below) Furthermore, more specific tool could be identified according to the data type it collects or processes. The data types involved in GPRS optimization normally has three types: performance data, field test data(Um interface data) and signaling trace data.
Firstly, we shall notice that the post-processing & analysis tools have fixed relation with the data collection tools, e.g. BATRANA2004 is used for BSS measurement data from OMC-B and Deskcat is used from Um test data from TEMS investigation. Therefore, before a data collection tool will be decided to be used, the availability of its related post-processing & analysis tool shall be checked. Secondly the GPRS optimization tools are not necessarily new tools and they are normally the upgraded version of classical GSM optimization tools that are enhanced with the embedded new GPRS functions. Therefore, in order to fully support GPRS measurement processing and analysis, please be careful with the version of tools you are going to choose. For example, for SBS performance measurement analysis, you should choose at least Batrana200 V3.1 or better Batrana2004; for drive tests, at least TEMS investigation v3 shall be chosen.
54/135
2005-09-12
Ethereal is a tool normally used to monitor the Ethernet interfaces, which is widely adopted in TCP/IP based data communication networks. However, it can be used in GPRS because its able to decode GPRS specific protocols such as GTP, GRE, etc. on Gn and Gi interfaces. It shall be noted that there are many choices for Um interface measurement tools: Ericsson TEMS Investigation, Agilent E64xx, Nemo Outdoor and so on. These tools are used for a common purpose but they are developed to be working with different GPRS test mobiles: Ericsson TEMS Investigation: Ericsson MS like R520sc Agilent E64xx: Sagem MS like OT190 Nemo Outdoor: Sagem or Nokia MS like 6310 Etc. Also please be aware that the performance, or more precisely KPI values of a GPRS network highly relies on the performance of GPRS mobiles. Field experiences also show that sometimes even GPRS test mobiles may have software bugs (e.g. incompatibility with GSM/GPRS specifications) which results in very bad or inaccurate measurement results. Therefore, keep dont forget to update your GPRS test mobiles regularly with the latest released firmware from the test mobile suppliers.
7 Main GPRS optimization skills

7.1 End-to-end GPRS performance tests
7.1.1 General introduction End-to-end GPRS test means the tester simulates the behavior of a real end-user of GPRS data service. The purpose of an end-to-end test is to collect the GPRS performance data in the field and get to know the real feeling of the end-users when they use the GPRS data service. In addition to the performance statistics data collected by OMC, the end-to-end performance test is the only way to collect the GPRS performance data and prepare important data resource for optimization input. As we know, end-to-end test is also made for voice service since long time ago. Voice CQT(call quality test) and DT(Drive test) are the most popular end-to-end tests during GSM voice optimization work. Comparatively the GPRS end-to-end test is much more complicated because the data service has much more applications and diversified performance concerns. Moreover, the specifications (like test method and test items) of GPRS end-to-end test might vary in different projects because different operators have their own requirements and concerns. Whereas the basic and most important GPRS performance focus is almost common: delay and throughput. In this document the most important and common GPRS end-to-end tests will be described and explained. GPRS end-to-end test includes two types: Stationary test and Drive test. Stationary test means that the test is made in some fixed locations such as office or hotel lobby. The purpose of the stationary test is to simulate the stationary GPRS end users; While Drive test means the test is made in a moving vehicle and the test route is designed beforehand to cover some important roads and places of GPRS
service area. The purpose of a drive test is to simulate the moving GPRS end users and check the GPRS data service performance under moving condition. Besides the different test purposes and methods of Stationary test and Drive test, the test items and test equipment might be different too. e.g. stationary test might require testing with mass market GPRS terminals from different suppliers, even GPRS PCMCIA card.
7.1.2 GPRS end-to-end test tools The tools used in GPRS end-to-end test include followings: GPRS test mobile (e.g. Ericsson R520sc) GPRS test software (e.g. TEMS investigation v3.2.5) Laptop PC with Microsoft windows Operation System like Windows2000 Test mobile cable (for connection between mobile and laptop PC) GPRS service enabled SIM card Dual-ports I/O card (for providing COM ports form PC side) FTP server (either public FTP server or dedicated test FTP server) FTP client software installed on the laptop PC Other GPRS mobile phones or GPRS PCMCIA card if required by test specifications Optional accessories are PC charger, test mobile charger In case of Drive test, more accessories are needed GPS (e.g. Garmin GPS) Car power converter Paper map of the target city One main characteristic of GPRS test tools is that the test mobile cable shall have two ports that are data port and trace port separately. The data port is the port for data modem connection and the trace port is for the measurement data collection. Accordingly the laptop PC will need two COM ports for the GPRS test mobile. Furthermore if the GPS is used in Drive test, totally 3 COMs shall be provided by the laptop PC. This explains why the dual-ports I/O card is needed for GPRS tests. The following chart shows the trace and data ports of SagemOT96 test mobile and the 3 COM ports of laptop PC.
GPS Trace
Data
Laptop PC
Agilent E7478A with SagemOT96
56/135
2005-09-12
7.1.3 Preparations before test Before going outside to the target test place, some work has to be carried out to check and verify whether the test tools are well prepared and surely ready for the infield test, because the GPRS test is much more complicated than voice test regarding the PC setting. Additionally some operations during the GPRS end-to-end test shall be practiced in office as well. The detailed preparation work are depicted and explained below: Install necessary software The necessary software in laptop PC include: o GPRS test software (TEMS investigation or Nemo outdoor or others) o FTP client software (like WS_FTP or CuteFTP or others) o Check the ping command is available in laptop PC (it is installed together windows OS by default. If the OS works well, ping shall work) Moreover, in the FTP server PC, the FTP server software (like SerU_ftp) shall be installed and some FTP account shall be created for test purpose by someone in charge. Install the data modem in laptop PC Because the test mobile acts as a data modem in the laptop PC, the modem must be installed in the PC firstly before the data application could be started from the PC. Remember that the data modem uses the data port of the test mobile. Depending on what test mobile is used, the appropriate modem driver shall be installed according to the relevant guide information from the mobile provider. Chart below (left part) gives a sreen shot after the modem is successfully installed in PC. Please note that the APN used for GPRS connection shall be set in the modem properties/advanced/Extra initialization commands as shown in the chart below (right part).
Set APN here
Modem of Ericsson R520
Properties page of modem Add the GPRS dial-up connection in laptop PC

57/135 2005-09-12
After the modem is installed, the GPRS dial-up connection using this data modem shall be added. The user name and password of GPRS dial-up connection are network dependent and shall be provided by network operator. The phone number of the dial-up connection formats in the way like *99***1#.
Dial-up connection property page
Verify setting and connections After the dial-up connection is added, following steps shall be carried out to verify the correctness of setting and make sure the tools are ready for test: 1. Verify the dial-up connection: connect the test mobile to the laptop PC and dial the connection. If the dial-up connection setting is correct, the dial-up connection shall be setup successfully. 2. Verify the FTP account: after dial-up connection is set up, run the FTP client software and try connecting to the FTP server according to the given FTP account. Because the file uploading and downloading need to be done as part of the test items, the existence of certain size (like 500kBypte) file shall be checked in the FTP server(for download) and laptop PC(for upload). Finally remember to try downloading and uploading some small size file in order to check the FTP account has been given the rights of reading and writing the FTP server. 3. Verify the test software: since the test software normally has license control, it is recommended to check the license availability and cables physical connection before the test in field. If the GPS will be used for DT, the GPS connection to the test software shall be verified too (there might be no latitude/longitude information available in house, but the physical connection can be verified by successful detection of the GPRS in test software) After all the setting and connection have been verified successfully, the tools are completely ready for GPRS end-to-end test. During the test, the GPRS test software will display and record lots of radio information of the GPRS test mobile over Um interface, and its log file will be postprocessed and analyzed later in GPRS optimization. Accordingly the using of the test software is important to the optimization engineer and will be discussed briefly in following chapters. As the matter of the fact that TEMS investigation is the most popular and powerful Um interface test software, so only the using of TEMS
58/135
2005-09-12
investigation is describer in this document. As for the other types of test software, please refer to their manuals.
7.1.4 Using TEMS investigation TEMS investigation GSM from Ericsson had been widely used for GSM network Um interface measurement in GSM voice network optimization. After the upgrade to version 3.0, TEMS investigation GSM started to support GPRS measurement and has become one of the most popular test tools for GPRS Um interface. One of the main features of TEMS 3.0 or later is that its license control uses hardware license which is embedded in the test mobile like R520sc, while the earlier versions like TEMS invesitigatioin2.0 used software license in form of registration code. Moreover the mobile used in TEMS investigation has to be the special test mobile released by Ericsson other than ordinary commercial mobiles. Since the TEMS investigation GSM 3.0 or later are just the successive product of the earlier version for GSM optimization, the using of TEMS investigation will be very easy for the optimization engineer who get used to the earlier versions. If the test engineer is a new user to TEMS investigation, he shall firstly read the TEMS investigation manual in order to grasp the knowledge of basic operations. In this chapter, only the special operations of GPRS test with TEMS investigation will be described for reference: Hardware connection In TEMS, the connection cable of GPRS test mobile R520 has actually two cables bundled together, one is the data cable (marked with data label) and the other is trace cable (marked with TEMS label). As result, the cable marked with data shall be connected to the dial-up modem COM port and the other one marked with TEMS shall be connected to the TEMS investigation software for Um interface message collection. On the other hand, actually both of the two cables can be recognized with name MS and DC separately in TEMS investigation. MS stands for the TEMS cable and DC stands for the data cable. Whether connecting the DC port in TEMS investigation or not depends on how the test engineer wants to operate the dial-up connection: if the engineer wants to operate the dial-up connection inside of TEMS investigation (by the advanced operation - command sequence), he has to connect DC to TESM investigation; if the engineer wants to operate the dial-up connection in MS Windows other than in TEMS investigation, he doesnt need to care the DC cable in TEMS investigation. Hardware connection in TEMS investigation could be configured by two ways: manual or automatic. If the port number of the hardware is known, the connection could be easily set up by click the Add equipment button and then the hardware and its appropriate port number shall be selected in ComboBox. Picture below demonstrates the operation of manually addition of R520m on port COM3 as MS.
59/135
2005-09-12
If the port numbers of the hardware are unknown, the hardware connection could be set up in an automatic way: by clicking the Identify Equipment button, TEMS investigation will search all the COM ports available in the laptop PC and list the detected hardware to the user. After the hardware is successfully detected by TEMS investigation, the connection between TEMS investigation and mobile hardware will be set up by clicking the connect button. In the status bar on the bottom of TEMS investigation window, the status of hardware connection could be checked anytime. GPRS information display In the TEMS investigation, the measurement information could be viewed either in a real time mode or in playback mode. Real time mode means the information is displayed at the same time when the test is made in field, while playback mode means the information is displayed when the log file is replayed in TEMS investigation during post-processing and analysis phase. How the information is displayed in TEMS information is up to the engineer because the display in TEMS investigation is flexible to be changed (This is one of the main advantages of TEMS investigation). Furthermore the display method could be saved in a tig file via save workplace as menu command and be loaded for later use. Actually the default display setting of TEMS investigation after the first installation has already contained a lot of useful information items displayed in different windows. In this chapter, the display based on the default setting of TEMS investigation is discussed firstly, followed by the explanation how to change the display. Basically, the information shown in TEMS investigation has been split into 5 sheets by default and each of them has different types of information as implied by its worksheet name. The 5 worksheets of TEMS investigation are: RF contains the RF information like receive signal level Data contains the GPRS data service related information like throughput Signaling contains the signaling messages like layer3 message Map contains the GPS and map-based information like geographic location
Ctrl&Config contains the control and configuration options like handset control The GPRS information in TEMS investigation is mainly displayed in the Data worksheet. In Data sheet, there are 8 windows displayed by default. It is not always necessary to view all the 8 windows, so the engineer could close some window as he wants. If he closed one window and want to view it again later, he could open it again by two ways: access by menu command on main window or by open the relevant items in navigator. Both of the menu command and navigator are shown in chart below:
Menu command on main window Items in navigator
Among all the 8 available windows, 2 windows (HSCSD Throughput and GPRS Prediction) could be omitted in GPRS end-to-end test and the other 6 windows all have important GPRS information as explained below: GPRS PDP Context: information details of GPRS PDP context is shown in this window, including QoS parameter, APN and PDP address(the IP address of the test mobile assigned by GPRS network). GPRS Status: basic GPRS status information, including GMM state, RAC, coding scheme utilization ratio and others GPRS Throughput: throughput and error rate information, including RLC and LLC layer throughput in kbps and decode error rate in %. Notes that the RLC decode error rate in TEMS is identical to BLER (Block Erasure Rate) Data Session: data session information on application level, including application layer throughput, application layer data amounts and others Data Timeslots: the detailed information of radio timeslots used for GPRS data service, including the downlink and uplink timeslot number, timeslot type and others Data Line Charts: Line charts are the windows in TEMS where information is displayed in graphic way and Data Line Charts in Data worksheet is where the GPRS data information is displayed in graphics. By default some basic GPRS throughput information are shown in this window, including LLC throughput and others. As we know, the main advantage of TEMS investigation is that the information display can be configured flexibly and easily to meet different engineers willingness
61/135
2005-09-12
and preferring. The test engineer can always change the display in three different aspects: Worksheet level: add, remove, rename worksheets and others Windows level: add, remove or move windows in worksheet and others Information items level: add, remove or modify the information items in windows Following chart demonstrates the Data worksheet display during the GPRS end-toend test: the original 5 windows (GPRS PDPcontext, GPRS status, GPRS Throughput, Data Timeslots and Data Line Chart) are kept and the rest of the orginal windows in Data worksheet are closed. In order to view the RF channel information and message information at convenience, 3 extra windows (Serving+neighbours and Current channel windows from /Presentation/GSM and Layer3 windows from /Presentation/Signalling) are added into data worksheet. As we can see from the picture below, the GPRS test mobile is assigned IP address 10.185.133.145 with peak throughput 64kbps(from GPRS PDP Context window); the current radio time slots occupied by GPRS are TS1,TS2,TS3 in downlink and TS2 in uplink(from Data Timeslots window); the current serving cell for the test mobile is cell with CI 40952 and LAC 39232(from current channel window).
The information items level configuration can be accessed by right click anywhere in the area of the certain window to activate its popup menu, then click the properties on the bottom of popup menu. In the properties page related to the certain window, it is able to change (add, delete, modify) the information items shown in the window. Following picture shows an example to change the property of the Data Line Chart for GPRS test. In this example, the information items shown in the chart1 are being edited after clicking the edit chart button of the property page.
In summary, TEMS investigation can provide a lot of useful GPRS information and their display appearances can be configured easily depending on the engineers interest. The purpose of change the display property is always to make the good information presentation help us somehow for problem checking. Advanced operations Besides the basic operations, there are still more advanced operations in TEMS investigation that might be needed during test. For example, it is planned to test the performance of GSM900 network only while the operators network is a dual-band network. In this case, the test mobile has to be forced to lock on GSM900 network only. Then the band lock operation has to be made by ticking the band control options of the property page of Handset control window in Cont&Config worksheet. As for GPRS data test, the major advanced operation is to create the command sequence. Firstly, it shall emphasize that the command sequence is not mandatory for GPRS because its actions could be made alternatively out of TEMS investigation. For example, the ping command could be made in command mode of MSwindows, the FTP file get command could be made in FTP client software. Consequently the GPR end-to-end test could be finished without command sequence of TEMS investigation. Secondly, we shall notice that the command sequence is used to create a sequence of test commands and it will ease the work effort. Whereas in the
TEMS investigation GSM software, so far the GPRS attach and detach command have not been supported yet, which means the attach and detach test must be made in other place like HyperTerminal. Following chart give an example that the ping command is added in command sequence window in the Ctrl&Config worksheet.
It is recommended that the command sequence function used only by advanced user of TEMS investigation and the new user can neglect it. All the GPRS end-to-end still can be done in alternative ways (to be described in the chapter test items). Log files operation Log file from TEMS investigation is a file with extension name .log and this file is created with the Ericsson proprietary format. Accordingly this file is not readable with normal text editor tool like Wordpad. In order to read the measurement information, either dedicated analysis tools or log file format converting tools shall be used. The dedicated analysis tool for TEMS investigation log files are following: TEMS investigaton, Deskcat, Actix Analyzer. The log file format convering means that the log file can be processed to be readable file format. This processing function is provided by TEMS investigation itself and it could convert TEMS log files to plain text format. Picture below shows the example that the log file is going to be exported to Text file format and the information elements included in this target text file has been selected already.
64/135
2005-09-12
Format selection Selected items
After the log file is successfully exported(the default extension name for exported file is fmt and it is automatically put under a newly created export folder), the exported file could be opened by MS Excel or other tools and all the information is readable now
7.1.5 Test items Normally the test items of GPRS end-to-end test shall be defined already before the test work is carried out in field, the definition normally is the negotiation result of the network operator and equipment vendors. In this chapter the most popular test items of GPRS optimization work are depicted and explained.
7.1.5.1 GPRS Attach Test result GPRS attach successful rate: the number of the successful GPRS attach / total number of GPRS attach attempts GPRS attach average time: the average duration between GPRS attach request and GPRS attach accept
Test Operation In order to make the GPRS attach test, some number of GPRS attach requests need to be made manually and then the network response to these requests (accept or
not) will be checked subsequently. After successful GPRS attach, GPRS detach will be forced to the test mobile so that the mobile will be ready for attach again. Generally there are normally two ways to make the GPRS attach request: HyperTerminal or Test software. In Hyperterminal, special AT commands can be used to make GPRS attach and GPRS detach: GPRS attach (AT+cgatt=1) and GPRS detach (AT+cgatt=0). As for the test software, Nemo supports the GPRS attach command but TEMS investigation does not. During the GPRS attach test, the test software is always running to record the singalling messages, from which the GPRS attach successful rate and average time will be calculated. The relation of the AT command and relevant Um signaling message is shown in picture below:(the message is examples extracted from TEMS investigation)
As test reference, normally 10 times of the GPRS attach will be tried for one target place of stationary test. If the network is working properly, 100% GPRS attach successful rate are expected.
7.1.5.2 PDP Context Activation Test result PDP context activation successful rate: the number of the successful PDP context activation / total number of PDP context activation attempts PDP context activation average time: the average duration between message Activate PDP Context request and Activate PDP Context accept
Test Operation Following the same way of the GPRS attach test, some number of PDP context activation requests need to be made manually and then the network response to the requests (accept or not) will be checked subsequently. After successful PDP context
66/135
2005-09-12
activation, PDP context deactivation will be forced to the test mobile so that the mobile will be ready for PDP context activation again. Please note that the PDP context must be activated with a certain APN, for example APN name internet or WAP. Consequently the APN used in the PDP context activation test has to be defined in accordance with test specification. APN setting of the PDP context activation could be defined in the modem property (refer to the chapter preparation work before test) or AT command. The AT command used in HyperTerminal is shown in the same chart of GPRS attach chapter. For PDP context activation and deactivation, the AT commands are AT+cgact=1,1 (using predefined PDP context with CID=1) and AT+cgact=0 separately. Normally test software supports the PDP context activation and deactivation. In TEMS investigation, the data dial up and hang up commands shall be included in the command sequence to make the activation and deactivation. Besides the HyperTerminal and Test software, the PDP context activation request can also be made by manually dial the dial-up connection in MS Windows and the deactivation can be made by manually disconnect the dial-up connection. As reference, normally 10 times of the GPRS PDP context activation will be tried for one test place of stationary test. If the network is working properly, 100% PDP context activation successful rate is expected. To calculate the test results (successful rate and average time for both attach and PDP context activation) is part of the work after the test is finished and the test software has successfully recorded all the Um signaling message in log files. The simple and direct way is to check the message flow and make calculation by playing back the log files in the test software itself, e.g. in the layer 3 window of TEMS investigation. While it is recommended that some post-processing tool shall be used to ease the calculation work especially when there are too many log files to be calculated.
7.1.5.3 Ping Test result Ping successful rate: the number of ICMP echo request with successful reply / total number of ICMP echo requests Ping average time: the average duration between sending ping echo request and receiving its echo reply
Test Operation Ping is the traditional command popularly used to evaluate the network connectivity in IP-based network. In the GPRS end-to-end test, it is used to test the network RTT(Round Trip Time) delay. Since delay becomes one critical performance concern for data service, the Ping test result is one important part of the GPRS network performance. In brief, ping command initiates the echo request by sending a certain size packet from one end (resource address) to the other end(target address). After receiving the packet, the other end(target address) will reply to this echo request by sending the same size packet back to initiated end(resource address). If the reply packet arrives the resource address successfully, the total time it takes the packet to travel in the whole round trip will be given as the RTT delay. If the reply packet cant arrive
67/135
2005-09-12
the resource address successfully in the predefined timeframe (this is timeout), one ping failure will be counted. The parameter options description of the ping command and one example of the ping command in GPRS test are shown in the picture below:
As shown in the above command example ping 192.138.228.1 n 2 l 500 w 5000, the necessary parameters need to be set in the ping command are: -n how many echo request will be tried in this ping command -l the size of the packet. The bigger, the longer the delay is -w the timeout window in ms for the echo replay. The bigger, the more possibility of successful echo reply The execution of the ping command could be made in either MSwindows command mode (as shown in the picture above) or the test software. Moreover the test result of ping can be found at the end of ping command in the command mode window. In practical GPRS end-to-end test, the packet size normally is set to be 32Bytes or 500Bytes, and the echo request to be 10 times. As for the target IP address, it could be defined in the test specification.
7.1.5.4 FTP Test result Download throughput: the downloaded file size / downloaded session time on FTP layer Upload throughput: the uploaded file size / uploaded session time on FTP layer
68/135 2005-09-12
Test Operation Because throughput is the most important concern of the data end users, the throughput is always one of the main test items and optimization target as well. FTP test is designed exactly to evaluate the data transfer throughput on application layer in GPRS network. With the windows-based FTP client software like WS_FTP, the FTP downloading or uploading session could be easily operated. The FTP server could be either a dedicated PC located in GPRS backbone network or a public FTP server in internet, but the certain size of file for FTP download has to be prepared in the FTP server. Moreover, the FTP commands Get and Put can also be used to make the file downloading and uploading in either Ms Windows command prompt or the test software. After the FTP session is finished, the FTP client software will give the duration of the FTP session, File Size and the FTP throughput at the end, subsequently these will used for the FTP test result. Normally, the file size used for downloading could be 500kBytes or even bigger as 2Mbytes, while the uploading file is much smaller to be 100kBytes. The main reason is that the GPRS MS can support much more radio timeslot in downlink direction than uplink. In other words, how many radio timeslot will be assigned for the FTP session is one decisive factor for FTP throughput.
7.1.5.5 WAP Test result WAP page display successful rate: the number of successful WAP page display / total number of WAP attempts WAP first page display time: the duration between starting WAP attempt and the WAP first page display on Mobile screen
Test operation As we know, WAP service has become one of the most popular services of mobile data users. Actually WAP service based on CSD(Circuit Switched Data) technology was already available before GPRS implementation. After GPRS was introduced into mobile network, WAP service based on GPRS, so called WAP over GPRS, has become the dominant way how WAP data is conveyed through the mobile network. In order to test the performance of WAP over GPRS, some WAP tests are adopted in GPRS end-to-end test. Basically the WAP test operation shall be made manually with the mobile phone (it could be the test mobile or other commercial GPRS mobiles), and some WAP address (for example wap.yahoo.com) shall be given by test specifications and be pre-configured in the mobile phone. APN setting shall be taken care because the APN for WAP service is different from the APN for other applications like WWW surfing. Thats say that the correct WAP service APN shall be pre-configured in the WAP setting of the mobile phone as well as the WAP address. When the WAP page access is attempted by push some button(or menu) of the mobile phone, the PDP context will be activated firstly and then the WAP page data will be downloaded from the WAP address. If the PDP context could not be activated or the WAP page downloading could be finished in certain timeframe, WAP service failure will be displayed in the mobile screen. If the PDP context is activated
successfully and WAP page downloading is finished in time, the duration from staring WAP service attempt and WAP page complete display will be measured manually by clock. After the WAP page is completely displayed and the test result is recorded, the WAP page shall be exited firstly and then the WAP menu shall be exited (PDP context will be deactivated) subsequently. By this the mobile phone returns to the initial status and be ready for next WAP attempt. Besides, in mobile phone there is normally some cache to store the WAP page data, and this cache shall be cleared every time before the next WAP service attempt is started. Normally 10 times of the WAP service attempts could be tried for one test place, and then the successful rate and first page display time will be measured. Please note that the WAP performance includes the performance of both the GPRS network, as the bearer layer, performance and WAP network.
7.1.5.6 Drive test In drive test, the test mobile is moving around in the GPRS network and some performance aspects related to moving condition is possible to be observed as part of the GPRS overall network performance. Since the FTP and WAP services are the most popular application in GPRS network, normally FTP file download/upload or WAP service shall be carried out during GPRS Drive test. In addition to the FTP and WAP test result described in above chapters, some more test result from Drive test might draw the attention too: Test result Average cell reselection interval: the average time intervals between cell reselections in the whole process of GPRS Drive test
Time interval between two cell reselections
70/135
2005-09-12
GPRS drop: the number of GPRS service drop during the Drive test. GPRS service drop means GPRS data transferring is no more possible because the unexpected PDP context deactivation happened abnormally. This performance indicator is similar to Call Drop of voice Routing area update successful rate: the number of successful routing are update/ the total number of routing area update attempts Bad coverage ratio: the measurement sample of bad signal level / all the measurement samples of MS. It is used to evaluate the radio coverage of the network after the bad signal level threshold is predefined, e.g. less than 95dBm
Test operation The operation of GPRS Drive test is just quite same with the stationary test of FTP test or WAP test. The differences are that the log file is mandatory to be recorded and GPS is highly recommended to be used.
71/135
2005-09-12
7.2 GPRS radio measurement analysis with Batrana2004

7.2.1 Introduction Cell or BSC based GPRS radio network measurement analysis can help an optimizer to detect problems of packet data services in the radio network such as: PDCH/Abis pool congestion, high packet retransmission rate due to interference, high TBF breakdown ratio, etc. Due to the reason that a huge amount of data is generated by the radio network which needs to be processed offline, a powerful and easy-to-use software tool is a required. This tool must be able to support a huge database and still be able to make statistics with acceptable speed. Batrana2004 is an official Siemens BSS performance measurement analysis program developed by the Network Optimization Group of Siemens Shanghai Mobile Communications, Ltd. In the following part of the document, we use word Batrana to refer Batrana2004 for simplicity reason. From the traffic measurement result export file from Radio Commander (RC), or the binary file directly uploaded from LMT and/or Radio Commander, Batrana extracts useful information out and stores them in a project specified database using MySQL database engine. After all the necessary information has been imported, reports concerning different packet service performance aspects, such as throughput, retransmission rate, coding scheme usage rate, etc. can be calculated in a form of table and graph, each of which can be exported for reporting. Batrana2004 can process BR60 and BR70 data. At the starting of Batrana, the user will be asked to choose a release.
7.2.2 Using MySQL Batrana2004 is using MySQL server (http://www.mysql.com) as its data engine to store the data. When Batrana user is importing data, either network configuration information or traffic measurement result information, or others, all of them are stored in a MySQL server. Later when different analysis is applied, necessary data are retrieved from that MySQL server and certain calculation is done upon them. If you are familiar with SQL command, you can also write your own formula and apply them. One MySQL server can have many databases. For Batrana users, each database is regarded as a project, such as Shanghai, Beijing, UMTS etc. So later when we say project Shanghai, it also means a database named Shanghai in the desired MySQL server.
72/135
2005-09-12
7.2.2.1 The Client/Server structure MySQL has two working structures, client/server (C/S) structure and embedded structure. In C/S structure, a stand-alone MySQL server is installed either in your local PC (for user 1) or in a remote computer connected through network (for other users). One advantage of such structure is that the project data can be accessed by other Batrana user remotely through network.
In the above picture, all notebooks have installed Batrana. A MySQL server is also installed in the notebook of user 1, for him, he can access to the project Berlin, and other users can also access to project Berlin through network. Meanwhile, all the users can access to project EDGE in server 2 and/or project Shanghai in Server N through network, where no Batrana but a MySQL server is installed. And Project EDGE can also be transferred from Server 2 to Server N, either with Batrana build-in function, or with direct command of MySQL server. For the C/S structure, there will be an administrator of the MySQL server, who is in charge of the user access management. At the connection to the server, a login name and password will be required for security reason.
73/135
2005-09-12
7.2.2.2 Embedded Server Together with Batrana installation, an embedded server of MySQL will also be installed on your PC. Embedded Server does not provide the multi-user and network function. It only access to the database in the same pc, and only allows one user to access to the data. In the following configuration, user 2 can access to the project Hannover of his embedded server, and other users in other PC cannot access to that project through network. Meanwhile, user 2 can neither access to project EDGE in server 2 through network. It is the same for user N of his project Frankfurt.
For user 2, if he want to access to project EDGE, then he must choose the C/S mode in Batrana. Note: If you have already installed MySQL server in the same PC, and want to use the Embedded server, you must first stop the background service of MySQL. In the above example, for user 1, because there is a MySQL server installed already in his PC, where also Batrana is installed, he MUST stop the MySQL server if he want to use the embedded server delivered together with Batrana.
74/135
2005-09-12
7.2.3 Getting Data from project 7.2.3.1 Getting BSC database Before importing the traffic measurement, necessary network configuration information must be stored in advance in the database. The user should import the BSC database first, which is in ASCII format. To prepare the ASCII file, please following the following procedure: Save the BSC database into the hard disk of BSC. Upload the BSC database from BSC into RC or LMT. Use DBAEM to convert the binary database into ASCII format. Rename the ASCII file (*.asc) into the format of [bscname].asc Each BSC should have its own ASCII file, and the name of the ASCII file is regarded as the BSC Name. For example, the BSC who has its ASCII database file name as Shanghai1.asc is regarded as BSC named Shanghai1. It is important that later the BSC traffic measurement from that BSC should also be named as Shanghai1, which is to have the one-to-one matching. It is suggested to have the BSC name planned at very beginning of the project, with meaning context, and be kept constant throughout the whole project.
7.2.3.2 Prepare Cellname information To retrieve data for a particular bts, the target cell can be addressed by BSCName-BTSM number-BTS number Lac-ci Cellname As for a particular cell at a fixed physical position, the first two information might have been changed due to cut-over, re-configuration, new radio planning etc, Batrana have introduced a new method of bts addressCellname. The cellname is a unique string defined by the user to distinguish a cell. With this function, the data from the same cell can be correctly retrieved even there are change of lac-ci and/or the bscname-btsm number-bts number for that cell in different project history period. The cellname information should be prepared and imported before the import of the traffic measurement. An ASCII text file with such following format should be prepared with TAB spacing: BSC BTSM BTS Site_Name CELL_ID LAC CI BSC051 27 0 railway_1 G051101 6208 4353 BSC051 28 0 railway_2 G051102 6208 4354 In the above example, btsm24:bts0 in BSC named BSC051 is named as railway_1, and btsm28:bts0 in BSC named BSC051 is named as railway_2.
The CELL_ID and LAC, CI information is kept for future usage. For CELL_ID we suggest you to put the related cell identifier in the radio planning information.
7.2.3.3 Scanner Creation At this version of Batrana, it is required to have the scanner granularity set to 60 minutes in order to guarantee the accuracy and correctness of the formulas. Such limitation will be removed in a future release. A scanner creation tool are included in Batrana to help you automatically create the scanners after BSC database file (asc file) analysis. It is recommended to include all measurement types of an object class during scanner creation. What happens if some scanners are not created? If some scanners are not created and therefore the collected measurement file doesnt contain any data in the related counters, Batrana will put NULL values for these counters and the formulas which contain these counters will give NULL results as well. In the program nothing will be displayed. So if you have some formulas which give no result, please check whether all the related counters are available in the measurement files and whether the scanners are created in a correct way. Note: Such behavior is a standard SQL behavior for NULL data handling. Due to the nature and project experience, Batrana do not assign any default value for missing counters.
7.2.3.4 Getting BSC traffic measurement Batrana supports measurement files in both binary format and the ASCII format. Binary file The binary files are either directly uploaded from BSC through LMT, or directly retrieved from RC. In some cases the binary files may be compressed by the system, you should uncompress them first before importing them into the Batrana database. Some system setting may add different headers to the binary file and change their format. Please choose the correct type during the import. ASCII file ASCII files are generated by RC. There are two kinds of them: ascii file and exp file. We recommend to use ascii file instead of exp file, which is much better for the processing speed. The ascii file is automatically generated by RC and normally stored under RC directory: /commanArea/OMC/DATA/OMP/PMG/asciiPmg The format of the ascii file can be set by command setMeasConvMode.
76/135
2005-09-12
The exp file is generated by the manual exporting or executing of command exportMeas on the RC.
7.2.4 MySQL connection and configuration 7.2.4.1 MySQL Configuration The first step of using Batraan2004 is connecting to the MySQL server. In order to make a successful connection, correct configuration is very important.
The Server/Client setting is only valid if you have installed a C/S MySQL server on your PC. If you have not installed any C/S MySQL server, then you can only specify a directory for data storage of your embedded server in the datadir field. The basedir is a folder where your MySQL server is installed, and the datadir is the folder where your MySQL stores the data. Here are some suggestion of the datadir setting: 1. We recommend to use a disk formatted in NTFS type. Using a disk in FAT32 type is of no problem, but using NTFS you will have less limitation on file size to store huge project or long-term data. 2. Choose a disk, which is not used frequent by other software. Please avoid to use a disk where Windows stores temporary files. 3. Ensure there will be enough free disk space left for MySQL to hold data.
77/135
2005-09-12
Default-character-set If you intend to have language special characters for BSC name and/or cell name other than the 26 English characters, then for a correct sort result, you should set the default-character-set to a correct value. Latin1: Latin1_de: Gb2312: Gbk: Big5: the general and default setting, fit for most of the cases with German character support with Chinese character set in gb2312 with Chinese character set in gbk with Chinese character set in big5.
7.2.4.2 MySQL connection By clicking connect, an connection will be established between Batrana and the selected MySQL server. If it is not an embedded server which is going to be connected, a login procedure is required. The user will be asked to input his username, password and the desired server.
With a successful connection to the server, Batrana display the name of the connected server in the status bar on the bottom left. If the embedded server is connected, the program displays localhost:
7.2.5 Project Operation After a successful connect to the MySQL server, you can open a project on that server. Each project is a database on that server, which is in a subfolder with the name of the project inside the datadir specified in the configuration. In each project, internal tables are used to store different kinds of data. Different BR releases have different tables. Batrana automatically deal with the correct tables
78/135
2005-09-12
according to the BR release you have selected at the start of Batrana tool. Therefore it is allowed, for example, for a project Shanghai to have both BR60 and BR70 data. When opening the project, Batrana checks whether all the desired internal tables exist. If not (e.g. incorrect internal table version, one or more internal tables were deleted by the user from Explorer, etc.) , these internal tables will be created again. In the latter case, the existing tables will be flushed. What is Project | save as? When your database is in the same PC as Batrana, which means you are using a Embedded server, or a C/S server which is installed in the same PC, you can save the currently opened project into a new name. All internal files and tables in the current project will be copied to a new project, which is a new subfolder in the datadir of the chosen MySQL server. How to transfer the project from one PC to another one? The simplest way of transferring MySQL database from one server to another server is just copying the whole subfolder of the database to the datadir of the other server. How to copy a remote project into my own PC? There can be cases that you want to copy a project from a remote server to your own PC, such as copying data from a central database server, so later you can make analysis along without network connection.
7.2.6 Import Data 7.2.6.1 Import BSC database The BSC database should be imported prior to the traffic measurement import. Necessary information will be extracted and stored into project database.
7.2.6.2 Import Cell name To establish the link between cell name and the bts numbering, the cellname information should be imported. Each time there is a change in such linkage, then the updated cellname information must be imported before the traffic measurement data is imported. Each cell name must be unique all through the life period. Such uniqueness is mandatory for Batrana to keep the trace of the old non-existing cell. For example, a cell of BSC001 of BTSM:0/BTS:0 was named railway_1, later when BSC001 of BTSM:0/BTS:0 was renamed to Station_1, the old railway_1 will still exist and retrievable. The cellname table can browsed as following:
7.2.6.3 Import BSS Measurement Both binary format and ascii format of the BSC measurement file can be imported. Certain rules apply for the file name extension: The binary file must have the extension .log or without any extension. The ascii file must have the extension of .ascii, or exp or without any extension. All the files whose extension is not in such form will be ignored. The options for importing:
BSC Name During importing, the BSC name of each traffic measurement file must be determined. There are two ways:
Source file name as BSC name: each file name without its postfix is the BSC name. For example, BSC001.ascii has BSCName as BSC001. Path name as BSC Name: The traffic measurement files from the same BSC are put into a folder with the name of the BSC. For example, all the traffic measurement file inside the folder /BSC001 will be regarded as from a BSC named BSC001.
Open Mode Open Source Files: The Batrana User must select the traffic measurement files one by one. Open Directory: The Batrana user can select a directory, all the files inside this directory will be scanned and opened according to their file name extension. The subfolder in that directory may also be scanned.
7.2.7 Network Performance Analysis Network performance analysis are based on imported traffic measurements which are stored in MySQL database. Queries are performed via standard SQL language.
7.2.7.1 Report Start Report function by following menu item:
The report window is split into five areas:
81/135
2005-09-12
Formula: Select KPIs to be used for analysis. There are three formula groups: Standard: The Siemens standard formula. Raw data: the imported traffic measurement data. User define: the formula defined by user. In the formula area, you can GPRS_PERFORMANCE_ANALYSIS find predefined GPRS formulas in group
Object: Select objects are to be analyzed. Time: Select the time duration within which the performance will be analyzed. There are three options: Whole Day: All the data in the same day are considered Busy Hour: Only the data in the most busy hour, which has the highest TCH traffic load, are considered Time Range: Only the data within the defined hour range are considered Chart: Graphical presentation of KPI curves Table: Numerical presentation of calculated KPI values How to make user defined formulas:
82/135
2005-09-12
Although Batrana provides many predefined formulas, there may be cases where the user wants to make performance analysis based on their own formulas. It is possible to do so via the following menu item:
The Formula Definition window are split into six areas:
The formula is defined by standard SQL commands. All user defined formulas are saved into an .INI file under the Batrana installation folder: BRxx_UserDefine_Formula.ini. This file can be copied to backup the user definitions. There are three types of data tables: Data table: to store imported traffic measurement data, e.g. the bts table above Configuration table: to store network configuration information, e.g. the tgtbts table above Temporary table: to store temporary data during performance analysis, e.g. the adjc table above. Search and filter function are provided while viewing these tables. Note: For accuracy reason, some counters are multiplied by 100 (or other numbers) and stored as an integer rather than a floating number.
7.3 GPRS core measurement analysis with Netan2000

7.3.1 Introduction to NETAN2000 NETAN2000 is a powerful and easy to use traffic measurement Analyzer, fully programmed with BORLAND C++. There is no need for any additional programs like MS EXCEL to run NETAN2000. NETAN2000, the NETwork ANalyzer for SIEMENS MSC/VLR/HLR SGSN/GGSN/SSNC is the tool if you want to be informed about the Grade Of Service of the network look for problems with the help of traffic measurement detect traffic bottlenecks as soon as possible make a traffic forecast, essential for demand-oriented expansion be informed about the actual traffic model and 2G/3G
All these tasks can be performed by using the integrated traffic measurement functions in a SIEMENS 2G or 3G core network. The various measurements can deliver more than 2000 different counters, which is of course a huge amount of information. To manage this information easily, to differ between important and unimportant counters and to set the different counters in a correct relation the NETAN2000 tool was developed. In particular, NETAN2000 offers the following evaluation for GPRS / UMTS: MP load: The SSNC consists of several Main Processors (MP). A main processor can handle several software processes, such as packet processing, protocol handling, performance measurement and OAM functions. Here the load of each MP is shown. The MP load Evaluation is based on the measurement CR MPLDME as well as some DISP commands. C7 Link Load: The CCS7 Link Load analysis calculates the load of each CCS7 link. The result shows the design specific usage of the SSNC with its allocated HW-units (MP) and the relevant load in Erlang. The C7 link load Evaluation is based on the measurement CR SIGLME as well as some DISP commands. Grade of Service: The 'Grade Of Service' (GOS) contains the most significant values, which indicate the performance and quality of a SGSN. The GOS is based on the SGSN measurements CR GMEALL. The measurements are subdivided in Registration (SLR) SMS-CAMEL (SMS and CAMEL subscribers) PDP Context (act, deact, mod, upd Contexts) Throughput (2G) (LLC and SNDCP frames, packets, octets) Overload (discarded and rejected packets) Routing Area Update (RAU) 2G (cluster change) Routing Area (attaches and detaches) Miscellaneous (Gr/Gs if, authentications, identifications)
Traffic model: The Traffic Model Evaluation contains the most significant values, which are important for the capacity of a SGSN.
7.3.2 NETAN Basics 7.3.2.1 Starting traffic measurement jobs Following SSNC/SGSN measurements can be evaluated within NETAN2000: 36190 Measurement: CR MPLDME Create MP Load Measurement CR SIGLME Create Signaling Link Meas. CR SPME Create SCCP Measurements CR SMEALL Create SGSN Measurements All
PM SCANREP SGSN dd0hhmm.spf
36140 Measurement: 36140 in SGSN: PM_36140_SGSN_dd0hhmm.spf 36140 GGSN: PM_36140_GGSN_dd0hhmm.spf These measurement files shall be transferred to a PC where NETAN is installed.
7.3.2.2 Quickstart Copy the measurement files of the NE's you want to analyze into a directory on your PC. Choose 'START EVALUATION' from the 'EVALUATION' menu. The 'Availability Check' window will appear.
Select the 'Source Directory' where the measurement files are stored. Select a 'Result Directory' where you want to store the result files.
85/135
2005-09-12
Alternatively, the box Map RESULT to SOURCE directory can be ticked which automatically sets the result directory the same as the source directory. Paths in 'Source Directory' and 'Result Directory' can contain a maximum of 50 characters.
The availability check will be performed. This means all files in the Press specified 'Source Directory' will be scanned for the following information: - Measurement file, which can be used for an evaluation? - Date and NE name of the measurement file - All necessary measurement files included? After finishing the availability check the window will change and present all possible evaluations:
Here you will get a choice of all possible evaluations found in the specified directory. You can now select all evaluations, all evaluations of a common type or only a single evaluation. When you have chosen the evaluations you want to perform click on the button.
86/135
2005-09-12
In the 'EVALUATION WINDOW' you can observe the progress of the current evaluations. After finishing the evaluation you can close the window and open a result file by choosing 'OPEN RESULT FILE' from the 'FILE' menu in the main window.
7.3.3 NETAN2000 Result Files 7.3.3.1 General Handling Main features: The result files contain lots of predefined formulas Several different Result Files can be opened at the same time in different windows The user can easily switch between the Result Files and arrange all the files horizontally, vertically or cascade them Result Files can be printed (one page or all pages) Result Files can be saved to Bitmap format The size of the Result Files on screen can be changed by the user to his requirements (resolution of the monitor) The user can select the Busy Hour, Busy Quarter, or any hour/quarter of the day Additionally, the evaluated interval can also chosen to be the whole day. In this case an average of all parameters over the whole day are calculated. The Time Line Chart functions can show each counter and formula in a result file
There are two basic working modes in the NETAN tool: the summary evaluation mode and the detailed graphical analysis.
87/135
2005-09-12
The summary evaluation mode (see figure below) is the default working mode after opening a result file. All calculated KPIs values are shown in tabular formats according to the selected measurement interval.
Fig. The summary evaluation mode
and switch to graphical mode. Double click on blank graph You can press button and the pop-up checkbox allows you to display a predefined KPI from Statistics(Calculated) or from Counters.
88/135
2005-09-12
Fig. The detail evaluation - graphical analysis to copy the picure to clipboard and use it for Here you can click the button reports. Following picture is an example of a copied picture.
Figure: MP Evaluation, Clipboard

7.3.3.2 SAVE TO TEXT functionality 1) Save to text from the summary evaluation mode: This feature allows the user to save the complete database from one result file into a text file which can be easily imported into MS EXCEL. One can save either the COUNTER DATABASE which are all raw counter for the whole day or the STATISTIC DATABASE. The STATISTIC DATABASE contains the calculated statistic values from all intervals.
2) Save to text from the graphical mode In the graphical mode, you can save the complete table to a comma separated txt file which can easily be imported to other application like MS EXCEL. Choose the counters/statistics you need and press the button SAVE TO TXT CURRENT or select FILE
90/135
2005-09-12
7.3.4 Evaluations of SSNC/GPRS 7.3.4.1 MP Load Prerequisites: CR MPLDRP CR MPLDME DISP MP Create MP Load Measurement Reporting Create MP Load Measurement display of the types of MP
For calculation of LOAD A (%) we use the following procedure: The actual Load (total idle - audit) represents sort of LOAD B. This value we divide by 1.25 to get LOAD A (~700 mErl). These 700 mErl represent 100% LOAD A. Evaluation:
Figure: MP Evaluation With the very first Icon (open result file) you are able to load the wanted result files. In this example the MP-Load result file was loaded. Interval (below the title line MP Evaluation): If Busy quarter/hour was selected, then the indicated time is the quarter/hour with the highest load of all MPs, taken as average. LOAD A % column:
The percentage is referring to a base value of 700 mErl (~(1000 mErl 100 mErl Reserve) /1.25). It should give a first hint when coming to values, where it is necessary to think about extension (extra MP, new SGSN) Type SW/HW: As these values are taken from DISP commands, please verify on the exchange, if HW type is correct. For Example: SM/C means Signaling Manager/ MPU:C you can store the contents of MP evaluation as a text file, which can Via the icon be easily loadable into an EXCEL table. Or press button and switch to graphical mode:
Selection on Statistic (formula used) and/or counter contents
Figure: MP Evaluation, Graphical Statistic Checkbox An example is given below to display load-A value of the MP:SA over the day.
92/135
2005-09-12
Unit - load type / HW type
contents Figure: MP Evaluation, Graphical Statistic Example You can select Counters worksheet to display all MP counters:
Figure: MP Evaluation, Counters Detailed
93/135
2005-09-12
7.3.4.2 MTP/SCCP load Prerequisites For analysis of the C7-Link Load we need for each SSNC the files from CR MTPSCHD Create MTP Scheduling CR SIGLME Create Sign. Link Meas. CR SPMESCHD Create Sccp Meas.Scheduler CR SPME Create Sccp Measurements Additionally we need for every SSNC the MLOG files for DISPSIGDL; DISPSIGLINK:Display names=Yes; Evaluation: 1) MTP measurements per MP/Linkset/LIC MTP measurement shows link information HW-oriented (per MP), additionally the total load per MP and SSNC.
Figure: MTP Measurement Per MP Netan automatically shows the Datalinkname, DataLink ID, Linkset Name, Linkset ID and Link Code in one row. Without Netan you have to correlate the Linkset Name and the Linkset ID manually.
94/135
2005-09-12
By selecting different worksheet on bottom of the window, you can also see MTP measurements for linksets and for LICs and therefore you can have the information about: Whether the load sharing within one link set works. How many Links are connected to a LIC, and the Load of the LICs.
Important remark: SSNC description shows that the SSNC can handle a permanent load of 0.4 mErl. This is internal handling and has nothing to do with the ITU recommendation of 0.2 mErl maximum load. Thus a provider may apply the 0.4 mErl limit internally in his network if the counterpart is a SSNC as well; if the traffic is going out of the network he should restrict to the ITU recommendation (<0.2mErl. load per link). 2) MTP Measurement for High Speed Links: The High-Speed Signaling Links (HSL) feature increases the bandwidth between SS7 network nodes. High-speed signaling links can, for example, be used to set up a very powerful SS7 overlay network consisting of (fully or partially) standalone signaling transfer points (STP). The signaling end points (SEP) of the SS7 network are connected to the assigned mated pairs of standalone STPs via conventional signaling links. High-Speed Signaling Links can also be used to connect the service control points (SCP) of an intelligent network (IN) to the standalone signaling transfer points (STP).
You can also display all high speed links in the same window by selecting: High Speed Links (Summary) worksheet.
95/135
2005-09-12
3) SLT Statistics SLT Statistics is a general overview about links.
4) MTP counters View all MTP counters 5) SCCP counters SCCP counters with formulas for STP/SEP function and global title translation.
Fig. SCCP counters

6) Graph mode Same like the introduction in the MP evaluation, you can switch to graph mode and view SCCP related measurements on graphical basis.
7.3.4.3 SGSN evaluation 1) Registration:
N/A indicates that this value cannot be calculated as there are no counters. Registered subscribers: subscribers in SLR (In above picture: Total amount) Attached subscribers: subscribers in ready/standby state or from the mean number of attached subscribers (should be same) Active subscribers: Subscribers which have PDP context active
97/135
2005-09-12
2) SMS, CAMEL
3) PDP Context
98/135
2005-09-12
Note: In GR3 there are no counters available to count PDP activation rejections due to different cause values. This problem will solved in GR5. 4) Throughput (2G only)
Throughput shows LLC frames and SNDCP packets and octets. Throughput exists only for 2G! 3G throughputs must be counted in SP:GTPU. Please do not use the information Statistics per subscriber per second if only a cluster is selected: As subscriber (= registered subscriber) is SGSN wide
99/135
2005-09-12
based, the throughput is however cluster information, the values in this box are only valid if on top cluster = total is selected! 5) Overload
Overload information is MP oriented and not separated for 2G and 3G. 6) Routing area update 2G
Routing Area Update (RAU) here is subdivided into : - Cluster change SGSN internal (= from one to another cluster inside this SGSN) - Incoming Change, i.e. a subscriber comes into this SGSN area newly This measurement if for 2G only!
100/135
2005-09-12
7) Routing area
101/135
2005-09-12
8) Miscellaneous
102/135
2005-09-12
9) Traffic model
Traffic model shows the activities in the system broken down to one subscriber that will serve as a typical subscriber for planning purposes. 10) Graph Mode In SGSN evaluation, you can switch to the graph mode at any time. Based on the measurement group (worksheet) from which the graph mode is activated, the user will have different choices for KPIs and counters based on routing area, cluster, MP:PD and SGSN. The following figure gives an example of routing area based graph of number of subscribers in ready and standby state:
103/135
2005-09-12
7.3.5 Long Term Statistics The LTS program can be opened within NETAN2000 in the menu 'TOOLS' LTS 'LAUNCH LTS' or by double-click on LTS.EXE in the NETAN2000 directory. This tool gives you two possibility: - Monitoring and evaluation of performance parameters from one exchange over several evaluation dates - Comparing performance parameters from different exchanges and projects The program reads the measurement parameters for the busy hour or a pre-selected interval of each day into a long-term text oriented database and displays the parameters in a table or chart format. Following NETAN result files regarding packet switched services can be used with the LTS program: - MP evaluation results - SSNC evaluation results - GPRS evaluation results It is possible to add multiple result file in one step. When adding a result file you will be asked for a project name. This allows keeping the results of different projects in one LTS file.
104/135
2005-09-12
7.4 GPRS Signaling Analysis (Call analyzer, K1205)

7.4.1 Introduction Signaling trace and analysis are proved to be an indispensable method for the optimization of GPRS network. The reasons are: Due to the nature of available performance measurements: they provide quality indicators from the network element point of view rather than from individual subscribers point of view, very often we could see a big picture of the network quality but find it difficult to analyze single transactions in detail. Many cause values in unsuccessful GPRS procedures (e.g. GMM/SM procedures) are not counted by any available PM counters and the only way of quantifying them would be making statistics on related signaling messages. Very often in an optimization project we have to face a multi-vendor network. In this case we cant analyze other vendors performance measurements and have to rely on signaling analysis on standard interfaces.
With help of a protocol tester (e.g. K1205, K1297, K1103 or OCEAN), we are now in a position to see exactly which kind of messages and parameters are sent between different network entities and it is therefore possible to determine the reason of unsuccessful procedures. However, this is not the end of the story. It has to be pointed out that although the protocol-tester allows you to search and to count single messages, it does not provide any means to detect a defined sequence of messages (e.g. PDP activation with a given result). The only way to obtain the outcome of a certain dialog is to manually search known messages and to display all related messages by means of the ZOOM function. This procedure is very time-consuming and is not efficient if a large number of procedures is to be analyzed. The PC-based program Call Analyzer has been designed to overcome this problem, i.e. to automatically detect, categorize and to count all possible predefined messagesequences (i.e. call flows or message flows). In the following contents of this chapter, we will give some introduction for the most commonly used protocol tester: K1205/K1297 and some general idea about the signaling analysis software: Call Analyzer. 7.4.2 K1205 introduction K1205 can be used to trace interfaces in the circuit switched network as well as in the packet switched network. For GPRS, two kinds of interface cards are commonly equipped: E1/DS1 interface boards (PRIMO) to trace PCM based interfaces such as Abis, Gb, Gr, etc. and Ethernet interface boards to trace LAN based interfaces such as Gn, Gi, etc. The K1297 has additional simulation functions which are necessary for test, development purposes. However, a pure software update from K1205 to K1297 (based on the K1205 hardware platform) is allowed to enhance the K1205 software
in order to support advanced tracing functions such as Siemens Flexible Abis Allocation Strategy since BR7.0. Except these advanced functions, the basic operational procedures are same for K1205 and K1297. They will be introduced generally in the following contents. Procedures to monitor and record signaling messages: 1. Switch the device on. The operating system boots automatically. The K1205 application is started automatically. The Status Window shows status messages for the individual modules. If the modules boot without errors, the message K1205 boot OK will be displayed. 2. Load system configuration The following dialog will appear when the application starts successfully.
Click on the Default button if you want to start with a new configuration. Click on the Autoconfiguration button if you have incomplete information about the connected lines you want to monitor. If you have already configured before, and you want to use the configuration, you can open the saved configuration file, and click Load. The Data Flow Window appears after this dialog. 3. Data Flow Window
106/135
2005-09-12
To carry out measurements you need to configure data sources, filters and optionally, real-time applications. All these configurations can be made in the Data Flow Window. In general, you can complete the configuration within 5 steps. Step-1. Configure scenarios Various pipeline branches can be assigned to one source to carry out different measurements on the received data. A measurement scenario bundles all measurement tasks (symbolized by pipeline branches) that should be executed on a single data source.
107/135
2005-09-12
1. Click an entry to select, add, or edit a scenario. 2. Add a pipeline to the selected scenario. Click on the appropriate button in the pipeline toolbar; then drag and drop the pipeline branch into the green window.
Pipeline branches
Step-2. Configure hardware Select Cards Overview on the bottom left to switch to the Cards Overview pane for hardware configuration.
1. Click to switch to the hardware configuration widow
2. Select the card and click to open the Ports Setup dialog.
108/135
2005-09-12
Set Up Ports: In the Ports Setup dialog configure the ports for the type of data you will receive during the trace. For different interface modules, the dialog offers individual setup options.
1. Select the ports you want to configure 2. Specify the port settings
There are two different start mode: Standard Mode and Special Mode. The Standard Mode monitors data channels and timeslots of successive bits with bit rates up to 64 Kbps. Provides up to 32 bi-directional channels, depending on general system resources. The Special Mode monitors data channels and timeslots of separate bits or with bit rates higher than 64 Kbps. Provides up to 8 bi-directional channels, depending on general system resources. The Special Mode is used only when bit rates higher than 64 Kbps are needed or data channels consist of separate bits, such as for the fractional mode of GPRS interface. A practical example is: When Siemens Gb interface is implemented via direct connections (i.e. Gb Frame Relays are built on a direct physical connections between BSC and SGSN.), the frame relay bandwidth can exceed 64 Kbps by bundling timeslots on the Gb interface PCM links. In this case the Special mode is mandatory for tracing the Gb interface. You can also modify the setting from menu item: Tools | DS1/E1 Default Start Mode:
109/135
2005-09-12
In the Impedance field, please set proper impedance values for the K1205 Primo E1/DS-1 ports according to the actual impedance value of your measuring socket . There are normally three options: 75Ohm, 120Ohm and High Impedance. In case you are not sure about the actual impedance, please select High Impedance. Please be aware that if the impedance configured for K1205 Primo ports is lower than the actual impedance of measuring socket, lower layer failure may happen to make the link down, severe service outage will happen! Step-3. Configure data source Select on the Src. Online scenario box in the Measurement Scenarios panel to configure data sources. Here you need to set up the logical links from which signaling messages will be collected.
110/135
2005-09-12
2. Click to set up an online data source
1. Click to open the measurement scenario panel
Set Up Logical Links: In the Logical Link Setup dialog you configure the logical links of a data source. A logical link is a channel with an assigned protocol stack and a specified channel decoding method (level 2 method).
1. Select the link type: single link or pair. Create a pair if you want to monitor both directions (RX and TX)
2. Assign a logical link to the card and port from which data will be received. 3. Select protocol stack. This defines the interpretation rules for the processing of the received data
4. Define Timeslot, subslot, bandwidth, Level 2 method, etc.
111/135
2005-09-12
A stack is a set of protocols with definitions of links among them. The links give the conditions under which a proper upper layer protocol will be chosen to decode signaling messages contents. These conditions are in fact parameters of the protocol layer staying directly below. The structure of an Gb-interface stack is displayed below as an example.
The user of K1205/K1297 is allowed to modify or create stack files by himself. However this results experience and good knowledge about relative signaling protocols. In the installation directory of K1205/K1297, you will find two folder with name: protocols and stacks. Files in both folders are well organized according the technical specification bodies, countries or functions they belong to. Typical stacks used in GPRS network optimization: Gb interface stack: /K1297/stacks/gprs/gprs_Gb_31.stk Siemens Abis PCU downlink stack: /K1297/stacks/gprs/abis_sie_smg31_d.stk Siemens Abis PCU uplink stack: /K1297/stacks/gprs/abis_sie_smg31_u.stk Gr interface stack: /K1297/stacks/gprs/gprs_Gc_Gf_Gr_TSG10_R99.stk Gn interface stack: /K1297/stacks/gprs/gprs_GnEth_TSG10_PPP.stk Gi interface stack: /K1297/stacks/ip_gprs/iw_Gn_eth.stk
The figures below give practical examples of Gb interface and Abis interface data source setting.
Fig. Gb uplink data source setting. If Monitoring Pair is selected, you dont configure downlink Gb interface separately.
Fig. Abis uplink data source setting.

Because different stacks are used for Abis uplink and downlink. Single LL shall be selected and Abis uplink and downlink shall be configured separately. Step-4. Configure data sinks A data sink defines the output of the received data. It is symbolized by the last box of a pipeline branch. This example shows all possible output devices for an online measurement. The received data of this source can be: a. Recorded b. Viewed in the Monitor Window c. Evaluated statistically
setup the data sink as Recording File
or feed data into statistics
or display in the monitor
Step-5. Configure processing elements Each pipeline contains different modules necessary to process a measurement. These modules are called processing elements. By configuring the processing elements, you define your individual criteria for receiving and evaluating data.
114/135
2005-09-12
a. b. c. d. e. f. g. h.
Data source: data links from which signaling messages are collected. Trigger: defines the starting point for the measurement. Switch: starts, interrupts and terminates the measurement. Filter-1: Capture Filter. Selected messages will be stored in the Capture RAM. Application: call a program for real-time statistics. Capture RAM: work as a buffer for collected signaling data. Filter-2: View Filter. Selected messages will be displayed in the Monitor from the RAM buffer. Monitor: as a data sink, collected signaling messages are directed to the screen.
Practical examples of Filter setting Filter 1 (Capture Filter) 1) Gb link Settings on Network Service layer for not to capture nasty regular NS-ALIVE and NSALIVE-ACK PDUs:
115/135
2005-09-12
Settings on Frame Relay layer for not to capture nasty regular Status and Status Enquiry messages:
Note: If the Gb links show general performance problems, bypass these filter settings and verify first if the lower layers of these links perform as intended.
2) Abis link Abis PCU down, step-1:
Step-2:
117/135
2005-09-12
Step-3:
Press Edit button and set Uplink State Flag > 0 for PDDCB blocks. Abis PCU up, step-1:
118/135
2005-09-12
step-2:
Filter 2 (View Filter)
For a better overview it is recommended to display the Gb- and Abis-links only. If you need to see the PCU-frames in special situations, then just bypass filter 2.
4. Connect cables Connect the Protocol Testers PRIMO interface boards to the measuring sockets of the data line, using the appropriate cables. The interfaces on the PRIMO boards are 9-pin Sub-D sockets. 5. Record the signaling data Signaling messages collected from the data source can be recorded on K1205 hard disk. Following recording pipeline shall be created:
How to start recording: 1) click open in the Recording File box, give record file name and recording options:
2) Press OK to confirm your settings. The name of the recording file will appear in the data sink box. 3) To actually start recording, set the ON/OFF switch of the Recording pipeline branch to ON.
Now, the signaling messages collected from the activated data source, applying filtering options, will be written to the recording file until the disk is full, the file reaches predefined maximum file size or the ON/OFF switch is set to OFF manually.
120/135
2005-09-12
The record file can be opened later in the offline scenarios. For complete information about using K1205 for GPRS signaling traces, the following training document is recommended: MN1888: GPRS SBS System Specialist TAC 07_mn1888eu09mn_0001_practical_examples_traces.pdf
7.4.3 Call Analyzer introduction 7.4.3.1 Overview After a large trace file has been recorded by means of a protocol-tester, Call Analyzer can easily process it and quickly produce useful information and statistics based on your own definitions (flow definitions). From these accurate results you can conclude very important insights concerning the status of your network. Entire callflows can be examined and certain outcomes (flow-events) can be detected easily. Call Analyzer has already been used in different networks worldwide and it soon became an indispensable tool for network specialists. The tool does not only help to detect and solve very sophisticated problems but can also generate very detailed measurement results. It should be noted, that Call Analyzer provides an endless flexibility as the entire behavior and even the decoding of single messages can be defined by the user. However, once a certain scenario had been defined and stored within your specific setup-file, less experienced users can conduct the analysis without knowing the technical details behind.
121/135
2005-09-12
7.4.3.2 What do we get from Call Analyzer 1. Total counters Message counters DTAP AUTREQ...........................17404 AUTRES...........................17079 LUREQ.............................8673 LUACC.............................8608 SETUP...........................8374 CMSREQ..........................6971 CONNECT........................5079 CONACK..........................5063 PAGRES..........................1756 CMSREJ.........................60 LUREJ..............................21 . . BSSMAP ASREQ.................................8328 ASCMP.................................8263 HORQD................................7328 PAGIN...............................5311 HOCMD...............................1383 HOREQ..............................970 HORAC..............................960 HOFAI...............................754 HORRJ.................................678 CLREQ...............................642 . .
2. User defined flow results
It should be clear, that it is not enough to just count basic messages but rather to detect and count messages which have a defined relation to other messages. This is exactly the main strength of Call Analyzer. Flow-event counters can be automatically exported to EXCEL. Groups, colors, formatting, comments etc. are generated without additional macros within EXCEL. The table above also shows the average delay of the message. This information allows to analyze and compare setup times for different call types.
3. Flow graphic
4. Interval counters
Each of the counters (i.e. messages, message-events, flow-events) can not only be obtained as a total but also for a user-defined interval so that the tendency of counters can be shown over the time.
123/135
2005-09-12
5. Timing analysis
Timing distribution: From APCR to APAC
Faulty mobiles found
6. Export to database (CDR database)
For a given flow-event, any parameter of the related message can be exported to a standard database (MS ACCESS) so that very specific Call Records can be generated. Those CDR's can be displayed and analyzed with simple queries.
124/135
2005-09-12
Call Analyzer allows you to define your own CDR structure and contents for any parameter of any messages within your own pre-defined transaction. Optionally, certain fields of the database can be declared as key-fields which will trigger Call Analyzer to automatically generate summary tables. With this function, you can group the result instantly with e.g. IMSI, LAC, timeslot, etc. An example is given in the picture below:
7. Auto HTML report generation
For complete information about Call Analyzer, please visit: http://www.netcare-international.net/
125/135
2005-09-12
8 GPRS optimization case studies

8.1 GPRS radio optimization cases
8.1.1 Increase throughput by improving C/I
126/135
2005-09-12
8.1.2 Activating GPRS forced intra cell handover
8.1.3 Peak throughput management Part 1:
127/135
2005-09-12
Part 2:
8.1.4 Reduce delay on Gb interface
128/135
2005-09-12
8.1.5 Cell reselection parameter: CELLRESH
8.1.6 Reconfigure number of Abis subslots under FAAS
129/135
2005-09-12
8.2 GPRS core optimization cases

8.2.1 GMM/SM procedure analysis by call analyzer
130/135
2005-09-12
8.2.2 Investigation on data transfer pause
131/135
2005-09-12
8.2.3 WAP performance analysis
132/135
2005-09-12
133/135
2005-09-12
8.2.4 Mis-Configuration of VLAN
134/135
2005-09-12
8.2.5 Detect radio problems from core network side
135/135
2005-09-12

GPRS Optimization Handbook

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

GPRS Optimization Handbook

Загружено:

Авторское право:

Доступные форматы

GPRS Network Optimization

Optimization Handbook (Preliminary)

For internal use only

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

Changes: First release

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

INTRODUCTION OF COMMON GPRS NETWORK OPTIMIZATION TOOLS ................... 54

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

Opt. of new services

1.2 Network optimization process

Audit Audit Report

Solution Change Request Implementation Stage Report Verification

New problem Found?

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

1.3 Aim of this handbook

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

2.2 GPRS Network system architecture

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

2.3 GPRS protocols

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

G PRS Transm ission Planes

SND CP LLC RLC

Logical Link Control M edium Access Control Radio Link Control

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

BSSGP RLC BSS GP

S u b se t o f BSSAP fu n c tio n a litie s

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

2.4 GPRS procedures basic

Mobility Management State Transitions

Cancel Location, GPRS Detach IDLE

MM State Model of SGSN

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

Functional PDP State Model

INACTIVE no routing and mapping of PDP PDUs possible no data transmission

Deactivate PDP Context MM state change to IDLE

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

R o u tin g A r e a U p d a te R e q u e s t (o ld R A I, U p d a te T y p e ) S e c u rity F u n c tio n s

R o u tin g A r e a U p d a te A c c e p t (P -T M S I, P -T M S I R e a llo c a tio n ) R o u tin g A r e a U p d a te C o m p le te (P -T M S I)

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

2.5 GPRS radio network

LLC RLC M A C G SM RF G SM RF L 1 b is RLC M A C G SM RF

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

CS-1 CS-2 CS-3 CS-4

1/2 2/3 3/4 1

181 268 312 428

456 588 676 456

high middle low no

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

RLC BSSGP MAC

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

2.7 GPRS core network

Protocol & Interfaces

Fault & Maintenance Management

SIEMENS Shanghai Mobile Communications, Ltd. MN SE NP GPRS Network Optimization Handbook

SP:GTP SP:GTP MCP

MP:MM MP:MM MP:SM