Core Network Planning, Optimization and Forecasting in GSM/GPRS Networks

Transcription

1 Core Network Planning, Optimization and Forecasting in GSM/GPRS Networks C.N.Konstantinopoulou, K.A.Koutsopoulos, G.L.Lyberopoulos and M.E.Theologou National Technical University of Athens (NTUA) Dept. of Electrical Eng. and Computer Sci. 9, Heroon Polytechniou Str , Zographou, Athens, GREECE Keywords: GSM/GPRS Core Network Planning, Network Dimensioning, Optimization, Network Evolution, Evaluation Algorithm Abstract Despite the wide theoretical and technical knowledge about the capabilities of Global System for Mobile communications (GSM) elements, there is no specific methodology for designing from scratch either the Network Sub-System (NSS) nor the Base Station Subsystem (BSS) parts or optimizing and expanding the existing NSS/BSS architecture, by forecasting future requirements in terms of network elements/resources. Thus, for an evolving network operator who wishes to have a network running at lower cost, offering a competitive Quality of Service (QoS) to its subscribers, a Planning, Optimization & Forecasting Tool (POFTool) is of pivotal importance. This paper aims at identifying the main features of a proper POFTool that efficiently supports dimensioning and optimization studies for combined GSM / General Packet Radio System (GPRS) networks and to propose an evaluation methodology which enables the prioritization of alternative core network configurations according to operator-driven criteria. 1 Introduction The revenue increase is of pivotal importance for every mobile network operator. Operators strategies focus on capturing new subscribers, reducing churn, increasing customers loyalty, increasing subscriber satisfaction in service provisioning, reducing fraud, etc. Among the methods used are: the enrichment of service portfolio with new, innovative services attractive and relevant to the user, service differentiation offerings, handset subsidization, QoS improvements, etc. QoS improvement can be achieved by expanding the coverage area and by optimally utilizing the available resources on both the access and the core network parts. An optimized network: increases network availability and efficiency, increases subscriber satisfaction by improving the overall network quality, increases competitiveness, reduces operational costs and therefore, shall constitute one of the core network planning and dimensioning activities. Despite the wide theoretical and technical knowledge about the capabilities of GSM elements [1], there is no specific methodology for designing from scratch either the NSS nor the BSS parts or optimizing and expanding the existing NSS/BSS architecture, by forecasting future requirements in terms of network elements/resources. In most cases, the operator instinctively chooses one of the alternative configurations, mainly based on previous experience, availability and ease. It is obvious that this is not the proper way for operating an optimized network. This task requires intense and persistent effort, each time the network configuration needs to be changed. Things are getting more complicated for network designers due to the introduction of the GPRS related elements and the incorporation of IP-based backbone data network to the existing GSM infrastructure. GSM/GPRS architecture [1-5] is illustrated in Figure 1.

2 MSC/VLR A E Gs Gb Signalling Interface SMS-GMSC SMS-IWMSC Gd D TE MT BSS SGSN GGSN PDN TE Gn R Um Gn Gf Gp EIR SGSN GGSN Other PLMNs C Gr Signalling and Transfer Interface HLR SM-SC Gc Figure 1: GSM/GPRS Network Architecture The BSS part is composed of the Base Station Transmitter (BTS) and the Base Station Controller (BSC). The BSC includes the so called Packet Control Unit (PCU), which supports all relevant GPRS protocols for communication over the air-interface. The NSS part comprises: The functional entities of GSM: Mobile Switching Center (MSC) Visitor Location Register (VLR) Home Location Register (HLR) Short Message Service Centers (SMSCs) enhanced by additional interfaces for interworking with GPRS. The two new GPRS nodes: Serving GPRS Support Node (SGSN), which switches the packets to the correct BSS. Its task includes ciphering, authentication, session and mobility management and logical link management to the Mobile Station (MS). Gateway GPRS Support Node (GGSN), which is the gateway node between the GPRS and external packet data networks (IP) or Packet- Switched (PS) data networks (X.25). Additionally its task is to assign the correct SGSN for an MS depending on its location. Gi As far as medium- and long-term network expansion is concerned, the operator need to timely predict the required number of network elements (e.g., number of MSCs, HLRs, SMSCs, SGSNs, etc.) for a given subscriber and traffic distribution. As long as, the number of subscribers and the MSC/SMSC/ Voice Mail System (VMS) -types available in the market increase, the candidate scenarios (number of MSCs per MSCtype) increase dramatically. To select the optimum scenario (as well as the evolution steps) a methodology to identify, evaluate and prioritize them, based on certain operator-driven criteria, is mandatory. For an evolving network operator who wishes to have a network running at lower cost, offering a competitive QoS to its subscribers, a POFTool, fulfilling many diverse requirements (see section 2), is of pivotal importance. The purpose of this paper is to identify the main features of a proper POFTool that efficiently supports dimensioning and optimization studies for GSM/GPRS networks and to propose a methodology, which enables the identification of candidate network evolution scenarios and their prioritization according to operatordriven criteria. 2 Generic Operators Requirements A network planning tool shall enable the operator to import the existing network architecture (including GSM access and core network elements, GPRS network elements, ATM/IP switches), the relevant Points of Interconnection (POIs) (e.g., VMS platform, prepaid platform, SMSC, PSTN, PBXs, other PLMNs, internet, corporate intranets), the routing plan as well as traffic, subscriber and network element related information through a user-friendly Graphical User Interface (GUI). Another potential feature of the tool shall be the possibility of predicting future network dimensioning requirements, based on traffic/subscriber related data, and selecting (through an appropriate methodology) the proper evolution path based on certain evaluation criteria. This will enable the operator to: Identify all possible configurations of network elements (i.e., MSCs, number of PCUs per BSC, SGSNs, GGSNs, Charging Gateway (CG), Border Gateway (BG), ATM/IP infrastructure, etc.) taking into consideration the available network elements types. Select the most preferred scenario(s) (i.e., network elements combination) based on certain criteria driven by the operator such as, cost, simplicity, transferability, re-usability, etc. Proceed with and evaluate alternative network configurations, so as to identify the optimum scenario. A POFTool shall contribute to network optimization studies 1. Additionally a POFTool shall enable the 1 Network optimization shall aim at: (a) increasing the internal traffic per MSC/SGSN/GGSN and (b) increasing the terminating (to connected BSS) traffic coming from external POIs. In both cases, the inter-element traffic and therefore the demand in inter-element Pulse

3 operator to alter the network configuration and observe the impact onto the network performance. Example changes are given below: The subscribers' distribution per BSC, MSC, SGSN and GGSN (as a result of BSC rehomings). The distribution of originating and terminating traffic. The interconnections between network elements. The POIs each network element serves. The routing plan. The incorporation of new technologies, such as, Voice over IP (VoIP) and Voice over ATM (VoATM). 3 POFTool Description The POFTool functionality can be decomposed in two main sub-tools: (a) The Network Dimensioning Sub-tool and (b) the Network Evolution Sub-tool. The outcome of the Network Dimensioning Sub-tool is the evaluation of the network configuration under study and the computation of network parameters like the number of PCM lines, the throughput of SGSNs/GGSNs, the Erlangs/Busy Hour Call Attempts (BHCA) per MSC, etc. The outcome of the Network Evolution Sub-tool is the selection of the optimum network configuration for a given traffic and subscriber distribution and for a specific time period. 3.1 Input And Output Input The input data can be categorized into: Configuration-related data The location of GSM/GPRS network elements (e.g., X, Y coordinates). GSM/GPRS network elements' names, installation date, notes, etc. GSM/GPRS core network configuration and network elements' interconnections to external networks/systems (e.g. PSTN, VMS, Internet, other PLMN). Restrictions to e.g., max. number of PCM lines per route. Number of cells/trxs per BSC. Average number of Packet CHannels (PDCHs) per cell. Routing Plan: The tool shall enable the operator to import its own routing plan. Service-based and destination-based routing per network element (where applicable) shall be supported 2. Traffic-related data Total number of subscribers (contract and prepaid) per network element (MSC/BSC). Originating/terminating traffic (in Erl) per contract/prepaid subscriber per service (voice, data) and per destination. Portion of prepaid subscribers traffic. Number of incoming/ outgoing calls per service and per BSC/MSC during the busy hour. Average time for a mobile call, (PLMN towards PSTN, from PSTN towards PLMN, etc.). Requested bitrate per GPRS subscriber and per service during busy hour, or average bitrate (Mbits per day per subscriber and per service). Network elements' related data BSC related data Number of cells/trxs per BSC. Max. number of BSCs/MSC. Max. number of PCUs/BSC. MSC/VLR related data Max. VLR capacity. Max. load per MSC (in Erl). Max. BHCA per MSC. SGSN/GGSN related data Max. number of simultaneously attached users per SGSN. Max. number of Packet Protocol (PDP) context activations per GGSN. Max. throughput (Mb/sec, packets/sec) for SGSN and GGSN. Max. number of BSCs that can be connected to an SGSN Output A non-exhaustive list of the POFTool output data is given below: Required number of PCM lines per internal and external route (POI). Originating traffic load per network element and for specific destinations. Code Modulation (PCM) lines will be reduced decreasing thus the related cost. Keeping the number of PCM lines unchanged the traffic load (utilization) of the existing PCM lines will be reduced and the overall performance under high load conditions will be improved. 2 In certain cases, the operator shall route the prepaid originated traffic, before reaching the terminating MSC, to the PrePaid platform for charging purposes.

4 Outgoing, incoming and total traffic load per internal and external route (POI). Terminating traffic per network element. Indication of element(s) that have reached their capacity limits. Cost Indication. Indication of the max. traffic increase (with the same distribution) that can be supported by the current configuration. Number of Frame Relay (FR) timeslots required in the G b interface. Number of required PCUs per BSC. Number of SGSNs, GGSNs, CGs, BGs. 3.2 Network Dimensioning Sub-Tool A non-exhaustive list of the most desirable POFTool s capabilities regarding network dimensioning are described below: Definition, storage and retrieval of a core/switching network configuration via a userfriendly GUI. A library or a set of libraries containing candidate network elements from various vendors (e.g., BSCs, MSCs/VLRs, VMSs, SMSCs, SGSNs, GGSNs, ATM/IP switches, etc.). The elements characteristics (e.g., max. number of subscribers/msc, max. BHCA, max. attached users per SGSN, max. PDP context activations per GGSN) shall automatically be incorporated to the model. Possibility to: (a) modify the characteristics of existing elements, (b) define new network elements and store them at the respective library, (c) delete network elements from the library. Possibility to alter the network configuration by introducing additional elements with drag & drop capability directly from the respective library. Network configuration/modification with drag & drop capabilities (e.g., ability to connect a POI to another MSC, SGSN, GGSN). Possibility to the operator to select among a variety of alternative input data (statistics), depending on their availability. Incorporation of the operator's specific routing plan. The tool shall be able to evaluate the network configuration under study i.e., to identify routing problems, etc. The tool shall keep track of network changes. The tool shall enable the computation of: (a) the traffic coming from different call/traffic types (voice, data, SMS, fax, etc.)- per route, (b) the required number of PCM lines per route, (c) the number of signalling links/link-sets per route, (d) the Erlangs/BHCA per MSC, (e) the throughput per SGSN/GGSN (Mbps, packets/sec), (f) the number of PDP context activations per GGSN, (g) the number of attached users per SGSN, etc.. The results shall be provided in both textual and graphical form. The operator shall be able to impose restrictions where applicable e.g., the number of max. PCM lines to certain routes. Depiction of network elements performance (capacity, rate of requests handled, etc.); indication of those that have reached their bounds (along with the relevant parameter). Indication of the max. traffic increase (with the same distribution) that can be supported by the current configuration. A graphical representation of the network dimensioning sub-tool s capabilities is given in Figure 2. Define/Retrieve/Store Network Configuration Assign Destinations/Routes per Network Element Results Depiction/Storage, etc. GUI Netw Element Libraries Libraries Calculations Results Formulation Input Configuration related Traffic related Network Element -related Figure 2: POFTool: Network Dimensioning Sub-tool Capabilities. 3.3 Network Evolution Sub-Tool The POFTool shall be able to provide the means for the identification of future network configuration alternatives, the selection of the preferred scenario(s) and finally, the assessment of the optimum network configuration. More specifically, the sub-tool s functionality can be decomposed into (see Figure 3): 1. The candidate scenarios identification phase. The aim of this phase is to obtain a rough estimation of the number and the type 3 of the required network 3 Different scenarios result from the fact that different MSC types (i.e., transit, partial transit, pure MSC) with different characteristics in terms

5 elements so as to fulfil traffic and subscriber-related requirements. 2. The evaluation phase. During this phase, the alternative scenarios are being evaluated based on operator-driven criteria and a prioritized list is generated. 3. The optimum network configuration selection phase. The top-ranked scenarios of the prioritized list are being inserted to the POFTool, alternative configurations are being investigated and the optimum network configuration option is being selected, using the network dimensioning sub-tool. Note: To achieve the best possible result (e.g., reuse of existing infrastructure) for medium and large-scale networks, it should be given the possibility to divide the service area into large Geographical Areas (GAs) - based on population distribution and mobility characteristics- and then to apply the phases 1 & 2 to each GA. SCENARIOS IDENTIFICATION Apply Evaluation Methodology OPTIMUM SCENARIO SELECTION Traffic Estimation Identify # of Required Network Elements Prioritized List of Candidate Scenarios Dimensioning Sub-Tool Apply Dimensioning Rules SCENARIOS INPUT EVALUATION DATA Traffic Traffic Net. Net. Elem. Elem. Evaluation Evaluation Criteria Criteria Values Values Weighting Weighting Factors Factors Config. Config. Figure 3: Proposed Evaluation Methodology Candidate Scenarios Identification Phase The aim of this phase is the identification of the number and type of required GSM/GPRS elements, based on the corresponding dimensioning rules (see network-elements related data). The GSM dimensioning requires the estimation of the BSS traffic 4 and the transit traffic 5, while the GPRS of number of subscribers, max. BHCA, etc. can be utilized. The same applies for SGSNs in terms of the number of attached subscribers and the throughput. 4 This traffic stream represents the sum of the originating and the terminating traffic (in Erl) from/to the BTSs. dimensioning includes the identification of the number of PCUs/BSC, the number of FR timeslots in the Gb interface, the number of SGSNs and GGSNs required taking into account: (a) the number of GPRS subscribers having active PDP contexts, (b) the requested bitrate per service, (c) the number of traffic channels that can be used for GPRS, (d) the number of cells per BSC, etc Evaluation Phase The proposed methodology can be decomposed into the following subtasks: Identify an exhaustive list of evaluation criteria against which the candidate scenarios will be evaluated (section ). Define the algorithm according to which the evaluation will be performed and assign proper values to the weighting factors (section ). Apply the algorithm for each scenario. The outcome will be a prioritized list of scenarios (section ) The Evaluation Criteria An exhaustive list of criteria according to which the identified scenarios will be evaluated is given below: Cost (Q1): This factor will be highly determined by the number of new elements or upgrades of existing ones (MSCs, SGSNs, GGSNs, ATM/IP switches) required. The higher the re-use of the existing elements the lower the cost. It should be stressed that the re-use shall be considered in a networkwide approach (i.e., transfer of material to another GA). Reusability (Q2): This factor concerns the degree of reusing the existing infrastructure (network elements) in the GA under consideration so as to fulfil the network evolution requirements. Transferability (Q3): This factor concerns the possibility of transferring redundant network elements -not considered for future use- from the GA under study to another GA fulfilling the latter GA s network evolution requirements. Performance (Q4): This factor is mainly determined by the capacity/load offset. It can be decomposed into three sub-criteria: (Q41) Performance with regard to the number of subscribers. (Q42) Performance with regard to the BSS traffic handled. 5 This traffic stream represents the mobile-originating/terminating traffic destined/ coming to/from POIs (PSTN, other PLMNs, internet) connected to the G-MSCs and GGSNs.

6 (Q43) Performance with regard to the transit traffic handled. Simplicity (Q5): This criterion concerns the minimum number of required network elements to handle sufficiently the total traffic. The minimum the number of network elements is, the lower the number of interconnections/routes/links required and therefore the lower the network complexity is. New Generation Elements (Q6): This criterion will be of high value in supporting the operator's decision to use new generation network elements types in the future Algorithm Description The general guidelines of the algorithm are the following [6]: Identify tradable and not-tradable criteria. Tradable quantities are added together, so that it is the sum of the question responses that contributes to the overall figure of merit. Not-tradable quantities (question responses) are multiplied. Not-tradable criteria contribute independently to the quality of a proposed scenario. Multiplying independent criteria (or independent groupings of tradable criteria) emphasizes the importance of each one individually and minimizes the potential selection of scenarios that are extremely poor performers in any one independent category. To weight the criteria so that they reflect the desired emphasis, tradable question responses in the same added grouping are each multiplied by a weighting factor to reflect their relative level of importance. Multiplied question responses in the same grouping are each raised to an exponent to reflect their relative level of importance. Weights in the same added (or multiplied) group are then independently normalized. In our case, the tradable quantities are (see Figure 4): Cost versus (Simplicity, Reusability and Transferability). The application of the algorithm (according to the guidelines described above) results in the following formula: 1 13 w11 w12 w31 w32 w33 w Score = { Q4 Q6 {[ W21Q 1 + W22( Q2 Q3 Q5 )] 1} (1) 4 The quantity 1 is abstracted to shift the range from zero to four. Then the entire expression is divided by four to scale it between zero and one. The weighting factors are shown in Table 1. Note that the weighting factors values shall be among 1 and 5. Criterion Factor Performance W 11 Group 1 New Generation Network Elements W 12 Reusability-Transferability-Simplicity W 13 Group 2 Cost W 21 Reusability-Transferability-Simplicity W22 Group 3 Performance NewGenerat MSCs Reusability W 31 Transferability W 32 Simplicity W 33 Table 1: Weighting Factors Cost Reusability Transferabilty Simplicity Figure 4: Tradable and not-tradable Criteria Finally the top-ranked scenarios of the prioritized list are being inserted to the POFTool, alternative configurations are being investigated and the optimum network configuration option is being selected Application Example We assume that we have to evaluate five different network configuration scenarios. The values assigned for each factor (serving here as operator choices), are depicted in Table 2. The weighting factors that have been chosen are shown in Table 3. Applying the formula (1), scenarios E and C (see Table 2) appear to be the most promising ones and therefore the operator shall proceed with more detailed dimensioning-related studies for specific network configurations (inter-connections, assignment of POIs to MSCs, use of partial-transit MSCs, routing plan, etc.). Scenario Q 1 Q 2 Q 3 Q 4 Q 5 Q 6 Scoring A B C D E Table 2: Application of Evaluation Algorithm and Results

7 Criterion Factor Value Normalized value Performance W New Generation MSCs W Reusability-Transferability- Simplicity W Cost W Reusability-Transferability- Simplicity W Reusability W Transferability W Simplicity w Conclusions Table 3: Weighting Factors' Values In this paper the main features of a POFTool enabling mobile operators to optimize their network and forecast future requirements concerning network elements and resources have been presented. In addition a network evolution methodology based on which the operator can evaluate alternative network configuration scenarios and prioritize them according to specific criteria. Finally an application example of this methodology was given. References [1] M. Mouly, M.-B. Pautet, The GSM system for mobile communications, published by the authors, Palaiseau, France, [2] J. Cai, D.J. Goodman, General packet radio service in GSM, IEEE Commun. Mag., Vol. 35, No. 10, Oct [3] R. Kalden, I. Meirick, M. Meyer, Wireless Internet access based on GPRS, IEEE Personal Commun., Vol. 7 No. 2, April [4] ETSI GSM 3.60 version Release 1998 "Digital Cellular Telecommunications System (Phase 2); General Packet Radio Service (GPRS); Service Description; Stage 2". [5] C.Bettstetter, H. Vogel, J. Eberspacher, "GSM Phase 2+ General Packet Radio Service GPRS: Architecture, Protocols, and Air-Interface", IEEE Communications Survey, vol. 2 no. 3, 3rd Quarter 1999, [6] Anne DePiante Henriksen and Ann Jensen Traynor, "A Practical R&D Project Selection Scoring Tool", IEEE Trans. on Engineering Management, Vol. 46, No.2, May 1999.