Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
R. Shalak, Dr. K. Sandrasegaran, Dr. J. Agbinya, S. Subenthiran University of Technology, Sydney (UTS) Faculty of Engineering I Broadway, Ultimo Sydney, N.S.W, Australia Email: rshalak, kumbes, agbinya, ssuben@eng.uts.edu.au
AbsIrnet
deployed worldwide. For the service provider, upgrading from thcir existing second generation network or deploying a new third generation network brings with it many considcrntionr which will prove vital to their ultimnte success. In thee formative years, perhaps the most contentious issue is UMTS network planning and optimisation.
I n this article, planning considerations are discussed for the core network. A model is generated and propasnls are made on how the operator can marimise rfiieienry as well as provide the user with thc required QoS.
this balance by exposing how core network planning strategies can effectively increase profits and efficiency and reduce wasteful expenditure Although, this paper essentially focuses on UMTS core network planning, it is important to recognise that the RAN has a direct bearing on core network design. With this in mind. it would he neglectful not to overview various RAN configurations and their impact on the core network. An integral part of the UMTS network development process is network deployment and evolution planning. By deployment, we are referring to a carrier who is rolling out a greenfield UMTS network, whilst evolution refers to a 2G operator who is upgrading to UMTS. Both cases primarily involve the evaluation and consequent purchase of new network hardware. followed by the integration of that equipment into the existing network. This paper recognises that on most occasions this selection of vendor equipment is rarely done on a merit basis but more oRen due to a service providers strategic alliance and ties with a particular supplier. These long built relationships can often be counter-productive. as they breed an air of automatic acceptance of a vendors product devoid of sound judgement. Furthermore, it is important to recognise that 3GPP standards were devised to produce global compatibility of multi-vendor equipment. These international standards enable service providers to utilise equipment produced by a multitude of vendors subsequently reaping benefits such as greater bargaining powers and competitively priced equipment. lnitially.it may seem wasteful to allocate time and resources to the process of determining which supplier can provide the most suitable equipment to satisfy a catriers current and future needs. However, considering that the decisions subsequent effects are often long lasting and vety costly to reverse adds funher weight to the process of selection. This paper addresses this issue by comparing and assessing the strengths and weaknesses of several current market UMTS core network components developed by a multitude of vendors. The research presented in this paper, is condensed into a core network planning model, This model utilises several key planning inputs to produce an optimised configuration output. The inputs include the breakdown of traffic forecasts into voice,
network Serving GPRS Support Nodc, Gateway CPRS Support Node. build-ahead.
1. Introduction
Rapid growth in traffic volume combined with a multitude of new services has begun to alter the structure of wireless networks. To cope with increased demands, service providers must upgrade to flexible networks with high data rate capabilities and QoS guarantees. In short they must adopt Third Generation technologies. The UMTS network can be segregated into two main subsections, the Core Network and the Radio Access Network (RAN). The Core Network, which is the primary concern of this article, comprises of Mobile Switching Centres, Serving CPRS Support Nodes (SGSN), Gateway CPRS Support Nodes (GGSN), Home Location Registers and Visitor Location Registers. This nucleus is responsible for the main network functions including switching of traffic, providing of QoS, mobility management, network security and billing. Consequently, its meticulous planning and careful dimensioning are imperative in ensuring overall efficiency and reliability. To date, almost all planning research, publications and literature concentrate on the Radio Access Network and overlook the Core Network. Our ongoing work, aims to shift
685 -
data and subscriber usage, excess dimensioning based on buildahead requirements and interface dimensioning based on several factors including RAN configuration. In addition, the paper evaluates and underlines various considerations taken in dimensioning an efficient UMTS core network. Section I gives a brief outline of UMTS network I components and architecture. The functionality o f each element is reviewed including a description of how these components function collectively. Section 111 outlines the differences in network shllcture between GSM and UMTS. The migration from 2G to 3G is outlined with special attention given to the new packet handling components. Section IV discusses core network planning considerations, highlighting the dimensioning limitations of Core Network elements. Our novel core network planning model is presented. This model aims to increase overall network efficiency by targeting and improving core network structure. It goes further to underline the impact of the radio access network on interface dimensioning and core network. Section V compares different vendor product offerings, providing insight into current capacity limitations for the various core network components. Finally, proposals are made on how an overator can maximise efiiciencv as well as provide a subscribe; with the required QoS.
Key: ......
--
Sign.lllng 1nterr.ce
ComNehvork Boondarin
Generally, there is a many to many relationship between SGSNs and GGSNs. A GGSN is the interface to external packet data networks for several SGSNs and a SGSN may mute its packets over different GGSNs to reach different packet data networks.
- 6 68
existing MSCNLR and HLR are retained but upgraded. Two additional nodes have been added. The SGSN and GGSN provide the network with its packet handling capabilities. The SGSN is responsible for the delivery of data packets to and from mobile nodes within its service area, whilst the GGSN provides network connectivity to several external Packet Data Networks (PDN) including the Internet. UMTS ability to process high data rates i s a direct result of its new W-CDMA radio access technology. However, in contrast to the core network. the existing GSM radio network will remain relatively unaltered. Instead, UMTS designers have opted to develop the UMTS Terrestrial Radio Access Network (UTRAN) lo attach and run in parallel with existing Base Station Controllers (BSCs) and Base Transceiver Stations (BTSs). The strategy developed aims to allow the UTRAN to handle the delivery o f packet switched data lo and from the mobile nodes, whilst existing GSM infrastNChrre maintains control ofthe circuit switched traffic. Terminal manufacturers are also following the seamless transition and have committed to producing dual - mode UMTSIGSM terminals enabling users to roam freely concurrently utilising both 2G and 3G services. These dualmode devices are expected to cost up to 30% less than those o f any other technology [2].
Traffic forecasts are separated into two main categories: data and voice. Voice usage forecasts are further disseminated into mobile to land, land to mobile, mobile l o mobile and mobile to voice mail usages. In evaluating data usage forecasts, planners must dimension the network according to the nature of data traffic and its intended destination. For example, the volume o f Internet traffic passing to and from mobile users has a direct bearing on the dimensioning o f SGSN and GGSN interfaces. Alternatively. the volume ofdata traffic terminating on an email server inside the network directly affects the interface dimensioning o f that server as well as SGSN. This process should be reitecated for a l l types of data traffic passing through the network.
In calculating the required number ofeach type ofcore network node, i t is important to initially understand the capacity limitations of each node. By contrasting node capacity limitations with expected traffic volumes, i t i s possible to clearly determine the basic network node rcquirements. I t i s imperative to recognise that these calculations represent the minimal network operating requirements. To meet any expected growth in demand, networks must be over-dimensioned. This i s referred to as build-ahead141.
Build-ahead accounts for traffic growth over a period of time. In its infancy, the operators network will be dimensioned according l o traffic forecasts alone, therefore i t i s wise to dimension for a longer build-aheadperiod o f around 12 months. This ensures that the network will cope with any excess demand as well as allow the operators traffic to grow without needing to constantly redimension their network. As the operators understandingof traffk trends increases, it will become possible to safely dimension for a much shorter build-ahead period increasing overall network efficiency.
To determine projected growth i n traffic, carriers rely on widely available government statistics involving population types, incomes, distribution o f wealth. taxation and spending habits. There i s also a need for statistics depicting the existing penetration o f mobile voice services and average Internet usage in the market [SI.
The radio access network has a direct bearing on core network dimensioning. The lu-CS and Iu-PS links are dimensioned to reflect expected RNC loading. Taking the case where a particular RNC is expected to process 40Mbps ofdata traffic, then the bandwidth o f the lu-PS should be 20% -30% more to allow for GTP and A T M overhead. lu-CS dimensioning can be similarly determinedby estimating the amount o f Erlangs processed by !he RNC.
A SGSN with a maximum throughput ofaround 250Mbps can support a maximum of five RNCs carrying 40Mbps. If more than five RNCs are required i a particular service area then n another SGSN will need to be deployed.
687-
All other network interfaces can be dimensioned according to expected user rates with a little extra capacity for overhead and bursty periods.
Only Ericsson, Nokia and Siemens specified the Quality of Service of their SGSN. Ericsson's SGSN are compliant with the GSM 3.60 class. It supports the reliability classes 2 and 3, Delay classes I to 4 for subscriber data. Nokia's QoS was based on priority, interactive and background classes while Siemens' QoS are configurable to meet the needs of different types of users. Most of the vendors collect charging information in relation IO the time and date, QoS, duration and volume ofthe packet data transferred. Nokia also collect information in regards to the radio resources used for re-transmitting.
The GGSN by Cisco. Ericsson, Nokia, None1 and Siemens has the following functions: Authentication and authorisation Ciphering Relay and routeing Address translation and mapping Encapsulation and tunnelling Mobility management
Only the GGSN by Ericsson, Lucent, Motorola, Nokia and Nortel have the charging information collection function. The number of subscribers supported ranges from 400,000 to I 5 000.Lucent's GGSN supports the most while Ericsson's combined platform of SGSNIGGSN supports the least. Ericsson's platform can be scaled to support more users. The number ofPDP contexts varies from the I Million supported by Nortel to the 180,000 supported by Cisco.
The supported data throughput varies from 2Gbps by Nortel's Univity platform to 100 Mbps by Siemens' avantage.
Most vendors allocate IP addresses statically via the HLR and dynamically using a RADIUS server. The Cisco GGSN supports QoS negotiation and handling and its QoS classes a e x mapped to Internet Differentiated Service Classes. Nortel's GGSN supports traffic management functions such as queuing, policing and shaping ofthe trafflc while Ericsson's model complies with the reliability and delay classes ofthe GSM 3.60 standard. The charging information collected includes time taken between the commencement and termination of PDP contexts, volume of data exchanged, and packet network destinations. It should be noted that the above-mentioned comparisons are void of pricing considerations. Equipment pricing and support play a vital role in vendor product selection. On most occasions,
- 688 -
vendors are hesitant to provide equipment pricing unless a nondisclosure contract is signed. Conclusion Increased services and subscriber numbers are pushing service providers lo acquire Third Generation wireless technologies. As the 2C network stmcture begins to change, current planning strategies become dated and new strategies must be adopted. The aim of these strategies is to maintain network eficiency throughout the transition period and beyond. This paper outlines core network changes experienced by service providers upgrading from CSM to UMTS. Modifications, mainly involve the addition of packet handling nodes and a requirement for increased interface dimensioning due to the extra capabilities of the W-CDMA radio network. In addition, the paper presents a core network planning model yielding long term planning strategies as well as performing a comparison of current market vendor products.
REFERENCES
[I] Van der Spiegrl. Accessing the development of UMTS and 3G K. UMTS Forum, Institute for InternationalRaearch Conference. Sydney. May 2M)I 121 Siemens URL. UMTS Overview, http://w.riemenr.com. [3] Clinl Smith P.E. Daniel Collins, 3 G Wireless NerworW, McGraw-Hill, 2 . m 141 J.P. Crrrro, The UMTS Nework ond Radio A r ~ e i rTechnologv. published by Wiley.NcrvYork.NY.2WI [SI UMTS World, Planning Baricr. URL hrrp://w.umtsworld.com.
(IY~O-.
689 -