As mobile networks evolve the capacity of many network elements is becoming increasingly dynamic, creating a big challenge for network financial planners and engineering teams alike. Until recently, it has been relatively straightforward to decide how much equipment a network requires and to develop the associated capital plan based on a few simple calculations – This involves dividing the capacity of each network element by forecast demand to determine the number of elements required and then multiplying by the cost per element to determine ongoing expenditure requirements. When the capacity of each element becomes dynamic however the process breaks down and much deeper analysis is required.

Capacity in mobile networks is increasingly driven by transaction processing (July 30^th article) that in turn is heavily influenced by subscriber mobility.  Subscriber mobility drives capacity erosion across many network elements as a result of service proliferation across multiple overlay technologies that in turn gives rise to a significant increase in signaling and hence CPU processing requirements. In modern GSM/UMTS networks the principal elements with (traffic) capacity limitations that are driven by transaction processing are RNC’s, SGSN’s and switches, including MSC’s/ Call Servers and Media Gateways.  The capacity of these elements can be extremely dynamic because of the signaling and associated processing they must provide to network in order for it to operate, including the support of: 

§      Multiple, concurrent service connections with multiple COS/QOS requirements

§      Multiple overlay technologies with multiple signaling protocols and message stacks (SS7/C7, IPV4 and IPV6)

§      Multiple network domains with multiple hand-off types 

To overcome the challenge of planning a network with dynamic building blocks, both engineering and financial planning groups, need the capability to model the impact of transactions in real-time for each system element.  Historically this methodology has been used to determine the capacity of MSC’s to support voice traffic, by developing a call profile that specifies the amount of time the MSC CPU spends on each service that it provides to each subscriber. This profile is multiplied by the CPU occupancy time for each service to determine the operating capacity of the switch in terms of the number of subscribers it will support.

By applying the same principals to RNC and SGSN equipment dimensioning, the team at Cerion has developed similar modeling capabilities for these system elements as well as conventional MSC’s/Call Servers/Media Gateways To ensure the best possible accuracy, the profiles used for these system elements go beyond voice and address the dynamics listed above – Including the modeling of capacity for combined voice and data services delivered over shared 2G and 3G infrastructure.

Experience of applying this approach in live networks, shows that he capacity of MSC’s, Call Servers, Media Gateways, SGSN’s and RNC’s can vary by as much as 60% depending on the profile – Based on this, it is absolutely critical for Mobile Operators to embrace the new approach to managing the capacity of their networks. Only by doing this will they be able to ensure that they are operating the leanest possible network with best possible performance.