Call Admission Control and Dynamic Pricing in a GSM/GPRS Cellular Network

Abstract

In the past decade, the wireless communications market has experienced tremendous growth, and this growth is likely to continue in the near future. In addition to an increase in the number of users, ever more demanding applications will appear, resulting in ever greater resource requirements. The limited radio frequency spectrum available can no longer support this increasing demand, and the required quality of service will no longer be attainable if an efficient solution is not found. The easiest approach to solve this problem is to increase the capacity of the network, but this is uneconomic and not really practical. Indeed, at their current size, the networks are already under utilized most of the time, even if they can not accept every incoming call during congested peak periods. Increasing their capacity still further may solve the congestion problem for a while, but at the cost of an even higher global under utilization of resources. Other solutions have emerged to alleviate this problem, but none of them is really effective when the degree of congestion becomes too high. An alternative solution is to keep the current network capacity and to make the users? demand fit this limited capacity. This is the basic principle which leads to dynamic pricing: the price of making a call depends on the network load, it can be very high when congestion occurs or very low to encourage users to make calls during off-peak periods. As a result, the congestion is decreased while the overall utilization of the communications channels is improved. Dynamic pricing has already been applied successfully in several domains, but has only recently been considered for use in cellular networks. This project aims to look at how the different solutions mentioned above perform to solve the problem of congestion in the case of both GSM and GSM/GPRS networks, and in particular, at whether or not a combination of dynamic pricing and more traditional approaches can give better results. For this purpose, a detailed traffic model for both GSM and GPRS networks is given and implemented in an event-driven simulator. The effect of several dynamic pricing and admission control combinations is then analysed, and the importance of some incoming traffic parameters is highlighted.