Determining traffic and control channels for a packet based cellular system having time-varying traffic

In this paper, we study the problem of determining the optimal number of traffic and control channels in a packet based cellular system. We only consider systems where a given (total) number of available channels are divided into traffic and control channels. We first show how to compute the ratio of required traffic and control channels based on the parameters of the system. However, it may be difficult to determine the parameters for a practical system, particularly, when the traffic is non-stationary. We observe that the maximum utilization of channels can be achieved when we split the total number of channels such that a data transfer request that passes the availability test of one of them does not get rejected at the other. Based on this observation, we propose a dynamic algorithm for splitting the available channels among data and control channels. The algorithm does not require any system parameters and does not assume any traffic characterization. Irrespective of the number of mobile stations, it gives best-effort utilization of the available resources by dynamically distributing channels among traffic and control channels. We study the performance of the algorithm using a simulation model. We consider non-stationary traffic patterns based on 'time of day' behavior.