Optimal MAC State Switching for cdma2000 Networks

This paper deals with the performance modeling of the various MAC states as defined by the cdma2000 protocol. Our method uses a composite performance metric which has the capability of proportionally combining three basic parameters: channel utilization, waiting time and the saving in the signalling overhead. The scheduler at the base station is not only responsible for admitting new services into the system but also for switching the MAC states of a service depending on its activity. Since the true nature of the wireless data traffic is yet unknown, we use a mix of Poisson-distributed voice packets and Pareto-distributed data packets. We derive analytical expressions and also conduct simulation experiments to study the nature of the performance curve and thus compute the optimal values of expiration timers at which the MAC states should be switched such that the system performance is maximized. We show how our model can be made suitable for different systems by tuning the scaling functions (or weights) for each of the three performance parameters considered.