Near-Optimal Channel Reservation for Cellular Phone System

In this paper, optimal management of requests for channel usage in a cellular system is formulated as an access control problem for a multi-server-loss-queue network. Using an inductive approach of dynamic programming, the optimal channel reservation policy is induced to be functions of the number of ongoing cellular calls in each cell. However, the scale of cellular networks is generally very large. So, the computational burden of finding an optimal policy may be too heavy due to its involvement with the states for all cells in the network. Hence, the idea of bottleneck modelling and aggregation method in stochastic process is used to downsize the system and find a near-optimal channel reservation policy. Focusing on the target cell, we first reduce the scale of loss-queue network to a seven-cell model with the compensation of system parameters. Furthermore, an aggregation method is proposed to further reduce the seven-cell model to a two-queue system. Numerical results show that the near-optimal control policy based on the reduced two-queue model demonstrates close performance to the optimal reservation policy.