Optimal Admission Control in Multimedia Mobile Networks with Movement Prediction: A Sensitivity Analysis

We study the problem of optimizing admission control policies in mobile multimedia cellular networks when predictive information regarding movement is available and we evaluate the gains that can be achieved by making such predictive information available to the admission controller. We consider a general class of prediction agents which forecast the number of future incoming handovers. For the optimization process we propose a novel Reinforcement Learning approach based on the concept of after states. The main objectives of the paper are twofold. First to quantify the impact that different ways of aggregating the movement prediction information referred to different service classes have on system performance. Second, to evaluate the sensitivity of the optimal anticipation time with which the admission controller needs to know the predictive information with different system parameters. Numerical results show that with the proposed aggregation of the predictive information we obtain policies as good as those obtained with the conventional representation of the system state, but at a lower computational cost. Besides, although the performance gain varies with system parameters, the location of its maximum remains almost invariant with the anticipation time in all the studied scenarios.