Optimal Admission Control in Mobile Cellular Networks with Movement Prediction

In this paper we study the impact of incorporating handover prediction information into the session admission control process in mobile cellular networks. The comparison is done between the performance of optimal policies obtained with and without the predictive information. A prediction agent classifies mobile users into two classes, those that will probably produce a handover to and/or from the cell under study and those that probably will not produce such handover. Moreover, the time instant has also been used in the prediction by means of a deterministic size window. Two different approaches to compute the optimal admission policy were studied: dynamic programming and reinforcement learning. Results show significant performance gains when the incoming predictive information is used in the admission process, and that higher gains are obtained when temporal information is used.