A Handover Prediction Model and its Application to Link Layer Triggers for Fast Handover

Abstract

Cross layer handover schemes are expected to provide seamless services to users on the move. To achieve this goal, a link layer must timely trigger handover protocols of upper layers so that they can finish necessary handover procedures before current wireless link terminates. Therefore, a mobile node needs an appropriate model to predict impending handover to generate link layer triggers in time. In this paper, we propose a mobility model that does not assume any radio propagation environments and movement patterns of a mobile node. Through statistical analysis of the received signal strength index (RSSI) datasets measured in various radio propagation environments, we validate that an adaptive autoregressive process can be used as a handover prediction model. With the proposed model, we also propose a link layer triggering scheme. Since the prediction process inevitably introduces errors, we devise a statistical compensation method to initiate link layer triggers in a timely manner considering the given handover signaling delay bound. Simulation results show that the error compensation method can significantly decrease the rate of late link layer triggers with the reasonable increase in the false alarm rate.