Predictive Timely Trigger and Performance Optimization for Deterministic Handovers

Abstract

Pre-registration as an important feature in FIMPv6, ensures providing the continuous mobility services for the end-users roaming across heterogeneous wireless networks, and is achieved by obtaining the timely and precise link layer (L2) trigger in the handover preparation. Firing the L2 trigger in the timely manner for deterministic handovers hence is crucial for guaranteeing the handover performance, but difficult to be achieved in the practical application scenarios. The late L2 trigger causes a missed detection deteriorating the quality of service perceived by the mobile node. To address this issue, taking the measured received signal strength (RSS) as the evaluated index, a predictive model based on the combination of Empirical Mode Decomposition and Support Vector Regression is proposed to predict the RSS values for timely firing the event of handover trigger, and then the performance optimization is further designed with the assistant of the predictive trigger. With an appropriate implementation of the predictive trigger model, the accuracy is compared with the representative time series predictive model, and the efficiency has been verified by the simulations conducted on the NS-2 platform.