A Predictive Handoff Approach for Mobile IP

We propose a predictive handoff mechanism which can predict the occurrence of link layer handoff precisely. It can be used as a trigger for mobile IP to start network layer handoff ahead or pre-allocate resources. Using this approach, link layer of MN (mobile node) detects signal strength changes of base stations to calculate the probability of coming handoff and triggers network layer handoff based on mobility model and RFM (recursive forgetting memory linear square). In this paper we illustrate the mechanism of this approach in detail and provide experiments to evaluate the performance. The results show that this approach can provide an exact trigger with low rates of false alarm and missed detection. It can significantly shorten handoff latency and reduce packet loss in mobile IP.