An Inteligence Aided Efficient Handover Strategy For PMIPv6

Through the customary explanation and formal analysis of MN's movement habit, considering the handoff latency, packet data continuity and packet costs, this paper proposes a method using machine learning algorithm to predict handover to improve the handover efficiency of PMIPv6. In order to reduce the burden of MAG and LMA, network-oriented mobility management is supported by importing intelligence handover agent(IHA). Using IEEE 802.21 MIH event services, command services and information services, training samples are collected independently of network media from MAGs. At the same time, both the two way of collecting training data, dynamic feedback and static impotation, are supported. The IHA is organically integrated into PMIPv6 network by well designing of function unit deployment and information exchange processes. When MN moves, it almost doesn't breake up the continuity of data, only some interaction between the new MAG and the original one happens to complete the handover. During handover intervals, IHA predicts where and when the MN would move. MAG and LMA prepare for the next handover by managing prepare bingding entry list. Data-driven model is used to wakeup the bi-tunnel to reduce the control signaling overhead. The success of handover forecasting can increase effective time for packet queuing cache, route optimization and resource reservation.Using the BP artificial neural network to predict handover, neural network input and output parameters and network structure must be well designed. A variety of BP algorithms are used train the neural network and better ones are enrolled. Through the simulation using MATLAB neural network toolbox,we find that a variety of training algorithm must exist. NS2 software is expanded to meet the simulation requirements PMIPv6 and IAH. Simulation results show that when the access network exists a large proportion of the regular movement, IAH can effectively reduce the handover latency, improve IP continuity, and to a certain extent, reduce the cost of control signaling.