A novel genetic algorithm approach to mobility prediction in wireless networks

Abstract

Wireless networks are required to support nomadic computing, providing seamless mobility without call drops. The number of mobile users must be known apriori for allocating the channel bandwidth, which is a function of the number of mobile users already admitted into the system and direction of their movement. If the user's next movement can be predicted, it may be used to allocate a channel in the neighboring cell, thereby reducing the call dropping rate and leading to seamless mobility. Most of the mobility prediction algorithms offer reduced performance as a result of being movement history based and also do not give accurate results when users move towards the corner of the hexagonal cell structure but for the sectorized ones. Substantial improvement is demonstrated with the use of the proposed novel genetic algorithm based approach presented in the paper. Simulation results show that the proposed algorithm gives accurate results independent of the user's movement pattern. © 2009 Springer Berlin Heidelberg.