A Markov Model-Based Location Prediction Scheme for Mobile Ad-Hoc Networks

摘要

Mobile ad-hoc networks (MANET) are multihop networks that are capable of establishing communication in the absence of any pre-existing infrastructure. Due to frequent node mobility and unreliable wireless links, the network is characterized by unpredictable topological changes. For more robust and reliable communications, it is important that a mobile node anticipates address changes and predicts its future routes in the network. This paper describes a novel prediction-based mobility management scheme for supporting mobile ad-hoc networks. We propose a Markov model-based mobility management scheme that provides an adaptive location prediction mechanism. We used simulation to evaluate the prediction accuracy as well as the probability of making the correct predictions. The results indicated that higher-order Markov models have slightly greater prediction accuracy than lower-order Markov models. Moreover the probability of making a correct prediction increases both -with an increase in number of repeated predictions made, and in the order of Markov model.