MOBILITY PREDICTION IN WIRELESS ADHOC NETWORK USING ADABOOST-MARKOV MODEL

Estimation of node mobility in an ad hoc wireless network will positively support to ensure the quality of service in the network. Embedding the mobility prediction model at the network layer of the node will lead to several challenges in the design of an efficient routing protocol but when integrated at the application level the information from the user's profile can be exploited to provide a better service to the user. The aim is to develop a mobility prediction model for accurate estimation of next location of a wireless node. As the next location of a node depends on the current location and few earlier locations in the trajectory the entire mobility pattern or the trajectory of a node can be modeled as a Markov process. To strengthen the prediction accuracy the adaboost algorithm was combined with markov model. A multi-order fusion markov model will be generated by combining weak models. The weight coefficients of the models are estimated using the adaboost algorithm. When compared to the general markov model the adaboost generated multi-order markov model yields better results.