Mobility pattern recognition in mobile ad-hoc networks

A Mobile Ad hoc Network (MANET) is a collection of wireless mobile nodes forming a self-configuring network without using any existing infrastructure. Network nodes in a mobile Ad-hoc network move in some motion patterns called mobility models. The mobility models play a very important role in determining the protocol performance in MANET. Thus, it is essential to study and analyze various mobility models and their effect on MANET protocols. If we can recognize the mobility pattern of motion of mobile nodes in our environment we can customize our network protocols to deal with that existing mobility model. In this paper we introduce a new method for classification and pattern recognition of mobility traces into mobility models in mobile Ad-hoc networks. This method uses a simple learning based classification method to recognize the existing mobility model in raw mobility traces which was collected from real motion of mobile Ad-hoc nodes or mobility traces generated by mobility simulators. Our simulation results prove ability of our proposed method to accurately classify given unknown mobility traces into various mobility models.