Mobility prediction in mobile ad hoc networks using neural learning machines

Recent advances in wireless and mobile computing have paved the way for an unprecedented demand growth for mobile services and applications. These services and applications communicate and exchange information using wireless local area networks (WLANs) and mobile ad hoc networks (MANETs). However, new design challenges emerge due to the error-proneness, self-organization and mobility nature of these networks. This paper proposes a neural learning-based solution to the problems associated with the mobility of MANET nodes where future changes in the network topology are efficiently predicted. Using synthetic and real-world mobility traces, the proposed predictor does not only outperform existing mobility prediction algorithms but achieves accuracy scores higher by an order of magnitude. The attained accuracy enables the proposed mobility predictor to improve the overall quality of service in MANETs.