Impact of mobility prediction on the temporal stability of MANET clustering algorithms

Scalability issues for routing in mobile ad hoc networks (MANETs) have been typically addressed using hybrid routing schemes operating in a hierarchical network architecture. Several clustering schemes have been proposed to dynamically identify and maintain hierarchy in MANETs. To achieve significant performance gains, it is important that the underlying clustering scheme is able to identify stable clusters such that the cost associated with maintaining the clustered architecture is minimized. In this paper, we study the impact of mobility prediction schemes on the temporal stability of the clusters obtained using a mobility-aware clustering framework. We investigate the performance of the prediction schemes with respect to Gauss-Markov, Random Waypoint, and Reference Point Group mobility models under varying network and mobility conditions. Our results indicate that while mobility prediction significantly improves temporal stability of the clusters, an accurate mobility tracking algorithm need not always lead to an accurate mobility prediction scheme.