Contact modeling and prediction-based routing in sparse mobile networks

Mobile ad-hoc networks (MANETs) provide highly robust and self-configuring network capacity required in many critical applications, such as battlefields, disaster relief, and wild life tracking. In this paper, we focus on efficient message forwarding in sparse MANETs, which suffers from frequent and long-duration partitions. Asynchronous contacts become the basic way of communication in such kind of network instead of data links in traditional ad-hoc networks. Current approaches are primarily based on estimation with pure probability calculation. Stochastic forwarding decisions from statistic results can lead to disastrous routing performance when wrong choices are made. This paper introduces a new routing protocol, based on contact modeling and contact prediction, to address the problem. Our contact model focuses on the periodic contact pattern of nodes with actual inter-contact time involved, in order to get an accurate realization of network cooperation and connectivity status. The corresponding contact prediction algorithm makes use of both statistic and time sequence information of contacts and allows choosing the relay that has the earliest contact to the destination, which results in low average latency. Simulation is used to compare the routing performance of our algorithm with three other categories of forwarding algorithm proposed already. The results demonstrate that our scheme is more efficient in both data delivery and energy consumption than previously proposed schemes.