Statistical estimation of link availability and its impact on routing in wireless ad hoc networks

In this paper, an analytical framework is developed and validated via simulation for statistical estimation of the evolution of the separation between a pair of mobile nodes in an ad hoc network. Simulation results demonstrate that path selection based on minimization of the product of the link statistic significantly outperforms minimumhop and fixed threshold-based 'path-stability' schemes. A hierarchical mobility model integrating the dynamic effects of velocity, group movement and geographic scope is used to generalize the results. Another significant result is the performance enhancements hold in large networks irregardless of the assumptions used for statistical estimation. The effect of merging many independent groups appears to restore independent mobility. Finally, results show that at the highest mobility levels, 90% of the longest surviving paths fail within 60s. None of the strategies approach this optimal value. This important result suggests that optimal predictive mechanisms alone are insufficient to ensure scalable routing in ad hoc networks.