On Optimal Connectivity Restoration in Segmented Sensor Networks

This paper investigates the optimal connectivity restoration in segmented sensor networks, where mobile/relay nodes are optimally placed to form “bridges” among segments, such that both the average path length from nodes to the sink and the number of mobile nodes used are minimized. We formulate the optimal connectivity restoration as a multi-objective optimization problem and develop centralized and distributed algorithms for solving it. Given global network topology information, our centralized algorithm (i.e., Cut Restoration Genetic Algorithm or CR-GA) produces a Pareto Optimal set consisting of multiple non-dominated solutions. For scenarios where the global network topology is unknown (e.g., due to unexpected network segmentation) we develop a Distributed Connectivity Restoration algorithm (i.e., DCR). DCR restores network connectivity with lower overhead (when compared with CR-GA), at the cost of a suboptimal solution (i.e., the average path length and/or number of mobiles used). Through theoretical analysis, we prove that the worst case performance of DCR is bounded. We also show the effectiveness of our solutions through extensive simulations and a proof-of-concept system implementation and evaluation.