Redundancy and coverage detection in sensor networks

Reviewer: Maytham Hassan Safar

Several constraints contribute to the complexity of designing efficient algorithms on a distributed sensing network (battery power, communication bandwidth, computing power, and so on). Some work has explored the relationship between the coverage lifetime and connectivity of sensor networks; however, this work focused only on extending the coverage lifetime of wireless sensor networks by developing efficient algorithms for redundant sensor detection. Many works have focused on efficient techniques using the idea of turning off selected sensors to extend the network's lifetime. This reduces to the problem of detecting the redundant sensors, and selecting the maximum number of redundant sensors that can be safely turned off simultaneously. In this work, the authors present a low-cost algorithm, based on a Voronoi diagram with a complexity of O(n log n), for detecting redundant sensors and safely turning them off (where n is, on average, a constant number of Voronoi neighbors of a sensor). Their idea is based on the fact that a coverage boundary (including sensors on the outer periphery of the network and holes) can be used to optimize sensor placement. To determine if a sensor is redundant or not, the authors check to see if its removal will impact the coverage boundary of the network. The authors also provide efficient algorithms, with a complexity of O(log n), to maintain the solution in cases of sensor failures and new sensor deployments. This is done by recomputing the Voronoi cell locally where the deployment or failure occurred, using the information stored in every sensor about its neighbors in the network. In order to make the proposed technique scalable with the number of sensors, the authors developed algorithms that allow sensors to localize their view of the network, and build a complete map of the environment, without any initial knowledge of the network structure. They also explain how to construct and update the used Voronoi diagrams using only local information. The only reasonable assumptions used in this work are that each sensor knows its two-dimensional location, and that there is no simultaneous failure of related sensors. The paper is well written, technically sound, and highlights the advantages of using the proposed algorithm over other commonly used state-of-the-art algorithms. The work contains sufficient information to objectively compare the authors' approach to competing schemes. The authors have provided a compelling reason to believe that the new proposed technique offers some real advantages, in terms of scalability, reliability, computational complexity, and efficiency over existing schemes.