Sensor localization in concave environments

Abstract

In sensor network localization, multihop based approaches have been proposed to approximate the shortest paths to Euclidean distances between pairwise sensors. A good approximation can be achieved when sensors are densely deployed in a convex area, where the shortest paths are close to straight lines connecting pairwise sensors. However, in a concave network, the shortest paths may deviate far away from straight lines, which leads to erroneous distance estimation and inaccurate localization results. To solve this problem, we propose an improved multihop algorithm that can recognize and filter out the erroneous distance estimation, and therefore achieve accurate localization results even in a concave network.