Convex Programming Based Robust Localization in NLOS Prone Cluttered Environments

Abstract

In a large variety of industrial scale processes, fixed or mobile sensors are typically deployed in large-scale vessels to monitor parameters such as temperature, pressure and chemical concentration. When these vessels are cluttered with obstacles, e.g. large cooling ponds cluttered with nuclear waste containers, it becomes increasingly difficult for the sensors to estimate their position. The acoustic ranging signals used for estimating distances between each sensor node and reference nodes fixed to the vessel infrastructure can suffer from Non-Line-Of-Sight (NLOS) signal propagation and thus introduce large positive errors in some of the estimated distances. In this paper we present a robust localization algorithm for localizing sensors in cluttered NLOS environments. We show that if the number of erroneous range measurements is less than half, it is possible to accurately estimate these NLOS errors at each sensor node by solving a convex optimization problem. Each sensor node can then use its estimate of NLOS errors to accurately localize itself. Our approach is completely independent of the physical hardware used to perform range measurements and thus can be used to localize sensor nodes in any NLOS prone environment. We demonstrate this with the help of experimental results with three different hardware platforms each employing a different ranging mechanism.