Exploiting Constrained Rotation for Localization of Directional Sensor Networks

Constrained rotation may change the configuration of a directional sensor network for improving the sensing performance while incurring low resource overhead. One of the premises is that localization has been completed before the network embarks on the reconfiguration process. In this paper we consider a localization problem of 2-D directional sensor networks that contain a subset of anchor nodes with fixed positions. All sensors are assumed to recognize the global north exactly through the operation of equipped compass. We introduce a distributed strategy to calibrate the position of non-anchor sensors. In our approach, sensors with directional sensing capability measure relative range and bearing with respect to line-of-sight neighboring nodes. Moreover, sensors enable constrained rotation (pan) to compensate the confined field of view and improve relative bearing measurements. We discuss the localization accuracy as a function of deployment density and bearing granularity. Using noisy distance and bearing (angular) information allows sensors to compute position estimate that asymptotically approaches ground truth given sufficiently high deployment density. We evaluate the performance through simulations.