Exploiting a constellation of narrowband RF sensors to detect and track moving targets

This paper presents a novel approach to detecting and tracking moving targets using a constellation of narrowband radio frequency (RF) sensors. Our methodology is an innovative combination of nonlinear estimation and information theoretic sensor placement. The nonlinear filtering approach fully exploits bistatic Doppler measurements made by the sensors without thresholding or linear/Gaussian assumptions, thereby improving the detection/false alarm tradeoff and lowering tracking error. The information theoretic sensor placement algorithm, which is based on minimizing the Cramer-Rao bound (CRB) on localization variance, selects sensor positions that lead to the best estimation performance, thereby maximally exploiting the finite sensing resources. We illustrate the efficacy of the algorithm by showing how well we can track a target using sensors placed using the CRB optimal method as compared to sensors placed randomly.