Acoustic localization of mortar and small arms fires

Abstract

This paper compares three methods that estimate the location of an acoustic event based on measurements of its time-of-arrival (TOA) and direction-of-arrival (DOA) at a set of microphone arrays. We propose first a Least-Square (LS) estimator for source location for this combined DOA-TOA measurement model. We then look at the Maximum Likelihood (ML) estimator, comparing both estimators to the Cramer-Rao lower bound (CRB). Our third estimator is based on the Maximum A Posteriori (MAP) formulation and is designed to handle the association problem, where detections at different arrays must be matched if they correspond to a single event. Simulations show that the LS estimator performs slightly better than the ML estimator when the observation noise is not the expected one. Both methods exhibit a bias in the range estimate, which accounts for most of the square error. The MAP estimator, applied to live fire data, was accurate and successfully resolved multiple targets from outlier and multipath noise.