ML-PMHT Tracking of Multiple Wideband Sources by Fusing Data From a Distributed Acoustic Vector Sensor Array

Acoustic vector sensors provide a measurement of both the scalar acoustic pressure as well as the acoustic particle velocity vector in the orthogonal Cartesian directions, allowing for direction of arrival estimation of an acoustic source. Typically, acoustic vector sensors are used in array configurations in place of hydrophones to increase the overall array gain and directivity. However, for this work we explore the use of more sparsely distributed singular acoustic vector sensors for the purpose of target tracking, not relying on a traditional array configuration. Measurement fusion of individual direction of arrival estimates from each of the various sensors is performed using the Maximum-Likelihood Probabilistic Multi-hypothesis Tracker, which estimates the target motion parameters that maximize the likelihood based on a batch of measurements. The Cramér-Rao lower bounds for angular error variances are computed about the estimates and used as the associated measurement error variances in the likelihood function. We evaluate the efficacy of angular measurement extraction and accuracy of tracking estimates for a variety of scenarios, including both single and multiple target cases.