Multistatic sonar target tracking

Active sonar tracking using measurements from multistatic sensors has shown promise: there are benefits in terms of robustness, complementarity (covariance-ellipse intersection) and of course simply due to the increased probability of detection that naturally accrues from a well-designed data fusion system. It is not obvious what is the placement of the sources and receivers that gives the best fused measurement covariance for any target – or at least for any target that is of interest. In the first part of this research work, we investigate the problem as one of global optimization, in which the objective is to maximize the information provided to the tracker. The optimal placements that result are consistent with our intuition, suggesting that our placement strategy may provide a useful tool in more complex scenarios where intuition is challenged. The multitarget tracking (MTT) problem aims to estimate the states of an unknown number of targets by processing of measurements without knowing with certainty their origins. Many of the well-established MTT algorithms attack the problem in two phases, as by answering the question: Assuming there are N targets, where are they? followed by What is N? . In the second part of this dissertation, we propose a multitarget tracking algorithm that approaches the MTT problem by asking a different question: Is there a target at a given point? . Our approach is via a model of the surveillance region using small bins wherein a target can (or may not) lie. Then, by considering the limiting case in which the volume of these bins goes to zero, we reach the continuous form of this bin-occupancy filter, which has the capability of providing estimates for both the number of targets and their locations in a Bayesian fashion. The resulting prediction and update equations for the bin-occupancy filter are immediately recognizable as that of the probability hypothesis density (PHD) filter, which was originally proposed by Mahler in a sequence of papers. The bin occupancy filter can be extended to incorporate a target number stochastic process. We also show that the resulting filter is identical to the cardinalized PHD filter. The strength of our derivations is that they are based on a physical bin model: They do not use tools from finite point processes (FPP), do not require set-derivatives or other concepts as in, e.g., [40] and require no approximations. The Gaussian Mixture CPHD (GMCPHD) filter is a computationally efficient version of the CPHD filter that approximates the surface it represents with a set of Gaussians. In the last two chapters of this dissertation we report our results of the implementation of the Gaussian Mixture CPHD tracker, the GMCPHD filter augmented with a track management scheme, to two different real-world applications: The first if the problem of tracking groups of ground moving targets. In particular, we focus on the issue of integrating digital road-maps into the algorithm and the problem of modeling the clutter notch of ground moving target indicator (GMTI) radar sensors for low-Doppler targets. The second is the problem of tracking underwater targets using multistatic sonar networks. For the latter, we employed data sets from the multi-laboratory initiative called Multistatic Tracking Working Group (MSTWG) and from SEABAR\u2707 — an international sea trial led by NATO Undersea Research Centre.