On the Performance of Multi-Target Passive Sonar Tracking in Clutter

The problem of tracking multiple targets with passive bearings-only measurements is extremely challenging and rarely addressed explicitly in the literature. The difficulty arises mainly due to the low information content and non-linearity of the received measurements. Furthermore, effects such as high clutter density, closely spaced targets, and crossing targets are commonplace in passive tracking scenarios, adding to the difficulty of maintaining continuous and accurate target tracks. While many algorithms for multi-target tracking in clutter have been proposed, results on their application to passive tracking are scarce. The aim of this paper is to review a selection of state-of-the-art multi-target tracking techniques, apply them to the problem of bearings-only multi-target tracking in clutter, and evaluate and compare their performance under challenging scenarios. The types of algorithms addressed here include Sequential Monte Carlo (SMC) methods and Markov Chain Monte Carlo (MCMC) methods, as well as the more traditional Joint Integrated Probabilistic Data Association (JIPDA), and Multiple Hypothesis Tracking (MHT).