Joint Target Detection and Tracking in Multipath Environment: A Variational Bayesian Approach

摘要

Different from traditional point target tracking systems assuming that atarget generates at most one single measurement per scan, there exists a classof multipath target tracking systems where each measurement may originate fromthe interested target via one of multiple propagation paths or from clutter,while the correspondence among targets, measurements, and propagation paths isunknown. The performance of multipath target tracking systems can be improvedif multiple measurements from the same target are effectively utilized, butsuffers from two major challenges. The first is multipath detection thatdetects appearing and disappearing targets automatically, while one target mayproduce $s$ tracks for $s$ propagation paths. The second is multipath trackingthat calculates the target-to-measurement-to-path assignment matrices toestimate target states, which is computationally intractable due to thecombinatorial explosion. Based on variational Bayesian framework, this paperintroduces a novel probabilistic joint detection and tracking algorithm(JDT-VB) that incorporates data association, path association, state estimationand automatic track management. The posterior probabilities of these latentvariables are derived in a closed-form iterative manner, which is effective fordealing with the coupling issue of multipath data association identificationrisk and state estimation error. Loopy belief propagation (LBP) is exploited toapproximate the multipath data association, which significantly reduces thecomputational cost. The proposed JDT-VB algorithm can simultaneously deal withthe track initiation, maintenance, and termination for multiple multipathtarget tracking with time-varying number of targets, and its performance isverified by a numerical simulation of over-the-horizon radar.