Multi-target Bayesian Tracking -- 24-007

Our approach will use a Gaussian Mixture probability hypothesis density (PHD) recursive algorithm to estimate the time-varying number of targets and their states jointly.  The resultant output is a collection of contact followers (position, course, and speed estimates) derived from MAC and MAC-E snippet data that maintain integrity, even in scenarios where tracks are temporally and spatially close.  Our approach seeks to optimize track initiation to provide rapid detection while minimizing the false alarm rate and is not limited by the single target hypothesis.    We intend to partner with Applied Ocean Sciences (AOS) on this effort.  AOS will support the development implementation of an environmental surface into the multi-target tracking algorithm.  Seeding the tracker with signal excess (SE) information will enable the tracker to rapidly identify and discern clutter tracks derived from bottom interference, especially when paired with alternating waveforms or other Doppler discriminating capabilities.  We will leverage AOS’ Energy Flux model to keep computational complexity at a level suitable for use aboard fleet computing infrastructure.  We believe an environmentally informed tracker will produce more valid tracks with fewer returns; however, our approach will not entirely depend upon environmental data input and calculations.