An Information Theoretic Radar Tracker Performance Model
Chris Kreucher, Matthew P. Masarik, John R. Valenzuela, Rebecca Malinas · 2019
This paper describes a new information theoretic approach for modeling the performance of a radar Tracker. Our approach is based on computing the Posterior Cramér-Rao Lower Bound (the so-called “Tracker bound”) as a function of radar capabilities, locations, and target state. The main contribution of this paper is an extension to more closely model fielded trackers which exhibit a series of track-drop and re-initialization events during the target lifetime. We account for this phenomena by marginalizing the Posterior (Bayesian) information matrix with respect to the tracker start time. We show that this new approach allows us to very closely predict the behavior of a radar tracker by comparing the new performance prediction to Monte Carlo runs of a tracker on a model 3-radar tracking problem with centralized fusion.