A Joint Multitarget Estimator for the Joint Target Detection and Tracking Filter

This paper proposes a joint multitarget (JoM) estimator for the joint target detection and tracking (JoTT) filter. An efficient choice to the unknown JoM estimation constant (i.e., hypervolume around target state estimate) is proposed as a Pareto-optimal solution to a multi-objective nonlinear convex optimization problem. The multi-objective function is formulated as two convex objective functions in conflict. The first objective function is the information theoretic part of the problem and aims for entropy maximization, while the second one arises from the constraint in the definition of the JoM estimator and aims to improve the accuracy of the JoM estimates. The Pareto-optimal solution is obtained using the weighted sum method, where objective weights are determined as linear predictions from autoregressive models. In contrast to the marginal multitarget (MaM) estimator, the “target-present” decision from the JoM estimator depends on the spatial information as well as the cardinality information in the finite-set statistics (FISST) density. The simulation results demonstrate that the JoM estimator achieves better track management performance in terms of track confirmation latency and track maintenance than the MaM estimator for different values of detection probability. However, the proposed JoM estimator suffers from track termination latency more than the MaM estimator since the localization performance of the JoTT filter does deteriorate gradually after target termination.