Joint Detection Threshold Optimization and Illumination Time Allocation Strategy for Cognitive Tracking in a Networked Radar System

Abstract

In this paper, a joint detection threshold optimization and illumination time allocation (JDTOITA) strategy was developed for multi-target tracking in an asynchronous networked radar system under cluttered background. The basis of this strategy is to facilitate detection and tracking using the prior target information in the tracking recursive cycle. The information reduction factor in the Bayesian detection framework is derived, optimized, and incorporated in the posterior Cramer-Rao lower bound (PCRLB), which is then utilized to serve as the optimization metric. Due to the asynchronous data and cluttered environment, the objective function needs to be recursively deduced and is nonlinear and nonconvex. We propose an efficient solver integrating the convex relaxation with the local search technique for this problem solving. Simulation results demonstrate the superiority of the JDTOITA strategy compared with the benchmarks with no optimization or optimization of either the illumination time allocation (ITA) or detection threshold alone. The results also imply that the target reflectivity and sampling interval of local radars are two important factors that influence the resource optimization.