Distributed Multitarget Tracking With Overlapping Sensor Fields of View Using Jump-Markov Probability Hypothesis Density

This article addresses the challenge of information fusion in multisensor multitarget tracking systems, focusing on the limitations of the arithmetic average (AA) fusion process. Through an in-depth analysis of information content changes during complete information fusion and AA fusion, we reveal how inaccuracies in correlation estimates or partial sensor coverage can lead to information loss and suboptimal estimates. We propose a fusion framework designed for the jump-Markov probability hypothesis density filter to overcome these issues, incorporating the measurement information cross-correlation (MIC) method. This approach enhances target position detection efficiency in an overlapping sensor field of view (FoV) by prefusing measurement data. In the process of MIC fusion, to improve efficiency, the method uses detected targets from a single local sensor as the index for matching, rather than simply aggregating all neighboring sensor data. This approach significantly enhances clustering speed. Simulations validate the effectiveness of our proposed framework in scenarios with multitarget nonlinear movements and overlapping sensor FoV.