Belief-Propagation-Based Resolvable Group Target Tracking With Unknown Detection Probability

The problem of <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">resolvable group target tracking</i> (RGTT) with unknown detection probability is considered in this article. To adaptively match the detection probability used in the filtering process with the practical tracking environment, we propose a belief-propagation-based robust RGTT algorithm, named BP-R-RGTT. Specifically, the target state, existence variable, and detection variable are first used to construct the joint state vector of all potential targets. Then, given the measurements, the joint posterior <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">probability density function</i> is derived. Next, the recursion process of the proposed BP-R-RGTT is described in detail, where the marginal posterior distributions of variables are effectively obtained via the BP algorithm. Lastly, the <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">Beta-Gaussian mixture</i>-based implementations of the proposed BP-R-RGTT is proposed by using the GM and Beta distribution to represent the target state and detection probability, respectively. Simulation experiments are provided in single-sensor and multisensor RGTT scenarios to verify the superiority and robustness of the proposed BP-R-RGTT.