Joint Multi-Target Tracking and Identification for Distributed Radars using Bayesian Binary Test

This paper introduces a joint multi-target tracking and identification (J-MTT-I) approach in distributed radar systems. In the tracking phase, we utilized the extended Kalman filter, cascading to covariance intersection fusion, for distributed estimation of target states. Addressing the target identification/recognition issue involves considering and formulating true-false target classification as a binary test problem. Subsequently, the naive Bayesian classifier was employed to model the posterior probability density function concerning target types and to extract target type information from tracking results. The proposed approach offers a practical method for training target classifiers and is anticipated to be applicable to real-world radar systems. Numerical simulations confirm that utilizing multi-dimensional state estimation information derived from the target tracker can further enhance the accuracy of target identification.