JIE for a class of MD systems

Abstract

In multi-detection (MD) systems including over-the-horizon radar (OTHR) systems, the forward-based receivers systems and passive coherent location systems, resolvable multiple detections of one target due to the multi-path propagation will be received, in which both identification (including measurement association and propagation mode/transmitting origin identification) and estimation (including path-conditional state estimation and multi-path track fusion) are deeply coupled and affect each other. Up to present, all corresponding methods fall into the scope of sequential identification orientated or estimation orientated processing, and hence the solutions are not satisfactory. Here the joint identification and estimation scheme for the MD systems (MD-JIE) is developed based on the generalized Bayes risk considering both identification and estimation performance. The likelihood-ratio function and the conditional probability density function as the identification cost and estimation cost are recursively calculated, respectively. Taking the data-association constraints into consideration, the proposed MD-JIE scheme is implemented via online constrained optimization technology. Compared to the identification-thenestimation (ITE) method and the estimation-then-identification (ETI) method, the proposed MD-JIE method prevails in the joint performance measure with the higher correct identification rate than the ETI method and the smaller root MSE than the ITE method. Besides, the robustness of the proposed method is verified.

Tracking multiple extended targets with multi‐Bernoulli filter

“…For tracking the point target, several methods [1][2][3][4] have been developed to track targets with a fixed and time-invariant number. In order to deal with the tracking situation with a varying number of the targets, the random finite set (RFS) [5] is introduced to describe the state set of which the cardinality of the set may change with time.…”