Ground target tracking with RCS estimation utilizing probability hypothesis density filters

The knowledge on the radar cross section (RCS) of a ground target can support classification and identification tasks. In addition, it might also contribute to the resource management of the radar system because in general less energy needs to be emitted towards larger targets in order to obtain a detectable target return compared to small targets. The focus of this work, however, is to distinguish closely-spaced targets by first determining the mean RCS of the individual moving objects and then using this additional target attribute information to improve the track continuity in such a challenging environment. The RCS of a ground moving target can be estimated based on signal strength measurements. For this method to work, the RCS fluctuations are assumed to follow the analytically tractable Swerling-I and Swerling-III cases. The estimation scheme of the target RCS is incorporated into the Gaussian mixture variants of the probability hypothesis density (PHD) and cardinalized probability hypothesis density (CPHD) filters. The performance of these algorithms is analyzed based on a multi-target simulation scenario using a modified version of the optimal subpattern assignment (OSPA) metric that also accounts for labeling errors.