Use of radar cross section in track-to-truth assignment for assessment of multiple target tracking algorithms

Abstract

A critical part of the assessment of multiple target tracking (MTT) algorithms is the assignment of tracks to truth objects. For real-time experiments and some computer simulations, kinematic assignment is the only option for track-to-truth assignment. However, kinematic assignment is often made to objects without regard to the ability of the sensor to detect and track the objects and this leads to misassociations of tracks to truth objects. In this paper, the use of radar cross section (RCS) to reduce the errors in the kinematic assignment of tracks to truth objects is investigated. Two approaches are considered. In the first approach, the tracking requirements for objects with smaller RCSs are exploited to develop a probability of tracking for each truth object from the RCS of the object. This probability of tracking is then included in the cost calculations for the track-to-truth assignment. In the second approach, the short-term RCS estimate for the track is treated as a Gaussian feature and included in the track-to-truth assignment likelihood as a Gaussian random variable. The extraction of the RCS for the truth object and a novel gating scheme are discussed in the paper. As a point of reference simulation results for both track-to-truth assignment algorithms are given in the paper along with results that do not make use of RCS.