Using Kolmogorov-Smirnov Tests to Detect Track-Loss in the Absence of Truth Data

Of significant interest in the practical application of data association algorithms to target tracking in cluttered environments is how to determine track-loss in the absence of truth data. An approach is laid out for Kalman filter based data association algorithms where sample gated measurement count distributions are compared to theoretical measurement count distributions of the "tracking" or "track-lost" regimes to determine the regime of filter operation for single-target tracking applications. The comparisons are done via a pair of Kolmogorov-Smirnov tests. Among the advantages of this method are that confidence intervals are associated with the track regime tests, and that the number of samples required to discriminate between regimes can be determined adaptively. Simulation results for the method are provided.