DIFFUSION ANALYSIS OF TRACK LOSS IN CLUTTER

In tracking a target through clutter, the selection of incorrect measurements for track updating causes track divergence and eventual loss of track. By modeling the plot-to-track association algorithm as a Markov process and the tracking error as a diffusion process, one can study the mechanism of track loss analytically, without recourse to Monte Carlo simulations. The method is illustrated for nearest-neighbor association in two space dimensions. Explicit results are presented showing the dependence of various performance parameters, such as mean time to lose track and track half-life, on the clutter spatial density. The results indicate the existence of a critical density region, in which the tracking performance degrades rapidly with increasing clutter density. An optimal gain adaption procedure is proposed, which significantly improves the tracking performance in the critical region.