A secure threshold of dynamic programming techniques for track-before-detect

Track before detect (TBD) algorithm with dynamic programming (DP) has been introduced to track target in low signal-to-noise ratio environments. But its performance is difficult to evaluate since we can't give the merit statistic an exact probability distribution. In this paper, we construct an simple-approximated upper bound of the merit statistic to give a secure threshold of the merit statistic produced by the DP-TBD algorithm according to some fixed false alarm rate. (3 pages)