A track-before-detect algorithm for the detection of a Markov target in the presence of missing observations

In this paper we address the problem of multi-frame detection (MFD) of a Markov target observed through noisy measurements. To limit the system complexity, we gate the detector with a pre-processing stage which discards unreliable observations in each frame. A novel dynamic programming algorithm is introduced, which applies a track-before-detect (TBD) logic to declare the presence of the target and jointly estimate its position. The closed-form complexity analysis and the numerical examples show that advantageous complexity-performance tradeoffs can be obtained with respect to the single-frame detector and with respect to other strategies already present in the literature.