A Computationally Efficient Dynamic Programming Based Track-Before-Detect

Dynamic programming based track before detect (DP-TBD) is a batch processing method. It exploits the space-time correlation among several consecutive frames of measurements and jointly processes all the measurements in these frames. Due to this batch processing manner, DP-TBD suffers heavy computational load since it involves processing a large volume of data. Moreover, in order to track long continuous time target trajectories, sliding window (SW) processing is required which further increases the computational burden. The SW moves forward by one step at each measurement time to include the latest N frames of measurements and then jointly processes these frames and updates target states. This means that N frames have to be processed at every measurement time, and each frame needs to be repeatedly processed N times by N consecutive SWs, which is highly inefficient. To overcome this limitation, we present a computationally efficient DP-TBD algorithm in this paper. Similar to the recursive filter, the proposed method only processes the latest frame at each measurement time, and each frame only needs to be processed once. Our analysis suggests that the proposed method achieves significant reduction of computational cost, especially when N is large. Finally various numerical examples are used to demonstrate the robustness and efficiency of the proposed method.