A novel unscented filter TBD algorithm FOR weak radar target

In this paper, the unscented particle filter (UPF) is adopted to realize the tracking before detection (TBD) of weak radar target. In order to construct an accurate approximation to true proposal distribution, the state at each time scan is predicted according to the unscented Kalman filter (UKF). Unfortunately, in the framework of TBD, the UPF is used to handle with the raw measurements, which are unthresholded. In this situation, how to select the only measurement to compute the innovation is an important and difficult issue. Thus, it is difficult to perform the state update, and ineffective selection will lead to poor performance or even fail to detect targets. To address this problem, a modified UPF-TBD algorithm (MUPF-TBD) is proposed, in which the observation innovation is generated by a dummy observation. And the dummy observation can be generated from the measurement prediction of the state estimation at last step. Thus, using the feedback information, this new proposed method can obtain an accurate approximation to the system and as a result, improve the estimation performance. An application example is given to draw a comparison between this new algorithm and the existing algorithm. Simulation results illustrate the effectiveness of this approach.