A New Tracking Algorithm for Maneuvering Targets

In target tracking for radar with Kalman filter, the process noise covariance matrix is usually selected experientially and assumed to remain unchanged throughout the tracking process. Although this method is effective on steady moving targets, in some practical situation, especially in the case of large maneuvering targets, we will meet some unsuccessful examples and fail to track those targets. In this paper, an improved target tracking algorithm based on the law of large numbers for maneuvering targets is proposed. During the process of Kalman filtering, the sliding window is used to select the acquired target trajectory data to estimate the process noise covariance matrix according to the law of large numbers. It means that the process noise covariance matrix can change adaptively with the movement of the target so that the filter can track the trajectory of the target more accurately. The simulation results show that the proposed tracking algorithm can produce smaller tracking errors than classical Kalman filter for targets with different motion models.