Passive Sensor Target Tracking And Data Association Under Complicated Interference Environment

This paper studied the target tracking and data association algorithm of multi-station bearing-only passive localization system under a complex interference environment. The complex interference environment means the incomplete measurement information and noise related problems due to dense targets, high clutter rate, leak detection of sensor and so on. The multi-station bearing-only passive localization system now is facing the nonlinear measurement information, data correlation and some other problems, and the key point of this paper is to solve the measurement loss, noise related, high clutter rate problems and to associate the sensor data with targets effectively in the complex interference environment.Under a complex interference environment, the sensors miss some measurements, which results in filter divergence problem. Stochastic process theory is used to describe this phenomenon and a conversion measurement Kalman filter algorithm is introduced with incomplete observations. This algorithm solves the nonlinear measurement problem with the measurements lost, which can be applied to a more realistic environment.The traditional convert measurement Kalman algorithm directly discard the correlative noise information of the pseudo measurement, which will reduce the tracking accuracy of the algorithm. This paper proposed an observation noise decorrelation algorithm based on eigenvalue decomposition. After dealing with the correlated noise of the pseudo measurement, the algorithm get a new noise-unrelated pseudo measurement which improves the accuracy of the algorithm while ensuring real-time performance.The multidimensional assignment algorithm is an effective data association algorithm based on passive location. And the cost function describes the different measurement division relative to the probability distribution of the target location which approximately obtained by least squares algorithm. The traditional cost function should take observation noise into account. In this paper an improved cost function is proposed. By considering the error in the target position estimation with least squares algorithm, we can get better position estimation and thus improve the association accuracy. To improve the real-time performance passive sensor data association, a(S + 1)-D dynamic assignment algorithm is proposed. The algorithm uses the trajectory information to form observation threshold. Whether the current measurement falls into the threshold is checked, and we can effectively eliminate the redundancy partitioning, which greatly reduce the calculation burden and improve the real-time performance of the algorithm.