Track Clustering Based Data Association for Sky-wave Over-the-horizon Radar

The multipath propagation and low measurement precision in sky-wave Over-the-horizon Radar (OTHR) pose new challenges to data association. Aiming at the formation of ships, this paper presents a new method of data association. Firstly the plane measurement model is established, and the transformation between radar and ground coordinates is deduced according to the given ionospheres state, which leads to different propagation modes. Then measurements observed from radar are transformed to ground coordinates considering all possible propagation modes, and the transformed points in ground coordinates for each measurement are called hypothetic points. Max-min distance clustering method is adopted to divide these hypothetic points into multi-classes. The feasible classes are selected using the restriction of propagation modes, and the geometry center of clustering can be calculated by averaging the points of feasible class. Finally, the nearest neighbor method is used to associate the geometry centers of clustering among different scans and further find the correlation of measurements, propagation mode; and targets. Simulation results indicate that the proposed algorithm can associate the tracks in the formation of ships effectively, and the influence of radar detection probability on the algorithm is also analyzed.