Key Technologies of Multistatic Sonar Fusion Detection

To solve the fusion detection problem for multistatic sonar, this paper discusses three key technologies, i.e., measurement model, data association, and tracker performance model. The measurement model includes the error distributions of time, azimuth, position and sound velocity, as well as the localization expression of a target. The expression of each error factor and the expression of localization error are given. Data association is an important part of fusion detection to solve the assignment problem between measurements and known targets. The methods of nearest neighbor data association and multiple hypothesis data association are introduced in detail. The two data association methods use the same data model, measurement model and motion model. Simulations on the nearest neighbor data association and multiple hypothesis data association show that the target can be well tracked. The tracker performance model is used to evaluate the output quality of the tracker. The output quality is evaluated according to three performance parameters, i.e., tracking probability of detection, tracking fragmentation, and false alarm rate if tracking. According to the performance parameters from simulation, the output quality of the multiple hypothesis data association tracker is significantly better than that of the nearest neighbor data association tracker.