A knowledge-based data fusion strategies for airborne multisensor surveillance system

Tracking of maneuvering multitarget in clutter is a very difficult task, and generally tracks of the targets are estimated and up dated by using the kinematic data measured by the airborne radar only. One of the most difficult problems of this task is “Plot-Track Association”, especially in cluttered environment, because there are too many false returns within the correlation gate. In this paper, a knowledge-based data fusion strategies is presented, not only the kinematic data, but also the attributes of the targets measured by the radar and other sensors (IFF,IR,ESM) are used. Based on an operator expert system, these attributes are used to filter out those false returns first, then the type of the targets producing the remained “true returns” are estimated. Combine with the kinematic data of the targets, these ID are then used to assign each return to the proper track with high confidence. An interacting model (IMM) tracking filter with data integration (radar and IR) is used for prediction and track smoothing. Simulation result shows that this approach is very effective in dense cluttered environment