THE USE OF TRACK-TO-TRACK CORRELATION AND NEGATIVE INFORMATION IN MULTI-SENSOR DATA FUSION

THE PROPOSED RESEARCH CONSISTS OF A FEASIBILITY STUDY AND PROTOTYPE ALGORITHM DEVELOPMENT FOR USE OF TWO PREVIOUSLY UNTESTED CONCEPTS IN NAVY SURFACE CONTACT DATA FUSION PROBLEMS. THE FIRST CONCEPT IS NEGATIVE INFORMATION, WHICH MEANS ACCOUNTING FOR ALL SENSOR'S DETECTION CAPABILITIES AND TRAJECTORIES IN ADDITION TO THE CONTACTS WHICH THEY GENERATE. THE SECOND CONCEPT IS TRACK-TO-TRACK CORRELATION, WHICH MEANS THAT A NEW CONTACT IS ASSUMED TO CORRELATE WITH A STREAM OF PAST CONTACTS CALLED A TRACK WITH PROBABILITY 1, AND THIS TRACK MAY CORRELATE WITH ONE OR MORE OTHER TRACKS WITH PROBABILITY <1. IN TYPICAL AFLOAT CORRELATION STRATEGIES, EACH CONTACT IS TREATED AS AN INDEPENDENT EVENT AND IS COMPARED TO ALL EXISTING HYPOTHETICAL TRACKS WITHOUT REGARD TO PREVIOUS CONTACTS. OUR APPROACH IS FIRST TO DEVELOP PRELIMINARY MATHEMATICAL MODELS OF THE SURFACE PICTURE FUSION PROBLEM INCLUDING OPINTEL, ELINT, ESM, RADAR, AND SONAR SOURCES. NEXT, WE WILL DEVELOP A SET OF SCENARIOS REPRESENTING VARIOUS COMBINATIONS OF SURFACE TARGETS, DATA SOURCES, AND MOTION. FINALLY, WE WILL IMPLEMENT PROTOTYPE ALGORITHMS FOR TESTING THEM AGAINST THE BENCHMARK SCENARIOS.