Efficient data fusion for multi-sensor management

This paper is concerned with the development of a general framework for the management of multiple sensors in tracking a single target. To achieve this aim we draw on concepts from data fusion, particle filtering and heuristic optimization. Previous work gave the Multi-Sensor Fusion Management algorithm which provided a rigid scheme under which sensors were placed to maximize the probability of detecting the target. We present an adaptation to this scheme in which sensor placements are chosen to minimize a measure of uncertainty in the target position. We demonstrate the algorithm in an Anti-Submarine Warfare scenario in which we use passive sonobuoys to generate bearings and frequency (Doppler) data. We show that the quality of the track increases dramatically with the combined use of the two data sources and that the new sensor management algorithm further improves the track, and uses significantly fewer sensors in the process.