A sparse undersea sensor network decision support system based on spatial and temporal random field
Bo Ling, Michael Zeifman, Mike Traweek, Thomas A. Wettergren · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2007
In a sparse sensor network, the sensor detection regions are often not overlapped. The traditional instantaneous detection scheme is less effective due to the fact that targets may not be detected by any sensors at certain sampling instances. To detect the moving targets in a sparse sensor network, we have developed a new system suitable for multiple targets detection and tracking. An optimization based random field estimation method has been developed to characterize spatially distributed sensor reports without making any assumptions of their underlying statistical distributions. FBMM (Forward & Backward Mapping Mitigation) technology is developed to reduce the false detections resulted from the random field estimation. To further reduce the false detections, the refined random field is clustered using gap statistics. STLD (Spatial & Temporal Layering Discrimination) method is developed to individual clusters and true sensor detections are determined based on both spatial and temporal patterns. Simulation results have shown that our system can effectively detect multiple target tracks in a large surveillance region.