Anomaly Detection in Ground Moving Target Indicating (GMTI) Radar

UAV-based GMTI radar offers the capability to detect anomalous vehicle and personnel activity in real-time to cue narrow field of view EO/IR sensors for identifying enemy targets in the presence of normal vehicle and foot traffic. Data mining is the process of extracting useful information from large complex data sets, and is ideal for application to GMTI. TSC proposes to significantly enhance our existing GMTI Data Mining techniques, which have been successfully applied to detect convoys in dense background traffic in a classified Joint STARS GMTI database from Operation Iraqi Freedom. In Phase 1, TSC will develop innovative temporal, spatial and hybrid data mining techniques to analyze traffic patterns and detect suspicious vehicle behavior in challenging UAV surveillance scenarios of interest to the Army. We will employ extensive in-house target and background models and GMTI simulation tools to evaluate the feasibility of these algorithms on simulated data that is representative of near-term UAV systems. In Phase 2, TSC will refine the most promising techniques and develop a software architecture that will be utilized to fully test the algorithms on both simulated and collected GMTI data. TSC will also investigate methods to detect anomalous activities of personnel using GMTI measurements.