Graphical Analysis for MTI Exploitation (GAME)

The ability of airborne GMTI radars to observe motion persistently over large regions provides an unsurpassed potential to detect linkages between events or facilities for forensic analysis. However, realizing this potential is currently a time-consuming, manually intensive task requiring days or weeks to produce actionable intelligence. To reduce the analysis time and the analyst workload we propose to develop a family of graphical analysis tools for automated GMTI link discovery. Our approach extracts key track-to-track relationships from MTI (or other INT) data to produce a large directed graph, enabling the application of computationally efficient algorithms which have been developed to perform link analysis on very large graphs. Because our approach transforms the data from tracks to a discrete nodes and links, it simplifies difficult problems of multi-INT correlation and fusion, avoids the need for long track lifetimes, and still retains the information needed for forensic analysis. We also apply recent extensions of scan statistic methods to graphical data to detect interesting relationships from noisy graph data. These methods allow users to incorporate information about data uncertainty and to get back quantitative confidence measures for the discovered linkages, helping users to determine if there is adequate information for making decisions.