All-Source Track and Identity Fusion

Abstract : This work addresses the association of moving-target indicator (MTI) tracks, EO and SAR imagery (IMINT) tracks, and signal intelligence (SIGINT) tracks, and the fusion of the corresponding report-level kinematic and identity information. Our fusion algorithm is based on hypothesis-management logic which recursively processes incoming frames of data from upstream trackers. The logic includes hypothesis generation, scoring, and pruning components. These components are based on track information, kinematic-state information, and vehicle identity information. Key track-level performance metrics for our fusion algorithm include the probability of correct track-to-track association and track fragmentation. We study the performance of the algorithm with simulated single-sensor tracks, for two scenarios of interest. The first scenario is based on a data collection for a set of 30 GPS-equipped targets, while the second is based on simulated ground truth for a set of 8 scattering targets.