An ML-MHT approach to tracking dim targets in large sensor networks

Abstract

-Poor individual sensor performance as well as a large number of sensor scans per time interval are two challenges for multi-target tracking is large sensor networks. We introduce a two-stage processing scheme (ML-MHT) to address the former issue, and another to address the latter issue (MHT 2 ). We consider as well the combination of these two techniques (ML-MHT 2 ). Simulation results are encouraging. Future work will include application of these techniques to more challenging multi-sensor datasets characterized by extremely poor detection and localization performance.

Contact clustering and classification using likelihood-based similarities

“…Coraluppi et al propose the use of two stages of maximum likelihood multihypothesis tracking (ML-MHT) [2]. The first stage is run on all contacts from all receivers on a single ping.…”