SPECSweb Multistatic Tracking on a Truth-Blind Simulated Scenario of the MSTWG

摘要

Effective fusion and tracking of multistatic active sonar contacts is challenging, due to high levels of false alarm clutter present on all sonar nodes. Such false alarms often overload the sensor-to-fusion-center communications links and fusion/tracking processes producing too many false tracks. The Specular-Cued Surveillance Web (SPECSweb) multistatic tracker mitigates these problems through implementation of two amplitude thresholds: a high threshold used to identify the occurrence of high-strength specular detection cues for track initiation and a low threshold for selective extraction of additional detections within retrieval snippets for track extension forward and backward in time. This approach can significantly reduce the data rate at the input to the fusion/tracking algorithm, and reduce node-to- fusioncenter communication link throughput requirements. This paper provides performance results of this tracking algorithm on simulated multistatic data sets from the Multistatic Tracking Working Group (MSTWG). These included a ('truth-blind') data set for which the target truth information was not provided in advance of the time of algorithm application. The results show effective tracking performance using this approach, yielding high quality target tracks with no or few false tracks. The method is shown to have excellent potential in reducing the overloading of the communication links, the automated tracking algorithm, and the ultimately, the operator.