Broadband Active Sonar Swimmer Detection and Identification

• Published in: The 2006 IEEE International Joint Conference on Neural Network Proceedings pp. 2600 - 2605
• Main Authors: Jae-Byung Jung, Denny, G.F., Tilley, J.W., Kulinchenko, A.B., Simpson, P.K.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 2006
• Subjects: Computerized monitoring; Detectors; Marine animals; Neural networks; Pulse compression methods; Security; Sonar detection; Target tracking; TCPIP; Underwater tracking
• ISBN: 9780780394902; 0780394909
• ISSN: 2161-4393; 2161-4407
• DOI: 10.1109/IJCNN.2006.247137

This paper proposes a viable solution to the port security swimmer detection/identification problem using a broadband active sonar system with TCP/IP interface. Broadband active sonar, characterized by compressed pulse and wide-ranging spectral features, offers the ability to locate an underwater target accurately and classify it as a swimmer (diver) apart from fish or marine mammals, thus considerably reducing the false alarm rate compared with existing energy-based detection systems. Coupled with tracking algorithms, the fusion output detector can further reduce false alarms, allowing high-probability automatic alerts to security forces that can't continuously and consistently monitor this sector. Applications of conventional back-propagation algorithms or probabilistic neural networks in cooperation with track-based averaging technique show promising performance to this real-world classification problem. Several testing efforts conducted in both Alaska and Washington state waters have demonstrated the applicability of the technology and highlighted aspects of deployment around facilities such as ferry terminals and cruise ship docks for homeland security.