Techniques for Automatically Exploiting Passive Acoustic Sonar Data Navy SBIR FY2006.2 Sol No.: Navy SBIR FY2006.2 Topic No.: N06-138 Topic Title: Techniques for Automatically Exploiting Passive Acoustic Sonar Data Proposal No.: N062-138-0843 Firm: Lakota Technical Solutions, Inc. PO Box 1180 Laurel, Maryland 20725 Contact: J. Pence Phone: (301) 725-1700 Web Site: www.lakota-tsi.com Abstract: Anti-submarine warfare (ASW) has long been a capability critical to the survivability of surface and sub-surface naval assets. Due to their ability to detect threats without revealing an asset's location, passive acoustic sensor technologies have proven to be an invaluable component of ASW operations. However, passive ASW operators have carried a disproportionate share of the burden of detecting, classifying, and localizing acoustic signatures by searching through Passive Broadband (PBB) and Passive Narrowband (PNB) waterfall displays. This searching process is labor intensive and its effectiveness is limited to a Cold War environment. As the Navy endeavors to maintain superiority into the 21st century, it must be capable of performing ASW operations in littoral environments, where passive ASW operators will experience a dramatic increase in workload due to an increase in the number of acoustic sources, which may compromise asset survivability. To counter the increase in operator workload, PBB and PNB data must be exploited to assist in target detection. Lakota proposes a Feature-Aided Track-While-Detect approach to automatically confirm potential targets using PBB and PNB feature data. The proposed solution provides a low controllable false alarm rate, as well as target existence confidence. Benefits: The proposed solution is applicable to a wide variety of commercial applications that use multi-modal sensing technologies. Defense and Homeland Security applications include: mine detection, dismounted troop tracking, and border incursion detection. Medical applications include diagnostic Hyperspectral Imaging (HSI) as well as Cell Migration Analysis (CMA) for biological cell classification. In all of these applications, multi-model sensor data is exploited to more accurately detect, classify and track objects of interest. Although the objects of interest are different for each of these examples (e.g., mines, people, and biological cells), the proposed Feature-Aided Track-While-Detect technique is generally applicable to sensor systems that produce multiple, independent, heterogeneous features. Return