Autonomous Classification of Acoustic Signals

Actionable situational awareness in cluttered and littoral environments with a passive sensor network requires a cost effective system capable of a high probability of detection of low-level undersea sound sources in large shallow water areas. Distributed passive arrays and autonomous sensor platforms have the potential for persistent monitoring of surface and subsurface acoustic targets. However these sensor platforms generate a tremendous amount of data that would require a great deal of operator supervision and detailed understanding of target signatures. 3 Phoenix, Inc. (3 Phoenix) has teamed with the Integrity Applications Inc. (IAI) to develop a robust suite of detection, classification, and localization (DCL) algorithms that will improve automated target recognition (ATR) of surface and subsurface contacts in high clutter littoral environments. We propose novel feature extraction methods in tandem with an efficient nonlinear adaptive kernel elastic net (AKEN) classification. The proposed DCL engine will be optimized for situational awareness within an operating scenario consisting of cluttered littoral environments. Efficient methods of implementation will be derived to enable real-time algorithm operation on existing hardware/firmware platforms such as the Persistent Littoral Undersea Surveillance (PLUS) processor.