System and Algorithm Concepts for Automatic Detection and Classification of Nontraditional Acoustic Signals

Under this SBIR project, ORINCON proposes to develop and demonstrate algorithms to perform automated detection and classification of cavitational signatures. Several detection and classification algorithms will be used to ¿decompose¿ broadband energy envelopes. Selected regions of the broadband spectrum associated with targets of interest will first be detected with a broadband energy detector and then classified as being ¿cavitation-like¿ or ¿non-cavitation-like¿ energy. Algorithms to be demonstrated under this program include techniques designed to exploit signature components contained in the amplitude, phase, and frequency components associated with the modulation of broadband energy sources, such as acoustic cavitation. Multiharmonic detection processing will be used to extract classification information. Broadband ambiguity functions between pairs of sonobuoys will be generated from time series containing nontraditional signals such as modulated or nonmodulated cavitation. Families of Receiver Operating Characteristic (ROC) curves for signal-to-noise (SNR) values ranging from +10 to -20 dB will be generated using both simulated and real data to measure performance at both the algorithm and system levels.