Mitigation of Biologically Induced Active Sonar Reverberation in Littoral Regions

The proposed effort develops and evaluates features exploiting the swim bladder resonance observed in broadband echoes from fish for automatic screening, reducing mid-frequency active sonar clutter. Real world data from shallow water is used to develop and evaluate features for discriminating between the broad peaks characteristic of an aggregate echo from a school of fish and the comparatively flat echo from target and target-like scatterers. Exploitation of this feature is important because biologics can produce high level echoes, move, and are not amenable to other sensing modalities. Because the frequency and sharpness of the resonances depend strongly on the relative density of fish species and their depth, physically motivated features of the spectral shape and auto-regressive coefficients from speech recognition are leading candidates for investigation. Another product of the work is an understanding of the system bandwidth required to achieve reliable automatic screening of fish echoes without significantly reducing target detections. Beyond the benefits of reliable screening, the developed features themselves offer the potential to improve associations in automatic tracking. This project will demonstrate the feasibility of exploiting fish swim bladder resonances to improve automatic screening and tracking performance of U.S. Navy mid-frequency active sonar systems.