Unconventional Signal Processing Using the Cone Kernel Time Frequency Representation

Early detection, localization, and classification of submarine threats is an on-going mission within Anti-Submarine Warfare (ASW) programs. To counter the increasingly quiet passive signatures of these threats, the overall Figures of Merit of tactical and strategic surveillance systems have been improved. But as the resulting detection ranges and search volumes increase, so does the frequency and number of detections. Increased contact loading forces the operator to classify contacts more rapidly, reducing his effectiveness. This study addresses the contact loading problem directly by investigating the applicability of the Cone Kernel Time Frequency Representation (CK-TFR) detection and feature extraction technique to ASW. The study will also explore intuitive man-machine interface (MMI) techniques that assist the sonar operator in managing the contact workload. Our approach uses a real-- time signal processing testbed and display system to evaluate this alternative technique, identify new classification methods and clues, and explore advanced MMI concepts. Specific issues to be examined include characterizing the performance of the CK-TFR algorithm in a real ASW signal environment, identification of additional potential classification clues, such as unique time-frequency patterns of transients, and user displays and controls which optimize operability.