LFA and CFLA Acoustic Sensors

This Phase I project will develop an innovative signal processing and information processing concept for improved detection in acoustically noisy littoral waters using existing towed arrays or hull-mounted arrays. These relatively shallow waters have many sources of noise and interference which make it difficult to track the multiple targets with classical filtering approaches such as particle filters or Kalman filtering. It is a multisource problem because of background noise sources and the natural tendency of the multiple signals to overlap. This project proposes an evolutionary approach by building on currently-used detection algorithms using a new probabilistic framework to determine the instantaneous localization (azimuth, range, and depth) of multiple targets and their trajectories in the short-term by spatio-temporal clustering. Autonomous detection and localization using long-term signal clustering with the Bayesian Information Criterion allows targets to be tracked as they move through the noisy littoral space. Outputs from existing systems can also be directly processed by this framework to further improve sonar performance. The Phase I research objectives are to evaluate with underwater acoustic sounds and establish feasibility of the system for improved performance in active clutter reduction that reduces false alarms and improves performance in detection, classification, tracking, and displays.