Model-based tracking for autonomous arrays

Over the last 30 years, model-based approaches to array signal processing have evolved considerably to the point where initial skepticism has been replaced by casual acceptance. Still challenges remain. In this work the authors are progressing to a fully autonomous implementation in which all of the processing is done within the array. Target tracks are then passed to the surface using acoustic links. This requirement is a big challenge in that the algorithms must be fast to operate in real-time on modest embedded computers. They also must be robust enough to function without human invention. They describe one approach and its performance in an experimental sea test with a prototype autonomous array.