Otimização convexa aplicada à redução de ruído impulsivo em sistemas de comunicação acústica sob a água

Abstract

In warm shallow waters it is common the presence of an impulsive-like acoustic noise generated by snapping shrimps. This noise degrades the performance of underwater acoustic communication systems if the detectors of these systems are designed for the pure Gaussian noise channel. Aiming at improving the robustness of such systems against the snapping shrimp noise, this dissertation proposes a noise reduction technique based on convex optimization. The underwater acoustic communication system was simulated using the Matlab software tool, and the proposed noise reduction technique was treated by the CVX, which is a system for modeling and solving convex optimization problems, developed to work under the Matlab platform. Techniques for noise reduction based on nonlinear detectors were also considered for comparison purposes. The results unveil a significant performance improvement brought by the proposed technique in all considered scenarios, with no need for adaptation or knowledge of any parameter. When compared with the nonlinear detector, the proposed noise reduction has unveiled more efficacy in the high signal-to-noise regime, again for all considered scenarios.