A Hybrid Fuzzy Logic-Neural Network Approach For Multi-path Separation Of Underwater Acoustic Signals

TitleA Hybrid Fuzzy Logic-Neural Network Approach For Multi-path Separation Of Underwater Acoustic Signals
Publication TypeConference Paper
Year of Publication2019
AuthorsLee-Leon A., Yuen C., Herremans D.
Conference Name89th IEEE Vehicular Technology Conference
Date Published04/2019
Conference LocationKuala Lumpur, Malaysia
Abstract

Abstract—Underwater acoustic channels are generally recognized as one of the most difficult communication media in use today. One of the most important constraints of underwater communications is the acoustic propagation of the signals. The aim of a multi-path separator is to extract individual correlated signals from their mixtures. We present an algorithm, Tag Receiver, that is similar to the RAKE receiver, but interpreted from the neural network viewpoint. First, the received signal is split into frames. Next, the frames are “tagged” with a predicted number of multi-path in the framed segment via fuzzy logic and feature extraction. Finally, the frame and “tag” number are inputted into a neural network, which separates the signals. The analysis of the proposed algorithm shows an improvement from the conventional RAKE receiver in terms of BER.