A Feed-forward Switched Adaptive Filtering configuration for Underwater Acoustic Signal Denoising Technique with low-complexity

The Underwater acoustic communication is difficult by virtue of the noise that exists undersea. This can be caused by both natural phenomena, like waves and currents, as well as artificial sources such as ships and boats. In order to reduce or eliminate this interference, it is necessary to process the signal before further processing. One way of doing this is through denoising. However, because noise constantly fluctuates in intensity, predicting its future behavior becomes almost impossible. To address this problem, we propose a new method for filtering underwater communications using an superlative feedfoward switched adaptive filter model using Signed Error variant of Least Mean Square (SELMS) and Signed Data variant of Least Mean Square (SDLMS) algorithm that takes into account signed form LMS values. The effectiveness of the filter is tested using a clean fish sound that has been tainted by noises from underwater vessels from the ShipsEar database.