Design Of Feedback Control System For Noise Removal From Underwater Acoustic Data

Abstract

In order for underwater acoustic signals to convey valuable information, the highly dynamic broadband noise must be separated adequately from the narrowband gated signal pulses. It is intricate for a single system with a non-adaptive filter to recover the narrowband signals with varying frequencies. In an orthodox setup, an adaptive filter is reliant on a reference signal. In this paper, we propose a system, based on the leaky normalized least means square (leaky NLMS) adaptive filter, to accurately retrieve the gated acoustic signal pulses. The design works by exploiting the correlation of the required signal with its time-delayed version by using its delayed version as the reference signal. The broadband noise, produced by the highly vibrant underwater environment, has a sizeable variance inhibiting the filter coefficients from adapting to it. Moreover, down sampling the signal shrewdly reduces the high filter order obligations. The fixed- and floating-point results from the MATLAB and HDL designs attest that signals are retrieved with a high Signal to Noise Ratio (SNR), brisk convergence rate, and a negligible quantization error validating the performance of the designed system.