A hybrid algorithm based on neuron-fuzz and wavelet transforms for wideband sonar detection in a reverberation-limited environment

Abstract

A fast hybrid de-noising algorithm is developed to enhance the performance of wideband acoustic signal detection in a reverberation-limited environment. Making use of the hybrid algorithm, the active sonar echolocation detector is able to estimate motion parameters (radial range and velocity) of a moving target in an effective and efficient manner with very low level of signal-to-reverberation ratio (SRR). The hybrid algorithm is composed of two parts: adaptive noise reduction (ANR) part based on an adaptive intelligent fuzzy system in the continuous wavelet transform (CWT) domain, and target motion estimation (TME) part based on recursive fast wavelets transform. In the ANR operation which serves as prefiltering of the noisy signal, the SRR of a noisy wideband signal is drastically improved by adopting the technique of adaptive neuro-fuzzy inference system (ANFIS). The pre-filtered signal is transformed to the CWT domain and then processed using the recursive TME operation, a combination of discrete wavelet denoising (WDeN) and CWT techniques. Simulation results demonstrate that the proposed hybrid algorithm is not only effective in accurately predicting the motion parameters, but also is more efficient in terms of computational time consumption than the fuzzy detector previously developed on the basis of the ANR operation.