•  
  •  
 

Abstract

Wavelet transform is one of methods that utilizes the dilation and contraction of prototype wavelet basis function to form a set of fully flexible windows in the time-frequency plane. It exhibits a good effect in both the analysis of high and low frequency components of signal. By applying such a transform method to traditional least-mean-square algorithm (LMS), we can obtain an orthogonally transformed signal from a time-domain input signal after imposing wavelet transform on it. In addition, we can improve the convergence speed of learning curve by passing the transformed signal through weight which is adapted with LMS. The improved method is called the wavelet-based least-mean-square algorithm (WLMS). In this paper, we impose the WLMS on the traditional adaptive beamformer, the one that used the time-domain input signal with LMS algorithm (TLMS) to adjust the system weights. Instead of TLMS, we adapt WLMS here to improve the convergence speed of adaptive beamformer. Besides, we further demonstrate theoretical analysis and computer simulations to verify the performance of our transform method.

Download
COinS