The Application of the Wavelet-Domain LMS Algorithm to Adaptive Beamformer

• Main Authors: Chin-Chang Chang, 張晉彰
• Other Authors: Shun-Hsyung Chang
• Format: Others
• Language: zh-TW
• Published: 1995
• Online Access: http://ndltd.ncl.edu.tw/handle/27625351994142430668

碩士 === 國立海洋大學 === 電子工程學系 === 83 === Wavelet transform is one way which utilize dilation and contraction of prototype wavelet basis function, and forming a set of windows which is full of flexibility on the time-scale plane. It has a good effect on the analysis of high or low component of signal. By such a transform method for traditional least-mean-square algorithm (LMS), we can obtain orthogonally transformed signal by time-domain input signal after wavelet transform. As a result, we can improve the convergence speed of learning curve by passing the transformed signal through weight adapted with LMS. The improved method is called wavelet-domain least-mean-square algorithm (WLMS). In this thesis, we try to impose WLMS on the inputted signals of both stationary and nonstationary, and make a comparison between WLMS and time- domain LMS (TLMS). In addition, two kinds of convergence factors are used, including (1)constant convergence factor and (2)time- varying convergence factor. We use theoretical analysis and computer simulation to explain the performance of this transform method. It is very important to design the adaptive beamformer in radar, sonar and signal processing system. The traditional adaptive beamformer uses the time- domain input signal with LMS algorithm to adjust the system- weights so as to reduce the component of jammers and receive the desired signal exactly. However, time-domain input signal with LMS algorithm has the defects of slow convergence speed. Instead of TLMS, we adopt WLMS to improve the convergence speed. By passing through the multiresolution analysis of wavelet transform, it will promote the speed for adjusting weights in the beamformer. It will also exactly suppress jammers in time so as to improve the efficiency for traditional adaptive beamformer.