A New Variable Step-Size BLMS Algorithm Based on Discrete Wavelet Transforms

摘要

The wavelet transforms domain LMS algorithm is integrated with variable step-size LMS algorithm and BLMS algorithm, from which a new wavelet transforms domain variable step-size BLMS adaptive algorithm is presented. The algorithm has advantages of the above three algorithms, it can reduce the cross-correlation of input signals effectively and can overcome the conflict between high convergence rate and low steady-stare maladjustment in LMS algorithm which leading by fixed step-size, it also has well real time performance. The computer simulation results indicate that the new algorithm has higher convergence rate and lower maladjustment noise, it can be applied effectively in the adaptive systems.