A VARIABLE STEP SIZE LMS ADAPTIVE FILTERING ALGORITHM BASED ON THE NUMBER OF COMPUTING MECHANISMS

This paper proposes a novel variable step size LMS algorithm based on the number of computing mechanisms by analyzing the basic principles of the variable step LMS adaptive filtering algorithms. The algorithm establishes a different nonlinear relation-ship between step factor and error signal based on the different sampling time, which not only can control step size by using sampling time indirectly, but also change the parameters to meet the requirement of fast convergence. Compared with the other variable step algorithms, this algorithm is proved to have faster convergence speed and higher stability.