Adaptive Variable-step Size LMS Filtering Algorithm and Its Analysis

In order to improve the performance of the variable step-size LMS(Least Mean Square)adaptive filtering algorithm,a novel algorithm based on the Sigmoid nonlinear functional relationship between the step-size and the error signal was proposed.The step-size of the algorithm increases adaptively at the beginning of the algorithm or when the channel is varying with time,and it would be smaller during the steady state.And the algorithm avoids the effects of the irrelevant noise.Furthermore,the proposed algorithm overcomes the deficiency of Sigmoid functional relationship in the process of step-size change of adaptive steady state.Computer simulation results verify the theoretical analysis and indicate that the algorithm outperforms the former algorithms.