A Variable Step-Size Strategy Based on Error Function for Sparse System Identification

• DOI: 10.1007/s00034-016-0344-1
• Publication Date: 2016-05-30T06:05:02Z
• Authors: Tao Fan; Yun Lin
• Keywords: 0202 electrical engineering, electronic engineering, information engineering; 02 engineering and technology

The well-known reweighted zero-attracting least mean square algorithm (RZA-LMS) has been effective for the estimation of sparse system channels. However, the RZA-LMS algorithm utilizes a fixed step size to balance the steady-state mean square error and the convergence speed, resulting in a reduction in its performance. Thus, a trade-off between the convergence rate and the steady-state mean square error must be made. In this paper, utilizing the nonlinear relationship between the step size and the power of the noise-free prior error, a variable step-size strategy based on an error function is proposed. The simulation results indicate that the proposed variable step-size algorithm shows a better performance than the conventional RZA-LMS for both the sparse and the non-sparse systems.