A novel variable step-size normalized subband adaptive filter based on mixed error cost function

摘要

A novel variable step-size algorithm is proposed for normalized subband adaptive filter. The proposed algorithm is based on the mixed error cost function. By assuming the time-averaging estimate of the priori and posteriori errors equals the variance of subband noise, the step-size is obtained. Therefore, the proposed algorithm has more effective approach to the optimum solution. The power of noise-free subband priori error is obtained by using the shrinkage denoising method. Using the energy conservation method, the mean-square convergence performance analysis is presented. The analysis result shows this algorithm is stable and effective. The simulation results demonstrate the performance of proposed algorithm distinctly outperforms other conventional variable step-size algorithms in both steady-state error and abrupt tracking performance.