A robust variable step-size frequency-domain adaptive filter algorithm for acoustic echo cancellation

In acoustic echo cancellation (AEC), frequency-domain adaptive filter (FDAF) algorithm is commonly used to improve the computational efficiency and the convergence rate. However, the choice of step size has to compromise between the convergence rate and the steady-state performance. A variable step-size algorithm was presented by Nitsch to tackle this problem, where the spectral convergence state is estimated to control the step size. In this paper, a new approach for the estimation of the optimal step size is proposed. The relationship between the time-domain system mismatch and the frequency-domain spectral convergence state is derived using Parsevals theorem. Moreover, an artificial delay is inserted into the signal path to calculate the time-domain system mismatch. The proposed algorithm achieves both fast convergence rate and low steady-state misadjustment. Simulation results demonstrate the effectiveness and the robustness of the proposed method.