In this communication, fast adaptive algorithms are suggested to enhance the performance of the Fast- Normalized Least Mean Square (FNLMS) algorithm in Acoustic Echo Cancellation (AEC) applications with a sparse system. We propose two new algorithms, the first one is the Zero-Attracting (ZA) FNLMS which gives a better performance when the unknown system is extremely sparse. However, by decreasing the sparsity of the system, the Mean Square Error (MSE) got significantly worse than that of the FNLMS algorithm. To overcome this issue, another algorithm named Reweighted Zero-Attracting FNLMS (RZA-FNLMS) algorithm is proposed in this paper. Simulation results with stationary and non-stationary inputs under different Signal to Noise Ratio (SNR) values of additive noise and change in the impulse response lengths show an improvement in the convergence speed.
This communication deals with the problem of Acoustic Echo Cancellation (AEC) by reduced complexity adaptive filtering algorithms. In this context, we propose to use the Set-Membership filtering concept with the Fast Normalized Least Mean Square Algorithm (SM-FNLMS). The proposed SM-FNLMS algorithm shows a superior convergence speed performances compared with the classical Set-Membership NLMS (SM-NLMS) and NLMS algorithms.