Adaptive filtering in acoustic echo cancellation systems — A practical overview

This paper presents a practical overview on the adaptive algorithms used in acoustic echo cancellation (AEC) systems. It describes first the Least Mean Square (LMS) family and the variable step size (VSS) corresponding versions. It includes also the Recursive Least Squares (RLS) family, focusing on the benefits introduced by this class of adaptive algorithms (e.g., faster convergence). Finally, it discusses the QR decomposition (QRD) used for real-world implementations, which leads to the QRD-RLS algorithm. The conclusions are based on simulations considering speech signals and the echo path models from G168 ITU-T Recommendation.