Fast Recursive Least Squares Algorithm for Acoustic Echo Cancellation Application

Adaptive filtering is used in a wide range of applications including echo cancellation, noise cancellation and equalization. In these applications, the environment in which the adaptive filter operates is often non-stationary. For satisfactory performance under non-stationary conditions, an adaptive filtering is required to follow the statistical variations of the environment. Tracking analysis provides insight into the ability of an adaptive filtering algorithm to track the changes in surrounding environment. The tracking behavior of an algorithm is quite different from its convergences behavior. While convergence is a transient phenomenon, tracking is a steady-state phenomenon. Over the last decade a class of equivalent algorithms such as the Normalized Least Mean Squares algorithm (NLMS) and the Fast Recursive Least Squares algorithm (FRLS) has been developed to accelerate the convergence speed. In acoustic echo cancellation context, we propose in this paper to use numerically stable Fast Recursive Least Squares algorithm to improve the quality and the intelligibility of the enhanced speech.