JOINED LMS AND LMF THRESHOLD TECHNIQUE FOR DATA ECHO CANCELLATION

The goal of this paper is to introduce the joined LMS and LMF adaptive algorithm (LMFSF) for data echo cancellation. The idea is to use an error estimate as smooth as possible, and after comparison with the selected thresholds to decide whether for the next step we apply LMS or LMF. In order to avoid oscillations from LMS to LMF and back, we use two hysteresis loops. It will be shown through simulations in the data echo cancellation framework that the new algorithm resulted is better than LMS and LMF from both the convergence speed or residual error point of view.