Delayless subband echo cancelers using computationally efficient parallel Kalman filters

Conventional delayless subband filters use LMS or NLMS algorithm for coefficient adaptations in subband domain. It is known that. Kalman filter (KF) algorithm shows better convergence than the conventional LMS/NLMS algorithm but with the cost paid in computational complexity. In this work, we show that introducing parallel Kalman filters (PKFs) in the subband domain performance can be made better both in terms of ERLE and computational complexity than that of the LMS/NLMS algorithms.