Normalized Least Mean EE' Algorithm and Its Convergence condition

摘要

the normalized least mean square (NLMS) al- gorithm has thedrawback that the convergence speed of adap- tive filter coefficientsdecreases when the reference signal has high auto-correlation. Atechnique to improve the convergence speed is to apply thedecorrelated reference signal to the calculation of the gradientdefined in the NLMS algorithm.