Abstract
As previously shown, the normalized least mean square (NLMS)algorithm has superior convergence properties than the least meansquare(LMS) algorithm. However, the weight noise effect of the NLMS algorithm is large so that the steady state residue power islarger than that for the LMS algorithm. A generalized NLMSalgorithm is developed based upon the pseudoinverse of anestimated covariance matrix. A preliminary evaluation indicatesimproved performance can be attained but the implementationcomplexity might be high.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have