Title :
Natural gradient algorithm for cyclostationary sources
Author :
Jafari, M.G. ; Chambers, J.A. ; Mandic, D.P.
Author_Institution :
Centre for Digital Signal Process. Res., King´´s Coll., London, UK
fDate :
7/4/2002 12:00:00 AM
Abstract :
A new approach to blind source separation of cyclostationary sources is introduced which incorporates a cyclic pre-whitening operation within the learning rule, and thereby provides a new member of the family of natural gradient algorithms. The technique improves the convergence properties of the natural gradient algorithm for complex valued, cyclostationary signals. Simulations show the improved convergence speed of the approach
Keywords :
convergence of numerical methods; gradient methods; signal processing; signal sources; blind source separation; convergence properties; cyclic pre-whitening operation; cyclostationary signal; learning rule; natural gradient algorithm;
Journal_Title :
Electronics Letters
DOI :
10.1049/el:20020503
