Title :
Convergence properties of the frequency domain block-LMS Volterra adaptive filter
Author :
Wilhoyte, Michael E. ; Ogunfunmi, Tokumbo
Author_Institution :
Appl. Signal Technol. Inc., Sunnyvale, CA, USA
Abstract :
Summary form only given, as follows. This paper investigates the convergence properties of the constrained frequency-domain block LMS (CFBLMS) algorithm and the unconstrained frequency-domain block LMS (UFBLMS) algorithm when applied to truncated Volterra systems. As shown by previous work, the range of stability for the UFBLMS algorithm for linear systems is 1/2 of that with the CFBLMS algorithm. This work verifies that this relationship between the UFBLMS algorithm and the CFBLMS algorithm is valid for truncated Volterra systems as well. This paper further shows through simulation results that the performance of the UFBLMS algorithm is comparable to the CFBLMS algorithm if the step size is adaptive
Keywords :
Volterra series; adaptive filters; convergence of numerical methods; filtering theory; least mean squares methods; Volterra adaptive filter; adaptive step size; constrained LMS algorithm; convergence properties; frequency-domain block LMS algorithm; stability; truncated Volterra systems; unconstrained LMS algorithm; Adaptive filters; Convergence; Frequency domain analysis; Least squares approximation; Linear systems; Stability;
Conference_Titel :
Circuits and Systems, 1995., Proceedings., Proceedings of the 38th Midwest Symposium on
Conference_Location :
Rio de Janeiro
Print_ISBN :
0-7803-2972-4
DOI :
10.1109/MWSCAS.1995.510261
