Title :
Necessary and sufficient condition on convergence and robustness for a class of iterative algorithms with applications in adaptive filtering
Author_Institution :
Inst. of Math. Sci., Acad. Sinica, Chengdu
Abstract :
The following class of recursive algorithm is considered: [X n+1=Xn+an (B +Cn)Xn+bn ]. Under weaker assumptions about {Cn} than those used for previous results, the necessary and sufficient condition on {bn} for the convergence of this algorithm is established, and its robustness is examined. These results are then applied to an adaptive filtering algorithm in order to obtain better convergence and robustness results than those previously given by the authors (1991)
Keywords :
adaptive filters; convergence of numerical methods; filtering and prediction theory; iterative methods; recursive functions; adaptive filtering; convergence; iterative algorithms; recursive algorithm; robustness; stochastic algorithms; Adaptive filters; Algorithm design and analysis; Convergence; Eigenvalues and eigenfunctions; Filtering algorithms; Iterative algorithms; Robustness; Signal processing algorithms; Stochastic processes; Sufficient conditions;
Conference_Titel :
Decision and Control, 1992., Proceedings of the 31st IEEE Conference on
Conference_Location :
Tucson, AZ
Print_ISBN :
0-7803-0872-7
DOI :
10.1109/CDC.1992.371683
