Title :
A theoretical analysis of the median LMF adaptive algorithm
Author :
Bysted, Toinnzy K. ; Rusu, Corneliu
Author_Institution :
Dept. of Inf. Technol., Tech. Univ., Lyngby, Denmark
Abstract :
Higher order adaptive algorithms are sensitive to impulse interference. In the case of the LMF (Least Mean Fourth), an easy and effective way to reduce this is to median filter the instantaneous gradient of the LMF algorithm. Although previous published simulations have indicated that this reduces the speed of convergence, no analytical studies have yet been made to prove this. In order to enhance the usability, this paper presents a convergence and steady-state analysis of the median LMF adaptive algorithm. As expected this proves that the median LMF has a slower convergence and a lower steady-state error compared to the LMF
Keywords :
adaptive filters; adaptive signal processing; approximation theory; convergence of numerical methods; filtering theory; impulse noise; median filters; sensitivity analysis; convergence analysis; higher order adaptive algorithms; instantaneous gradient; least mean fourth algorithm; median LMF adaptive algorithm; median filter; steady-state analysis; steady-state error; Adaptive algorithm; Adaptive filters; Algorithm design and analysis; Convergence; Equations; Filtering algorithms; Information technology; Laboratories; Least squares approximation; Steady-state;
Conference_Titel :
Electronics, Circuits and Systems, 1999. Proceedings of ICECS '99. The 6th IEEE International Conference on
Conference_Location :
Pafos
Print_ISBN :
0-7803-5682-9
DOI :
10.1109/ICECS.1999.813198