Title :
A new 2-D LMS adaptive algorithm
Author :
Ohki, Makoto ; Hashiguchi, Sumihisa
Author_Institution :
Fac. of Eng., Yamanashi Univ., Kofu, Japan
Abstract :
A novel adaptive LMS (least mean square) algorithm is presented in which the algorithm can update the filter coefficients along both the horizontal and the vertical directions on a 2-D plane. Both the conventional algorithm and the new 2-D LMS algorithm are applied to the identification of unknown 2-D systems with stationary or nonstationary characteristics. The learning curves and the mean square errors show that the new 2-D LMS adaptive algorithm is particularly suitable for processing 2-D nonstationary signals. However, it converges slowly for stationary inputs
Keywords :
adaptive filters; identification; least squares approximations; signal processing; two-dimensional digital filters; 2-D LMS adaptive algorithm; 2-D nonstationary signals; 2-D systems; adaptive filters; filter coefficients update; identification; learning curves; least mean square; mean square errors; nonstationary characteristics; signal processing; stationary inputs; Adaptive algorithm; Adaptive filters; Anisotropic magnetoresistance; Convergence; Degradation; Least squares approximation; Mean square error methods; Noise reduction; Signal processing; Two dimensional displays;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1991. ICASSP-91., 1991 International Conference on
Conference_Location :
Toronto, Ont.
Print_ISBN :
0-7803-0003-3
DOI :
10.1109/ICASSP.1991.150823
