Title :
Constrained total least squares: signal and system modeling
Author :
Cadzow, J.A. ; Wilkes, D.M.
Author_Institution :
Dept. of Electr. Eng., Vanderbilt Univ., Nashville, TN, USA
Abstract :
The concepts of least squares and total least squares as applied to a system of linear equations are standard techniques in mathematical analysis. They have become an indispensable tool in the development of signal processing algorithms. The notion of constrained total least squares is introduced and its applications to the problems of exponential data modeling and system identification are briefly developed. It is shown that a significant improvement in performance is achieved over conventional methods
Keywords :
constraint theory; identification; least squares approximations; signal processing; constrained total least squares; exponential data modeling; signal processing; system identification; Least squares methods; Linear systems; Mathematical analysis; Mathematical model; Modeling; Nonlinear dynamical systems; Nonlinear equations; System identification; Vectors;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1992. ICASSP-92., 1992 IEEE International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-0532-9
DOI :
10.1109/ICASSP.1992.226516
