Title :
Total least squares system identification and frequency estimation for overdetermined model orders
Author :
Davila, Carlos E.
Author_Institution :
Electr. Eng. Dept., Southern Methodist Univ., Dallas, TX, USA
Abstract :
A method of obtaining accurate parameter and frequency estimates using the total least squares (TLS) method when the model order is overdetermined is described. The proposed method finds the linear combination of the noise subspace eigenvectors which yields the sought-after unknown system parameters. The proposed approach was demonstrated to perform better than the minimum-norm method in a parameter identification and a frequency estimation experiment.<>
Keywords :
eigenvalues and eigenfunctions; least squares approximations; parameter estimation; signal processing; frequency estimation; noise subspace eigenvectors; overdetermined model orders; parameter identification; total least squares;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1993. ICASSP-93., 1993 IEEE International Conference on
Conference_Location :
Minneapolis, MN, USA
Print_ISBN :
0-7803-7402-9
DOI :
10.1109/ICASSP.1993.319696
