A Globally Convergent Matricial Algorithm for Multivariate Spectral Estimation

Author

Ramponi, Federico ; Ferrante, Augusto ; Pavon, Michele

Author_Institution

Autom. Control Lab., ETH Zurich, Zurich, Switzerland

Volume

54

Issue

10

fYear

2009

Firstpage

2376

Lastpage

2388

Abstract

In this paper, we first describe a matricial Newton-type algorithm designed to solve the multivariable spectrum approximation problem. We then prove its global convergence. Finally, we apply this approximation procedure to multivariate spectral estimation, and test its effectiveness through simulation. Simulation shows that, in the case of short observation records, this method may provide a valid alternative to standard multivariable identification techniques such as Matlab´s PEM and Matlab´s N4SID.

Keywords

approximation theory; autoregressive moving average processes; convergence of numerical methods; mathematics computing; convex optimization; globally convergent matricial algorithm; matricial Newton-type algorithm; multivariable identification techniques; multivariable spectrum approximation problem; multivariate spectral estimation; Algorithm design and analysis; Approximation algorithms; Automatic control; Computer languages; Convergence; Entropy; Filters; Frequency; Interpolation; Testing; Convex optimization; Hellinger distance; global convergence; matricial Newton algorithm; multivariable spectrum approximation; spectral estimation;

fLanguage

English

Journal_Title

Automatic Control, IEEE Transactions on

Publisher

ieee

ISSN

0018-9286

Type

jour

DOI

10.1109/TAC.2009.2028977

Filename

5256185