Statistical analysis of subspace-based estimation of reduced-rank linear regressions

Author

Gustafsson, Tony ; Rao, Bhaskar D.

Author_Institution

Dept. of Electr. & Comput. Eng., California Univ., San Diego, La Jolla, CA, USA

Volume

50

Issue

1

fYear

2002

fDate

1/1/2002 12:00:00 AM

Firstpage

151

Lastpage

159

Abstract

A number of signal processing and system identification problems include linear regressions with a reduced-rank regression matrix. A typical step in "subspace-based" algorithms is to apply the singular value decomposition (SVD) to compute a low-rank factorization. However, it is not clear how certain weighting matrices should be defined for best possible accuracy. We present a statistical analysis of the estimate of the reduced-rank regression matrix, and we discuss a couple of approaches for finding weighting matrices

Keywords

identification; parameter estimation; signal processing; singular value decomposition; statistical analysis; SVD; low-rank factorization; reduced-rank linear regressions; reduced-rank regression matrix; signal processing; singular value decomposition; statistical analysis; subspace-based algorithms; subspace-based estimation; system identification; weighting matrices; Eigenvalues and eigenfunctions; Linear regression; Matrix decomposition; Noise measurement; Random processes; Signal processing algorithms; Singular value decomposition; State-space methods; Statistical analysis; System identification;

fLanguage

English

Journal_Title

Signal Processing, IEEE Transactions on

Publisher

ieee

ISSN

1053-587X

Type

jour

DOI

10.1109/78.972491

Filename

972491