DocumentCode
1022157
Title
An ESPRIT-like method for quadratic phase coupling estimation
Author
Parthasarathy, Harish ; Prasad, Surendra ; Joshi, Shiv Dutt
Author_Institution
Dept. of Electr. Eng., Indian Inst. of Technol., Bombay, India
Volume
43
Issue
10
fYear
1995
fDate
10/1/1995 12:00:00 AM
Firstpage
2346
Lastpage
2360
Abstract
Two algorithms are proposed for estimating the quadratically coupled frequency pairs (QC pairs) in a signal consisting of complex sinusoids in white noise. Three matrices are constructed from the complex third-order cumulants of the noisy signal, the second and third being time shifted versions of the first. The list of coupled frequencies is obtained from the rank reducing numbers of the matrix pencil formed from the first matrix and either of the latter two. The first algorithm then pairs these components by relating quadratic coupling to the intersection of generalized eigenspaces corresponding to two of these frequencies. The coupling strengths are also obtained in terms of generalized eigenvectors in this intersection space. The second algorithm constructs a two-parameter matrix pencil using all the three matrices. The rank reducing pairs of this pencil on the unit circle yield the QC pairs and the associated generalized eigenvectors: the coupling strengths
Keywords
eigenvalues and eigenfunctions; frequency estimation; higher order statistics; matrix algebra; phase estimation; signal processing; white noise; ESPRIT-like method; QC pairs; complex sinusoids; complex third-order cumulants; coupling strengths; generalized eigenspaces; generalized eigenvectors; intersection space; matrix pencil; noisy signal; phase coupling estimation; quadratically coupled frequency pairs; rank reducing numbers; rank reducing pairs; two-parameter matrix pencil; white noise; Biomedical signal processing; Data analysis; Electroencephalography; Frequency estimation; Gaussian noise; Information analysis; Phase estimation; Signal analysis; Signal processing; Signal processing algorithms;
fLanguage
English
Journal_Title
Signal Processing, IEEE Transactions on
Publisher
ieee
ISSN
1053-587X
Type
jour
DOI
10.1109/78.469855
Filename
469855
Link To Document