Signal analysis by triangularization and diagonalization of covariance matrix
Abstract
Factor analysis is applied to signal analysis with signal as the sum of factors. Two methods of factorization are examined in detail: the triangularization of the covariance matrix, which in signal analysis corresponds to linear prediction methods, and the diagonalization of the covariance matrix which corresponds to the KarhunenLoeve transformation. Both methods are equivalent to the reduction of a quadratic form to its canonical form (sum of squares).
 Publication:

L'Onde Electrique
 Pub Date:
 September 1977
 Bibcode:
 1977LOEle..57..525C
 Keywords:

 Covariance;
 Factor Analysis;
 Matrices (Mathematics);
 Signal Analysis;
 Autoregressive Processes;
 Linear Prediction;
 Transformations (Mathematics);
 Vector Analysis;
 Weighting Functions;
 Communications and Radar