Adaptive algorithms for eigenstructure based spectral estimation and filtering

This paper considers the problem of adaptive estimation of the complete set of eigenvectors and eigenvalues of a data covariance matrix. After considering some applications of eigenstructure algorithms to spectral estimation and constrained filtering problems, the paper discusses a new approach to data adaptive eigenestimation. A new technique based on an adaptive implementation of the Q-R algorithm is presented. This algorithm offers a strategy for recursive estimation of the complete set of covariance matrix eigenvectors and eigenvalues, which is computationally efficient to implement.