Highly Parallel Eigenvector Update Methods With Applications To Signal Processing
R.D. DeGroat, Richard A. Roberts · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1986
Updating the eigenvalue decomposition (EVD) of the estimated correlation matrix arises in many signal processing problems. In this paper, we present an improved, highly parallel, linear rank-one EVD update algorithm which is based on a quadratic rank-one method due to Bunch, et al [1]. Both methods are evaluated and compared. A number of simulations involving the frequency estimation of non-stationary narrow-band signals are also performed in which linear prediction techniques are used with eigenvector pre-processing.