High performance SVD-like procedure for spectral estimation using Rayleigh function estimates
Document typeConference report
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
Rights accessOpen Access
The author describes how Rayleigh estimates can be viewed as a method which performs singular-value decomposition (SVD) procedure without doing it. As a short cut to get principal-component reduction, Rayleigh quotients allow the resolution of frequency detectors yet preserve the asymptotic behavior of the actual power spectral density. In a filtering framework the estimate is extended to adaptive schemes and 2-D spectral estimation. The resulting estimate provides the means for adaptive processing with low computational complexity. It avoids also the crucial decision between signal subspace and noise subspace which promotes undesired distortion and false peaks in spectral estimation applications
CitationLagunas, M. High performance SVD-like procedure for spectral estimation using Rayleigh function estimates. A: IEEE International Conference on Acoustics, Speech, and Signal Processing. "ICASSP 1988: ICASSP 88: 1988 International Conference on Acoustics, Speech, and Signal Processing: New York: 11 April 1988-14 April 1988". Seattle, Washington: Institute of Electrical and Electronics Engineers (IEEE), 1998, p. 2368-2371.