作者: B.D. van Veen , L.L. Scharf
DOI: 10.1109/29.57585
关键词: Statistics 、 Estimation theory 、 Algorithm 、 Frequency band 、 Covariance matrix 、 Rank (linear algebra) 、 Covariance 、 Noise (signal processing) 、 Mathematics 、 Estimation of covariance matrices 、 Signal subspace
摘要: A close relationship between low rank modeling and multiple window spectrum estimation is demonstrated by using maximum likelihood estimates of structured covariance matrices. The power in a narrow spectral band estimated estimating the variances signal plus noise model. This model swept through entire frequency to obtain an estimate as function frequency. resulting are given weighted combinations eigenspectra. Each eigenspectrum results from projecting data onto orthogonal component subspace squaring. Thomson (1982) correspond particular choice for matrix framework also used derive center noise. obtained combination >