Comparison of spatially smoothed weighted subspace methods
Bhaskar D. Rao · 2002
The performances of spatially smoothed weighted MUSIC and weighted state space methods/ESPRIT are analyzed, and compared for the direction of arrival (DOA) estimation problem. A key outcome of the analysis is the observation that, when a forward-backward spatially smoothed approach is used, optimally weighted MUSIC and optimally weighted state space methods/ESPRIT have identical asymptotic behavior. It is also shown that the forward-backward covariance matrix is better conditioned than the forward-only covariance matrix, providing additional justification for the use of the forward-backward approach. The analysis is shown to be extendable to the time-series frequency estimation problem, thereby providing a unified framework for dealing with DOA estimation and time-series frequency estimation.>