Threshold extension based on a new paradigm for MUSIC-type estimation
M. Kaveh, A. Bassias · International Conference on Acoustics, Speech, and Signal Processing · 2002
Certain functionals of two vectors are proposed as models for the formulation of MUSIC-type parameter estimators. The two vectors are the projections of the steering vector onto (1) the subspace spanned by the noise and the least significant signal eigenvector and (2) the least significant signal eigenvector. It is shown that both MUSIC and minimum-norm estimators are based on special cases of these functionals. It is further shown that the choice of the functional significantly affects estimator performance, with some providing resolution and mean-squared error thresholds similar to rooting techniques which are, only applicable to linear uniform arrays.>