Advances in high-resolution spectral estimation

We review a new approach to spectral estimation, based on the use of filter banks as a means of obtaining spectral interpolation data. This data replaces the standard covariance estimates used in traditional maximum entropy spectral estimation. The new method is based on our recent theory of analytic interpolation with degree constraint and produces suitable pole-zero (ARMA) models, for which the choice of the zeros (MA-part) of the model is completely arbitrary. By suitable choices of filter-bank poles and spectral zeros the estimator can be tuned to exhibit high resolution in targeted regions of the spectrum. A convex optimization approach is presented, which is based on a generalized concept of entropy.