Rational Spectral Estimation: A High Performance-Fast Algorithmic Procedure

摘要

It is widely recognized that spectral estimation serves as a powerful digital signal processing tool in such diverse areas as radar, array processing, adaptive filtering, seismology, and, speech processing. In this paper, a novel method of rational spectral estimation shall be presented which possesses a number of admirable properties: (1) it has an elegant algebraic structure, (2) its spectral estimation performance has been empirically found to be superior to such contemporary methods as the maximum entropy and Box-Jenkins methods, and (3) it is implementable by a 'fast algorithm, which is computationally competitive with recently developed fast LMS algorithms. Taken in combination, these properties mask this new method as being a primary spectral estimation tool in those challenging applications requiring high performance spectral estimation in a real time setting. The new method is predicated on selecting the parameters of a rational spectral model so that the underlying Yule-Walker equations are best approximated over an extended time span.