Introducing the spectral diversity.

The use of linear time-invariant IIR filters in the estimation of the autocorrelation function in time-varying environment overcomes the inherent tradeoff problem between temporal and spectral resolution associated with a single estimator. The filter allows simple recursive and nonrecursive generati...

Full description

Bibliographic Details
Main Author: Amin, Moeness G.
Format: Villanova Faculty Authorship
Language:English
Published: 1993
Online Access:http://ezproxy.villanova.edu/login?url=https://digital.library.villanova.edu/Item/vudl:173378
Description
Summary:The use of linear time-invariant IIR filters in the estimation of the autocorrelation function in time-varying environment overcomes the inherent tradeoff problem between temporal and spectral resolution associated with a single estimator. The filter allows simple recursive and nonrecursive generation of a set of power spectrum estimators (PSE) which vary in performance and characteristics. This variation depends on the locations of the poles and zeros of the employed filter as well as its realization form. The proper choice of the filter singularities provide simultaneous access to a class of distinct PSE. Estimators with high spectral resolution can be used during periods of stationarity to detail the power distribution of the data over frequency, while others with fast convergence can be examined for detection of transients in signal statistics. This paper focuses on the cascade realization form and shows the changes in the estimator characteristics as it travels through the structure. In this context, a desired order of cascading the poles and zeros of a given filter must lead to a high spectral diversity, i.e., allow the filter structure to enclose fine spectral resolution and temporal resolution estimators.