Detection and Time Delay Estimation of Non-Gaussian Signals in Noise
Abstract
Gaussian time series have the property that their higher order spectra are identically zero. In many signal processing applications, the noise fields are predominantly Gaussian. Thus if the signal has a nonzero higher order spectra (i.e., the signal is non-Gaussian), then higher order spectral processing of the combined signal plus noise field may provide processing gains over more traditional processing methods that rely on lower order properties of the signal. However, any practical signal processing applications of higher order spectra rely on estimates of the higher order spectra made from a finite number of data samples, and thus the statistical properties of various estimators of the higher order spectra determine the processing gains that may be achieved in practice for specific signals and noise. Therefore, to determine the utility of higher order spectral processing it is necessary to determine estimators that are appropriate for specific signal processing applications and determine the performance of those estimators for the signal and noise characteristics of the specific application based on the derived statistical properties of the estimators. This report summarizes the work performed under Contract N00014-87-K-0785 in which statistical estimators for the detection and time delay estimation of non-Gaussian signals were defined and their statistical properties in presence of noise were derived. (rh)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 31, 1990
- Accession Number
- ADA227046
Entities
People
- Gary R. Wilson
Organizations
- University of Texas at Austin