SIGNAL PROCESSING USING HIGHER-ORDER STATISTICS AND WAVELETS

PDF

Authors
  1. Swingler, D.N.
Corporate Authors
Defence Research Establishment Atlantic, Dartmouth NS (CAN);Saint Mary's Univ, Halifax NS (CAN) Div of Engineering
Abstract
The document comprises two separate reports, each on a different facet of modern signal processing. One concerns Higher-Order Statistics, the other, Wavelets. These reports were produced under the same DREA contract and are bound together for convenience. The section on Higher-Order Statistics comprises an introduction to some basic theory associated with the field, followed by a detailed review of four papers in the unclassified literature whose content is of interest to DREA. The overall conclusion is that while Higher-Order processing is of considerable academic interest, due primarily to its blindness to additive Gaussian noise, there are problems associated with the statistical stability of the estimates of highedr-order statistical quantities from finite data. This suggests practical application of the technique should be approached with caution. The material on Wavelet processing is a collection of simulation experiments based on a few of the more common time-frequency analysis techniques, including Wavelets, applied to the problem of detecting damped exponential transients in noise. It is demonstrated that due to a closeness between Wavelet processing and Matched Filtering, the former can, in appropriate circumstances, be the optimum method. However, in more realistic scenarios the Short-Time Fourier Transform method remains competitive.
Keywords
Wavelets
Report Number
DREA-CR-92-440 — Contractor Report
Date of publication
01 Jun 1992
Number of Pages
86
DSTKIM No
93-03625
CANDIS No
132539
Format(s):
Hardcopy;Originator's fiche received by DSIS;Document Image stored on Optical Disk

Permanent link

Document 1 of 1

Date modified: