Spectral Estimates of Band-Limited Signals.

Abstract

We consider the problem of estimating the spectrum of a band-limited signal perturbed by additive white noise. Sharp bounds on the mean square errors of linear spectral estimates are computed and expressed as functions of time-bandwidth product sand signal-to-noise power ratios. When the data are sampled at the Nyquist rate, no linear spectral estimator can produce significantly more accurate results than the conventional discrete Fourier transform. In the more general case, consistent spectral estimators can be obtained only if both the data rate and the length of the observation window are increased without bound. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 02, 1979
Accession Number
ADA072776

Entities

People

  • William B. Gordon

Organizations

  • United States Naval Research Laboratory

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Data Rate
  • Discrete Fourier Transforms
  • Equations
  • Estimators
  • Extrapolation
  • Fourier Series
  • Integrals
  • Military Research
  • New York
  • Noise
  • Spectra
  • White Noise

Readers

  • Calculus or Mathematical Analysis
  • Radar Systems Engineering.