Filtering of Time Correlated Data,

Abstract

Convential filter theory neglects autocorrelation of errors in input data. The assumption that this can be done lead to erroneous asymptotic behavior of filter smoothing factors at high data rates. This has been demonstrated by introducing a more realistic correlation model into the theories of the moving arc least squares filter and a simple recursive filter. The smoothing factor of the former in the revised theory is asymptotically independent of rate, whereas the conventional theory predicts 1/(square root of r) behavior. The theory of recursive filter exhibits a finite optimum rate. The optimum rate is infinite if correlation of input data errors is neglected.

Document Details

Document Type
Technical Report
Publication Date
Jan 01, 1972
Accession Number
AD0750320

Entities

People

  • Richard H. Duncan

Tags

DTIC Thesaurus Topics

  • Autocorrelation
  • Correlation Techniques
  • Data Rate
  • Data Science
  • Filters
  • Filtration
  • Information Science
  • Lepidoptera
  • Mathematical Analysis
  • Mathematical Filters
  • Mathematics
  • Recursive Filters
  • Square Roots

Readers

  • Approximation Theory.
  • Electromagnetic Wave Scattering and Antenna Radiation Engineering