An Effective ARMA Modeling Method.

Abstract

The ability to model random time series plays a prominent role in a variety of applications as exemplified by seismic data analysis, doppler radar processing, speech processing, adaptive filtering, and, array processing. Undoubtedly, two of the more popular procedures for effecting such time series models are the classical Fourier (MA) approach and the maximum entropy (AR) method. In this paper, a theoretical comparison of these contemporary procedures with a more general ARMA method will be made. It will be demonstrated that the spectral estimation performance of the ARMA method typically exceeds that of its more special-method typically exceeds that of its more specialized MA and AR counterparts. With this supremacy thus established, a recently developed method for effectively generating ARMA model estimates from time series observations will be then presented. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Apr 01, 1981
Accession Number
ADA101871

Entities

People

  • James A. Cadzow

Organizations

  • Virginia Tech

Tags

Communities of Interest

  • Ground and Sea Platforms
  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Autocorrelation
  • Coefficients
  • Data Analysis
  • Electrical Engineering
  • Engineering
  • Equations
  • Frequency
  • Observation
  • Power Spectra
  • Probability
  • Sequences
  • Signal Processing
  • Spectra
  • Standards
  • Statistical Analysis
  • White Noise

Fields of Study

  • Engineering

Readers

  • Statistical inference.
  • Systems Analysis and Design