Estimation of Parameters of Non-Gaussian Non-Zero Mean Autoregressive Processes with Application to Optimal Detection in Colored Noise

Abstract

The problem addressed in this paper is that of estimating signal and noise parameters from a mixture of Non-Gaussian autoregressive (AR) noise with partially known deterministic signal. Two models are considered in order to examine different kinds of additive mixing. The Cramer-Rao bounds to the joint estimation of the signal amplitude and the noise parameters are presented. A computationally efficient estimator, which was previously proposed for estimation in the absence of signal, is extended for the two models under consideration. The proposed method essentially consists of two stages of least squares (LS) estimation which is motivated by the maximum likelihood estimation (MLE). The technique is then applied to the problem of detecting a signal known except for amplitude in colored non-Gaussian noise. Two slightly different mixing models are used and a generalized likelihood ratio test (GLRT), coupled with the proposed estimation scheme, is used to solve the problems. The results of computer simulations are presented as an evidence of the validity of the theoretical predictions of performance.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1988
Accession Number
ADA199919

Entities

People

  • Debasis Sengupta
  • Steven Kay

Organizations

  • University of Rhode Island

Tags

DTIC Thesaurus Topics

  • Computer Simulations
  • Computers
  • Data Science
  • Detection
  • Detectors
  • Electrical Engineering
  • Equations
  • Estimators
  • False Alarms
  • Gaussian Noise
  • Gaussian Processes
  • Information Science
  • Probability
  • Security
  • Simulations
  • Statistics
  • Warning Systems

Fields of Study

  • Engineering

Readers

  • Statistical inference.
  • Systems Analysis and Design