Multichannel Detection Using the Discrete-Time Model-Based Innovations Approach
Abstract
This report makes several contributions. First, an approach is developed to synthesize multichannel autoregressive (AR) random processes allowing for the control of temporal and cross-channel correlation of the processes subject to specific constraints for realizable correlation sequences. Second, analytic expressions are developed for the error variance of time- averaged correlation function estimators for discrete, complex baseband processes. These expressions reveal the functional dependence of the error variance, not only on the window size of the observation interval, but also on fundamental characteristics of the observed processes. Third, model-based likelihood ratios are developed for the multichannel binary detection problem using error vector processes obtained through predicational error filtering operations. These ratios are derived assuming wide-sense stationarity of the baseband Gaussian processes. However, a more general likelihood ratio applicable for nonstationary bandpass processes is also developed. Fourth, the more general likelihood ratio utilized a more powerful estimator than previously noted.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1991
- Accession Number
- ADA241331
Entities
People
- James H. Michels
Organizations
- Rome Laboratory