A Pseudo Bayes Approach to Digital Detection and Likelihood Ratio Computation

摘要

The research investigates the generalized problem of detecting an arbitrary random process in the presence of additive Gaussian noise. The problem is considered in the discrete domain, and computationally feasible algorithms are derived for the likelihood ratio which optimally solves the problem. The likelihood ratio algorithm is expressed as a function of the one state prediction minimum variance estimate of the signal process, suggesting that the optimum digital detector involves implementation of an estimation algorithm prior to computation of the likelihood ratio. In order to realize algorithms which are applicable to the very general class of problems considered, certain judicious approximations are made. Justifications are given for all such approximations, and the statistical properties of the resulting algorithms are investigated. In addition, several example problems are presented to demonstrate the efficacy of the results. (Author)