Bayesian estimate of the doppler frequency shift with reduction of the data dimensionality

Abstract

The paper considers the possibility of reducing the amount of calculations of the Bayesian estimate of the Doppler frequency shift and the initial phase of a quasi-deterministic signal observed in the presence of white-noise distortions due to introduction of the stage of reduction of the dimensionality of the observed data. The minimization of the loss of the Fisher information on the estimated parameter is used to control the correctness of the dimensionality reduction procedure; the estimation accuracy depends on this procedure. The computer simulation using test signals proves the possibility of a considerable reduction in the amount of calculations upon realization of Bayesian procedures for the same estimation accuracy.