Bayesian Parameter Estimation Using Periodic Cost Functions

摘要

In this paper, a new method for Bayesian periodic parameter estimation is derived using periodic cost functions. The method, named parameter estimation via root finding (PERF), is based on Fourier series representation of the Bayes periodic-risk functions. The PERF method is implemented for minimum cyclic error, minimum absolute periodic error, and minimum mean-square-periodic-error (MSPE) estimators and the corresponding estimators are derived. The periodic estimators are applied to direction-of-arrival and phase estimation problems and compared with the minimum mean-square-error and maximum a posteriori probability estimators, and the periodic Ziv-Zakai lower bound in terms of MSPE.