Bayesian Estimation with Successive Rejection and Utilization of A Priori Knowledge

A method for synthesis of estimation algorithms using the Bayesian approach with successive consideration of a priori knowledge is presented. The lack of knowledge on the form of the a priori distribution or the rejection of this knowledge due to the impossibility of deriving a Bayesian estimation algorithm can be simulated by replacing the true a priori distribution with another distribution having a higher entropy; in this case, well-known nonBayesian estimation algorithms can be obtained using the classical Bayesian rule. An approach to the synthesis of an estimation algorithm is proposed, in which a part of a priori knowledge is successively rejected, then taken into account. Examples of practical use of this approach for the synthesis of a signal demodulation algorithm in communication systems with a MIMO channel are presented.