On design of optimal robust estimation algorithms

A technique is proposed for desing of iterative robust estimation algorithms which relies on approximate methods of optimal nonlinear filtering of Markovian processes. The application is illustrated with designing an algorithm for estimating the mean value of a random quantity.