The minimax posterior Wonham filtering/identification

Abstract The paper investigates a filtering/identification problem of finite-state Markov processes given continuous and/or counting observations. All the transition intensity matrix, observation plan and counting intensity are parameterized by a random vector with uncertain distribution belonging to a known class. An assertion concerning existence of saddle point in the considered minimax optimization problem as well as a form of corresponding estimate is presented. Monitoring of TCP link status under uncertainty is proposed as an illustrative numerical example of application of obtained theoretical results.