Distributed simultaneous estimation of states and unknown inputs

Abstract

In this paper, the problem of distributed estimation in linear discrete-time large-scale stochastic systems is explored. The objective is to estimate the states as well as the unknown inputs of the system under partial observation of the system states by utilizing a number of estimators. The main assumption is that the estimator at each node of the system can exchange its estimated state with the neighboring estimators through communication links. With this consideration, a recursive distributed filter is introduced and its unbiasedness and minimum variance are examined. Furthermore, the necessary and sufficient conditions for the stability and convergence of the proposed distributed filter are investigated. Finally the performance on a numerical example is presented.