Comments on “Robust state estimation for uncertain discrete-time stochastic systems with missing measurements” [Automatica 47 (2011) 1520–1524]

We show that some results presented in the aforementioned article are rather limited. One of the major restrictions on applicability of the obtained results is the ergodic requirement on the received plant output measurements, which is generally not satisfied by a time-varying system. Another major restriction is that the comparisons are not appropriate for its numerical simulations, that is, using the state estimation results of an estimator that utilizing current and past observations to compare with those of a one-step state predictor is not appropriate. To evaluate the filter performances objectively, we do a numerical simulation with the robust state estimator and the well-known Kalman filter that does not take either parametric errors or random measurement loss into account; and the simulation results show that there is no significant difference in filtering performance between these two estimators.