Optimal timing of observations for state estimation in a one-dimensional linear continuous system (Q2265983)

This paper considers the optimal timing of observations for the state estimation in a one-dimensional linear continuous stochastic system which is corrupted by white Gaussian noise. The criterion adopted is a min-max one. The optimal observation times are analytically shown to be located periodically including the terminal time, irrespective of the values of the variances of both the observation noise and the system noise. Finally, the relation between the min-max solution and MSE (mean-square error) solution is considered, and the method of obtaining a suboptimal MSE solution with a little computational cost is presented.