Individually optimal solutions to a remote state estimation problem with communication costs

In this paper, we discuss a remote state estimation problem that consists of a linear plant driven by process noise, a sensing unit that assesses the state of the plant, and a remotely located estimator. At each time step, the sensing unit decides whether to transmit the state of the plant to the estimator. The purpose of this work is to design a sensing unit and an estimator that minimize a cost which combines the expectation of the squared estimation error and a communication cost. We establish the existence of a jointly optimal solution to the problem, and we propose an iterative method, which generates a sequence that approaches the set of individually optimal solutions¹.