Optimal Information Control in Cyber-Physical Systems

In this paper, we address optimal information control in cyber-physical systems. In particular, we study optimal closed-loop policies for transmission of measurements of a stochastic dynamical system through a communication channel given estimation and communication costs. We develop a framework for optimizing an aggregate cost function that incorporates the estimation and the communication costs over a finite time horizon. We obtain the optimal closed-loop policy, and show that it can be expressed directly in terms of the value of information. In addition, we propose an approximation algorithm that yields a suboptimal closed-loop policy. Numerical and simulation results are presented for a simple system.