Signalling of information in networked stochastic dynamical control systems

Signalling of information is made feasible from one controller to another controller, in a networked control system consisting of two interconnected control systems. In this application, controller 2 of control system 2 (CS-2) has access to feedback information from its output, while controller 1 of control system 1 (CS-1) does not have access to feedback information from its output. The methodology is based on computing the Control-Coding (CC) Capacity of CS-2, and then transforming the randomized control strategy which achieves the CC Capacity of CS-2, into a controller-encoder, which simultaneously controls the output of CS-2, and encodes the output of CS-1, then decodes it and applies it to the controller of CS-1, to minimize the pay-off of CS-1. The paper demonstrates that randomized strategies of control systems simultaneously control and encode information, and signal information from one processor to another processor, through the control system, provided the rate at which information is signalled through the control system is below the CC Capacity of the control system.