This work addresses private communication with distributed systems in mind. We consider how to best use secret key resources and communication to transmit signals across a system so that an eavesdropper is least capable to act on the signals. One of the key assumptions is that the private signals are publicly available with a delay---in this case a delay of one. We find that even if the source signal (information source) is memoryless, the design and performance of the optimal system has a strong dependence on which signals are assumed to be available to the eavesdropper with delay. Specifically, we consider a distributed system with two components where information is known to only one component and communication resources are limited. Instead of measuring secrecy by "equivocation," we define a value function for the system, based on the actions of the system and the adversary, and characterize the optimal performance of the system, as measured by the average value obtained against the worst adversary. The resulting optimal rate-payoff region is expressed with information theoretic inequalities, and the optimal communication methods are not standard source coding techniques but instead are methods that stem from synthesizing a memoryless channel.

Allerton 2010, 6 pages, 1 figure, uses IEEEtran.cls