The Common-Information Approach to Decentralized Stochastic Control

TLDR: This chapter presents the common-information approach to decentralized stochastic control, and describes this approach for a general model and illustrates it by examples from real-time communication, networked control systems, paging and registration in cellular systems, and multi-access broadcast systems.

Abstract: Decentralized stochastic control arises in multi-stage decision-making with multiple decision-makers having different information and a common objective. Examples include cyber-physical systems, communication networks, sensing and surveillance systems, transportation systems, etc. In this chapter, we present the common-information approach to decentralized stochastic control. The key idea behind this approach is to formulate an equivalent centralized stochastic control problem from the point of view of a fictitious coordinator that observes only the information that is commonly available to all decision-makers. The optimal control problem for the fictitious coordinator is shown to be a partially observable Markov decision process (POMDP) which can be solved using techniques from Markov decision theory. We describe this approach for a general model and illustrate it by examples from real-time communication, networked control systems, paging and registration in cellular systems, and multi-access broadcast systems.