Discrete-Event Coordination Design for Distributed Agents

This paper presents new results on the formal design of distributed coordinating agents in a discrete-event framework. In this framework, agents are modeled to be individually equipped with a coordination module, through which they interact and communicate. In terms of existing control-theoretic concepts, we first define the concept of a coordinable language and show that it is the necessary and sufficient existence condition of coordination modules for distributed agents to achieve conformance to a prespecified interagent constraint language. Following, we present a synthesis algorithm to compute near-optimal coordination modules. An example is provided to illustrate the design synthesis using the proposed algorithm. Finally, a discussion with related work distinguishes our coordination design problem from related problems in the literature. Note to Practitioners-Multiagent coordination presents a key approach to developing complex systems. In this approach, the basic idea is to model a system as a network of interacting agents, and design for each agent a coordination module by which the agents can interact and communicate to manage the interdependencies arising due to system needs or limitations. In this paper, we propose a novel approach for coordination design of distributed agents. By modeling coordinating agents as discrete-event processes, we formulate and address the problem of synthesizing for each agent a coordination module to achieve conformance to a given interagent constraint. Importantly, by showing that our multiagent coordination problem shares the same algorithmic foundation with existing problems in the literature, we are able to adapt existing techniques to develop a new coordination synthesis algorithm, without reinventing the wheel. The coordination modules synthesized by our algorithm are proven to be minimally interventive with the agents' local plans and ensure that communication among the agents is made only when necessary. Potential applications of our work can be found in domains where discrete-event modeling has proven to be suitable and effective, and these include manufacturing, transportation, and logistics systems.