Coevolutionary Multi-agent Optimization of Distributed Supply Networks

We present a distributed coevolutionary multi-agent optimization approach for real-world large-scale supply networks within global enterprises which rely on a distributed network of suppliers and manufacturers for producing complex engineered systems. In these environments, time and cost efficient material movement through a global network is based on decisions made on parts, suppliers, and manufacturing assignments. The distributed optimization mirrors the human-based multi-agent network observed in the real-world. In this distributed multi-agent system, each supplier is an autonomous network-connected decision-making agent that maximizes a utility function of relevance to the supplier agent while supporting the planning goals of the supply network. A coevolutionary algorithm is naturally suitable to such a distributed supply network, and in this paper we make a strong argument for its consideration and application deployment.