Today’s global economy is characterized by changing market demands, short product lifecycles, and fine-grained customization. This context demands a capability to identify promising supply chain partners—and adapt existing relationships—in response to dynamic conditions. One major impediment is the multiplicity of considerations that bear on supply chain decisions. For example, a material supply contract might specify price, quantity, delivery time and mode, quality, warranty, and many other features. Most of today’s auction systems support one-dimensional negotiations—where price and quantity are the only features determined dynamically.

This project aims to support negotiation over many attributes in dynamic, multilevel supply chains. To this end, it is developing flexible languages for expressing multiattribute offers, general two-sided matching mechanisms supporting both iterative and one-shot negotiation, and real-time decision support based on summarized assessments of production states. The new techniques will be validated using an agent-based supply chain simulation testbed, as well as analytical models.

Currently, our primary scenario is the Supply Chain Management Trading Agent Competition. TAC/SCM presents a challenging dynamic market game, where agents playing the role of PC assemblers compete for customer orders and component supplies, while managing their manufacturing capacity to maximize overall profit. The TAC SCM game was designed as part of this project, with assistance from the Swedish Institute of Computer Science, which implemented the game environment and operates the competition.

The resulting methods will enable a future where companies routinely form and adapt supply chain relationships for a broad range of productive activities. The systemic ability to respond dynamically to changing conditions can ultimately lower costs, while delivering greater value through increased customization and responsiveness to novel opportunities.