Collaborative agent-based negotiation in supply chain planning using multi-behaviour agents
Pascal Forget, Thibaud Monteiro, D'Amours, Sophie, Frayret, Jean-Marc · PolyPublie (École Polytechnique de Montréal) · 2008
In order to work efficiently, supply chain partners must coordinate their actions. When planning is distributed instead of being centralized (as in most cases), it is necessary to use specific coordination protocols between partners in order to act in a coherent manner. On many occasions, partners negotiate in order to act mutually to satisfy their needs. Negotiation is used as a coordination mechanism to find an acceptable agreement between partners or to collectively search for a coordination solution. Agent- based supply chain planning systems can integrate automated negotiation in order to implement negotiation capabilities. While various negotiation mechanisms (or behaviours) can be used in many situations, planning agents using case-based reasoning abilities can learn which to select under specific conditions. This paper proposes to study the performance of the use of a variety of negotiation behaviours and to compare this strategy to the use of a single one. A review of automated negotiation in general and specifically adapted to supply chain planning is first presented, followed by an empirical analysis from simulations of one-to-one collaborative negotiation behaviours. Partners have equivalent negotiation powers and are fully cooperative. These heuristic-based negotiation behaviours are implemented in an agent-based supply chain planning platform, using Multi-behaviour planning agents. Simulations are based on a study case from the lumber supply chain.