Skip to Main Content
In this paper, we describe the multiagent supply chain simulation framework MACSIMA. This framework allows the design of large-scale supply network topologies consisting of a multitude of autonomous agents representing the companies in the supply network and acting on their behalf. MACSIMA provides all agents with negotiation and learning capabilities so that the co-evolution and adaptation of the price negotiation strategies of the business agents that exchange goods over an electronic B2B-market can be simulated and evaluated. Thereby MACSIMA supports a fine-tuning of the parameterization of the learning mechanism of each individual business agent and additionally enables the agents to exchange information about finished negotiations with other cooperating agents. We outline evaluation results with a first focus on the emergence of niche strategies within a group of cooperating agents. After that we center a second focus on the coordination efficiency, i.e. on the effects of the application of different learning mechanism parameterizations on the overall turnover and profit of supply networks. Our simulation results show that depending on the parameter setting of the learning mechanism the outcome (e.g. the overall turnover) of such a supply network varies significantly.