Collaborative single and parallel machine scheduling by autonomous agents

Collaborative interorganizational systems is a relevant and promising research domain. When autonomous agents associated with different enterprises negotiate about rival resources, coordination is a major challenge. Information provided by the agents is a fruit of a poisonous tree because the selfish agents have an incentive to lie strategically which makes the revealed information unreliable and misleading for a central authority. Moreover, the agents might not be willing to reveal information due to privacy concerns. Thus, there is a need for sophisticated mechanisms in collaborative systems that facilitate the coordination of non-cooperative agents. This paper deals with rival machine scheduling of one or more homogeneous machines by autonomous agents. Overall objectives are the minimization of the collective total weighted tardiness and Pareto efficient outcomes. We have developed a mediated negotiation protocol drawing on quotas for the acceptance of contract proposals. The protocol ensures that the agents accept sufficient proposals such that they can overcome individual local optima and achieve Pareto superior outcomes. For evaluation purposes, we have conducted several computational experiments. The experiments show that the proposed protocol achieves beneficial outcomes.