Degree of Local Cooperation and Its Implication on Global Utility

Abstract

In a cooperative multi-agent system that is situated in an evolving environment, agents need to dynamically adjust their negotiation attitudes towards different agents in order to achieve optimal system performance. In this paper, we construct a statistical model for a small cooperative multi-linked negotiation system. It presents the relationship between the environment, the level of local cooperation and the global system performance in a formal and clear way that allows us to explain system behavior and predict system performance. The analysis results in a set of design equations that can be used to develop distributed mechanisms that optimize the performance of the system dynamically. It helps us more concretely understand the important issue of distraction and provides us with the local attitude parameter to handle distraction effectively. This research demonstrates that sophisticated probabilistic modelling can be used to understand the behaviors of a system with complex agent interactions, and provide guidelines for the development of effective distributed control mechanisms.