Game Engineering a Multiagent Systems Perspective

Abstract : A central goal for the field of distributed control is to develop an underlying theory for the design and control of multiagent systems. Achieving this goal is fundamentally challenging stemming from the underlying complexity associated with a potentially large number of interacting agents and the analytical difficulties of dealing with overlapping and partial information. Game theory is beginning to emerge as a promising new direction for achieving this goal. The reason for this emergence is the similarity between the prevalent decision making architecture in social systems and the desired decision making architecture in distributed engineering systems. Accordingly, many existing game theoretic tools for analyzing behavior in social systems are immediately accessible as design tools for prescribing desirable behavior in distributed engineering systems. The applicability of game theoretic tools for this new prescriptive agenda has led to significant breakthroughs in the analysis and design of multiagent systems; however, the broader research effort connecting game theory to distributed control has ultimately fallen short of providing a systematic design methodology for multiagent coordination. The proposed research effort will address these deficiencies through the development of game engineering methodologies using an improved game theoretic framework that is better suited to handle the challenges inherent to multiagent coordination.