A multi-agents contractual approach to incentive provision in non-cooperative networks

Recent years have witnessed the increasing importance of exchanging information over computer networks or distributed systems. Two end nodes wishing to communicate often rely on independent intermediate nodes to relay messages. In consideration of the rational nature of both the end nodes and intermediate nodes, we have to accommodate two inherently coexistent games: one played between the end nodes and the intermediate nodes and the other played among the intermediate nodes. This is particularly challenging due to the well-known hidden information and the hidden action issues. In this paper we propose a holistic approach PMAC to address the two games, exploiting the principal and multi-agents model creatively. In PMAC, the end nodes make contracts with each intermediate node. The contracts together produce good system properties which are twofold. First, it is guaranteed that the utility of the end nodes is maximized. Second, it is proved that the cooperation of the intermediate nodes can be induced since there exists a Nash equilibrium for the intermediate nodes. However, one serious issue that there may be other Pareto superior Nash equilibriums inevitably hinders the unique implementation of the contracts. We also adopt technique without incurring any additional cost to the end nodes. By knocking out the other redundant Nash equilibriums in the intermediate nodes' game, we ensure that the equilibrium most desired by the end pair is successfully achieved.