Repeated-Game Modeling of Cooperation Enforcement in Wireless Ad Hoc Network

Abstract

在缺乏集中控制的无线自组网络中,节点在转发过程中所表现出的自私行为将严重影响其网络服务的可靠性.在节点理性假设的基础上,针对自组网络节点的预期收益及其协作交互过程建立了一个重复博弈模型,提出了一个激励一致性条件,在此条件下,节点将迫于惩戒机制威慑而自愿采取合作策略;并分析了节点对将来利益重视程度、机制参数和作弊检测效率对协作效果的影响.仿真结果表明,通过合理选择惩戒机制参数,能够有效抵御网络规模的增长及节点合作意愿、作弊检测效率的降低所导致的协作性削弱,进而提高存在自私节点时的整体网络性能.;Due to the absence of centralized authority, the service reliability of wireless ad hoc network is seriously affected by selfish actions of the rational nodes during the packet forwarding. This paper proposes a repeated-game model of node behavior that takes account of the selfish nodes' future payoff expectations and their long-term desires for profit. An incentive-compatible condition under which the selfish one will be deterred from cheating by the subsequent punishments and then turn to cooperate is shown analytically. The impacts on the selfish nodes' behaviors, which are induced by their willingness for future collaboration, the parameter settings of punishment mechanism and the efficiency of misbehavior detection, are also discussed. Simulation results show that, the increase of network scale, the deterioration of node's collaborative patience and the low misbehavior detection efficiency will motivate entities toward self-interested action, but this tendency can be neutralized by a careful configuration of the punishment mechanism in the model.