联盟合作是自组织网络中一个热点研究领域,Agent间的信息具有不确定性、不完全性和局部性等特征,个体Agent节点出于风险的考虑会在不同的联盟间迁移,因此在联盟的演化研究中需要融入信任因素.文中研究群体随机运动环境下可信联盟的演化机制,从微观和宏观上对个体Agent运动行为和联盟规模结构进行分析,将基于历史交互的信任信息融入演化过程中,提出演化规则,界定了演化稳定性的标准.从微分动力系统的视角建立约束演化方程并通过概率计算对方程进行转化求解,分析了演化计算的复杂性,并进一步分析了Agent异构性和局部信息感知对联盟演化的影响.最后通过模拟实验对演化进行讨论,并从信息论角度对联盟的有序性进行分析,讨论信任对演化过程的影响,刻画了群体随机活动中可信联盟的构建与演化过程.%Coalition cooperation is a hot field in self-organizing network. Information among agents is uncertain, incomplete, and localized. Agent continuously moves among coalitions for risk considerations. So trust factor should be integrated into the evolution of coalition. The paper depicts evolution mechanism of truthful coalition from random motion. Firstly the moving behaviors of agent and structure of size of coalition are analyzed by microcosmic and macroscopic view, and then the trust information from historical interconnection is blended in evolution. The evolution rules and standard of evolution stability are given. The restricted evolution equation is got by the view of power system, and is solved by reducing probability computation. The paper also analyzes the complexity of evolution, the further impact to evolution made by heterogeneous type and perceptual locality information are discussed too. Finally a simulation experiment is given and the orderliness of coalition is analyzed. The order of coalition is analyzed by information theory, the impact of trust on evolution process is discussed. The construction and evolution process of credible coalition in crowd random motion are also depicted.
展开▼
