摘要
目前MAS信任模型研究关注的焦点是信任管理的微观机制设计,而从这些微观机制到系统的宏观特性,并不是显而易见的.因此,这些研究都采用了仿真分析.但是这些仿真实验的设计大多缺乏理论指导,往往不能正确反映系统的宏观特性.引入系统科学中的涌现视角来审视MAS,探讨基于涌现视角的MAS信任模型仿真分析方法.这一方法从相对清晰的微观层面入手,建立仿真运行模型;在充分考虑的相关宏观约束下,研究系统宏观层面的问题以及微观和宏观之间的联系.总结分析了信任模型微观机制的设计图式,在此基础上探讨了仿真Agent的设计框架;讨论了信任模型仿真分析中应考虑的宏观问题,包括宏观约束、威胁模型、评价指标以及仿真执行中的问题,最后实例验证了此方法的有效性.
The research attention of MAS （multi-agent system） trust model is now focusing on the devising of trust mechanisms. But the macro properties implied by these micro mechanisms are not obvious. So, all of the researches apply modeling and simulation for analyzing the macro properties generated by trust mechanisms. However, lots of these simulation experiments lack proper guidance by simulation theory and sometimes can not exactly reflect the properties. This paper introduces the notion of emergence in system science to investigate MAS, and develops a simulation approach based on this notion for analyzing the MAS trust model. Starting with the relatively clear behavior of MAS micro level （i.e. agent＇s behavior）, considering sufficient MAS macro constraints, and building up a runnable simulation system, this approach studies the macro problems and the micro-macro relations of the MAS trust model. Accordingly in this paper, at the micro level, the design scheme of trust mechanisms is analyzed and summarized, and based on this, a framework for implementing simulation agent is proposed. At the macro level, several system issues that must be taken into account in simulation-based MAS analysis are discussed, including macro constraints, threat model, evaluating metric, and issues in simulation running. Finally, this method is validated via several simulation experiments.
出处
《计算机研究与发展》
EI
CSCD
北大核心
2010年第12期2090-2099,共10页
Journal of Computer Research and Development
