网格环境下层次信任模型研究

Research on hierarchical trust model in grid environment

信任管理作为网格研究的核心内容,受到研究人员的高度关注。目前设计的模型中,实体域的信任值由直接信任值和推荐信任值合成,但引入推荐信任就必须对网格域进行层次划分。现有的划分方法只是主观地根据网格域之间的交易情况来划分层次,这种划分方式得到的推荐值的可靠性难以保证,并且不能防止联合欺骗的行为。基于对网格实体域行为的深入研究,给出了一个基于实体域行为的直接信任值动态量化模型,在此基础上,引入匹配指标和粗糙度的概念,给出了一个层次划分模型,利用该模型很好地解决了推荐信任值的计算问题,给出了一个信任评估模型,理论分析和实验结果表明,该模型能够保证推荐信任值的可靠性,并且防止了推荐过程中的联合欺骗,有效地解决了网格环境中信任评估问题。

Trust management is the central part of grid computing system and the model is a prime concern.In the present models,trust values are based on the combination of direct trust and recommend trust.For obtaining the recommend trust,grid domains must be classified.The current classification methods are subjectively based on the transactions in the grid domains.This can＇t ensure the reliability of recommends trust and it＇s hard to resist the effect of malicious recommendation.By the research of grid entitative behavior,dynamic quantification model for direct trust based on entitative behavior trust is presented.Match Index and roughness as two new concepts are introduced,and a classification model is given,which solve the problem of how to calculate the recommend trust value.The results of simulation experiments show that the model ensures the reliability of the recommend trust and prevent the malicious recommendation ,effectively evaluate the trust of entities.