基于信任合成的云服务动态组合机制研究

摘要

With flexible SLA (Service Level Agreement) as the core business characteristic, cloud computing provides users with different value-added services through dynamic integration of various types of cloud services. However, in the traditional approach to service composition, with QoS (Quality of Service) as the core constraint, trust degree of the service can not be well regulated. This paper proposes a new approach to autonomous and flexible composition and management of could services with trust integration as its core feature. In this approach, the trust property of a cloud service consists in the combination of basic trust and experience trust. The basic trust is defined as the basic elements of the object with the Bayes reasoning analysis, while the experience trust defined as the behavior history between objects with its confidence given by Chebyshe and Bernstein inequality equation. A series of experiments on a prototype are conducted to verify the accuracy and validity of the mechanism proposed. The results show that the proposed approach is effective in composing and selecting services on the cloud.%云计算为代表的新型计算模式以灵活的"服务合约"为核心商业特征,通过动态整合各类云服务为用户提供不同粒度的增值服务.但是传统以QoS(Quality of Service,服务质量)为核心约束的服务组合方法,无法满足用户对服务服务质量的深层要求,即对服务信任程度的规范.为此本文提出了一种基于信任合成的云服务动态组合方法,该方法通过定义云服务的信任属性,将其分解为基础信任和经验信任的集合.将基础信任评价问题建模为云服务属性判断问题,利用Bayes合情推理分析了属性不可穷举情况下基础信任的审定与反驳.将经验信任评价问题建模为云服务交互行为判断问题,利用Chebyshe和Bernstein定理给出经验信任的置信度,进而为经验信任的量化提供依据.实验结果表明,本文所提出的方法可以在持续变化的云环境下有效地组织和提供云服务,进而满足新型计算模式动态多样化应用需求.