信任关系存储和信任发现机制分析-analysis of trust relationship storage and trust discovery mechanism.docxVIP

摘要信息技术及其产业的高速发展推动了人类社会和经济的发展，也导致了很大安全风 险，信息安全问题备受关注。可信计算技术通过硬件结构和底层软件为基础的综合措施 来提高信息系统的安全性，在计算结点之间建立信任以构建有序的分布计算环境。信任管理是可信计算的关键性技术之一，对于解决分布式系统中的信任问题具有重 要意义。现有信任管理模型不能很好的适应网络安全环境的多变性和不确定性，无法满 足安全环境的动态变化，安全策略验证的能力和效率较低。针对现有信任管理模型存在的问题，基于信任领域概念和 Agent 技术，利用 Manager-Agent、Monitor-Agent、Entity-Agent 三层代理构建层次化体系结构，对结点 实体进行可信度量，同时管理结点间的信任关系，允许结点自主协同工作，降低了管理 开销，提高了管理效率。基于 MMA 模型，建立了基于有序二叉树的信任关系存储机制。使用有序二叉树来存 储结点实体的信任度等属性，并根据需要动态调整结点属性值和有序二叉树的形态，从 而实现了信任关系的动态管理。信任发现是信任管理的重要内容之一。针对现有信任发现机制存在的问题，建立了 基于半环理论的信任发现机制。把代数学中半环思想引入信任领域建立可信半环，进而 通过在可信图中寻找最短路径的方法来发现两个陌生结点之间的信任，并更新信任关系 及其存储。本文的研究在信任关系存储和信任发现机制方面具有一定的理论价值和实践意义。 关键词 可信计算 信任关系存储 信任发现 有序二叉树 半环AbstractThe rapid evolvement of information technology and information industry not only boosts the development of society and economy, but also results in great security risks. Therefore, information security is receiving much concern now. The comprehensive measures based on hardware structure and bottom software can enhance information security by building trust among computing nodes to establish an orderly distributed computing environment.Trust management, one of the key technology in trusted computing, is of great significance to the settlement of the trust problems in distributed systems. The existing trust management model cannot adapt to the variability, the dynamic changes and uncertainty of network security due to its inefficiency in verifying the security strategies.To solve the problems in trust management model, layered architecture is constructed with the help of Manager-Agent、Monitor-Agent、Entity-Agent, based on trust area conception and Agent technology. This structure can reduce the management expenses and increase themanagement efficiency by conducting trust measurement of entity nodes，managing the trustrelationship between management nodes and allowing the autonomous collaboration of nodes.The storage mechanism of trust relationship which is set up on the basis of MMA model employs sequential binary tree to store the