项目名称: 分布式环境下信任管理的可视分析关键技术研究
项目编号: No.61202401
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 计算机科学学科
项目作者: 彭帝超
作者单位: 浙江大学
项目金额: 22万元
中文摘要: 以物联网、云计算、传感器网络等为代表的分布式计算环境对信任管理提出了更高的挑战。现有的信任管理模型大都建立在机器智能之上,利用计算机高速批量处理数据的能力对信任关系实施自动管理。然而仅仅依赖机器的智能常常难以完美的解决信任管理问题,特别是当攻击者针对信任模型的设计弱点采取复合攻击策略时,我们往往很难直观的了解系统面临着什么威胁,从而开展有针对性的防御措施。 本项目独辟蹊径,提出使用可视分析的方法来研究信任管理系统:通过使用节点链接图、信任关系矩阵等可视化方法,让攻击模式与可视化结果的视觉特征一一对应起来。进一步的,我们运用人机交互技术,让用户的智慧参与到信任管理过程中来:通过对信任关系网络的可视化呈现和交互,让用户对机器的分析和判断进行反馈,对机器分析的参数和策略进行干预和调整。经过这一过程的反复迭代求精,探索和发现攻击规律,对系统面临的攻击模式作出正确的判断。
中文关键词: 信任关系;信任管理;可视分析;;
英文摘要: In the age of modern distributed computing, the introduction of Internet of Things, Cloud Computing and Sensor Networks poses new challenges to the task of trust management. Traditional trust management frameworks are largely relying on machine intelligence: utilizing computers' exceptional data processing ability to achieve automatic trust management. However, computers alone can sometimes generate disappointing results, especially when malicious users adopt complex attack schemes to target the weakness of trust models. In such cases, it can be difficult for us to intuitively understand the threat that we are facing, not to mention initiating meaningful countermeasures. In this project we propose a novel approach to bridge the gap between computer and human intelligence by the means of visual analysis. With visualization schemes such as edge bundling in node-link diagram and spectral clustering of trust matrix, our approach can map specific attack models onto visual patterns. In addition, we propose to incorporate the users' wisdom into trust management process by the means of human-computer interactions: with the vivid visualization and convenient interaction, the users are granted with the power to provide professional feedbacks to the automated trust management and adjust the analysis parameters & scheme
英文关键词: Trust relations;Trust management;Visual analysis;;