新型互联网计算模式下的信任关系建模及其机理研究

Trust model that is usually regarded as a relatively flexible soft security measurement mechanism, is a hotspot of academic research in the new Internet computing. This project intends to start with the multi-dimensional trust evidence, builds a multi-dimensional trust evidence model and trust evidence collection mechanism to provide an effective evidence supporting technology for trust model. The formalizing representation and evaluation methods of trust relationship are established based on entity context and time stamp. The forming and propagation mechanisms of trust relationship are analyzed through the recommended network. By exploring the impact of interaction factors on the accuracy of trust evaluation, for example environment, time, context, the number of interactions, entity stability, and reliable recommended et al, a dynamic adaptive trust evaluation strategy is build based on interaction-aware technology to improve accuracy of trust evaluation. The decision-making mechanism of trusted entity is established based on trust reasoning and load balancing to provide strategy foundation for collaborative entity selection and malicious entities protective. An effective recognition and protection mechanisms of cheating behavior is proposed by analyzing the features of complex co-cheating behavior, which provide support technology for security and robustness of the trust model. The research results of project will help to reveal laws of the forming and propagation of dynamic trust relationship in Internet computing environment, and provide theoretical and technical support for the construction of the trust model.

在新型互联网计算模式下，信任模型作为一种相对柔性的"软安全"度量机制，成为了学术界研究的热点。本项目拟从多维信任证据入手，构建多维度信任证据模型及其证据收集机制，为模型获取评估证据提供有效支撑技术；建立一种基于上下文和时间戳的信任关系形式化表示和评估方法，基于推荐网络探索信任关系形成和传播机制；探索环境、时间、上下文、交互次数、实体稳定度、可靠推荐实体数等交互因子对信任评估准确性的影响，建立基于交互感知的动态自适应的信任评估策略，提高模型评估信任的准确性；构建一种基于信任推理和负载均衡的可信实体决策机制，为协作实体选择和恶意实体防护提供策略基础；分析复杂隐蔽协同作弊行为的特征，构建一种有效的作弊行为识别和防护机制，为模型的安全性和鲁棒性提供支撑技术。研究成果有助于揭示互联网计算环境下实体动态信任关系形成和传播的规律，为信任模型的构建提供理论和技术依托。

传统的授权机制和安全措施已无法完全满足动态开放网络环境下实体安全防护和跨域协作的需求，而动态信任机制为这类应用的安全防护和实体跨域协作提供了有效的保障，是当前新一代互联网技术研究的热点。.项目围绕适用于新型互联网计算模式的动态信任关系评估和演化的理论方法及具体技术展开研究。提出了评估实体信任关系的多维度信任证据模型，并针对不同类型的证据分别给出了对应的证据收集方法。建立了具有细粒度的信任等级模型以及一套完整的信任等级评估策略，为以后信任关系建模提供了理论依据和评价标准。建立了基于交互感知的动态自适应的信任评估机制，该方法能够动态感知实体在交互过程中所收集的各维度证据的变化情况，依据变化后的证据动态自适应的调整评估策略，从而更加有效的提高了信任关系评估的准确性。建立了一种基于层状硬币模型的信任证据收集机制，在信任证据收集过程中通过数字签名和加密机制对转发证据进行层层验证，有效防止了恶意实体和共谋实体的攻击行为。建立了一种基于多属性变化推理的 Web 服务QoS协同预测选择方法，实现了网络性能的稳定和负载均衡。利用所得到的理论、方法和技术成果设计开发了情境感知的Web服务协同推荐平台的原型系统IWebService，进行了成果扩展与应用，并取得了一系列进展。基金资助发表了文章14篇，其中被SCI收录2篇，EI收录7篇。.研究成果揭示了互联网计算环境下实体动态信任关系形成和传播的规律，为信任模型构建提供理论和技术依托。