CAREER: CIF: Anonymous Networking with Guaranteed Quality of Service: Towards a Theoretical Foundation

网络系统中的匿名性是指保护通信各方的身份和数据流的路径。暴露此类信息不仅是对用户隐私的侵犯，而且还会使恶意攻击者得以发动强大的攻击。虽然人们对信息安全和利用通信内容保护网络活动的方法给予了极大的关注，但对网络对对抗性时间分析的脆弱性和风险及其对隐私和社会互动的影响知之甚少。在本研究中，开发了一个严格的理论基础，为网络系统提供任何期望的匿名级别，并提供可证明的网络服务质量保证。本研究建立在一个信息论框架上来衡量匿名性，这产生了第一个匿名性的定量模型，该模型考虑了对手可用的完整信息以及网络的资源和拓扑约束。传统方法通过将对手建模为无所不知来规避网络与对手之间的相互依赖关系，而研究者将匿名网络建模为网络节点与对手之间的博弈，并研究参与者的纳什均衡行为，即没有人有偏离这种行为的动机。利用数学框架，描述了匿名性和网络性能之间的基本权衡。因此，可以深入了解网络对手的可能行为。在更广泛的背景下，这种严格的分析基础将对向一系列网络系统（包括计算机网络、医疗保健、交通和智能能源分配系统）提供可证明的隐私产生重大影响。此外，针对初高中学生的计划和远程学习机会扩大了社会对网络隐私漏洞的认识。

Anonymity in networked systems refers to protecting the identities of communicating parties and paths of data flow. Exposing such information not only is a violation of user privacy, but also empowers a malicious adversary to launch powerful attacks. While a great deal of attention has been paid to information security and methods of protecting network activity using contents of communication, far less is known about vulnerabilities and risks of a network to adversarial timing analysis and its implications on privacy and social interactions. In this research, a rigorous theoretical foundation is developed for delivering any desired level of anonymity to networked systems with provable guarantees on network quality of service.This research is built on an information theoretic framework to measure anonymity which yields the first quantitative model for anonymity that takes into account the complete information available to an adversary and the resource and topological constraints of a network. While conventional approaches circumvent the interdependence between network and adversary by modeling the adversary as omniscient, the investigators model anonymous networking as games between network nodes and the adversary and study Nash equilibrium behavior of the players, behavior from which none has the incentive to deviate. Using the mathematical framework, the fundamental trade-offs between anonymity and network performance is characterized. Consequently, insights are derived into likely behavior of network adversaries.In a broader context, this rigorous analytical foundation will have significant impact in delivering provable privacy to a range of networked systems including computer networks, healthcare, transportation and smart energy distribution systems. In addition, the programs for middle and high school students and distance learning opportunities broaden the awareness of privacy vulnerabilities of networks in society.