TTP: Small: Collaborative: Defending Against Website Fingerprinting in Tor

人们使用互联网越多，他们分享不想让别人知道的信息的风险就越大。Tor是一项每天帮助数百万人保护他们在线隐私的技术。Tor用户——从普通公民到拥有宝贵知识产权的公司——在其在线信息和活动的内容，以及他们与谁互动和何时互动方面获得了保护。在大多数情况下，Tor是非常安全的。然而，它有一个被称为网站指纹的攻击的已知漏洞。攻击者，比如窃听无线连接的人，可以分析网站发送给访问者的数据中可识别的模式，从而潜在地确定用户访问了哪些网站。研究表明，在某些情况下，这种攻击的成功率高达90%。Tor的开发人员和一组研究人员将合作建立和部署针对网站指纹的新防御，这将大大降低攻击的准确性，而不会降低用户体验或增加运行系统的成本。通过广泛部署这样的防御，Tor的安全性大大提高。这也为其他研究人员提供了一个实验平台和数据，可以用于他们自己的实验，有助于进一步加强Tor。以前防御网站指纹识别的方法增加了延迟，因此网页加载速度比正常情况慢2到4倍。为了设计一种有效的防御，不会使浏览速度变慢或增加带宽开销，该项目建立在一种称为自适应填充的技术之上，该技术在初步工作中显示出有希望的结果。自适应填充的工作原理是注意到网站和Tor用户之间的数据包在网络上流动的独特空隙，并用额外的数据包填充这些空隙，以混淆网站的指纹。研究团队首先开发了一个平台来评估自适应填充和Tor中的其他防御，允许在不完全部署或危及用户隐私的情况下在实时Tor网络中进行安全实验。利用这个平台，该团队对网站指纹防御和攻击进行了广泛的实验，并最终部署。该项目侧重于可以实际部署并显著提高用户安全性的防御。由于满足这些要求的防御措施到目前为止还没有得到充分的研究，因此该项目增强了这一领域的知识。最后，该项目的研究结果完全转移到Tor，使互联网浏览更安全，更有保障。

The more people use the Internet, the more they risk sharing information they don't want other people to know. Tor is a technology that every day helps millions of people protect their privacy online. Tor users -- ranging from ordinary citizens to companies with valuable intellectual property -- gain protection for the content of their online messages and activities, as well as whom they interact with and when. For the most part, Tor is very secure. However, it has a known vulnerability to an attack called website fingerprinting. An attacker, such as someone eavesdropping on a wireless connection, can analyze recognizable patterns in data that websites send to visitors, potentially determining which websites a user visits. Studies have shown that, in some situations, this attack can be successful up to 90% of the time. Developers of Tor and a team of researchers will collaborate to build and deploy new defenses against website fingerprinting, which substantially lower attack accuracy without degrading user experience or increasing the costs of running the system. By widely deploying such a defense, Tor's security is greatly improved. This also provides an experimental platform and data that other researchers can use for their own experiments, helping to further strengthen Tor.Previous approaches to defending against website fingerprinting increased latency, so that web pages load 2 to 4 times slower than normal. To design an effective defense that doesn't unacceptably slow browsing or increase bandwidth overhead, this project builds on a technique called Adaptive Padding, which has shown promising results in preliminary work. Adaptive Padding works by noticing distinctive gaps in data packets flowing over the network between a website and Tor user and filling these gaps with extra packets to obfuscate the website's fingerprint. The research team first developed a platform for evaluating Adaptive Padding and other defenses in Tor, allowing security experiments in the live Tor network without full deployment or endangering user privacy. Leveraging this platform, the team performs extensive experiments on website fingerprinting defenses and attacks, leading towards deployment. The project focuses on defenses that can be practically deployed and significantly improve user security. Since defenses meeting these requirements have been understudied to date, this project enhances knowledge in this area. Finally, the findings of the project are fully transitioned to Tor to make Internet browsing safer and more secure.