CRII: SaTC: New Side-Channel Techniques in Support of Global Monitoring for Network Disruptions

互联网的规模和复杂性使得人们很难理解网络数据包在从源到目的地的传输过程中发生了什么。互联网服务提供商和政府有时会干预用户的在线活动，这使这项任务进一步复杂化。该项目通过开发远程测量互联网流量操纵的新方法，如试图审查、篡改和监控用户的在线活动，促进了对网络干扰检测的科学理解。新提出的技术可以在全球范围内实现前所未有的网络操作细节。网络干扰和审查损害了美国的经济利益、国家安全和受审查国家公民的人权。通过对用于检测和监控此类干扰的技术进行更好的科学理解，该项目可以保护互联网用户免受无理的网络操纵。过去的研究使用了行星实验室、虚拟专用网（vpn）和终端用户机器等资源来测量网络中断。这些有利位置的规模很差，地理覆盖范围有限。它们难以在实践中部署，也无法持续收集数据。相比之下，Spooky、Augur和Iris等系统（在主要研究者的初步工作中开发）允许测量系统推断远程系统是否与另一个远程系统具有网络连接，而无需控制其中任何一个。为此，这些系统利用现有Internet协议和基础设施中的特性与远程系统进行交互，并从它们的响应中了解它们是否能够与其他目标主机进行交互。这些技术克服了有利位置和招募参与者的问题。

The Internet's size and complexity make it difficult to understand what happens to network packets as they travel from source to destination. Internet service providers and governments sometimes interfere with users' online activities, further complicating this task. This project advances the scientific understanding of network interference detection by developing new methods to remotely measure the manipulation of Internet traffic, such as attempts to censor, tamper with, and monitor users' online activities. The newly proposed techniques can achieve an unprecedented level of detail about network manipulation around the world. Network interference and censorship harms U.S. economic interests, national security, and the human rights of citizens in censored countries. By working towards a better scientific understanding of the techniques used for detecting and monitoring such interference, this project can defend Internet users from unwarranted network manipulation. Past research has used resources such as PlanetLab, Virtual Private Networks (VPNs), and end-user machines to measure network disruptions. These vantage points scale poorly and have limited geographical coverage. They are difficult to deploy in practice and are unable to continuously collect data. In contrast, systems such as Spooky, Augur, and Iris (developed in the principal investigator's preliminary work) allow measurement systems to infer whether a remote system has network connectivity to another remote system without needing control of either. To do this, these systems exploit features in existing Internet protocols and infrastructure to interact with remote systems, and learn from their responses whether or not they were able to interact with other targeted hosts. These techniques overcome the problems of vantage point location and of recruiting participants.