CIF: Medium: Anonymous Broadcasting over Networks: Fundamental Limits and Algorithms

University Of Illinois At Urbana-Champaign, Urbana IL

Investigators

Abstract

In a free society, people have the right to live without being unduly monitored or surveilled. When that right is violated, anonymity-preserving technologies can provide a safety net. Anonymity-preserving systems are specifically designed to prevent adversaries from linking users to their actions within the system; most online services today are not designed to protect anonymity. This project studies the anonymity properties of networked systems in which users broadcast sensitive information over a network. It considers two representative applications: anonymous microblogging and cryptocurrencies. Anonymous microblogging refers to social networks where users broadcast messages over a social graph; examples include Yik Yak and Whisper. Cryptocurrencies are distributed, digital currencies that allow a community to collectively, cryptographically verify the legitimacy of transactions; Bitcoin is the most successful example. Both anonymous microblogging platforms and cryptocurrencies broadcast sensitive messages over a network and are generally perceived as systems that protect user anonymity. This project aims to understand, from first principles, the vulnerabilities of existing broadcasting protocols against statistical deanonymization attacks, and whether such attacks can be prevented by carefully designing new protocols. Additionally, the project will develop open-source implementations to demonstrate the key theory including a distributed anonymous messaging application and a lightweight upgrade to Bitcoin's networking stack. The project also includes training of graduate and undergraduate students in this important research area. The project is conducted in three steps. The first step is model-building which extracts simple, canonical mathematical models that capture the essence of these complex systems. The second step is source detection, where an adversary performing statistical inference to deanonymize the source of a broadcast is considered, and the mathematical models from step 1 are used to analyze the vulnerability of current protocols. The final step focuses on source hiding where the aim is to design novel broadcasting and network-management protocols that give provably-optimal anonymity guarantees while maintaining high utility for the target applications.