Project B5

Anonymous and Censorship-Resistant

Principal Investigators

Project Summary

Anonymous communication is essential to protect the privacy of users online. The academic community has developed several anonymous communication networks over the years, which can be classified in semi-decentralized (like Tor, which uses a central directory) and fully decentralized (like I2P) networks. Today, some of them have reached an unprecedented degree of maturity and popularity: Tor is used daily by over four million users worldwide, I2P by thirty thousand, and their user base is constantly increasing.
Despite this success, anonymous communication networks are affected by a fundamental problem, which undermines the functionality they offer: ISP providers, and thus governmental agencies, can determine which traffic enters an anonymous communication network, which enables them to censor undesired communication, as reported in China or Syria. The same techniques have also been used to identify the users in such network, with the goal of monitoring their online behavior (the second German, after the prime minister, to be reported subject of monitoring by NSA is an Erlangen student running a Tor entry node), and they can also be used more generally to perform traffic analysis and effectively deanonymize users. For all these reasons, developing censorship-resistant communication techniques is of paramount importance not only for the freedom of speech but also, more generally, for the privacy of users in the digital society. This project aimed at devising a broad range of techniques to make anonymous communication networks censorship-resistant. First, we explored whether censorship events can be reliably observed using public observations, which allows us to assess the impact of the censoring problem and to react to censorship events. Second, we explored censorship-resistant communication methodologies that solve the challenge of reliably bootstrapping to the network, in particular decoy routing systems. Third, we investigated how the general technique of credit networks can be used to model trust in a censorship setting. Finally, we investigated if emerging new name resolutions protocols such as DNS-over-HTTP and DNS-over-TLS solve censorship threats by hiding domain names from passive adversaries.

Role Within the Collaborative Research Center