There is a growing concern about privacy in centralized social networks, which led to the advent of decentralized alternatives. A effective way to provide privacy in a decentralized social network is to leverage the pre-existing social trust relations and mapping them into communication links. However, such communication overlays are very fragile if there is churn in the system. In [4] we have proposed a protocol that improves robustness of trust-based overlays. The solution relies on pseudonym systems (such as TOR rendezvous nodes) which allow two non-trusting members to form a communication link between them while preserving their privacy requirements. The work did not consider any specific pseudonym system, it rather abstracted the properties of the pseudonym system to describe the overall solution.
This master project aims to analyze different pseudonym systems from the privacy and reliability point of view. It requires the student to implement the protocols proposed in [4] using different pseudonym systems. Student also has to evaluate the reliability of the implementation through experiments. An optional extension for the interested student would be to analyze the level of privacy provided by a pseudonym system using the information metric proposed in [2].
The project should consider both DHT-based and server-based pseudonym systems. In DHT-based systems, a DHT-key plays the role of a pseudonym identifier. In contrast, in server-based systems (such as TOR or I2P), the IP address and a unique identifier at the server play that role. An overview of information leakages in DHT-based systems can be found in [3], while [2] discusses the challenges in server-based systems.
References:
[1] M. Rogers and S. Bhatti. How to Disappear Completely: A Survey of Private Peer-to-Peer Networks.
[2] A. Serjantov and G. Danezis. Towards an Information Theoretic Metric for Anonymity. Privacy Enhancing Technologies 2003.
[3] P. Mittal and N. Borisov. Information Leaks in Structure Peer-to-Peer Anonymous Communication Systems. ACM Transactions on Information and Systems Security. 15, 1, Article 5 (March 2012).
http://doi.acm.org/10.1145/2133375.2133380
[4] A. Singh, G. Urdaneta, M. van Steen and R. Vitenberg. Robust overlays for privacy preserving data dissemination over a social graph. ICDCS 2012.
http://dx.doi.org/10.1109/icdcs.2012.57