Trial lecture - time and place
29th of January, 10:15 AM,
“Gossip-based algorithms for dissemination and structure in large scale systems”
Conferral summary
In recent years, there have been many instances of leakage of users’ private information from popular messaging platforms, such as Twitter. There is an urgent need to design better platforms that prevent such information leakages. This thesis describes the design of a robust platform that allows a large number of users to exchange messages quickly in many user groups without compromising users’ privacy.
Main research findings
The goal of my thesis is to design a privacy-preserving platform that allows users to exchange Twitter-style micro-news messages in many groups. I use a decentralized design for the platform, as a centralized design has a single point of failure. This decentralized design is based on combining P2P protocols with an existing low-anonymity system.
The novelty of my approach is the way these systems are combined. I ensure that this combination does not compromise users’ privacy, scalability of the platform, high availability of messages, or low-latency message dissemination. We design protocols to ensure that a member of a group can disseminate a micro-news message quickly to all members of the group even when members’ availability is low. The protocols are designed such that two users have similar resource consumption even if they participate in different number of groups. This prevents an adversary monitoring users’ resource consumption from learning the number of groups in which a user is participating.
We demonstrate the practicality of our design by implementing the proposed platform using Tor, a popular anonymity system. To prevent information leakages due to Tor’s weak anonymity guarantees, we design algorithms for intelligent selection of resources in Tor’s infrastructure. A user employs these algorithms to ensure that her communication remains hidden from non-trusted entities even when she uses Tor for the communication.
Contact information to Department: Mozhdeh Sheibani Harat