More is Merrier: Relax the Non-Collusion Assumption in Multi-Server PIR
Tiantian Gong, Ryan Henry, Alexandros Psomas, Aniket Kate
IEEE Symposium on Security and Privacy 2024 · Day 3 · Continental Ballroom 6
This talk, "More is Merrier: Relax the Non-Collusion Assumption in Multi-Server PIR," presented by Tiantian Gong, delves into a critical security challenge within **Private Information Retrieval (PIR)** schemes, specifically those utilizing multiple servers. PIR allows a client to retrieve an item from a database without the server(s) learning which item was requested. While multi-server PIR offers significant efficiency benefits over single-server approaches, these advantages traditionally come at the cost of a strong **non-collusion assumption** – specifically, that the servers will not conspire to reveal the client's query.
AI review
This research fundamentally shifts the paradigm for multi-server Private Information Retrieval (PIR) by replacing the naive non-collusion assumption with a practical rationality model. Through a clever game-theoretic mechanism, leveraging commitments and Zero-Knowledge Proofs, it disincentivizes server collusion, making efficient PIR robust for real-world deployment. A critical advancement for privacy-preserving systems operating in untrusted environments.