Duke Computer Science Seminar

Practical Single-server Private Information Retrieval

September 6, -
Speaker(s): Ling Ren
Abstract

In this talk, I will present our recent results on practical single-server private information retrieval (PIR). I will start with OnionPIR, a single-server PIR protocol whose response overhead is a small constant. OnionPIR utilizes recent advances in leveled fully homomorphic encryption (FHE) and carefully composes different lattice-based FHE schemes and homomorphic operations to control the noise growth and response size.

State-of-the-art single-server PIR schemes (OnionPIR included) are faced with fundamental barriers that limit their efficiency. In the second part of the talk, I will present two efficient amortized PIR protocols when the client has many entries to retrieve. The first protocol makes use of vectorized FHE to handle a batch of queries at a similar cost for a single query. The second protocol introduces a new hint system to obtain a simple and lightweight stateful PIR protocol that fulfills a query within milliseconds.

Lunch

Lunch will be served.

Speaker Bio

Ling Ren is an assistant professor in the Department of Computer Science at the University of Illinois at Urbana-Champaign. Prior to joining the University of Illinois, he obtained his Ph.D. from MIT and worked at VMware Research. He has won several awards including NSF CAREER Award, Google research scholar award, VMware Early Career Faculty Grant, Top Picks in Hardware and Embedded Security, and Best Paper Runner-Up and Best Student Paper at CCS. 

 

Contact

Kartik Nayak