Alexandra Henzinger
March 22, 2023 at 11:00 AM on Zoom / Soda Hall
Simple and Fast Single-Server Private Information Retrieval
Abstract:
Private-information-retrieval protocols allow a client to query a database server while hiding its query from the server. With private information retrieval (PIR), a client could, for example, fetch an article from Wikipedia without revealing which article it is interested in. In spite of the vast potential applications of PIR---to everything from private database search, to credential breach reporting, to privacy-preserving advertising---there have been markedly few real-world deployments of the technology. PIR has been deemed too impractical for use at scale, as existing schemes require (a) large server-side computation costs, or (b) two or more non-colluding database servers.
This talk will present SimplePIR, a new single-server PIR scheme that is over 30x faster than state-of-the-art, comparable protocols. In addition, SimplePIR is easy to implement and simple to describe; our complete open-source implementation consists of 1,400 lines of Go, plus 200 lines of C, and uses no external libraries. With this talk, I hope to demonstrate that the costs of PIR have fallen to a point that makes real-world deployments feasible for many applications, and I will sketch one such application to the task of private auditing in Certificate Transparency.
This talk is based on joint work with Matthew M. Hong, Henry Corrigan-Gibbs, Sarah Meiklejohn, and Vinod Vaikuntanathan.
Bio: Alexandra is a third year PhD student in the PDOS and CSS research groups at MIT, advised by Henry Corrigan-Gibbs. She is interested in computer systems, security, and cryptography. Recently, much of her work has focused on constructions, implementations, and applications of private information retrieval. Before coming to MIT, she graduated from Stanford with a BS with honors in Computer Science.