Combining Multi-Server and Single-Server Private Information Retrieval for Epidemiological Modelling
Bachelor Thesis
Motivation
Private Information Retrieval has various applications for any sensitive lookup of data (e.g. medical or financial data), privately retrieving software updates, improving privacy-guarantees of messaging services or improving the scalability of anonymous communication protocols like TOR. However, for many use-cases both computational PIR (cPIR) [1, 2] and information-theoretic PIR (IT-PIR) [3, 4] are not efficient enough for practical use, especially for large database sizes and/or small database entries. While cPIR makes use of homomorphic encryption (HE) and suffers from high computation on the server-side, IT-PIR has large communication costs for databases with many entries.
Goal
The goal of this thesis is to implement the hybrid PIR scheme of [5] making use of both cPIR and IT-PIR (using RAID-PIR [3, 4] and further optimizations) on mobile devices and analyse the performance improvements compared to using only cPIR or only IT-PIR for different database sizes and network settings (including LTE, WiFi, and WAN).
Requirements
- Good programming skills in C/C++
- Basic knowledge in Android development
- At least basic knowledge of cryptography
- High motivation + ability to work independently
- Knowledge of the English language, Git, LaTeX, etc. goes without saying
References
- [1] Sebastian Angel, Hao Chen, Kim Laine, and Srinath Setty. (opens in new tab). In S&P, 2018. PIR with compressed queries and amortized query processing
- [2] Craig Gentry and Shai Halevi. (opens in new tab). In TCC, 2019. Compressible FHE with applications to PIR
- [3] Daniel Demmler, Amir Herzberg, and Thomas Schneider. (opens in new tab). In CCSW, 2014. RAID-PIR: Practical multi-server PIR
- [4] Daniel Demmler, Marco Holz, and Thomas Schneider. (opens in new tab). In ACNS, 2017. OnionPIR: Effective Protection of Sensitive Metadata in Online Communication Networks
- [5] Casey Devet and Ian Goldberg. (opens in new tab). In PETS, 2014. The best of both worlds: Combining information-theoretic and computational PIR for communication efficiency
Supervisor
