An implementation of LMW23-style doubly efficient PIR

This is the code accompanying our paper OPPW23. It is written in Rust, based only on the feanor-math library. Currently, the asynchronous reading from an SSD is only implemented on Windows.

Computing the preevaluation databases

In addition to the main implementation in Rust, we accelerate the computation of the evaluation database using GPUs. This is the code in poly-batch-eval. When running it, make sure that the parameters d and m match the ones of the interpolation polynomial. The implementation is quite naive, since the performance of this step is not the focus of our work.

References

[LMW23] W. Lin, E. Mook, D. Wichs. “Near-Optimal Private Information Retrieval with Preprocessing”. In: IACR Cryptol. ePrint Arch. (2022), p. 830.

[OPPW23] Hiroki Okada, Rachel Player, Simon Pohmann, and Christian Weinert. Towards practical doubly-efficient private information retrieval. Cryptology ePrint Archive, Paper 2023/1510, 2023.