Paper 2025/1313

Two-Server Sublinear PIR with Symmetric Privacy and Statistical Security

Shuaishuai Li, Zhongguancun Laboratory
Anyu Wang, Tsinghua University
Cong Zhang, Tsinghua University
Xiaoyun Wang, Tsinghua University
Abstract

The field of private information retrieval (PIR) has made significant strides with a recent focus on protocols that offer sublinear online time, ensuring efficient access to public databases without compromising the privacy of the queries. The pioneering two-server PIR protocols developed by Corrigan-Gibbs and Kogan (EUROCRYPT 2020) enjoy the dual benefits of sublinear online time and statistical security. This allows their protocols to provide high efficiency and resist computationally unbounded adversaries. In this work, we extend this seminal work to the symmetric PIR (SPIR) context, where the protocol must ensure that the client is privy only to the requested database entries, with no knowledge of the remaining data. This enhancement aligns with scenarios where the confidentiality of non-requested information is as critical as the query itself. Our main result is the introduction of the first two-server SPIR protocols that achieve both sublinear online time and statistical security, together with an enhancement for achieving sublinear amortized time. Our protocols require a pragmatic level of shared randomness between the servers, which however is necessary for implementing statistical security in two-server SPIR, as showed by Gertner et al. (STOC 1998).

Metadata
Available format(s)
PDF
Category
Cryptographic protocols
Publication info
Preprint.
Contact author(s)
liss @ zgclab edu cn
anyuwang @ tsinghua edu cn
zhangcong @ tsinghua edu cn
xiaoyunwang @ tsinghua edu cn
History
2025-07-19: approved
2025-07-18: received
See all versions
Short URL
https://ia.cr/2025/1313
License
Creative Commons Attribution
CC BY

BibTeX

@misc{cryptoeprint:2025/1313,
      author = {Shuaishuai Li and Anyu Wang and Cong Zhang and Xiaoyun Wang},
      title = {Two-Server Sublinear {PIR} with Symmetric Privacy and Statistical Security},
      howpublished = {Cryptology {ePrint} Archive, Paper 2025/1313},
      year = {2025},
      url = {https://eprint.iacr.org/2025/1313}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.