Paper 2026/2950

LNRPIR: High Throughput PIR with Flexible Communication-Computation Trade-offs

Liren Xu, Xidian University
Zhe Li, Xidian University
Abstract

Private information retrieval (PIR) is a powerful cryptographic primitive with numerous applications. Recent works have largely pursued two different directions: minimizing online communication at the expense of high online computation costs, or achieving high server throughput while incurring substantial online communication overhead. In this work, we introduce LNRPIR, a PIR scheme that achieves high server throughput with moderate online communication overhead. LNRPIR also provides flexible trade-offs between communication and computation, addressing a gap in the trade-off space left by prior works. Our evaluation shows that LNRPIR retrieves a $256$-byte record from an $8$ GB database $2.541\times$ faster than VIA-C (S&P 2026), a state-of-the-art protocol in the client-hint model, with only $10.473$ KB of online communication. Furthermore, compared with InsPIRe (S&P 2026), a state-of-the-art protocol in the silent-preprocessing model, LNRPIR achieves $17.074\times$ lower online communication overhead for an $8$ GB database, while incurring only $1.885\times$ the online computation time of InsPIRe. LNRPIR primarily relies on three novel ciphertext conversion techniques: LWE-to-NTRU conversion, LWE-to-VNTRU conversion, and NTRU-to-RLWE conversion.

Metadata
Available format(s)
PDF
Category
Cryptographic protocols
Publication info
Preprint.
Keywords
Private Information RetrievalHomomorphic EncryptionCiphertext Conversion
Contact author(s)
24151111259 @ stu xidian edu cn
lizh0048 @ e ntu edu sg
History
2026-11-22: approved
2026-11-19: received
See all versions
Short URL
https://ia.cr/2026/2950
License
Creative Commons Attribution
CC BY

BibTeX

@misc{cryptoeprint:2026/2950,
      author = {Liren Xu and Zhe Li},
      title = {{LNRPIR}: High Throughput {PIR} with Flexible Communication-Computation Trade-offs},
      howpublished = {Cryptology {ePrint} Archive, Paper 2026/2950},
      year = {2026},
      url = {https://eprint.iacr.org/2026/2950}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.