Paper 2026/808

Efficient Two-Online-Round Secure Comparison in the Plain Model

Koji Nuida, Kyushu University, National Institute of Advanced Industrial Science and Technology
Satsuya Ohata, Kyushu University, Byerlis Inc.
Abstract

Secure comparison is a fundamental building block frequently employed in various applications of secure multiparty computation, such as secure machine learning. Such protocols based on secret sharing (SS) typically excel in throughput, but they historically suffer from higher (online) round complexity. To address this issue, Attrapadung et al. (AsiaCCS 2022) proposed a protocol with two online rounds in the client-aided model (i.e., assuming a trusted third party in the offline phase), and Lu et al. (USENIX Security 2025) proposed a protocol with three online rounds in the plain model. In this paper, we propose a plaintext comparison protocol with two online rounds in the plain model. While keeping the lowest online round complexity, our protocol also reduces the total bit complexity. For example, for the case of 64-bit integers, the total bit complexity of our protocol is 25.7% of Lu et al.'s protocol and 23.3% of Attrapadung et al.'s protocol adjusted to the plain model.

Metadata
Available format(s)
PDF
Category
Cryptographic protocols
Publication info
Preprint.
Keywords
secure multiparty computationinteger comparison protocolconstant-round protocol
Contact author(s)
nuida @ imi kyushu-u ac jp
satsuya ohata @ byerlis jp
History
2026-08-19: last of 2 revisions
2026-04-24: received
See all versions
Short URL
https://ia.cr/2026/808
License
Creative Commons Attribution
CC BY

BibTeX

@misc{cryptoeprint:2026/808,
      author = {Koji Nuida and Satsuya Ohata},
      title = {Efficient Two-Online-Round Secure Comparison in the Plain Model},
      howpublished = {Cryptology {ePrint} Archive, Paper 2026/808},
      year = {2026},
      url = {https://eprint.iacr.org/2026/808}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.