Paper 2026/973

Asynchronous Lagrange-Based Threshold FHE with Smaller Modulus Overhead

Won Kim, Korea University
Changmin Lee, Korea University
JeongHwan Lee, Korea University
Alain Passelègue, Cryptolab Inc.
Damien Stehlé, Cryptolab Inc.
Abstract

We study $t$-out-of-$n$ threshold fully homomorphic encryption (ThFHE) based on Shamir secret sharing (SSS) in the asynchronous setting. A central bottleneck for SSS-based ThFHE is that Lagrange reconstruction during distributed decryption can amplify noise, forcing a substantially larger ciphertext modulus to maintain correctness. In this work, we revisit SSS-based ThFHE and give a rigorous analysis of the correctness and simulation-security constraints that govern parameter choices. We then compare families of Lagrange interpolation points through the lens of these constraints. Our main contributions are analytic bounds that closely track empirical behavior and significantly reduce the modulus overhead required for distributed decryption. For example, for $n = 512$, our analysis reduces this modulus overhead (in bits) by 30% for $t = n/2$ and by up to 90% for $t$ close to $n$, compared to prior parameterizations.

Metadata
Available format(s)
PDF
Category
Cryptographic protocols
Publication info
Published by the IACR in CRYPTO 2026
Keywords
Homomorphic encryptionThreshold cryptography
Contact author(s)
munsanwon2 @ korea ac kr
changminlee @ korea ac kr
hwani0814 @ korea ac kr
alain passelegue @ cryptolab co kr
damien stehle @ gmail com
History
2026-05-27: revised
2026-05-16: received
See all versions
Short URL
https://ia.cr/2026/973
License
Creative Commons Attribution
CC BY

BibTeX

@misc{cryptoeprint:2026/973,
      author = {Won Kim and Changmin Lee and JeongHwan Lee and Alain Passelègue and Damien Stehlé},
      title = {Asynchronous Lagrange-Based Threshold {FHE} with Smaller Modulus Overhead},
      howpublished = {Cryptology {ePrint} Archive, Paper 2026/973},
      year = {2026},
      url = {https://eprint.iacr.org/2026/973}
}
Note: In order to protect the privacy of readers, eprint.iacr.org does not use cookies or embedded third party content.