Paper 2025/1007 Original: eprint.iacr.org/2025/1007 ↗

Scalable Multiparty Computation from Non-linear Secret Sharing

Sanjam Garg
Abhishek Jain
Pratyay Mukherjee
Mingyuan Wang
Abstract

A long line of work has investigated the design of scalable secure multiparty computation (MPC) protocols with computational and communication complexity independent of the number of parties (beyond any dependence on the circuit size). We present the first unconditionally-secure MPC protocols for arithmetic circuits over {\em large fields} with total computation $\mathcal{O}(|C|\log|F|)$, where $|C|$ and $|F|$ denote the circuit and field size, respectively. Prior work could either achieve similar complexity only in {\em communication}, or required highly structured circuits, or expensive circuit transformations. To obtain our results, we depart from the prior approach of share packing in linear secret-sharing schemes; instead, we use an ``unpacking'' approach via {\em non-linear} secret sharing.

Metadata
Available format(s)
PDF
Category
Foundations
Publication info
Preprint.
History
2025-05-31: last updated
2025-05-31: published
Short URL
https://ia.cr/2025/1007
License
https://creativecommons.org/licenses/by/4.0/

BibTeX

@misc{cryptoeprint:2025/1007,
      author = {Sanjam Garg and Abhishek Jain and Pratyay Mukherjee and Mingyuan Wang},
      title = {Scalable Multiparty Computation from Non-linear Secret Sharing},
      howpublished = {Cryptology {ePrint} Archive, Paper 2025/1007},
      year = {2025},
      url = {https://eprint.iacr.org/2025/1007}
}
Note: In order to protect the privacy of readers, this archive does not use cookies or embedded third party content.