Preprocessing for Life: Dishonest-Majority MPC with a Trusted or Untrusted Dealer

Elette Boyle, Niv Gilboa, Matan Hamilis, Yuval Ishai, Ariel Nof

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

We put forth a new paradigm for secure multi-party computation (MPC) in the preprocessing model, where a feasible one-time setup can enable a lifetime of efficient online secure computations. Our protocols match the security guarantees and low costs of the cheapest category of MPC solutions, namely 3-party protocols (3PC) secure against a single malicious party, with the qualitative advantages that one party communicates data sublinear in the circuit size, and can go offline after its initial messages. This '2+ 1'-party structure can alternatively be instantiated between 2 parties with the aid of an (untrusted) dealer. Within such existing protocols, we provide comparable online performance while improving the storage and offline dealer-to-party communication requirements by more than 3 orders of magnitude. At the technical level, we build on the Fully Linear Interactive Oracle Proof (FLIOP)-based protocol design of Boyle et al. (CRYPTO 2021). We provide an extensive assortment of algorithmic and implementation-level optimizations, design efficient distributed proofs of well-formedness of complex FLIOP correlations, and make them circuit-independent. We implement and benchmark our end-to-end system against the state of the art in the 2+1 regime, a dealer-aided variant of SPDZ for Boolean circuits. We additionally extend our techniques to the (n+1) party setting, where a dealer aids general dishonest-majority MPC, and provide a variant of the protocol which further achieves security with 'identifiable abort.'

Original languageEnglish
Title of host publicationProceedings - 46th IEEE Symposium on Security and Privacy, SP 2025
EditorsMarina Blanton, William Enck, Cristina Nita-Rotaru
Pages2433-2452
Number of pages20
ISBN (Electronic)9798331522360
DOIs
StatePublished - 2025
Event46th IEEE Symposium on Security and Privacy, SP 2025 - San Francisco, United States
Duration: 12 May 202515 May 2025

Publication series

NameProceedings - IEEE Symposium on Security and Privacy

Conference

Conference46th IEEE Symposium on Security and Privacy, SP 2025
Country/TerritoryUnited States
CitySan Francisco
Period12/05/2515/05/25

Keywords

  • multi-party computation

All Science Journal Classification (ASJC) codes

  • Safety, Risk, Reliability and Quality
  • Software
  • Computer Networks and Communications

Cite this