CryptoDB
Cheater Identification on a Budget: MPC with Identifiable Abort from Pairwise MACs
| Authors: |
|
|---|---|
| Download: |
|
| Presentation: | Slides |
| Conference: | CRYPTO 2024 |
| Abstract: | Cheater identification in secure multi-party computation (MPC) allows the honest parties to agree upon the identity of a cheating party, in case the protocol aborts. In the context of a dishonest majority, this becomes especially critical, as it serves to thwart denial-of-service attacks and mitigate known impossibility results on ensuring fairness and guaranteed output delivery. In this work, we present a new, lightweight approach to achieving identifiable abort in dishonest majority MPC. We avoid all of the heavy machinery used in previous works, instead relying on a careful combination of lightweight detection mechanisms and techniques from state-of-the-art protocols secure with (non-identifiable) abort. At the core of our construction is a homomorphic, multi-receiver commitment scheme secure with identifiable abort. This commitment scheme can be constructed from cheap vector oblivious linear evaluation protocols based on learning parity with noise. To support cheater identification, we design a general compilation technique, similar to a compiler of Ishai et al. (Crypto 2014), but avoid its requirement for adaptive security of the underlying protocol. Instead, we rely on a different (and seemingly easier to achieve) property we call online extractability, which may be of independent interest. Our MPC protocol can be viewed as a version of the BDOZ MPC scheme (Bendlin et al., Eurocrypt 2011) based on pairwise information-theoretic MACs, enhanced to support cheater identification and a highly efficient preprocessing phase, essentially as efficient as the non-identifiable protocol of Le Mans (Rachuri \& Scholl, Crypto 2022). |
BibTeX
@inproceedings{crypto-2024-34254,
title={Cheater Identification on a Budget: MPC with Identifiable Abort from Pairwise MACs},
publisher={Springer-Verlag},
doi={10.1007/978-3-031-68397-8_14},
author={Carsten Baum and Nikolas Melissaris and Rahul Rachuri and Peter Scholl},
year=2024
}