search

data61 / MP-SPDZ

Versatile framework for multi-party computation
Other
869 stars 272 forks source link

Robust MPC #532

Closed lu562 closed 2 years ago

lu562 commented 2 years ago

Hello,

May I know if MP-SPDZ has implemented some protocols supporting robust/fair MPC?

Thanks!

mkskeller commented 2 years ago

No, there aren't any. The closest might be Rep4 because the security model would allow adding robustness relatively easy.

lu562 commented 2 years ago

Thank you for the confirmation!

lu562 commented 2 years ago

btw may I know which paper does "malicious-shamir-party.x" implement?

I'm searching for a fast malicious-secure secret-sharing-based protocol (in terms of multiplication and reconstruction), which one do you think could be a good option?

Thanks!

mkskeller commented 2 years ago

Malicious Shamir uses Beaver multiplication with triples that are generated using the semi-honest Shamir protocol and then sacrificed. The description by Lindell and Nof (https://eprint.iacr.org/2017/816) comes close but we don't use triple generation according to Damgård and Nielsen.

The best option depends on the number of players and corrupted players. ATLAS is probably better than malicious Shamir, I'd use sy-rep-ring for 3 parties and rep4-ring for 4. With a dishonest majority, protocols based on homomorphic encryption usually come out on top.

lu562 commented 2 years ago

Thanks!