Welcome to the resource topic for 2023/096
Title:
MPC With Delayed Parties Over Star-Like Networks
Authors: Mariana Gama, Emad Heydari Beni, Emmanuela Orsini, Nigel P. Smart, Oliver Zajonc
Abstract:While the efficiency of secure multi-party computation protocols has greatly increased in the last few years, these improvements and protocols are often based on rather unrealistic, idealised, assumptions about how technology is deployed in the real world. In this work we examine multi-party computation protocols in the presence of two major constraints present in deployed systems. Firstly, we consider the situation where the parties are connected not by direct point-to-point connections, but by a star-like topology with a few central post-office style relays. Secondly, we consider MPC protocols with a strong honest majority (n \gg t/2) in which we have stragglers (some parties are progressing slower than others). We model stragglers by allowing the adversary to delay messages to and from some parties for a given length of time.
We first show that having only a single honest rely is enough to ensure consensus of the messages sent within a protocol; secondly, we show that special care must be taken to describe multiplication protocols in the case of relays and stragglers and that some well known protocols do not guarantee privacy and correctness in this setting; thirdly, we present an efficient honest-majority MPC protocol which can be run on top of the relays and which provides active-security with abort in the case of a strong honest majority, even when run with stragglers. We back up our protocol presentation with both experimental evaluations and simulations of the effect of the relays and delays on our protocol.
ePrint: https://eprint.iacr.org/2023/096
See all topics related to this paper.
Feel free to post resources that are related to this paper below.
Example resources include: implementations, explanation materials, talks, slides, links to previous discussions on other websites.
For more information, see the rules for Resource Topics .