Paper 2020/1137
MOTION - A Framework for Mixed-Protocol Multi-Party Computation
Lennart Braun and Daniel Demmler and Thomas Schneider and Oleksandr Tkachenko
Abstract
We present MOTION, an efficient and generic framework for mixed-protocol secure multi-party computation (MPC). Our framework is built from the ground up and incorporates several important engineering decisions such as full communication serialization which enables MPC over arbitrary messaging interfaces and removes the need of owning network sockets. It is available under the liberal MIT license and independent of external MPC libraries, which often have stricter licenses. MOTION is extensive and thoroughly tested: it currently consists of more than 36000 lines of code, 20% of which are unit and component tests. It is built in a user-friendly, modular, and extensible way, intended to be used as tool in MPC research and to increase adoption of MPC protocols in practice. MOTION incorporates several novel performance optimizations that improve the communication complexity and latency, e.g., 2x better online round complexity of precomputed correlated Oblivious Transfer (OT). We instantiate our framework with protocols for $N$ parties and security against up to $N-1$ passive corruptions: the MPC protocols of Goldreich-Micali-Wigderson (GMW) in its arithmetic and Boolean version and oblivious transfer (OT)-based BMR (Ben-Efraim et al., CCS'16), as well as novel and highly efficient conversions between them, including a non-interactive conversion from BMR to arithmetic GMW. Moreover, we design a novel garbling technique that saves 20% of communication in the BMR protocol. MOTION is highly efficient, which we demonstrate in our experiments by measuring its run-times in various network settings with different numbers of parties. For secure evaluation of AES-128 with $N=3$ parties in the high-latency network setting from the OT-based BMR paper, we achieve a 16x better throughput of 16 AES/s using BMR. This shows that the BMR protocol is much more competitive than previously assumed. For $N=3$ parties and full-threshold protocols in the LAN setting, MOTION is 10x-18x faster than the previous best passively secure implementation from the MP-SPDZ framework, and 190x-586x faster than the actively secure SCALE-MAMBA framework. Finally, we show that our framework is highly efficient for privacy preserving neural network inference.
Metadata
- Available format(s)
- Category
- Implementation
- Publication info
- Preprint. MINOR revision.
- Keywords
- secure multi-party computationhybrid protocolsefficiencyoutsourcing
- Contact author(s)
- lennart braun @ stud tu-darmstadt de,demmler @ informatik uni-hamburg de,schneider @ encrypto cs tu-darmstadt de,tkachenko @ encrypto cs tu-darmstadt de
- History
- 2022-04-01: last of 3 revisions
- 2020-09-21: received
- See all versions
- Short URL
- https://ia.cr/2020/1137
- License
-
CC BY