CompGC: Efficient Offline/Online Semi-Honest Two-Party Computation
Abstract
We introduce a new technique, component-based garbled circuits, for increasing the efficiency of secure two-party computation in the offline/online semi-honest setting. We observe that real-world functions are generally constructed in a modular way, comprising many standard components such as arithmetic operations and other common tasks. Our technique allows circuits for these common tasks to be garbled and shared during an offline phase; once the function to compute is specified, these pre-shared components can be chained together to create a larger garbled circuit. We stress that we do not assume that the function is known during the offline phase only that it uses some common, predictable components. We give an implementation, Comp GC, of this technique and measure the efficiency gains for various examples. We find that our technique results in roughly an order of magnitude performance improvement over standard garbled circuit-based secure two-party computation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jul 06, 2016
- Accession Number
- AD1033490
Entities
People
- Adam Groce
- Alex J. Malozemoff
- Alex Ledger
- Arkady Yerukhimovich
Organizations
- MIT Lincoln Laboratory