Functional Secret Sharing for Two-Party Secure Thresholded Distance Evaluation.
- Languages: C (99+), Python (3.8+, for the wrapper)
- Platforms: Linux, Windows, MacOS (any system with a C compiler)
- Dependencies:
libsodium(optional, for fast and secure random number generation) - Code Author: Alberto Ibarrondo
- License: GNU GPLv3 (any code derived from this must be open source)
- Version: 1.1.0
Funshade is a library that implements a protocol to securely compute a distance metric between two vectors, followed by a secure comparison to a threshold. It is extremely lightweight, making use of cheap primitives such as AES and requiring a single intermediate round of communication and just two numbers sent per distance computation in the preprocessing model (besides input sharing and output reconstruction, common to all secret-sharing MPC schemes).
Consider citing the paper if you use it in your research!
The library is written in plain C to make it portable. It has been tested in Linux and Windows, but it should work in any system with a C compiler. To compile it, you can:
- Use the provided CMakeLists.txt with
cmake(mkdir build && cd build && cmake .. & cmake --build .) - Directly call your compiler with the
-msse -msse2 -maesflags for faster hardware-based AES acceleration (and consider-O3 -march=nativefor further optimizations).
For conveniency and for seamless integration with higher-level languages, we also provide a Python wrapper. Install it with:
pip install .
As an optional dependency, it uses libsodium for fast and secure random number generation.
The library is designed to be used as a black-box, with a simple API.
Check the bottom of fss.h for the available functions, test_fss.c for some uses in C, or test_funshade.py for a step-by-step Python example.
- Additive secret sharing (refer to the paper, page 22 for further explanation on this scheme)