PhantomFHE: A CUDA-Accelerated Fully Homomorphic Encryption Library

Important

This is a research project and is not intended for production use. We are actively working on improving the performance and usability of this library. If you have any questions or suggestions, please feel free to open an issue or contact us.

Warning

This project has been tested on Tesla A100 40G/80G, GTX 3080Ti/3090Ti/4090, AGX Xavier. Other GPUs may have compatibility issues and may not give correct results.

Documentation

Please read https://encryptorion-lab.gitbook.io/phantom-fhe/ for detailed instructions and explanations.

Features

• Native GPU acceleration (for NVIDIA GPUs)
• Support word-wise schemes including BGV, BFV, and CKKS (without bootstrapping)
• SOTA performance in most operations
• Easy to integrate with applications (PPML, etc.)

License

This project (PhantomFHE) is released under GPLv3 license. See LICENSE for more information.

Some files contain the modified code from Microsoft SEAL. These codes are released under MIT License. See MIT License for more information.

Some files contain the modified code from OpenFHE. These codes are released under BSD 2-Clause License. See BSD 2-Clause License for more information.

Citation

If you use Phantom in your research, please cite the following paper:

@article{DBLP:journals/tdsc/YangSDZLZ24,
  author       = {Hao Yang and
                  Shiyu Shen and
                  Wangchen Dai and
                  Lu Zhou and
                  Zhe Liu and
                  Yunlei Zhao},
  title        = {Phantom: {A} CUDA-Accelerated Word-Wise Homomorphic Encryption Library},
  journal      = {{IEEE} Trans. Dependable Secur. Comput.},
  volume       = {21},
  number       = {5},
  pages        = {4895--4906},
  year         = {2024},
  url          = {https://doi.org/10.1109/TDSC.2024.3363900},
  doi          = {10.1109/TDSC.2024.3363900},
  timestamp    = {Fri, 20 Sep 2024 14:01:59 +0200},
  biburl       = {https://dblp.org/rec/journals/tdsc/YangSDZLZ24.bib},
  bibsource    = {dblp computer science bibliography, https://dblp.org}
}

If you are exploring BFV optimizations, please also cite the following paper:

@article{PhantomFHE_BFV,
    author={Shen, Shiyu and Yang, Hao and Dai, Wangchen and Zhou, Lu and Liu, Zhe and Zhao, Yunlei},
    journal={IEEE Transactions on Computers},
    title={Leveraging GPU in Homomorphic Encryption: Framework Design and Analysis of BFV Variants},
    year={2024},
    volume={73},
    number={12},
    pages={2817-2829},
    doi={10.1109/TC.2024.3457733},
}

Roadmap

We are planning to support the following features in the future:

• support bit-wise schemes TFHE/FHEW
• support scheme switching between word-wise schemes and bit-wise schemes
• support bootstrapping for BFV/BGV/CKKS