FedAvg-FHEMK: Aprendizado Federado Seguro com Criptografia Homomórfica Multi-Chave Eficiente
Resumo
A adoção do FL expõe riscos de vazamento de informação, pois ainda é possível recuperar os dados a partir das atualizações do modelo compartilhadas durante o protocolo. A criptografia homomórfica (FHE) traz segurança ao permitir computação sobre dados cifrados, mas sua aplicação em FL, especialmente no cenário multi-chave, é limitada por altos custos computacionais, sobrecarga de comunicação e protocolos interativos. Este trabalho propõe o FedAvg-FHEMK, um protocolo de aprendizado federado seguro baseado em FHE com arranjo multi-chave modificado, que reduz custo e complexidade ao exigir apenas uma rodada de comunicação por iteração e concentrar o processamento em operações quase lineares. O esquema é mais adequado a cenários cross-silo estáveis, pois requer uma fase inicial mais custosa e não tolera abandono de clientes. Ele garante proteção contra servidor semi-honesto mesmo sob colusão de até P − 2 clientes, onde P é o número total de clientes participantes da federação. Experimentos mostram que o FedAvg-FHEMK preserva acurácia do modelo, com aumento moderado do custo computacional e redução significativa do custo de comunicação em relação a abordagens de FHE de chave única, indicando um compromisso prático entre privacidade forte e eficiência em aprendizado federado seguro.Referências
Bell, J. H., Bonawitz, K. A., Gascón, A., Lepoint, T., and Raykova, M. (2020). Secure single-server aggregation with sublinear overhead. In Proceedings of the 2020 ACM SIGSAC Conference on Computer and Communications Security, pages 1253–1269.
Bonawitz, K., Ivanov, V., Kreuter, B., Marcedone, A., McMahan, B., Patel, S., Ramage, D., Segal, A., and Seth, K. (2017). Practical secure aggregation for privacy-preserving machine learning. In Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security, pages 1175–1191.
Chen, H., Dai, W., Kim, M., and Song, Y. (2019). Efficient multi-key homomorphic encryption with packed ciphertexts with application to oblivious neural network inference. In Proceedings of the ACM SIGSAC Conference on Computer and Communications Security (CCS), pages 395–412.
Correia, P., Silva, I., Amorim, I., Maia, E., et al. (2025). Federated learning: An approach with hybrid homomorphic encryption. arXiv preprint arXiv:2509.03427.
Geiping, J., Bauermeister, H., Dröge, H., and Moeller, M. (2020). Inverting gradients-how easy is it to break privacy in federated learning? Advances in neural information processing systems, 33:16937–16947.
Hu, C. and Li, B. (2024). Maskcrypt: Federated learning with selective homomorphic encryption. IEEE Transactions on Dependable and Secure Computing, 22(1):221–233.
Jin, W., Yao, Y., Han, S., Joe-Wong, C., Ravi, S., Avestimehr, S., and He, C. (2024). Fedml-he: An efficient homomorphic-encryption-based privacy-preserving federated learning system. arXiv preprint arXiv:2303.10837.
Korkmaz, A. and Rao, P. (2025). A selective homomorphic encryption approach for faster privacy-preserving federated learning. arXiv preprint arXiv:2501.12911.
López-Alt, A., Tromer, E., and Vaikuntanathan, V. (2012). On-the-fly multiparty computation on the cloud via multikey fully homomorphic encryption. In Proceedings of the forty-fourth annual ACM symposium on Theory of computing, pages 1219–1234.
McMahan, B., Moore, E., Ramage, D., Hampson, S., and y Arcas, B. A. (2017). Communication-efficient learning of deep networks from decentralized data. Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS).
Mignotte, M. (1982). How to share a secret. In Workshop on cryptography, pages 371–375. Springer.
Mouchet, C., Troncoso-Pastoriza, J., Bossuat, J.-P., and Hubaux, J.-P. (2021). Multi-party homomorphic encryption from ring-learning-with-errors. Proceedings on Privacy Enhancing Technologies, 2021(4):291–311.
Omar, A. A., Yang, X., Choo, E., and Ardakanian, O. (2025). Efficient privacy-preserving cross-silo federated learning with multi-key homomorphic encryption. arXiv preprint arXiv:2505.14797.
Sav, S., Pyrgelis, A., Troncoso-Pastoriza, J. R., Froelicher, D., Bossuat, J.-P., Sousa, J. S., and Hubaux, J.-P. (2020). Poseidon: Privacy-preserving federated neural network learning. arXiv preprint arXiv:2009.00349.
Wu, R., Chen, X., Guo, C., and Weinberger, K. Q. (2023). Learning to invert: Simple adaptive attacks for gradient inversion in federated learning. In Uncertainty in Artificial Intelligence, pages 2293–2303. PMLR.
Zhang, C., Li, S., Xia, J., Wang, W., Yan, F., and Liu, Y. (2020). BatchCrypt: Efficient homomorphic encryption for Cross-Silo federated learning. In 2020 USENIX annual technical conference (USENIX ATC 20), pages 493–506.
Zhu, L., Liu, Z., and Han, S. (2019). Deep leakage from gradients. Advances in neural information processing systems, 32.
Bonawitz, K., Ivanov, V., Kreuter, B., Marcedone, A., McMahan, B., Patel, S., Ramage, D., Segal, A., and Seth, K. (2017). Practical secure aggregation for privacy-preserving machine learning. In Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications Security, pages 1175–1191.
Chen, H., Dai, W., Kim, M., and Song, Y. (2019). Efficient multi-key homomorphic encryption with packed ciphertexts with application to oblivious neural network inference. In Proceedings of the ACM SIGSAC Conference on Computer and Communications Security (CCS), pages 395–412.
Correia, P., Silva, I., Amorim, I., Maia, E., et al. (2025). Federated learning: An approach with hybrid homomorphic encryption. arXiv preprint arXiv:2509.03427.
Geiping, J., Bauermeister, H., Dröge, H., and Moeller, M. (2020). Inverting gradients-how easy is it to break privacy in federated learning? Advances in neural information processing systems, 33:16937–16947.
Hu, C. and Li, B. (2024). Maskcrypt: Federated learning with selective homomorphic encryption. IEEE Transactions on Dependable and Secure Computing, 22(1):221–233.
Jin, W., Yao, Y., Han, S., Joe-Wong, C., Ravi, S., Avestimehr, S., and He, C. (2024). Fedml-he: An efficient homomorphic-encryption-based privacy-preserving federated learning system. arXiv preprint arXiv:2303.10837.
Korkmaz, A. and Rao, P. (2025). A selective homomorphic encryption approach for faster privacy-preserving federated learning. arXiv preprint arXiv:2501.12911.
López-Alt, A., Tromer, E., and Vaikuntanathan, V. (2012). On-the-fly multiparty computation on the cloud via multikey fully homomorphic encryption. In Proceedings of the forty-fourth annual ACM symposium on Theory of computing, pages 1219–1234.
McMahan, B., Moore, E., Ramage, D., Hampson, S., and y Arcas, B. A. (2017). Communication-efficient learning of deep networks from decentralized data. Proceedings of the 20th International Conference on Artificial Intelligence and Statistics (AISTATS).
Mignotte, M. (1982). How to share a secret. In Workshop on cryptography, pages 371–375. Springer.
Mouchet, C., Troncoso-Pastoriza, J., Bossuat, J.-P., and Hubaux, J.-P. (2021). Multi-party homomorphic encryption from ring-learning-with-errors. Proceedings on Privacy Enhancing Technologies, 2021(4):291–311.
Omar, A. A., Yang, X., Choo, E., and Ardakanian, O. (2025). Efficient privacy-preserving cross-silo federated learning with multi-key homomorphic encryption. arXiv preprint arXiv:2505.14797.
Sav, S., Pyrgelis, A., Troncoso-Pastoriza, J. R., Froelicher, D., Bossuat, J.-P., Sousa, J. S., and Hubaux, J.-P. (2020). Poseidon: Privacy-preserving federated neural network learning. arXiv preprint arXiv:2009.00349.
Wu, R., Chen, X., Guo, C., and Weinberger, K. Q. (2023). Learning to invert: Simple adaptive attacks for gradient inversion in federated learning. In Uncertainty in Artificial Intelligence, pages 2293–2303. PMLR.
Zhang, C., Li, S., Xia, J., Wang, W., Yan, F., and Liu, Y. (2020). BatchCrypt: Efficient homomorphic encryption for Cross-Silo federated learning. In 2020 USENIX annual technical conference (USENIX ATC 20), pages 493–506.
Zhu, L., Liu, Z., and Han, S. (2019). Deep leakage from gradients. Advances in neural information processing systems, 32.
Publicado
25/05/2026
Como Citar
ROSA, Gabriel S.; PEREIRA, Hilder V. L..
FedAvg-FHEMK: Aprendizado Federado Seguro com Criptografia Homomórfica Multi-Chave Eficiente. In: SIMPÓSIO BRASILEIRO DE REDES DE COMPUTADORES E SISTEMAS DISTRIBUÍDOS (SBRC), 44. , 2026, Praia do Forte/BA.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 786-799.
ISSN 2177-9384.
DOI: https://doi.org/10.5753/sbrc.2026.19794.
