FedG2L: a privacy-preserving federated learning scheme base on “G2L” against poisoning attack

Figures & data

Table 1. Summary of existing poisoning attack defense schemes in FL.

Scheme	PGP	WBG	PoP
Anti-poisoning attacks	✗	✗	✗
(Chen et al., 2021b; Levine & Feizi, 2020; Wen et al., 2021b)
FL against poisoning attacks	✗	✗	✗
(Awan et al., 2021; Shejwalkar & Houmansadr, 2021; Sun et al., 2021)
PEFL (Liu et al., 2021)	✓	✗	✗
ShiledFL (Ma et al., 2022)	✓	✗	✗
Blockchain-based FL	✗	✗	✗
(Hou et al., 2021; Li et al., 2022; Yang et al., 2023)
PBFL (Miao et al., 2022)	✓	✗	✗
Our scheme: FedG2L	✓	✓	✓

Note: PGP: Protect gradient privacy; WBG: Without benign gradient; PoP: Period of poisoning. ✗ Considered. ✓ Not considered.

Table 2. Notations.

Notation	Description
pk	The public key used to encrypt gradients.
sk	The secret key used to decrypt gradients.
$G_k$	Sub-gradient.
n	The number of nodes in the blockchain.
$G_{avg}$	The average gradient calculated by model aggregation.
ε	The search radius.
MinPts	The minimum limitation of gradients in a cluster.
l	The epoch of model training.
$C_{max}$	The gradient set which clusters most gradients.
$\mathbb{E}$	The expectation.
η	The learning rate.
ACC	The maximum model accuracy.
$\wedge$	The empty set.
R	The random vector.
$L_{node}$	The loss function of the node task model.

Figure 1. The PBFT consensus protocol.

Figure 2. The process of DBSCAN. (a) First step (b) Second step.

Figure 3. The architecture of GAN.

Figure 4. System model.

Figure 5. The process of FedG2L.

Figure 6. Comparison of accuracy with different epochs and byzantine percentage. (a) Epochs. (b) Byzantine percentage.

Figure 7. The ASR with different Byzantine percentages on 2 datasets. (a) Performance on MNIST dataset. (b) Performance on EMNIST dataset.

Figure 8. Comparison of accuracy with different schemes.

Figure 9. Comparison of accuracy with different numbers of generated data on different datasets.

Chen, J., Zhang, X., Zhang, R., Wang, C., & Liu, L. (2021b). De-pois: an attack-agnostic defense against data poisoning attacks. IEEE Transactions on Information Forensics and Security, 16(2021), 3412–3425. https://doi.org/10.1109/TIFS.2021.3080522

 Google Scholar

Levine, A., & Feizi, S. (2020). Deep partition aggregation: Provable defense against general poisoning attacks. arXiv preprint arXiv:2006.14768.

 Google Scholar

Wen, J., Zhao, B. Z. H., Xue, M., Oprea, A., & Qian, H. (2021b). With great dispersion comes greater resilience: efficient poisoning attacks and defenses for linear regression models. IEEE Transactions on Information Forensics and Security, 16(2021), 3709–3723. https://doi.org/10.1109/TIFS.2021.3087332

 Google Scholar

Awan, S., Luo, B., & Li, F. (2021). Contra: defending against poisoning attacks in federated learning. In European symposium on research in computer security (pp. 455–475). Springer.

 Google Scholar

Shejwalkar, V., & Houmansadr, A. (2021). Manipulating the byzantine: optimizing model poisoning attacks and defenses for federated learning. In NDSS. ISOC.

 Google Scholar

Sun, J., Li, A., DiValentin, L., Hassanzadeh, A., Chen, Y., & Li, H. (2021). Fl-wbc: enhancing robustness against model poisoning attacks in federated learning from a client perspective. In Advances in Neural Information Processing Systems (Vol. 34, pp. 12613–12624). MIT Press.

 Google Scholar

Liu, X., Li, H., Xu, G., Chen, Z., Huang, X., & Lu, R. (2021). Privacy-enhanced federated learning against poisoning adversaries. IEEE Transactions on Information Forensics and Security, 16(2021), 4574–4588. https://doi.org/10.1109/TIFS.2021.3108434

 Google Scholar

Ma, Z., Ma, J., Miao, Y., Li, Y., & Deng, R. H. (2022). Shieldfl: mitigating model poisoning attacks in privacy-preserving federated learning. IEEE Transactions on Information Forensics and Security, 17(2022), 1639–1654. https://doi.org/10.1109/TIFS.2022.3169918

 Google Scholar

Hou, D., Zhang, J., Man, K. L., Ma, J., & Peng, Z. (2021). A systematic literature review of blockchain-based federated learning: architectures, applications and issues. In 2021 2nd Information communication technologies conference (ICTC) (pp. 302–307). IEEE.

 Google Scholar

Li, D., Han, D., Weng, T.-H., Zheng, Z., Li, H., Liu, H., Castiglione, A., & Li, K.-C. (2022). Blockchain for federated learning toward secure distributed machine learning systems: a systemic survey. Soft Computing, 26(9), 4423–4440. https://doi.org/10.1007/s00500-021-06496-5

 PubMed Web of Science ®Google Scholar

Yang, Z., Shi, Y., Zhou, Y., Wang, Z., & Yang, K. (2023). Trustworthy federated learning via blockchain. IEEE Internet of Things Journal, 10(1), 92–109. https://doi.org/10.1109/JIOT.2022.3201117

 Web of Science ®Google Scholar

Miao, Y., Liu, Z., Li, H., Choo, K.-K. R., & Deng, R. H. (2022). Privacy-preserving byzantine-robust federated learning via blockchain systems. IEEE Transactions on Information Forensics and Security, 17(2022), 2848–2861. https://doi.org/10.1109/TIFS.2022.3196274

 Google Scholar