Efficient Federated Learning with Encrypted Data Sharing for Data-Heterogeneous Edge Devices

Status: Accepted by ICWS 2025 (IEEE International Conference on Web Services)

Paper Here: arXiv

This repository contains the official implementation of the paper, "Efficient Federated Learning with Encrypted Data Sharing for Data-Heterogeneous Edge Devices". The paper proposes FedEDS, a novel federated learning framework designed for edge devices with heterogeneous data. By leveraging encrypted data sharing between clients and training models with both local and shared encrypted data, FedEDS effectively mitigates the challenges of data heterogeneity and accelerates convergence. Experimental results demonstrate that FedEDS significantly improves model performance and training efficiency compared to traditional methods, making it well-suited for real-world edge applications.

How to run:

Use the following command to run the code.

# dataset_name: CIFAR10 or FashionMNIST
python main.py --dataset_name [dataset_name]

The code will automatically save the running results to the results folder.

The different datasets used by the code are saved in the datasets folder.

The saved_splits folder saves the data division method.

Citation

If our work is useful for your research, please consider citing:

@INPROCEEDINGS{11169706,
  author={Li, Hangyu and Wu, Hongyue and Fan, Guodong and Zhang, Zhen and Chen, Shizhan and Feng, Zhiyong},
  booktitle={2025 IEEE International Conference on Web Services (ICWS)}, 
  title={Efficient Federated Learning With Encrypted Data Sharing for Data-Heterogeneous Edge Devices}, 
  year={2025},
  volume={},
  number={},
  pages={76-86},
  keywords={Performance evaluation;Training;Data privacy;Federated learning;Stochastic processes;Data models;Cryptography;Servers;Protection;Convergence;Federated Learning;Edge Computing;Encrypted Data Sharing;Privacy Protection;Data-Heterogeneous Edge Devices},
  doi={10.1109/ICWS67624.2025.00019}}

@misc{li2025efficientfederatedlearningencrypted,
      title={Efficient Federated Learning with Encrypted Data Sharing for Data-Heterogeneous Edge Devices}, 
      author={Hangyu Li and Hongyue Wu and Guodong Fan and Zhen Zhang and Shizhan Chen and Zhiyong Feng},
      year={2025},
      eprint={2506.20644},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2506.20644}, 
}