Enhanced Edge Data Security Scheme for Smart Grids Based on Federated Learning and Blockchain

Abstract

The authors propose a data security enhancement scheme for smart grids based on Federated Learning and Blockchain to address issues of data privacy leakage and tampering in edge computing environments within smart grids. In edge computing scenarios, traditional centralized model training is vulnerable to attacks and struggles to protect data privacy, with an attack success rate as high as 54.3%. By integrating Federated Learning and Blockchain technologies, the authors have constructed a secure architecture that enables the system to protect users’ electricity consumption data privacy without sharing raw data, while ensuring the integrity and security of the data transmission process. Experimental results show that the privacy leakage rate of the Federated Learning+Blockchain scheme is only 0.9%, compared to 60.2% for schemes without privacy protection. Additionally, the Federated Learning+Blockchain approach outperforms traditional solutions in terms of data transmission, processing time, Central Processing Unit (CPU) and memory consumption, and security, achieving a balance between performance and security.

Cite this article as: J. Wang, J. Xu, J. Li, C. Xu and H. Xie, "Enhanced edge data security scheme for smart grids based on federated learning and blockchain," Electrica 25, 0102, 2025. doi: 10.5152/electrica.2025.25102.