A Multi-source Log Big Data Detection System for Mining Ransomware Attack Chains in Power Grid Dispatch Systems via Integrated Graph Neural Networks and Federated Learning

View at publisher Summarise with VIRAT
Authors: Zhaoxuan Zhang, Yangguang Dong, Zixi Feng, Bingchen Liu, Xianjin Xu, Yangfei Li, Yubao Li, Yubo Hei
Cited by: 0
References: 15

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This study constructs a ransomware detection model for power grid dispatch systems by integrating graph neural networks (GCNs) with federated learning. Building upon traditional GCN architectures, it incorporates federated aggregation and differential privacy mechanisms to achieve structured modeling and secure training of distributed multi-source logs. Experimental results demonstrate that the model achieves 95.8% accuracy on both UNSW-NB15 and real-world power grid log datasets, representing a 3.5% improvement over traditional GCNs while reducing communication latency by 18.6%. This approach significantly enhances attack chain identification capabilities while ensuring privacy protection.

Fields of Study & Subjects

Network Security and Intrusion Detection Advanced Malware Detection Techniques Privacy-Preserving Technologies in Data

Publication details

Journal / source
IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
Pages
1187–1190
Publication date
28 January 2026
Language
EN