GridWatch: a smart network for smart grid

Conference item

The adoption of decentralized energy market models facilitates the exchange of surplus power among local nodes in peer-to-peer settings. However, decentralized energy transactions within untrusted and non-transparent energy markets in modern Smart Grids expose vulnerabilities and are susceptible to attacks. One such attack is the False Data Injection Attack, where malicious entities intentionally inject misleading information into the system. To address this threat, this paper proposes GridWatch, an effective real-time in-network intelligent framework to detect false data injection attacks. Gridwatch operates in a hybrid model. It deploys inference model in the programmable network devices and also on the server to detect false data injection attacks. GridWatch was evaluated using a real-world dataset from Austin, Texas, and can detect false data injection attacks with 94.8% accuracy. GridWatch on average performs 4 billions transactions per second in less than 1.8 microsecond latency.

Authors: Hemmatpour, M; Zheng, C; Zilberman, N; Ha, PH

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: IEEE
• Pages: 141-146
• Publication date: 2024-11-04
• Acceptance date: 2024-07-09
• Event title: International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm 2024)
• Event location: Oslo, Norway
• Event website: https://sgc2024.ieee-smartgridcomm.org/
• Event start date: 2024-09-17
• Event end date: 2024-09-20
• DOI: 10.1109/SmartGridComm60555.2024.10738103
• EISSN: 2474-2902
• ISSN: 2373-6836
• Language: English
• Keywords: False Data Injection Attack (FDIA); Peer-to-Peer (P2P) trading; security; Smart Grid; in-network machine learning