Skip to Main content Skip to Navigation
Conference papers

Toward Resilient Smart Grid Communications Using Distributed SDN with ML-Based Anomaly Detection

Abstract : Next generation “Smart” systems, including cyber-physical systems like smart grid and Internet-of-Things, integrate control, communication and computation to achieve stability, efficiency and robustness of physical processes. While a great amount of research has gone towards building these systems, security in the form of resilient and fault-tolerant communications for smart grid systems is still immature. In this paper, we propose a hybrid, distributed and decentralized (HDD) SDN architecture for resilient Smart Systems. It provides a redundant controller design for fault-tolerance and fail-over operation, as well as parallel execution of multiple anomaly detection algorithms. Using the k-means clustering algorithm from the machine learning literature, it is shown that k-means can be used to produce a high accuracy (96.9%) of identifying anomalies within normal traffic. Furthermore, incremental k-means produces a slightly lower accuracy (95.6%) but demonstrated an increased speed with respect to k-means and fewer CPU and memory resources needed, indicating a possibility for scaling the system to much larger networks.
Complete list of metadata

Cited literature [13 references]  Display  Hide  Download
Contributor : Hal Ifip Connect in order to contact the contributor
Submitted on : Friday, August 23, 2019 - 11:16:00 AM
Last modification on : Friday, May 21, 2021 - 6:12:02 PM
Long-term archiving on: : Saturday, January 11, 2020 - 1:23:17 AM


Files produced by the author(s)


Distributed under a Creative Commons Attribution 4.0 International License



Allen Starke, Janise Mcnair, Rodrigo Trevizan, Arturo Bretas, Joshua Peeples, et al.. Toward Resilient Smart Grid Communications Using Distributed SDN with ML-Based Anomaly Detection. International Conference on Wired/Wireless Internet Communication (WWIC), Jun 2018, Boston, MA, United States. pp.83-94, ⟨10.1007/978-3-030-02931-9_7⟩. ⟨hal-02269741⟩



Record views


Files downloads