Related Experiment Video
Updated: Jul 4, 2025
11:54
Microfluidic Platform with Multiplexed Electronic Detection for Spatial Tracking of Particles
Published on: March 13, 2017
9.3K
Detection, differentiation and localization of replay attack and false data injection attack based on random matrix.
Yuehao Shen1, Zhijun Qin2,3
1Guangxi Key Laboratory of Power System Optimization and Energy Technology, Guangxi University, No. 100, Daxue Road, Xixiangtang District, Nanning, 540003, Guangxi, People's Republic of China.
Scientific Reports
|February 2, 2024
Summary
This study introduces a novel method to detect and locate hybrid cyber-attacks in power systems. The approach uses random matrix theory and a SVD-CNN classifier to differentiate false data injection attacks from replay attacks, enhancing grid security.
Area of Science:
- Cybersecurity
- Power Systems Engineering
- Data Science
Background:
- Supervisory Control and Data Acquisition (SCADA) systems are vulnerable to cyber-attacks like replay attacks and false data injection attacks (FDIA).
- These attacks falsify meter measurements, disrupting power system operations.
- Existing defense mechanisms may not effectively address hybrid attacks combining both attack types.
Purpose of the Study:
- To develop a systematic methodology for defending against hybrid cyber-attacks in power systems.
- To propose a detection method using random matrix theory to identify hybrid attacks and distinguish FDIA from replay attacks.
- To localize falsified measurements with improved accuracy.
Main Methods:
- Short-term load and renewable power generation forecasting to obtain predicted measurements.
- Calculation of random variables by comparing forecasted and real-time measurements.
- Construction of a random matrix and analysis of its eigenvalue statistics using a sliding time window for hybrid attack detection.
- Development of a novel multi-label classifier (SVD-CNN) combining SVD decomposition, eigenvalue analysis, and convolutional neural networks to distinguish and localize FDIA.
Main Results:
- The proposed random matrix theory-based method effectively detects hybrid attacks by analyzing changes in eigenvalue statistics.
- The SVD-CNN classifier demonstrates high accuracy in distinguishing FDIA from replay attacks.
- The method shows strong detection capabilities, effectively filtering measurement noise.
- Simulations on IEEE 14-bus and IEEE 57-bus systems validate the proposed approach's performance.
Conclusions:
- The developed methodology provides a robust defense against hybrid cyber-attacks in SCADA systems.
- The SVD-CNN approach significantly enhances the accuracy of FDIA localization.
- This research contributes to improving the security and reliability of modern power grids against sophisticated cyber threats.
