A Framework for Detecting False Data Injection Attacks in Large-Scale Wireless Sensor Networks

Area of Science:

• Computer Science
• Electrical Engineering
• Cybersecurity

Background:

• False data injection attacks (FDIAs) pose significant risks in sensor networks by corrupting data and leading to incorrect decisions.
• The increasing scale of sensor networks amplifies the challenge of detecting these sophisticated attacks.

Purpose of the Study:

• To propose a novel framework for the distributed detection of FDIAs in large-scale sensor networks.
• To enhance the security and reliability of sensor network data.

Main Methods:

• Extracting spatiotemporal correlation information from sensor data to group sensors.
• Utilizing autoregressive integrated moving average (ARIMA) models to capture temporal correlations within groups.
• Establishing a consistency criterion to identify anomalous sensor nodes indicative of FDIAs.

Main Results:

• The proposed framework effectively categorizes sensors and identifies abnormal nodes.
• Validation using a U.S. smart grid dataset demonstrated the framework's capability against both simple and stealthy FDIAs.
• The method shows promise in detecting FDIAs in complex, large-scale environments.

Conclusions:

• The developed distributed detection framework is effective for identifying FDIAs in large-scale sensor networks.
• The approach leverages spatiotemporal correlations and time-series analysis for robust anomaly detection.
• This research contributes to securing critical infrastructure reliant on sensor networks.