An optimized anomaly detection framework in industrial control systems through grey wolf optimizer and autoencoder...

Area of Science:

• Cybersecurity
• Artificial Intelligence
• Industrial Control Systems

Background:

• Reliable Internet connectivity is crucial for Industrial Control Systems (ICS) real-time monitoring and anomaly detection.
• Current anomaly detection methods in ICS face challenges like high computational complexity, dataset limitations, and high false-positive rates.

Purpose of the Study:

• To develop a novel collaborative data processing framework for enhanced anomaly detection in ICS.
• To integrate and optimize the Grey Wolf Optimizer (GWO) with Autoencoders (AE) for improved performance.

Main Methods:

• The proposed approach optimizes GWO through enhanced prey selection, encircling, and initial population generation.
• Autoencoder (AE) dropout functionality is improved for better model generalization.
• The framework employs a two-stage process: GWO for feature selection and AE for anomaly detection.

Main Results:

• Experimental validation on SWaT and WADI datasets showed superior performance compared to existing methods.
• Significant improvements were observed in accuracy, precision, recall, and F1-score.
• The model effectively identifies relevant features and reduces feature errors.

Conclusions:

• The proposed GWO-AE framework demonstrates significant potential in addressing limitations of current ICS anomaly detection systems.
• The approach offers a more accurate and reliable solution for real-time monitoring and anomaly detection in ICS environments.