Malware Identification Method in Industrial Control Systems Based on Opcode2vec and CVAE-GAN.

Area of Science:

• Cybersecurity
• Machine Learning
• Industrial Control Systems

Background:

• Industrial Control Systems (ICSs) are increasingly vulnerable to malware due to internet integration.
• Existing machine learning (ML) malware identification methods perform suboptimally in ICS environments.

Purpose of the Study:

• To propose an innovative ML-based method for enhanced malware identification specifically tailored for ICS environments.
• To improve the stability and robustness of malware classifiers in ICS.

Main Methods:

• Integration of opcode2vec with preprocessed features.
• Utilizing a conditional variational autoencoder-generative adversarial network (CVAE-GAN).
• Employing Convolutional Neural Networks (CNNs) for malware classification.

Main Results:

• Achieved high performance metrics: 97.30% accuracy, 92.34% precision, 97.44% recall, and 94.82% F1-score.
• Demonstrated improved stability and robustness of malware classifiers in ICS.
• Validated the efficacy of the proposed method through extensive experiments.

Conclusions:

• The proposed method significantly enhances malware identification performance in ICS.
• The tailored approach offers a robust solution for ICS cybersecurity challenges.
• The achieved results represent the highest reported values in the experimental context.