Enhancing Industrial Protocol Security with GAN-based Fuzz Testing and Self-Attention Mechanism

This study introduces a novel deep learning framework for identifying vulnerabilities in Industrial Control Protocols (ICPs) through intelligent fuzz testing. Leveraging a Wasserstein GAN with Gradient Penalty (WGAN-GP) and self-attention mechanisms, the proposed model generates realistic yet diverse protocol message sequences to effectively detect potential security threats. By bypassing the need for manual protocol specifications, the approach improves test case generation and accelerates anomaly detection. Extensive evaluations on MQTT and Modbus- TCP protocols demonstrate superior performance in both accuracy and vulnerability detection compared to traditional and deep learning-based fuzzing methods.