Malicious Traffic Classification via Edge Intelligence in IIoT

The proliferation of smart devices in the 5G era industrial IoT (IIoT) produces significant traffic data, some which is encrypted malicious traffic, creating a problem for detection. Malicious classification one most efficient techniques detecting traffic. Although it labor-intensive and time-consuming process to gather large labeled datasets, majority prior studies on use supervised learning approaches provide decent results when substantial quantity data available. This paper proposes semi-supervised approach classifying IIoT utilizes encoder–decoder model framework classify even with limited amount We sample normalize during data-processing stage. In model-building stage, we first pre-train unlabeled dataset. Subsequently, transfer learned weights new model, then retrained using small also offer an edge intelligence that considers aspects such as computation latency, transmission privacy protection improve model’s performance. To achieve lowest total latency reduce risk leakage, create privacy-protection models each local, edge, cloud. Then, optimize overall level. study classification, experimental demonstrate our method reduces training time 97.55% accuracy; moreover, boosts factor.