Research on Encrypted Malicious 5G Access Network Traffic Identification Based on Deep Learning

The 5G small stations are widely deployed to improve the capacity of the 5G communication system. However, increasing malicious attacks are hidden in the traffic of small stations. Therefore, it is of great significance to identify the encrypted malicious traffic in the 5G access network. However, the traffic transferred through the backhaul of 5G small base station is usually encrypted. To this end, a deep learning-based method to identify the signaling hijacking traffic on the access network is proposed. Firstly, a 5G signaling hijacking system is developed to address the vulnerabilities of small stations and generate practical malicious traffic. To identify encrypted malicious traffic from 5G backhaul links, a 1D-CNN recognition model based on data packets is constructed. Finally, the 1D-CNN model is tested and validated multiple dimensions. The extensive experiment results reveal that the proposed method can achieve a recognition accuracy of over 99.95%.