DL-SecureNet: A Transformer-based Cybersecurity Framework for Threat Detection in 5G and Beyond

The development of the 5G networks has given a new direction for the faster communication but it has brought new security threats related to its efficacy. In the presented work, DL-SecureNet, a deep learning transformer-based approach is introduced that aim at identifying cybersecurity threats in 5G networks. Based on the CIC-DDoS dataset, the model was developed and tested with different 5G attributes such as packet size, packet delay, protocol, and transfer rate. However, DL-SecureNet uses positional encoding and multi-head attention to model temporal and contextual relationships of packets. Evaluation of the model produced satisfactory results and granted an overall accuracy of 98.4% with precision of 98.5%, recall 98.3%, and F1 score of 98.4% concerning numerous types of attacks such as DDoS, Botnet, SQL Injection, and Malware Communication. The comparison analysis showed that DL-SecureNet was superior to those standard algorithms like CNN, RNN, LSTM, and XGBoost. Also, it is augmented with threat intelligence mapping in real-time using CVE and MITRE ATT&CK frameworks. They demonstrate that DL-SecureNet can be used to implement a proactive network security model for 5G.