Explainable-AI for DoS Attacks Detection in 5G Network Using Deep Learning Models
Amjad Albashayreh, Yahya M. Tashtoush, Abdallah Aldosary, Omar Darwish, Firas Ali Albalas · 2024
With the emergence of the fifth-generation (5G) net-work, numerous revolutionary applications are enabled, including low-latency and machine-type communications. This great in-crease creates a broader security threat concerns, such as denial-of-service (DoS) attacks, which can disrupt network functionality. The complexity and decentralization of 5G networks create new vulnerabilities for adversaries, necessitating comprehensive security procedures to identify, mitigate, and prevent DoS attacks in 5G networks. This paper introduces a novel approach for DoS detection in 5G networks, utilizing deep learning and machine learning models, along with Local Interpretable Model-Agnostic Explanations (LIME), to interpret model predictions and identify the significant role of data features in detecting DoS attacks. The results revealed that the random forest model demonstrated superior recall of 99.98, while BiLSTM demonstrated exceptional performance with a recall of 98.02.