AI-Driven Intrusion Detection Framework for 5G and Beyond Networks

Abstract

The evolution of wireless networks into the fifth generation (5G) and beyond has enabled unprecedented connectivity, very low latency, and an enormous density of devices. As much as these advances are positive to the new applications of smart cities, autonomous vehicles, and industrial IoT, it has also introduced more attack surfaces and vulnerabilities. The dynamic, high-throughput and decentralized nature of the 5G network is almost impossible to be supported by the legacy-based intrusion detection system (IDS). As a result, AI-based intrusion detecting systems are increasingly being considered as the future of cybersecurity systems. The current paper is a review and analysis of AI in IDS 5G and above networks. It explains the implementation of machine learning (ML), deep learning (DL), and reinforcement learning (RL) into intrusion detection to make it more adaptable, scalable, and more effective at detecting intrusions. The paper recognizes the following frameworks: anomaly-based IDS, federated learning-based IDS to maintain privacy, and hybrid AI-IDS models, that incorporates the signature-based and behaviours-based detection. Furthermore, it discusses how AI-based IDS can be used to slice networks, software-defined networking (SDN), and edge computing-enabled 5G networks. The findings indicate that artificial intelligence-based IDS can detect zero-day attacks, distributed denial-of-service (DDoS) and advanced persistent threats significantly better than conventional systems. Nevertheless, interpretability of models, inaccessible training data, the cost of computation and AI adversarial attacks remain a problem. Explainable AI (XAI), sparse models to support edge devices, and cross-layer collaborative intrusion detection are other issues that need to be studied to guarantee safe and reliable next-generation networks. In this paper, we find that AI-powered intrusion detection systems are both practical and warranted in providing cybersecurity in 5G and beyond ecosystems, the mainstay of secure digital infrastructure.