Edge-Based AI for Real-Time Threat Detection in 5G-IoT Networks: A Comparative and Architectural Review
Alexandre Sousa, Luís Correia, Manuel J. C. S. Reis · 2025
The convergence of 5G and the Internet of Things (IoT) has created unprecedented opportunities for ultra-low latency, high-throughput, and massive device connectivity. However, this transformation introduces complex cybersecurity challenges, especially in scenarios where real-time threat detection is critical. Traditional cloud-based solutions are increasingly unsuitable due to latency constraints, bandwidth overhead, and privacy concerns. This paper investigates the role of edge-based Artificial Intelligence (AI) as a decentralized, responsive, and scalable solution for real-time anomaly detection in 5G-enabled IoT networks. We present a critical analysis of recent advancements in lightweight AI models deployed at the network edge, including federated learning, behavior-based intrusion detection, and AI-enhanced threat analytics. The paper evaluates these approaches in terms of latency, detection accuracy, computational efficiency, and adaptability to resource-constrained environments. Furthermore, we discuss key implementation challenges such as model drift, secure model updates, and hardware limitations. Our findings highlight the potential of edge-AI to fortify the 5G-IoT ecosystem, while underscoring the need for continued research to bridge existing performance and scalability gaps.