Hybrid Edge Intelligence for Real-Time Intrusion Detection in Advanced Traffic Management Systems

Advanced Traffic Management Systems (ATMS) are critical to urban mobility. Still, they are increasingly vulnerable to cyberattacks due to their reliance on interconnected Internet of Things (IoT) and Vehicle-to-Everything (V2X) infrastructure. Existing intrusion detection systems (IDS) struggle with real-time performance, data scarcity, and resource constraints on edge devices. This paper proposes a hybrid framework that combines rule-based filtering with machine learning powered by transfer learning (TL) optimized for edge deployment. Our framework aims to achieve 95% accuracy and <20ms latency on a Raspberry Pi 4, bridging deterministic rules and adaptive Artificial Intelligence (AI) for secure, explainable, and resource-efficient cybersecurity in ATMS.