Proactive Threat Hunting and Anomaly Detection in Intelligent Green Transportation: A Federated Edge-AI Framework for In-Vehicle Consumer Networks

Connected consumer vehicles are central to next-generation intelligent transportation systems, yet face significant cybersecurity threats targeting their internal networks. Traditional cloud-centric security architectures face latency and privacy risks that are incompatible with the real-time, safety-critical requirements of consumer vehicles. This study proposes IDIV-FL, a Federated Edge-AI framework for proactive threat hunting and anomaly detection in in-vehicle networks. A Convolutional Neural Network (CNN) is utilized at a central server, distributed to consumer vehicles, and iteratively trained on local data. Consumer vehicles share only updated weights, which are aggregated using the FedAvg algorithm to form a global model. Using the CICIoV2024 dataset (in binary, decimal, and hexadecimal formats), IDIV-FL achieved near-perfect accuracy on binary data, attributed to its simple 0/1 representation, and up to 99.6% attack detection on decimal and hexadecimal data, with sub-category classification accuracy reaching 99.01% (decimal) and 99.16% (hexadecimal). This demonstrates a scalable, low-latency solution for privacy-preserving intrusion detection in connected consumer vehicles.