Federated Learning-Based Intrusion Detection for Vehicular Networks Using Robust LSTM

In recent years, advancements in the Internet of Things (IoT) and its integration into vehicles have given rise to a promising new paradigm: the Internet of Vehicles (IoV). The IoV enables the sharing of information among vehicles through onboard communication technologies, thereby enhancing both traffic safety and efficiency. However, due to its openness and highly dynamic nature, the IoV is particularly vulnerable to network attacks. Existing intrusion detection systems typically rely on centralized models, which are prone to issues such as data silos, privacy leakage, and computational bottlenecks. Therefore, it is especially important to develop advanced intrusion detection technologies to address these threats.Federated learning, as a distributed machine learning approach, allows collaborative training without sharing raw data, offering an efficient and secure intrusion detection mechanism for the IoV. In this paper, we propose an IoV intrusion detection method based on federated learning. Our approach leverages an optimized LSTM model and incorporates the latest FedAware aggregation algorithm to address the challenge of non-independent and identically distributed (non-IID) data across vehicle nodes, thus improving detection accuracy and efficiency. Experimental results on the publicly available VeReMi Extension IoV intrusion detection dataset validate the effectiveness of the proposed method.