Leveraging Semi-supervised Learning for Enhancing Anomaly-based IDS in Automotive Ethernet

Intelligent Connected Vehicles (ICVs) rely on highly interconnected automotive components, with automotive Ethernet enabling high-bandwidth in-vehicle networking and facilitating the transmission of sensor data among electronic control units. However, the increasing connectivity and potential vulnerability inheritance in connected and autonomous vehicles expose them to security risks. To address this challenge, an anomaly-based intrusion detection system (termed AE-TW) is proposed in this paper, which focuses on attacks stemming from the automotive Ethernet on in-vehicle networks. We employ a semi-supervised machine learning method, AutoEncoder (AE) with time windowing, to train the normal profile for detecting anomalies. The proposed approach is implemented in a real-world vehicle testing environment. We evaluate the performance of the proposed intrusion detection system (IDS) using a synthetic dataset called EFA-IDS, which we generated, and the well-known TOW-IDS automotive Ethernet intrusion dataset. The experimental results demonstrate that our approach achieves high detection performance across different datasets and manifests low computation cost, making it highly applicable for real-time anomaly detection.