Attention-CNN-LSTM based intrusion detection system (ACL-IDS) for in-vehicle networks

Modern vehicles rely on electronic control units (ECUs) communicating through the controller area network (CAN) bus protocol. However, increased connectivity through Wi-Fi, Bluetooth, and onboard diagnostics (OBD) ports has heightened cybersecurity risks due to the CAN bus protocol’s inherent security vulnerabilities. Addressing this challenge requires an in-vehicle intrusion detection system (IDS) with high accuracy and minimal false alarms. Existing IDSs for in-vehicle networks often fall short due to inadequate extraction of network traffic features’ dependencies in a time-series context.To overcome this limitation, this study presents ACL-IDS, a hybrid intrusion detection system for in-vehicle network traffic. Leveraging deep learning techniques like CNN, LSTM, and attention mechanisms, ACL-IDS effectively captures short-term and long-term dependencies within network traffic, enhancing intrusion detection accuracy. Extensive experiments on a real benchmark dataset demonstrate ACL-IDS’s superior performance compared to individual, ensemble, and state-of-the-art methods. With detection accuracy reaching up to 99%, ACL-IDS emerges as a robust solution for analyzing and detecting intrusions in in-vehicle network traffic.