A reliable cyber-attack detection architecture for cyber-physical systems in SDN-enabled internet of vehicles

Abstract Today, the Internet of Vehicles (IoV) is a new and very attractive technology in automotive communications. IoV networks can combine Internet-connected devices to store, process, and analyze real-time data in intelligent transportation. However, detecting cyberattacks in this environment is inevitable and remains a major challenge, as malicious threats can disrupt vehicle communications, leading to network congestion and safety risks. To enhance security in IoV networks, software-defined networking (SDN) provides a centralized and flexible framework for managing traffic flow and implementing security measures. In this study, we propose a novel Intrusion Detection System (IDS) for SDN-enabled IoV environments. Our proposed Genetic Algorithm-Ensemble Bagging Trees (GA-EBT) hybrid model employs the Message Queuing Telemetry Transport (MQTT) protocol for secure data transmission and integrates a hybrid machine-learning model to predict and detect cyber threats in IoV networks. Using the IoT_SDN-IDS and MQTT-IoT-IDS2020 datasets, we complete a comprehensive case study to evaluate various machine learning algorithms. Our findings indicate that our hybrid GA-EBT model performs more effectively than previous models. Simulation results show accuracy up to 99.9931% and 99.997% on the IoT_SDN-IDS and MQTT-IoT-IDS2020 datasets, respectively. The results prove that our hybrid SDN-based cyber-attack detection model effectively detects cyber-attack threats in IoV environments. Moreover, the proposed GA-EBT model provides secure data interaction and improves vehicular communication reliability.