Leveraging LSTM and GRU-based deep neural coordination in intelligent transportation to strengthen security in the Internet of Vehicles

• Published in: INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS
• Main Authors: Yanmin, Cai; Sarkar, Arindam; Zain, Jasni Mohamad; Bhar, Arindam; Noorwali, Abdulfattah; Othman, Kamal M.
• Format: Article; Early Access
• Language: English
• Published: SPRINGER HEIDELBERG 2024
• Subjects: Computer Science
• Online Access: https://www-webofscience-com.uitm.idm.oclc.org/wos/woscc/full-record/WOS:001329069500001
• ISSN: 1868-8071; 1868-808X
• DOI: 10.1007/s13042-024-02401-2

Concerns about security and safety are escalating as additional smart cars become part of the Internet of Vehicles (IoV) and link up to the Internet via Internet of Things (IoT) applications. The rising amount of connected vehicles brings about intricate cybersecurity issues, such as unauthorized entry, data breaches, and potential vehicle misuse. Although research has delved into utilizing Machine Learning (ML) techniques for Intrusion Detection (ID) in IoT networks to avoid vehicle accidents and cyberattacks, there are still major hurdles in accurately detecting unauthorized access and attaining real-time effectiveness. This study tackles these obstacles by suggesting a new Intrusion Detection System (IDS) for the Internet of Vehicles (IoV), utilizing the collective capabilities of Deep Learning (DL). This research addresses existing shortcomings by employing GRU and LSTM models to capture short- and long-term dependencies in attack patterns, improving system detection accuracy and responsiveness. In addition, we present a secure key exchange technique for the IoT network using Artificial Neural Networks (ANNs) to enable mutual learning and enhance key distribution among connected devices, ensuring a strong defense against unauthorized access and data breaches. The proposed strategy's effectiveness is proven by intense training and testing of DL models on various datasets like CSE-CIC-IDS 2018, CI-CIDS 2017, CIC DoS, and a specialized vehicle exploitation dataset, ensuring generalizability and robustness. The tests demonstrate that the IDS has exceptional precision, detecting 99.9% of network attacks and 95.9% of car hacks. Additional evaluation using recall, F1-score, and accuracy measures affirms the efficiency and trustworthiness of the suggested approach. This study enhances IoT security with a novel ANN-based key exchange method and proposes a comprehensive IDS system for IoV using advanced DL methods. The proposed research makes significant progress in network security and intrusion detection accuracy, greatly improving the safety and security of smart cars and the Internet of Vehicles ecosystem, leading to a safer and more secure future in intelligent transportation systems.