Cooperative and Decentralized Defense in Autonomous Vehicle Networks

Thumbnail Image
Anwar, Anika
Autonomous Vehicle Security
In the era of the transportation revolution, the emergence of Autonomous Vehicle Networks (AVNs) represents a remarkable feat of technological sophistication. It embodies a seamlessly interconnected system of Autonomous Vehicles (AVs) that operate with unparalleled precision, exhibit cohesive communication capabilities, and engage in effortless collaboration. This network has the potential to revolutionize transportation efficiency and elevate safety standards through its diverse array of services. These services aim to optimize traffic flow, minimize the occurrence of accidents, and significantly enhance passenger convenience. The accuracy and reliability of the data sensed and transmitted by AVs are essential for the effective functioning of an AVN. However, the integrity of vehicle communications is susceptible to various cyber threats, including spoofing attacks. Traditional centralized security solutions based on cryptography and authentication are insufficient in providing the necessary scalability and adaptability to effectively protect AVNs due to their extensive and dynamic nature. Moreover, the expanding scale of future AVNs necessitates an innovative approach to ensure the authenticity of traffic information exchanged within the network. This thesis offers a cooperative and decentralized Defense-in-Depth (DiD) framework integrating game theory for securing inter-vehicular communications within AVNs. This framework incorporates three defense parts, including prevention, detection, and mitigation, to ensure comprehensive protection against attacks. We propose a novel decentralized threat prevention system as the first part of the DiD framework, aiming to enhance security within AVNs. The system is specifically designed to dynamically assess risks in real-time and facilitate secure message routing in Vehicle-to-Vehicle (V2V) communications. Next, we present a decentralized Cooperative Intrusion Detection Network (CIDN) to facilitate the efficient sharing of intrusion information among AVs and improve the overall defense capabilities. The CIDN promotes collaboration both within and between vehicle groups to enhance scalability within AVNs, enabling large-scale deployments of AVs without compromising security. Finally, the mitigation of data integrity attacks is addressed through a trust-based vehicle collaboration model incorporating a secure data aggregation procedure. This model minimizes the impact of data integrity attacks by leveraging cooperative strategies and establishing trust among participating vehicles. By integrating this decentralized DiD framework, AVNs can enhance their resilience against security attacks, ensuring a safer and more reliable autonomous transportation ecosystem. The thesis experimentally demonstrates the effectiveness of the proposed techniques in strengthening the overall resilience of AVNs while achieving scalability.
External DOI