Location Hijacking Attack in Software-Defined Space-Air-Ground-Integrated Vehicular Network

Internet of Vehicles (IoV) is an emerging technology in automotive field, in which vehicles can communicate with other vehicles and roadside infrastructures to improve information acquisition ability as well as obtain various services to elevate the security and comfort level. To cope with the increasingly complex vehicular network, software-defined networking (SDN) architecture with advantages of centralized management and flexible control becomes a promising solution. However, in application scenarios, the security of SDN is rarely concerned. If attackers exploit the vulnerabilities of SDN to hijack the network location of the servers or vehicles, vehicles may not be able to access the services they need timely and effectively, which will pose a great threat to the benefit of vehicle users. In light of this, we focus on location hijacking attack against SDN in vehicular network. We perform this attack on five mainstream SDN controller platforms and analyse its impacts from multiple perspectives. As far as we know, this is the first study of such attack in vehicular network. Furthermore, using the advantages of the software-defined space-air-ground-integrated vehicular network and the characteristics of high altitude platform (HAP), such as wide coverage and high load capacity, we put forward the attack recovery scheme based on deep Q-learning (DQL) to supplement existing defence mechanisms that always have counter attacks and endow the vehicular network with a certain resilience.

Automated summary

Internet of Vehicles (IoV) is an emerging technology in the automotive field that allows vehicles to communicate with each other and roadside infrastructures to improve information acquisition and enhance security and comfort. Software-defined networking (SDN) architecture, with its centralized management and flexible control, offers a promising solution for the increasingly complex vehicular network. However, the security of SDN in application scenarios is often overlooked. This study focuses on the location hijacking attack against SDN in vehicular networks and proposes an attack recovery scheme based on deep Q-learning (DQL) to increase the network's resilience.