Security threat assessment methods for unmanned vehicles in a smart city

This paper presents a comprehensive cybersecurity model for autonomous vehicles (AV) system in framework of the smart city. We consider a problem of information flow description for a real AV system in the frame of discretional security models. The definition of safe data transfer for some types of information exchange between the system assets is presented. Some techniques for asset classification in frame of Biba and Bell-LaPadula models taking into account information links between the assets are introduced. We discuss the complexity and ambiguousness of the «integrity» definitions in relation to cybersecurity provision and emphasize the significance of its extension to the equipment and data processing methods. This approach is considered to shift the focus to the system analysis and system vulnerability removal and to develop the protection upon the system internal properties. Dynamical methods of cybersecurity architecture synthesis for transport systems with AVs are considered. The methods are shown to lessen computational complexity of system modelling. Instead of analyzing the complete security graph, one may analyze an induced subgraph for a given AV. The vertices (assets) of full security graph are excluded from the consideration if they are at a great «distance» from the analyzed AV. Formulae for the distance calculation are provided for some cases. Some examples of cybersecurity architecture analysis and synthesis for autonomous vehicles are presented.