Funded through the Graduate School in Computer Science (CUGS)
Involved persons: Felipe Boeira (PhD student), Mikael Asplund (PI)
Advanced driver assistance systems (ADAS) are becoming increasingly sophisticated and connected. Emerging applications include vehicular platoons, collision avoidance, and emergency vehicle awareness. The next generation of vehicular applications need to consider a holistic perspective on traffic flows in order to meet the increasing demands on sustainable transport and user safety and comfort. However, such coordinated mobility between vehicles requires robust and secure communication primitives.
Current vehicular communication standards provide some support for security mechanisms such as the ability to sign messages. However, the increasing amount of software and network interfaces of modern vehicles make them potentially vulnerable to malware and other forms of cyberattacks. Therefore, establishing real trust between vehicles remains an open problem. This is a critical issue since the safety of the participating vehicles might depend on the accuracy and truthfulness of information from other entities.
Chen et al. [1] presented an example of what can occur if the design of connected vehicular applications has not sufficiently accounted for the possibility of false information. The authors analyse the Intelligent Traffic Signal System (I-SIG), which is currently being launched as a pilot application by the US department of transportation. The authors show that by spoofing messages, a single attacker is able to disturb the signalling control algorithm, resulting in massive traffic jams. Moreover, our work (Boiera et al. 2017) demonstrates that the ability to create fake identities coupled with false message contents can cause severe collisions for vehicular platooning scenarios.
Common for these works is that they consider attacks that use incorrect information regarding speed and location of one or multiple vehicles in the area. Counteracting such attacks require some form of location verification. There is a wide range of works that try to tackle this long-standing problem using various assumptions. For example, the work by Yan et al. [2] use antenna arrays to verify that claimed locations are consistent with the angle at which signals arrive to a base station. However, given the inherent uncertainty associated with radio propagation and the potentially complex attack scenarios, collaborative algorithms are needed to differentiate false alarms from real attacks. Existing work for collaborative location verification such as that by Zhu and Cao [3] are not adapted for the high-speed scenarios associated with vehicular applications.
The purpose of this project is to investigate secure communication mechanisms that allow trust to be established between vehicles. Examples of such mechanism are the verification of the physical location of nearby vehicles and also the sharing of that information within the group to enable trust propagation. Key challenges that will be considered include limited and congestion-prone communication channels, lack of ubiquitous communication infrastructure, and the possibility of reputation attacks from malicious entities.
The first steps to analyse the problem at hand is to understanding the potential impact of position falsification attacks in platooning applications. The immediate next phase is to establish mechanisms that are able to detect and mitigate such location falsification attacks. This requires making use of a combination of input data such as local sensors, infrastructure capabilities (e.g., the localisation capabilities provided by 5G technology), and the sharing of that information among the vehicles.
Moreover, much more work is needed to understand the performance and impact of such security mechanisms on the communication system and its ability to enhance vehicular safety. It is clear that cryptographic primitives will have an important role to play to provide trustworthy communication between vehicles. The project will investigate the trade-off between the security and the frequency and range of messages that can be exchanged between vehicles (especially in dense scenarios) due to the extra overhead induced by these methods.
The mechanisms developed in the project will be implemented and evaluated using simulation-based tools for vehicular communication. Moreover, fully understanding how groups of vehicles can establish mutually verifiable information requires formulating suitable theoretical modelling frameworks for this purpose.