A Simulation-based Methodology to Assess the Impact of Jamming Attacks on Interconnected Automated Road Vehicles

Licentiate thesis, 2024

This thesis addresses security benchmarking of Cooperative Driving Automation (CDA) applications, focusing on simulation-based assessment of the consequences of jamming attacks. CDA systems are expected to improve the safety, fuel efficiency, and passenger comfort of future road vehicles. These systems rely on data received wirelessly from other vehicles and roadside installations and must, therefore, be resilient to attacks conducted via the wireless channel.
We propose a framework for benchmarking the resilience of CDA applications against various types of jamming attacks through simulations. To this
end, we have developed a simulation engine for communication-based fault and attack injection experiments called ComFASE, which utilizes four existing
simulators: Plexe, Veins, OMNet++, and SUMO.
We illustrate our benchmarking approach by conducting a series of simulations where we study the impact of different types of jamming attacks on a
longitudinal control algorithm for platooning, which is provided in the Plexe framework. We propose and investigate simulation models for five types of
jamming attacks: delay attack, denial-of-service (DoS) attack, barrage jamming, deceptive jamming, and destructive interference. We implement these models in a simulation model of the physical layer of the IEEE 802.11p standard provided
in Veins.
We emphasize that the work presented in this thesis constitutes a first step towards defining a framework for security benchmarking of CDA applications.
Such a framework must consider various aspects related to system design, environmental conditions, attack types, and system use cases. We address only a few of these aspects, specifically for jamming attacks: the driving scenario, the attack model, and the attack model parameters.
Our attack models are based on three primary parameters: attack duration, attack start-time, and attack value. The first two are defined in relation to
the time axis of the driving scenario, while the attack value determines the nature or strength of the attack. Our results show that a significant number
of the simulated attacks caused collisions among the vehicles in the platoon. They also show that the outcome of the simulations is highly dependent on the driving scenario, the attack types, and the attack model parameters.