Author: Christina Malliou
Connected Autonomous Vehicles (CAVs) rely heavily on data received from their environment and their sensors. The information from the traffic lights or the roadside units (RSUs) allows to handle traffic scenarios, adapt routes, avoid congestion and enhance traffic efficiency or even reduce CO2 emission. Through their sensors, CAVs can identify objects, pedestrians, or other vehicles and determine their next step of action under any circumstances and weather conditions. The communication capabilities between vehicles (V2V) allow them to react in real-time and in many cases to provide the desired responses to dangerous situations long before their users would.
Any abnormalities or cyber-attacks in those components might lead to accidents with disastrous consequences for the users, the passengers of other vehicles, or pedestrians. Although those attacks have a severe impact, the nature of the CAVs does not allow easy test and validate real-world scenarios that will include traffic or road infrastructure. Those tests might be too expensive or too difficult to implement, with high risk, or in some cases even prohibited by law. As a consequence, anomalies in autonomous vehicles might not be detected as they are not included in easily accessible data sets for training. However, to increase the public's trust and the acceptance of CAVs, security should be undisputed. Contrary to the real-world tests, tests could be implemented with low cost and with low risk in a simulated environment.
A lot of simulators have been proposed and developed to simulate the behavior of CAVs. However, the vast majority of those simulators are only able to represent a single aspect of an autonomous vehicle, whether this is the communication or the sensors. Even when simulators can capture more than one aspect of CAVs, they are not able to simulate both the normal behavior and the behavior under attack without any modifications by the user.
A complex simulation tool should be able to simulate intricate infrastructure, with various traffic scenarios and car-following models, to consider the imperfection of the drivers and non-driving areas, which might block the communication between CAVs and infrastructure such as parks. In addition to this, to execute realistic scenarios that include CAVs, a configurable set of sensors is paramount. The set of sensors along with a realistic representation of the vehicle and its components will provide the necessary flexibility to the simulation tool allowing its use for training authorized personnel and cybersecurity officers. A crucial aspect of any simulation that includes CAVs is the communication between the vehicles and the available infrastructure in an extensible and modular way that will incorporate at the same time different communication protocols and support various platforms.
To achieve the above, different simulators have to be combined and integrated. A vehicle simulator for the sensors and the physical representation of the vehicle, a traffic simulator with the realistic approach in complex traffic phenomena, and a network simulator, flexible enough to adapt to any changes in the infrastructure and the communication protocols. It is worth mentioning, that even after the challenging integration process of the abovementioned simulators, only the normal operation of the CAVs is simulated. To achieve the simulation of cyber-attacks, further, development is needed to ensure the realization of the different attack scenarios.
Figure 1: Integrating different simulators to simulate CAVs
As nIoVe project aims to provide innovative solutions for a secure future of the IoV ecosystem, complex tools that will be able to simulate the demanding aspects of CAVs and help mitigate or even prevent cyber attacks are being implemented.
 CARLA. (December 22, 2020) Open-source simulator for autonomous driving research Retrieved June 6, 2021, from https://carla.org
 Eclipse SUMO. (April 6, 2021). Simulation of Urban Mobility. Retrieved June 5, 2021, from https://www.eclipse.org/sumo/
 OMNET++. (January 22, 2021). (OMNET++) Retrieved June 7, 2021, from https://omnetpp.org/