Vehicle-to-vehicle (V2V) communications can be extremely useful in improving the safety of all road users. Connected, Cooperative and Automated Mobility approaches, such as those promoted by the CCAM consortium, allow the data received from on-board vehicle sensors to be complemented with information from vehicles outside the range of the sensors, thus improving the perception of ADAS systems and their ability to react promptly to dangerous conditions and prevent damage.

All main standards for V2V communications (WAVE/DSRC, ETSI-ITS, C-V2X) require vehicles to transmit the current position, speed, acceleration, and the possible occurrence of emergency maneuvers (such as sudden braking or loss of grip). This information can be actively used by ADAS systems to slow down when a vehicle in front of us initiates emergency braking, even if this vehicle is out of range of our sensors or masked by other vehicles or obstacles that prevent direct perception.

This sounds great, but it also exposes ADAS systems to possible cyberattacks where malicious actors send false messages. What would happen if an attacker on the side of a highway used Software Defined Radio to send fake messages about non-existent vehicles pretending to be in front of us and initiating emergency braking? Would it be enough to block traffic on a major road artery? Or, could an attacker convince vehicles around him to “lead the way” simply by sending modified messages claiming to be an emergency vehicle proceeding with sirens blaring?

Some of these attack scenarios are mitigated by security countermeasures already included in the standards, which include the possibility of digitally signing messages, certifying their authenticity. However, an attacker could abuse a TCU (telecommunication unit) equipped with valid digital certificates to send V2V messages with incorrect information to nearby vehicles. For this reason it is important to envisage the use of systems for the analysis of V2V messages capable of identifying anomalies and potentially incorrect behavior.

Traditional detection approaches, such as the one proposed by the Framework For Misbehavior Detection (F2MD)1, rely on the analysis of received V2V messages. However, these systems can be deceived by sophisticated attackers, who create "plausible" message sequences, explicitly designed not to activate the mechanisms for detecting anomalies and impossible situations2. Such attacks can only be accurately detected using innovative approaches, which fuse and compare the data received from V2V communications, the data collected by on-vehicle sensors, and the data collected by sensors at the infrastructure level. The FRODDO project will take all these aspects into account, resulting in the design of safer and more secure V2V communications.