The development of self-driving vehicle networks that collaborate and communicate with each other or infrastructure to make decisions is an exciting advancement in the field of transportation. However, a recent study led by the University of Michigan has highlighted the vulnerability of these networks to data fabrication attacks. These attacks could potentially lead to serious safety risks for passengers and other drivers on the road.
The study, presented at the 33rd USENIX Security Symposium, focused on the concept of collaborative perception, which allows connected and autonomous vehicles to enhance their sensing capabilities by sharing information with each other. While this collaborative approach offers many benefits, it also opens up opportunities for hackers to introduce fake objects or manipulate perception data, leading to potentially dangerous situations on the road.
One of the key findings of the study was the introduction of sophisticated, real-time attacks that were tested in both virtual simulations and real-world scenarios at U-M’s Mcity Test Facility. These attacks involved falsified LiDAR-based 3D sensor data that appeared realistic to the system but contained malicious modifications. The use of zero-delay attack scheduling, a high-risk cyber attack that introduces malicious data without delay, further demonstrated the potential security vulnerabilities in collaborative perception systems.
In virtual simulated scenarios, the attacks were highly effective, with success rates reaching 86%. On-road attacks conducted at the Mcity Test Facility triggered collisions and hard brakes, highlighting the real-world implications of these security threats. To address these vulnerabilities, the researchers developed a countermeasure system called Collaborative Anomaly Detection, which leverages shared occupancy maps to cross-check data and quickly detect abnormal data.
The Collaborative Anomaly Detection system achieved a detection rate of 91.5% with a false positive rate of 3% in virtual simulated environments. In the real-world scenarios at Mcity, the system successfully reduced safety hazards caused by data fabrication attacks. These findings provide a robust framework for improving the safety and security of connected and autonomous vehicles, not only in transportation but also in other sectors such as logistics, smart city initiatives, and defense.
By open-sourcing their methodology and providing comprehensive benchmark datasets, the researchers aim to set a new standard for research in this domain. This will help foster further development and innovation in autonomous vehicle safety and security, ultimately ensuring the protection of passengers and other road users from potential cyber threats. The study serves as a valuable contribution to the ongoing efforts to advance the safety and security of self-driving vehicle networks in the future.
