By Daniel Browning
Our society is fast entering the age of driverless vehicles. Propelled by hefty capital investment and research from technology leaders like Google, Tesla, and others, the coming years will see steering wheels and pedals traded for touchscreens and infotainment systems. There will be no need for driver education classes. Traffic fatalities, which totaled more than 36,000 deaths in 2019, will likely be close to zero. And policymakers will have to rethink the role of transportation agencies like the DMV, as people trend toward cheaper, more convenient self-driving services.
As mass adoption of the so-called internet car inches closer to reality, security experts remind us that the underlying technologies powering the driverless age present new attack vectors ripe for exploitation. Compounding the challenge, we must remember that when dealing with vehicles, loss is measured in terms of human life and limb rather than data.
In this article, we’ll consider the major problem areas and systemic risks of the self-driving era as well as explore the vulnerabilities within systems such as GPS, LiDAR, and VANET.
Risks and Consequences of Driverless Vehicles
Malicious Traffic Jams
Research from Georgia Tech shows how “randomly stalling 20 percent of cars during rush hour would mean total traffic freeze,” in a metropolis such as New York City. Hackers could use such traffic jam attacks as a prelude to other criminal activity, or to simply cause mayhem in a city, as emergency personnel and first responders would be overwhelmed by the congestion.
Kidnapping / Hostage Events
We also have to consider the threat posed by hostage taking and kidnap events. Prominent figures in society – celebrities, political figures, business executives – are arguably most at risk. When a hacker has control of an autonomous vehicle, occupants are at the complete mercy of that person.
Vehicle-related terrorism is yet another area of concern. In recent years, society has witnessed terror groups increasingly turn to vehicles to commit mass murder across the globe. The death toll from some of these attacks measures in hundreds of lives lost. Society can’t allow autonomous vehicles to become yet another tool of destruction for these groups.
In addition to security worries, policymakers, universities, and manufacturers must also wrestle with the complex moral issues that we’ll face along the journey. For instance, much of the decision making will be handled by neural nets like the IoT. When life or death scenarios arise, how will these algorithms respond? Who will be responsible for those actions? What does accountability look like?
Vulnerabilities in Sensing and Visual Systems
Modern vehicles that offer semi-autonomous features rely on sensors like LiDAR, camera systems, and radar to learn about the surrounding environment. The accuracy of the information that these sensors send to control systems is paramount to the safe operation of the vehicle. Inaccurate or misinterpreted data about the surroundings can easily cause an accident.
Each of these aforementioned sensor technologies has a unique set of vulnerabilities. Research shows that graffiti, stickers, or subtle fading can cause camera-based sensors to misinterpret road signs, even when “discreet enough to be undetectable to casual observers.”
LiDAR also has its well-known pitfalls, including vulnerability to spoofing attacks that can cause sensors to see vehicles or road debris that don’t exist. Researchers at the University of Michigan demonstrated this through two attacks they called the “emergency brake attack” and the “AV freezing attack.” The break attack shows “how an attacker can suddenly halt a moving vehicle by tricking it into thinking an obstacle appeared in its path.” The AV attack uses “a spoofed obstacle to fool a vehicle that had been stopped at a red light to remain stopped after the light turns green.”
Vulnerabilities in Positioning Systems
Spoofing attacks also threaten the integrity of positioning systems. Hackers can use off-the-shelf satellite simulation software to deceive a victim GPS by sending bogus position, velocity, and temporal information. Security professionals have demonstrated how this erroneous data can easily be used to hijack a vehicle, as hackers can send spoofed data to effectively reprogram the destination.
GPS is susceptible to these attacks because it lacks trust procedures. In other words, a receiver will simply assume that incoming data is authentic. And since it’s a low power system, an attacker can easily overwhelm a victim if in control of the stronger signal. Adding to the issue, spoofing attacks are extraordinarily cheap, requiring just a “$400 radio and a laptop,” for one research team.
Given that GPS is essential to full self-driving, security teams are working to develop methods that better protect systems from attacks. New research from the University of Texas, San Antonio says GPS can be complemented with “Doppler radar speedometers, radar cruise control, and radar-based obstacle detection systems … to deliver additional active location validation.”
Vulnerabilities in Vehicle Ad Hoc Networks
Vehicle Ad Hoc Networks or VANETs are designed to improve road safety through vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) data transmission. As VANET technology reaches its full potential in the coming years, we’ll rely on these spontaneous networks to intelligently inform each other about traffic maneuvers, destinations, hazards, and information about the surrounding environment.
Unsurprisingly, since we’re dealing with networks, VANETs face many of the same problems as any other network system. Attacks range from denial of services, eavesdropping, man-in-the-middle, spam, location trailing, session hijacking, key-certification replication, bogus messages, message saturation, and impersonation
Although we have several major milestones to reach before automakers begin to roll out fully autonomous vehicles, today we are closer than ever. But as highlighted in this piece, the driverless age will introduce new hazards and technical challenges that must be addressed to ensure public safety. However, once those problems are solved, society will be rewarded greatly as roads become smarter and safer for all.
Daniel Browning is the Business Development Coordinator at PDF Electric & Supply. He enjoys writing about automation, AI, and new technology.