It doesn’t take much to turn rush hour traffic in Manhattan into a nightmare. According to a new study, hackers could bring traffic there to a complete standstill if they could randomly paralyze roughly 20 percent of all cars on the road at the height of rush hour.
For years, computer scientists have shown they can hack into cars remotely via the Internet. For example, in 2015, two researchers showed they could hack into a Jeep's brakes, steering, transmission, radio, air conditioning and wipers from afar using the Internet connection its entertainment system makes through Sprint's mobile network.
Extrapolating from these examples, physicist Peter Yunker at the Georgia Institute of Technology and his colleagues wanted to explore what might happen if hackers attacked not just single cars, but multiple vehicles simultaneously.
“What if many cars across the city within a very short amount of time suddenly stopped, blocking traffic?” Yunker says.
Using computer simulations, the scientists analyzed the movements of cars on streets with varying numbers of lanes, using known models for the way moving cars typically change speed or switch lanes in response to other vehicles. They next assumed that some cars were hacked to stop, blocking some lanes.
The researchers, who detailed their findings in 30 July online edition of the journal Physical Review E., found that whether traffic halted or not followed the physics of percolation. This is a phenomenon often exploited in materials science to see if a desirable quality such as a specific level of rigidity will spread throughout a material to make the final product uniformly stable. In this case, stalled cars made traffic go from flowing to stuck.
"Percolation recurs frequently in physics; there's a lot of established mathematics behind it," says Yunker.
Maps of Manhattan show the effect of hacking and stranding 20 percent of vehicles on the streets at varying times of day. The fainter the street, the slower the traffic. Streets completely faded out are gridlocked to a standstill, and very faint streets are practically unusable. The simulations are conservative as they do not factor in spillover traffic from blocked roads, delivery trucks, and normal traffic obstructions.Image: Yunker Lab
Using the mathematics of percolation, the scientists could rapidly calculate the probability whether traffic on any road would come to a halt based on the number of cars on the road and the number of cars stalled due to hacking. They focused on Manhattan for these simulations because a lot of data was available for its traffic patterns.
The researchers predicted that randomly stalling 20 percent of all cars during rush hour in Manhattan would result in total gridlock.
“This isn't just bad traffic where you are an hour late. It becomes impossible to get from point A to point B,” says Yunker. “You are completely surrounded by blocked roads. You have complete fragmentation of the city. This is especially concerning for emergency responders such as EMTs and firefighters who cannot reach their destinations to respond to an emergency.”
Not all cars on the road would need to be self-driving and Internet-connected for such paralysis to occur. For example, if 40 percent of all cars on the road in Manhattan were online and autonomous, hacking half of those would suffice.
The researchers note that other cities might suffer even worse harm from such hacking. The streets of Manhattan and Chicago are arranged as grids that make traffic more efficient. But cities without large grids—Atlanta, Boston, and Los Angeles, to name a few—were more vulnerable to gridlock from such attacks.
Yunker and his colleagues cautioned that they considered only static situations where roads were either blocked or not blocked. Future research with more dynamic models would likely show that blocked roads would spill traffic over into other roads. Given such effects, it might be possible to trigger gridlock by stalling much less than 20 percent of all cars, Yunker says.
The scientists hope their findings will start conversations about how to mitigate the risk of cities falling prey to such worst-case scenarios. For instance, Yunker notes that in the United States, each of the top five automakers have about 10 percent of the market. “If two of those automakers shared the same security protocols, then hacking them could lead to the worst-case scenario we identified,” he says. “On the other hand, if individual companies each had multiple approaches to security—say, if one [carmaker] had five different protocols for its vehicles—then hackers would have to find vulnerabilities in each one of those approaches to [have a shot at gaining control of] 10 percent of the vehicles on the road.”