The 5G Network Probably Won’t Be Good Enough for Self-Driving Cars
Car-to-car meshnets could enable self-driving cars where there’s no internet.
A remote road in Arizona. Image: Nicholas A. Tonelli/Flickr.
If you’re super jazzed about getting an autonomous vehicle to chauffeur you around town while you catch up on Netflix or take a nap, we’ve got some bad news: The current geographical disparities in internet access mean it’s probably not going to happen for at least another 15–20 years.
“People always ask me, ‘How long until I can buy an autonomous vehicle?’ And my answer is, ‘Well, where do you want to go?’” Ross McKenzie, managing director of the Waterloo Centre for Automotive Research (WatCAR), told me in a phone interview. Connected/autonomous vehicles (CAVs) need ubiquitous, reliable, and extremely fast wireless networks to communicate with other vehicles, infrastructure, and devices while in drive. Any kind of delay or latency in transmitting vital information to a self-driving car could be extremely dangerous.
Shuttling around town with autonomous-compatible infrastructure could be done in the next two decades, McKenzie said. But intercity travel is complicated by vast dead zones where you can’t even get a cell signal, let alone count on your car to not drive you off a cliff.
The problem is, huge parts of the world have terrible cell and internet coverage and it’s uncertain whether the 4G cellular network’s successor, 5G, will be enough to power the self-driving industry. That means CAV makers might have to get creative in proposing new and potentially revolutionary networking solutions, like mesh networking.
Before you get mental images of self-driving cars careening into crowds due to a dropped connection, you should know there’s a wireless spectrum band (5.9 GHz) currently reserved for CAVs. That spectrum is primarily used for dedicated short-range communications (DSRC), a wireless network that operates separately from 4G/5G and Wi-Fi to enable vehicle-to-vehicle and vehicle-to-infrastructure signalling.
But as the name implies, it’s meant to be a short-range solution. It typically taps out at one kilometre (about 0.6 miles)—so we’d still potentially need an overarching network to communicate past that range.
If the impending 5G network had the right infrastructure to connect remote areas to fast wireless internet, it could potentially solve some of the communication problems for CAVs.
“But then you get into an economics issue,” said Robert Fitts, interim president of a Canadian consortium of telecoms and networking companies known as CENGN. Making a truly connected road network would cost telcos a hell of a lot of money, he noted. “Who’s going to pay for it?”
Telcos have little financial incentive to deliver stable, low-latency, high-speed coverage along the entire road network: It’d cost them billions of dollars to build the infrastructure alone to service relatively few individuals. Instead, they’re more interested in delivering better service to high-density areas where they’re likely to find a critical mass of customers.
Part of the holdup is that it’s not immediately clear how wireless network providers will make money off of CAVs. So in the current scenario, the task—and the tab—typically falls to the government.
Mesh networking, or letting people use their internet connection to provide access to others, could potentially alleviate some of the communication problems the CAV industry is facing, said Jason Ernst, chief networking scientist for mesh-network maker RightMesh who holds a PhD in mesh networking and heterogeneous wireless networks.
“The idea with the mesh is to enable phone-to-phone-to-phone communication,” said Ernst—or, in this case, car-to-car-to-car.
You can visualize it as a chain: The first link has a Wi-Fi connection, and all the others linked to it piggyback off of its connection to access the internet rather than each link connecting to the cell tower individually. In the CAV context, that connection could be Wi-Fi, 4G/5G or DSRC, or switch between them.
British Columbia-based RightMesh is currently developing these networks in population-dense parts of the developing world such as India, Bangladesh, and Latin America, where Wi-Fi and cellular infrastructure are lacking. “Every person who’s directly attached to cell phone tower can become their own mini-ISP (internet service provider),” said Ernst. The technology is also being used in the United States in communities that have been passed over by telecom companies.
In a CAV scenario, all wireless internet connections—including connections at households and businesses near the road—could be used as vehicular communication devices.
To incentivize subscribers to open up their wireless connections to passersby, Ernst proposed a data-reselling program where people can earn money for unused plan data, like the blockchain-powered program RightMesh uses for mobile mesh networks. It’d save telcos money on infrastructure and electricity (Wi-Fi consumes about 23 times less power than 4G, according to Ernst)—but it would also invariably endanger the subscription model telcos have based their entire business model on.
“All these normal telecom companies are coming from what they know. They know how to monetize it on subscriptions,” said Ernst. “It’s hard to monetize a car-to-car network.”
Get six of our favorite Motherboard stories every day by signing up for our newsletter .