Cars

Can cellphones handle vehicle-to-vehicle comms better than radio networks?

It's going to be a long time before DSRC-equipped vehicles are common on our roads.

Jonathan M. Gitlin – Jul 31, 2017 11:50 am | 63

Nexar's iOS and Android app uses the phone's forward-looking camera, accelerometers, and GPS. It will recognize the backs of cars ahead and provide forward collision warnings. Credit: Nexar

NEW YORK—Vehicle-to-Vehicle communication (aka V2V) sometimes feels like the automotive world's Duke Nukem Forever. The idea of vehicles communicating with each other over short distances to warn drivers of potential obstacles or dangers is compelling. But it may as well be vaporware. Nearly 20 years after the Federal Communications Commission allocated radio spectrum for it, we're still waiting. Today, it may be more than a decade before enough V2V-equipped cars are on our roads to make a difference. But we might all be walking around with the solution in our pocket—smartphones. At least, that's the impression I walked away with after a demo of Nexar's technology one recent morning in Manhattan.

A V2V traffic jam?

Dedicated Short-range Radio Communications (DSRC) networks communicate with each other at ranges of up to 984 feet (300m), operating in a band around 5.9GHz. And so the FCC set aside radio spectrum for V2V communications way back in 1999. But it took another 11 years for a wireless protocol—802.11p—to get nailed down, finally happening in 2010. From that point, it was another four years before the National Highway Traffic Safety Administration (NHTSA) announced that it was ready to start the rule-making process to get V2V approved for cars. NHTSA finally released a draft rule in December 2016, and we believe it to be a regulation that the current administration is keen to see implemented. Once on the books, it would give car makers two model years—plus a little extra leeway for older models—to make the technology standard across their fleets.

Even assuming that happens as planned, it's going to be quite some time before V2V has sufficiently penetrated the US light vehicle fleet for it to be of much use. After all, if there are no other V2V-equipped cars within 300m, there's nothing for the system to do, right?

It's the implementation, not the idea

Consequently, not everyone is as enamored with V2V as you might think. Compare it to Automatic Emergency Braking (AEB): the OEMs are racing to make that a standard feature well before NHTSA can mandate it as such. Yet there's no such push happening with DSRC, in part because the march of technology may well have overtaken this particular flavor. Like that other mid-'90s automotive legacy system—the CANbus—DSRC was a product of an earlier time. Within a couple of years it will be impossible to buy a new car without an embedded LTE modem. And with very low latency 5G LTE on the way, companies like BMW, Daimler-Benz, and others think a technology-agnostic V2V mandate is the way to go.

Even once a final specification—be it a specific radio system and protocol or a cloud-based interoperable standard—is agreed upon and so-equipped cars start filtering out of the factories, that still leaves the hundreds of millions of legacy vehicles already on our streets. On average we're keeping our vehicles for longer than before, and cars are built to take more punishment than they used to cope with. Absent schemes like "Cash for clunkers," the future—like the present—is going to be a mix of new and older machines sharing the roads.

But there's hope for those of us who still drive something old-school or for the (seemingly significant) demographic of Ars readers who don't want to drive a vehicle permanently connected to the outside world: hope that looks something like this smartphone-based dashcam app from an Israeli startup called Nexar. Eran Shir, one of the co-founders and the company's CEO, told me that Nexar came about in reaction to DSRC's troubled history. "Really, in networking terms it's a relic from the past. We've shown you can provide warnings with physical LTE under 50ms, and that's extremely relevant for many use cases today," Shir said.

The tech is already here—and in your pocket

Nexar's app uses the phone's camera to provide forward collision warning (FCW) alerts, but on top of that it will communicate with other Nexar users within close range to give them relevant safety alerts as well. "If you're just using sensors running on an individual vehicle you'll never get there. What we did was use machine vision and sensor fusion to track everyone around you—akin to MobilEye—but will connect you to other vehicles in your vicinity. So you can see beyond line of sight," he explained.

During a demo on the streets of Manhattan's Financial District one Monday morning, the system did just that. I rode passenger in a car running the Nexar app on an iPhone 6. (Nexar has apps for iOS and Android.) Running in debug mode, I was able to see the image recognition system pick up and bound the different vehicles in front of us with squares; green for vehicles directly in our path and therefore a potential hazard, black for vehicles in other lanes. (You won't see this on the images in the gallery, as they show the view actual Nexar users get.)

This being Manhattan traffic, we didn't need to wait long before a delivery driver slammed on the brakes in front, triggering a FCW alert. As the system isn't integrated with the vehicle at all—there's no OBD2 port required or anything else that could be a vector from the Internet into your car's hardwired network—those alerts are merely audio and visual. On its own, that might not have been any more interesting than a demo of any other dashcam that does image recognition and collision warnings.

But accompanying us on our drive was another Nexar-equipped car. After allowing several other road users to merge into our lane (thereby putting several cars between us), and when it was safe to do so, this car made an emergency stop. (The phone's accelerometers are used to detect this; given a sufficient G-threshold it knows you're braking hard.) Despite not having a visual on the second Nexar car, the app gave us a near real-time warning of an emergency braking action in our lane. Both vehicles were using 4G LTE, albeit on different network providers, and even with the need for the data to be coordinated in the cloud, the time from braking event to warning us was under 100ms.

During the 20-minute drive around downtown Manhattan, I noticed the app was picking up more than just our other demo driver. Nexar tells me it has about 5,000 daily users in the Big Apple, mainly taxis, delivery vehicles, and other drivers who spend their days at the wheel. Its cloud-based platform keeps tabs on all those drivers, and when they're sufficiently close to each other—within two to three blocks—it allows them to send alerts between them.

Qui bono?

As a driver, there are a few benefits to running Nexar's system other than just collision alerts: insurance discounts for one, plus the camera will automatically upload video up to and after an incident of heavy braking (which needn't just be a crash; it could be someone cutting you off). In this event, it sends the preceding 20 seconds and following 10 seconds to the cloud and uses character recognition to identify the number plate of the vehicle "at fault," which Nexar says at some point could be shared to other users.

It's not just the drivers using Nexar on their phones who benefit from the data they're capturing, either. "We see all of Manhattan 100 times a day. That allows us to detect things happening, capture that, and create a fire hose of events which we can provide," Shir told me. That's a data platform (called CityStream) that lets municipalities and city agencies have access to a near real-time view of their street network. "We can see obstacles, dynamically detect school buses or lane closures, then within seconds share that through the fire hose. It's really about the ability to make a decision that something is of interest, uploading and analyzing it, and providing it as a light GIS layer. Not just mapping lane markers but providing a high-level overview of what actually happens so you can act on it."

Of course, a proprietary V2V network has some of the same problems as an ineffectually rolled-out V2V network: what's the point if only a small percentage of cars are covered? Although Nexar said it has no plans to open up its platform to others just yet, it did seem open to the idea of interoperability further down the line. Then again, we could also see the company being acquired and its tech being folded into an existing product—something like Waze springs to mind.

Regardless of the eventual fate of Nexar, one thing did seem crystal clear by the end of the demo drive: V2V needn't be shackled by an out-of-date technological mandate. NHTSA is showing that it's prepared to take a tech-agnostic approach to self-driving vehicles, so maybe it will be able to do so with V2V as well.

Jonathan M. Gitlin Automotive Editor

Jonathan is the Automotive Editor at Ars Technica. He has a BSc and PhD in Pharmacology. In 2014 he decided to indulge his lifelong passion for the car by leaving the National Human Genome Research Institute and launching Ars Technica's automotive coverage. He lives in Washington, DC.

63 Comments