Monday, August 29, 2016

Will Vehicle-to-Vehicle Communications Ever Get On the Road?


With all the recent attention on self-driving cars occasioned by the first fatality involving such vehicles, the advent of "talking cars"—cars that communicate wirelessly via vehicle-to-vehicle (V2V) communications—has taken a back seat, so to speak.  But V2V ultimately promises to be a vital link in the chain of technologies that will make driverless vehicles possible, as well as making ordinary human-driven cars safer. 

The basic idea is this.  Each V2V-equipped vehicle has a transmitter and receiver that operate in a 5.9-GHz (microwave) wireless band.  By one proposed standard, each car transmits its location, speed, direction, and other relevant data ten times a second to any other car in a thousand-yard (~910-meter) radius.  Other cars equipped with V2V can use this data to keep pace as a following vehicle, or to avoid a collision with a car that is still out of visual sight—around a corner, for instance—but on a collision course.  Some government experts estimate that if every vehicle on the road was equipped with V2V, the number of accidents not related to impaired drivers (alcohol, etc.) could be reduced by as much as 80%.  So what's the holdup? A couple of things.

First it turns out that, according to a recent Associated Press report, the main federal agency boosting V2V and prescribing an industry standard for it is the National Highway Traffic Safety Administration (NHTSA), which is now locked in a battle with another agency, the Federal Communications Commission (FCC).  The bone in this dogfight is the microwave band that V2V needs to use.  The FCC, leaned on by powerful wireless-comm companies, wants to reallocate that part of the spectrum for wireless internet users.  But a recent technical paper examined the tradeoffs involved in reducing the bandwidth used by V2V, and showed that even the currently contemplated 75 MHz of spectrum might not be wide enough to allow virtually error-free transmission, which is what is needed in this safety-critical application.

Aside from the radio-spectrum issue, there is a question of security.  The NHTSA has had enough imagination to build in a complex security protocol for V2V.  You can easily think of ways to use V2V nefariously.  For example, crooks in an escape car being chased by cops could use a false V2V signal to tell the cop car that it's about to have a head-on collision, and the cops would slam on the brakes—if they trusted what the car told them.  This assumes that the V2V information is used only as warnings to the driver, but sooner or later automakers are going to take the driver out of the loop and allow the V2V information to be used directly by the car's control mechanisms—brakes, steering, accelerator, etc. 

So in order to prevent such shenanigans, the NHTSA has devised a complicated security system that involves digital certificates, public-key infrastructure encryption, and a lot of other things that apparently have never been combined in such an elaborate way before.  It's nice that they have thought to make each car anonymous and to ensure that potential hackers will have lots of trouble hijacking the system, but even the NHTSA itself admits they haven't worked all the bugs out of this security process yet.

The closest analogy I can think of between the proposed V2V system and anything we have now is the air-traffic control system that uses active transponders on each aircraft.  The need for security in air-traffic control is a lot less, because there are a lot fewer planes than there are cars, the Federal Aviation Administration is looking over the airlines' shoulder all the time, and there was already an extensive radar-based air-traffic control system in place before the transponders were added.  With V2V, there is no centralized control, only a lot of cars talking with each other, so the technical challenge is harder.

Even if the automakers started selling V2V-equipped cars tomorrow, it would be twenty years, by some estimates, before nearly all cars on the road would be so equipped.  And until then you couldn't count on doing things with V2V such as traveling in closely-spaced packs or caravans on freeways, because even one non-V2V car in the pack would throw everything off. 

Still, if auto insurers find that V2V-equipped vehicles really do get involved in accidents at a significantly lower rate, they're likely to offer insurance discounts for such cars.  And while consumer behavior is not entirely predictable, buying a car that automatically lowers your insurance rate would be a strong incentive for car buyers to upgrade to V2V sooner rather than later.  However, the insurance companies aren't going to do that until they have a few years of data to base their price tables on.  And that won't happen till there's a significant deployment of V2V-equipped cars.  So we have a chicken-and-egg problem. 

Close to thirty years ago now, right after digital ICs capable of generating voice-quality audio came out, the car makers experimented with another kind of talking car.  If you sat down and didn't fasten your seat belt, this woman's voice came out of nowhere and nagged you to fasten it.  That kind of talking car quickly disappeared.  The V2V idea shows promise of making cars a lot safer without a lot of complexity added, except for the system issues involved with spectrum allocation and security.  I hope that the two fighting executive-branch agencies can work out a reasonable compromise so that people can both stream video as much as they want (or are willing to pay for), and drive in safer cars.  But so far, we're not there yet.

Sources:  The Associated Press article "Auto tech industries clash over future of talking cars" was carried in the print edition of the Austin American-Statesman of Aug. 27, 2016, and appeared in other venues such as the Aug. 25 online edition of the Los Angeles Times at http://www.latimes.com/business/autos/la-fi-hy-talking-cars-20160825-snap-story.html.  I also referred to a technical paper by Lei Shi and Ki Wong Sung, "Spectrum Requirement for Vehicle-to-Vehicle Communication for Traffic Safety," available at https://www.metis2020.com/wp-content/uploads/publications/VTCSpring_2014_Shi_etal_SpectrumRequirementForV2VCommunication.pdf, and articles on V2V security at https://www.contrastsecurity.com/security-influencers/v2v-communications

No comments:

Post a Comment