Feature Story

Mobile network technology for smarter cars

February 22, 2011

By Stephen H. Wildstrom

As you are about to drive into an intersection, a piercing alarm and a flashing red warning light cause you to hit the brakes hard. You narrowly avoid a serious collision with a car running a red light.

Today, a relatively few cars equipped with expensive radar- and camera-based collision-avoidance systems could give you this warning. But if a coalition of automakers backed by the U.S. Department of Transportation has its way, much cheaper warning systems would be installed in all new cars before the end of the decade. The key is replacing expensive sensors with cheap and ubiquitous global positioning system (GPS) and Wi-Fi wireless technology.

Cars are full of electronic sensors and computers that monitor all phases of vehicle operation. GPS, already present in cars with built-in navigation systems and inexpensive to add to others, can provide data on precise location, acceleration, speed, and direction. Wi-Fi provides an easy way to broadcast the data so other cars in the vicinity can use the information.

A car equipped with the system broadcasts data 10 times per second using a WiFi variant known as dedicated short-range communications. Other cars in the vicinity pick up the broadcasts and feed the data to onboard computers, which use the information to calculate the relative position of other cars and figure out which pose potential hazards.

"We use low-cost automotive GPS," says Joseph Stinnett, a research engineer with Ford Motor Active Safety, who demonstrated the system in a mostly empty parking lot at Washington's Robert F. Kennedy Stadium. "Since we are just locating two vehicles relative to each other, the GPS positioning is quite good.  If we wanted to position the vehicle on a map of the earth, we would have three to five meters of error. But positioning relative to another vehicle, it's actually sub-one meter."  That's good enough for the critical task of resolving which car is in what lane.

In the real world, it will take some sophisticated software to make the system work effectively. On a crowded highway, a car might be receiving signals from dozens of vehicles within Wi-Fi range at any time. The software has to filter out information from irrelevant cars—for example, those heading in the other direction on a divided highway—to focus on those that pose a potential threat. With 1,000 or more signals arriving every second, that takes a fair amount of processing power. Fortunately, processing, like GPS and Wi-Fi, has gotten very cheap.

Meanwhile, high prices continue to pose a major barrier to adoption of existing systems, effectively limiting their use to high-end cars. For example, a relatively limited radar-based blind-spot monitoring system is part of a package that adds more than $3,000 to the price of a Ford Taurus.

Of course, existing systems have one huge advantage: their radar and computer vision sensors don't depend on other vehicles being similarly equipped to do their job. But the GPS plus Wi-Fi approach only works if all nearby vehicles are broadcasting data. A car in your blind spot that isn't sending out position information is as invisible to your computer as it is in your mirrors. That's why the members of the third-generation Vehicle Safety Communications Consortium--Ford, General Motors, Daimler, Volkswagen Audi. Kia Hyundai, Toyota, Nissan, and Honda—are working with DoT and theNational Highway Traffic Safety Administration toward requiring installation of the systems once they become cheap and reliable enough.

Widespread use of the system could provide additional benefits. Existing systems for monitoring traffic flow mostly rely on loop sensors embedded in the pavement. They are expensive to install and maintain, generally limiting their use—and the availability of real-time traffic reports--to a few major roads. Monitoring the Wi-Fi broadcasts of passing cars, though it would raise some privacy issues that must be addressed, could provide much more extensive data at much lower cost.