CNET también está disponible en español.

Ir a español

Don't show this again

Tech Industry

Why telematics is stuck in neutral

QNX exec Paul Leroux writes that the era of talking, thinking autos may be fated to take a long detour unless Silicon Valley can come up with far more reliable software.

    Ever notice how car commercials stress style over substance? Brand X, we are assured, builds excitement. Brand Y makes you want to tear off your necktie and play hooky. And Brand Z is so sporty looking that young women love to run their hands along its...well, you get the idea.

    Of course, it often makes marketing sense to sell the sizzle, not the steak. But as it turns out, automakers have little choice. Competing brands of automobiles can have so much in common that, in many cases, a car's styling really is the only differentiator worth flaunting.

    Imagine, then, if a technology could help automakers add real--and distinctive--value to their products. Such is the promise of in-vehicle telematics. Daimler-Chrysler certainly sees the potential: They've recently unveiled a hands-free telematics system that allows drivers to operate a cell phone using natural voice commands. This one feature makes communicating from your car both safer and more convenient--not a bad differentiator.

    The Daimler-Chrysler system is only the beginning. The same push for product differentiation that spawned this product is driving other automakers to combine cellular technology, Internet access, GPS (Global System for Mobile Communications) and dynamic navigation into their own unique in-car systems. In fact, it's estimated that over 20 million telematics-enabled cars and light trucks will be on the road in the United States by 2006.

    This convergence of technologies could change driving dramatically. Lost your car key? Just dial a number on your cell phone, enter a password and presto!--your door opens. Accident? An onboard computer could immediately dial 911 and provide the dispatcher with your exact GPS coordinates. Engine trouble? The same computer could automatically locate the nearest service center and, if you'd like, book a service appointment (after it has checked the scheduler on your PDA, of course).

    The software deployed in cars is going to get very complex--more sophisticated, in fact, than many of the applications on your desktop PC.
    All these features mean one thing: The software deployed in cars is going to get very complex--more sophisticated, in fact, than many of the applications on your desktop PC. But the software will also have to be a lot more reliable.

    Think about it: What do you do when your desktop operating system crashes? You might curse a blue streak, but you'll probably still buy the next version of the OS. But if your dashboard crashes? I don't know about you, but my brand loyalty would take a dive. That's a huge issue in the auto industry, where it takes an average of 18 years to win a customer back.

    Of course, automakers will be extremely careful about software testing--safety and regulatory issues give them no choice. Unfortunately, once software gets complex enough, no amount of testing can eliminate every bug. That's going to present problems when the software may be deployed in thousands of vehicles.

    More to the point, a car offers a relatively hostile environment. Desktop PCs are rarely exposed to excessive radio frequency or electromagnetic interference, but, within the car, stray interference near power lines or transformers can affect hardware to the point that a software driver will fail.

    This fault tolerance can't apply only to applications. It has to go deeper, right down to the device drivers and protocol stacks at the heart of any telematics system.
    Automakers must assume such problems may occur, and must design their systems to recover quickly and automatically, without affecting the car's occupants in any way. It's a tall order!

    In effect, they need to deploy high-availability systems. By this, I don't mean conventional designs, which typically recover from software failures by using redundant backup systems. That isn't an option in the car market, where the cost of every bolt counts. So, rather than use redundant hardware, high-availability designs for automobiles have to be implemented where most problems can occur in the first place: the software.

    Virtually any software process must be able to fail without affecting services provided by other processes. Moreover, the system should be able to restart any process automatically. For example, if a media player faults, the system would restart it instantly, without the driver even knowing there was a problem. Mind you, this fault tolerance can't apply only to applications. It has to go deeper, right down to the device drivers and protocol stacks at the heart of any telematics system.

    Can automakers really do this? Definitely, provided they use the right operating-system technology. They need to look closely at the operating system they choose and ensure that it can provide memory protection not just for applications (the desktop approach), but for every software driver, file system and protocol. The operating system must also offer a high-availability framework that can automate software recovery, without the need for a reboot. Otherwise, the phrase "car crash" may take on a whole new meaning.

    It remains to be seen just how much consumers will embrace this brave new era of talking, thinking cars. But my fear is that without high-availability operating-system technology, it won't get past the starting line.