Imagine a way for your car to know as you’re driving–about a huge pothole just ahead. Or a ginormous traffic snarl. Or both. Would that information be of interest to you?
Bhaskar Krishnamachari, a researcher at the University of Southern California thinks it might be. He is working on technology that would enable a peer-to-peer network of properly-sensored automobiles to share and then discard that information as they travel.
The research is early stage, but Krishamachari, an associate professor at USC’s electrical engineering and computer science departments, is working with General Motors on the technology. If things go well, it could surface in five to ten years, he said at the Emtech 2011 conference at MIT today.
A big part of the foundation is the IEEE’s 802.11p standard freeing up Wifi spectrum for car-to-car communication. GigaOM’s Katie Fehrenbacher reported on a Ford Motor Company demo of the technology last June. And yesterday, Ryan Kim reported that Waze, the smart crowd-sourced traffic information application, picked up $30 million in more venture funding
Krishmachari’s speciality is low-power sensor networks that enable “smart buildings” and potentially smarter cars. Applying that technology to moving targets like cars poses its own challenges but the upside is big. Say a car is on an icy road. It detects that and shares that information with on-coming or nearby vehicles that could react accordingly.
“The sensors and network would propagate that information to more nodes and use vehicular networks to see that content and those nodes would propagate that information virally” to other nodes in close proximity–ie. the only nodes that might need to know that information.
“Vehicular networks need to be information-centric, not address-centric, where the content and the way it is processed is organized based on what it is, where it is and when and sent only to those for whom it is relevant,” Krishnamachari told conference attendees on Monday.
Such networks also require fast, in-memory algorithms and wireless-enabled smart systems–sensors–in the vehicles themselves.
One hurdle is the cost-conscious nature of the auto makers. If sensors add $100 to the cost of a car–multiple millions of dollars in aggregate– the cost-benefit analysis has to be fully worked out, he said. Car companies may be technology companies, but “they are not IT companies,” he said.
It would probably speed things up if an add-in model could be developed — if say, Google with its Google’s self-driving car project or other third parties — would adopt and promote the technology, he said
Most of the car-networking discussion now focuses on car-cloud communication, but there is a need for this more localized connectivity as well, Krishnamachari said.