Mass autonomous cars project lets 3000 vehicles talk

Hal Hodson, technology reporter

(Image: US Department of Transportation)

On Tuesday this week, something unprecedented happened in a medium-sized town just east of Detroit, Michigan - 3000 cars, trucks and buses started talking to each other.

They are part of a US Department of Transport (DOT) pilot project which aims to make driving safer. For a year, 3000 volunteer motorists in Ann Arbor will have equipment installed in their vehicles which wirelessly exchanges speed and location data with others in the area, as well as with infrastructure such as traffic lights.

A small loudspeaker warns drivers of danger in their area, such as a sudden deceleration on the road ahead, or a sudden lane change nearby. The vehicles store all the data they collect for researchers at the University of Michigan and the DOT to analyse at the end of the year.

The pilot is testing the real-world performance of vehicle-to-vehicle communications systems. These have never been tried on such a grand scale, and naturally, security and privacy are a concern.

In a risk assessment paper last year, the DOT's Intelligent Transport Systems office outlined the security risk to the system's 3000 users, suggesting that a sufficiently funded organisation might be able to steal the encryption keys that protect the communication channels between the vehicles and use these to broadcast erroneous messages into the system. That kind of attack would require physical access to the car.

On the privacy side, the document describes how personally identifying information can be scrubbed from the signals, making it impossible for individual drivers to be tracked across the city via their vehicle's wireless signal. This "anti-linking" technology changes all the identifying data sent out by the pilot cars every 5 minutes, including their encryption certificates, so no one number or data point is continuously associated with any one vehicle.

The paper ends with a list of pros and cons about the project. On the plus side, it says, the system is scalable to more than 250 million users, and is difficult to attack with significant knowledge of the system.

On the downside, the paper notes that there is no clear financial model to keep the project going. Also noted, somewhat disturbingly, was the fact that the system cannot instantly identify misbehaving drivers, saying that there is inevitably a "delay in identification and [a] delay in removing misbehaving actors from system".

Source: http://feeds.newscientist.com/c/749/f/10897/s/22a7289b/l/0L0Snewscientist0N0Cblogs0Conepercent0C20A120C0A80Cmass0Eautonomous0Ecars0Eproject0El0Bhtml0DDCMP0FOTC0Erss0Gnsref0Fonline0Enews/story01.htm

gmail down tim lincecum ryan oneal file taxes online tupac shakur sledge hammer tax day freebies