A preliminary December 2012 study conducted by the Massachusetts Institute of Technology and the University of California has identified the source of traffic jams in the Boston and San Francisco area by tracking vehicles using three weeks of anonymous cellphone and GPS signals.
This study is notable in that is it the first traffic study to track travel using anonymous cellphone data rather than traffic surveys or censuses.
Over the course of three weeks, researchers collected cellphone logs from 680,000 commuters to obtain information about travel routes in Boston and San Francisco. The logs identified the towers used to transmit calls, and allowed researchers to trace each individual’s commute from origin to destination. A driver’s home neighbourhood was inferred based on the regularity of a travelled route and the volume and location of calls made between 9 PM and 6 AM. This information was combined with population density and capacity of road networks to determine which neighbourhoods contributed the most drivers on each road segments.
However, cell tower data provides good location about location and routing, but imprecise at positioning. Thus the data collected was checked by obtaining traffic speed from GPS sensors in taxis, from which congestion levels were determined.
It was found that during rush hour, 98% of roads in Boston were below traffic capacity. The 2% of roads over capacity included highways, a number of downtown streets, and some suburban arteries, all of which were key to connecting different parts of the metropolitan area. Additionally, only a small number of drivers from a small number of neighbourhoods were responsible for congestion on major routes. Delaying 1% of trips across an entire road network would reduce congestion delays by 3%, but cancelling trips in carefully selected neighbourhoods reduces travel times across a road network by up to 18%.
Study authors note that although it is easy to point fingers at which suburbs are responsible for the traffic snarl (why hell-o, Streetsville), it is still difficult for policy makers to decide whether and how to target specific geographical areas.
The authors anticipate that their methodology could be applied to almost any urban area where population density, topological road network information, and cellphone data are available, especially in the developing world, where traffic surveys are not in widespread use.
Leave a Reply