A wide-ranging investigative project from MIT and Belgium’s Louvain University has found that mobile phone records identify users even more accurately than their own fingerprints.
A new paper published in Nature called “Unique in the Crowd: The Privacy Bounds of Human Mobility,” claims that 95% of mobile phone users can be identified based entirely on their patterns of movement. Mobile phones routinely ping cell phone antennas as customers travel from place to place, even if the phone is not being used–and even if the phone is turned off. The only way to prevent a phone from pinging antennas is to physically remove the battery.
It’s old news that your mobile phone records aren’t public and that law enforcement authorities can easily triangulate your daily habits through carrier records. But now
Researchers Yves-Alexandre de Montjoye, Cesar Hidalgo, Michel Verleysen, and Vincent Blondel used big data analysis techniques to parse through 15 months of mobility data from 1.5 million anonymized mobile phone records from an unnamed mobile provider in a “small European country”; the dataset contained information on the location of connecting antennas when users made or received telephone calls and SMS text messages between April 2006 and June 2007. The study was funded by the Belgian government–Belgium, of course, being a small European country.
Once these anonymized records were parsed by the researchers, they discovered something surprising: 95% of the anonymous customers whose records they analyzed had distinct patterns of movement. Every time they traveled, their phones pinged nearby cell phone towers whenever they were used.
An image below summarizes three findings from the study. The first box shows the trace of one anonymous phone user from the study, with the dots representing the times and locations where they made or received phone calls. The second box shows the user’s trace according to a correlated mobility database kept by the phone provider. Meanwhile, the third box shows the user’s trace once filtered through a spatial and temporal aggregation by the researchers. This technique, which is frequently used by law enforcement and intelligence agencies, highly accurately predicts at what time during the day customers will be in a certain neighborhood or town.
As previously noted, the study only included data collected in 2006 and 2007. In the post-iPhone era, smartphones with highly enhanced geolocation data have become everyday tools in much of the world. Even if smartphone owners opt not to use voluntarily self-geotagging apps like Foursquare, Facebook, Instagram, Twitter, and Yelp, their location data–which is triangulated through nearby mobile antennas and pings to random nearby Wi-Fi routers–is still recorded and parsed by a host of location-based apps. The dirty secret of the mobile app economy is that anonymized location-based information is the unofficial currency of the mobile web. Geography-based information on a user’s most intimate habits can be translated into targeted advertisements sent to them on both their phones and computers (thanks to the magic of cookies). The type of pre-school a user drops their kids off at, the location of the grocery stores where they shop, and the movie theatres or restaurants where they go on dates are all valuable breadcrumbs for mobile advertising firms to monetize. These anonymized records, of course, are easily accessible by law enforcement authorities and intelligence agencies–and can be merged with the plain call records analyzed by the MIT-Louvain team without too much of a headache.
These location records, which showed each anonymous user’s motions throughout the day, created individual fingerprints which were as unique as fingerprints for the 95% of users who moved around during the day. The other 5%, presumably, were either homebodies or people savvy enough to remove their mobile phone batteries when their device was in disuse.
[Top Image: Flickr user epsos.de, Bottom Image: Yves-Alexandre de Montjoye, Cesar A. Hidalgo, Michel Verleysen, Vincent D. Blondel]