Scrubbed geo-location data not so anonymous after all
Your commute = your fingerprint
Anonymized data collected from GPS-enabled devices may not be as anonymous as you think, according to researchers who show that knowing someone's general home and work locations can be enough to identify an individual uniquely.
The findings, by Philippe Golle and Kurt Partridge of PARC, or the Palo Alto Research Center, are significant, given the proliferation of devices that monitor a user's geographic location using global positioning system and other technologies. At the same time, a growing number of websites monitor user location to offer restaurant recommendations and other services.
Frequently, those offering services based on geo-location give assurances that users' privacy is safe because identifying characteristics will be scrubbed from any data stored. Meanwhile, their cell phones, vehicle navigation systems, and even laptops continue to log their comings and goings almost continuously and divulge them to online services.
Of course, one of the two most common locations any person frequents are those of his home and work. What Golle and Partridge found is that attempts to anonymize, or obfuscate, personally identifying information may fall woefully short if a user's residence and office can be deduced.
"Obfuscation techniques which prevent re-identification based on (approximate) home location alone may not be adequate if the subject's (approximate) work location is also known," they write. "In fact, we show that home and work locations, even at a coarse resolution, are often sufficient to uniquely identify a person."
Once a subject's home and work are known, snoops can use data compiled by the US Census Bureau's LEHD, or Longitudinal Employer-Household Dynamics, program, which tracks where people live and work. Although only "privacy-preserving synthetic data" is publicly available for download, Bayesian techniques can be used to work around this limitation.
The findings are reminiscent of the wake-up call that resulted in 2006 when AOL released 20 million search queries from 658,000 users. Although the company took care to remove names and other personal information, the disclosure proved a debacle after privacy advocates showed the data could still be used to identify the people making the searches.
Geo-location data, it seems, might end up being a similar land mine, according to the 33 Bits of Entropy blog, which provides further analysis of the findings. A PDF of the original paper is available here. ®
Sponsored: Becoming a Pragmatic Security Leader