Longitudinal global positioning system travel data and breach of privacy via enhanced spatial and demographic analysis

Longitudinal Global Positioning System (GPS) travel data provide a wealth of information related to travel behavior and on-road vehicle behavior that is very valuable to researchers. Sharing the data publicly allows researchers to explore the data and create new knowledge beyond the initial research objectives. However, if any data are to be used outside a secure server, the data must be processed in such a manner that ensures that the confidentiality of the data will not be breached. High-resolution GPS data (e.g., second-by-second speed and location information), when associated with the individual households or drivers, compromise privacy and have a significant potential to harm human subjects. This paper explores how data from the Commute Atlanta study in Georgia could be processed to make it useful to researchers while participants' privacy is protected. The research developed and assessed methodologies designed to identify the individual participant's home location from processed data and then tested analytical data sets for breach of privacy. The research effort found that the home location could be identified to within reasonably small neighborhoods; when the household demographic information was included in the data sets (which was necessary for researchers), exact households could be identified. Although some new data-processing approaches might be used to eliminate privacy concerns, until such systems are developed and proved to be unbreachable through rigorous analysis, the Georgia Institute of Technology team has determined that researchers should access the high-resolution data in controlled secure labs and that the data sets should not be made public without additional efforts to ensure that home locations cannot be identified when external data sources are leveraged in the analyses.