Traffic engineering applications often benefit from collecting vehicle GPS traces in well-defined locations. As a motivating example, we will consider a traffic signal performance evaluation technique which requires GPS traces of vehicles traversing intersections. Since these applications do not require vehicle identity information, they are a good candidate for data de-identification techniques. Prior techniques can either provide data from specific locations or guarantee a high degree of anonymity under light traffic conditions but do not achieve both. In this paper, we propose a virtual trip line zone-aware path cloaking algorithm which combines these features. Zones where data should be retained can be predefined over the intersections of interest and the path cloaking algorithm uses entropy-estimates to decide whether the data can be revealed. Result obtained from a traffic simulator show that the application success rate increased from 39 to 82% compared to a zone-unaware path cloaking algorithm, while achieving a similar degree of privacy.
The different versions of the original document can be found in: