Spatio-Temporal Location Privacy Quantification for Vehicular Networks

Connected vehicles continuously reveal their location information to the potential observers that then track vehicles over time and space. However, safety applications require disseminating such location information to the vicinity, which essentially urges to develop the awareness ability of the exposed location-privacy quantity. We propose an analytic framework to quantify location privacy from the perspective of temporal and spatial correlation, which permits connected vehicles to flexibly configure their preferential location-privacy sensitivity extent. We offer a novel approach so as to make the disclosure of location privacy more human-understandable and independent of the suffered attack pattern. This paradigm enables users to customize the location-privacy protection mechanisms for adapting to the frequently varying traffic context. Moreover, we put forward an adaptive lived-term pseudonym scheme for exampling how to utilize the presented quantitative framework. Finally, we move from theory to practice by applying the proposed theoretical framework to a simulation vehicular mobility data set TAPASCologne composed of more than two hundred and fifty thousand vehicles and a real trace set of more than ten million location sample points obtained from the Roman taxis. We investigate the accuracy and applicability of the presented quantification model and effects of combinations of various concerned parameters. The results show that the quantification model can identify the real-time exposure of location privacy efficiently and effectively as vehicles move and then provide quantitative control feedback to privacy protection mechanisms, which facilitates restrict the location privacy to a target value and tradeoff between the location privacy and service enjoyment.