EPLA: efficient personal location anonymity

A lot of researchers utilize side-information, such as the map which is likely to be exploited by some attackers, to protect users’ location privacy in location-based service (LBS). However, current technologies universally model the side-information for all users and don’t distinguish different users. We argue that the side-information is personal for every user. In this paper, we propose an efficient method, namely EPLA, to protect the users’ privacy using visit probability. We select the dummy locations to achieve k-anonymity according to personal visit probability for users’ queries. In EPLA, we use AKDE(Approximate Kernel Density Estimate), which greatly reduces the computational complexity compared with KDE approach. We conduct the comprehensive experimental study on the two real Gowalla and Foursqure data sets and the experimental results show that EPLA obtains fine privacy performance and low computation complexity.