Trajectory And Friendship Based Location Privacy Preserving

In recent years, with the rapid development of location-based social network, location privacy issue has drawn increasing attention. Large-scale real social network data has been generated, which not only directly discloses the location information, but also implies the habit, health and other private information of users, therefore, location privacy of users in location-based social network need be protected right now.To protect location privacy, this paper considers not only the online interactions between users but also the offline relationships between users and locations, and an attack model and protection mechanisms based on the nature of location-based social networking are proposed.Firstly, this paper analyzes the effect of friend relationships and historical trajectories on location privacy, thereby a location attack model based on historical trajectories and friend relationships is established. And then factors that affect the result of attack model are further analyzed.Secondly, based on analysis of the attack model, location privacy protection mechanisms are established. This paper assumes that the knowledge of location attack is known at first and takes the quality of service and strength of privacy into consideration, location privacy problem is modeled as a multi-objective programming problem. On this basis, a non-cooperative game model is established to solve the privacy protecting problem under the situation that the knowledge of location attack is uncertain.Finally, in order to verify the effectiveness of protection mechanisms and attack model, this paper designed and implemented experiments based on Foursquare data set. The results show that (a) the location attack proposed by this paper is better than other location attacks, (b) the privacy protection mechanisms we proposed achieve good protective effects in the aspects of accuracy, correctness and certainty, (c) this paper verified the restriction between quality of service and strength of privacy in the privacy protection mechanisms we proposed.