Research On Privacy Preserving High-dimensional Trajectory Data Publishing Method Against Subsequence Attack

With the development and popularization of mobile location technology and location aware devices,people are increasingly relying on mobile devices to obtain services.Service providers,research institutions and even individuals have a growing number of trajectory datasets.Trajectory data contains abundant spatiotemporal information,publishing trajectory data and mining information is the foundation of the development and update of many applications in the current market.However,the data usually contains a large amount of sensitive information of the users,which will bring a series of privacy security problems if directly published on the Internet.How to anonymize the data before releasing it to protect the privacy of users and guarantee the availability of data to the greatest extent,has become a hot issue for domestic and foreign scholars.This thesis focuses on the problem of privacy disclosure in high-dimensional trajectory publishing,aiming at the situation that an attacker steals the privacy information of the user by using trajectory sequence and its combination with sensitive attributes under the condition of partial background knowledge,studies on the existing trajectory privacy protection technology and proposes novel methods to achieve the goal of protecting user privacy.The main work of this thesis is as follows:1.In view of the current high-dimensional trajectory privacy protection algorithm using a single trajectory privacy protection technology,which makes it difficult to accurately measure the anonymous location,resulting in high data loss rate and poor privacy protection,this thesis proposes a trajectory privacy protection method based on single point of gain.First,the algorithm finds out the set of trajectories with the risk of privacy disclosure,then measures the privacy benefits brought by suppressing trajectories or adding dummy trajectories with single point gain,finds the most appropriate anonymous location(or sequence).Then,uses the anonymous method with larger single point gain to anonymize the location(or sequence)till the user's privacy tolerance is satisfied,there is no trajectories in the dataset that may reveal privacy.Finally,the effectiveness of the proposed algorithm is proved by comparing the proposed algorithm with other algorithms through simulation experiments in the simulation experiment.2.In the privacy protection of high-dimensional trajectories,as the distance between trajectories are difficult to measure,and it is also hard to converge when clustering highdimensional trajectory using traditional clustering algorithm.This thesis proposes a trajectory clustering algorithm aiming at the generalization of high-dimensional trajectory sensitive attributes.Firstly,the algorithm uses a common subsequence to calculate the similarity between trajectories.Then establishes the adjacency table and obtain the density of each trajectory.At last,in order to get rid of the randomness of selecting the center trajectories,the trajectories with high density are selected as the initial cluster center.3.In view of the fact that the attackers have mastered part of the trajectories as background knowledge,and data exchange has been carried out among multiple attackers.A personalized privacy-preserving method for trajectory data publication based on sensitive attribute generalization and location perturbation is proposed.First,frequent patterns are mined by clustering algorithm,and the sensitive attributes carried by trajectory are personalized and generalized in frequent pattern group.Then,the trajectory sequence is interfered based on the common subsequence to resist the attack launched by the attacker.The simulation results show that this method can protect the privacy of trajectory and effectively solve the problem of privacy disclosure in trajectory data publishing under the joint attack of attackers.