Trajectory Privacy Preserving Approach Based On Markov Distance

With the development of mobile Internet and the widespread application of positioning technology,a large amount of space-time trajectory data have produced.Analyzing and mining the trajectory data can benefit to our live,such as urban transport planning.However,the direct release of untreated trajectory data may result in the disclosure of personal privacy of the mobile object.Therefore,in the data publishing how to ensure the requirement of user's privacy requirements and how to make the data published with high availability is one of the hotspots of track privacy preserving.Trajectory k-anonymous technology achieves a better balance in data availability and privacy protection,However,most of existing method of trajectory k-anonymous assume the moving object do uniform linear motion between adjacent points of position,this assumption is too ideal.To solve this problem,in this paper,Piecewise cubic Hermite interpolation is used to trajectory reconstruct.Piecewise cubic hermite interpolation avoid the ideal assumption,first derivative of piecewise cubic hermite interpolation polynomial is consecutively which make sure the velocity of moving object will not mutate in sampling point and piecewise cubic hermite interpolation avoid the intense oscillation of high order interpolation.In order to construct trajectory k-anonymous collection,first,putting trajectories which have same start and end time into a same equivalent class set.Then considering the overall distribution of trajectory sampling points,this paper propose using Markov distance to calculate the distance of two trajectories.According to this measurement method,we construct space-time similar trajectory k-anonymous set of the trajectory in the equivalent class.Finally we anonymous the trajectory of the k-anonymous set by disturbing algorithm based on publishing atomic trajectory.The trajectory in k-anonymous set is not distinguishable,so the goal of trajectory privacy protection can be achieved.Using different data sets design and implement experiment.the results show that the proposed TPP-MD method is feasible and effective in different data sets.