Research On Privacy Preserving Technique In Trajectory Data Publishing Based On Segment Clustering

In recent years,the growing popularity of location technologies and location-based applications,bring great convenience to people.At the same time,application providers and research institutions accumulate a large number of user's trajectory data.The analysis of trajectory data has important application value for government and enterprise,and also has widely application prospect in traffic monitoring,city planning,mobility management and marketing strategy.As the trajectory itself contains abundant privacy information such as user attributes,the privacy preserving issues for user tracks become a hot research problem in trajectory data publishing.Due to the characteristics of trajectory data such as large-scale,high-dimension,and rich-background,the research on privacy preserving issues is facing severe challenges.This thesis mainly focuses on the privacy protection issue for the publishing of large-scale trajectory data.Firstly,considering the uncertainty of the trajectory in real life,introuduce the(k,?)-anonymity model,a spatio-temporal editing perturbation algorithm is proposed to solve the trajectory anonymity problem.On this basis,a trajectory privacy preserving algorithm based on segment clustering is proposed and implemented.In the trajectory privacy preserving algorithm,the raw database is partitioned into blocks firstly.And then,the blocks of the trajectories are partitioned into segment based on the MDL principle.Then,the segments of the trajectories are clustered by using LSTD distance function.Finally,these segments are anonymized with cluster-constraint strategy.The problem of lack of diversity in the anonymous group of traditional trajectory privacy preserving algorithm is solved,and the re-clustering attacks against the characteristics of publishing data can be prevented,the difficulty in anonymous and the low usability of data are solved.Finally,in the section of experiment,the performance of the proposed privacy preserving algorithms are assessed,in terms of the data quality and data efficiency,and compare the proposed algorithm with classic NWA algorithm.The experiment results show that the(k,?)-anonymity algorithm based on spatio-temporal editing has smaller anonymous cost in most case.The trajectory privacy preserving algorithm based on segment clustering can realize the anonymity of large-scale trajectory database,and can greatly improve the level of privacy preserving with a minimum data quality cost.