A Dynamic Privacy Preserving Method On Protecting Trajectory Data

The trajectory data can depict people's mobile behavior.It riches in significant spatio-temporal properties,and lurks great values to be excavated.This type of data has become fatal resource of competition in all walks of life.In the research of data mining field,to excavate the rich fund of knowledge from trajectory data has always been in the leading edge.The trajectory model caries sensitive information such as the pattern of working and living for the people,which can be regarded as a typical privacy data.One of the most important problems in data mining of trajectory is how to protect the privacy security of people's mobility behavior.The research on trajectory privacy preserving is a frontier hot topic in the field of privacy protection.However,the past achievements principally concentrated on snapshot trajectory data and has already been difficult to meet the demand for dynamic characteristics of the trajectory data.There may be a close connection between different historical versions of trajectories.An attacker can undertake to conjoint analysis and reasoning from different historical versions of trajectories and disclose people's privacy information.In this scenario,it is fail to protect the privacy for the traditional preseving approaches with static policy.To solve the problem of privacy leaks caused by group related attack and historical joint strike in the process of trajectory release,this thesis researches a trajectory privacy protection method,which is taking into account the dynamic characteristics of trajectory.The dominating works of this study include three aspects.Firstly,this study introduced the maximum information coefficient said the latest achievements of data correlation to depict with nonlinear characteristics of the correlation degree between different historical versions of trajectories.This novel coefficient avoids the shortcomings of traditional methods which lack of nonlinear description ability.Secondly,a depth model for detecting associated trajectories was proposed in order to improve the pertinence and efficiency of trajectory protection.Futhermore,an implementation of this model was given based on the pseudo relevance feedback ideology.By this detection model,the trajectories with high sensitivity can be detected for preserving pertinently.Thirdly,a dynamic anonymous algorithm was proposed for preseving the privacy of trajectory data.This algorithm is based on the dynamic characteristics of trajectory and can preseve the privacy security of high-sensitive trajectories which are the vulnerable trajectories attacked by the group related attack and historical joint strike.In addition,this study also conducted simulation experiments on real data sets and synthetic data sets to validate the preseved data availability and algorithmic efficiency.