Personalized Trajectory Privacy Protection Method Based On Differential Privacy

As the third wave of informatization construction is coming,people start an intelligent stage characterized by in-depth data mining and fusion applications.Through data mining,operators can provide people with better life services,but they also face the risk of privacy leakage.In particular,the trajectory data is highly sensitive.Through the trajectory data,an attacker can learn sensitive information such as the user's living habits,religious beliefs,and health status.Such sensitive information can easily damage the user's reputation,physical and mental health and endanger the user's vital interests.Therefore,making the data more usable while protecting the privacy of user trajectories from being leaked has become a current problem to be solved.Among the current trajectory privacy-protection technologies,differential privacy technology is favored by scholars because it is not affected by the attacker's background knowledge and has a strict mathematical definition.However,the traditional differential privacy trajectory protection technology still has some shortcomings.The first problem is that directly adding noise to the overall trajectory will result in low data availability and fail to meet the individual needs of users.The second problem is the road network environment.Trajectory privacy has its particularity,and the release of one location point may reveal other locations of the user.This thesis proposes a personalized trajectory privacy protection method based on differential privacy in response to the above problems.For the first problem,I use the TFIDF algorithm to filter out the user's sensitive locations,add noise to sensitive locations,publish non-sensitive locations directly.While meeting the user's personalized privacy protection needs can significantly improve the availability of data.For the second question,characterize the privacy risk of each non-sensitive location.Noise is also added to nonsensitive locations with high privacy risks to protect the privacy of users' trajectories further.Experiments on real trajectory data set GeoLife and T-drive show that the algorithm in this thesis can maintain better data availability while protecting the privacy of user trajectories.