Research On Staying Points Mining And Privacy Protection Based On Vehicle Trajectory

With the development of GPS and the popularization of vehicle-mounted intelligent terminal devices,massive vehicle trajectory data has been generated.Mining and analyzing these trajectory data can promote the development of intelligent applications such as vehicle pattern recognition,path planning,and trajectory anomaly detection.However,the trajectory data,especially the stopping point,usually contains a lot of privacy information of the user,and if the data is not protected for privacy and used directly,it will easily leak the user’s personal sensitive information and bring certain security problems.Firstly,an effective vehicle stop mining algorithm is proposed based on clustering technology,and a location privacy protection algorithm is proposed based on the location privacy protection of users in the process of data release.The main research content includes the following three parts:(1)A stopping point perturbation algorithm based on differential privacy technology is proposed.Firstly,use the trajectory compression algorithm based on synchronous Euclidean distance for trajectory preprocessing.In the preprocessing,combined with the spatiotemporal attributes of the trajectory data,the trajectory feature points are screened out by calculating the synchronous Euclidean distance between the trajectories,and connect the feature points to represent the real trajectory to realize the preprocessing of the trajectory data.The proposed stop point mining algorithm is mainly aimed at the preprocessed trajectory data.This paper combines the characteristics of the large latitude and longitude span of the vehicle trajectory data,and considers that only the distance threshold and time threshold are not accurate enough to mine the stop points.Therefore,the trajectory is introduced.speed property.In clustering,by calculating the average velocity of the trajectory,setting the velocity threshold of the trajectory,further judging the stay point,and narrowing the identification range of the stay point,the stay point can be better divided.(2)A stopping point perturbation algorithm based on differential privacy technology is proposed.Firstly,calculate the distance between the mined stay points and the center of the region of interest,and mark the distance as a privacy budget label.Then allocate a privacy budget for each stay point according to the size of the privacy budget label value,add different amounts of noise according to different privacy budgets,and generate a disturbance area for each stay point.After that,calculate the average semantic similarity of the trajectory points in each perturbation area,and eliminate the trajectory points whose similarity is greater or much smaller than the average semantic similarity.Finally,randomly selected the points in the perturbation area to replace the stay points,so as to improve the data availability of the postperturbation trajectory and achieve the purpose of location privacy protection.Through experimental verification on two real vehicle datasets,compared with the comparison algorithm,the stop points mining algorithm and privacy protection algorithm proposed in this paper have certain advantages in performance measurement,which can effectively mine the vehicle stop point location information while protecting its privacy from being leaked.