Research On Location Privacy Mining And Protection Technology For Location Services

With the continuous development of mobile Internet and the widespread popularity of mobile intelligent devices,location-based services(LBS)are increasingly popular among people.Location-based services are mainly based on location service providers accessing users’ geographical location information through GPS and other positioning technologies,and providing users with corresponding services,such as path navigation,point of interest query and weather query.However,there are still some problems with LBS in reality.On the one hand,some malicious users will cover up their real location by using virtual location programs to generate fake locations,so as to achieve their own profit,such as remote check-in and regional concessions.On the other hand,since the service is usually an untrusted entity,the service may analyze and trade the obtained user location information,which will eventually lead to the leakage of the user location information.Therefore,how to mine users’ location information and how to protect their location privacy are urgent problems to be solved in the actual operation of LBS.Aiming at these two problems,the thesis makes the following research work:(1)Aiming at the problem of malicious users using fake location to hide real location,we proposed a trajectory similarity hybrid networks(TSHN)to mine the real location of users.Specifically,this method firstly uses an improved trajectory-to-image(i T2I)encoding scheme to convert historical trajectories into image data,which is used to extract movement features of trajectories,such as trajectory shape,trajectory direction and acceleration.Then,the individual characteristics of users are extracted from the historical trajectory data,such as the point of interest,starting and ending time and commuting time.Finally,the moving characteristics and individual characteristics of the trajectory are taken as input,and the similarity between the historical trajectory and the target trajectory is calculated using convolutional neural network and fully connected network,so as to judge the real location of the user.The experimental results show that the recognition rate of fake trajectory by TSHN model reaches 97%.(2)Aiming at the problem that traditional location privacy protection schemes cannot resist data mining attacks,we proposed and implemented a location privacy protection system based on fake trajectory,called Another Me.In this system,the point of interest recognition algorithm is used to identify the point of interest from the historical trajectory,and the real user is modeled based on the point of interest information.Then,based on Amap API,the point of interest mapping algorithm is used to generate k fake users similar to real users but with different geographical locations,so as to hide the location information of real users.Then,through the fake trajectory generation algorithm,the fake trajectory generated by different fake users of k is generated,and the movement characteristics of the fake trajectory are similar to the real trajectory,so as to cover up the real trajectory.Finally,the proper location and the real location are selected from the fake trajectory and sent to the server at the same time to achieve the acquisition of location services.Experiments show that the recognition rate of fake trajectories generated by Another Me system is only 55%,which improves the degree of privacy protection.