Research On Local Differential Privacy Protection Method For Dynamic Continuous Location

The wide application and analysis of location big data have accelerated the development of intelligent terminals.Although the real-time interaction of location big data provides users with a variety of convenient services,there are also potential risks of personal location information leakage.How to make the location service information as accurate as possible while protecting the user’s location privacy is a key issue that needs to be considered.User location data is highly correlated in time and space,and dynamic location privacy protection can be realized more clearly by modeling and analyzing this correlation.In this thesis,the local differential privacy method is combined with the Hidden Markov Model and the privacy budget absorption method respectively in the process of real-time location data publishing to realize the privacy protection and publishing of dynamic location data.Dynamic location big data has temporal and spatial correlation,and its potential correlation can be extracted by some time series models.To solve these problems,this thesis proposes a local differential privacy dynamic location protection method based on Hidden Markov Model.Firstly,combined with the spatial-temporal correlation in the process of dynamic location changing,the time series and privacy protection security region based on the Hidden Markov Model were constructed,and the optimization of local differential privacy perturbation region after the user’s location update was realized.Secondly,a dynamic position continuous perturbation algorithm based on Hidden Markov Model and a local differential privacy random response mechanism are designed to perturb the position points in the preferred area,and achieve dynamic local differential privacy protection of the user’s location.Finally,the experiment and analysis on the actual location trajectory data set prove that the proposed method can achieve higher aggregation accuracy and statistical availability on the premise of realizing the local differential privacy protection of dynamic location data.In order to capture the relevant features of dynamically published data and improve the publishing availability of dynamic location data stream,this thesis proposes a local differential privacy dynamic protection method for location data stream.Firstly,the sliding window model and the local differential privacy perturbation mechanism RAPPOR were used to perturb the location data in the window,which realized the local differential privacy protection of w-events.Secondly,the local privacy budget absorption method is used to realize the dynamic privacy budget allocation process through the privacy decision mechanism and privacy disturbance mechanism.Finally,the privacy proof of the proposed method is carried out,and the experiment and analysis on the real spatio-temporal location data set prove that the proposed method is better in the perturbation result deviation and the accuracy of the location aggregation data.