Research On A Differential Privacy Histogram Publishing Technology Based On Staircase Distribution

With the rapid development of smart terminal equipment and 5G network technology,related applications based on geographic location services have been widely used in various fields of social life production.Through the relevant modeling of the shared data uploaded by the user,the user can bring many conveniences to him in the process of using the application after loading the model.But with the enhancement of people's personal privacy protection awareness,how to protect sensitive information in location data from malicious intrusion has become an urgent problem to be solved.Differential privacy can solve the problem of insufficient traditional privacy protection.Differential privacy quantitatively analyzes the degree of privacy risk through mathematics,and adds random noise to the published results,so that the relevant statistical characteristics of the entire data set have little effect,and the attacker cannot pass the The data reversely pushes out individual user information.The data publishing forms based on differential privacy are mostly in the form of histograms.When performing interval query on the histogram,the long span of the interval will lead to excessive accumulated noise,which will reduce the availability of published data.This paper studies the optimization of interval grouping to improve the published data Availability.The content of this article is as follows:(1)Adding Laplacian noise is the standard method of differential privacy.Randomly disturb the data before publishing the numerical data.The Laplacian mechanism is currently widely used as the most basic method of differential privacy.There is no relevant research work on optimizing the Laplace mechanism.This article introduces a method that can replace Laplace,the stepwise segmented distribution mechanism.Given the same privacy budget,the running time of this mechanism is significantly reduced,and the performance is effectively improved.(2)This paper takes static trajectory data as the research object,and proposes a differential privacy histogram publishing algorithm DPHR based on step distribution.In order to improve the similarity within the group,a scoring function that fully considers the counting distance and azimuth distance is designed,and this scoring function is used to sort the initial data using an exponential mechanism that meets the difference privacy,and then cluster the sorted histogram Grouping,without specifying the number of groups,greedy clustering grouping is based on the principle of reducing the overall error,adding random noise generated by the stepped distribution mechanism to the grouped clusters,and using differential privacy theory to prove that the method meets the difference Privacy definition.Finally,simulation comparison experiments were performed on the two data sets.The experimental results show that the algorithm has high data availability.