Adaptive Grid Decomposition Model Based On Standard Deviation Circle Radius

With the rapid development of the Internet and the diversification and popularity of mobile terminals,user geographic location information interacts with these terminal moments.When these data are obtained by illegal elements,data analysis and mining of these data sets can be used to infer the leakage of the user's private information,such as the user's home address,health status,hobbies and other privacy information.Therefore,the protection of private data when users publish data sets is a challenging hot issue.This paper focuses on the current privacy protection of spatial datasets based on differential privacy.Through the review and analysis of existing differential privacy protection models and algorithms,it is pointed out that in the distribution of spatial datasets based on meshing.There are still some improvements.In the current research,the distribution characteristics of the data set are often neglected or not fully considered;thus,in the stage of adding noise,the privacy protection requirements of the data set are not taken into consideration,and a uniform-scale noise is often added,which is likely to generate large noise.The error,which leads to a large relative error,reduces the query accuracy of the data set,and does not take into account the user's query granularity.In the multi-layer meshing,large query errors may occur.In this paper,an adaptive meshing model based on standard deviation circle radius is proposed for the above problems.The model fully considers the distribution characteristics of the dataset and the user's query granularity.In the noise phase,according to the dataset grid of different distribution features,Corresponding noise is added,filtering and binning are used to reduce noise error in multi-layer meshing,and post-processing is used to improve the accuracy of range query.The main research work of the thesis is as follows:(1)In order to fully consider the distribution characteristics of the data set,we can find a quantitative description of the data set distribution characteristics.In this paper,the degree of dispersion of the data set is described by calculating the standard deviation circle radius of the data set in each grid after meshing,and then the quantitative calculation of the distribution characteristics of the data set is realized.(2)In order to realize the on-demand allocation of privacy budget,this paper introduces the concept of privacy protection requirements,which is represented by the ratio of the standard deviation circle radius of the data grid to the sum of the standard deviation circle radii of all data grids of the layer data layer.The privacy protection needs,and then portrays the privacy protection requirements of the data grid,so as to realize the on-demand distribution of privacy budget according to the privacy protection requirements,and finally realize the data grid with different distribution characteristics to dynamically add different scale noise according to different privacy budgets.(3)In order to reduce the noise error and consider the user's query granularity,the mesh is filtered in the multi-layer meshing.If the original count of the mesh is 0,the added noise is rounded to 0;then the mesh is performed.The buckets are divided into similar buckets into the same bucket,and noise is added to each bucket's privacy budget to reduce noise errors.Finally,in order to improve the query accuracy of the data set,this paper proposes a post-processing method,which enhances the accuracy of the query result by constraining the processing operation.Ultimately,the effect of improving the usability and query accuracy of the data set is achieved.(4)Based on the above theory,this paper proposes an adaptive grid decomposition model based on standard deviation circle radius.The performance comparison of the algorithm model was carried out through experiments.The experimental results show that the algorithm model can effectively reduce the relative error and improve the query accuracy and usability of the data set.