Research On Privacy Protection Algorithm Of Two-dimensional Spatial Data Based On Differential Privacy

With the extensive use of geographic location acquisition technology in mobile devices,the user's behavior habits analyzing from the user's mobile trajectory data(forming spatial datasets)has become a hot point.Attacking the spatial datasets with location information will lead to personal interests and hobbies,behavior patterns,social habits,health status and other privacy information exposed.Therefore,it is a challenging problem for users to share his location information.The differential privacy is a very strict attack model,and it gives a rigorous quantitative definition and proof for the privacy disclosure risk.Therefore,the differential privacy is better than the traditional privacy preserving algorithms,and has been widely used in the privacy preserving of spatial datasets.However,there are some improvements in the spatial data release based on grid hierarchical division.For example,the current research often neglects the uneven distribution of the position data in each grid,and can not fully consider the distribution of data and Data privacy protection requirements;In the noise phase,the query results of grid on each level mixed the same size noise,not according to the privacy requirements in each grid data,which is easy to produce relative error,and thus reduce the data query accuracy.Therefore,this paper presents a noise dynamic mixing algorithm model based on the privacy protection requirement of spatial datasets.The algorithm model quantitatively describes the distribution of data by measuring the degree of data dispersion,and then quantitatively describes its privacy protection requirements.Finally,according to the privacy protection requirement of each data in the grid,the datasets is mixed the corresponding scale noise.In this paper,we take into account the distribution of the data in each grid fully,and calculate the degree of discretization of the data in each lattice by calculating the standard deviation radius of the data in each grid.Then we quantitatively describe the data distribution in each grid.Secondly,in the data is mixed the noise phase,the privacy protectionrequirement of the data is quantified according to the proportion of the standard deviation radius of all the grids in the same level.Set the different privacy needs to dynamically mix the corresponding privacy budget,to achieve different distribution of data dynamically mix different noise values.To achieve the relative error reduction,improve the usefulness of the data.Finally,based on the above theory,a noise dynamic allocation algorithm model(SDC-DP)based on the privacy protection requirement of spatial datasets is designed and implemented.The performance of the algorithm model is verified by simulation experiment.Experimental results show that the algorithm model can effectively describe the data privacy requirements,and can effectively reduce the relative errors.