Research On Enhanced Differential Privacy Protection Technology For User Sensitive Data

With the rapid development of network information technology,various types of user data show explosive growth,and various big data analysis technologies bring convenience to people's lives,but also pose a serious threat to user privacy protection.Differential Privacy protection technology provides a new way to solve the problem of user sensitive information leakage caused by data release and analysis.The main idea is to add random noise to the query output through the output disturbance,so that the attacker can not obtain accurate individual information by observing the calculation result even if it has the greatest background knowledge.However,the existing differential privacy protection technology also has some inadequacies,which reduce the privacy protection for user sensitive data when facing the new scene and the various requirements.For example,(1)The existing differential privacy protection models only provide a uniform level of privacy protection for all query users,so that it cannot provide different privacy and availability data information for different query users;(2)The non-memory of the received query request makes the security protection weakened when faced with repeated query attacks;(3)The existing differential privacy protection mechanism only controls the risk of data leakage to a certain extent,attackers can analyze and obtain accurate user sensitive data by differential attack and probability inference attack when the added random noise falls in a very small interval.In view of the above problems,this paper relies on The National Science and Technology major project “ "5G security protection technology research and verification"(No.2018ZX03002002)and the National Science Foundation of The Research on virtual resource management technology for 5G network slicing(No.61801515).To solve the defects and shortcomings of existing differential privacy protection technology,the paper has carried out many researches on improving the ability of differential privacy protection.from the perspectives of hierarchical query control,data availability analysis,privacy protection specification measurement and probability inference attack analysis.The main research work are as follows:1?A differential privacy protection method based on hierarchical query control is proposed.Firstly,this method calculates the query trust value based on the querier's reputation value and access rights.Then,the query security trust degree is quantitatively analyzed and mapped to different trust levels and privacy protection parameters according to the dynamic nature of the data privacy attribute.Finally,to achieve differential privacy protection and provide different data availability,Laplace random noise obeying the corresponding distribution geatures is added to query results according to the detemined privacy protection budget parameters.The simulation experimental results demonstrate that the proposed method could provide protected data with error rates ranging from 0.1% to 30% for different levels of query users,which releasing the important limitation of differential privacy providing only a uniform level of privacy protection,and solving the privacy protection problem of data query of multi-trust level users.2?A personalized differential privacy protection method for repeated queries is proposed.Aiming at the security risks that the privacy protection effect of existing differential privacy protection technology is weakened in dealing with repeated query attacks,we firstly propose a new privacy protection specification according to the data privacy protection requirements,query user privilege and the number of repeated queries.And then we construct a personalized differential privacy protection algorithm for repeated queries to improve the choice of privacy protection budget parameters.Finally,by adding random noise with different distribution characteristics,the query returns a different degree of generalization,so that the data availability of the returned result data of the subsequent repeated query is not higher than the first query result.The simulation experiments show that the proposed method make the relative error value of the returned results increases with the number of queries,and can provide fine-grained and differentiated privacy protection flexibly for aggregated queries of massive data when responding to repeated query requests.So that it can solve the privacy leakage threat caused by repeated query attacks effectively,3?A differential privacy protection parameter configuration method based on confidence level is proposed.Firstly,we analysis the confidence level of the attacker under the probabilistic inference attack model,and deduce the correspondence between the attacker's confidence with the confidence interval,the noise distribution location parameter and scale parameter.Then,according to the difference of different query user query permissions,the privacy protection budget parameter selection range is determined based on rigorous mathematical derivation according to the privacy probability threshold setting,which can configure reasonable privacy protection parameters to reduce the risk of user sensitive data being compromised.The simulation experimental analysis shows that the proposed method analyzes the correspondence between attacker confidence level and privacy protection parameters based on query privilege,noise distribution characteristics and data privacy attributes,and derives the configuration formula of privacy protection parameters,which configure the appropriate parameters without violating the privacy protection probability threshold.