Research On Data Masking Technology For Internet Of Things

With the rapid development of the Internet of Things(Io T)technology,the Io T needs to collect and link user data to provide personalized experiences and services.This has led to frequent occurrences of privacy leaks in the Io T in recent years.Leakage of sensitive data will cause all kinds of troubles to users,such as receiving various spam messages,using sensitive data for fraud and so on.How to protect the data privacy of the Io T users and make the services provided by the Io T comply with relevant privacy protection laws is an urgent issue to be solved.Data masking can achieve reliable protection of sensitive data and protect user privacy.Differential privacy is a type of data masking technology that can realize the quantification of privacy protection,and can resist powerful background knowledge attacks,and is more suitable for application in Io T scenarios.Due to that differential privacy may disturb the data and reduce data availability for the Io T to provide users with intelligent services,it is necessary to propose the method of improving data availability of differential privacy on the basis of protecting user data privacy.Make it efficiently applied to Io T scenarios.Considering the application scenarios of wearable devices in the Io T,this thesis proposes a personalized local differential privacy method.When the data aggregation server is not trusted,the user's sensitive data is processed to protect the privacy of the user in the wearable device.According to the different needs of different users for sensitive data,personalized privacy protection is realized.Then,the numerical data in the wearable device are perturbed through calculation and reasonable combination of the random response mechanism and the piecewise mechanism to reduce the noise variance in the worst case.For the classification data of wearable device,Progressive k-ary Randomized Response mechanism is proposed by calculating the optimal solution of frequency estimation variance.The mechanism can be used to perturbate the classification data.Solve the problem that the size of attribute domain has a great influence on frequency estimation in k-ary Randomized Response mechanism.In addition,the method is extended to multi-dimensional data,and the random sampling technique is used to calculate the sample value m,which balances the data availability and the degree of privacy protection.On the basis of protecting the user's data privacy,the data availability is improved to the maximum extent.Finally,through mathematical analysis and simulation experiments,the effectiveness and availability of personalized local differential privacy method for wearable devices are analyzed and verified from two aspects of numerical data and categorical data respectively.