Research On Privacy-preserving Methods Based On Local Differential Privacy In Edge Computing

In recent years,with the development of the Internet of Things,the delay caused by network transmission will lead to inefficient data processing.The emergence of edge computing can effectively reduce the delay phenomenon in the process of data transmission and improve the data processing capabilities.Nowadays,due to the rapid rise of many sciences and technologies,it has become easier to process large amounts of data quickly to extract useful information from it.However,at the same time,in the environment of edge computing,because user-related data is often handed over to a semi-trusted authorized entity for processing,the user's privacy will be leaked.Therefore,how to use an effective privacy protection mechanism to protect private data from being leaked has become a major research issue in the Artificial Intelligence(AI)environment.Aiming at the data-sensitive nature of edge computing,its privacy protection has become the main research direction of this thesis.Privacy-preserving is an important research issue in edge computing.Its research content mainly includes three aspects,namely identity privacy-preserving,data privacypreserving,and location privacy-preserving.This thesis mainly studies the protection mechanism of location privacy,and designs edge computing location privacy-preserving algorithm.According to the shortcomings of previous research methods,how to protect privacy information from the perspective of centralization and decentralization is researched.The main research contents of this thesis are summarized as follows:1.In order to reduce the loss of service quality caused by the location privacypreserving algorithm,a centralized differential privacy-preserving mechanism based on Gaussian white noise is proposed.This privacy-preserving mechanism interferes with malicious attackers' guessing of the user's true location by adding random white noise that obeys the Gaussian distribution.In the real data set,it is compared with Gaussian colored,Laplace noise-based differential privacy mechanism and k-anonymity mechanism,respectively.The experimental results show that the proposed Gaussian white mechanism is effective in privacy-preserving.Compared with the existing privacy-preserving methods,this mechanism not only can better meet the privacy needs of users,but also has more stable data interference.2.In order to solve the problem of untrustworthiness of semi-trusted authorized entities in edge computing environment,a decentralized user location data disturbance method is proposed to protect user privacy.First,the Delaunay method is used to divide the road network space constructed by the data set,and the Voronoi diagram is drawn to determine the location of the edge node.Then,the local differential privacy mechanism is used to allow mobile users to pass through the dimensionality where they are located.Nuoge randomly generates multiple false locations to confuse its true location.Finally,compared with the existing privacy-preserving methods,this mechanism not only can better meet the privacy needs of users,but also has higher data availability.