Research On Location Differential Privacy Protection Based On Edge-Assisted Connection

Edge computing is a new computing paradigm that disposes communication,computing,control and storage services at the edge of the network and manages resources in near-mobile devices,sensors,drivers and terminal devices.Edge computing can transfer some storage and computing tasks from the cloud data center to the edge of the network.However,in the process of data processing at the edge nodes,the issue of personal data privacy protection and security is particularly prominent,which may bring many challenges related to security and privacy.For the leakage of mobile users' location information,the following problems need to be solved.Firstly,the coverage of edge nodes is small.In the process of data processing,users usually establish a connection with the nearest node,so that attackers can predict the location of users through the node location.Secondly,in order to protect the user's location information,it is necessary to add noise to the original data,which will inevitably reduce the availability of the original data.Through the above analysis,in order to solve the problem of user location privacy leakage in the application of edge computing,a local difference privacy location fuzzy framework for edge data analysis is proposed,which is mainly composed of three parts:(1)Edge node query,the main function is to query the connection of node information in the process of data processing,so as to prevent attackers from inferring the location of users based on node information.According to the distribution characteristics of edge data center,this paper considers it as an undirected graph,and users can query the connection information between different terminals and edge nodes according to the query model.Finally,the query model of edge nodes is evaluated from two aspects of algorithm execution efficiency and privacy protection measurement.The results show that the query structure of edge nodes can guarantee the protection of connection information between nodes to a certain extent.(2)Data confusion and local difference privacy are mainly used to ensure the privacy of transmitted data and avoid the disclosure during transmission.The coverage area of the edge node is generally small,and some location-sensitive applications will have great deviation,so they cannot adapt to the edge auxiliary structure well.Based on the geo-indistinguishable difference privacy,the Laplace distribution is used to generate the noise location.(3)Data reconstruction.The main function of this part is to reconstruct the data from disturbance and improve the availability of data.After the query is established,edge nodes need to generate probabilistic fuzzy matrix based on historical data.Considering that the geographical indistinguishable threshold can be expressed by linear constraint,this paper proposes to reduce the selection problem of the optimal probabilistic fuzzy mechanism K to a linear optimization problem,and then use the standard method of linear programming to solve the problem.Using the compressed sensing principle,the missing matrix data values are inserted and the data matrix is reconstructed.One of the preconditions of compressed sensing is to satisfy the low rank,so the rank of the approximate matrix needs to be minimized.However,two parameters,rank bound r and tradeoff coefficient A have great influence on the final estimation quality.In order to obtain the best performance of the algorithm in terms of estimation error,the optimal parameters need to be determined,and the sorting method based on genetic algorithm and the solution method of optimal parameter of tradeoff coefficient are used.The results show that data reconstruction can improve data accuracy to some extent.