Location Privacy Preservation Through Kernel Transformation

The mobile internet and positioning technology are developing rapidly today,and smart devices are becoming increasingly popular,which has greatly promoted the development of Location Based Service(LBS).Generally,the location service requires users to disclose their exact location to the location service providers,which gives attackers the opportunity to successfully infer personal sensitive information by using location information,and thus pose corresponding risks to personal privacy.Therefore,the location privacy information of users in location services needs to be protected.At present,the privacy protection methods mainly include spatial anonymity protection methods,differential privacy-based methods,and encryption methods.However,in the existing methods,it is difficult to ensure the accuracy of the location query while ensuring the location privacy of the user,or it has a large calculation and communication burden.This paper proposes a new framework for location privacy protection based on fog computing,which uses kernel function mapping to provide location privacy protection for users in location services.This framework mainly includes three parts: location server,user and fog computing device.The location server uses a kernel function to generate a private map for each user,and then the user directly submits a location query to the fog computing device in the area.Without conspiracy,neither the LBS server nor the fog computing device can infer the user's real location.In most existing location service frameworks,the location server is at the center of the service framework and bears the burden of interaction with users and third parties.Therefore,it often becomes a performance bottleneck for the entire service system,or a single point from where an attacker can obtain data or break the system.The framework of this paper is a distributed service framework.The fog computing device and server share the computing and communication load of the system to ensure the load balance of the entire system.This paper first applies kernel functions to the field of privacy protection,and designs a new spatial mapping method based on kernel functions.The user uses the parameter vector selected by themselves to map the position of the user and the point of interest from the two-dimensional space to the high-dimensional space while retaining the relative distance between the transformed points.A point of interest query is performed,but the attacker cannot infer the original location based on the transformed high-dimensional coordinates.At the same time,an earth projection method suitable for the framework of this paper is also designed in this paper to reduce the errors caused by traditional earth projection methods used for position conversion.Finally,this paper uses real data set SimpleGeo Place and Yelp commercial data set for experiments,studies the effect of privacy protection parameters on the performance of the proposed framework,and verifies the superiority of this method by comparing the effect of location query with other privacy protection methods.