Research On Location Data Privacy Preservation For Mobile Internet

Along with the fast development of the mobile Internet technology and the booming popularization of the smart mobile devices,mobile phones,personal digital assistants(PDAs),and in-vehicle navigation systems,which are based on the telecommunications of the mobile Internet have become the necessaries of modern daily life.The smart mobile devices are in light weight and small volume,so that the users could carry them in their daily life,and in the course of their daily life with these devices,a large amount of the users' personal location data will be collected.The mobile applications providers exploit the location data collected in real time,and develop various kinds of location based services(LBSs),such as Gaode Map,Meituan,Mo Mo,etc.These LBSs has greatly facilitated people's lives and become the star products in the applications' market.Besides,in academic research,mining the mobility data could promote the research on human behavior,even provide solutions to the global problems,like controlling the COVID-19 epidemic.However,sensitive information,such as home address,occupations,or even the physical conditions and religious belief are involved in the mobility data,therefor,exposing such mobility data directly would cause privacy issues,increase social instability and decrease the users' desire in using the LBSs,and even hold back the development of big data era.This paper studies the privacy issue existing in the mobile Internet from four aspects: quantifying the utility of the released mobility datasets,quantifying the privacy preservation of the released mobility datasets,improving the data transmission efficiency and raising the sampling rate of the raw mobility datasets,and proposes four complementary privacy preserving methods for mobility datasets,the main content involving:Firstly,in order to solve the problem that the current location based services applications obtain the users' location data collection permission in a single way,which is easy to be attacked by the malicious,a user authentication method based on the user's breath rate is proposed to enhance the privacy preservation strength of the process of individual users granting the permission to collect location data.This method uses the front camera sensor of the mobile intelligent devices to collect the users' face images,and obtains their breath rate throuth them,which could reduce the process of human-computer interaction.At the same time,combined with compressive sensing theory,this method reduces the frequency of data acquisition and alleviates the problem of limited computing power of mobile intelligent decives.In addition,a sparse dictionary training method is proposed to recover the compressed data to normal frequence.Finally,the feasibility and effectiveness of this method are verified by simulation experiments.Besides,aiming at the problem that excessive attention is paid on the privacy preserving capacity when protecting the mobility datasets,while the utility of the released datasets is ignored,we propose a method to protect the privacy of the released mobility datasets based on the data structure Count-Min Sketch(CMS).This method utilizes the CMS to aggregate the frequencies of different locations records existing in the raw mobility datasets and releases the estimated population distributions from the CMS to protect the privacy of the raw mobility datasets.As false positive occurs when aggregating mobility data with CMS,there will be difference between the released estimated and real population distributions,and the released population distributions could achieve privacy preservation.Meanwhile,this method could quantify the utility of the released datasets by computing the probability of false positive happened in CMS,so that,this method could satisfy different utility requirements by tuning the value of parameters.The results of the simulation experiments verify the feasibility and effectiveness of this method.Moreover,to balance the tradeoff between the utility loss and privacy preserving capacity of the released mobility datasets,a differential privacy based method for preserving the privacy of the mobility datasets is proposed.This method firstly aggregates the raw mobility data with CMS,and then,employing Laplace mechanism to add noise to the CMS,which could make the CMS satisfy-differential privacy.And in the last step of this method,the privacy preserving population distributions aggregated by the differentially private CMS is released to support various LBSs,and achieving the goal of protecting the privacy of the mobility datasets.Both CMS and Laplace mechanism are employed in this method to preserve the privacy of mobility datasets,and thus the excessively strong correlation between the privacy preservation and utility of the released datasets could be weaken.In this method,the users also could tune the value of the parameters of Laplace mechanism to support different privacy requirements.Finally,the simulation experiments' results show that comparing to other state of the art methods,this method could achieve better privacy preserving capacity under the same utility loss of the released datasets,and verifying the feasibility of this method.The last point,as the energy and computation capacity of these smart mobile devices are limited,improving the data transmission efficiency has become a hot issue.Focusing on such issue,a differentially private double sketches method for preserving the privacy of mobility datasets is proposed.For the CMS based privacy preserving method for mobility datasets,the size of the sketch effects the utility preservation of the released datasets directly,and thus it is inappropriate to improve the data transmission efficiency by reducing the size of sketch.This method introduces two collaborative sketches,global sketch and temporal sketch.The temporal sketch is not involved in the data transmission stage,and therefore,its size could be set relatively large to guarantee the utility of the data,which will be aggregated by the global sketch.With the help of temporal sketch,the size of the global sketch could be set smaller,which could achieve the goal of high utility and efficient data transmission.Meanwhile,this method employs Laplace mechanism in the data transmission stage,which makes the transmitted global sketch satisfy differential privacy,and provides double protection to the mobility datasets.At last,the simulation experiments results illustrate that under the same utility preservation,the proposed method could improve the data transmission efficiency,and verify its feasibility and effectiveness.