Research On Trajectory Privacy Based On Differential Privacy

With the rapid development of intelligent terminal and mobile communication technology,location-based services(LBSs)applications such as navigation,query and recommendation of interest points,takeout,check-in and finding nearby people are growing,which brings great convenience to people's daily life.When users get services,they need to provide location information to service providers.However,location information is related to sensitive information,and attackers can get sensitive information such as home/work address,consumption level,health status and living habits from location information.Therefore,how to obtain location service without disclosing location privacy is a key problem to be solved in LBS applications.In recent years,location differential privacy overcomes the shortcomings of traditional k-anonymity model,which dependents on the attacker's background knowledge and leads to its security can not be strictly proved,provides a strict privacy quantification model,and can control the level of privacy by adjusting the privacy parameters,so it attracts the attention of scholars.On the one hand,the concept of geo-indistinguishability is proposed,and a noise(false position)generation mechanism is designed to meet this characteristic.The closer any two positions are,the more similar the probability of output the same noise is.Therefore,the attacker cannot distinguish the real location,and the user's location information is protected.On the other hand,from the attacker's point of view,the attacker's differential privacy is proposed,which quantifies the degree of the attacker's inspiration by observing the output of the disturbance mechanism as the privacy leakage.However,it is still in the early stage to solve the privacy problem of trajectory by using the location differential privacy model,and there are still many problems such as balance of privacy and availability,location correlation,accumulation of privacy cost and quantification of privacy of trajectory.Therefore,the location differential privacy technology is further improved to better solve the problems of three LBS trajectory privacy protection technologies.The specific research contents are as follows:Firstly,aiming at the problems of balance between privacy and availability and location correlation in the privacy protection technology of expanding query area,this paper proposes a differential privacy trajectory protection mechanism based on similar map.Firstly,the?-similar map model is proposed to balance privacy and usability.Secondly,a delay query scheme based on differential privacy is proposed to solve the location correlation of continuous queries.Secondly,aiming at the problems of distribution correlation and balance between privacy and availability for independent applying the geo-indistinguishability in the continuous queries,a differential privacy trajectory protection mechanism based on distribution correlation is proposed.Firstly,the existence of distribution correlation is found,and the minimum probability density of noise distribution is used as the overlap density to calculate it.Secondly,based on the distribution correlation,an improved query mechanism of geo-indistinguishability is proposed.This mechanism uses the predicted historical publishing location to replace the noise location with strong distribution correlation,and reduces the accumulated privacy cost.Finally,we prove that each query location in the trajectory satisfies?-DP_d and existes a lower bound for the availability.Finally,in order to solve the problems of trace privacy quantification and worse availability and privacy in the prediction mechanism,a differential privacy trace protection mechanism based on prediction and sliding window is proposed.Firstly,markov chain and exponential perturbation method are used to predict the location which satisfies the differential privacy and spatio-temporal security,and service similarity map is introduced to detect the availability of the location.On this basis,the trajectory privacy budget allocation mechanism based on w sliding window is designed to ensure that any continuous w queries in the trajectory meet the?-differential privacy and solve the trajectory privacy problem of continuous queries.In addition,a privacy customization strategy is designed based on the sensitivity map.By customizing the privacy sensitivity of semantic location,the privacy budget can be customized to improve its utilization.