可预测的差分扰动用户轨迹隐私保护方法
|
摘要
连续查询时由于轨迹位置间的相关性,满足差分隐私定义要求的拉普拉斯混淆机制在查询次数较少时可以很好地起到位置隐私保护作用,但独立向每个真实位置添加噪声导致隐私预算水平迅速消耗.针对这一问题,提出一种可预测的差分扰动用户轨迹隐私保护方法,由预测函数、测试函数和噪声机制三部分构成,同时使用预算管理器配置每步所需参数.如果预测函数生成的干扰位置通过测试函数,则直接使用该位置请求服务,否则通过噪声机制重新生成一干扰位置.实验结果表明,该方法比单独向每个位置添加噪声在轨迹偏移度和隐私预算消耗率上均具有较高优势,实现了隐私保护度与服务质量的平衡.
During continuous query,the Laplacian obfuscation mechanism that meets differential privacy definition can play a good role in position privacy protection when the number of queries is small,but the independent noise is added to each real position results in a privacy budget level quickly consumes due to trajectory position correlation. To solve this problem,this paper proposes a predictable differential perturbation user trajectory privacy protection method consisting of a prediction function,a test function and a noise mechanism.At the same time,a budget manager is used to configure the parameters required for each step. If the perturbation position generated by the prediction function passes the test function,it is directly used to request the service,otherwise,an perturbation position is regenerated by the noise mechanism. The experimental results show that this method has higher advantages in track offset degree and privacy budget consumption rate than independently adding noise to each location and achieves a balance between privacy protection and service quality.
During continuous query,the Laplacian obfuscation mechanism that meets differential privacy definition can play a good role in position privacy protection when the number of queries is small,but the independent noise is added to each real position results in a privacy budget level quickly consumes due to trajectory position correlation. To solve this problem,this paper proposes a predictable differential perturbation user trajectory privacy protection method consisting of a prediction function,a test function and a noise mechanism.At the same time,a budget manager is used to configure the parameters required for each step. If the perturbation position generated by the prediction function passes the test function,it is directly used to request the service,otherwise,an perturbation position is regenerated by the noise mechanism. The experimental results show that this method has higher advantages in track offset degree and privacy budget consumption rate than independently adding noise to each location and achieves a balance between privacy protection and service quality.
引文
[1] Wu Zhen-gang,Sun Hui-ping,Guan Zhi,et al. Overview of location privacy protection for continuous spatial query[J]. Journal of Computer Applications,2015,32(2):321-325.
[2] Xu T,Cai Y. Location anonymity in continuous location-based services[C]//ACM Internationa Symposium on Advances in Geographic Information Systems,ACM,2007:1-8.
[3] Palanisamy B,Liu L. Attack-resilient mix-zones over road networks:architecture and algorithms[J]. Mobile Computing IEEE Transactions on,2015,14(3):495-508.
[4] Xu Z,Zhang H,Yu X. Multiple mix-zones deployment for continuous location privacy protection[C]//Trustcom/bigdatase/ispa,IEEE,2017:760-766.
[5] Niu B,Gao S,Li F,et al. Protection of location privacy in continuous LBSs against adversaries with background information[C]//International Conference on Computing,Networking and Communications,IEEE,2016:1-6.
[6] Dwork C. Differential privacy[C]. Proc of the 33rd Inter-national Colloquium on Automata,Languages and Progra-mmig,Berlin:Springer,2006:1-12.
[7] Xiao Y,Xiong L. Protecting locations with differential privacy under temporal correlations[C]//ACM Sigsac Conference on Computer and Communications Security,ACM,2015:1298-1309.
[8] Wang Hao,Xu Zheng-quan,Xiong Li-zhi,et al. CLM:differential privacy protection method for track release[J]. Correspondence Journal,2017,38(6):85-96.
[9] Li C,Palanisamy B,Joshi J. Differentially private trajectory analysis for points-of-interest recommendation[C]//IEEE International Congress on Big Data,IEEE,2017:49-56.
[10] Hu De-min,Zhan Han. Homogeneous incremental near-est neighbor query method based on differential perturbation for location privacy protection[J]. Journal of Chinese Computer Systems,2018,39(7):1482-1486.
[1]吴振刚,孙惠平,关志,等.连续空间查询的位置隐私保护综述[J].计算机应用研究,2015,32(2):321-325.
[8]王豪,徐正全,熊礼治,等. CLM:面向轨迹发布的差分隐私保护方法[J].通信学报,2017,38(6):85-96.
[10]胡德敏,詹涵.差分扰动的均衡增量近邻查询位置隐私保护方法[J].小型微型计算机系统,2018,39(7):1482-1486.
[2] Xu T,Cai Y. Location anonymity in continuous location-based services[C]//ACM Internationa Symposium on Advances in Geographic Information Systems,ACM,2007:1-8.
[3] Palanisamy B,Liu L. Attack-resilient mix-zones over road networks:architecture and algorithms[J]. Mobile Computing IEEE Transactions on,2015,14(3):495-508.
[4] Xu Z,Zhang H,Yu X. Multiple mix-zones deployment for continuous location privacy protection[C]//Trustcom/bigdatase/ispa,IEEE,2017:760-766.
[5] Niu B,Gao S,Li F,et al. Protection of location privacy in continuous LBSs against adversaries with background information[C]//International Conference on Computing,Networking and Communications,IEEE,2016:1-6.
[6] Dwork C. Differential privacy[C]. Proc of the 33rd Inter-national Colloquium on Automata,Languages and Progra-mmig,Berlin:Springer,2006:1-12.
[7] Xiao Y,Xiong L. Protecting locations with differential privacy under temporal correlations[C]//ACM Sigsac Conference on Computer and Communications Security,ACM,2015:1298-1309.
[8] Wang Hao,Xu Zheng-quan,Xiong Li-zhi,et al. CLM:differential privacy protection method for track release[J]. Correspondence Journal,2017,38(6):85-96.
[9] Li C,Palanisamy B,Joshi J. Differentially private trajectory analysis for points-of-interest recommendation[C]//IEEE International Congress on Big Data,IEEE,2017:49-56.
[10] Hu De-min,Zhan Han. Homogeneous incremental near-est neighbor query method based on differential perturbation for location privacy protection[J]. Journal of Chinese Computer Systems,2018,39(7):1482-1486.
[1]吴振刚,孙惠平,关志,等.连续空间查询的位置隐私保护综述[J].计算机应用研究,2015,32(2):321-325.
[8]王豪,徐正全,熊礼治,等. CLM:面向轨迹发布的差分隐私保护方法[J].通信学报,2017,38(6):85-96.
[10]胡德敏,詹涵.差分扰动的均衡增量近邻查询位置隐私保护方法[J].小型微型计算机系统,2018,39(7):1482-1486.