面向轨迹数据发布的KSDP方案  

KSDP scheme for trajectory data publishing

作　　者：张俊[1] 刘德安 申自浩[1] 王辉[2] 刘沛骞 ZHANG Jun;LIU Dean;SHEN Zihao;WANG Hui;LIU Peiqian(School of Computer Science and Technology,Henan Polytechnic University,Jiaozuo 454000,Henan Province,P.R.China;School of Software,Henan Polytechnic University,Jiaozuo 454000,Henan Province,P.R.China)

机构地区：[1]河南理工大学计算机科学与技术学院,河南焦作454000 [2]河南理工大学软件学院,河南焦作454000

出　　处：《深圳大学学报（理工版）》2023年第2期236-243,共8页Journal of Shenzhen University(Science and Engineering)

基　　金：国家自然科学基金资助项目(61300216);河南省高等学校重点科研资助项目(23A520033);河南理工大学博士基金资助项目(B2022-16,B2020-32)。

摘　　要：轨迹隐私保护中使用k-means算法进行聚类时,对初始值敏感,且聚簇数目的选择具有一定的盲目性,为解决该问题并提高聚类结果的可用性,提出一种结合k-shape和差分隐私的轨迹隐私保护方案KSDP(k-shape differential privacy).首先,对轨迹数据进行划分切割预处理,利用轨迹的时间属性和空间属性对轨迹切割划分,从而提高聚类泛化的质量.其次,使用设定的效用函数对预处理后的轨迹数据进行评判,并对过滤后数据进行聚类泛化操作.最后,在泛化后的数据中加入Laplace噪声,使其满足差分隐私保护模型,进一步保护轨迹隐私.实验仿真结果表明,与传统差分隐私k-means聚类方案对比,KSDP方案有效提高了聚类结果的可用性,并具有一定的性能优势,更好地实现了轨迹数据发布和隐私保护.For clustering applications in the field of trajectory privacy protection,the k-means algorithm is sensitive to initial values and the number of clusters may be somewhat arbitrary.To address these issues and further improve the usability of clustering results,a trajectory privacy protection scheme combining k-shape and differential privacy(KSDP)is proposed.Firstly,the trajectory data is partitioned and preprocessed based on the temporal and spatial attributes of the trajectory to improve the quality of clustering generalization.Secondly,a utility function is used to evaluate the preprocessed trajectory data,and the clustering generalization is performed after filtering the data.Finally,Laplace noise is added to the generalized data to satisfy the differential privacy protection model,so as to further protect the trajectory privacy.The experimental simulation results show that compared with the traditional differential privacy k-means clustering scheme,the KSDP scheme effectively improves the availability of clustering results and achieves better trajectory data publishing and privacy protection.

关 键 词：数据安全与计算机安全 轨迹隐私 差分隐私 k-shape 隐私保护 数据发布 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]