基于图划分的个性化轨迹隐私保护方法  

Personalized trajectory privacy preserving method based on graph partition

作　　者：杨静[1] 张冰[1] 张健沛[1] 谢静[1] 

机构地区：[1]哈尔滨工程大学计算机科学与技术学院,黑龙江哈尔滨150001

出　　处：《通信学报》2015年第3期1-11,共11页Journal on Communications

基　　金：国家自然科学基金资助项目(61370083;61073043;61073041);高等学校博士学科点专项科研基金资助项目(20112304110011;20122304110012);哈尔滨市科技创新人才研究专项基金资助项目(2011RFXXG015)~~

摘　　要：针对用户对轨迹匿名数据的个性化需求,提出一种基于轨迹间夹角和位置重合的(s,λ)-覆盖个性化轨迹间关联构建方法,并根据轨迹间距离和方向度量轨迹间边权,以构造规模可变的个性化轨迹图模型。同时,将轨迹k-匿名集的构建转化为轨迹图划分问题,提出了一种基于贪心策略寻找近似最优的k条轨迹构建轨迹k-匿名集的方法。通过对比实验,在合成轨迹数据集上验证了所提算法的有效性和合理性。A （s,λ）-overlap was proposed which based on the angle and location between two trajectories to construct the relationship between them, and also measure the weight between two trajectories by trajectory angle and distance, to- gether to construct the personal trajectory graph model. Then a greedy partition method to was proposed construct trajec- tory k-anonymization sets by finding the approximate optimal k trajectories. Experiment results on synthetic dataset show the effectiveness and reasonableness of proposed method.

关 键 词：隐私安全 个性化 图划分 (s λ)-覆盖 轨迹间夹角 

分 类 号：TP309.2[自动化与计算机技术—计算机系统结构]