保持数据可用性的细粒度轨迹隐私保护方案

Fine-grained privacy-preserving framework while ensuring data usability in trajectory databases

轨迹数据的隐私保护近年来越来越受到重视,现有的工作很少考虑不同的隐私敏感位置之间的区别,也较少考虑不同的轨迹应用之间的区别(例如保险推销和紧急救助).鉴于轨迹数据用途的多样性以及用户个性化的隐私需求,本文提出了一种细粒度的基于标签的轨迹数据隐私保护方案,此方案能让用户够灵活自主地控制不同隐私敏感的轨迹片段对不同轨迹应用的访问授权.此外,考虑到大部分的隐私敏感位置都与轨迹停留相关,为了合理地隐藏轨迹中不可见的采样点,本文提出了一种将不可见的隐私敏感轨迹片段中的位置采样点,合理散布到周围频繁访问的多个位置中的方法.实验结果表明,本文提出的方法能够在有效保护轨迹隐私的同时只引入较小的额外计算负担.

The privacy of trajectories has aroused a wide concern. In previous works, rarely have the differences between different sensitive locations been discussed, nor the differences between different applications （eg： for advertising and for emergencies）. While in fact, some sensi- tive locations are more important and some applications ought to be granted the access. In this paper, to meet different privacy requirements and data utility requirements, we propose a finegrained privacy-preserving framework which allows the users to specify which locations arc visible to some applications and invisible to others at the same time. In addition, since most sensitive locations are relevant to stay points and a significant stay in a sensitive place may last longer than the ordinary places, we also propose an efficient approach to distribute invisible location samples along the nearby popular visit sequences. Experiment results indicate that our framework performs efficiently without introducing significant performance penalties.