一种基于隐私拆分的轨迹隐私保护方法

A Trajectory Privacy Preserving Algorithm Based on Privacy Splitting

针对连续位置服务中的位置隐私问题,提出一种基于隐私拆分的轨迹隐私保护方法.首先,系统分析轨迹中基于时空关联的位置攻击方法,提出历史、空间和行进3种位置相关性,并利用差分隐私模型,建立单点位置的发布对查询轨迹的前向和后向隐私风险评估机制.在此基础上,提出一种基于多服务器的交替查询机制,通过拆分查询轨迹,消除轨迹中位置间的相关性,提高轨迹的隐私安全.最后,基于真实数据集的仿真实验验证模型的有效性和可行性.

Aiming at the location privacy problem in continuous location-based services,a trajectory privacy protection method based on privacy splitting is proposed.Firstly,the location attack method based on spatiotemporal correlation in trajectory is analyzed systematically,and three kinds of location correlations,namely history,space and marching,are proposed.Then,a forward and backward privacy risk assessment mechanism for query trajectory is established by using differential privacy model.On this basis,an alternative query mechanism based on multi server is proposed.By splitting the query track,the correlation between the positions in the track is eliminated,and the privacy and security of the track are improved.Finally,we verify validity and feasibility of the model based on real data sets.