一种基于扰动的轨迹数据隐藏发布方法

A perturbation-based privacy preserving trajectory publication method

针对轨迹数据发布中的隐私保护和数据可用性问题,结合统计学的概念,提出一种基于扰动的轨迹数据隐藏发布方法.首先定义一种隐私泄露检测机制,当该检测机制发现攻击者依赖所掌握部分轨迹能以较大概率推测出某隐私节点时,基于统计方法,寻找出现频率最低的同类隐私节点,若存在且用其替换有隐私泄露风险的隐私节点后不会出现新的隐私泄露,则执行替换操作;否则在拥有该隐私节点的所有轨迹中,选择最佳的那条轨迹,将该隐私节点移除.这样,就能降低隐私节点的隐私泄露概率,保证发布后的轨迹数据满足用户的隐私需求.理论分析和实验结果表明,所提出的方法能有效避免基于部分轨迹推测剩余隐私节点的攻击,有效保持原有轨迹数据中不同种类节点间连接关系的可用性.

Considering the problems of privacy preserving and data utility in trajectory publication, a perturbation-based privacy preserving trajectory publishing method is proposed from the view of sta- tistics. First, a mechanism of privacy leak detection is devised to identify whether an attacker can use partial trajectories as quasi-identifier to infer the rest privacy-aware trajectory nodes with high probability. When it detects privacy leak, a statistic-based approach is used to find homogeneous pri- vacy-aware nodes of the lowest occurrence frequency, and replace private leak nodes with them, on condition that there is no new privacy disclosure after replacement. Otherwise, the method suppres- ses the private leak node of appropriate trajectory. In this way, the breach probability could be de- creased effectively, and it can be assured that the published trajectory datasets meet the users＇ privacy demands. Theoretical analysis and experimental results testify that the proposed method can prevent the attack of inferring sensitive locations by partial trajectory, meanwhile keep the utility of linkage relation amon~ different kinds of nodes in the original trajectory datasets well.