Abstract: K-anonymity for location privacy is widely applied in LBS privacy protection. However, most of the researches based on K-anonymity mechanism do not consider the attackers’ background knowledge. Thus, a dynamic anonymity algorithm for resisting the trajectory prediction attack is proposed. The similar historical trajectories of the user’s base trajectory are picked out using the sliding-window constraint. Then the trajectories are used to predict the next location of the user. Further, neighboring historical data is added to the vulnerable locations to resist the prediction attack. Compared with the similar algorithm, experimental results show that the algorithm has better prediction performance, and reduces the privacy disclosure risk under the same privacy requirements.

Key words: K-anonymity, privacy protection, prediction attack, sliding-window

摘要： 位置K-匿名技术被广泛应用于LBS隐私保护中，然而大多数基于K-匿名机制的研究缺少对攻击者背景知识的考虑，针对此，提出了一种抵御基于历史轨迹预测攻击的动态匿名算法。该方法以滑动窗口约束的方式挑选出与用户基轨迹相似的历史轨迹对用户位置进行预测，并对存在预测风险的位置动态添加历史数据以抵御预测攻击。与同类算法相比，实验结果表明该算法具有更好的预测性能，且在同等隐私需求下降低了用户的隐私披露风险。

关键词: K-匿名, 隐私保护, 预测攻击, 滑动窗口