【摘要】 现有的本地化差分隐私位置保护方法大多针对用户的静态位置进行保护，未考虑到用户位置动态变化的应用场景，且存在算法复杂度高、扰动结果可用性低等问题.针对上述问题，提出一种基于隐马尔科夫模型(hidden Markov model, HMM)的本地化差分隐私动态位置保护方法.首先，结合位置动态变化过程中的时空相关性，构建基于隐马尔科夫模型的时间序列和隐私保护安全区域，实现用户位置更新后本地化差分隐私扰动区域的优选.其次，设计基于隐马尔科夫模型的动态位置连续扰动算法和本地化差分隐私随机响应机制，对优选区域内的位置点进行扰动处理，实现用户位置的动态化本地差分隐私保护.最后，在实际位置轨迹数据集合上进行实验和分析，证明所提方法能够在实现动态位置数据本地化隐私保护的前提下，达到更好的聚合准确度和统计可用性.更多还原

【Abstract】 The existing localized differential privacy location protection methods mainly focus on protecting the static locations of users without considering the dynamic scenarios of location updating. Additionally, these methods often suffer from high algorithm complexity and low availability of perturbation results. In order to solve the above problems, a local differential privacy dynamic location protection method based on hidden Markov model(HMM) was proposed in this paper. Firstly, a time-series model and a privacy-preserving safe region based on the hidden Markov model and the privacy protection safety area were constructed by incorporating the spatial-temporal correlations of the dynamic changes of users’ locations, so as to optimize the local differential privacy perturbation regions after users’ location updates. Then, a hidden Markov model-based continuous perturbation algorithm and a random response mechanism of local differential privacy were designed to perturb the location points within the optimized region, realizing dynamically local differential privacy protection of users’ locations. Finally, experiments and analyses were carried out on the actual location trajectory dataset. The results demonstrate that the proposed method can achieve better aggregation accuracy and statistical availability on the premise of realizing dynamic local differential privacy protection of locations.更多还原