基于轨迹多特性的隐私保护算法

现有基于聚类的轨迹隐私保护算法在衡量轨迹间的相似性时大多以空间特征为标准,忽略了轨迹蕴含的其他方面的特性对轨迹相似性的影响。针对这一情况可能导致的匿名后数据可用性较低的问题,提出了一种基于轨迹多特性的隐私保护算法。该算法考虑了轨迹数据的不确定性,综合方向、速度、时间和空间4个特性的差异作为轨迹相似性度量的依据,以提高轨迹聚类过程中同一聚类集合中轨迹之间的相似度；在此基础上,通过空间平移的方式实现同一聚类集合中轨迹的k-匿名。实验结果表明,与经典隐私保护算法相比,在满足一定隐私保护需求的前提下,采用所提算法实施隐私保护之后的轨迹数据整体具有较高的数据可用性。

Privacy Protection Algorithm Based on Multi-characteristics of Trajectory

Most of existing trajectory privacy protection algorithms based on trajectory clustering use spatial features as the standard when measuring the similarity between trajectories,ignoring the influence of other temporal and spatial characteristics of trajectories on trajectory similarity.In view of the fact that this situation may lead to the problem of low availability of anonymous data,a protection algorithm based on integrated spatiotemporal characteristics of trajectory was proposed.The proposed algorithm combines the uncertainty of trajectory data,and uses the difference of 4 aspects of direction,speed,time and space to measure similarity between trajectories,in order to improve the similarity between the trajectories in the same cluster set.And then the trajectories of the same clustering set are spatially shifted to achieve the k-anonymization of the trajectories in the same clustering set.The experimental results show that compared with the classical privacy protection algorithm,the trajectory data protected by proposed algorithm as a whole has higherdata availability under certain privacy protection requirements.