一种基于K-匿名聚类的可穿戴设备数据重发布方法

A wearable data re-dissemination method based on K-anonymity clustering

近年来,可穿戴设备被广泛地被应用于日常生活。用户量增加造成的可穿戴设备数据重发布是导致隐私泄漏的一个重要原因。为此,数据匿名化重发布方法受到了广泛关注。然而,现有的数据匿名化重发布方法存在两个方面的不足:一方面,现有的数据匿名化重发布算法可能会造成严重的信息损失或用户隐私数据的泄漏;另一方面,现有的数据匿名化重发布算法在兼顾保护用户隐私和减少信息损失的情况下会造成较高的发布成本。为了兼顾隐私安全和数据可用性,并且提高数据重发布算法的效率,结合可穿戴设备自身的特点,提出基于聚类的数据匿名化重发布算法,该算法直接对增量数据进行基于聚类匿名化操作,使数据匿名化重发布更为高效。此外,在数据量较大的应用场景中,基于聚类的数据匿名化重发布算法可以有效减少信息损失。实验结果表明,基于聚类的数据匿名化重发布算法能够在保证用户隐私安全的前提下减少信息的损失并且提高执行效率。

Nowadays, wearable devices are widespread in our daily life. The rapid growth of users＇ number results in data redissemination by data holders. Nevertheless, improper redissemination methods can lead to unacceptable information loss or privacy leakage. In addition to privacy preserving concern, the data republishing methods for wearable devices should also be efficient enough due to the rapid increase of the number of wearable devices. In order to enhance the efficiency under the premise of information security of users and acceptable information loss, we propose a wearable data redissemination method based on clustering Kanonymity. The proposed method jointly considers data privacy, information loss and the overheads. Specifically, the proposed method processes the incremental data directly to enhance the efficiency and the clustering Kanonymity can limit information loss. The proposed method can reduce information loss when the amount of the data is huge. Experimental results demonstrate its effectiveness.