Abstract:

The existing privacy preservation methods in location-based services only focus on the protection of user location and identification information. It produces the truth of sensitive homogeneity attack when the queries in a cloaking set are sensitive information. To solve this problem, a personalized (k,p)-sensitive anonymization model was presented. On the basis of this model, a pruning tree-based cloaking algorithm called PTreeCA was proposed. The tree-type index in the spatial database has two features. The one is that mobile users are roughly partitioned into different groups according to the locations of mobile users; the other one is that the aggregated information can be stored in the intermediate nodes. By utilizing the two features, PTreeCA could find the cloaking set from the leaf node where the query user is and its sibling nodes, which are benefit for improving efficiency of the anonymization algorithm. The efficiency and effectiveness of PTreeCA are validated by a series of designed experiments on the simulated and real data sets. The average success rate is 100%, and the average cloaking time is only about 4ms. The experimental results show that PTreeCA is effective in terms of success rate, cloaking time, and the anonymization cost when the privacy requirements levels are low or medium.

摘要：

基于位置服务中的隐私保护方法存在只关注保护用户位置和标识信息的问题，当匿名集中提出的查询均属于敏感查询时,将产生敏感同质性攻击。针对此问题，提出了个性化(k,p)-敏感匿名模型。并基于此模型，提出了基于树型索引结构的匿名算法——PTreeCA。空间数据库中的树型索引具有两大特点：1)空间中的用户已根据位置邻近性在树中被大致分组；2)在树的中间节点中可以存储聚集信息。利用这两个特点，PTreeCA可以从查询用户所在叶子节点和其兄弟节点中寻找匿名集，提高了匿名算法的效率。最后，在模拟和真实数据集上进行了实验，所提算法平均匿名成功率可达100%，平均匿名时间只有4ms。当隐私级别较低和适中时，PTreeCA在匿名成功率、匿名时间和匿名代价方面均表现出良好性能。