Abstract:

Aiming at the problem of semantic inference attack caused by not considering semantic information in the anonymous set constructed by K-anonymous method in location-based service, a semantic diversity location privacy protection method in road network environment is proposed. According to the number of user accesses in different semantic locations, the method uses Euclidean distance to select other semantic location types with similar characteristics to construct an optimal semantic location type set. According to the proportion of semantic locations belonging to this type of set, the optimal segment is chosen to construct the anonymous set, which not only satisfies the semantic diversity, but also increases the uncertainty of users’ semantic location. The experimental results show that compared with the LSBASC algorithm, the method improves the average anonymous time by 27%, and the SDA algorithm performs better. The relative spatial granularity has decreased by 21%, and the privacy leakage has been reduced by 3%. That is, the SDA algorithm provides higher quality of service and privacy protection with a smaller anonymous space, and can effectively protect users’ semantic location privacy.

Key words: location privacy, privacy protection, semantic diversity, semantic location type similarity

摘要：

针对位置服务中基于K-匿名方法构造的匿名集因未考虑语义信息导致语义推断攻击问题，提出了一种路网环境下的语义多样性位置隐私保护方法。该方法根据不同语义位置用户访问数量，利用欧氏距离选择具有相似特性的语义位置类型，构建最优语义位置类型集合。根据路段上属于该类型集的语义位置所占比例，选择最优路段构建匿名集，使得匿名集不仅满足语义多样性，而且增加了用户语义位置的不确定性。实验结果表明，与LSBASC算法相比，该方法在平均匿名时间上提高了27%，SDA算法的执行效率更好。在相对空间粒度上减小了21%，隐私泄露程度上降低了3%，SDA算法以更小的匿名空间提供更高的服务质量和隐私保护程度，能有效地保护用户语义位置隐私。

关键词: 位置隐私, 隐私保护, 语义多样性, 语义位置类型相似度