面向LBS的结合k-匿名和本地差分隐私的位置隐私保护

Study on Location Privacy Protection for LBS Combining K-anonymityand Local Differential Privacy

针对现有位置隐私保护方案大多没有充分考虑隐私保护中攻击者背景知识的问题,提出一种结合k-匿名和本地差分隐私的位置保护方案。该方案选择的匿名候选集是由与真实用户位置的查询概率所接近的兴趣点形成,然后使用Voronoi图对该匿名集进行划分,再采用RAPPOR算法对分割后的候选集进行扰动,最后通过计算候选位置与真实位置之间的欧式距离,筛选出一个较为分散的位置集。实验结果表明:所提出的模型在有效保护位置隐私的同时,还能更进一步保证LBS服务的质量。

Aiming at the problem that most existing location privacy protection schemes do not fully consider the attacker′s background knowledge in privacy protection,a location protection scheme combining k-anonymity and local differential privacy is proposed.The anonymous candidate set selected in this article is formed by interest points that are close to the query probability of the real user′s location.Then,Voronoi diagrams are used to partition the anonymous set,and RAPPOR algorithm is used to perturb the segmented candidate set.Finally,a more dispersed position set is selected by calculating the Euclidean distance between the candidate position and the real position.The experimental results indicate that the model proposed in this article can effectively protect location privacy while further ensuring the quality of LBS services.