基于布隆过滤器的轻量级隐私信息匹配方案  

Bloom filter-based lightweight private matching scheme

作　　者：万盛[1] 何媛媛[2] 李凤华[2,3] 牛犇[2] 李晖[1] 王新宇[2] 

机构地区：[1]西安电子科技大学综合业务网理论及关键技术国家重点实验室,陕西西安710071 [2]中国科学院信息工程研究所信息安全国家重点实验室,北京100195 [3]北京电子科技学院信息安全系,北京100070

出　　处：《通信学报》2015年第12期151-162,共12页Journal on Communications

基　　金：国家自然科学基金-广东联合基金资助项目(U1401251);国家高技术研究发展计划("863"计划)基金资助项目(2012AA013102);教育部重点基金资助项目(209156)~~

摘　　要：针对智能终端用户私有数据匹配中的隐私保护问题,基于布隆过滤器和二元向量内积协议,提出一种新的综合考虑用户属性及其偏好的轻量级隐私信息匹配方案,包括建立基于Dice相似性系数的二维向量相似度函数、设置参数、生成布隆过滤器、计算二元向量内积、计算相似度和确定匹配对象6个部分。该方案采用基于布隆过滤器的相似度估计和基于混淆方法的二元向量内积协议,在不依赖于可信第三方的前提下,大幅度降低计算开销,且能够有效抵御蛮力攻击和无限制输入攻击。实验结果表明,该方案与典型代表方案相比,计算效率得到明显提升。With rapid developments of mobile devices and online social networks, users of proximity-based mobile social networks（PMSN） could easily discover and make new social interactions with others, but they enjoyed this kind of conveniences at the cost of user privacy and system overhead, etc. To address this problem, a third party free and lightweight scheme to privately match the similarity with potential friends in vicinity was proposed. Unlike most existing work, proposed scheme considered both the number of common attributes and the corresponding priorities on each of them individually. The Bloom filter-based common-attributes estimation and the lightweight confusion binary vector scalar product protocol reduce the system overhead significantly, and can resist against brute force attack and unlimited input attack. The correctness, security and performance of overhead of proposed scheme are then thoroughly analyzed and evaluated via detailed simulations.

关 键 词：隐私信息匹配 Dice相似性系数 布隆过滤器 二元向量内积协议 

分 类 号：TN929[电子电信—通信与信息系统]