基于服务相似性的隐私保护k近邻查询方法  

Privacy-preserving k-Nearest Neighbor Query Method Based on Service Similarity

作　　者：张学军[1] 李佳乐 杨依行 黄海燕 许陈 李桢 杜晓刚 ZHANG Xue-jun;LI Jia-le;YANG Yi-xing;HUANG Hai-yan;XU Chen;LI Zhen;DU Xiao-gang(School of Electronic and Information Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China;School of Electronic Information and Artificial Intelligence,Shaanxi University of Science and Technology,Xi′an710021,China)

机构地区：[1]兰州交通大学电子与信息工程学院,兰州730070 [2]陕西科技大学电子信息与人工智能学院,西安710021

出　　处：《兰州交通大学学报》2023年第1期44-53,61,共11页Journal of Lanzhou Jiaotong University

基　　金：国家自然科学基金(61762058,61901201,61861024);兰州交通大学“百名青年优秀人才培养计划”;甘肃省自然科学基金(21JR7RA282,20JR5RA387);甘肃省教育厅产业支撑计划项目(2022CYZC-38)。

摘　　要：如何同时保护用户的位置和查询隐私而又不妨碍服务质量和增加系统处理开销是当前位置服务隐私保护系统面临的一个重要挑战.为了解决这一挑战,基于位置查询的服务相似特性和Paillier密码系统的同态特性,提出了一种能同时保护位置隐私和查询隐私的k近邻查询方法,实现不依赖可信第三方便可实现用户位置和查询内容的隐私保护以及兴趣点的精确查询;同时,通过构造服务相似地图生成扰动位置,解决已有方法查询处理开销大的问题,并保证查询结果的准确性.利用真实数据集从查询准确性和系统开销等方面对所提方法进行了实验验证.实验结果与安全性分析表明:和已有方法相比,所提出的方法在兼顾位置和查询隐私保护的同时,有效地保证了服务质量,降低了系统处理开销.How to protect the user′s location and query privacy at the same time without compromising the quality of services and increasing system overhead is a special challenge for privacy-preserving location-based services.To address this challenge, a privacy-preserving k-nearest neighbor query method is proposed based on the service similarity of location query and the homomorphism of Paillier cryptosystem in this paper, which can preserve both location and query privacy for users.The method not only realizes the privacy-preserving of user′s location and query content but also obtains the accurate retrieval of point of interest without relying on a trusted third part.Meanwhile, by constructing service similarity map to generate perturbed location, the issue of high resource cost is solved and the accuracy of query results is guaranteed.The comprehensive experiments are carried out based on real dataset from the aspect of accuracy and system overhead. Experimental results and security analysis show that, compared with the state-of-the-art approaches, the proposed method preserves both location privacy and query privacy when the service quality is effectively guaranteed and the system overhead is reduced.

关 键 词：服务相似性 位置隐私 查询隐私 同态加密 K近邻查询 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]