车联网中抵御位置数据中毒攻击的隐私保护机制研究

The server, in Internet of vehicles, collects and analyzes a larger number of users' location data, and the analysis results are expected to improve applications. However, the server is vulnerable to the location poisoning attacks when malicious users send poisoning locations, resulting in the server's data privacy leakage. This project intends to conduct the first attempt towards the privacy-preserving scheme against the location poisoning attacks. This project will utilize the random walk and feature learning to explore the acquired characteristics of users' movement and construct the inferred social graph. Then, it protects the data privacy utilizing the structural relationships between the social network graph and the inferred social graph. Thereafter, this project will propose the location data aggregation algorithm with the help of both Diffie-Hellman and secret sharing, aiming to guarantee the data utility. Finally, this project will study the performance quantification model based on the differential privacy, and moreover investigate the performance against the location poisoning attacks via the algorithm simulations and the small-scale prototype implementations. The research findings of this project will facilitate the enrichment and improvement of the privacy-preserving theories and technologies in Internet of vehicles, and furthermore have very realistic theoretical significance and application values in promoting the sustained and healthy development as well as decreasing the risks of disclosing data privacy in Internet of vehicles.

在车联网中，中央处理器收集、分析大量车联网用户的位置数据，并将分析结果用于改善相关应用服务，在此过程中，恶意车联网用户发送有毒位置数据，将会导致中央处理器遭受位置数据中毒攻击，即中央处理器的数据安全受到威胁。本项目拟研究车联网中抵御位置数据中毒攻击的隐私保护机制。首先，拟通过随机步游和特征学习揭示用户运动的外源属性、构造推测社交图，根据社交网络图与推测社交图之间结构关联性，保证数据隐私；然后，拟结合Diffie-Hellman和密匙共享设计车联网用户位置数据收集算法，保证数据有用性；最后，从理论层面研究基于差分隐私的算法性能量化模型，并从仿真、实验层面搭建算法仿真和小规模实验平台验证算法性能。本项目研究成果能够推动车联网中隐私保护相关理论体系不断丰富和完善；对进一步降低车联网中隐私泄露风险、促进车联网持续健康稳定发展具有重要的理论意义和应用价值。

本项目研究车联网中抵御位置数据中毒攻击的隐私保护机制。首先通过随机步游揭示用户运动的外源属性、构造推测社交图；然后根据社交网络图与推测社交图之间结构关联性保证数据隐私；最后从仿真、实验层面搭建算法仿真和小规模实验平台验证算法性能。本项目取得了一系列创新性成果，推动演进中的车联网数据安全理论体系不断丰富和完善，促进车联网健康稳定发展；以第一或通信作者发表有标注的SCI论文15篇，其中SCI 1区论文4篇、SCI 2区论文5篇、CCF A类论文2篇、CCF B类论文1篇；以第一发明人授权国家发明专利5项；培养硕士研究生10名，其中毕业生4名；举办国际学术会议2次，参加国际学术会议8人次。