异构车联网中容忍入侵的隐私评估方法研究

The generalized heterogeneous vehicle network is mainly composed by Vehicle-to-Vehicle(V2V), Vehicle-to-Infrastructure (V2I) and Vehicle-to-Cloud, which is a complex network included vehicle node, internet, cloud, and the data saved in the cloud. Thus the privacy data become more important and serious if they are leaked. But there are three main difficult problems in privacy assessment: the related factors of real-time interaction of vehicles, the uncertainty of privacy behavior judgment caused by insufficient vehicle information, and the deliberate malicious vehicles registered as normal inside nodes. To address these problems, firstly, a joint-entropy based method will be proposed to evaluate the correlation of privacy. Secondly, we present the Bayesian network to tolerate the observation variables in time series to reduce the uncertainty caused by information asymmetry. Finally, the consistency of historical data from the cloud platform and real-time communication behaviors are used to predict the deliberate internal attack nodes. Through these innovative ideas, this project is expected to significantly reduce the relevance, uncertainty, and impact from inside attacks to privacy assessment. The results of the project can be used to exclude the low-level privacy protection nodes from participating in the information interaction, thereby improving the privacy protection level of the entire networks.

广义的异构车联网包含车-车、车-路和车-云网络，复杂庞大的网络结构和数据长期保存的云存储，使得原本可视范围内的车辆信息放大到互联网和云平台的存储周期，因此使车辆信息及其包含的驾乘人员隐私泄露问题日益严重。车辆隐私保护能力的评估有三个方面的难点：车辆实时交互的关联因素导致隐私泄露、车辆历史信息不充分带来的隐私行为判定的不确定性及恶意车辆注册到车联网内部蓄意获取并泄露其它车辆隐私信息。本项目拟用联合熵来评估隐私的关联影响，用贝叶斯网络在时序上对观察变量不充分性的容忍来降低因信息不对称带来的评估的不确定性，并以基于云平台的历史行为和实时通信的预估和一致性判定区分蓄意的内部攻击节点。项目研究可望通过评估方法的创新大幅度降低车联网隐私评估中的关联性、不确定性和内部攻击影响。研究结果可用于排除低等级隐私保护节点参与车联网信息交互，进而提高整个网络的隐私保护水平。

异构车联网中存在大量的车与车、车与路边设施的数据通信，而汽车的高速和动态性，使得车联网中通信信道的存活时间极短，而传统的基于加密的通信信道需要消耗大量时间构建，这将加剧车联网中通信延时。针对车联网中身份认证中存在的单点失效、隐私易暴露和跨域漫游困难等问题，研究基于区块链的车联网分布式数字身份系统。针对智能网联下异构多域网络环境中的车数量众多、接入频繁、移动高速等特点，提出异构车联网中跨域安全身份接入体系。研究基于零知识证明、隐私计算等密码技术的可验证凭证，实现最少信息披露与可验证性的平衡。研究基于消息的大规模安全低延时通信，实现大规模的车联网数字身份认证系统。提出面向车联网跨域漫游的零知识证明身份认证机制。在认证效率上，扩展现有认证框架建立单次传输完成身份和消息的双重认证的模型；在隐私保护上，基于零知识证明构建车辆盲化证书保证匿名性和不可链接性；在系统安全和可靠性方面，采用主侧链协同机制保证了车辆认证系统的正确性和可验证性；在性能方面，采用UTXO多输入多输出机制，优化漫游认证与存储开销，进一步减少用户计算量。理论分析和实验都验证了项目在安全和效率方面得到了提升。