| 基于交通业务特征理解的车路协同可信交互方法 |
| |
| 引用本文： | 上官伟,查园园,付瑶,郑四发,柴琳果.基于交通业务特征理解的车路协同可信交互方法[J].交通运输工程学报,2022,22(4):348-360. |
| |
| 作者姓名： | 上官伟 查园园 付瑶 郑四发 柴琳果 |
| |
| 作者单位： | 1.北京交通大学 电子信息工程学院，北京 1000442.北京交通大学 轨道交通控制与安全国家重点实验室，北京 1000443.中国移动研究院，北京 1000534.清华大学 车辆与运载学院，北京 100084 |
| |
| 基金项目： | 国家重点研发计划2018YFB1600600中国国家铁路集团有限公司科技研究开发计划N2021G045 |
| |
| 摘 要： | 为保障车路协同环境下信息的可信交互，分析了车车、车路协同信息交互流程和不同模式下的交互需求，设计了车路协同可信交互架构；构建了车辆行为状态推演模型与路径扰动因子量化模型，设计了车辆主体可信度计算方法与等级评估规则，实现了车辆主体行为可信认证；通过对交通业务的有效特征理解构建了消息紧急度量化模型，利用低分辨率筛选策略初步过滤了消息报文，基于支持向量机(SVM)对消息内容进行了深度理解，形成了多分辨率交互内容认知方法；使用包含OMNeT++和SUMO仿真模拟器的Veins搭建了仿真测试环境，针对不同网联自动驾驶车辆(CAV)渗透率下的开放道路和交叉口场景开展了仿真试验，对提出的车路协同可信交互方法进行了测试验证。研究结果表明:结合交通业务特征理解能够有效改善车路协同信息交互的可信度判别，提出的方法对信标位置消息的平均认知正确率可以达到90.91%，相比基于时效性检测的可信交互方法提高了8.68%；在安全效率消息可信交互验证试验中，随着恶意车辆比例的增加，传统基于投票机制的车路协同可信交互方法逐渐失效，而提出的方法在保证单次认证时延小于13 ms的条件下，平均正确率达到94.96%，较传统基于反向传播(BP)神经网络的方法提高了3.05%，且CAV渗透率越大，可信交互检测结果的准确率越高，漏报率越低，能够满足车路协同可信交互需求。
|
| 关 键 词： | 车路协同 可信交互 交通业务特征理解 行为状态推演 路径扰动因子量化 多分辨率认知 |
| 收稿时间： | 2021-12-12 |
Vehicle-infrastructure cooperative credible interaction method based on traffic business characteristics understanding |
| |
| SHANGGUAN Wei,ZHA Yuan-yuan,FU Yao,ZHENG Si-fa,CHAI Lin-guo.Vehicle-infrastructure cooperative credible interaction method based on traffic business characteristics understanding[J].Journal of Traffic and Transportation Engineering,2022,22(4):348-360. |
| |
| Authors: | SHANGGUAN Wei ZHA Yuan-yuan FU Yao ZHENG Si-fa CHAI Lin-guo |
| |
| Affiliation: | 1.School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing 100044, China2.State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University, Beijing 100044, China3.China Mobile Research Institute, Beijing 100053, China4.School of Vehicle and Mobility, Tsinghua University, Beijing 100084, China |
| |
| Abstract: | For the credible information interaction in vehicle-infrastructure cooperative environments, the processes of vehicle-vehicle and vehicle-infrastructure cooperative information interaction and the interaction requirements of different modes were analyzed, and a vehicle-infrastructure cooperative credible interaction framework was designed. A model of vehicle behavior state deduction and one of path perturbation factor quantification were constructed. A credibility calculation method for the vehicle object and level evaluation rules were designed. The credible authentication of vehicle object behavior was thereby achieved. A quantification model for the message urgency was built by understanding the effective traffic business characteristics. The low-resolution filtering strategy was used to preliminarily filter the message, and the message content was deeply understood on the basis of the support vector machine (SVM), thereby obtaining a multi-resolution interactive content cognition method. The Veins with OMNeT++ and SUMO simulators was used to build a simulation test environment. Simulation tests were carried out in open roads and intersection scenarios with different penetration rates of connected and automated vehicles (CAVs). The proposed vehicle-infrastructure cooperative credible interaction method was tested and verified. Research results show that the credibility identification for the vehicle-infrastructure cooperative information interaction can be effectively improved by understanding the traffic business characteristics. The average cognitive accuracy for the beacon location message achieved by the proposed method is 90.91%. It is 8.68% higher than that of the credible interaction method based on the timeliness detection. In the credible interaction verification experiment on the safety efficiency message, as the proportion of malicious vehicles increases, the traditional vehicle-infrastructure cooperative credible interaction method based on the voting mechanism is gradually held invalid. In contrast, an average accuracy of 94.96% is achieved by the proposed method under the condition that the single authentication delay is less than 13 ms. It is 3.05% higher than that of the traditional method based on the back propagation (BP) neural network. Moreover, a higher accuracy rate and a lower false negative rate of the credible interaction detection results can be obtained with a higher CAV penetration rate. Therefore, the needs of vehicle-infrastructure cooperative credible interaction can be met by the proposed method. |
| |
| Keywords: | |
|
| 点击此处可从《交通运输工程学报》浏览原始摘要信息 |
|
点击此处可从《交通运输工程学报》下载全文 |
|
