Li Hantao, Liu Feng, Zhao Zhongliang, Karimzadeh Mostafa
School of Electronic and Information Engineering, Beihang University, Beijing 100190, China.
Sensonic, 4780 Scharding, Austria.
Sensors (Basel). 2022 Mar 31;22(7):2686. doi: 10.3390/s22072686.
Exploring data connection information from vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications using advanced machine learning approaches, an intelligent transportation system (ITS) can provide better safety services to mitigate the risk of road accidents and improve traffic efficiency. In this work, we propose an end-edge-cloud architecture to deploy machine learning-driven approaches at network edges to predict vehicles' future trajectories, which is further utilized to provide an effective safety message dissemination scheme. With our approach, the traffic safety message will only be disseminated to relevant vehicles that are predicted to pass by accident areas, which can significantly reduce the network data transmission overhead and avoid unnecessary interference. Depending on the vehicle connectivity, our system adaptively chooses vehicle-to-vehicle (V2V) or vehicle-to-infrastructure (V2I) communications to disseminate safety messages. We evaluate the system by using a real-world VANET mobility dataset, and experimental results show that our system outperforms other mechanisms without considering any predicted vehicle trajectory density information.
利用先进的机器学习方法探索车对车（V2V）和车对基础设施（V2I）通信中的数据连接信息，智能交通系统（ITS）可以提供更好的安全服务，以降低道路事故风险并提高交通效率。在这项工作中，我们提出了一种端到边缘到云的架构，在网络边缘部署机器学习驱动的方法来预测车辆的未来轨迹，并进一步利用该轨迹提供有效的安全消息传播方案。通过我们的方法，交通安全消息只会传播给预计会经过事故区域的相关车辆，这可以显著减少网络数据传输开销并避免不必要的干扰。根据车辆连接情况，我们的系统会自适应地选择车对车（V2V）或车对基础设施（V2I）通信来传播安全消息。我们使用真实世界的车载自组织网络（VANET）移动数据集对系统进行评估，实验结果表明，在不考虑任何预测车辆轨迹密度信息的情况下，我们的系统优于其他机制。
