Design and Evaluation of Remote Driving Architecture on 4G and 5G Mobile Networks
Design and Evaluation of Remote Driving Architecture on 4G and 5G Mobile Networks
复制标题
4G和5G移动网络上的远程驾驶架构设计与评估
DOI：
10.3389/ffutr.2021.801567
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
I. Tafur Monroy
中科院分区：
文献类型：
--
作者：
José den Ouden;Victor Ho;Tijs van der Smagt;G. Kakes;S. Rommel;I. Passchier;Jakub Juza;I. Tafur Monroy
尽管在过去的十年里，自动驾驶汽车的发展取得了一定的进展，但要以经济的价格达到大规模部署所需的可靠性水平，并结合安全要求，还有很长的路要走。在某些用例中，例如自动班车和出租车，甚至不再需要方向盘和踏板，远程驾驶可以弥补这一差距；在自动化系统难以确定下一步行动的情况下，远程操作员可以控制车辆。在物流业，它甚至可以通过提供更灵活的工作条件和更少的卡车停留时间来解决卡车司机短缺等已经很紧迫的问题。远程驾驶的一个重要方面是远程站与车辆之间的连接。随着5G移动技术在全球许多国家的推出，远程驾驶的实施离大规模部署更近了一步。在这项技术的部署中，5G可能会改变游戏规则。在这项工作中，我们研究了远程驾驶在最近部署的sub-6 ghz商用5G独立（SA）移动网络上的应用和远程驾驶的网络级性能。评估了移动边缘计算（MEC）集成、本地突破、延迟等5G架构对远程驾驶应用性能的影响。我们描述了一种远程驾驶解决方案的设计、开发（基于硬件在环仿真）和性能评估，并使用两种不同的车辆和两个不同的远程站点在5G和4G移动SA网络上进行了测试。定义了两个测试用例来评估应用程序和网络性能，并基于位置精度、相对反应时间和距离感知进行评估。结果表明，该网络的性能足以满足相对较低速度（<40 km/h）的远程驾驶应用。与4G相比，网络延迟已降至一半。延迟与远程驾驶性能之间的强相关性尚不清楚，需要进一步评估，并考虑到用户界面的影响。
Despite the progress in the development of automated vehicles in the last decade, reaching the level of reliability required at large-scale deployment at an economical price and combined with safety requirements is still a long road ahead. In certain use cases, such as automated shuttles and taxis, where there is no longer even a steering wheel and pedals required, remote driving could be implemented to bridge this gap; a remote operator can take control of the vehicle in situations where it is too difficult for an automated system to determine the next actions. In logistics, it could even be implemented to solve already more pressing issues such as shortage of truck drivers, by providing more flexible working conditions and less standstill time of the truck. An important aspect of remote driving is the connection between the remote station and the vehicle. With the current roll-out of 5G mobile technology in many countries throughout the world, the implementation of remote driving comes closer to large-scale deployment. 5G could be a potential game-changer in the deployment of this technology. In this work, we examine the remote driving application and network-level performance of remote driving on a recently deployed sub-6-GHz commercial 5G stand-alone (SA) mobile network. It evaluates the influence of the 5G architecture, such as mobile edge computing (MEC) integration, local breakout, and latency on the application performance of remote driving. We describe the design, development (based on Hardware-in-the-Loop simulations), and performance evaluation of a remote driving solution, tested on both 5G and 4G mobile SA networks using two different vehicles and two different remote stations. Two test cases have been defined to evaluate the application and network performance and are evaluated based on position accuracy, relative reaction times, and distance perception. Results show the performance of the network to be sufficient for remote driving applications at relatively low speeds (<40 km/h). Network latencies compared with 4G have dropped to half. A strong correlation between latency and remote driving performance is not clearly seen and requires further evaluation taking into account the influence of the user interface.