远程遥控驾驶场景下基于网络信息开放的QoS预测

doi:10.11959/j.issn.1000-0801.2023011

电信科学 ›› 2023, Vol. 39 ›› Issue (3): 153-161.doi: 10.11959/j.issn.1000-0801.2023011

远程遥控驾驶场景下基于网络信息开放的QoS预测

贾宇航¹^,², 雷艺学¹^,², 张翼鹏¹^,², 张云飞¹^,²

¹ 腾讯云计算（北京）有限责任公司，北京 100191
² 腾讯科技（深圳）有限公司，广东深圳 518000

修回日期:2023-01-08 出版日期:2023-03-20 发布日期:2023-03-01
作者简介:贾宇航（1997- ），男，腾讯科技（深圳）有限公司、腾讯云计算（北京）有限责任公司未来网络实验室研究员，主要研究方向为5G 网络、XR、云游戏、网络信息开放及其在智慧交通和云化多媒体领域的应用等
雷艺学（1978- ），男，博士，腾讯科技（深圳）有限公司、腾讯云计算（北京）有限责任公司未来网络实验室专家研究员，主要研究方向为 5G/6G 网络演进及实时数字孪生、云游戏与XR技术、网络信息开放与跨层优化、V2X智能网联技术及应用等
张翼鹏（1989- ），男，腾讯科技（深圳）有限公司、腾讯腾讯云计算（北京）有限责任公司未来网络实验室研究员，主要研究方向为智慧交通、车联网技术与标准研究
张云飞（1977- ），男，腾讯科技（深圳）有限公司、腾讯云计算（北京）有限责任公司未来网络实验室主任，腾讯智慧交通首席科学家，国家高层次人才特殊支持计划特聘专家，北京交通大学、上海交通大学兼职教授，5GAIA 副理事长，主要研究方向为 5G网络、云计算、分布式系统、流媒体、车联网和移动终端等

Network information exposure based QoS prediction in tele-operated driving

Yuhang JIA¹^,², Yixue LEI¹^,², Yipeng ZHANG¹^,², Yunfei ZHANG¹^,²

¹ Tencent Technology (Beijing) Co., Ltd., Beijing 100191, China
² Tencent Technology (Shenzhen) Co., Ltd., Shenzhen 518000, China

Revised:2023-01-08 Online:2023-03-20 Published:2023-03-01

摘要/Abstract

摘要：

5G系统的飞速发展支持许多车联网用例对服务质量（quality of service，QoS）的苛刻要求。但网络和应用的适配仍然存在很多问题。网络信息开放是一种潜在的解决方案，旨在实现网络向应用程序实时提供蜂窝无线网络信息，从而帮助服务提供方实现更好的策略控制并改善用户体验。提出一种基于网络信息开放的服务质量预测（predictive QoS，PQoS）方法，通过提前预测即将发生的网络变化来支持应用做出提前响应，提高用户的体验质量（quality of experience，QoE）。介绍了网络信息开放及PQoS的背景，并介绍了PQoS的国内外研究、标准及落地现状；提出一种远程遥控驾驶（tele-operated driving，ToD）场景下基于网络信息开放的QoS预测方法；对实际测试的数据进行分析，评估验证了PQoS的可行性。结果表明，基于网络信息开放的QoS预测技术能够良好地支持包括5G ToD在内的车联网应用，为5G系统在智慧交通行业的落地提供了参考。

关键词: 服务质量预测, 体验质量, 远程遥控驾驶, 网络信息开放

Abstract:

The rapid development of 5G systems supports the stringent quality of service (QoS) requirements for many Internet of vehicles (IoV) use cases.However, there are still many problems in the adaptation of networks and applications.Network information exposure is a potential solution that also enables the network to provide real-time cellular wireless network information to applications, thereby helping service providers achieve better policy control and achieve an experience for users.A predictive QoS (PQoS) method based on network information exposure was proposed.Applications could respond in advance and improve users’ quality of experience (QoE) by predicting upcoming network changes.Firstly, the background of network information exposure and PQoS was introduced, and the research, standards, and implementation status of PQoS both at home and abroad were introduced.Then, a QoS prediction method based on network information exposure in tele-operated driving (ToD) was proposed, and experiments were carried out through actual test data and the feasibility of PQoS was verified through the evaluation and analysis.The results show that the QoS prediction based on the network information exposure can well support the application of some IoV, including 5G remote driving, which provides a reference for implementing the 5G system in the smart transportation industry.

Key words: PQoS, QoE, ToD, network information exposure

贾宇航, 雷艺学, 张翼鹏, 张云飞. 远程遥控驾驶场景下基于网络信息开放的QoS预测[J]. 电信科学, 2023, 39(3): 153-161.

Yuhang JIA, Yixue LEI, Yipeng ZHANG, Yunfei ZHANG. Network information exposure based QoS prediction in tele-operated driving[J]. Telecommunications Science, 2023, 39(3): 153-161.

图/表 8

图1

表1

图2

表2

图3

图4

表3

图5

参考文献 19

[1]	3GPP. New WID:5G system enhancement for advanced interactive services[EB]. 2021.
[2]	3GPP. System architecture for the 5G system (5GS)[EB]. 2021.
[3]	XIE Y X , FAN Y , KYLE J . PBE-CC:congestion control via endpoint-centric,physical-layer bandwidth measurements[C]// Proceedings of the Annual Conference of the ACM Special Interest Group on Data Communication on the Applications,Technologies,Architectures,and Protocols for Computer Communication. New York:ACM Press, 2020.
[4]	JIA Y H , ZHANG Y F , YANG R ,et al. MoWIE for network aware application[S]. 2020.
[5]	ZHANG Y F , LI G , XIONG C S ,et al. MoWIE:toward systematic,adaptive network information exposure as an enabling technique for cloud-based applications over 5G and beyond[C]// Proceedings of the Workshop on Network Application Integration/CoDesign. New York:ACM Press, 2020: 20-27.
[6]	5GAA. Making 5G proactive and predictive for the automotive industry[S]. 2019.
[7]	BARMPOUNAKIS S , MAGOULA L , KOURSIOUMPAS N ,et al. LSTM-based QoS prediction for 5G-enabled connected and automated mobility applications[C]// Proceedings of 2021 IEEE 4th 5G World Forum (5GWF). Piscataway:IEEE Press, 2021: 436-440.
[8]	KOUSARIDAS A , MANJUNATH R P , PERDOMO J ,et al. QoS prediction for 5G connected and automated driving[J]. IEEE Communications Magazine, 2021,59(9): 58-64.
[9]	GUTIERREZ-ESTEVEZ M A , UTKOVSKI Z , KOUSARIDAS A ,et al. A statistical learning framework for QoS prediction in V2X[C]// Proceedings of 2021 IEEE 4th 5G World Forum (5GWF). Piscataway:IEEE Press, 2021: 441-446.
[10]	PERDOMO J , KOUSARIDAS A , ZHOU C ,et al. Deep learning-based QoS prediction with innate knowledge of the radio access network[C]// Proceedings of 2021 IEEE Global Communications Conference (GLOBECOM). Piscataway:IEEE Press, 2021: 1-6.
[11]	MASON F , DRAGO M , ZUGNO T ,et al. A reinforcement learning framework for PQoS in a teleoperated driving scenario[C]// Proceedings of 2022 IEEE Wireless Communications and Networking Conference (WCNC). Piscataway:IEEE Press, 2022: 114-119.
[12]	BOBAN M , GIORDANI M , ZORZI M . Predictive quality of service (PQoS):the next frontier for fully autonomous systems[J]. arXiv preprint, 2021,arXiv:2109.09376.
[13]	SCHIPPERS H , SCHULER C , SLIWA B ,et al. System modeling and performance evaluation of predictive QoS for future tele-operated driving[J]. arXiv preprint, 2022,arXiv:2201.11432.
[14]	PALAIOS A , VIELHAUS C , KüLZER D F ,et al. Effect of spatial,temporal and network features on uplink and downlink throughput prediction[C]// Proceedings of 2021 IEEE 4th 5G World Forum (5GWF). Piscataway:IEEE Press, 2021: 418-423.
[15]	PALAIOS A , GEUER P , FINK J ,et al. Network under control:multi-vehicle E2E measurements for AI-based QoS prediction[C]// Proceedings of 2021 IEEE 32nd Annual International Symposium on Personal,Indoor and Mobile Radio Communications (PIMRC). Piscataway:IEEE Press, 2021: 1432-1438.
[16]	KüLZER D F , DEBBICHI F , STA?CZAK S ,et al. On latency prediction with deep learning and passive probing at high mobility[C]// Proceedings of 2021 IEEE International Conference on Communications. Piscataway:IEEE Press, 2021: 1-7.
[17]	LOSSI L , ALEMANY P , HIDALGO J F ,et al. 5GCroCo barcelona trial site results:orchestration KPI measurements and evaluation[C]// Proceedings of 2022 Joint European Conference on Networks and Communications ＆ 6G Summit (EuCNC/6G Summit). Piscataway:IEEE Press, 2022: 387-392.
[18]	HETZER D , MUEHLEISEN M , KOUSARIDAS A ,et al. 5G connected and automated driving:use cases and technologies in cross-border environments[C]// Proceedings of 2019 EuropeanConference on Networks and Communications (EuCNC). Piscataway:IEEE Press, 2019: 78-82.
[19]	KüLZER D F , KASPARICK M , PALAIOS A ,et al. AI4Mobile:use cases and challenges of AI-based QoS prediction for high-mobility scenarios[C]// Proceedings of IEEE Vehicular Technology Conference (VTC Spring). Piscataway:IEEE Press, 2021.

特征符号	含义	单位
time	时间	s
Throughput	上行吞吐量	bit/s
Uplink delay	上行时延	ms
RSRP	参考信号接收功率	dB
RSRQ	参考信号接收质量	—
SINR	信号与干扰加噪声比	dB
RSSI	接收信号强度指示	dB

参数	参数含义	预设值
n_in	滞后时间	15 s
n_out	超前预测时间	9s
n_vars	特征数量	6
train_proportion	拆分训练集的比例	0.8
n_neoron	神经元数量	100
n_batch	小批量数量	80
n_epoch	迭代次数	100
Repeats	平均次数	5
loss_function	DNN损失函数	‘mae’
optimizer_function	DNN优化函数	‘adam’
Index_col_name	时间序列索引值	‘Timestamp’
output_col_name	待预测特征	‘SINR’
cols_to_drop	需要删除的特征	[‘’]
D	遗忘值	0.01
draw_loss_plot	是否画出训练过程	True

时间	平均相对误差
预测第一秒	0.017 453 517 211 292 71
预测第二秒	0.025 728 424 501 522 14
预测第三秒	0.040 747 431 467 749 104
预测第四秒	0.047 477 053 756 471 12
预测第五秒	0.051 100 039 211 740 35
预测第六秒	0.054 721 370 995 354 825
预测第七秒	0.058 356 737 121 058 45
预测第八秒	0.060 536 148 206 996 68
预测第九秒	0.062 588 531 297 455 44

远程遥控驾驶场景下基于网络信息开放的QoS预测

Network information exposure based QoS prediction in tele-operated driving

在线阅读

PDF下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 8

参考文献 19

相关文章 6

Metrics

推荐阅读 0

[1]	张世行, 罗雅文, 张敏, 孙奇福, 阳小龙. 面向移动内容分发网络环境的移动视频预取技术研究进展[J]. 电信科学, 2021, 37(1): 8-21.
[2]	贺甜甜,刘彦凯,宋利. 视频用户体验理论与实践[J]. 电信科学, 2017, 33(8): 67-76.
[3]	季石宇,唐良瑞,李淑贤,杜施默. 基于用户体验质量和系统能耗的异构网络联合接入选择和功率分配策略[J]. 电信科学, 2017, 33(11): 47-55.
[4]	林浩凌,吴奕生,郑伟旭,刘昕. 基于移动互联网应用的感知评估体系研究[J]. 电信科学, 2014, 30(Z1): 1-5.
[5]	翁颐,姚良,奚溪. 基于单播流的IPTV频道切换提速方案研究[J]. 电信科学, 2014, 30(9): 159-163.
[6]	杨燕. 浅析移动通信网络中的QoE[J]. 电信科学, 2007, 23(8): 34-38.