面向低延时实时视频的多维跨层带宽预测

doi:10.12141/j.issn.1000-565X.220705

摘要/Abstract

摘要：

5G网络的可用带宽是影响实时视频业务的关键要素之一，但如何在低延时实时视频业务下完成准确预测依旧是个难题。传统可用带宽预测算法通常依靠业务层的数据指标，根据发包策略完成预测。但此类算法在网络频繁变化的复杂场景下会出现预测滞后问题，严重影响了用户的接收视频质量。为解决此问题，文中提出了一种基于跨层多维参数的可用带宽预测算法，该算法综合考虑了业务层、物理层、网络层等的相关数据指标，并通过多个维度参数提升无线网络带宽探测的准确性。文中采用深度强化学习作为模型框架，针对不同的运动场景，通过跨层多维度的数据模型学习实现离线预测和在线预测的融合。同时，将网络丢包率、图像质量评估和端到端时延等链路影响因素作为约束条件，以实现预测模型在传输过程中的实时调整和优化。在半物理平台上的实验结果表明：文中所提算法的预测性能优于传统的预测算法，预测曲线与实际曲线的拟合程度高达95.8%以上；相比于单层预测算法，所提算法在步行、驾驶场景下的丢包率分别降低了47.3%和30.9%，接收视频质量提升了12%。

关键词: 跨层多维参数, 实时视频流, 5G网络, 可用带宽预测, 深度强化学习

Abstract:

The available bandwidth of 5G networks is a pivotal factor influencing real-time video services. However, how to achieve accurate predictions in the context of low-latency real-time video services remains a difficult problem. Conventional algorithms for predicting available bandwidth typically rely on data metrics at the application layer and complete the forecast according to the packet transmission strategies. Such algorithms can lead to prediction lag in complex scenarios, thereby significantly impairing the received video quality for users. To address this concern, this study proposed a novel available bandwidth prediction algorithm based on cross-layer multi-dimensional parameters. This algorithm comprehensively integrates pertinent data metrics from the application, physical, and network layers, utilizes multiple dimensions of parameters to enhance the precision of wireless network bandwidth detection. In this paper, deep reinforcement learning was adopted as the model framework to integrate offline prediction and online prediction through cross-layer and multi-dimensional data model learning for different motion scenarios. Furthermore, network packet loss rates, image quality assessments, end-to-end delays, and other link-related factors were introduced as constraints, to realize the real-time adjustment and optimization of the prediction model during the transmission process. Experimental results conducted on a semi-physical platform show that: the prediction performance of the proposed algorithm is better than the traditional prediction method, and the fitting degree of the prediction curve and the actual curve is more than 95.8%; compared with the single-layer prediction algorithm, the packet loss rate of the proposed algorithm in walking and driving scenarios decreases by 47.3% and 30.9%, respectively, and the quality of received video is improved by 12%.

Key words: cross-layer multi-dimensional parameter, real-time video stream, 5G network, available bandwidth prediction, deep reinforcement learning

中图分类号:

TN929.5

陈锋, 毛豪滨, 蔡吉玲, 等. 面向低延时实时视频的多维跨层带宽预测[J]. 华南理工大学学报(自然科学版), 2023, 51(11): 18-27.

CHEN Feng, MAO Haobin, CAI Jiling, et al.. Multidimensional Cross-Layer Bandwidth Prediction for Low-Latency Real-Time Video[J]. Journal of South China University of Technology(Natural Science Edition), 2023, 51(11): 18-27.

图/表 8

图1

表 1

图2

图3

图4

图5

图6

表 1

参考文献 32

1	ARSAN T ．Review of bandwidth estimation tools and application to bandwidth adaptive video streaming［C］∥Proceedings of High Capacity Optical Networks and Emerging/Enabling Technologies．Istanbul：IEEE，2012：152-156．
2	JIANG J， SEKAR V， ZHANG H ．Improving fairness，efficiency，and stability in HTTP-based adaptive video streaming with festive［J］．IEEE/ACM Transactions on Networking，2014，22（1）：326-340．
3	YIN X， JINDAL A， SEKAR V，et al ．A control-theoretic approach for dynamic adaptive video streaming over HTTP［C］∥ Proceedings of the 2015 ACM Confe-rence on Special Interest Group on Data Communication.New York：ACM，2015：325-338．
4	HE Q， DOVROLIS C， AMMAR M ．On the predictabi-lity of large transfer TCP throughput［J］．Computer Networks，2007，51（14）：3959-3977．
5	MIRZA M， SOMMERS J， BARFORD P，et al ．A machine learning approach to TCP throughput prediction［J］．IEEE/ACM Transactions on Networking，2010，18（4）：1026-1039．
6	SUN Y， YIN X， JIANG J，et al ．C S2P：improving video bitrate selection and adaptation with data-driven throughput prediction［C］∥ Proceedings of the 2016 ACM SIGCOMM Conference．New York：ACM，2016：272-285．
7	HIGUCHI T， SHINGU H， SHIMIZU N，et al ．Real-time video streaming based on TFRC using communication logging for 5G HetNet［J］．IEICE Transactions on Communications，2019，E102.B（8）：1538-1546．
8	KHANGURA S K， AKIN S ．Online available bandwidth estimation using multiclass supervised learning techniques［J］．Computer Communications，2021，170：177-189．
9	BENTALEB A， BEGEN A C， HAROUS S，et al ．Data-driven bandwidth prediction models and automated model selection for low latency［J］．IEEE Transactions on Multimedia，2020，23：2588-2601．
10	POKHARKAR S R， WAGH S J， DESHMUKH S N ．Machine learning based predictive mechanism for internet bandwidth［C］∥ Proceedings of 2021 the 6th Inter-national Conference for Convergence in Technology．Maharashtra：IEEE，2021：1-4．
11	MEI L， HU R， CAO H，et al ．Realtime mobile bandwidth prediction using LSTM neural network and Bayesian fusion［J］．Computer Networks，2020，182：107515/1-11．
12	MEI L， GOU J， CAI Y，et al ．Realtime mobile bandwidth and handoff predictions in 4G/5G networks［J］．Computer Networks， 2022，204：108736/1-11．
13	RIBEIRO V J， RIEDI R H， BARANIUK R G，et al ．PathChirp：efficient available bandwidth estimation for network paths ［R］．Washington D C：United States Department of Energy，2003．
14	KURDOGLU E， LIU Y， WANG Y，et al ．Real-time bandwidth prediction and rate adaptation for video calls over cellular networks［C］∥ Proceedings of the 7th International Conference on Multimedia Systems．New York：ACM，2016：1-11．
15	SAMBA A， BUSNEL Y， BLANC A，et al ．Instantaneous throughput prediction in cellular networks：which information is needed？［C］∥ Proceedings of 2017 IFIP/IEEE Symposium on Integrated Network and Service Management．Lisbon：IEEE，2017：624-627．
16	TAANI B， ZIMMERMANN R ．Spatio-temporal analysis of bandwidth maps for geo-predictive video streaming in mobile environments［C］∥ Proceedings of the 24th ACM International Conference on Multimedia.New York：ACM，2016：888-897．
17	SLIWA B， FALKENBERG R， LIEBIG T，et al ．Machine learning based context-predictive car-to-cloud communication using multi-layer connectivity maps for upcoming 5G networks ［C］∥ Proceedings of 2018 IEEE the 88th Vehicular Technology Conference．Chicago：IEEE，2018：1-7．
18	YUE C， JIN R，SUH K，et al ．LinkForecast：cellular link bandwidth prediction in LTE networks［J］．IEEE Transactions on Mobile Computing，2017，16（7）：1582-1594．
19	MINOVSKI D， ÖGREN N， MITRA K，et al ．Throughput prediction using machine learning in LTE and 5G networks［J］．IEEE Transactions on Mobile Computing，2023，22（3）：1825-1840．
20	RIKIĆ B， SAMARDŽIJA D， ČADOVSKI O，et al ．Cellular network bandwidth prediction in consumer applications［C］∥ Proceedings of 2021 IEEE International Conference on Consumer Electronics．Las Vegas：IEEE，2021：9427757/1-3．
21	ČADOVSKI O， ILIĆ V， RIKIĆ B，et al ．Proposal for bandwidth prediction using hybrid division model［C］∥Proceedings of 2021 IEEE International Conference on Consumer Electronics．Las Vegas：IEEE，2021：9427751/1-2．
22	NARAYANAN A， RAMADAN E， MEHTA R，et al ．Lumos5G：mapping and predicting commercial mmWave 5G throughput［C］∥ Proceedings of the ACM Internet Measurement Conference．New York：ACM，2020：176-193．
23	RACA D， LEAHY D， SREENAN C J，et al ．Beyond throughput， the next generation：a 5G dataset with channel and context metrics［C］∥ Proceedings of the 11th ACM Multimedia Systems Conference．New York：ACM，2020：303-308．
24	HAARNOJA T， ZHOU A， HARTIKAINEN K，et al ．Soft actor-critic algorithms and applications［EB/OL］．（2019-01-29）［2022-10-20］．．
25	CHRISTODOULOU P ．Soft actor-critic for discrete action settings［EB/OL］．（2019-10-18）［2022-10-20］．.
26	迈思源．Holowan网络损伤仿真仪［EB/OL］．（2020-01-12）［2022-10-20］．.
27	ZHOU Z H， FENG J ．Deep forest［J］．National Science Review，2019，6（1）：74-86．
28	HAYKIN S ．Adaptive filter theory［M］．4 ed．Delhi：Pearson Education，2002．
29	WAHAB A， AHMAD N， SCHORMANS J ．Variation in QoE of passive gaming video streaming for different packet loss ratios［C］∥ Proceedings of 2020 the Twelfth International Conference on Quality of Multimedia Experience．Athlone：IEEE，2020： 1-4．
30	LAGHARI A A， LAGHARI R A， WAGAN A A，et al ．Effect of packet loss and reorder on quality of audio streaming［J］．EAI Endorsed Transactions on Scalable Information Systems，2020，7（24）：1-7．
31	ADEYEMI-EJEYE A O， ALRESHOODI M， AL-JOBOURI L，et al ．Impact of packet loss on 4K UHD video for portable devices［J］．Multimedia Tools and Applications，2019，78：31733-31755．
32	AHMAD N， WAHAB A， SCHORMANS J ．Importance of cross-correlation of QoS metrics in network emulators to evaluate QoE of video streaming applications［C］∥ Proceedings of 2020 the 11th International Conference on Network of the Future．Bordeaux：IEEE，2020：43-47．

编辑推荐 0

Metrics

阅读次数

全文

243

HTML			PDF

最新录用	在线预览	正式出版	最新录用	在线预览	正式出版
0	0	17	3	0	223

来源	本网站	其他网站

次数	106	137
比例	44%	56%

摘要

139

最新录用	在线预览	正式出版

21	0	118

来源	本网站	其他网站

次数	61	78
比例	44%	56%

数据指标	相关系数
数据指标	步行场景	驾驶场景
Latitude	0.018	0.042
Longitude	-0.130	0.048
Speed	-0.045	-0.380
Compass	0.023	0.049
RSRP	0.470	0.450
SINR	0.470	0.470
RSRQ	-0.150	0.350
Tower_id	0.130	0.160

时延/ms	视频序列	RMSE		R²
时延/ms	视频序列	步行场景	驾驶场景	步行场景	驾驶场景
0	City	10.141	35.430	0.987	0.960
	Bridge	12.128	36.549	0.975	0.967
	Highway	13.707	39.629	0.972	0.958
	Harbour	10.389	32.512	0.994	0.962
50	City	14.774	100.899	0.852	0.831
	Bridge	14.178	92.122	0.860	0.846
	Highway	14.709	104.417	0.865	0.820
	Harbour	16.380	98.242	0.851	0.843
100	City	20.146	137.509	0.741	0.661
	Bridge	17.286	130.625	0.798	0.675
	Highway	20.210	139.996	0.752	0.654
	Harbour	19.318	131.605	0.787	0.669

[1]	郑娟毅, 董嘉豪, 张庆珏, 等. 基于残差密集网络的智能超表面信道估计算法[J]. 华南理工大学学报(自然科学版), 2024, 52(3): 102-111.
[2]	张广驰, 乐文英, 庞海舰, 等. IRS辅助认知无线携能通信网络的发射功率最小化算法[J]. 华南理工大学学报(自然科学版), 2023, 51(3): 110-123.
[3]	刘娇蛟, 陈阿粤, 马碧云. 基于直扩超声波宽带人体通信多点接入的自适应速率调整[J]. 华南理工大学学报(自然科学版), 2023, 51(3): 91-97.
[4]	吴皓威, 黄风娇, 闫莲, 等. 面向可疑中继通信网络的合法窃听方案[J]. 华南理工大学学报(自然科学版), 2022, 50(10): 70-79.
[5]	李松, 余意, 孙彦景, 等. 面向移动VR的部分缓存内容共享方法[J]. 华南理工大学学报（自然科学版）, 2021, 49(6): 122-130,140.
[6]	陈志坚蔡敏. 抗干扰高线性射频前端设计[J]. 华南理工大学学报（自然科学版）, 2016, 44(1): 37-43.
[7]	刘娇蛟马碧云. 基于联盟形成博弈论的异构无线传输速率分配[J]. 华南理工大学学报（自然科学版）, 2015, 43(9): 27-33.
[8]	潘伟锵胡少东刘靖. 混合业务下基于能效准则的基站开/关选择策略[J]. 华南理工大学学报（自然科学版）, 2015, 43(5): 30-34.
[9]	郎改平徐玉滨马琳. 基于非合作博弈论的异构网络选择算法[J]. 华南理工大学学报（自然科学版）, 2014, 42(5): 29-35.
[10]	顾军仲元红张振宇曾凡鑫. 用于多小区多用户 MIMO 系统的干扰抑制方法[J]. 华南理工大学学报（自然科学版）, 2013, 41(6): 42-46.
[11]	刘娇蛟曹燕. 基于无线随机接入代价函数的协作速率分配[J]. 华南理工大学学报(自然科学版), 2012, 40(6): 22-28.
[12]	师晓晔葛建华李靖高洋. 全分集协作系统C-HARQ 方案的设计及性能[J]. 华南理工大学学报（自然科学版）, 2011, 39(9): 108-114.
[13]	周晓明赖声礼. 旋转人头模型与单极天线手机相互作用的数值模拟[J]. 华南理工大学学报(自然科学版), 2004, 32(5): 30-33.
[14]	周晓明赖声礼. 单极手机与分层人头模型作用系统的数值模拟[J]. 华南理工大学学报(自然科学版), 2003, 31(12): 5-8.
[15]	朱宁西张齐. GSM网络短信息无线通信SCADA 系统的实现[J]. 华南理工大学学报(自然科学版), 2003, 31(12): 9-12.