Vibra: neural adaptive streaming of VBR-encoded videos

Gangqiang Zhou; Run Wu; Miao Hu; Yipeng Zhou; Tom Z. J. Fu; Di Wu

doi:10.1145/3458306.3460993

Vibra: neural adaptive streaming of VBR-encoded videos

Gangqiang Zhou, Run Wu, Miao Hu^*, Yipeng Zhou, Tom Z. J. Fu, Di Wu

^*Corresponding author for this work

School of Computing

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

5 Citations (Scopus)

Abstract

Variable Bitrate (VBR) video encoding can provide much high quality-To-bits ratio compared to the widely adopted Constant Bitrate (CBR) encoding, and thus receives significant attentions by content providers in recent years. However, it is challenging to design efficient adaptive bitrate algorithms for VBR-encoded videos due to the sharply fluctuating chunk size and the resulting bitrate burstiness. In this paper, we propose a neural adaptive streaming framework called Vibra for VBR-encoded videos, which can well accommodate the high fluctuation of video chunk sizes and improve the quality-of-experience (QoE) of end users significantly. Our framework takes the characteristics of VBR-encoded videos into account, and adopts the technique of deep reinforcement learning to train a model for bitrate adaptation. We also conduct extensive trace-driven experiments, and the results show that Vibra outperforms the state-of-The-Art ABR algorithms with an improvement of 8.17%-29.21% in terms of the average QoE.

Original language	English
Title of host publication	NOSSDAV 2021 - Proceedings of the 2021 Workshop on Network and Operating System Support for Digital Audio and Video, Part of MMSys 2021
Place of Publication	New York, NY
Publisher	Association for Computing Machinery, Inc
Pages	2-8
Number of pages	7
ISBN (Electronic)	9781450384353
DOIs	https://doi.org/10.1145/3458306.3460993
Publication status	Published - 2021
Event	31st ACM Workshop on Network and Operating Systems Support for Digital Audio and Video, NOSSDAV 2021 - Virtual, Online, Turkey Duration: 28 Sept 2021 → 1 Oct 2021

Conference

Conference	31st ACM Workshop on Network and Operating Systems Support for Digital Audio and Video, NOSSDAV 2021
Country/Territory	Turkey
City	Virtual, Online
Period	28/09/21 → 1/10/21

Keywords

adaptive bitrate
reinforcement learning
VBR-encoded video
video streaming

Access to Document

10.1145/3458306.3460993

Cite this

@inproceedings{51dd7c6261d548a2b5f41f1588494cbf,

title = "Vibra: neural adaptive streaming of VBR-encoded videos",

abstract = "Variable Bitrate (VBR) video encoding can provide much high quality-To-bits ratio compared to the widely adopted Constant Bitrate (CBR) encoding, and thus receives significant attentions by content providers in recent years. However, it is challenging to design efficient adaptive bitrate algorithms for VBR-encoded videos due to the sharply fluctuating chunk size and the resulting bitrate burstiness. In this paper, we propose a neural adaptive streaming framework called Vibra for VBR-encoded videos, which can well accommodate the high fluctuation of video chunk sizes and improve the quality-of-experience (QoE) of end users significantly. Our framework takes the characteristics of VBR-encoded videos into account, and adopts the technique of deep reinforcement learning to train a model for bitrate adaptation. We also conduct extensive trace-driven experiments, and the results show that Vibra outperforms the state-of-The-Art ABR algorithms with an improvement of 8.17%-29.21% in terms of the average QoE.",

keywords = "adaptive bitrate, reinforcement learning, VBR-encoded video, video streaming",

author = "Gangqiang Zhou and Run Wu and Miao Hu and Yipeng Zhou and Fu, {Tom Z. J.} and Di Wu",

year = "2021",

doi = "10.1145/3458306.3460993",

language = "English",

pages = "2--8",

booktitle = "NOSSDAV 2021 - Proceedings of the 2021 Workshop on Network and Operating System Support for Digital Audio and Video, Part of MMSys 2021",

publisher = "Association for Computing Machinery, Inc",

note = "31st ACM Workshop on Network and Operating Systems Support for Digital Audio and Video, NOSSDAV 2021 ; Conference date: 28-09-2021 Through 01-10-2021",

}

Zhou, G, Wu, R, Hu, M, Zhou, Y, Fu, TZJ & Wu, D 2021, Vibra: neural adaptive streaming of VBR-encoded videos. in NOSSDAV 2021 - Proceedings of the 2021 Workshop on Network and Operating System Support for Digital Audio and Video, Part of MMSys 2021. Association for Computing Machinery, Inc, New York, NY, pp. 2-8, 31st ACM Workshop on Network and Operating Systems Support for Digital Audio and Video, NOSSDAV 2021, Virtual, Online, Turkey, 28/09/21. https://doi.org/10.1145/3458306.3460993

Vibra: neural adaptive streaming of VBR-encoded videos. / Zhou, Gangqiang; Wu, Run; Hu, Miao et al.
NOSSDAV 2021 - Proceedings of the 2021 Workshop on Network and Operating System Support for Digital Audio and Video, Part of MMSys 2021. New York, NY: Association for Computing Machinery, Inc, 2021. p. 2-8.

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding contribution › peer-review

TY - GEN

T1 - Vibra

T2 - 31st ACM Workshop on Network and Operating Systems Support for Digital Audio and Video, NOSSDAV 2021

AU - Zhou, Gangqiang

AU - Wu, Run

AU - Hu, Miao

AU - Zhou, Yipeng

AU - Fu, Tom Z. J.

AU - Wu, Di

PY - 2021

Y1 - 2021

N2 - Variable Bitrate (VBR) video encoding can provide much high quality-To-bits ratio compared to the widely adopted Constant Bitrate (CBR) encoding, and thus receives significant attentions by content providers in recent years. However, it is challenging to design efficient adaptive bitrate algorithms for VBR-encoded videos due to the sharply fluctuating chunk size and the resulting bitrate burstiness. In this paper, we propose a neural adaptive streaming framework called Vibra for VBR-encoded videos, which can well accommodate the high fluctuation of video chunk sizes and improve the quality-of-experience (QoE) of end users significantly. Our framework takes the characteristics of VBR-encoded videos into account, and adopts the technique of deep reinforcement learning to train a model for bitrate adaptation. We also conduct extensive trace-driven experiments, and the results show that Vibra outperforms the state-of-The-Art ABR algorithms with an improvement of 8.17%-29.21% in terms of the average QoE.

AB - Variable Bitrate (VBR) video encoding can provide much high quality-To-bits ratio compared to the widely adopted Constant Bitrate (CBR) encoding, and thus receives significant attentions by content providers in recent years. However, it is challenging to design efficient adaptive bitrate algorithms for VBR-encoded videos due to the sharply fluctuating chunk size and the resulting bitrate burstiness. In this paper, we propose a neural adaptive streaming framework called Vibra for VBR-encoded videos, which can well accommodate the high fluctuation of video chunk sizes and improve the quality-of-experience (QoE) of end users significantly. Our framework takes the characteristics of VBR-encoded videos into account, and adopts the technique of deep reinforcement learning to train a model for bitrate adaptation. We also conduct extensive trace-driven experiments, and the results show that Vibra outperforms the state-of-The-Art ABR algorithms with an improvement of 8.17%-29.21% in terms of the average QoE.

KW - adaptive bitrate

KW - reinforcement learning

KW - VBR-encoded video

KW - video streaming

UR - http://www.scopus.com/inward/record.url?scp=85111421748&partnerID=8YFLogxK

UR - http://purl.org/au-research/grants/arc/DE180100950

U2 - 10.1145/3458306.3460993

DO - 10.1145/3458306.3460993

M3 - Conference proceeding contribution

AN - SCOPUS:85111421748

SP - 2

EP - 8

BT - NOSSDAV 2021 - Proceedings of the 2021 Workshop on Network and Operating System Support for Digital Audio and Video, Part of MMSys 2021

PB - Association for Computing Machinery, Inc

CY - New York, NY

Y2 - 28 September 2021 through 1 October 2021

ER -

Vibra: neural adaptive streaming of VBR-encoded videos

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Building Intelligence into Online Video Services by Learning User Interests

Cite this

Vibra: neural adaptive streaming of VBR-encoded videos

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Projects

Building Intelligence into Online Video Services by Learning User Interests

Cite this