End-to-end video adaptation using frame rate optimization and TCP-friendly rate control

In this research, we focus on two areas of video adaptation: video rate control and frame rate control. The first part is on video rate control, which aims to adjust video bitrate to meet the network constraint. To that end, we propose a joint source rate and congestion control scheme called video TCP-friendly rate control (VTFRC) that incorporates the video bit rate characteristic into the TFRC rate. VTFRC uses a frame complexity measure and the rate gap between the TCP and TFRC rates to opportunistically encode the video at a higher rate. Experiments show that VTFRC improves video quality over existing scheme while maintaining TCP-friendliness. However, VTFRC needs the encoder meet a target bitrate and provide a frame complexity measure. To do this, we propose a complexity-based rate control scheme using edge energy. We show that this scheme can describe the individual complexities of the frames without needing any information on the whole video. Experiments show that the scheme produces a video stream that is closer to the target bitrate while improving on its video quality over existing schemes. The second part of this thesis is on frame rate control, which aims to maximize the video quality and continuity using the frame generation rates. To approach this, we propose a frame rate optimization framework that characterizes the frame rate control problem and use Lyapunov optimization to systematically derive decoupled optimization policies. Results show that the framework reduces the prebuffering requirements significantly with a modest tradeoff in video quality. We then examine different ways of improving the frame rate optimization framework. Firstly, we reformulated the framework based on a discontinuity penalty virtual buffer, which is the cumulative difference between the receiving interval and playout interval. We show that this discontinuity penalty correlates to the discontinuity of the video and enables a wider range of frame quality functions to be used with the framework. Secondly, we introduce a constraint on the playout slowdowns using a virtual buffer that records the cumulative delays. Results show that this provides a superior tradeoff between the video quality and the delay introduced compared to the existing scheme. Lastly, we show how the optimization policies can be derived in the presence of feedback delays. Analyses show that the delayed feedbacks had a minimal impact on the optimization policies.

Grantor dc:publisher

UNSW, Sydney

Year dc:date

2011

Author dc:creator

• Tan, Evan Li Shen

• Stochastic optimization
• Multimedia communications
• Rate control
• Tcp-friendly
• Video adaptation
• Adaptive media playout
• Lyapunov drift analysis
• h.264

Statement dc:rights

• open access
• CC BY-NC-ND 3.0
• free_to_read

Licence

• https://purl.org/coar/access_right/c_abf2
• https://creativecommons.org/licenses/by-nc-nd/3.0/au/

Language dc:language

EN

Identifier

https://doi.org/10.26190/unsworks/14991

OAI identifier oai:identifier

oai:unsworks.library.unsw.edu.au:1959.4/51348