Truong Cong Thang Univ. of Aizu, Aizu-Wakamatsu, Japan
Le, H.T. ; Nguyen, H.X. ; Pham, A.T. ; Jung Won Kang ; Yong Man Ro
Journal of Communications and Networks, Vol. 15, No. 6, Dec. 2013
Labmeeting - 20151013 - Adaptive Video Streaming over HTTP with Dynamic Resource Estimation
1. NTUST - Mobilizing Information Technology Lab
Adaptive Video Streaming over HTTP with
Dynamic Resource Estimation
Truong Cong Thang Univ. of Aizu, Aizu-Wakamatsu, Japan
Le, H.T. ; Nguyen, H.X. ; Pham, A.T. ; Jung Won Kang ; Yong Man Ro
Journal of Communications and Networks, Vol. 15, No. 6, Dec. 2013
Advisor:Jenq-Shiou Leu
Student:Bing-Syuan Wang
Date:2015/10/13
National Taiwan University of Science and Technology
2. NTUST - Mobilizing Information Technology Lab
Outline
• Introduction
• Overview of Adaptive HTTP Streaming
• Throughput Estimation
• Experiments
• Video Bitrate Estimation
• Experiments
• Conclusion
2
3. NTUST - Mobilizing Information Technology Lab
Introduction
• Hypertext transfer protocol (HTTP) streaming has become a cost effective
means for multimedia delivery.
• Client-based approach.
• Bandwidth estimation.
Bitrate estimation.
• Constant bitrate (CBR) video/Variable bitrate (VBR) video.
• This solution may enable CBR-streaming even though the video is encoded in
VBR mode
3
4. NTUST - Mobilizing Information Technology Lab
Overview of Adaptive HTTP Streaming
• Dynamic adaptive streaming over HTTP (DASH)
• Metadata: media presentation description (MPD)
4
5. NTUST - Mobilizing Information Technology Lab
Throughput Estimation
• Based on previous segment throughputs / last segment (aggressive method).
5
6. NTUST - Mobilizing Information Technology Lab
Throughput Estimation
• 𝑇𝑑
𝑒
(𝑖) : The estimated download throughput
• 𝐷 𝑒
(𝑖) : The estimated download duration
6
7. NTUST - Mobilizing Information Technology Lab
Experiments
• Segment duration 6s; target buffer level 12s.
• Client behavior with segment duration of 6 s and throughput estimation using:
(a) Aggressive method and (b) proposed method.
7
11.5s
6s
8. NTUST - Mobilizing Information Technology Lab
Experiments
• Comparison of the aggressive method and proposed method with segment
duration of 6 s using: (a) CDF of bitrate and (b) CDF of buffer level.
8
9. NTUST - Mobilizing Information Technology Lab
Video Bitrate Estimation
• The bitrate is not constant.
• The client will be able to dynamically select the highest possible bitrate.
• Quantization parameter (QP)
(1-unit QP → 12%)
• A. Inter-Stream Bitrate Estimation
• B. Intra-Stream Bitrate Estimation
9
10. NTUST - Mobilizing Information Technology Lab
Experiments
• Segment duration 2s; target buffer level 4s.
• Preparing the video alternatives in VBR mode.
10
11. NTUST - Mobilizing Information Technology Lab
Experiments
• Based on average bitrate (ABR), highest bitrate (HBR), estimated bitrate (EBR)
11
12. NTUST - Mobilizing Information Technology Lab
Conclusion
• The proposed method can capture quickly the changes of throughput, and
then adjust video bitrate accordingly.
• The estimation methods essentially do not affect the playback quality (less
than 1 ms).
• The proposed solutions were effective to maintain a stable buffer under
fluctuations of bandwidth and video bitrate.
12
Notas do Editor
Feature extraction: parameter
Controller: decide to adjust the computation model