1. DYNAMIC ADAPTIVE STREAMING
OVER HTTP @ ITEC
Christopher Müller, Stefan Lederer and Christian
Timmerer
Alpen-Adria Universität Klagenfurt (AAU) Faculty of Technical Sciences (TEWI)
Institute of Information Technology (ITEC) Multimedia Communication (MMC)
19.03.2012
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 1
2. AGENDA
(Short) Introduction to DASH & Motivation
Dataset & DASHEncoder
Peer-Assisted DASH
DASH under Vehicular Mobility
DASH @ ITEC
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 2
3. MOTIVATION
HTTP Streaming has become very popular on
the Internet
Media encoded a several bitrates, resolutions etc.
Clients request portions of the media due to
bandwidth conditions on-demand
Easy to use existing CDN structure
No NAT/Firewall issues due to HTTP
Various technologies
BUT: no standard in use!
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 3
4. DASH ARCHITECTURE
I. Sodagar, “The MPEG-DASH Standard for Multimedia Streaming Over the
Internet”, IEEE Multimedia, IEEE MultiMedia, October–December 2011, pp. 62–
67.
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 4
5. AGENDA
(Short) Introduction to DASH & Motivation
Dataset & DASHEncoder
Peer-Assisted DASH
DASH under Vehicular Mobility
DASH @ ITEC
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 5
6. DATASET
Dataset with DASH Content
Long sequences in high quality
Various segment-length versions
Free available for DASH experiments
PSNR values per frame
Problem: Content Rights
CC-Attribution 2.0 Generic (CC-BY 2.0) License or similar
Free to Share, Free to Remix
Note: YouTube introduces CC-BY in June 2011!
Negotiation with content owner
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 6
7. DATASET SEQUENCES
Name Source Quality Length Genre
Big Buck Bunny 1080p YUV 09:46 Animation
Elephants Dream 1080p YUV 10:54 Animation
Red Bull Playstreets 1080p, 6 Mbit H.264 01:37:28 Sport
The Swiss Account 1080p, 6 Mbit H.264 57:34 Sport
Valkaama 1080p, 6 Mbit H.264 01:33:05 Movie
Of Forest and Men SD 10:53 Movie
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 7
8. DASH DATASET
SEQUENCES
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 8
9. DASH CONTENT TYPES
Bitrates from 50 kbit/s. to 8 Mbit/s.
Segment Size:
Seconds: 1, 2, 4, 6, 10, 15
File Organization
Segmented
One file per representation, Byte Range Requests
e.g.: Big Buck Bunny
120 Encodings needed
Only 6 DASH Encoder runs
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 9
10. DASH ENCODER
• h.264: x264 / ffmpeg
• AAC: ffmpeg
Encode • [VP8/Webm encoding]
• MP4Box: Video / Audio / Video + Audio
• [Webm segmentation]
Container
• Generate one MPD
• Subfolder Organization
• MPD Variation (Byte Range Requests,etc.)
MPD
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 10
12. BENEFITS OF THE DATASET?
Public available and free DASH content
Common basis for evaluations
DASH Implementations, Stream Switching Algorithms, Network
and Cache Configurations, ...
Enables objective comparison of research results
Also used in the Peer Assisted Streaming evaluation
Provides usefull hints and practices of DASH content
generation
Publication:
S. Lederer, C. Müller and C. Timmerer, “Dynamic Adaptive Streaming over HTTP
Dataset”, In Proceedings of the ACM Multimedia Systems Conference 2012, Chapel
Hill, North Carolina, February 22-24, 2012.
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 12
13. AGENDA
(Short) Introduction to DASH & Motivation
Dataset & DASHEncoder
Peer-Assisted DASH
DASH under Vehicular Mobility
DASH @ ITEC
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 13
14. PEER ASSISTED STREAMING
Clients offer their downloaded segments
Segment requests are monitored by server
Integration in DASH MPD for future clients
Reduction of server load: Goal 10 - 20 %
Peer Traffic
Unsymmetrical network connection
Bottleneck: low upload resources
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 14
18. SIMULATION RESULTS - SERVER
- 15 % - 25 %
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 18
19. MAJOR FINDINGS
First evaluation simulation:
Up to 25 % bandwidth savings
Directly convertable to CDN cost reduction
Much more possibilities
Intelligent client clustering in larger scale
environments
Peer management & download algorithm
improvements
MPD update improvements
S. Lederer, C. Müller and C. Timmerer, “Peer-Assisted Dynamic Adaptive Streaming
over HTTP - System Design and Evaluation”, Packet Video Workshop 2012 (PV
2012), München, Germany, May 10-11, 2012 (to appear).
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 19
20. AGENDA
(Short) Introduction to DASH & Motivation
Dataset & DASHEncoder
Peer-Assisted DASH
DASH under Vehicular Mobility
DASH @ ITEC
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 20
21. METHODOLOGY
Experiment 1 / Track 1 (601 seconds)
Drive on the freeway A2, passing by the city of Villach in the
direction to Klagenfurt.
Experiment 2 / Track 2 (575 seconds)
From the Alpen-Adria-Universität Klagenfurt on the freeway A2 until
the service area around Techelsberg.
Experiment 3 / Track 3 (599 seconds)
From the service area around Techelsberg on the freeway A2 to the
exit of Klagenfurt.
Christopher Mueller An Evaluation of DASH in Vehicular Environments 21
22. EXPERIMENTAL SETUP
Bandwidth Shaping Node
Ubuntu 11.04
Linux Hierarchical Token Bucket (htb)
Available bandwidth will be adjusted every 2 seconds
Network Emulation Node
Emulates a round trip time of 150ms
Server
Server based on Windows Server 2008 and IIS
Client
Windows or Linux depending on the evaluation system
Christopher Mueller An Evaluation of DASH in Vehicular Environments 22
23. MICROSOFT SMOOTH STREAMING
Few Switches with a good average bitrate
Nevertheless close to unsmoothness at second 300
Christopher Mueller An Evaluation of DASH in Vehicular Environments 23
24. ADOBE DYNAMIC STREAMING
High number of unsmooth seconds
Rather binary and unpredictable
Christopher Mueller An Evaluation of DASH in Vehicular Environments 24
25. APPLE HTTP LIVE STREAMING
Very few switches with a lower bitrate
Large buffer for energy awareness
Christopher Mueller An Evaluation of DASH in Vehicular Environments 25
26. MPEG – DASH
Non stepwise switching
Good average bitrate and stable buffer
Christopher Mueller An Evaluation of DASH in Vehicular Environments 26
27. COMPARISON
Name Average Bitrate Average Switches Average Unsmoothness
[kbps] [Number of Switches] [Seconds]
Microsoft 1522 51 0
Adobe 1239 97 64
Apple 1162 7 0
MPEG – DASH
1464 166 0
Pipelined
Christopher Mueller An Evaluation of DASH in Vehicular Environments 27
28. MAJOR FINDINGS
Microsoft Smooth Streaming: performs best
Altough they don„t use pipelining, maybe specific TCP
implementation?
Adobe Dynamic Streaming: not usable, interesting
buffer
Apple HTTP Streaming: interesting features
MPEG-2 TS, Large Buffer, Conservative Approach (Energy?)
MPEG-DASH: good start with rather simple algorithm
Pipelining brings improvement + integrate further features:
parallel downloads, TCP modification, other Protocols (e.g.
SPDY)
C. Müller, S. Lederer and C. Timmerer, “An Evaluation of Dynamic Adaptive
Streaming over HTTP in Vehicular Environments”, ACM Workshop on Mobile
Video, Chapel Hill, North Carolina, February 24, 2012.
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 28
29. AGENDA
(Short) Introduction to DASH & Motivation
Dataset & DASHEncoder
Peer-Assisted DASH
DASH under Vehicular Mobility
DASH @ ITEC
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 29
31. STATISTICS
Besucher pro Woche
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 31
32. DASH @ ITEC
News:
24 Posts about new features, conferences, etc.
~ 150 comments + a huge number of emails
Cooperations
VideoLAN VLC
Mozilla
DASH-PG
Tools:
DASH VLC Plugin DASH Dataset
libDASH
DASH MPD Validator (Markus
DASHEncoder
Waltl)
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 32
33. DASH RESEARCH @ ITEC
Publications
Stefan Lederer, Christopher Müller and Christian Timmerer, “Peer-Assisted Dynamic
Adaptive Streaming over HTTP – System Design and Evaluation“, In Proceedings of the
IEEE International Packet Video Workshop 2012, Munich, Germany, May 10-11, 2012. (to
appear)
Christopher Müller, Stefan Lederer and Christian Timmerer, “An Evaluation of Dynamic
Adaptive Streaming over HTTP in Vehicular Environments”, In Proceedings of the ACM
Multimedia Systems Conference 2012 and the 4th ACM Workshop on Mobile Video, Chapel
Hill, North Carolina, February 24, 2012.
Stefan Lederer, Christopher Müller and Christian Timmerer, “Dynamic Adaptive Streaming
over HTTP Dataset”, In Proceedings of the ACM Multimedia Systems Conference 2012,
Chapel Hill, North Carolina, February 22-24, 2012.
Christopher Müller and Christian Timmerer, “A VLC Media Player Plugin enabling Dynamic
Adaptive Streaming over HTTP”, In Proceedings of the ACM Multimedia 2011 , Scottsdale,
Arizona, November 28, 2011.
Christopher Müller and Christian Timmerer, “A Test-Bed for the Dynamic Adaptive Streaming
over HTTP featuring Session Mobility”, In Proceedings of the ACM Multimedia Systems
Conference 2011, San Jose, California, February 23-25, 2011.
Christian Timmerer and Christopher Müller, “HTTP Streaming of MPEG Media”, In
Proceedings of the Streaming Day 2010, Udine, Italy, September 16-17, 2010.
Patents
Christopher Müller, Yuwen He, James Crenshaw, Bandwidth Adaptation for Dynamic
Adaptive Transfering of Multimedia, U.S. Provisional Application No.: 61/576,334.
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 33
34. THANK YOU FOR YOUR ATTENTION
http://dash.itec.aau.at
Christopher Müller and Stefan Lederer Dynamic Adaptive Streaming over HTTP 34