Dynamic Adaptive Streaming over HTTP Dataset

17
ACM Multimedia Systems 2012 Feb. 22-24 2012 hapel Hill Dynamic Adaptive Streaming over HTTP (DASH) Dataset Stefan Lederer, Christopher Müller and Christian Timmerer Alpen-Adria-Universität Klagenfurt (AAU)

description

This is the presentation of our paper "Dynamic Adaptive Streaming over HTTP Dataset" at the ACM Multimedia Systems 2012.

Transcript of Dynamic Adaptive Streaming over HTTP Dataset

Page 1: Dynamic Adaptive Streaming over HTTP Dataset

AC

M M

ulti

med

ia

Syst

ems

20

12

F e b . 2 2 - 2 42 0 1 2

C h a p e l H i l l

Dynamic Adaptive Streaming over HTTP (DASH)Dataset

Stefan Lederer, Christopher Müller and Christian Timmerer

Alpen-Adria-Universität Klagenfurt (AAU)

Page 2: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 2

Motivation

• HTTP Streaming has become very popular on the Internet – Various different bitrates of the video are available on the

server --> choose the best version to load – Easy to use existing CDN structure– No NAT/Firewall issues due to HTTP – Various technologies

• BUT: no standard in use!

Page 3: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 3

DASH

• Dynamic Adaptive Streaming over HTTP (DASH)– Will be Part 6 of MPEG-B– Existing DASH Plugin for VLC

Page 4: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 4

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

Page 5: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 5

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

Page 6: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 6

DASH Dataset Sequences

Page 7: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 7

Bitrates and Resolutions

# Animation Sport Movie1 50 kbit/s, 320x240 100 kbit/s, 320x240 50 kbit/s, 320x2402 100 kbit/s, 320x240 150 kbit/s, 320x240 100 kbit/s, 320x2403 150 kbit/s, 320x240 200 kbit/s, 480x360 150 kbit/s, 320x2404 200 kbit/s, 480x360 250 kbit/s, 480x360 200 kbit/s, 480x3605 250 kbit/s, 480x360 300 kbit/s, 480x360 250 kbit/s, 480x3606 300 kbit/s, 480x360 400 kbit/s, 480x360 300 kbit/s, 480x3607 400 kbit/s, 480x360 500 kbit/s, 854x480 400 kbit/s, 480x3608 500 kbit/s, 480x360 700 kbit/s, 854x480 500 kbit/s, 854x4809 600 kbit/s, 854x480 900 kbit/s, 854x480 600 kbit/s, 854x480

10 700 kbit/s, 854x480 1,2 Mbit/s, 854x480 700 kbit/s, 854x48011 900 kbit/s,1280x720 1,5 Mbit/s,1280x720 900 kbit/s,1280x72012 1,2 Mbit/s,1280x720 2,0 Mbit/s,1280x720 1,2 Mbit/s,1280x72013 1,5 Mbit/s,1280x720 2,5 Mbit/s,1280x720 1,5 Mbit/s,1280x72014 2,0 Mbit/s,1280x720 3,0 Mbit/s,1920x1080 2,0 Mbit/s,1920x108015 2,5 Mbit/s,1920x1080 4,0 Mbit/s,1920x1080 2,5 Mbit/s,1920x108016 3,0 Mbit/s,1920x1080 5,0 Mbit/s,1920x1080 3,0 Mbit/s,1920x108017 4,0 Mbit/s,1920x1080 6,0 Mbit/s,1920x1080 4,0 Mbit/s,1920x108018 5,0 Mbit/s,1920x1080   5,0 Mbit/s,1920x108019 6,0 Mbit/s,1920x1080   6,0 Mbit/s,1920x108020 8,0 Mbit/s,1920x1080    

Page 8: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 8

DASH Content Types

• 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

Page 9: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 9

DASHEncoder

• DASH Content Generation Tool– Encoding + Multiplexing + MPD generation– Generates isoffmain profile compliant MPDs– Fully configurable using a config-file– Enables batch processing– Currently uses x264 and GPAC‘s MP4Box– Easy extensible to further

encoders & multiplexers

Page 10: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 10

DASH Encoder

Encode

• h.264: x264 / ffmpeg• AAC: ffmpeg• [WebM, etc.]

Container

• MP4Box: Video / Audio / Video + Audio• [e.g. WebM/MKV Segmenter]

MPD

• Generate one MPD• Subfolder Organization• MPD Transformation

Page 11: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 11

Connection Type Evaluation

• Peristent vs. non-persisten connections– Bandwidth variations in high delay network– Influence of segment length to streaming

performance– Apache Web Server + DASH VLC Plugin (AAU)

Page 12: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 12

Non-/Persistent Conn. Segment Length: 2 sec.

Page 13: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 13

Non-/Persistent Conn. Segment Length: 15 sec.

Page 14: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 14

Evaluation Results

Page 15: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 15

Conclusion

• Public available and free DASH dataset• Common basis for evaluations

– DASH Implementations, Stream Switching Algorithms, Network and Cache Configurations, ...

– Enables objective comparison of research results

• Basic content generation tool: DASHEncoder• Fully compatible to DASH VLC Plugin of ITEC/AAU• Evaluation showing influence of segment length

Page 16: Dynamic Adaptive Streaming over HTTP Dataset

ACM MMSys 2012, Feb. 22-24, Chapel Hill/NC/USA Slide 16

Future Work

• Dataset– Further profiles and MPDs– Further media segment formats

• Evaluation of mobile scenarios– Under vehicular & pedestrian mobility

• Peer-assisted DASH– Reduce server bandwidth requirements– Inter-peer synchronisation & communication

Page 17: Dynamic Adaptive Streaming over HTTP Dataset

Thank You!

http://dash.itec.aau.at

Ste

fan

Led

erer

C

hri

sto

ph

er M

ülle

rC

hri

stia

n T

imm

erer