Quality impact of scalable video coding tunneling for media aware content delivery

2,988 views

Published on

0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
2,988
On SlideShare
0
From Embeds
0
Number of Embeds
1,230
Actions
Shares
0
Downloads
27
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Quality impact of scalable video coding tunneling for media aware content delivery

  1. 1. QUALITY IMPACT OF SCALABLE VIDEO CODING TUNNELING FOR MEDIA-AWARE CONTENT DELIVERY Michael Grafl , Christian Timmerer, Hermann Hellwagner Alpen-Adria-Universität Klagenfurt, Austria Workshop on Multimedia-Aware Networking 2011 – ICME'11 M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  2. 2. OUTLINE <ul><li>Motivation </li></ul><ul><li>SVC for IPTV </li></ul><ul><li>SVC Content Delivery & Tunneling in ALICANTE </li></ul><ul><li>Test Setup </li></ul><ul><li>Results </li></ul><ul><ul><li>PSNR </li></ul></ul><ul><ul><li>Bandwidth Requirements </li></ul></ul><ul><li>Future Work </li></ul><ul><li>Conclusions </li></ul>M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  3. 3. MOTIVATION <ul><li>Scalable Video Coding (SVC) can save bandwidth for multicast content delivery </li></ul><ul><li>But many devices (e.g., smart phones) don't support SVC natively </li></ul><ul><li>Content often available in MPEG-2 (e.g., DVD) </li></ul><ul><li>Proposed architecture: </li></ul><ul><ul><li>Legacy formats at sender and receiver sides, SVC tunnel only for delivery </li></ul></ul><ul><ul><li>Transcoding at ingress and egress points </li></ul></ul><ul><ul><li>Can bandwidth savings outweigh quality loss??? </li></ul></ul>M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  4. 4. SVC FOR IPTV <ul><li>Scalable Video Coding (SVC) </li></ul><ul><ul><li>Layered video coding, Extension of H.264/AVC </li></ul></ul><ul><ul><li>Spatial, temporal, and quality (SNR) scalability </li></ul></ul><ul><ul><li>Coding gain over MPEG-2: ~ 59% [5], [6] </li></ul></ul><ul><li>Multicast use case for IPTV </li></ul><ul><ul><li>Multicast with heterogeneous terminals (e.g., different resolutions) </li></ul></ul><ul><ul><li>In-Network Adaptation: Adapt the stream within the network at Media Aware Network Elements (MANEs) </li></ul></ul><ul><ul><li>Simulcast (for AVC) vs. SVC mode </li></ul></ul>M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  5. 5. SVC CONTENT DELIVERY IN ALICANTE <ul><li>ALICANTE: </li></ul><ul><ul><li>&quot;Medi a Ecosystem Dep l oyment through Ubiqu i tous C ontent- A ware N e t work E nvironments&quot; </li></ul></ul><ul><li>FP7-ICT project </li></ul><ul><li>Goal: New Home-Box layer and CAN layer with cross-layer adaptation enabling cooperation between providers, operators, and end-users </li></ul><ul><ul><li>http://ict-alicante.eu </li></ul></ul>M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  6. 6. SVC TUNNELING IN ALICANTE M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling Source: [2]
  7. 7. TEST SETUP <ul><li>Three steps: </li></ul><ul><ul><li>Encode raw video to MPEG-2 </li></ul></ul><ul><ul><li>Transcode video to SVC (full transc.: decode & re-encode ) </li></ul></ul><ul><ul><li>Transcode video back to MPEG-2 (full transcoding) </li></ul></ul><ul><li>Videos: Two standard test sequences: </li></ul><ul><ul><li>&quot;Mobile&quot; and &quot;Foreman&quot; (CIF@30fps) </li></ul></ul><ul><li>Objective quality measurement: </li></ul><ul><ul><li>Peak Signal-To-Noise Ratio (PSNR) </li></ul></ul><ul><ul><li>Quality measured at several fixed target bitrates (same bitrate for all 3 representations) </li></ul></ul>M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  8. 8. RESULTS: MPEG-2  SVC  MPEG-2 M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling Software: ffmpeg (SVN-r25599) GPL MPEG-1/2 DirectShow Decoder Filter (v0.1.2) MainConcept SVC/AVC/H.264 Video Encoder & Decoder (v1.0.0.236699)
  9. 9. RESULTS: MPEG-2  SVC  MPEG-2 M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  10. 10. RESULTS: BJONTEGAARD DELTA <ul><li>Bjontegaard Delta (BD) [3], [4]: </li></ul><ul><ul><li>Average change of PSNR (  PSNR) and bitrate (  bitrate) between two r-d curves as scalar values </li></ul></ul><ul><ul><li>Calculation: </li></ul></ul><ul><li>Average results for repeated transcoding: </li></ul>M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling BD-PSNR BD-bitrate between 1 st and 2 nd curve (MP2  SVC) -0.8 dB 15 % between 2 nd and 3 rd curve (back to MP2) -1.3 dB 25 % between 1 st and 3 rd curve (MP2  SVC  MP2) -2.1 dB 43 %
  11. 11. MULTI-RATE MULTICAST SCENARIOS M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling Scenario 1: MPEG-2 Simulcast Scenario 2: SVC Tunneling Scenario 3: MPEG-2-to-SVC PDT Traditional MPEG-2 delivery chain. Simulcast of multiple resolutions. &quot;SVC Tunnel&quot; Server & terminals: MPEG-2 Network: SVC Server: MPEG-2 Network & terminals: SVC Transcoding High res. MPEG-2 Low res. MPEG-2 SVC SVC SVC MPEG-2 MPEG-2 SVC MPEG-2 MPEG-2 MPEG-2 MPEG-2
  12. 12. BANDWIDTH REQUIREMENTS: MULTI-RATE MULTICAST (SAME QUALITY AT SERVER) M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  13. 13. FUTURE WORK <ul><li>Comparison of SVC implementations (wrt. quality, performance): JSVM, MainConcept, bSoft </li></ul><ul><li>(Ultra-) High definition test sequences </li></ul><ul><li>In-depth analysis of transcoding scenarios (SVC Tunnel, only MPEG-2-to-SVC, etc.) </li></ul><ul><li>Improved encoder configuration </li></ul><ul><li>Further video formats </li></ul><ul><li>Subjective quality evaluation </li></ul><ul><li>TBD </li></ul>M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  14. 14. CONCLUSIONS <ul><li>ALICANTE architecture: </li></ul><ul><ul><li>SVC tunneling, supporting heterogeneous terminals </li></ul></ul><ul><ul><li>Reduces bandwidth requirements but impairs quality </li></ul></ul><ul><li>Objective quality measurement for repeated transcoding </li></ul><ul><li>PSNR drop: 2.1 dB for SVC tunnel </li></ul><ul><li>Bandwidth reqs: -31% wrt. MPEG-2 simulcast </li></ul><ul><li>Future work: performance, HD sequences, subjective quality </li></ul>M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  15. 15. LITERATURE <ul><li>[1] Z. Avramova et al., &quot;Comparison of simulcast and scalable video coding in terms of the required capacity in an IPTV network,&quot; Proceedings of PV, 2007. </li></ul><ul><li>[2] European Commission, &quot;ALICANTE, Annex I – Description of Work&quot;, FP7-ICT-2009-4, Grant agreement no. 248652, 2009. </li></ul><ul><li>[3] G. Bjontegaard, &quot; Calculation of average PSNR differences between RD-curves ,&quot; ITU-T SG16/Q6, 2001. </li></ul><ul><li>[4] G. Bjontegaard, &quot;Improvements of the BD-PSNR model,&quot; ITU-T SG16/Q6, 2008. </li></ul><ul><li>[5] T. Wiegand et al., &quot;Rate-constrained coder control and comparison of video coding standards,&quot; Circuits and Systems for Video Technology, IEEE Transactions on, vol. 13, no. 7, pp. 688-703, 2003. </li></ul><ul><li>[6] M. Wien et al., “Performance Analysis of SVC,” Circuits and Systems for Video Technology, IEEE Transactions on, vol. 17, no. 9, pp. 1194-1203, 2007.   </li></ul>M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling
  16. 16. THANK YOU FOR YOUR ATTENTION! Questions? M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling

×