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

QUALITY IMPACT OF QUALITY IMPACT OF SCALABLE VIDEO CODING TUNNELING SCALABLE VIDEO CODING TUNNELING

FOR MEDIA-AWARE CONTENT DELIVERYFOR MEDIA-AWARE CONTENT DELIVERY

Michael Grafl, Christian Timmerer, Hermann Hellwagner

Alpen-Adria-Universität Klagenfurt, Austria

Workshop on Multimedia-Aware Networking 2011 – ICME'11M. Grafl, C. Timmerer, H. Hellwagner 1Quality Impact of SVC Tunneling

OUTLINEOUTLINE

Motivation SVC for IPTV SVC Content Delivery & Tunneling in ALICANTE Test Setup Results

PSNR Bandwidth Requirements

Future Work Conclusions

M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling 2

MOTIVATIONMOTIVATION

Scalable Video Coding (SVC) can save bandwidth for multicast content delivery

But many devices (e.g., smart phones) don't support SVC natively

Content often available in MPEG-2 (e.g., DVD) Proposed architecture:

Legacy formats at sender and receiver sides, SVC tunnel only for delivery

Transcoding at ingress and egress points Can bandwidth savings outweigh quality loss???


SVC FOR IPTVSVC FOR IPTV

Scalable Video Coding (SVC) Layered video coding, Extension of H.264/AVC Spatial, temporal, and quality (SNR) scalability Coding gain over MPEG-2: ~ 59% [5], [6]

Multicast use case for IPTV Multicast with heterogeneous terminals (e.g., different

resolutions) In-Network Adaptation: Adapt the stream within the

network at Media Aware Network Elements (MANEs) Simulcast (for AVC) vs. SVC mode


SVC CONTENT DELIVERY IN SVC CONTENT DELIVERY IN ALICANTEALICANTE


ALICANTE: "Media Ecosystem Deployment through Ubiquitous

Content-Aware Network Environments"

FP7-ICT project

Goal: New Home-Box layer and CAN layer with cross-layer adaptation enabling cooperation between providers, operators, and end-users

http://ict-alicante.eu

SVC TUNNELING IN ALICANTESVC TUNNELING IN ALICANTE


Source: [2]

TEST SETUPTEST SETUP

Three steps:1. Encode raw video to MPEG-2

2. Transcode video to SVC (full transc.: decode & re-encode)

3. Transcode video back to MPEG-2 (full transcoding)

Videos: Two standard test sequences: "Mobile" and "Foreman" (CIF@30fps)

Objective quality measurement: Peak Signal-To-Noise Ratio (PSNR) Quality measured at several fixed target bitrates

(same bitrate for all 3 representations)M. Grafl, C. Timmerer, H. Hellwagner Quality Impact of SVC Tunneling 7

RESULTS: MPEG-2RESULTS: MPEG-2SVCSVCMPEG-2MPEG-2


22

24

26

28

30

32

34

36

38

40

1000 2000 3000 4000 5000 6000 7000

PSN

R in

dB

Bitrate in kbps

PSNR-Y "Mobile" CIF@30

MP2

SVC(MP2)

MP2(SVC(MP2))

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)

RESULTS: MPEG-2RESULTS: MPEG-2SVCSVCMPEG-2MPEG-2


RESULTS: BJONTEGAARD DELTARESULTS: BJONTEGAARD DELTA

Bjontegaard Delta (BD) [3], [4]: Average change of PSNR (PSNR) and bitrate

(bitrate) between two r-d curves as scalar values Calculation:

Average results for repeated transcoding:


BD-PSNR BD-bitrate

between 1st and 2nd curve (MP2SVC) -0.8 dB 15 %

between 2nd and 3rd curve (back to MP2) -1.3 dB 25 %

between 1st and 3rd curve (MP2SVCMP2) -2.1 dB 43 %

MULTI-RATE MULTICAST MULTI-RATE MULTICAST SCENARIOSSCENARIOS


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.

"SVC Tunnel"Server & terminals: MPEG-2

Network: SVC

Server: MPEG-2Network & terminals: SVC

TranscodingHigh res. MPEG-2Low res. MPEG-2SVCSVCSVCMPEG-2

MPEG-2SVCMPEG-2

MPEG-2MPEG-2MPEG-2

BANDWIDTH REQUIREMENTS: BANDWIDTH REQUIREMENTS: MULTI-RATE MULTICAST (SAME QUALITY AT MULTI-RATE MULTICAST (SAME QUALITY AT

SERVER)SERVER)


TBD

FUTURE WORKFUTURE WORK

Comparison of SVC implementations (wrt. quality, performance): JSVM, MainConcept, bSoft

(Ultra-) High definition test sequences In-depth analysis of transcoding scenarios

(SVC Tunnel, only MPEG-2-to-SVC, etc.) Improved encoder configuration Further video formats Subjective quality evaluation


CONCLUSIONSCONCLUSIONS

ALICANTE architecture: SVC tunneling, supporting heterogeneous terminals Reduces bandwidth requirements but impairs quality

Objective quality measurement for repeated transcoding

PSNR drop: 2.1 dB for SVC tunnelPSNR drop: 2.1 dB for SVC tunnel Bandwidth reqs: -31% wrt. MPEG-2 simulcastBandwidth reqs: -31% wrt. MPEG-2 simulcast Future work: performance, HD sequences,

subjective quality


LITERATURELITERATURE

[1] Z. Avramova et al., "Comparison of simulcast and scalable video coding in terms of the required capacity in an IPTV network," Proceedings of PV, 2007.

[2] European Commission, "ALICANTE, Annex I – Description of Work", FP7-ICT-2009-4, Grant agreement no. 248652, 2009.

[3] G. Bjontegaard, "Calculation of average PSNR differences between RD-curves," ITU-T SG16/Q6, 2001.

[4] G. Bjontegaard, "Improvements of the BD-PSNR model," ITU-T SG16/Q6, 2008.

[5] T. Wiegand et al., "Rate-constrained coder control and comparison of video coding standards," Circuits and Systems for Video Technology, IEEE Transactions on, vol. 13, no. 7, pp. 688-703, 2003.

[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.


THANK YOU FOR YOUR THANK YOU FOR YOUR ATTENTION!ATTENTION!

Questions?


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

Technology

svc svc svc mpeg

svc server

svc mode

svc pdt traditional

svc transcoding high

ndcurve mp2 svc

outline motivation svc

iptv svc content delivery