Scalable Media Delivery Chain with Distributed Adaptation

SCALABLE MEDIA DELIVERY CHAIN
WITH DISTRIBUTED ADAPTATION
PhD Thesis
Michael Grafl
1st Supervisor: Prof. Hermann Hellwagner
2nd Supervisor: Dr. Cyril Concolato
Michael Grafl 1Scalable Media Delivery Chain with Distributed Adaptation

OUTLINE
 Introduction
 Motivation
 Research Objectives
 Technical Background
 Scalable Video Coding Framework
 SVC Tunneling
 Distributed Adaptation and Media Transport
 Conclusions and Future Work
Michael Grafl Scalable Media Delivery Chain with Distributed Adaptation 2

MOTIVATION
 Increasing amount of video traffic
 Video traffic (of all forms) to amount to ~86% of
global consumer traffic by 2016 [Cisco VNI 2011-2016]
 Today's media delivery chains are not utilizing their
resources optimally
 Redundant video encodings
 Content-agnostic transport at network level
 Low end-user device support for scalable media coding formats
 Integrate scalable media coding with a content-aware
distributed adaptation approach for media delivery

RESEARCH OBJECTIVES
1) Evaluate Scalable Video Coding (SVC) encoding
configurations and scalability features
2) Develop guidelines for SVC encoding in the context of
adaptive media streaming
3) Investigate SVC tunneling for device-independent
access
4) Analyze scalability features and adaptation configurations
for content- and context-aware media delivery
5) Investigate distributed adaptation in content-aware
networks for different transport mechanisms
6) Evaluate distributed media adaptation in an end-to-end
streaming system

OUTLINE
 Introduction
 SVC, DASH, and ALICANTE
 SVC Tunneling

SVC, DASH, AND ALICANTE
 Scalable Video Coding (SVC) extension of H.264/AVC
 Base layer (AVC-compatible) + enhancement layers
• Temporal, spatial, and quality scalability
 ~10% bitrate overhead compared to AVC (per layer)
 Dynamic Adaptive Streaming over HTTP (DASH)
 Media segments (e.g., 2-10 sec) in different representations,
described in manifest file
 Sequential download based on avail. bandwidth
 FP7 Project
 New Media Ecosystem with enhanced home-gateways
(Home-Boxes) & content-aware in-network adaptation of SVC

OUTLINE
 Introduction
 SVC Encoding Guidelines
 High-Definition SVC Performance Evaluations
 Hybrid SVC-DASH
 SVC Tunneling
 Conclusions

SCALABLE VIDEO CODING FRAMEWORK
Quality
ResolutionFrame rate Which bitrates?
Which resolutions?
Number of layers?
Combination of layers?

SVC ENCODING GUIDELINES
 Prominent streaming solutions providing AVC
encoding recommendations
 Apple HTTP Live Streaming
 Adobe HTTP Dynamic Streaming
 Microsoft Smooth Streaming
 YouTube
 MTV
 Recommendations were analyzed, aggregated and
adjusted for SVC streaming
  7 common resolutions and recommendations for
2 and 4 bitrates

HD SVC ENCODING PERFORMANCE
 Evaluate different SVC layer configurations &
encoder implementations for high-definition content
 Rate control modes
(constant bitrate vs. fixed quantization parameter)
 Combination of spatial and quality scalability
(multiple resolutions & multiple quality layers)
 Number of quality layers
 Requantization, combination of quality scalability modes

HYBRID SVC-DASH
 One stream (with quality layers) per resolution
instead of a single stream with all resolutions
 Resolutions
for device
classes
 Quality layers
for dynamic
adaptation
  Higher
viewing quality
User1
User2HD-Ready TV
Full-HD
TV
Mobile
User3
Traditional SVC-DASH
Hybrid SVC-DASH
Enhancement Layer 3
Resolution 1
Resolution 2
Resolution 3
Enhancement Layer 2
Enhancement Layer 1
SVC Base Layer

OUTLINE
 Introduction
 SVC Tunneling
 Concept and Considerations
 Evaluations

SVC TUNNELING
CANCAN
Home-Box Layer
HB
MANE
HB HB
HB
MANE MANE MANE
Autonomous
System
End-to-End Multimedia Communication (MPEG-2, MPEG-4, AVC, SVC, ...)
...
...
SVC (Layered-Multicast) Tunnel
HB
Autonomous
System
Context-
Aware
Adaptation
Dynamic,
Network-Aware
Adaptation

CONCEPTS AND CONSIDERATIONS
 SVC (layered-multicast) tunnel
 Adaptation of scalable media resource at MANE
 At the border to the user (Home-Box),
transcoding modules are deployed for
device-independent access
 Bandwidth savings compared to simulcast
 Transcoding
 Quality loss through re-encoding
 Real-Time Constraints

SVC TUNNELING EVALUATIONS
 Evaluated trade-off between quality loss and
bandwidth savings in multicast scenario
 MPEG-2 as source and target formats
 Test-bed gradually refined during 3 evaluations

OUTLINE
 Introduction
 SVC Tunneling
 Scalable Media Coding for Content-Aware Networking
 Representation Switch Smoothing
 End-to-End Adaptive Streaming System

DISTRIBUTED ADAPTATION
AND MEDIA TRANSPORT
UltraHD TV
MANE2
BufferBuffer
MANE1
SVC-Base Layer
Enhancement Layer 1
Enhancement Layer 2
U1
U2
U3
HD-Ready
Mobile
R1
R3
Full-HD
TV
R2
S1
S2

SCALABLE MEDIA CODING FOR
CONTENT-AWARE NETWORKING
 Identified Content-Aware Networking (CAN)
challenges and potentials based on use cases
for scalable media delivery
 Flow processing, caching/buffering, QoS/QoE
management
 Transport mechanisms
• RTP Unicast
• RTP Multicast
• P2P
• DASH

REPRESENTATION SWITCH SMOOTHING
 Avoid abrupt quality
switches
 Smooth transition
between
representations
 Initial subjective
test results (n=18)
 Improves
viewing quality
Representations
min bitrate
& quality
max bitrate
& quality
Time
Abrupt
change of
playback
quality
Representations
min bitrate
& quality
max bitrate
& quality
Time
Original
quality of
segment
Smooth transition
between
representations

END-TO-END ADAPTIVE STREAMING
SYSTEM
 Integrated previous findings into an
end-to-end adaptive streaming system prototype
 SVC encoding guidelines
 SVC-to-MPEG-2 transcoding
 Dynamic in-network adaptation
 System validation and evaluation
 End-to-end delay for streaming
 Quality improvement through dynamic adaptation

OUTLINE
 Introduction
 SVC Tunneling
 Findings & Future Work
 Publications

CONCLUSIONS & FUTURE WORK
 SVC encoding guidelines established
 Configurations & encoders evaluated
 SVC Tunneling approach developed and tested
 Trade-off between quality loss & bandwidth savings
 Distributed adaptation architecture examined
 Theoretical considerations & practical prototype
 Future Work
 Performance analysis of upcoming Scalable
High-Efficiency Video Coding (SHVC) standard
 SVC tunneling for evaluations high-definition content
 Elaborate coordination of distributed adaptation

PUBLICATIONS (1/2)
 M. Grafl, et al., "Scalable Video Coding Guidelines and Performance
Evaluations for Adaptive Media Delivery of High Definition Content", Proc.
IEEE ISCC, 2013.
 M. Grafl, et al., "Hybrid Scalable Video Coding for HTTP-based Adaptive Media
Streaming with High-Definition Content", Proc. IEEE WoWMoM, 2013.
 M. Grafl et al., "Scalable Media Coding enabling Content-Aware
Networking", IEEE MultiMedia, 2013.
 M. Grafl et al., "Distributed Adaptation Decision-Taking Framework and
Scalable Video Coding Tunneling for Edge and In-Network Media Adaptation",
Proc. IEEE TEMU, 2012.
 M. Grafl, C. Timmerer, and H. Hellwagner, "Quality Impact of Scalable Video
Coding Tunneling for Media-Aware Content Delivery", Proc. IEEE ICME, 2011.
 M. Grafl, "SVC Tunneling for Media-Aware Content Delivery: Impact on Video
Quality", Proc. IEEE WoWMoM - PhD Forum, 2011.
 M. Grafl et al., "Scalable Video Coding in Content-Aware Networks:
Research Challenges and Open Issues", in: N. Blefari-Melazzi, G. Bianchi,
and L. Salgarelli (eds.), Trustworthy Internet, Springer, 2011.
...

PUBLICATIONS (2/2)
 C. Timmerer et al., "Scalable Video Coding in Content-Aware Networks:
Research Challenges and Open Issues", Proc. ITWDC, 2010.
 C. Timmerer et al., "A Metadata Model for Peer-to-Peer Media Distribution", Proc.
WISMA, 2010.
 P. Kudumakis et al., "MPEG-M: A Digital Media Ecosystem for
Interoperable Applications", accepted for publication in Signal
Processing: Image Communication, scheduled for publication in 2013.
 G. Gardikis, E. Pallis, and M. Grafl, "Media-Aware Networks in Future
Internet Media", accepted for publication in: A. Kondoz and T. Dagiuklas
(eds.), 3D Future Internet Media, Springer, scheduled for publication in
2013.
 M. Grafl and C. Timmerer, "Representation Switch Smoothing for Adaptive HTTP
Streaming", accepted for publication in Proc. PQS, 2013.
Open-Source Software:
 "SVC Demux & Mux", https://sourceforge.net/projects/svc-demux-mux/, 2013.
 "SVC RTP MST", https://sourceforge.net/projects/svc-rtp-mst/, 2013.

THANKS FOR YOUR ATTENTION!
Questions?

Scalable Media Delivery Chain with Distributed Adaptation

More Related Content

What's hot (15)

Similar to Scalable Media Delivery Chain with Distributed Adaptation (20)

Recently uploaded (20)

Scalable Media Delivery Chain with Distributed Adaptation