SlideShare a Scribd company logo
REPRESENTATION SWITCH
SMOOTHING
FOR ADAPTIVE HTTP STREAMING
Michael Grafl and Christian Timmerer
4th International Workshop on Perceptual Quality of Systems (PQS 2013),
September 2-4 2013, Vienna, Austria
M. Grafl and C. Timmerer 1Representation Switch Smoothing
OUTLINE
 Introduction & Concept
 Implementation Options
 Evaluation & Results
 Discussion
 Conclusions
M. Grafl and C. Timmerer Representation Switch Smoothing 2
INTRODUCTION
 DASH: Dynamic Adaptive Streaming over HTTP
 Client downloads segments sequentially in best fitting
representation (quality, resolution, frame rate)
 Dynamically switch between representations (e.g.,
based on available bandwidth)
 Representation switches annoying to viewers
 How to reduce the quality impact of
representation switches?
M. Grafl and C. Timmerer Representation Switch Smoothing 3
CONCEPT
 Avoid abrupt
quality switches
 Smooth
transition
between
representations
M. Grafl and C. Timmerer Representation Switch Smoothing 4
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
IMPLEMENTATION OPTIONS
 Pre-decoder
 Remove picture fidelity data (transform coefficients) before
the decoder
 Suitable for Scalable Video Coding (SVC)
 Causes motion compensation drift
 In-decoder
 Remove picture fidelity data inside the decoder
 Less drift but decoder-dependent
 Post-decoder
 Post-processing filter mimicking distortion
 No drift
 Coding format independent
M. Grafl and C. Timmerer Representation Switch Smoothing 5
IMPLEMENTATION OPTIONS
 In-decoder implementation option
for SVC
M. Grafl and C. Timmerer Representation Switch Smoothing 6
Motion
Compen-
sation
Inverse
Quanti-
zation
Inverse
Quanti-
zation
Inverse
Trans-
form
Decoded
Picture Buffer
Predict-
ion
Data
Base
Residual
Enhance-
ment
Layer
Residual
+
+
+
+ Decoded
Frame
Motion
Compen-
sation
Inverse
Quanti-
zation
Inverse
Quanti-
zation
Inverse
Trans-
form
Decoded
Picture Buffer
Predict-
ion
Data
Base
Residual
Enhance-
ment
Layer
Residual
Decoded
Frame
+
+
+
+
Inverse
Trans-
form
+
+
Representation
Switch Smoothing
EVALUATION
 Subjective evaluation of down-switching scenario
 2 test sequences (15 sec, from TearsOfSteel,
1280x720, H.264, no sound)
 Quality Switching (after 10 sec) vs.
 Representation Switch Smoothing (5-sec transition)
 18 participants
 Pair-wise comparison (may repeat versions)
 Rating: Version a, Version b, No difference
 Smoothing simulated through repeating full-sequence
encoding and extraction of relevant frame
 Issue: temporal noise due to moving blocking artifacts
M. Grafl and C. Timmerer Representation Switch Smoothing 7
EVALUATION
 Per-frame PSNR for test sequences
M. Grafl and C. Timmerer Representation Switch Smoothing 8
high motion low motion
SCREENSHOTS
M. Grafl and C. Timmerer Representation Switch Smoothing 9
Sequence 1 Sequence 2
RESULTS
M. Grafl and C. Timmerer Representation Switch Smoothing 10
Preferred
Version
Sequence
Quality
Switching
Representation
Switch
Smoothing
No
Difference
Sequence 1 5 7 6
Sequence 2 3 12 3
DISCUSSION
 Representation Switch Smoothing: significant
improvement for Sequence 2 (low-motion)
 Temporal noise may have impacted results
 Longer transitions (e.g., 10 sec) may improve QoE
 Possible influence factors:
 Base quality, bitrate difference, cuts, resolution, spatio-
temporal complexity, duration of low quality
 Alternative approach: limited steps below
just-noticeable difference
M. Grafl and C. Timmerer Representation Switch Smoothing 11
CONCLUSIONS
 Idea: reduce annoyance of abrupt quality switches
by a smooth transition
 Avoid viewer distraction in adaptive HTTP streaming
 Implementation options discussed
 Subjective evaluation for down-switching
 Possible influence parameters identified
 Future work:
 Improve implementation (avoid temporal noise)
 Analyze impact of influence parameters
 Evaluated up-switching scenario
M. Grafl and C. Timmerer Representation Switch Smoothing 12
THANKS FOR YOUR ATTENTION!
Questions?
M. Grafl and C. Timmerer Representation Switch Smoothing 13
http://itec.aau.at/~mgrafl | @MyKey_ – http://aau.at/tewi/inf/itec/mmc/ | @itecMMC

More Related Content

PPT
Презентация на тему: «Административные барьеры в процессах международной торг...
PPTX
SumitomoBlack Ant ID Pestech 2014
PDF
Section based hex cell routing algorithm (sbhcr)
PPTX
How to find foreclosures in nj
PPTX
Sd pertemuan 1 & 2
PPT
я і ураїна. тварини
PDF
Upvc fascia
PPTX
Gyngestolen.dk
Презентация на тему: «Административные барьеры в процессах международной торг...
SumitomoBlack Ant ID Pestech 2014
Section based hex cell routing algorithm (sbhcr)
How to find foreclosures in nj
Sd pertemuan 1 & 2
я і ураїна. тварини
Upvc fascia
Gyngestolen.dk

Viewers also liked (10)

PPT
система управления персоналом
DOC
Dr. musli vërbani kurbani
PDF
2012 prius vs. 2012 nissan leaf - north hollywood toyota, los angeles new use...
PDF
Daily newsletter-e no500 6-6-2014
PDF
Mbrship.93
PDF
Nonfiction Reading Strategies
PPT
Chapter 08
PPS
Cary Nadler, Phd.
PDF
Banner-v1
PDF
Programming Camp for an Adult 大人のためのプログラミングキャンプ at PHP Matsuri 2010
система управления персоналом
Dr. musli vërbani kurbani
2012 prius vs. 2012 nissan leaf - north hollywood toyota, los angeles new use...
Daily newsletter-e no500 6-6-2014
Mbrship.93
Nonfiction Reading Strategies
Chapter 08
Cary Nadler, Phd.
Banner-v1
Programming Camp for an Adult 大人のためのプログラミングキャンプ at PHP Matsuri 2010
Ad

Similar to Representation Switch Smoothing for Adaptive HTTP Streaming (20)

PPT
Quality impact of scalable video coding tunneling for media aware content del...
PPT
Development of a Multipurpose Audio Transmission System on the Internet
PPT
Advances in Network-adaptive Video Streaming
PPT
Evaluation of bandwidth performance for interactive spherical video
PPTX
AQM performance for VOIP
PPT
Amir Zmora’s Presentation at eComm 2009
PDF
A 15 bit third order power optimized continuous time sigma delta modulator fo...
PPT
Digital Video And Compression
PDF
Spectral Analysis of Sample Rate Converter
DOCX
mini project on CLASS D AUDIO POWER AMPLIFIER
PPT
High Speed Edge Card Transitions
PDF
Realtime, Non-Intrusive Evaluation of VoIP Using Genetic Programming
PPTX
Cross-Layer Design of Raptor Codes for Video Multicast over 802.11n MIMO Chan...
PDF
Pulse modulation, Pulse Amplitude (PAM), Pulse Width (PWM/PLM/PDM), Pulse Pos...
PDF
HIGH SPEED CONTINUOUS-TIME BANDPASS Σ∆ ADC FOR MIXED SIGNAL VLSI CHIPS
PDF
6360_Project2_Report
PDF
IDT2014 Chirp Presentation
PDF
PDF
Design of 17-Bit Audio Band Delta-Sigma Analog to Digital Converter
PPT
LTE Uplink Power Control
Quality impact of scalable video coding tunneling for media aware content del...
Development of a Multipurpose Audio Transmission System on the Internet
Advances in Network-adaptive Video Streaming
Evaluation of bandwidth performance for interactive spherical video
AQM performance for VOIP
Amir Zmora’s Presentation at eComm 2009
A 15 bit third order power optimized continuous time sigma delta modulator fo...
Digital Video And Compression
Spectral Analysis of Sample Rate Converter
mini project on CLASS D AUDIO POWER AMPLIFIER
High Speed Edge Card Transitions
Realtime, Non-Intrusive Evaluation of VoIP Using Genetic Programming
Cross-Layer Design of Raptor Codes for Video Multicast over 802.11n MIMO Chan...
Pulse modulation, Pulse Amplitude (PAM), Pulse Width (PWM/PLM/PDM), Pulse Pos...
HIGH SPEED CONTINUOUS-TIME BANDPASS Σ∆ ADC FOR MIXED SIGNAL VLSI CHIPS
6360_Project2_Report
IDT2014 Chirp Presentation
Design of 17-Bit Audio Band Delta-Sigma Analog to Digital Converter
LTE Uplink Power Control
Ad

More from mgrafl (7)

PDF
Scalable Media Delivery Chain with Distributed Adaptation
PPTX
Scalable Video Coding Guidelines and Performance Evaluations for Adaptive Me...
PDF
Evaluation of Hybrid Scalable Video Coding for HTTP-based Adaptive Media Stre...
PPTX
SVC Adaptation in Content-Aware Networks
PPTX
Distributed Adaptation Decision-Taking Framework and Scalable Video Coding Tu...
PPTX
Scalable Video Coding in Content-Aware Networks
ODP
Modelling Music Albums In Modern File Formats
Scalable Media Delivery Chain with Distributed Adaptation
Scalable Video Coding Guidelines and Performance Evaluations for Adaptive Me...
Evaluation of Hybrid Scalable Video Coding for HTTP-based Adaptive Media Stre...
SVC Adaptation in Content-Aware Networks
Distributed Adaptation Decision-Taking Framework and Scalable Video Coding Tu...
Scalable Video Coding in Content-Aware Networks
Modelling Music Albums In Modern File Formats

Recently uploaded (20)

PPTX
MYSQL Presentation for SQL database connectivity
PDF
Empathic Computing: Creating Shared Understanding
PDF
Machine learning based COVID-19 study performance prediction
PDF
Spectral efficient network and resource selection model in 5G networks
PDF
Video forgery: An extensive analysis of inter-and intra-frame manipulation al...
PDF
Optimiser vos workloads AI/ML sur Amazon EC2 et AWS Graviton
PDF
MIND Revenue Release Quarter 2 2025 Press Release
PDF
Diabetes mellitus diagnosis method based random forest with bat algorithm
PDF
Network Security Unit 5.pdf for BCA BBA.
PPTX
Programs and apps: productivity, graphics, security and other tools
PDF
Build a system with the filesystem maintained by OSTree @ COSCUP 2025
PDF
TokAI - TikTok AI Agent : The First AI Application That Analyzes 10,000+ Vira...
PDF
Building Integrated photovoltaic BIPV_UPV.pdf
PDF
cuic standard and advanced reporting.pdf
PPTX
20250228 LYD VKU AI Blended-Learning.pptx
PDF
Dropbox Q2 2025 Financial Results & Investor Presentation
PPTX
Tartificialntelligence_presentation.pptx
PDF
Accuracy of neural networks in brain wave diagnosis of schizophrenia
PDF
Unlocking AI with Model Context Protocol (MCP)
PPT
Teaching material agriculture food technology
MYSQL Presentation for SQL database connectivity
Empathic Computing: Creating Shared Understanding
Machine learning based COVID-19 study performance prediction
Spectral efficient network and resource selection model in 5G networks
Video forgery: An extensive analysis of inter-and intra-frame manipulation al...
Optimiser vos workloads AI/ML sur Amazon EC2 et AWS Graviton
MIND Revenue Release Quarter 2 2025 Press Release
Diabetes mellitus diagnosis method based random forest with bat algorithm
Network Security Unit 5.pdf for BCA BBA.
Programs and apps: productivity, graphics, security and other tools
Build a system with the filesystem maintained by OSTree @ COSCUP 2025
TokAI - TikTok AI Agent : The First AI Application That Analyzes 10,000+ Vira...
Building Integrated photovoltaic BIPV_UPV.pdf
cuic standard and advanced reporting.pdf
20250228 LYD VKU AI Blended-Learning.pptx
Dropbox Q2 2025 Financial Results & Investor Presentation
Tartificialntelligence_presentation.pptx
Accuracy of neural networks in brain wave diagnosis of schizophrenia
Unlocking AI with Model Context Protocol (MCP)
Teaching material agriculture food technology

Representation Switch Smoothing for Adaptive HTTP Streaming

  • 1. REPRESENTATION SWITCH SMOOTHING FOR ADAPTIVE HTTP STREAMING Michael Grafl and Christian Timmerer 4th International Workshop on Perceptual Quality of Systems (PQS 2013), September 2-4 2013, Vienna, Austria M. Grafl and C. Timmerer 1Representation Switch Smoothing
  • 2. OUTLINE  Introduction & Concept  Implementation Options  Evaluation & Results  Discussion  Conclusions M. Grafl and C. Timmerer Representation Switch Smoothing 2
  • 3. INTRODUCTION  DASH: Dynamic Adaptive Streaming over HTTP  Client downloads segments sequentially in best fitting representation (quality, resolution, frame rate)  Dynamically switch between representations (e.g., based on available bandwidth)  Representation switches annoying to viewers  How to reduce the quality impact of representation switches? M. Grafl and C. Timmerer Representation Switch Smoothing 3
  • 4. CONCEPT  Avoid abrupt quality switches  Smooth transition between representations M. Grafl and C. Timmerer Representation Switch Smoothing 4 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
  • 5. IMPLEMENTATION OPTIONS  Pre-decoder  Remove picture fidelity data (transform coefficients) before the decoder  Suitable for Scalable Video Coding (SVC)  Causes motion compensation drift  In-decoder  Remove picture fidelity data inside the decoder  Less drift but decoder-dependent  Post-decoder  Post-processing filter mimicking distortion  No drift  Coding format independent M. Grafl and C. Timmerer Representation Switch Smoothing 5
  • 6. IMPLEMENTATION OPTIONS  In-decoder implementation option for SVC M. Grafl and C. Timmerer Representation Switch Smoothing 6 Motion Compen- sation Inverse Quanti- zation Inverse Quanti- zation Inverse Trans- form Decoded Picture Buffer Predict- ion Data Base Residual Enhance- ment Layer Residual + + + + Decoded Frame Motion Compen- sation Inverse Quanti- zation Inverse Quanti- zation Inverse Trans- form Decoded Picture Buffer Predict- ion Data Base Residual Enhance- ment Layer Residual Decoded Frame + + + + Inverse Trans- form + + Representation Switch Smoothing
  • 7. EVALUATION  Subjective evaluation of down-switching scenario  2 test sequences (15 sec, from TearsOfSteel, 1280x720, H.264, no sound)  Quality Switching (after 10 sec) vs.  Representation Switch Smoothing (5-sec transition)  18 participants  Pair-wise comparison (may repeat versions)  Rating: Version a, Version b, No difference  Smoothing simulated through repeating full-sequence encoding and extraction of relevant frame  Issue: temporal noise due to moving blocking artifacts M. Grafl and C. Timmerer Representation Switch Smoothing 7
  • 8. EVALUATION  Per-frame PSNR for test sequences M. Grafl and C. Timmerer Representation Switch Smoothing 8 high motion low motion
  • 9. SCREENSHOTS M. Grafl and C. Timmerer Representation Switch Smoothing 9 Sequence 1 Sequence 2
  • 10. RESULTS M. Grafl and C. Timmerer Representation Switch Smoothing 10 Preferred Version Sequence Quality Switching Representation Switch Smoothing No Difference Sequence 1 5 7 6 Sequence 2 3 12 3
  • 11. DISCUSSION  Representation Switch Smoothing: significant improvement for Sequence 2 (low-motion)  Temporal noise may have impacted results  Longer transitions (e.g., 10 sec) may improve QoE  Possible influence factors:  Base quality, bitrate difference, cuts, resolution, spatio- temporal complexity, duration of low quality  Alternative approach: limited steps below just-noticeable difference M. Grafl and C. Timmerer Representation Switch Smoothing 11
  • 12. CONCLUSIONS  Idea: reduce annoyance of abrupt quality switches by a smooth transition  Avoid viewer distraction in adaptive HTTP streaming  Implementation options discussed  Subjective evaluation for down-switching  Possible influence parameters identified  Future work:  Improve implementation (avoid temporal noise)  Analyze impact of influence parameters  Evaluated up-switching scenario M. Grafl and C. Timmerer Representation Switch Smoothing 12
  • 13. THANKS FOR YOUR ATTENTION! Questions? M. Grafl and C. Timmerer Representation Switch Smoothing 13 http://itec.aau.at/~mgrafl | @MyKey_ – http://aau.at/tewi/inf/itec/mmc/ | @itecMMC