Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
Optimizing User QoE through Overlay Routing, Bandwidth Management and Dynamic Transcoding Maarten Wijnants , Wim Lamotte H...
Outline <ul><li>Introduction and Motivation </li></ul><ul><li>End-to-End QoE Optimization Architecture </li></ul><ul><ul><...
Introduction and Motivation <ul><li>Rising networked access of MM services </li></ul><ul><ul><li>Strict requirements on tr...
Introduction and Motivation <ul><li>Current networks often unable to provide MM users an acceptable usage experience </li>...
End-to-End QoE Optimization Architecture <ul><li>Proposed architecture employs 2-tier approach to achieve E2E QoE optimiza...
End-to-End QoE Optimization Architecture <ul><li>Overlay Server (OS) </li></ul><ul><ul><li>Deployed in network core </li><...
End-to-End QoE Optimization Architecture <ul><li>Network Intelligence Proxy (NIProxy) </li></ul><ul><ul><li>Deployed close...
End-to-End QoE Optimization Architecture <ul><li>Network Intelligence Proxy </li></ul><ul><ul><li>Network traffic shaping ...
End-to-End QoE Optimization Architecture 23/06/2008 ADAMUS2008 Resilient network core routing Overlay layer Network layer ...
H.264/AVC Video Transcoding <ul><li>Focus on bit rate reduction </li></ul><ul><li>Operates entirely in compressed domain <...
Evaluation Experimental Setup <ul><li>Experimental results produced on testbed </li></ul><ul><ul><li>10 Linux PCs: 3 OSs, ...
Evaluation Experimental Results <ul><li>Experiment </li></ul><ul><ul><li>2 H.264/AVC flows streamed to each client </li></...
Evaluation Discussion <ul><li>Findings </li></ul><ul><ul><li>Client’s last mile downstream capacity respected    Last mil...
Conclusions <ul><li>E2E QoE optimization platform </li></ul><ul><ul><li>Resilient overlay routing service circumvents erra...
Thank you for your attention! Any questions?
Upcoming SlideShare
Loading in …5
×

Optimizing User QoE through Overlay Routing, Bandwidth ...

615 views

Published on

  • Be the first to comment

  • Be the first to like this

Optimizing User QoE through Overlay Routing, Bandwidth ...

  1. 1. Optimizing User QoE through Overlay Routing, Bandwidth Management and Dynamic Transcoding Maarten Wijnants , Wim Lamotte Hasselt University - Expertise Centre for Digital Media Bart De Vleeschauwer, Filip De Turck, Bart Dhoedt, Piet Demeester Ghent University – IBCN - Department of Information Technology Peter Lambert, Dieter Van de Walle, Jan De Cock, Stijn Notebaert, Rik Van de Walle Ghent University – MMLab - Department of Electronics and Information Systems Optimizing User QoE through Overlay Routing, Bandwidth Management and Dynamic Transcoding
  2. 2. Outline <ul><li>Introduction and Motivation </li></ul><ul><li>End-to-End QoE Optimization Architecture </li></ul><ul><ul><li>Overlay Routing Components </li></ul></ul><ul><ul><li>Network Intelligence Proxy </li></ul></ul><ul><li>H.264/AVC Video Transcoding </li></ul><ul><li>Evaluation </li></ul><ul><ul><li>Experimental Setup </li></ul></ul><ul><ul><li>Experimental Results </li></ul></ul><ul><ul><li>Discussion </li></ul></ul><ul><li>Conclusions </li></ul>23/06/2008 ADAMUS2008
  3. 3. Introduction and Motivation <ul><li>Rising networked access of MM services </li></ul><ul><ul><li>Strict requirements on transportation network </li></ul></ul><ul><li>Service consumption environment has become highly heterogeneous </li></ul><ul><ul><li>Growing service dependability & adaptation requirements </li></ul></ul><ul><li>Current-gen networks often not capable of guaranteeing requirements are satisfied </li></ul><ul><ul><li>Internet routing service is best-effort </li></ul></ul><ul><ul><li>Constrained access network connections </li></ul></ul><ul><ul><ul><li>Insufficient last mile bandwidth  Congestion </li></ul></ul></ul>23/06/2008 ADAMUS2008
  4. 4. Introduction and Motivation <ul><li>Current networks often unable to provide MM users an acceptable usage experience </li></ul><ul><ul><li>More formally: Quality of Experience (QoE) </li></ul></ul><ul><li>Network architecture supporting full end-to-end QoE optimization needed </li></ul><ul><ul><li>Proposed by us in previous work </li></ul></ul><ul><li>We extended network architecture with a H.264/AVC video transcoding service </li></ul><ul><ul><li>Dynamic rate adaptation of H.264/AVC video </li></ul></ul><ul><ul><li>Enables further optimization of user QoE </li></ul></ul>23/06/2008 ADAMUS2008
  5. 5. End-to-End QoE Optimization Architecture <ul><li>Proposed architecture employs 2-tier approach to achieve E2E QoE optimization </li></ul><ul><ul><li>Enhance data dissemination in network core </li></ul></ul><ul><ul><ul><li>Through provision resilient overlay routing service </li></ul></ul></ul><ul><ul><li>Last mile user QoE optimization </li></ul></ul><ul><ul><ul><li>Network traffic shaping </li></ul></ul></ul><ul><ul><ul><li>Multimedia service provision </li></ul></ul></ul><ul><li>Consists of 3 types of components </li></ul><ul><ul><li>Overlay Server </li></ul></ul><ul><ul><li>Overlay Access Component </li></ul></ul><ul><ul><li>Network Intelligence Proxy </li></ul></ul>23/06/2008 ADAMUS2008 Resilient overlay routing Last mile QoE optimization
  6. 6. End-to-End QoE Optimization Architecture <ul><li>Overlay Server (OS) </li></ul><ul><ul><li>Deployed in network core </li></ul></ul><ul><ul><li>Maintain an overlay topology </li></ul></ul><ul><ul><ul><li>Perform active monitoring to obtain connectivity info </li></ul></ul></ul><ul><ul><ul><li>Info is used to construct overlay routing tables </li></ul></ul></ul><ul><li>Overlay Access Component (AC) </li></ul><ul><ul><li>Located near end-users </li></ul></ul><ul><ul><li>Decide when to forward traffic to overlay servers (based on quality direct IP connection) </li></ul></ul><ul><li>OSs exploit overlay routing tables to transport traffic to AC close to target node </li></ul>23/06/2008 ADAMUS2008
  7. 7. End-to-End QoE Optimization Architecture <ul><li>Network Intelligence Proxy (NIProxy) </li></ul><ul><ul><li>Deployed close to end-user </li></ul></ul><ul><ul><li>Improve user QoE by intelligently managing last mile content delivery to clients </li></ul></ul><ul><ul><li>Context introduction in transportation network </li></ul></ul><ul><ul><ul><li>Network awareness: Access channel conditions </li></ul></ul></ul><ul><ul><ul><li>Application awareness: E.g. stream significance </li></ul></ul></ul><ul><ul><li>Last mile network traffic shaping: Orchestrate last mile BW consumption of applications </li></ul></ul><ul><ul><ul><li>Prevent over-encumbrance of client's access link </li></ul></ul></ul><ul><ul><ul><li>Intelligently allocate available client downstream BW (based on application awareness) </li></ul></ul></ul>23/06/2008 ADAMUS2008
  8. 8. End-to-End QoE Optimization Architecture <ul><li>Network Intelligence Proxy </li></ul><ul><ul><li>Network traffic shaping operates by organizing network flows in a stream hierarchy </li></ul></ul><ul><ul><ul><li>Internal nodes: Implement BW distribution technique </li></ul></ul></ul><ul><ul><ul><ul><li>E.g. WeightStream </li></ul></ul></ul></ul><ul><ul><ul><li>Leaf nodes: Correspond to actual network flows </li></ul></ul></ul><ul><ul><ul><ul><li>Discrete : Toggle between discrete # of BW values </li></ul></ul></ul></ul><ul><ul><ul><ul><li>Continuous : Any rate in [0, max flow BW usage] </li></ul></ul></ul></ul><ul><ul><li>Multimedia service provision </li></ul></ul><ul><ul><ul><li>Perform computation/processing on network flows </li></ul></ul></ul><ul><ul><ul><li>Services can query and exploit NIProxy’s awareness </li></ul></ul></ul><ul><ul><ul><li>Implementation: Plug-in approach (dynamic loading) </li></ul></ul></ul>23/06/2008 ADAMUS2008
  9. 9. End-to-End QoE Optimization Architecture 23/06/2008 ADAMUS2008 Resilient network core routing Overlay layer Network layer Last mile QoE optimization
  10. 10. H.264/AVC Video Transcoding <ul><li>Focus on bit rate reduction </li></ul><ul><li>Operates entirely in compressed domain </li></ul><ul><ul><li>Only entropy decoding and encoding required </li></ul></ul><ul><ul><li># transformed coefficients are set to 0 based on dynamically changing cut-off frequency </li></ul></ul><ul><ul><li>Transcoder steered by rate control alg </li></ul></ul><ul><ul><ul><li>Ensures desired bit rate is achieved (Track buffer occupancy  Estimate bit budget current frame  Dynamically adjust cut-off frequency) </li></ul></ul></ul><ul><li>Integrated as plug-in for NIProxy </li></ul><ul><ul><li>Dynamically set desired bit rate H.264 flows </li></ul></ul><ul><ul><ul><li>Enables H.264 flow mgmnt using continuous leaves </li></ul></ul></ul>23/06/2008 ADAMUS2008
  11. 11. Evaluation Experimental Setup <ul><li>Experimental results produced on testbed </li></ul><ul><ul><li>10 Linux PCs: 3 OSs, 2 ACs, 2 NIProxies, 2 MM clients, video server, 2 Click impairment nodes </li></ul></ul><ul><ul><li>Click nodes emulate varying network condition </li></ul></ul><ul><ul><ul><li>Introduce random packet loss in core network </li></ul></ul></ul><ul><ul><ul><li>Enforce BW restriction on last mile </li></ul></ul></ul><ul><ul><li>Communication session server to each client </li></ul></ul>23/06/2008 ADAMUS2008
  12. 12. Evaluation Experimental Results <ul><li>Experiment </li></ul><ul><ul><li>2 H.264/AVC flows streamed to each client </li></ul></ul><ul><ul><li>Consisted of 5 intervals </li></ul></ul><ul><ul><li>Bit rates continuous leaf nodes enforced by H.264/AVC transcoder </li></ul></ul>23/06/2008 ADAMUS2008 Continuous leaf nodes Interval 1: Only 1 H.264/AVC flow; sufficient BW available to forward flow at maximal quality Interval 2: Introduction V2; V1 and V2 had identical weight and comparable max bit rate  received comparable BW budget Interval 3 + 4: Significance V1 increased  V1 is allocated more BW  V2 transcoded to lower bit rate Interval 5: Additional last mile BW available; used to upgrade quality V2 (V1 already at maximal quality)
  13. 13. Evaluation Discussion <ul><li>Findings </li></ul><ul><ul><li>Client’s last mile downstream capacity respected  Last mile congestion avoided </li></ul></ul><ul><ul><ul><li>Outcome = Optimal flow reception at client-side </li></ul></ul></ul><ul><ul><li>BW distribution captured stream importance </li></ul></ul><ul><ul><ul><li>Due to NIProxy’s application awareness </li></ul></ul></ul><ul><ul><li>H.264/AVC transcoding service enabled continuous video adaptation </li></ul></ul><ul><ul><ul><li>Optimal and full exploitation available last mile BW </li></ul></ul></ul><ul><li>Did not apply for the “unprotected” client! </li></ul><ul><ul><li>Degraded video playback at client-side </li></ul></ul><ul><ul><li>Clear difference in QoE provided to both clients! </li></ul></ul>23/06/2008 ADAMUS2008
  14. 14. Conclusions <ul><li>E2E QoE optimization platform </li></ul><ul><ul><li>Resilient overlay routing service circumvents erratic parts of network core </li></ul></ul><ul><ul><li>Last mile QoE optimization through bandwidth management and multimedia service provision </li></ul></ul><ul><li>Extended with H.264/AVC transcoding </li></ul><ul><ul><li>Enables continuous video adaptation </li></ul></ul><ul><li>Experimental results demonstrate positive impact on QoE optimization capabilities </li></ul><ul><ul><li>Full exploitation available last mile BW </li></ul></ul><ul><ul><li>More dynamic and effective BW distributions </li></ul></ul>23/06/2008 ADAMUS2008
  15. 15. Thank you for your attention! Any questions?

×