Institute of Computer Science
Chair of Communication Networks
Prof. Dr.-Ing. P. Tran-Gia

Socially-Aware Traffic
Managemen...
Agenda
1.

Traffic Management in the Internet

2.

New Internet Applications

3.

Social Awareness

4.

Socially-Aware Tra...
Traffic Management in the Internet





Broadband deployment makes Internet more efficient
But traffic volumes on the I...
New Internet Applications


Trend: Applications are moved into the “cloud”












Video/music on demand
Li...
Yesterday: Internet Structure and VoD Traffic
Tier 1 ISP

Tier 1 ISP

Tier 1 ISP

Tier 2
ISP

Tier 2
ISP

Content
Provider...
Yesterday: Internet Structure and VoD Traffic
Traffic from content
provider produces
transit costs and may
have high delay...
Changes: Internet Structure and VoD Traffic
Tier 1 ISP

Tier 1 ISP

Tier 1 ISP

Global Video CDN

Tier 2
ISP

Tier 2
ISP

...
Changes: Internet Structure and VoD Traffic
Tier 1 ISP

Tier 1 ISP

Content providers
expand to global CDN
to save transit...
Changes: Internet Structure and VoD Traffic
Tier 1 ISP

Tier 1 ISP

Tier 1 ISP

Global Video CDN

Tier 2
ISP

Tier 2
ISP

...
Changes: Internet Structure and VoD Traffic
Tier 1 ISP

Tier 1 ISP

Tier 1 ISP

Global Video CDN

Tier 2
ISP

Tier 2
ISP

...
Social Awareness


Social signals are ubiquitous
 Online Social Networks
– Friendships
– Interest
– Trust-relevant metad...
demand of item i λi(t)

Online Social
Network

λi(t)

Outlook: Towards Social Awareness


Use OSN
information to
predict ...
Socially-Aware Traffic Management


N. Sastry, E. Yoneki, J. Crowcroft, “Buzztraq: Predicting
geographical access pattern...
Socially-Aware Traffic Management


N. Sastry, E. Yoneki, J. Crowcroft, “Buzztraq: Predicting
geographical access pattern...
Socially-Aware Traffic Management


N. Sastry, E. Yoneki, J. Crowcroft, “Buzztraq: Predicting
geographical access pattern...
Socially-Aware Traffic Management


N. Sastry, E. Yoneki, J. Crowcroft, “Buzztraq: Predicting
geographical access pattern...
Outlook: Exploiting Social Awareness
Tier 1 ISP

Tier 1 ISP

Tier 1 ISP

Global Video CDN

Tier 2
ISP

Tier 2
ISP

Tier 2
...
Outlook: Exploiting Social Awareness
Tier 1 ISP

Tier 1 ISP

Tier 1 ISP

Global Video CDN

Tier 2
ISP

Tier 2
ISP

Tier 2
...
Outlook: Exploiting Social Awareness
Tier 1 ISP

Tier 1 ISP

Video can be
Tier 1 ISPtransmitted to cache
during night (off...
Outlook: Exploiting Social Awareness
Tier 1 ISP

Tier 1 ISP

Video can be
Tier 1 ISPtransmitted to cache
during night (off...
SmartenIT


Socially-aware Management of New
Overlay Application Traffic combined
with Energy Efficiency in the Internet
...
Discussion


Which services can benefit from socially-aware traffic
management?
 Traffic and requirement prediction
 Tr...
23

Michael Seufert
Upcoming SlideShare
Loading in …5
×

Socially-aware Traffic Management (Workshop Sozioinformatik)

834 views

Published on

Talk on "Socially-aware Traffic Management" given by Michael Seufert (http://www3.informatik.uni-wuerzburg.de/staff/michael.seufert/) at the workshop Sozioinformatik 2013 (http://www.sozioinformatik2013.de/, organized by Katharina Anna Zweig; held in conjunction with Jahrestagung der Gesellschaft für Informatik (INFORMATIK 2013)). The workshop addressed questions evolving around the interplay between social and technical systems, and bridged the gap from social sciences to computer sciences. The workshop talks gave an overview on different aspects of interactions between humans and IT-systems, and highlighted the need for a combination of social sciences and computer science in this field. The workshop showed that it is possible and sometimes necessary to integrate social studies into the design and application of IT-systems. This applies to SmartenIT especially in the context of socially-aware traffic management.

Michael Seufert, George Darzanos, Valentin Burger, Ioanna Papafili, Tobias Hoßfeld
Socially-Aware Traffic Management.
Workshop Sozioinformatik 2013, Koblenz, Germany, September 2013.

Abstract:
Socially-aware traffic management exploits social signals to optimize traffic management in the Internet in terms of traffic load, energy consumption, or end-user satisfaction. Several use cases can benefit from socially-aware traffic management and the performance of overlay applications can be enhanced. In the talk we show interdisciplinary efforts between communication networks and social network analysis. Specifically, we give an overview on existing use cases and solutions, but also raise discussions at the workshop on additional benefits from the integration of social information into traffic management.

Published in: Technology, Business
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total views
834
On SlideShare
0
From Embeds
0
Number of Embeds
78
Actions
Shares
0
Downloads
12
Comments
0
Likes
0
Embeds 0
No embeds

No notes for slide

Socially-aware Traffic Management (Workshop Sozioinformatik)

  1. 1. Institute of Computer Science Chair of Communication Networks Prof. Dr.-Ing. P. Tran-Gia Socially-Aware Traffic Management Michael Seufert, George Darzanos, Valentin Burger, Ioanna Papafili, Tobias Hoßfeld www3.informatik.uni-wuerzburg.de
  2. 2. Agenda 1. Traffic Management in the Internet 2. New Internet Applications 3. Social Awareness 4. Socially-Aware Traffic Management 5. EU FP7 Project SmartenIT 2 Michael Seufert 2
  3. 3. Traffic Management in the Internet    Broadband deployment makes Internet more efficient But traffic volumes on the Internet increase due to new applications or new service levels Traffic on the Internet has to be managed by Internet Service Providers (ISPs) in order to  Avoid congestion  Deliver services in an acceptable quality  Save resources, energy, and costs  Methods for traffic management      Prioritization Routing Bandwidth shaping Caching Offloading 3 Michael Seufert 3
  4. 4. New Internet Applications  Trend: Applications are moved into the “cloud”           Video/music on demand Live video/HD TV/music streaming Video conferencing/VoIP Remote desktop Collaboration File storage/sharing Instant messaging Gaming Online social networks Changes in Internet structure and content delivery 4 Michael Seufert 4
  5. 5. Yesterday: Internet Structure and VoD Traffic Tier 1 ISP Tier 1 ISP Tier 1 ISP Tier 2 ISP Tier 2 ISP Content Provider Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 3 Tier 3 Tier 3 5 Michael Seufert 5
  6. 6. Yesterday: Internet Structure and VoD Traffic Traffic from content provider produces transit costs and may have high delay Tier 1 ISP Tier 1 ISP Tier 1 ISP Tier 2 ISP Tier 2 ISP Content Provider Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 3 Tier 3 Tier 3 6 Michael Seufert 6
  7. 7. Changes: Internet Structure and VoD Traffic Tier 1 ISP Tier 1 ISP Tier 1 ISP Global Video CDN Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 3 Tier 3 Tier 3 7 Michael Seufert 7
  8. 8. Changes: Internet Structure and VoD Traffic Tier 1 ISP Tier 1 ISP Content providers expand to global CDN to save transit costs and get content close to users Tier 1 ISP Global Video CDN Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 3 Tier 3 Tier 3 8 Michael Seufert 8
  9. 9. Changes: Internet Structure and VoD Traffic Tier 1 ISP Tier 1 ISP Tier 1 ISP Global Video CDN Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Global Video CDN Cache Tier 3 Tier 3 Tier 3 9 Michael Seufert 9
  10. 10. Changes: Internet Structure and VoD Traffic Tier 1 ISP Tier 1 ISP Tier 1 ISP Global Video CDN Tier 2 ISP Tier 2 ISP Tier 2 ISP Global Video CDN Cache Tier 3 Tier 3 Tier 2 Tier 2 Content providers offer ISP ISP local caching services to bring content closer to users and to save transit costs Tier 3 10 Michael Seufert 10
  11. 11. Social Awareness  Social signals are ubiquitous  Online Social Networks – Friendships – Interest – Trust-relevant metadata  Applications – Call/messaging patterns – Interaction patterns – Content popularity  Sensors – Location  Social awareness  Harvest social signals  Extract and exploit information   Goal: Improvement of a service 11 Michael Seufert 11
  12. 12. demand of item i λi(t) Online Social Network λi(t) Outlook: Towards Social Awareness  Use OSN information to predict demand of video i λi(t)  Pre-fetch trending videos  Improve local cache hit ratios  Use unutilized bandwidth in offpeak periods t User File-Storage Physical Network VoD AccessNetwork Content Distribution Network InternetExchange Point local cache capacitiy Γ(t) Access-Network 12 Michael Seufert 12
  13. 13. Socially-Aware Traffic Management  N. Sastry, E. Yoneki, J. Crowcroft, “Buzztraq: Predicting geographical access patterns of social cascades using social networks“, Proceedings of the Second ACM EuroSys Workshop on Social Network Systems, 2009 13 Michael Seufert 13
  14. 14. Socially-Aware Traffic Management  N. Sastry, E. Yoneki, J. Crowcroft, “Buzztraq: Predicting geographical access patterns of social cascades using social networks“, Proceedings of the Second ACM EuroSys Workshop on Social Network Systems, 2009 14 Michael Seufert 14
  15. 15. Socially-Aware Traffic Management  N. Sastry, E. Yoneki, J. Crowcroft, “Buzztraq: Predicting geographical access patterns of social cascades using social networks“, Proceedings of the Second ACM EuroSys Workshop on Social Network Systems, 2009 15 Michael Seufert 15
  16. 16. Socially-Aware Traffic Management  N. Sastry, E. Yoneki, J. Crowcroft, “Buzztraq: Predicting geographical access patterns of social cascades using social networks“, Proceedings of the Second ACM EuroSys Workshop on Social Network Systems, 2009  Prediction of future access to user generated content (e.g. videos)  Generation of hints for replica placement and/or cache replacement  Usually replica placement is based on access history (locations of previous users)  Add locations of potential future users (i.e., OSN friends of previous users) to account for social cascades  Social cascade prediction improves history-based placement if content mainly spreads virally  Useful for content which is not (yet) globally popular, otherwise CDN 16 Michael Seufert 16
  17. 17. Outlook: Exploiting Social Awareness Tier 1 ISP Tier 1 ISP Tier 1 ISP Global Video CDN Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Global Video CDN Cache Tier 3 Tier 3 Tier 3 17 Michael Seufert 17
  18. 18. Outlook: Exploiting Social Awareness Tier 1 ISP Tier 1 ISP Tier 1 ISP Global Video CDN Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Global Video CDN Cache Tier 3 User posts a video which will be watched by his friend the next day Tier 3 Tier 3 18 Michael Seufert 18
  19. 19. Outlook: Exploiting Social Awareness Tier 1 ISP Tier 1 ISP Video can be Tier 1 ISPtransmitted to cache during night (offpeak hours) Global Video CDN Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP Global Video CDN Cache Tier 3 User posts a video which will be watched by his friend the next day Tier 3 Tier 3 19 Michael Seufert 19
  20. 20. Outlook: Exploiting Social Awareness Tier 1 ISP Tier 1 ISP Video can be Tier 1 ISPtransmitted to cache during night (offpeak hours) Global Video CDN Tier 2 ISP Tier 2 ISP Tier 2 ISP Tier 2 ISP When user wants to watch the video the next day, it is already in the cache Global Video CDN Cache Tier 3 User posts a video which will be watched by his friend the next day Tier 3 Tier 2 ISP Tier 3 20 Michael Seufert 20
  21. 21. SmartenIT  Socially-aware Management of New Overlay Application Traffic combined with Energy Efficiency in the Internet  EC FP7 STREP No. 317846, Duration: Nov 2012 – Oct 2015 10 European partners (academia, network and cloud operators) Incentive-compatible cross-layer traffic management for    Cloud service providers  ISPs  End-users  Exploitation of social awareness  User relations  Interests  Focus on energy efficiency 21 Michael Seufert 21
  22. 22. Discussion  Which services can benefit from socially-aware traffic management?  Traffic and requirement prediction  Trusted communication  …  Which social information is needed?  Detail level  Time scale  …  How to collect this social information?       Social signals from various platforms Collaboration with OSN Crawling of data Incentives for users to share their data … What about users„ privacy? 22 Michael Seufert 22
  23. 23. 23 Michael Seufert

×