1. Adaptive QoS for Service- Oriented Learning Environments Colin Allison, Martin Bateman, Ross Nicoll and Alan Ruddle School of Computer Science University of St Andrews mb@dcs.st-and.ac.uk

2. Overview Identify goals of Collaborative Learning Environments Educational goals -> QoS requirements How to meet QoS requirements Example: Finesse – Learning environment Conclusions

3. Pedagogical Goals Collaboration –Synchronous video audio –Asynchronous notebooks instant messenger Ubiquitous & heterogeneous –Network –Device Experiential and Active Learning Realism – ‘Live data’

4. Finesse Finance Education in a Scalable Environment Supports teaching of fund management Virtual portfolios at the core –Inspect historic data –Buy/sell shares –Try to make a profit Notebook for asynchronous collaboration

5. Finesse

6. Pedagogical aspects of Finesse Asynchronous Ubiquitous – 100% web Realism – London Stock Exchange Experiential

7. Finesse – To add Re-engineer to be GSDL based Synchronous communications –Video conferencing –Synchronous groupware Device independence

8. QoS Aspects User perception –Responsive –Timely –Looks good –Easy to use Network –Low delay –Low jitter –High bandwidth –Low packet loss

9. QoS Timeliness & Responsiveness QoS Require ment Interactive Resource Type Example Application BandwidthSample Rate DelayJitterLoss Tolerant Timelin- ess Continuou s Media Interactive video high15 fps250 msYes Interactive audio Low - medium 8000 hz250 msYesyes Respon- siveness Web- Server Based CGI/ Servlet Low-highN/A5 sNo

10. What does the Grid bring Common infrastructure –OGSA, WSDL, UDDI, etc Component sharing Dynamic Service Composition QoS based service discovery Registry Service Requester Service Provider 1 2 3

11. FiGS – Finesse Grid Services Browser Video User Web Servlets GS:ManagerGS:Manager GS: Notebook GS: Conferencing GS: Portfolio GS: Stock Data Source 1 GS: Stock Data Source 2 Finesse Services Grid Services Application VoiceML Video Phone

12. Internet QoS Approaches Resource Reservation –Eg IntServ Aggregate flows –Eg DiffServ Adaptive Best effort –Eg most applications Increasing Infrastructure

13. Adaptive QoS Provision Past network conditions -> statistics –Active: RTP/RTCP traffic –Passive: Traffic monitoring Estimate likely network path conditions –Temporal & spatial patterns in traffic Inform application at start Application adapts to changes

14. Location Information Server

15. Conference Controller Architecture for Video Conferencing

16. Adaptive QoS Advantages Network edge deployment –Under your administrative control –Manageable traffic rates Learns from traffic –Sharing traffic knowledge Utilises session knowledge –Duration, number of participants, type

17. Example Conference Configuration

18. Conclusions Collaborative Learning Environment –Achieve pedagogical goals Pedagogical goals -> QoS requirements Architecture for Adaptive QoS –Deployable now

19. Questions?

20. Traffic Patterns ResearchHome