P2PRG Session

Agenda bash

Summary of progress of drafts. Alto survey passed IESG review, japan and mythbustering are ready for RG review, taxonomy is under review.

4 new drafts have been submitted

Epi transport: Presenter Ricardo Bernardini
Presentation of slides:
* Started as a streaming multimedia approach, to minimize bandwidth. Became multicast for limited upload bandwidth. 
* Current status Ð stable core, open source, code in Ada
* Main approach: reduction procedure Ð reduce packets by a factor of R, send out, recombine and retransmit
* Example of the reduction function
* Data format presentation
* Network structure : push format
* Pseudo address approach
* Integration with existing protocols
* Conclusion. Not yet frozen, looking for community input
Questions:
* Richard Alimi: Can you compare with other network coding approaches such as random transformations? A: Are working on alternative reduction procedures, want to have the protocol finished. This approach may not be the best for reduction, just easy.
* David Dyson: Can you say a bit more about how this is multicast like? Looks like multiple unicast. A: Can request from server to multicast to some number of peers, and relay to additional peersÉ
* Volker: out of time, lets take discussion to list
 
Hirochika Asaim AS Relationships-aware Overlay Routing

Slides:
* Overview of AS structure in the Internet
* Motivation: Reduction in transit traffic/costs
* Approach: Be aware of relationship between AS. Like ALTO, but mainly on cross domain traffic without revealing relationships
* Detail of valley-free path model
* Presentation of requirements
* AS Relationship inference
* Slide on example of cost function
* Solution results
o Traditional/random approach is not ideal Ð show that new techniques reduce cost
o Question Martin Steimerling: Have you considered what happens if you get the AS information wrong? A: We are looking at improving based on AS. Q: Yes, but you need to look at what if you get it wrong Ð you are guessingÉ A: will consider
* Conclusions and Future work	
* Questions/comments
o Stefano: Looks like an interesting approach. Limit when you have inconsistent policies between service providers. This is challenging, and would be great if proven to work, but you will be rapidly faced with case where multiple providers are trying to solve the problem. May need to know the policy rather than infer then

Ove Strandberg : Information centric networking and P2P

Slides
* Motivation: currently host centric, looking to moving toward information centric
* Outline
* Secure naming characteristics : Self-certified ID, name persistence
* Self-certification
* Naming Scheme overview
* Metadata
* Evaluation
* NetInf Prototype
* Summary/Conclusion
* Questions
o Yunfei: Regarding the firefox plugin: Mapping between ID and URL? A: No, can use either, independent. Yunfei: Can you use both at same time? A: (co-author stands at mic) Map ID to URL, certifies content Yunfei: Can you then combine P2P and HTTP? If connection is good, use HTTP, but if it degrades, may reduce the transmission to be small transportationÉ
o Chair: Out if time Ð need to take this offline

Richard Alimi: RISE Ð robust streaming evaluation

Slides
* How to evaluate live streaming in a live situation?
* Motivation: critical algorithms in live streaming, developers continue to improve
* Limitations to current techniques: lab testing, model/sim, deploy test channel
* RISE objective: Test live system with real users
o Gather live data, ensure users have food quality system
* Key Techniques of RISE
* Scalable Protection: Run both in parallel
* Question: Volker: what if the different algorithms place pieces in different places? A: will be covered soon
* Virtual Player Platform
o Try running as if just experimental, if lose chunks, fall back to stable
* Virtual Player Analysis
o Some overhead lag
* Conclusion
* Questions to the mailing list