NWCRG Meeting Notes Notes mostly taken by Marie-Jose Monpetit Brian Adamson opens the meeting with the Agenda, and Note Well Agenda: start with reserach and experiences presentations, continue with discussion on Network Coding Architecture. 1. Dynamic Network Coding, policy-based NC, coded transport for many-to-many video streaming - Marie-Jose Monpetit, Vincent Roca, Jonathan Detchart http://www.ietf.org/internet-drafts/draft-montpetit-dynamic-nc-00.txt http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-0.pptx MJM: the goal is to have an open source protocol, able to change algorithms such as coding schemes depending on time and location. Morten Pedersen: does it have a framework? MJM: It's a code and a framework, an instantiation of a set of protocols. Aaron Falk: code points are already used by IANA. Brian Adamson: It can use code points from FECframe. Have FEC and encoding types. 2. IPR around Network Coding - Vincent Roca http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-3.pptx Victor F. - Q: patents the ones accessible today? V.R. yes there could be some more that are filed and not published yet Brian A.: DNC: exit ramp for the US market? MJM - yes Brian A.: patents are cost models for the developed work - IPR protection good for and at the same time stifling innovation - not need to review the information too often - but good to have the information - don’t be discouraged - the IPR issues are there and the presence of frameworks in nwcrg similar to what FECFRAME has done is good MJM comment: we have to be aware that there is more than just “research” and “commercial” but also people in the innovation economy who want to “try” NC Victor F. : good job, the list could be in the RG wiki and please continue upgrading 3. Update on Impact of Virtualization and SDN on Emerging Network Coding - Bhumip Khasnabish, Senthil Sivakumar, Evangelos Haleplidis, and Cedric Adjih http://datatracker.ietf.org/doc/draft-khasnabish-nwcrg-impact-of-vir-and-sdn/ http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-1.pptx Brian A: Publications/prototypes - B.K. nothing published and implementations still very early 4. Distributed Wireless Broadcast Protocols with Network Coding for Single/Multiple Sources - Cedric Adjih, Ichrak Amdouni, Oliver Hahm (Inria), Claudio Greco, Michel Kieffer (CentraleSupelec) http://www.ietf.org/archive/id/draft-amdouni-nwcrg-cisew-00.txt http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-2.pdf Brian A: are you going to highlight extra blocks needed in the architecture. C.A.: yes Frank F.: Proposed indexing is similar to what we did the broadcast work in the past? C.A.: yes Morten P.: I like the decoupling of the sliding window as this is also what we did. For the last part do you have a reference (the brute force)? C.A. The paper is to be presented to ICC in June. Brian A.: please review the draft 5. Combining TCP with coding in wireless network - Jinzhu Wang http://www.ietf.org/internet-drafts/draft-wang-tsvwg-tcp-coding-00.txt http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-4.pptx Morten P.: Do you have a systematic code? J.W. yes we have both systematic and non systematic; Morten P. how do you deal with short packets flows (a few packets)? J.W. I do not have an answer for that. Victor F. Coding is above TCP? J.W. Yes Brian A.: Are you thinking of a specific coding scheme? J.W. yes we are using a scheme where we ack early to avoid extra RTTs Victor F.: You are modifying TCP? J.W. Yes Morten P.: Simulated packet loss - very academic! J.W. we are also having real-world implementations. Morten P.: packet loss experienced? J.W. no information from the real world. David Black: There are severe congestion issues; there are other tsvwg discussions on this draft on the use of ECN in TCP which would erase the gains proposed; please have the curtesy tell the chairs when you propose a draft in 2-3 WG/RGs. J.W. Yes, will do, appologies, new to IETF, did not know. David Black: link error vs. congestion control is a hard problem Brian A.: there is work from RMT with ECN. That work could be presented here David Black: we encourage experimentation with ECN bits. 6. Latest Advancements and Use Cases for Network Coding - Morten Pedersen and Frank Fitzek - NC for software defined networks (SDN) - Reliable Synchronous Multicast with NC (and NORM) - RLNC for Storage and Cloud Solutions http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-7.pdf http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-9.pdf MJM: API In open source? M.P. yes Vincent Roca: yes for API and yes for a github repository to store it and make sure people can use it Brian A: good activity for us to support; full fledged use case Morten P.: would help having performance measures using the common API Victor F.: support this - how to connect this to the architecture before jumping to actual c code M.P. would be the FEC BB Morten P. Brian A.: RMT did not develop APIs - wish RMT would have been better at APIs MJM: APIs are a trend in TAPS 7. Frank Fitzek presents from his own computer; Use Cases for Network Coding slides to come Brian A.: synopsis on how you stack things? F.F. no - but you don’t want to to be swamped by feedback - goes back to 2008 Victor F.: where is the coding vis-a-vis the application - F.F. it is in the application (Morten P. confirms) Brian A.: opportunities for MANET - they are looking for charter items F.F. open the door to coding - not impose it Brian A.: motivation - home base for NC knowledge - and a forum - F.F. are we going Bhumip K.: 5G slides other categories? F.F. only those Bhumi K.: SND code done in any node open flow control protocol modifications F.F. at the moment it works but other developments could be good MJM: response to F.F. on how to do more dissemination - dissemination is good in IETF. F.F. it was a question Brian A.: work in MANET could profit more from the NC Victor F.: maybe a list Aaron F.: review other documents in the IETF in view of NC - there is a commercial vs. expert approach - expert approach is better as it gets the commercial elements out of the discussion Brian A.: identify the elements that could profit from NC - balance commercial vs. experts F.F.: it is easier to talk to new groups than well established one 8. Updates about our (patent free) on-the-fly network coding protocol draft - Jonathan Detchart https://tools.ietf.org/html/draft-detchart-nwcrg-tetrys-01 http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-5.pdf Bhumip K: what are the operations at each end J.D. this is coming Victor F.: changing the code dynamically is also what MJM presented J.D. it is related yes Victor F.: but your scheme would be compatible with DNC J.D. yes Morten P. changing the code means changing the coefficients? J.D. yes Network Coding Architecture, Building Blocks and Signaling - Group Discussion 9. Inputs to NC Architecture - Vincent Roca, Marie-José Montpetit, Jonathan Detchart http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-8.pptx Morten P.: I support finishing the taxonomy Victor F.: our group draft supersedes the old individual draft - I think we are close to be finished - people need to point out what is needed Brain A.: “downsizing” the set of names? Victor F.: I added all contributions to teh group draft - please send comments on the list Morten P.: abstract the symbol information to pass between the BB V.C yes Morten P.: interest in stage 4 - do the others in parallel - because a lot of good information will come out of it MJM: fully agree - the architecture will evolve - and there are transport vs. everything else - maybe we need more than one architecture Morten P.: architecture document may not be interesting Brain A.: audience is not developers it is people who do not know about NC 10. Network Coding Architecture and Building Blocks - Victor Firoiu, Brian Adamson, with inputs from Cedric Adjih, Jonathan Detchart, Marie-Jose Monpetit, Morten Pedersen, Vincent Roca http://www.ietf.org/proceedings/92/slides/slides-92-nwcrg-6.ppt Vincent Roca: I find this slide horrible - we should hide complexity instead of exposing - this is useless V.F. I also don’t like it maybe we need to make more than one MJM: this is complex because too generic - what are the use cases that could use all of this? Brian A: the block diagram is not that complex Victor F.: don’t concentrate on the text Brian A.: things identifed are example not the entire set? Victor F.: start to write things Brian A.: identify the building blocks that are Morten P.: I would prefer to identify the BB in the existing drafts instead of starting from a “vision” Victor F.: this is what I wanted to do MJM: we need to start from what is existing - not having this “vision” Brian A.: top down may not be useless - building blocks will be bottom up Bhumip K.: this will scare people off MJM: let the developers do the work - let them define what they need and are using Morten P.: the signaling will come from what is in from the draft Victor F.: still need to define common elements if not it will be a "smorgasborg". We'll continue to refine the draft and discuss on the mail list.