Mobile Ad hoc Networks (MANET) Minutes Meeting : IETF 87 Thursday August 1, 2013 Time : 1520-1650 Afternoon Session II Location : Tiergarten 1/2 Chairs : Joseph Macker Stan Ratliff Secretary : Ulrich Herberg Jabber : http://www.ietf.org/jabber/logs/manet/2013-08-01.html URLs : http://www.ietf.org/html.charters/manet-charter.html http://tools.ietf.org/wg/manet/ Minutes taker: Emmanuel Baccelli (with some additions by Ulrich Herberg) ========================================================= ---------------------------------------------------- Chairs ---------------------------------------------------- Updates about draft status (see slides) - Stan: Charter discussion planned at next IETF. Do we recharter? Please write proposals for new work to the mailing list. - Thomas Clausen (jabber): What's the hold-up for REPORT-MIB? - Ulrich: unclear, I will ask Robert Cole (the main editor of the draft). ---------------------------------------------------- Stan Ratliff talk about DLEP ---------------------------------------------------- - Stan: No new revision, still -04. We had a long discussion about DLEP and formed a short-tem (45-60 days) "research group", focused on getting a common set of meetings spanning across multiple radio types. - Chris Dearlove: Who is in the research group? - Stan: Henning Rogge, Teco Boot, Rick Taylor, John Dowdell, myself - Stan: Transport needs to be changed to TCP, unless we use UDP. - Ron: New to me that we can spawn a research group. Do you mean Design Team? - Stan: sure, I don't care - Thomas Clausen (jabber0: Is DLEP moving to the IRTF? - Stan: No - Stan: In the month of August, we will work on the metrics. DLEP-05 is planned for early/mid September. Push towards WG LC then. - Juliusz Chroboczek: Why do you switch to reliable transport? - Stan: Because that's what everybody else wanted to do. I would have preferred to leave it with UDP. The notion from the other people were that using a reliable transport keeps us out of using sequence numbers, stop-and-wait protocol. We still have to use some form of protocol ACKs / responses to requests. - Ulrich: Seems like a big change to me. Do you want to confirm consensus on the mailing list? - Stan: Will be confirmed on the mailing list. You have to start the discussion somewhere. - Ron: Will those not in the design team be able to follow on what is going on in the design team? - Stan: Right now it is a closed list. We are not trying to hide anything. We just want the work done. Nothing happens without consensus of the WG as a whole. - Adrian: Assuming this is a WG Design Team, send me an email with a short charter of the team (who is on it, why did you form it, what will you do, when will it deliver). Then I have enough information to grant the request. Then you request back to the WG according to the charter. There is benefits to gather in small teams from time to time. But there is also benefits from getting feedback from the WG. As long as there is free and open communication, nobody suffers. - Thomas (Jabber): what other issues were discussed during the 2-hour or 3-hour side meeting? TCP can't take that much time. IT would be very helpful if that "Research Group" would keep the WG in the loop and informed. - Stan: Discussion went from should we have used SNMP, or XML/XMPP, standards at IEEE/DSL forum, is there statistical reporting that needs to be done, went through the acronyms of Zigbee. Several topics that were considered and dropped. Output from the 2h discussion had to do with common set of metrics that could be applied to the core spec (3G, 4G, satellite, WiFi) + use of TCP. - Justin Dean (jabber): I would like to second Thomas' desire for greater transparency in document development. It need not stifle progress but greater information flow will be beneficial for all. - Stan: This is nothing more than a desire to get together the people who are maximally involved. - Rick Taylor: Our intention is to get some progress on DLEP. No intention to hide something or keep the WG out. ---------------------------------------------------- Charlie Perkins talk about AODVv2 ---------------------------------------------------- - Henning Rogge: Read the current draft. It's better, but still somewhat confusing to read. Jump between style/format for description between sections is confusing. Please choose a style and stick with it. There are a lot of acronyms. Some confusion between packet level processing, and IP level processing. Some part of the draft could also be outsourced to a PacketBB "best common practice" document. We still have to work on the presentation and structure of the draft, to jump less "back and forth" and streamline the document. - Charlie: I agree on segregating packet level processing, and IP level processing. Please send me your notes. The number of acronyms has decreased. - Stan: we can take acronyms case by case and maybe decrease more. - Juliusz Chroboczek: are you saying that if non-decreasing non-additive metrics are you used you will get loops? - Charlie: not necessarily. It is just more difficult to prove that you don't have loops. Please bring it up on the list! - Chris Dearlove: are there implementations of AODVv2? - Stan: there are none that I know of. We need to ask the list. In my WG participant opinion, we need implementations to standardize the protocol. - Charlie: We have extensive experience with AODVv1 RFC3561, which is essentially the same concepts at work. - Ulrich Herberg: I am disappointed. How long do we have to wait before this implementation pops up? - Jabber comment: some features of RFC3561 were proven harmful. - Charlie: we indeed identified some options that should be removed. - Ulrich: I believe we were promised a list of such options to be removed last IETF? - TIME'S UP for this slot ---------------------------------------------------- Henning Rogge talk on ETT ---------------------------------------------------- - Juliusz Chroboczek: I am pleased to see this draft. High time to see a non-trivial metric pop up in the IETF. However you need to rename with another name of ETT, because it is not strictly ETT, and it might be confusing Henning: would air-time metric be a better name? Julien: maybe, no firm opinion now. I would like to see evidence that this is working well, because this is not ETT, which is well studied. Furthermore there is no standard way to plug this in OLSRv2. We need to have that standard plugin. Chris Dearlove: yes, we could improvise that. Stan: The authors should revise the draft and come back to us in order to ask the working group if the document could be adopted as wg document. -------------------------------------------------------------------------------- Ulrich Herberg talk on MANET management -------------------------------------------------------------------------------- Ulrich: we are confused as to what the wg want with this draft. We would like some feedback on that, both from military side and civilian side. Stan: we lost energy on the author team. We need some help. Adrian Farrel (Routing AD): I'm willing to help, but I am clueless in MANET ;) But maybe there is another AD in the room that can help. Benoit Claise (OPs AD): I'm not pushing to have a draft from day 1 for every activity. But the time has come in MANET. How do you do fault management? How do you do config management? We need some answers. I'm ready to help. I don't know much about MANETs but I know the right questions to ask. Chris Dearlove: management in MANETs: As much as needed, little as possible and no single point of failure ;) Benoit Claise (OPs AD): In particular, don't waste time writing MIBs that won't be used. --------------------------------------------------------- Chris Dearlove talk on OLSRv2 cluster --------------------------------------------------------- Justin Dean (Jabber): Has this been studied in real systems. While more efficient depending on link dynamics it may make the network less robust. Have you looked at this? Chris: experiments show that is does not, it has been implemented and works. Stan: are you looking for WG acceptance for the documents? Chris: we have to ask the mailing list? Stan: how many people have read the document? (not some many people raise their hands). Let's take it to the list. --------------------------------------------------------- OPEN MIKE --------------------------------------------------------- Maaz Rehan: we have a new routing protocol for vehicular networks. We have compared to AODV. Will there be new items in the charter? Stan: I would like to see this work presented in a next MANET meeting. I'm interested in vehicular networks myself! Jiazi Yi (Jabber): I would like to announce the recent activities related to LOADng: On the loadng draft: the new revision -09 has been released. Compared to -08, two issues: one related to non-optimal routes (reported by Marius on the MANET mailing list), and one related to RERR message propagation (reported by G3-PLC Alliance) have been addressed. Two companion documents are also submitted: 1. Jitter consideration for reactive protocol in mobile ad hoc networks, to improve the efficiency of router discovery in reactive protocols. 2. Smart Route Request for LOADng: make use of existed routing information to reduce routing overhead in LOADng. Thomas Clausen (Jabber): could ETT draft be applied to other protocols too (not only for OLSRv2)? Henning: it is not OLSRv2 specific. It draws some values from it, but it is not tightly coupled to the protocol itself. We might need to adjust the range of values to the particular protocol that is targeted.