Mobile Ad hoc Networks (MANET) Minutes Meeting : IETF68 Wed March 21 1pm Location : Athens/Barcelona Chairs : Ian Chakeres ian.chakeres@gmail.com Joseph Macker joseph.macker@nrl.navy.mil Jabber : manet@jabber.ietf.org Audiocast: http://www.ietf.org/audio// URL : http://www.ietf.org/html.charters/manet-charter.html http://www.ianchak.com/manet http://tools.ietf.org/wg/manet/ Minutes : Jerome Haerri ========================================================= Administriva - Mailing list: manet@ietf.org - Presentations have been updated very recently. Please log on to download the latest version. - Scribe: Jerome Haerri (jerome.haerri@eurecom.fr) Justin Dean (justin.dean@nrl.navy.mil) on Jabber (??) - Blue Sheets: Please sign your name on the blue sheets - State your name at the microphone when asking questions - IPR - RFC3979 - Questions may be asked on Jabber. Justin Dean (??) is logged and will forward the questions to the group. ========================================================= Agenda bashing Speaker: Joe Macker Content: See slides ========================================================= WG Progress Speaker: Ian Chakeres Content: - Brief update on all documents. See slides. - Document relationship @ packetBB is closely used by all other documents @ IANADoc: Defines - a udp port - a link local multicast - a site local multicast @ SMF is an EXPERIMENTAL protocol @ Two new personal draft (Jitter, TimeTLV) need to be discussed in order to be included as WG documents - SHORT term goals @ LAST CALL targeted for PacketBB (standard track), IANA (standard track) and SMF (experimental track) @ NHDP is close to completion, but a last iteration is needed. Comments: (unkown): The relationship graph is very useful to understand the inter-linkage between the different documents. Where can it be found ? Ian Chakeres: It is on the Manet WG website - http://ianchak.com/manet ========================================================= MANET IANA Needs - draft-ietf-manet-iana-00.txt Speaker: Ian Chakeres Content: See slides @ Defines a UDP port that can be used by all protocols @ Defines a Link Local Multicast to transmit messages to neighbors more than one hop Comments: Bill Fenner: TBD3 and TBD5 should be aggregated. Thomas Clausen: The document Should be released as soon as possible ========================================================= Generalized Packet and Message Format Discussion - draft-ietf-manet-packetbb-04.txt Speaker: Thomas Clausen Content: See slides @ If anyone has anything to add to this document, please do it as soon as possible, as it is close to release. Comments: None. ========================================================= MANET Neighborhood Discovery Protocol Discussion - draft-ietf-manet-nhdp-03.txt Speaker: Chris Dearlove (CD) Content: See slides @ Since -00, - Dynamic local interface configuration - Link Quality and Hysteresis (OPTIONAL), but useful if used with 802.11 - Editorial Process: * Simplified Hellos @ Changes to be done: - address mapping issue. Example: 1) Mobile uses address A 2) Same mobile then uses address B 3) Same mobile then uses address A AND B The draft does not "address" this issue... But this should probably never happen. - Editorial process - Time TLVs and Jitters in external IDs - Normative reference to IANA document. @ Plan to issue a LAST CALL before the next meeting !! So please read the draft and send your comments Comments: (unknown): LC hysteresis: is it an option ? In the draft, it looks like it is mandatory. Thomas Clausen: It looks like it is mandatory, but changing the default value is not. Joe Macker: The default value for the LC hysteresis is also not clear. A static value should be defined. Chris Dearlove: It is defined in the reference in the last section of the draft. (same unknown): Yes, mandatory setting of the LC is useful, but for newcomers it is not clear how to use it. Chris Dearlove: We need to clarify this point. It is possible to leave it set, with the same effect of not using it. Joe Macker: For people implementing NHDP, if they can use LL information, they can use the LC. But it is possible to let the info come from other processes. ========================================================= OLSRv2 Discussion - draft-ietf-manet-olsrv2-03.txt Speaker: Thomas Clausen (TC) Content: @ Normative Dependencies to PacketBB and NHDP. @ A special case in OLSRv2: Local Attached Network (which could not be a MANET): OLSRv2 could attach a hop count or a distance measure. @ Fragmentation: partial update (incomplete TC) specified in this version of the draft @ Take the Jitters and Time TLVs out of the draft to external IDs. Comments: Joe Macker: Is "hop count" a container ? is it the only metric ? Thomas Clausen: We can add different cost metrics Joe Macker: Does the optimization aspect disallow non optimal cases ? Thomas Clausen If a node delegates a network advertisement to MPRs, the MPR MUST comply and do it. Ace: Is it a standard or an experimental draft ? Thomas Clausen: It is in the standard track. OLSRv2, DYMO are on the standard track, SMF is on the experimental track ========================================================= SMF Discussion - draft-ietf-manet-smf-04.txt Speaker: Joe Macker Content: @ Will only talk about changes between version -02 and -03. @ NHDP is doing some work of SMF. SMF is yet using the legacy neighborhood discovery. There is no need to use NHDP with SMF. @ Multiple gateways issues and dynamic extensions are purposely kept open for experimentation and future specifications. If someone gets confused, remember it is an on-going issue and SMF is in the EXPERIMENTAL track. @ Plan to add a TaggerID for IPv6-SMP-DPD headers. It helps to solve the multiple gateway problem. But it is OPTIONAL. @ CDS may need additional pruning to be efficient for SMF. Suggested E-CDS. @ S-MPR: Redundant for multicast. @ All CDS have been implemented and tested. A paper has been produced on the results. @ Need to clarify why we are doing SMF instead of group-based multicast. Need to discuss the different heuristics. @ Targeted a LAST CALL before next IETF meeting in Chicago. @ Future work on SMF if the exp RFC is mature. @ Possible BEST PRACTICE reference for the use of SMF and PIM @ Possible extension: "Filter Forwarding" for group multicast. Comments: Ian Chakeres: - TaggerID helps the multiple gateway problem, but it is only needed if the source has not tagged the packet. Thomas Clausen: - It is possible to run NHDP with SMF. But if not, the CDS selection is done based on any LL protocol that is able to provide neighborhood information. Joe Macker: - Indeed, if the LL protocol provides all information, we do not need to use NHDP. It is left open. The mechanisms are the same @ Duplicate packet detections @ Forwarding rules Both are specified. Yet, we might use a normative reference to NHDP for one mode, even if we have another mode without NHDP. ========================================================= - NRLSMF Update - http://pf.itd.nrl.navy.mil/smf/nrlsmf.html Speaker: Justin Dean Content: See Slides @ Forwarding protocols: - S-MPR: previous hop dependant - E-CDS, MPR-CDS: non previous hop dependant @ External control of NRLSMF for OLSR and OSPF packet forwarding Comments: Ian Chakeres: Skip slides with deep details on the SMF code implementation. Joe Macker: Those slides have been added because it makes people understand the basic framework for MANET protocol implementations. Joe Macker: QoS routing could also be done. It has been proven to be useful. NRLSMF is a USERSPACE prototype. One might want also have a KERNEL implementation. Also need to add a hook for ns-2, QualNet and OpNet simulations. Justin Dean: A basic hook exists for ns-2, but not fully functional. ========================================================= DYMO Discussion - draft-ietf-manet-dymo-08.txt Speaker: Ian Chakeres Content: See slides Comments: Thomas Clausen (TC): Changes X.TLV.Y to TLV.X.Y are hard to understand. What is exactly the notation ? Ian Chakeres: Will improve the notation and clarity ========================================================= Jitter - draft-clausen-manet-jitter-01.txt Speaker: Justin Dean Content: See slides @ Illustration of the dependences between documents, and how Jitter would impact. Comments: Joe Macker: Jitter will be a subclass of DYMO, and not a dependence. Thomas Clausen: Dash lines on the graph mean that the document MIGHT be useful. It is not a normative line, but instead a "useful" line. Joe Macker: Jitter may be useful for SMF and any timer dependant document. Joe Macker: On third slide, do not use "collision" but "contention", as it is specific to the MAC layer. Slides are not clear on what/which nodes are MPRs. Justin Dean: Jitter needed to avoid inconsistencies between diverse MPRs. Joe Macker: Is the MAC layer consistent with the TX range ? Justin Dean: No, not using 802.11. Chris Dearlove: Even for 802.11, we still have such problem Joe Macker: I want to take a poll to see who thinks this ID should become a WG document. It is not a requirement, but it would be interesting. Charles Perkins: It is too early to ask this, as almost no one has even read the document. Ian Chakeres: Actually, this technique has been implemented and tested on the OLSRv2 for a while. It works fine. Joe Macker: We could simply (not) agree on the idea, not the ground. Thomas Clausen: The text in the draft is as it is from the version in OLSRv2 and NHDP, and has been closely reviewed many times. Joe Macker: Then, we can simply remove the text from OLSRv2 and NHDP and add a normative reference in one place. If it is cited in three places in WG documents, the draft should also become a WG document. Thomas Clausen: Can we set a timeline for the document ? Chris Dearlove: As NHDP is close to LAST CALL, and in order not to have to reissue it, we need to be quick. Joe Macker: The idea is to remove the text (and the idea) from all document using it. Thus, it MUST be done before any LAST CALL planing of WG documents. This should be the week immediately after the 68th IETF meeting (questioning the use of this draft) Joe Macker: SMF and Jitters ? Brian thinks the equal dependencies between all references in the SMF draft should be discussed. SMF is indeed applicable to DiffServ applications. In that perspective, jitters are not useful. Thus, the dependancy between SMF and Jitters SHOULD be weakened. ========================================================= Time TLV - draft-clausen-manet-timetlv-00.txt Speaker: Thomas Clausen Content: See Slides @ We can either let each draft define how to transmit time or write a draft that define its once for all @ The dependence has therefore been extracted from OLSRv2 to an independent ID. Comment: Joe Macker: I agree on the approach. It is also interesting to show how to define TLVs for "time". But this should not be limited to source-based distance, but also extended to destination-based distance. Thomas Clausen: It is defined in OLSRv2. Ian Chakeres: It is important that people read the document in order to know if it should be passed as a WG document. Joe Macker: The strategy is to make it a WG document, as it is a high target. We indeed plan to remove it from other WG document implementing it before a LAST CALL. Thomas Clausen: We need to know if it is important, and put it in standard track before all other WG documents. Joe Macker: What is the estimate time for this process ? Thomas Clausen: Actually, we have what we need, as it has been extracted from previous deeply reviewed documents. It is already ready for LAST CALL. Joe Macker: OK, it MUST be declared LAST CALL before the LAST CALL of PacketBB, as it is dependent to TimeTLV. Chris Dearlove: It will be ready as soon as Joe Macker will be ready for the process. Thomas Clausen: We ask a commitment from the group to read it closely, review and comment immediately the document. ========================================================= Autoconf MANET Architecture - draft-ietf-autoconf-manetarch-01.txt Speaker: Thomas Clausen Content: see slides @ Spoke a long time during the last IETF meeting on the architecture. The authors in the autoconf WG are confident on the document. It is also important that the Manet WG feels confident on the document as it concerns us ! @ Please review it and send comments on the mailing list. Comment: Joe Macker: On Thursday, this document will be closely reviewed by the autoconf WG. Thomas Clausen: Therefore, the group should read the document closely and tonight and ask questions tomorrow. During the autoconf WG, there will not be a "tutorial" on the terminology as it had been the case last IETF in San Diego. ========================================================= Ian Chakeres: For the rest of this WG meeting, we do not have time to cover the rest of the slides. Speakers are coming forward and use only one slide. ========================================================= 6lowpan Routing Requirements - draft-dokaspar-6lowpan-routreq-00.txt Speaker: Dominik Kaspar Content: @ Requirement on the physical layer. We want to avoid fragmentation, as we need to reduce the energy consumption. @ Yet, manet protocols rely on fragmentation. Thus, we need feedback from the Manet WG on that matter. @ The 6lowpan WG is designing a Mesh routing protocol for 6LowPan. But they also want a feedback from the MANET WG. Comment: Unknown (6LowPan Chair (?)): The draft is NOT a WG document yet. But it covers the needs of 6LowPan. So, please read it !! ========================================================= Some simulation results on DYMO Speaker: Charles Perkins Content: @ Reduce the signaling overhead @ Path accumulation (P.A.): - more network information if protocols use P.A. instead of only RREQ/RREP @ But, packets are larger. We could loose in network performance. Comments: Ian Chakeres: A document on how to use Path Accumulation should be done. Charles Perkins: I will hopefully have more guidance in the future. Thomas Clausen: Is Path accumulation only related to DYMO, or is it also intended to other documents ? Charles Perkins: No, it is a problem specific to DYMO. ========================================================= MANEMO - MANET related MANEMO - http://www.mobileip.jp/MANEMO/ Speaker: Ryuji Wakikawa Content: see slides. - No time to present. Comment: None. ========================================================= Open Discussion, Related Work & Announcements - OSPF-MANET Speaker: None Content: No time for further discussion and announcements. Comment: none. ========================================================= Conclusion None. =========================================================