Minutes for the IETF manet WG meeting 12/11/00

Minutes taken by Vincent D. Park

1) Agenda Bashing and Announcements, J. Macker (NRL)

[Macker] Macker presented the candidate agenda for comment and bashing.  There will be one 2.5 hour meeting during this IETF. In addition to the agenda items there was a discussion of a few general WG issues upfront (not on agenda). Macker recommended that people experiment with available implementations considering the recent progress with a number of protocols and increased stability of documents. The group was reminded that there is "in place" a working group last call on AODV and DSR IDs with the idea of submitted them for experimental RFC consideration at the end of this comment period (Jan 15, 2001). Please read documents and give comments to authors and post issues to the mailing list. Another general items presented was the formation of proposed common design teams. Thomas Clausen is presently acting as a team leader within the proactive routing area for manet. Other possible team areas could include: addressing, broadcast, multicast, security, etc. [Perkins] noted a revision to broadcast support and existing doc on autoaddressing. [Templin/Ogier?] Will we consider both hard and soft state approaches in the proactive routing design team effort?  [Macker] Yes, all ideas will be considered but we need to move towards a common goal of a single document set to scope the work.  Along with on-demand routing, proactive routing is an important area for manet and all interested parties should get involved in the design team effort to input protocol ideas and implementation issues.

2) Updates of Current Drafts

a) DSR update, D. Johnson (Rice)

[Johnson] A review of DSR basic functionality was provided. There is a new version of draft that presents a base protocol design. Other features will be documented as other drafts. There a few minor protocol changes (see Dave's slides). There was also a general review of DSR simulation results and a review of DSR implementation and experiments. A commercial DSR implementation effort was described (see Dave's slides).

b) OLSR update, A. Qayyum/T. Clausen (INRIA)

[Clausen] The presentation highlighted a number of changes to OLSR draft.  Minor changes to the draft were described (see Thomas's slides). The future plan is to move advanced optional features of protocol to other drafts in the WG spirit of simplifying base specifications. Thomas also provided a review of the OLSR implementation status. There was a request from the OLSR implementers for increased outside experimentation and feedback.  The recent public release of their implementation should make this more feasible. [Johnson] A question was raised regarding the required OS for implementations. [Clausen] I was explained that most implementations presently work on Linux, one version also works/worked on Windows, but as a user space application porting should be relatively simple. There was a request for experimentation support to occur at the IETF meeting or future meetings (details to be announced). [Qayyum] There was also a presentation of OLSR results versus the DSR (framework) in mobile scenarios (see Amir's slides). [Macker] There was a question raised regarding the MAC model assumed. [Qayyum] The CSMA MAC model was used. [Johnson] There was a question raised regarding use of link layer notification over CSMA.  [Qayyum] The response was that the simulation method used was simply to show where the majority of loss occurs. [Johnson] There was another question and comment regarding the traffic model assumptions.  [Qayyum/Jacquet] Clarification was provided regarding the traffic models used for the results.

c) ZRP Update, M. Pearlman (Cornell)

[Pearlman] Marc provided an overview of a separated ZRP draft into multiple documents. There was a short review of a ZRP framework followed by a presentation of IARP conversion guidelines (see Mark's slides). There was then a presentation of IERP conversion guidelines (see Mark's slides).  There are now simulation models available. [Clausen] Where are drafts?  [Pearlman] We missed the update deadline, but they will be submitted, for now see the authors' website.

d) AODV IPv6, C. Perkins (Nokia Wireless)

[Macker] The following short presentation is not on the agenda but we are sliding it in now. [Perkins] Described a few minor changes to the AODV design to accomodate IPv6 (see Charlie's slides) [Johnson] Provided a comment regarding uniqueness of broadcast packets (agreement from WG).  [Perkins] A simple change can be added to fix that. [Johnson] Raised a question regarding time nodes must recall packets. [Oran] Provided a comment regarding prior relevant techniques such as rate limiting broadcast, and releasing of state when subsequent packets received from same source. [Jacquet] Raised another question regarding problem with unidirectional links. [Perkins] That is not covered in this presentation, but can be fixed by using ACK. There was a request to move further discussion to the mailing list. [Jacquet] Raised the question that this issue also exists for DSR. [Jacquet/Johnson/Perkins/etc] Lots of discussion ensued, but there was no resolution at the time.

3) New Draft Discussions

a) Landmark Routing Protocol (LANMAR), Gary Pei (UCLA)

b) Fisheye State Routing Protocol (FSR), Gary Pei (UCLA)

[Pei] A presentation of new IDs covering LANMAR and FSR was provided (see Gary's slides). [Johnson] Raised a question regarding of optimality of routes. [Pei] There was agreement from the presenter that optimality was not always achieved. [Ogier] Pointed out of what he felt was a potential looping problem. [Xiaoyan] The response was that this was not problem for link-state. There was disagreement from WG, and there was apparent agreement that temporary loops can exist. [Pei] Next, recent simulation results were described (see Gary's slides). [Johnson] Another comment on traffic models was raised. [Pei] While limited, the experiments were primarily designed to study scalability. [Johnson] In the authors' opinion which protocol presented is preferred? [Pei] Both protocols have advantages. [Ogier] We are planning to extend TBRPF to be hierarchical. Can your approach be done with hardstate? [Pei] I would think it should be ok.  [Johnson] Can any protocol truly be hardstate? Some philosophical responses from WG ensued. [Macker] Raised a question regarding the assumption of groups moving together for good performance. Is it expected to work well if this is not the case? [Pei] We are investigating that.

c) Multicast based on Zone Routing (MZR), Vijay Devarapalli (Nokia)

[Deva] Provided an overview of this protocol (see Vijay's slides).  Simulation results were described (see Vijay's slides). [Pearlman] What did the results show as the best zone radius? [Deva] Zone size of 2 found to be best.

4) Related Activity Announcements

[L. Miller (NIST)] Covered some DSR OPNET simulation models that are available. Indicated they are working on an AODV model as well (few months away).

[Macker] Requested status updates on implementations and simulation models.

[Telcordia?] There was an announcement of some information regarding a multicast routing draft (HLRM) that may be of interest to the WG.

[L. Feeney] Announced some work that they are doing to support on-demand routing.

[D.Johnson] Provided a MobiCom announcement.

[C. Perkins] Provided a MobiHoc announcement.

5) Open Discussion

[WG] There was an open discussion regarding group experiment with OLSR (at this IETF). There were issues raised regarding the use of WLAN cards within range of another operating network.  Recommendation to be careful with channel selection,etc for ad hoc experiments.

[Macker] Reminded the group of the need for WG input on last call documents.

[Perkins] Raised a comment regarding current proposed approaches for address selection.

[Macker] Proposed further detailed WG discussion.

[Ogier] Raised a question regarding commonality between protocols. Are simulations appropriate?

[Macker] Focus should be on commonalities of proactive approaches, but alternate ideas should be weighed in.  Suggested that interested parties get involved early in the proactive routing effort to contribute.