Minutes interim-2020-lpwan-01: Wed 16:00
minutes-interim-2020-lpwan-01-202001081600-02

Connection details
------------------
• Date: 7-8am US Pacific, 4pm CET:
https://www.worldtimebuddy.com/?qm=1&lid=100,12,5392171,1850147&h=100&date=2020-01-08&sln=15-16
Meeting link:
https://cisco.webex.com/cisco/j.php?MTID=me91d56b37454056b8c5ef3b102b9da50
Meeting number: 201 266 501 Password: txCGJTrS (89245877 from phones)

Agenda
------

[16:05] Administrivia                [ 5min]
    o    Note-Well, Scribes, Agenda Bashing
    o    Status of drafts

[16:10] Last updates of SCHC IP/UDP (Dominique)    [15min]
[16:25] SCHC YANG Data Model (Laurent)             [25min]
[16:50] AOB                                        [10min]

Minutes takers
-------------------
Ana Minaburo
Arunprabhu Kandasamy

Attendees
-------------

 - Pascal Thubert
 - Alexander Pelov
 - Ana Minaburo
 - Laurent Toutain
 - Ivaylo Petrov
 - Dominique Barthel
 - Olivier Gimenez
 - Diego Dujovne
 - Vincent Audebert
 - Arunprabhu Kandasamy
 - Juan Carlos Zuniga

Missing Past Attendees
-------------
 - Carles Gomez
 - Julien Catalano

Meetig minutes
------

[16:05] Administrivia                [ 5min]
    o    Note-Well, Scribes, Agenda Bashing
    o    Status of drafts

* Send the power point to chairs instead of PDF, Webex does not like pdf.

* Olivier will be added at the end, it will be added next time because
LoraAlliance meeting is today same time

* But the Lora Alliance is at the same time so we need to change the hour the
2nd wednesday, for next meeting we need to watch up this time

* Pascal give the Note Well

* Alex: CoAP is now in AD Evaluation, with Farell mail

* Ana: We have answer to Farell about his inputs

* Pascal: I think the draft is in the queue

* Alex: We have a new draft for PPP, do you think it will grow or it will be
kept like that?

* Pascal: I think it is done, the one think is to negotiate a compression
mechanism, and perhaps it give us SCHC over PPoE. SCHC over ethernet, PPP, wifi
can be addressed with this one single draft.

* Alex: Does it fit in the milestones?

* Pascal: Perhaps it could be because it is important to give this soution for
serial links, we can present in IntArea, what do you think Juan Carlos?

* JuanCarlos: INTArea open to new work , people will be more familiar about
lpwan

* Pascal: I can present it next time, actually we need to indicate where the
data model and the context is in the compression, probably through an URL.

* Alex: Yes, I think it is important to be presented next time

* Pascal: I send the meeting request for next IETF, if you have any conflict
please send to me an input

* Pascal: Next meeting deadlines, please take a look, for the early
registration and cutoff on early march. We will post a slot request, please
send an e-mail on march 11th to make the agenda

[16:19] Last updates of SCHC IP/UDP (Dominique)    [15min]

* Carsten provided extensive review of everything again.  Responded to all
points.

* Updated example in Appendix A.

* version 24 has editiorial improvements.

* Carsten commented we cannot decide if the profile allows multiple rule
matching or not.

* All0 and ACK Req must be distinguisable, clarified lifeitime of dtag in
receiver.

* -24 approved by Suresh.

* we don't have Misref? referencing a document that is still in draft.

* Pascal to make the transformation and send to Dominique , that will be a new
version.

* questions from authors for schc-over-foo drafts and implementers.

[16:33] SCHC YANG Data Model (Laurent)             [20min]

* Look at github directory to check the last version, we are working on. The
synchronisation will be finished after the meeting

* Pascal: Send a mail when it is done to the ML

* Laurent: I'm not a yang doctor, so I'll try to give a solution with the
things we understood.

* In the draft there is a definition of SCHC fields and MO and CDAs in yang,

* Level of granurality has been intriduced for some fields as traffic class or
code in coap.

* When the fields are define the question is until which details we go

* CoAP is in the same space as udp, ipv6. Carsten: reserve all the options
space as identifiers.

* Carsten propose to reserve the whole space but it could be a waste of numbers,

* openschc implementation uses 0xff as end option. Conflict arises if the core
uses this value for another option.

* In the model version is added, does it is a good idea? does version will be
use to identify the version we are using? discussion is needed

* Fragmentation has not been completely done, we need to define all the
fragmentation entries and we need input from the implementations

* Yang is define to give an ASCii value but for LPWAN is better to get a
numerical value, but yang is not done for this

* For target valule the next step is to give the highest and the lowest values

* Pascal: We need to talk with Benoit or people that talks yang if they have an
input

* Laurent: There are things that have not been well done,

* Pascal: yes, like IPv6 @

* Laurent: We have open questions about profiles, but we need to discuss and
see how to do it

* Limitation with several arguments, the question is who cares? do we need to
find a solution?

* Laurent: I will send a mail and ask some inputs to see the needs and to go
forward

* Pascal: We need to talk with a yang doctor,

* Alex: In Core this question was raised, so just use the ASCii representation
and then they change their mind, but I cannot remember to what

* Alex: suggest to use a string representing a Base64 encoding

* Laurent: possible but not as efficient as binary

* Pascal: Base64 increases computation load.

* Laurent: Numerical1 and Numerical2 to build large integer (typically IPv6
address).

* Alex: let's also investigate Bitarray

 [16:25]  LoraWan IID (Olivier)     [10min]
Alex: The time is gone, we can talk about IID next time
Olivier: or by mail
Pascal: Let's do it now
Olivier: We are using the half of the hash

[16:50] AOB                                        [5 min]

Discussion happens, currently-proposed IID computation looks good, need
Alliance validation that it does not leak too much information. Julien Catalano
proposed another mechanism that involves installing some context, therefore
needing a bit more control overhead. Olivier ot post on the ML in this and
follow up with Alliance