Hi Bastiaan,

Thank you for compiling the minutes, they are valuable reading for those 
who could not attend, including myself.

In the message you are asking for feedback.  Here is some feedback.

Please comment.

Le 27/08/2014 13:05, Wissingh, B.F. (Bastiaan) a écrit :
[...]

> Way forward basically two options:
> There are two options on moving forward with the Geonet activity:
>
> == Option 1: Try to create an IETF BOF/WG,

I agree with this option 1.

>
> by first modifying the
> following: ==
> 1) The scope is too large:
> Adjust the use-cases, align them more and make sure they all ³require² the
> same solution, or have the same problem statement. E.g, the precise goods
> tracking use case seems very different from the other use cases.
> Narrow further the scope

I agree about adjusting the use-cases.  Currently  there are 4 use-cases 
documented in draft-karagiannis-geonet-problem-statement-00:

1. Dissemination of IP packets to geographical areas
2. Precise tracking of package positions during a shipping process
3. Dissemination of ITS (Intelligent Transportation System) information
    to fixed and mobile RSUs via Internet
4. Tracking and communicating with people or objects located in certain
    geographical areas

There are some immediately clear similarities but also discrepancies. 
For example, (1), (3) and (4) may use the same mechanism, although the 
mobile RSU part of (3) may be a bit different because the RSU is mobile.

The (2) use-case - precise tracking - seems a bit more different from 
the others, although the use-case may read similar.  The difference lies 
in the fact that it is the IP address of the end-node that is associated 
with the coordinates of the assumed-to-be-RSU, whereas in use-cases (1), 
(3) and (4) it is the IP address of the RSU itself which is associated 
with the coordinates.

Given these aspects, I also think we need to refine the use-cases such 
as to stress the similarities and reduce the discrepancies.

> 2) show in detail how Geonet can be applied to one or more use cases,
> using message sequence diagrams, etc.

I agree.  A first message sequence diagram would be useful.  Could you 
please draw a message exchange geonet/its?  It should be something like 
this:

                                         RSU            Vehicle's MR
    Client         Server            at lat/long/alt    close to RSU
      |               |                   |                |
      |               |                   |                |
      |-------------->|                   |                |
      | IP address of |                   |                |
      | RSU at lat/lon|                   |                |
      |  /alt/prec?   |                   |                |
      |<--------------|                   |                |
      |    RSU-IP     |                   |                |
      |               |                   |                |
      |---------------------------------->|                |
      |      Message-to-RSU-IP            |        /       |
      |               |                   |-------/
      |               |                   |       \"broadcast"all in area
      |               |                   |        \


> 3) dissemination of packets within a geographical area should be layer 2
> technology agnostic

I agree.  We should be careful with the words:
- dissemination
- broadcast
- multicast

because they could easily clash.

Multicast can be "IP multicast", some-link-layer multicast (e.g. WiFi 
multicast).  But can not be "radio multicast".

Broadcast can be "IP broadcast, 255.255.255.255", sone-link-layer 
broadcast (e.g. ff:ff:ff:ff:ff:ff in WiFi) and can be "radio broadcast", 
or even "TV broadcast".

Dissemination is ambiguous with respect to all.  There is no such term 
as "IP dissemination", nor "some-link-layer dissemination".  HEnce, we 
could be free to define it.  We could say that geonet dissemination is a 
sequence of IP multicast packets which are radop-broadcasted in area no 
larger than 50 square kilometers.

What do you think?

> 4) show how geographical areas, including overlapping areas, can be
> possible mapped to IP addresses
> Ray Bellis mentioned that this is a very hard problem to solve and
> therefore it might be required to bring Geonet to IRTF and not IETF.

I fully agree.  It is hard in the computational sense in that decisions 
of member appartenance to sets is NP-complete.

> == Option 2: Try to go for a Researching Group instead of a Working Group.
> ==
> Advantage is that the problem statement, use cases, etc. can be more vague
> as it is a research topic. Take it to Lars?

This option could be explored as well.

What do you think?

Alex

[...]