Re: [nemo] Next steps in RO

[Chan-Wah NG <[e-mail]>]

Hello all,

On Wed, 2005-11-02 at 10:02 -0800, T.J. Kniveton wrote:
> On Oct 31, 2005, at 11:16 PM, Thierry Ernst wrote:
> 
> > I think authors of the drafts have done their part of the work, so now
> > it's our turn as a WG to decide where to go from now.
> 
> 
> Folks,
> 
> Please let's have some more discussion on this point. What are  
> important aspects of RO that NEMO should work on, and why?
> 
> Is this area still too undeveloped for standardization? Perhaps RO in  
> mobile networks should be turned over to MOBOPTS and the IRTF for  
> further research?
> 

I am contributing my two cents.

Judging by the number of proposals since begining, I think there are
definitely interest in having a standardized RO solution.  The danger of
moving this to the IRTF is that (1) the IRTF is not chartered to produce
standard, (2) it will take a lot longer.

I do not know what is IETF's basis for moving work to IRTF.  Can someone
shed a light on this?  When there are too many competing proposals?  Or
when there is not enough knowledge in the particular problem area for
there to be a sound engineering solution to the problem?

Now that the solution space is starting to be analyzed, people should
have a better understanding of the scope of RO work in NEMO.  I don't
know whether the work we did would suggest to people that we know the
solution space well enough to engineer a solution, or that we do not
know enough and more research work has to be done. :)  I peronally would
like to think its the former though :).  But that is the first thing the
WG should decide: Based on what is documented in the two RO drafts, do
the WG feel that there is more research work that needs to be done?  If
yes, what are they?  If not, I don't wee why the NEMO WG should not be
re-chartered to develop the RO solution.

If we are to go forward and devise an RO solution, I think it is
definitely not a single piece of solution.  Instead, I see a lot of
different pieces to the puzzle. 

My personal views are that the following 3 main pieces are needed:

(1) A mechanism for MNNs to send packet from where the NEMO is attached.

This mechanism would allow MNNs, however deeply they are nested, to
appear to a node outside the NEMO as if they are a simple mobile node.
attached to the point of attachement of the root-MR.  This mechanism
would allow a solution to the nested mobile HA problem, and the security
policy prohitbiting visitor traffic problem.  Not only that, it also
allows VMN to operate MIPv6 RO without being subjected to NEMO-BS
tunnel.

(2) A mechanism for MRs to offer route optimization to LFNs.

This is less important, but it's a mean to complete the entire RO
process.  The mechanism above will allowed modified nodes (i.e. VMN) to
achieve RO in a way compatible to the more "legacy" MIPv6 RO.  This
mechanism, will allow RO to be achieved for unmodified LFNs.

(3) A correspondent router solution.

The deployment of such an entity would allow two benefits: (1) perform
RO for non-modified CNs, and (2) the discovery of MR by other mobile
nodes, if this MR is treated as a CR.

Some of the above may require future research, and may encompass a lot
of smaller but significant issues to be solved.  One can in fact have
the NEMO WG re-chartered to work on only 1 or 2 of the above, and have
the rest moved to IRTF for futher research.

/rgds
/cwng