RE: [Manet-dt] OLSRv2 NHDP comments

> I am not quite
> sure how it gets neighbor info from 2hops away since HELLO is
> only for single hop.

In order that node A and node B can establish that there is a
symmetric link between them (and many protocols - including
of course OLSR - need this) node A must advertise B's address
in its HELLO message (and vice versa). Now node C, which also
hears node A's HELLO message, hears node B's address in that
message, and hence knows that node B is a two hop neighbour.
So you get two hop information "for free" as part of the
establishment of symmetric links. Of course there are some
additional complexities, in particular associated with multiple
interface nodes where things aren't quite free any more. But
an important thing to note is that although multiple interfaces
complicate matters, if you don't have any (in your network) you
don't pay that. (Note that even if you are single interface
yourself you must be able to cope with multiple interface neighbours
or you will be noncompliant as you can break things if you just
assume neighbours are not. We've seen this in OLSRv1.) And we have
done quite a bit of work on reducing the cost when you do have
them. (OLSRv1 had MID messages, OLSRv2 first had MA messages,
then these were incorporated into TC messages, then we got rid
of them altogether with a bit of extra information in HELLO
messages in NHDP.) And multiple interfaces are important in
that if you have them, they need to work.

> Also, the protocol seems a bit
> complex with symmetric, assymtetric times and interfaces - perhaps
> it's because it needs to support OLSRv2.
> I wonder whether it can be simplified to deal with only MANET
interfaces?

I'm not sure what you mean here. Supporting symmetric links is
important to many protocols, and that's most of the complexity.
(Obviously asymmetric links occur in a dynamic network as a
step towards establishing the symmetry of a link, as well as
some links actually being asymmetric at right now.)

But the basic thing is, it may look complicated, but that's because it's
(hopefully - if you think otherwise, let us know) fully specified in all
details (or at least the next draft coming before the next cutoff will
be, I hope). Getting that right, and allowing for all the ways people
might
do something unexpected that you need to handle, is where the pages come
from.

> I assume, 2 hop restriction is because of OLSRv2 support as the nodes
> are assumed to be multi-interfaced. Without that, it might be simple
to
> go for only link-local discovery of  MANET nodes.

As I note above, once you want symmetry, you get two hop information.
But two hop information is important - it allows you to establish
a means of optimised flooding, using MPRs by original design (in OLSR)
but also other options. And all three MANET protocols that may use
NHDP (OLSRv2, DYMO and SMF) want (as a future option in DYMO)
optimised flooding. (Because they may not all use MPRs, these are
not currently specified in NHDP, but left to protocols building on
NHDP.)

> Is there any particular
> advantage of knowing neighbors more than two hops away ?

I believe there are protocols that use this. But it's a law of
diminishing
returns, and (to the best of my knowledge) none of the protocols to use
NHDP at this time want it. But as NHDP is built on the extensible
packetbb
format.

Hope that helps.

Christopher Dearlove

********************************************************************
This email and any attachments are confidential to the intended
recipient and may also be privileged. If you are not the intended
recipient please delete it from your system and notify the sender.
You should not copy it or use it for any purpose nor disclose or
distribute its contents to any other person.
********************************************************************

_______________________________________________
Manet-dt mailing list
Manet-dt@ietf.org
https://www1.ietf.org/mailman/listinfo/manet-dt