Hi Carsten, Ron and all:
Agree, it is a good point that we did not clearly discuss the topology for
6lowpan architecture. During 6lowpan-nd discussion, we discussed and
agreed on different supporting topologies for 6lowpan ( both at interim 05
and at Dallas IETF meeting). Section 4 of 6lowpan-nd draft captures that
text (see below). If we decide to have a separate architecture document,
then we can use these portion of the text.
---------------------------------------------------------------------------------------------
4. Assumptions about Topology and Address Mapping
In order to minimize the multicast packets we need to make some
assumptions that tie together the L2 functional elements and the L3
functional elements. We also state our understanding of how IPv6
would map to a LowPan network:
o One PAN-ID defines one LowPan Network.
o Each LowPan corresponds to one IPv6 subnet as PAN-ID may be used
to determine a subnet.
o There is one PAN co-ordinator or PAN group leader per one LowPan.
o The IPv6 router which sits in the boundary of the LowPan is a PAN
co-ordinator.
o There can be multiple co-ordinators in the LowPan.
o When a device connects to the LowPan at layer 2, it finds out a
(unicast) layer 2 address for its co-ordinator.
o By recursive construction, the previous item implies that a co-
ordinator knows its co-ordinator from when it connected to the
LowPan, hence there is distributed knowledge of unicast addresses
that lead all the way to the PAN co-ordinator.
o The IPv6 router assigns addresses through prefix advertisement.
o The other FFDs in the network do not act as a IPv6 router, but
they generally route data packet in the L2 layer (Mesh layer
routing).
o Star topology assumes that each node is one hop away from the PAN
co-ordinator.
o This document defines a mesh topology (see diagram below). In
mesh topology, each node is capable of forwarding. Thus it can be
assumed as a network of full functional devices (FFDs) with one
PAN co-ordinator and multiple co-ordinators.
Chakrabarti & Nordmark Expires December 26, 2006 [Page 6]
Internet-Draft LowPan Neighbor Discovery Extensions June 2006
o FFDs may or may not be more than one hop away from the PAN co-
ordinator.
o We assume that the 64-bit EUI-64 addresses are used as link-layer
address in the lowpan, since these addresses never change for a
given node. Appendix A discusses some additional considerations
should we apply this to the 16-bit addresses.
This document currently supports the following topologies:
1) Star Topology
PAN-C
O
/ | \
o o o <------- RFDs or FFDs
2) Mesh Topology
Example-1 Example-2
O PAN-C O <-- PAN-C
/ | \ o / | \
/ | \ / / | \
FFD----FFD----- FFD FFD--FFD-----FFD
\ |\ / \ \ | \ /|
\ | \ /| o \ | \ / |
\ | \ / | \ | \/ |
\ | \ / | \ | /\ |
\ | \ | \| / \ |
\| / \ | |/ \|
FFD \ | FFD FFD
/ | \ FFD
o 0 o / \
o o <--RFDs
Each FFD may act as simple co-ordinators but there is only one PAN
co-ordinator in the mesh LowPan topology.
------------------------------------
Thanks,
-Samita
On 10/27/06, Carsten Bormann <[EMAIL PROTECTED]> wrote:
Ron,
good input.
Let me just pick up on one of these points:
The IETF is quite wary about "architecture" documents.
(The OSI reference model is the textbook example here:
ISO did an architecture ["reference model"], and then, only when
defining the protocols, found out that the architecture didn't make
sense.
But it was too late to fix the architecture, so the bad layers limped
on until OSI finally died.
The really funny thing, of course, is that, in 2006, people are
*still* citing the layers of the defunct reference model...)
In the IETF, we tend to standardize functional components and leave
it to the market how to build "architectures" out of these.
Of course, that is not always possible.
Also, it may be necessary to have an (really: one or more)
architectures in mind when building components.
So, I'd welcome discussion of architecture(s). I'm just not sure we
actually want to "standardize" them.
Gruesse, Carsten
_______________________________________________
6lowpan mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/6lowpan
_______________________________________________
6lowpan mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/6lowpan
