Date: Wed, 10 Jun 2009 22:53:17 +0300
From: Zach Shelby <[email protected]>
What do others think, is this a concern? Do we just run Ad-hoc LoWPANs
with no ER at all by default? If so, what is the simplest way to detect
this, and are you happy with how the network will function after this?
Zach,
I don't think it would make sense to not have an ER in an ad
hoc LoWPAN. As you said, some router must generate a ULA
prefix and disseminate it, so there should be an ER.
I am hoping that we can come up with a way to run large
ad hoc networks of (exclusively) small devices without
any major perturbations to the protocol.
The whiteboard is used for assigning 16-bit addresses and
for detecting collisions in OIIs (usually derived from
EUI64s). Even with a whiteboard, OII collision detection is
somewhat problematic in that there can be a significant
delay during which the colliding devices will be using the
same address(es). Would it be acceptable for the ER in an
ad hoc network to maintain only a partial whiteboard
containing some fraction of the networks OIIs? This would
increase the delay before an OII conflict was detected, but
the network would otherwise behave as if the ER had a full
whiteboard.
Having globally-unique 16-bit addresses helps with header
compression. Typically, all or most of the traffic in a
LoWPAN is either to or from one of a small number of nodes.
Ideally, in a large ad hoc network of small devices the ER
could assign globally-unique 16-bit addresses to only the
high-traffic devices (including itself). This could be done
by adding a 1-bit flag to address options in order to
indicate that the requesting host is a high- traffic device.
The flag would be only be set if the A (address generation)
flag were also set. A memory-poor ad hoc ER would use the
flag in deciding which address generation requests to honor.
All other ERs would ignore the flag and always generate an
address.
In summary, if an ad hoc ER lacks sufficient memory for a
full whiteboard, we would see the following:
- An increased delay in detecting OII collisions.
- Not all requests for address generation would
be honored; preference would be given for hosts
indicating that they were high traffic devices.
- Some loss in header compression for packets to
or from hosts that did not receive 16-bit
addresses.
- Local (one hop) detection and repair of 16-bit
addresses at the link layer, presumably done in
conjunction with link-layer security.
-Richard Kelsey
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
