At 11:44 -0500 11/11/05, Nelson, David wrote:
Phillip Hallam-Baker writes...
I think that what we should do is to send the IEEE 801.b/g group a
polite letter pointing out that if our people here at the IETF cannot
figure this stuff out then their less technically astute customers
be having some trouble as well.
I don't believe this is an 802.11 problem. That group standardizes PHY
and MAC (up to Layer 2) protocols. The usability problems with 802.11
networks are in the device drivers, operating systems and configuration
applications. It would be more effective to send mail to Microsoft,
Apple, et. al.
I disagree, I think. IETF, MPEG, large corporate conferences and so
on, they all have trouble running large 802.11 networks. They all
can run large wired networks. The difference is that even at
meetings run by and attended by supposed network experts, it's hard
hard hard to get an 802.11 network to run well. That is not right.
I do believe that there are (were) some operating systems that
switched to ad-hoc mode and made a network if it couldn't find the
network you asked to join. (I don't think it was OS X.) That's a
mistake. A big big mistake.
Guidelines on (a) network naming and (b) frequency selection from the
802.11 group would be useful. For example, maybe you need to do
something to claim to be an 'expert' to create an ad-hoc with a
'plain' name; otherwise your ad-hoc network would be (for example)
prefixed by "*" or something. And maybe OS's could diagnose
frequency problems ("there are several base stations in here all on
channel XX and they are interfering with each other" or whatever).
Dammit, a FAQ on <http://grouper.ieee.org/groups> would be a good
I've been at a meeting where a respected network equipment provider
provided the network. Because the base stations had an artificial
limit of 10 IP addresses for their NAT/DHCP, he setup 3 of them in
the room, next to each other, on the same channel and SSID. Result
-- they are all in very low-power mode, interfering like hell, and
the users if they get a signal can't choose from which box and so it
doesn't actually spread the load.
Finally, it's clear that at least some base stations get hopelessly
confused (sometimes I have even resorted to the technical term
"wedged") when there is an ad-hoc in range with the same SSID. Some
testing and robustness guidelines from the 802.11 group would also
Ietf mailing list