Japan can use all 14 WiFi channels (World/ETSI can use 13, US/FCC 11).
The IETF network was restricted to the US 11 channels, obviously so that users
of US cards would not be left cold.
On Monday, and for some time on Wednesday, there were problems with overlapping
channels. In WiFi, channel bandwidth is between 20 and 25 MHz, so there must be
a 4 to preferably 5 channel (5 MHz) gap between the channels that are actually
being used. Stations on the same channel coordinate. Stations on different but
overlapping channels essentially serve as efficient jammers to each other.
While this problem was solved by Monday dinner time using an appropriate
frequency plan (using only channels 1, 6, 11), the rogue stations that
repeatedly popped up at channels 3 and 10 probably did not help.
A lot of the problems that are seen in real wireless networks connected by
random Ethernet switches is that some Ethernet switches do not update their MAC
address tables fast enough to cope with fast roaming. I'd assume (but don't
know for sure) that we had our share of that.
I've been busy this week, so I have not attempted a full analysis of what
happened, but these are some data points.
I believe that IETF meetings are consistently the largest groups of WLAN users
in one network that ever happen, so they would make great research subjects (for
non-intrusive research, please).
Gruesse, Carsten