Hello Bruno,

There are some definite Duty Cycle (DC) Spikes in your trace. I believe that these are responsible for your Generic Wideband interference reports. 2414 - 2412 is pretty much center frequency of channel 1, and 2462 (the other one that I saw) is the center of channel 11. This is a dead giveaway. Your card is the first generation hardware, and this hardware see's very high wi-fi DC as a generic Wideband. It was a function of the hardware, and relative lack of CPU compared to the second generation cards. Interestingly back in 2005 when this hardware was released by Cognio, there where not a lot of WLAN's that would operate at DC above 30 % in a stable fashion, which is the trigger for this classification.

This is not a bug, but a hardware limitation - and one of the reasons why the second generation of hardware was released.

Now - as for your user issues. It looks like there are some AP's (very distant) on channels in between your 1-6-11 layout. These may have a bearing - in this trace they are not loud enough to see their DC.

I would check the lowest rates enabled on the AP's - try to eliminate 1,2,5 if you can, this will lower the overall duty cycle. DC as a whole on this network does not look bad at all - just in brief bursts.

As to your question on how DC afects a wifi network. 100% is the maximum number of timeslots available on the medium. Energy in the channel causes ED to go high, and CCA (clear channel assessment) to be set negative - which of course causes all our adapters to wait. If DC is bad enough - you will have nothing working.

Now think about how Wifi works - an AP is essentially a half duplex device and must spend at least as much time listening as it does talking. In a wifi network there are only two things that 802.11 can do to recover from a failed transfer attempt - it can retransmit the packet - which will turn on the radio a second time to send data that already was sent (increasing duty cycle) or rate shift to a slower rate - since the size of the data will not change - this will also increase the duty cycle significantly.

When things are really bad - you will see a heavily loaded network gradually approach 40-50% DC and then spike to 90-100%.

I'm not certain that this is what is happening on your network. The DC overall looks good on average. But there is something spiking it periodically, and depending on the clients this can be enough to cause issues at the application layer. Hence my suggestion to eliminate the lowest rates.

You might also try changing up some of the channels. From where this capture was taken - there was virtually no activity on channel 6 and very little on 11 - there was one artifact that I saw with High DC a couple of times in channel 11 - you will see this if you run the trace through and look at the Real Time DC plot with the max hold on....I'm not certain what that is - but possibly this could be causing you issues on the site as well.

As for tracking down the generic wideband - in the trace I saw it was not active long enough to really track it down. When an interference source quits transmitting - or we quit seeing it in classification you will see the time since last update counter at the top of the device finder tab start to increment. As long as you have the device in device finder - the classification event will remain true - up to 5 minutes after the last received transmission from the device.

Hope this helps.

Jim Florwick