Could someone clarify what noise is when we are talking about WiFi 802.11?
As we know the SNR is the difference between received signal and the noise floor. The noise floor seems to be all the signals in the measured channel, either 802.11 signal or ohter RF radiation on that channel.
I fully understand non-802.11 noise but WiFi moduled noise is where I'm strugling.
If we have two APs in same channel (plus associated clients) and I hear them both at -50 dBm, the SNR won't be 0? Is this because at the same point of time (when spectrum analyser takes the measurment point) there should be only one transmitting radio? Does this mean that if you have 802.11 noise, it's only from STAs where CSMA/CA doesn't work (due the reason x)?
Keep in mind that 802.11 is not truly constrained to the channel selected. That is, some energy bleeds over into adjacent channels. This adjacent channel interference (ACI) can be considered noise in relation to the physical energy detect algorithms of CCA (clear channel assessment). This is particularly true in 2.4 GHz, but also true to a lesser extent in 5 GHz with higher output power levels and close clients on adjacent channels.
In addition to visualize the "bleeding" of RF in 5 GHz in other channels,
see this picture of the CWNA class we just ran this week in Sydney.
(at the top you can see the AP) and you can see the green "RF" leakage of the Spectral Mask
at the left side (before channel 36 and into channel 52, 56 onwards)
FYI - The AP is doing 80 MHz 3x3 MIMO (Channel .11ac "42" (thus covering 36, 40, 44, 48)
(underscore 44 means that the "20 MHz' Beacon is send at channel "44")
Re 5 GHz: Bangkok, Thailand - this I captured last time, see how much 5 GHz is utilized on every channel:
(I can do a spectrum analysis here end of next week to see how much 5 GHz will spoil into each
other channels increasing the noise floors and spectral re-growth)
One important factor, often overlooked in these discussions, is the actual sensitivity of the radios in question.
Externally seen signal levels are of little relevance if the radios have poor sensitivity or are swamped by internally generated interference.
This can account for some devices (even of the same model) to require more signal than others, to be able to work properly.
Transmitter signal quality as defined by EVM values, also affects range, and is not predictable by simple spectrum plots.
These are a couple of reasons you need to allow a certain amount of margin in your link budgets.