Re: WiFi AP selection



While I was looking up how to spell "Johnson" I discovered that the Johnson Nyquist thermal noise is:
20 MHz
−101 dBm WLAN 802.11 channel
Is indeed ~100dBm for a 20MHz channel. Which makes the Linux limit of -110dBm for 'no bars' kinda dumb: the signal is 1/10th of the thermal noise level. As I have commented before, displaying 'signal strength' as a linear representation of the power in dBm also makes no sense. A signal of -40dBm is about as good as it gets with the radio within a foot or so of the AP transmitter so that is OK as the 100% level. At -70dBm you are down to 1/1000th of that level and that ignores whatever noise is around.

Red line: "signal strength percent, as reported by Linux" versus dBm


On 11/9/19 1:45 PM, Bill Riemers wrote:

I will bite.  How can a signal be too small to be considered a no signal?  You need a certain minimum to receive data over the background noise.  Anything smaller than that you want to consider no signal.


On Sat, Nov 9, 2019 at 3:55 PM Clive McCarthy via networkmanager-list <networkmanager-list gnome org> wrote:
I haven't given up in the hope that NM will make a better WiFi AP selection.

From what you have told me, a "supplicant", requests the NM to open a network connection. The NM manager then selects something from the various WiFi APs or wired connections. From what you have said so far, the NM manager prioritizes potential data bandwidth over signal strength. As a consequence, a 5GHz signal at -100dBm wins out over a -40dBm 2.4GHz signal, despite the fact that the 5GHz signal one millionth the power of the 2.4GHz signal.

So I began to research how much power -100dBm represents. My first thought was to compare the power to an ant. Not a good choice since an ant can put out 50mW, according to some people. A more useful comparison turns out to be a bacterial flagellar motor (E. coli has them, for example). They crank out 10⁻¹² mW which is  -120dBm. So for -100dBm you need 100 E. coli.

So the point of this? -100dBm is way too small to be considered no signal. The edge of an acceptable signal is around the power output 100,000 E. coli.





_______________________________________________
networkmanager-list mailing list
networkmanager-list gnome org
https://mail.gnome.org/mailman/listinfo/networkmanager-list
--

Bill C Riemers, PhD, CSM, CSD, SALESFORCE CPD I

Senior Software Engineer

Red Hat Canada Ltd

Enterprise Sales + Services (ESSA)






[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]