I DO KNOW this is an old thread, but no real sensible explanation inside, and this topic may still be relevant for the next couple of years.
We tend to overcomplicate stuff in pursuing that last bit of Mbps/dB or whatever else.
tipstir (post #2) was right and I would add that b is DSSS mode and g(+onwards) is OFDM.
thiggins (#3) - just to clear that up, that does not mean, it would slow down, it just wouldn't discard the packets as "noise or whatever.
RadoFan (#5) yep, to disable 802.11b, select Wireless Mode : N only (it's under Wireless/General)
sfx2000 :
that's just wrong! These "recommendations" come from "professionals" who read a paper back in 2000(no pun intended) and did not bother to re-read. These people are of course real professionals, but of the kind who buys this technology, not the kind who designs it, therefore it's not even harsh to say that all they say is just a guess, as good as anybody's.
That what you refer to might have been the case with some draft 11g or some other pioneering 11g equipment, but I think it could have been remnants of a bad draft implementation - to slow down for backwards compatibility or whatever. Even if it was the other way around this stuff was solved many years ago.
All these conflicts are in software and unlike hardware, where a new piece might be more delicate, software tends to get more robust. If 11b knows AdHoc, then rest assured it will not be a problem for the newer equipment.
It's the other way around: In the old days, when two 11b APs shared the same airspace, they had to be reduced to half power - this, in my book, implies that the new equipment might step on 11b's toes. Maybe it could take down a really poor 11g, but it's not taking down 11n - that will just not happen.
To support this, I have a "too new" 11b device (everyone was surprised a flagship phone had no 11g at that time) - Sony Ericsson P1i (it knows WPA2 and has a fairly nice range btw.) - and while my other devices show varying Tx/Rx in the ClientsList, this one shows a steady 1/11 (Mbps) - so this is clearly shows that the 11b is the "dumber" device, therefore that is the one you want to protect. Just like you wrote yourself in your last paragraph:
the 11b would crash - that's it, and yes, maybe the certifying authority decided (did not and will not look this up) it would be great for an 11n device to ASSUME that a noisy 2.4 Ghz spectrum means all the "archaic" 11g devices need to be ignored and as we know assumption is the mother of f-ups.
You know, when people talk about child labour in the factories, I look at the UI of these devices, and know that there certainly is child labour somewhere (I can imagine this is something Trump wouldn't say, yay), but really that checkbox could just show state and remain grey, not showing an illusion of a setting that "must" be tampered with.
When I look into my Intel wireless adapter settings, mixed mode protection is set to "CTS-to-self enabled" and all greyed out.
This basically is not a setting you are supposed to set, unless you use Wireless mode : Legacy, but that mode is just redundant, as it restricts you to 20MHz bandwidth only.
So long story short, aka
the long awaited TL;DR :
Wireless Mode : Legacy - allows setup (not sure if it does anything, didn't test)
Wireless Mode : N only - of course off, what would you use it for, anyways?
Wireless Mode : Auto - it can be ticked off, but ticks itself back on and it is true (tested with a 11b device), this mode makes the "Legacy" useless, as here you can set 40 MHz bandwidth (bonded channels), so what is the point of legacy? Using 11g devices without 11b protection or what?
I guess in case of "Auto", 11b and 11g devices communicate solely on the control channel and this is the one protected, but I guess it doesn't matter, beacause the older devices just ignore (are not aware of) the extension channel.
Then the performance of today's consumer AP HW is so high, there's
no reason to not let 802.11b devices on your network. If you have many modern clients, you just switch to N-only mode, but with no real performance benefit even in that case (and possibly trolling someone's legacy setup).
Sometimes it is just easier to go there, get hands dirty and test it, like I did
So this is my setup for RT-AC66u :
Wireless Mode : Auto
b/g protection : True (default, can't be changed)
Channel bandwidth : 40 MHz
If using 40 Mhz, try to use (whenever possible):
Control channel: 1
Extension channel: above,
as channels 12 and
13 (channel 13 used in: 13+below or 9+above) are rubbish (unless you only need a small coverage perimeter), also if you are in Europe and people around you adhere to another
ancient, so 2000 "pro tip" of 1-6-11 (instead of 1-5-9-13), you might get a lot of noise from channel 6, that's why you might want your beacon on 1 and only the extension on 5, not vice versa.
This is going strong; even with an 802.11b device on the network the .11n devices are up to highest expectation.
