Regarding UCI options I find zero support and I also do not find information regarding asking the capability of the radios. From the hostapd repo I found a parameter for the .conf (he_ofdma_disable) and I do not have pretty clear if in hostapd-phy*.conf I can find anything else.
Also I want to understand what goes about the requirement of having or not having OFDMA support in a WiFi 6 radio: is it the hardware or is it the driver?
For example: for TP-Link EAP615-Wall and EAP613 are indicated as OFDMA capables, while the Xiaomi Mi Router CR6606 with the same WiFi Chipset (MT7975DN + MT7905DAN) is not indicated as OFDMA capable (not explicitly in the table at least).
OFDMA is part of air packets that they contain multiple packets for multiple clients at once. It is not capabilty, it is radio discharging protocol data for everybody at once.
As I said, for hostapd.conf it is a capability, enabled by default (he_ofdma_disable).
For 802.11 and WiFi alliance OFDMA is also a capability reported in Beacon frames and other ones.(EDIT: sniffing beacon frames I do not find explicit capabilities regarding this, unfortunately the version I have of 802.11 is from 2013 so 802.11ax is not included to check the theory of this) At hand I have the EasyMesh specification where also OFDMA is an HE Capability:
In the table of capabilties I have found that the coexistence with Beamforming seems not to be an issue.
I never thought about it, unlike the coexistence between OFDMA and MU-MIMO which seems really complex to achieve from my point of view. In other hand I remember when Beamforming and MIMO (Single User) where new I found really difficult to believe that both can be done at the same time (since beamforming is using multiple antennas to transmit as one unique antenna/stream, and MIMO was using them to transmit multiple streams)
My general experience when trying to run OpenWrt on many different EAP (wifi5) devices is that the available opensource wifi drivers clearly lacks all the fancy dynamic performance that the original closed source drivers have.
The performance difference is so big I would say the opensource drivers actually behave like simple round radio on 5 GHz.
But the quality check here is pretty much only DL/UL speed test.
Go away 5-10-15m and redo the test to see if beam-forming actually work?
Or what happens if you put a wall between AP and device?
Or move laterally 90° and see the wifi strength change and come back.
It is these kind of tests that opensource drivers lack performance compared to closed source drivers, not DL/UL speed when sitting next to the device.
But it is impossible to know what it actually do without putting the device in a screened EMC room and measure the actual radiation pattern, but that won’t happen.
Well I think that all of this should be proven. I find that probably some devices, chipsets, etc... differ in performance.
I find some devices performing really well but for sure a measurement with appropiate tools and methods would be the right proof in each case and the source of objective comparations.
Regarding OFDMA I have been reading 802.11ax, sniffing beacon frames, etc and I am almost sure that there is no OFDMA capability for 802.11 (but it is in MultiAP/EasyMesh spec).
So probably the 802.11ax certification test DL and UL OFDMA as mandatory but in practice you could find devices where they simply do not "choose" to use OFDMA in DL,UL or both, without expressing it via explicit capabilities.
Anyway it will be really usefull for the community to know the actual status of OFDMA in every hardware, and better if existist an internal API to check the support over a running firmware. Also to check on live when OFMDA is being used or not (and how: which tones, how they are being assigned...)
