We talk like WiFi is single carrier QAM, but in reality it is OFDM (802.11ac) 
or OFDMA (802.11ax) with 64 or 256 subcarriers per 20 MHz channel.  Each 
subcarrier using QAM modulation.  (But unlike ADSL, each subcarrier uses the 
same modulation level)

 

So in my mind’s simplified view, with single carrier QAM (like most licensed 
radios), there would be an automatic equalizer that would do much of the heavy 
lifting (along with a timing recovery mechanism).  With OFDM, the multiple 
subcarriers do a lot of the work of the equalizer.  I was involved in a more 
technical way with ADSL in my former life, and you could actually do a pretty 
good single carrier simulation by calculating the multicarrier throughput and 
assuming a good equalizer would achieve the same thing.  But the ADSL twisted 
pair channel is quite different because it spans such a wide range of 
frequencies with drastically different attenuation and crosstalk, whereas 
microwave channels are fairly narrow and uniform.  (although the “puncturing” 
feature of 802.11be seems to address non uniform channels)

 

When we get into mitigating and even making constructive use of multipath 
propagation, I cannot wrap my head around how that works in a multicarrier 
system.  Does it operate at the subcarrier level, or is there some kind of 
signal processing equalizer operating over the entire channel?  And what does 
it use as a metric to drive its optimization?  Even a younger me would struggle 
with the concept.  Now I struggle with what to make for dinner, or remembering 
if I shut the garage door.  It feels like when I look at the Differential 
Equations textbook on my bookshelf and realize I haven’t used any of that since 
the final exam for that class.

 

Given that 802.11be supports up to 320 MHz channels (across non contiguous 
spectrum in different frequency bands), that would be what, 1920 subcarriers? 
(including pilot and guard tones).  No idea if some of this is done on chip in 
dedicated hardware, but that’s an incredible amount of signal processing, 
especially at the symbol rates involved for multigigabit data.

 

Even understanding how signal processing extracts V and H or +45 and -45 
polarizations when the antenna is rotated is more than I can grasp.  Although I 
probably have a digital communications textbook that would explain it.  An 
amazing amount of microwave theory has been around since the 1950’s.  But the 
processing power to handle stuff like 1920 or even 256 subcarrier OFDMA has 
obviously only existed recently.  Amazing to be able to buy it for a couple 
hundred bucks and hold it in your hand.

 

From: AF <[email protected]> On Behalf Of Chuck McCown
Sent: Thursday, October 9, 2025 9:10 PM
To: 'AnimalFarm Microwave Users Group' <[email protected]>
Subject: Re: [AFMUG] 16 spatial streams how is that possible

 

I can only presume that is referring to a particular pattern when phased in a 
certain way.  I know the radars in fighter jets used to move the aim around 
with beamforming so the antenna didn’t have to move.  Now a days I hear they 
have multiple main lobes that are independently steerable.  No idea how the 
separate the signals coming from and going to the different beams.  

 

From: AF [mailto:[email protected]] On Behalf Of Adam Moffett
Sent: Thursday, October 9, 2025 6:26 AM
To: 'AnimalFarm Microwave Users Group' <[email protected] 
<mailto:[email protected]> >
Subject: Re: [AFMUG] 16 spatial streams how is that possible

 

Since you're the best expert we have: 

 

Is "spatial signature—gain phase pattern" a real thing or is there a blowhard 
editing Wikipedia?

  _____  

From: AF <[email protected] <mailto:[email protected]> > on behalf 
of Chuck McCown <[email protected] <mailto:[email protected]> >
Sent: Wednesday, October 8, 2025 10:06 PM
To: 'AnimalFarm Microwave Users Group' <[email protected] 
<mailto:[email protected]> >
Subject: Re: [AFMUG] 16 spatial streams how is that possible 

 

I used to do lots of RF, lots of microwave, waveguide stuff, antennas etc.  
Constantly heard folks refer to it as black magic.  To me it was like anything 
else, once you get it digested it seems pretty simple.  Smith charts scare 
people way I think, I think they are an aesthetically pleasing representation 
of the complex plane.  I digress.  Pulling multipath signal artifacts out of 
the ether and reconstructing them is truly black magic to me.  

 

From: AF [mailto:[email protected]] On Behalf Of Bill Prince
Sent: Wednesday, October 8, 2025 6:57 PM
To: [email protected] <mailto:[email protected]> 
Subject: Re: [AFMUG] 16 spatial streams how is that possible

 

I have seen signal tests with a local operator deploying Tarana getting some 
remarkable throughput even in NLOS situations. The main feature is using 
multipath in time and polarization to enable signal reconstruction.

 

bp
<part15sbs{at}gmail{dot}com>

On 10/8/2025 4:04 PM, castarritt wrote:

The 450M did beam steering with phase delay that allowed it to shape the 
pattern, but only in a sort of sawtooth like preset pattern that it could scan 
left and right.  Tarana can dynamically beamform to shape the pattern towards 
and away from specific azimuths as needed, and perhaps most importantly, the 
Tarana client radios can do the same.

 

On Wed, Oct 8, 2025 at 5:34 PM Ken Hohhof <[email protected] 
<mailto:[email protected]> > wrote:

Over the years I’ve become suspicious of the term beamforming.

 

In the hands of marketing people, having multiple antennas and choosing the one 
with the strongest signal, becomes “beamforming”.  Even a Cambium 450m I don’t 
think does beamforming (in the sense of feeding multiple array elements with 
different phase delays to shape the antenna beam), it just has 14 narrow 
sectors inside.  But that allows it to talk to multiple SMs at the same time.  
I think cellular antennas may use actual beamforming, I don’t know.  Tarana 
talks like they use active beamforming, but it could just be the usual 
marketing hype, again I don’t know.

 

From: AF <[email protected] <mailto:[email protected]> > On Behalf 
Of Adam Moffett
Sent: Wednesday, October 8, 2025 5:11 PM
To: 'AnimalFarm Microwave Users Group' <[email protected] 
<mailto:[email protected]> >
Subject: Re: [AFMUG] 16 spatial streams how is that possible

 

I suppose there's electronic beam forming, but could that really work with an 
array of 16 little dipoles in a circle?

 

Since all the clients are 2x2, and the DSP magic to differentiate all those 
other chains has to come with a monetary and electrical cost that most 
consumers aren't going to pay, I'm guessing that a16x16 router would just be a 
bicycle with 14 extra wheels. 

 

  _____  

From: Adam Moffett <[email protected] <mailto:[email protected]> >

Sent: Wednesday, October 8, 2025 5:51 PM
To: 'AnimalFarm Microwave Users Group' <[email protected] 
<mailto:[email protected]> >
Subject: Re: [AFMUG] 16 spatial streams how is that possible 

 

What could you theoretically subtract from the raw received signal to leave you 
with one desired transmission distinct from the others?  It's the 21st century, 
so we don't need to worry about how computationally intensive it would be; we 
only need to worry about whether you could do it.  

 

For instance, if I have 16 different QAM constellations on top of each other, 
but I knew they each had a certain phase offset, then could I separate them?  
What about a time offset?  

 

And oh boy, what if there's one 802.11n client on the WLAN?  He's 2x2 and 
doesn't have the magic to handle 8 or 16 chains.  Whenever the AP sends a frame 
to that guy does it have to temporarily stop transmitting on all the other 
chains?  I bet it does.  

 

-Adam

 

 

 

 

  _____  

From: AF <[email protected] <mailto:[email protected]> > on behalf 
of Ken Hohhof <[email protected] <mailto:[email protected]> >
Sent: Wednesday, October 8, 2025 3:07 PM
To: 'AnimalFarm Microwave Users Group' <[email protected] 
<mailto:[email protected]> >
Subject: [AFMUG] 16 spatial streams how is that possible

 

In the list of supposed advantages of WiFi7 over WiFi6, I see 16 spatial 
streams vs 8.

 

Can someone explain to me the mechanism for using 16 spatial streams in a 
typical WiFi environment?  I have a hard time wrapping my head around anything 
more than 2 using V/H or dual slant polarization.

 

I was willing to believe that maybe you could get more (like maybe 4 spatial 
streams) due to reflections off furniture and stuff, and that somehow signal 
processing could magically separate out the streams (even though I don’t 
understand how it does that).  But 8 or 16 just sounds like crazy talk.

 

Maybe it’s like the rich people houses with 16 car garages, if I were rich 
people I would understand?

-- 
AF mailing list
[email protected] <mailto:[email protected]> 
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

 

-- 
AF mailing list
[email protected]
http://af.afmug.com/mailman/listinfo/af_af.afmug.com

Reply via email to