Why You Don't Need MU-MIMO

[thiggins]

The most recent testing under controlled conditions (octoScope system) shows total throughput gain with three STAs @ 90 degree separation.
It’s unlikely you would notice any improvement under normal usage unless you specifically tested for it.

 

[L&LD]

So, the benefits are real then.

 

[Tech9]

It’s unlikely you would notice any improvement under normal usage unless you specifically tested for it.

Indeed. I keep Beamforming Enabled, but MU-MIMO Disabled. I have 8 clients with MUBF support on my network and the only MU-MIMO effect I notice is cutting down the link rates from 866Mbps to 433Mbps when MU-MIMO is trying to do something. With MU-MIMO enabled and 2x concurrent transfers from my NAS the graph looks like hills and valleys. With MU-MIMO disabled both transfers have almost flat graph. Most of the time the transfers finish faster when MU-MIMO is disabled. Also, I have 4x access points and they serve multiple clients at the same time anyway. Folks with wired AiMesh also don't really need MU-MIMO enabled. Multiple radios do the job better, even using the same available bandwidth on the same channels.

 

[digital10]

Indeed. I keep Beamforming Enabled, but MU-MIMO Disabled. I have 8 clients with MUBF support on my network and the only MU-MIMO effect I notice is cutting down the link rates from 866Mbps to 433Mbps when MU-MIMO is trying to do something. With MU-MIMO enabled and 2x concurrent transfers from my NAS the graph looks like hills and valleys. With MU-MIMO disabled both transfers have almost flat graph. Most of the time the transfers finish faster when MU-MIMO is disabled. Also, I have 4x access points and they serve multiple clients at the same time anyway. Folks with wired AiMesh also don't really need MU-MIMO enabled. Multiple radios do the job better, even using the same available bandwidth on the same channels.

are you sure beamforming is ok? I think i tested it once and it gave a gain on 1 client but then totally deprived other clients from the bandwidth.

both MU MIMO and beamforming come disabled by default on Orbi which makes me very wary of it.

 

[Tech9]

are you sure beamforming is ok?

Explicit Beamforming is part of 802.11ac specifications. It is not supported by all AC class clients and some may have issues with it enabled. It all depends on what mix of clients do you have. In case you have issues with it, you can disable it. It doesn't work as pictured in some advertisements anyway.

I've seen better ones on router boxes like the router is shooting clients with a laser beam and 10 of them at the same time.

 

[thiggins]

the only MU-MIMO effect I notice is cutting down the link rates from 866Mbps to 433Mbps when MU-MIMO is trying to do something.

That is a known “feature” of Broadcom’s implementation.

 

[Tech9]

That is a known “feature” of Broadcom’s implementation.

I see this on my Ruckus R610 access points, Qualcomm hardware. They have 3x3 radios. I don't think there is other way to do MU-MIMO.

 

[thiggins]

I see this on my Ruckus R610 access points, Qualcomm hardware. They have 3x3 radios. I don't think there is other way to do MU-MIMO.

Two stream MU-MIMO is actually supported in the spec. But since MU-MIMO works with spatial streams, using 2 streams cuts the # of supported STAs in half.

 

[Tech9]

I don't know if it can do 2 + 1 streams on a 3-stream radio. My clients are 2-stream and if the link rate doesn't change, I don't know if it uses MU-MIMO or not. The fact is I see no difference in total throughput whatsoever. If there is a difference, it's not easily measurable.

 

[digital10]

I don't know if it can do 2 + 1 streams on a 3-stream radio. My clients are 2-stream and if the link rate doesn't change, I don't know if it uses MU-MIMO or not. The fact is I see no difference in total throughput whatsoever. If there is a difference, it's not easily measurable.

the question is, does it cause instability for other devices?

 

[thiggins]

Not under typical conditions, I.e. channel nowhere near capacity

 

[digital10]

Not under typical conditions, I.e. channel nowhere near capacity

so no harm if turning mu mimo and beamforming on?

 

[Tech9]

If you have devices that complain, there is no harm to turn both off as well.

 

[sfx2000]

That is a known “feature” of Broadcom’s implementation.

I thought that was sorted on the 2nd Broadcom silicon...

 

[Tech9]

I don't see it as an issue because the link rates jump back to 866Mbps right after the MU-MIMO attempt/event. I have MU-MIMO disabled on my APs. This was just an observation when doing some tests and it was quite some time ago. R610s are good as they are and not going anywhere any time soon.

 

[thiggins]

I thought that was sorted on the 2nd Broadcom silicon...

I'll double check with someone who has done more recent testing.

 

[jontalk]

Delays, missteps and over-aggressive marketing have made this once promising technology something you can live without.

Read on SmallNetBuilder

Wanted to thank you for sharing this article which shed light on a feature I knew nothing about. And as you probably know the default ASWRT settings enable all these features. Now that I've disabled all of them along with all types of beam forming, I'm finding much better throughput and range with my AX-68U which has been rock solid since the day I purchased it. I'm grateful for the education. Blessings