Why You Don't Need MU-MIMO

[thiggins]

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

Read on SmallNetBuilder

 

[sfx2000]

I would tend to agree - there are use cases where MU is of benefit...

It's been a marketing problem for most in the retail space - doesn't help that Wave 1 was very successful - and Wave 2 is an incremental update to 802.11ac in the market.

 

[thiggins]

It's been a marketing problem for most in the retail space

If you call selling futures on features that never show up and when they do don't work well, then ok, it's a "marketing" problem.

 

[iwod]

Well You really cant blame Apple for not supporting something when the feature is not properly implemented.

But this got me extremely worried with 802.11ax.

Are we going to repeat this again, basically constantly shipping updated Hardware Chips to the market.

 

[RMerlin]

While playing Buzzword Bingo, here's a new one I came accross recently, as part of Broadcom's new HND platform:

Adaptive Bandwidth Control

No idea what it does...

 

[thiggins]

Well You really cant blame Apple for not supporting something when the feature is not properly implemented.

I think the suits between Apple and Qualcomm also are at play...

But this got me extremely worried with 802.11ax.

Are we going to repeat this again, basically constantly shipping updated Hardware Chips to the market.

I wouldn't say the hardware updates have been "constant" for 11ac.

But yes, 11ax will take a few rounds to get right. The market has consistently shown it is willing to be a guinea pig for beta grade Wi-Fi.

 

[thiggins]

While playing Buzzword Bingo, here's a new one I came accross recently, as part of Broadcom's new HND platform:

Adaptive Bandwidth Control

11ac has as part of the standard the ability to change channel bandwidth on a frame-by-frame basis. 11ax has the same.

 

[Sarre9]

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

Read on SmallNetBuilder

Excellent article again! Thank you for unravelling the MU-MiMo mess the marketing geniuses foisted on us.

 

[Trentors]

Great article Tim!

Anyways I have claimed it before and will still do so. In order to increase capacity on a single AP/Router the triband solutions are the only valid option.

Even an old device as RT-AC3200 is better suited to handle multiple devices than 4x4 MU-MIMO routers. So even though many find triband routers overkill they do actually work unlike devices relying on MU-MIMO to increase capacity. And now Tim has proved this.

 

[thiggins]

Great article Tim!

Anyways I have claimed it before and will still do so. In order to increase capacity on a single AP/Router the triband solutions are the only valid option.

Even an old device as RT-AC3200 is better suited to handle multiple devices than 4x4 MU-MIMO routers. So even though many find triband routers overkill they do actually work unlike devices relying on MU-MIMO to increase capacity. And now Tim has proved this.

All I have shown is that MU-MIMO doesn't deliver the promised improvements,

"Smart Connect" is another can 'o worms. Yes, the additional radio helps increase total bandwidth, IF devices are properly assigned among the three radios. But that rarely happens automatically. Clients can often resist being band steered. When I have tested Smart Connect in past reviews, I've found it to be pretty stupid...

 

[Trentors]

All I have shown is that MU-MIMO doesn't deliver the promised improvements,

"Smart Connect" is another can 'o worms. Yes, the additional radio helps increase total bandwidth, IF devices are properly assigned among the three radios. But that rarely happens automatically. Clients can often resist being band steered. When I have tested Smart Connect in past reviews, I've found it to be pretty stupid...

I also completely agree. Smart Connect is not worth it IMO. But the fact that you still have an extra 5GHz network does increase capacity if you manually assign the proper devices to use. I basically split the two 5GHz networks between low performance devices on the first and a few high performance devices on the second. That way I can keep 1x1 devices from unnecessarily slowing down the 3x3/2x2 devices. And it works regardless of clients unlike MU-MIMO.

 

[Makaveli]

While playing Buzzword Bingo, here's a new one I came accross recently, as part of Broadcom's new HND platform:

Adaptive Bandwidth Control

No idea what it does...

That's an easy one merlin it uses a custom boardcom AI chip to intelligently provide bandwidth to applications that needs it the most.

And yes that it totally made up like most of the stuff coming from marketing departments these days.

 

[WiFiNemesis]

Are we going to repeat this again, basically constantly shipping updated Hardware Chips to the market.

Yes.

The IEEE ax Task Group did not approve Draft 1.0 of the standard as expected this year.
The final specification isn't expected until 2019, if all goes as planned.
Similations have indicated performance expectations are unlikely to be achieved (the solution: change the simulation assumptions).

That hasn't stopped the usual suspects from announcing products already. It will be a long, rough road, as Tim suggests.

 

[sfx2000]

I think the suits between Apple and Qualcomm also are at play...

Informed insight - has nothing to do with the Apple/Qualcomm stuff (which really doesn't have a place here - there are better forums for that discussion)

Much more has to do with benefit vs. risk to the end-user experience. Apple basically doesn't see a need for MU - and while Broadcom has scored a significant amount of design wins - Apple is not strongly bound to any WiFi chipset vendor.

 

[sfx2000]

11ac has as part of the standard the ability to change channel bandwidth on a frame-by-frame basis. 11ax has the same.

Yep - it's not observed in the wild often, but it's out there - mostly in enterprise/carrier grade AP's, and usually under heavy use - I've got a few PCAP's where it's been caught.

 

[sfx2000]

"Smart Connect" is another can 'o worms. Yes, the additional radio helps increase total bandwidth, IF devices are properly assigned among the three radios. But that rarely happens automatically. Clients can often resist being band steered. When I have tested Smart Connect in past reviews, I've found it to be pretty stupid...

Shareholders benefit from increased sales of WiFi NIC's that support "Smart Connect" - to the end user, it can introduce a level of complexity that offers little benefit.

OEM's like it as it can run up the MAX BANDWIDTH (with enhanced POW!) marketing numbers

 

[sfx2000]

If you call selling futures on features that never show up and when they do don't work well, then ok, it's a "marketing" problem.

It's a hard story to tell, and a hard story to sell - there is benefit, but it's not a single client benefit - and MU does work in certain use cases, esp. under heavy usage where client stations are relatively fixed.

Part of the problem with MU in 802.11ac - it's a nominal implementation - mostly for interop purposes.

One sees MU more common in LTE space, and there, the value can be more obvious when one has 200 users per cell (sector in old-school CDMA speak) - so if a Cell Tower has 600 mobiles attached - which is relatively common in 3G/4G networks, MU can add additional capacity - similar to VAMOS I/II that was used for GSM (phase rotation in same slot at a very high level) and Pseudo-Ramdom codes for CDMA, extending the Walsh codes from 64 to 128 - theory was 2x, but one does have to account for orthogonal issues there.

 

[thiggins]

The IEEE ax Task Group did not approve Draft 1.0 of the standard as expected this year.

I wasn't aware of that. And, of course, none of the companies announcing products have shared that information.
But it's right there in the January 2017 meeting report.

• The working group letter ballot #225 on the IEEE 802.11ax draft 1.0 was conducted from December 1st, 2016 to January 8th, 2017.
• The 802.11ax draft 1.0 did *NOT* meet the requirement of >= 75% approval ratio.
• Over 7300 comments received.
  • About 4700 out of 7300 comments are technical and general comments.

Looks like they've moved on to work on Draft 2.0.

 

[WiFiNemesis]

I wasn't aware of that. And, of course, none of the companies announcing products have shared that information.
But it's right there in the January 2017 meeting report.

Looks like they've moved on to work on Draft 2.0.

It gets better. Apparently some 18 companies in a prticipating SIG ("DensiFi") got caught colluding, which set things back a bit. Fun reading if you have nothing better to do. I see that the Wikipedia entry for 802.11ax has had the incident edited out.

 

[thiggins]

Thanks for that, WiFiNemesis. The investigation report has the details behind the allegations and investigation.

Basically, a small group of powerful companies had pre-meetings to discuss upcoming topics in the Working Group meetings to coordinate votes. The problem was that all interested parties were not allowed to join the group ("DensiFi"), so their ideas were not heard.

The technical issue at the core of this appears to be related to an interoperability mechanism that will have significant effect on the ratio of airtime legacy and AX devices get.

This stuff gets pushed to market too quickly, but consumers keep buying. So there are no negative consequences for the companies involved.