MU-MIMO in new ASUS Router

[azazel1024]

Isn't the up coming Asus router 4:4 ALSO MU:MIMO? If so that is dramatically better (if it works well) than any other setup on the market.

MU:MIMO is able to actually Tx/Rx to multiple clients at once spliting up the radios instead of being dependent entirely on CDMA and round robining it. So even if your clients were only 1:1, you can dedicate a radio and antenna to up to 4 clients and have all of the 4:4 802.11ac throughput goodness, instead of it cut down to 1/4th of its possible throughput. MU:MIMO, IMHO is the most interesting (if they can get it to work) technology to come to wifi so far. Much more interesting and possibly better than 11ac.

 

[L&LD]

Only Tx to multiple clients at once, Rx still limited to one at a time. This should still be huge though.

 

[azazel1024]

Hmmm, I didn't realize the Rx limitations...but then again, the bottleneck is often more felt on the Tx side of things anyway...so, yeah, will be awesome if they get the kinks ironed out and in shipping products.

I wonder if there is a future possibility to do mulitple Rx at a time?

One of the things along those lines, and maybe it is how they do multi-Tx...but why not multichannel possibly? I know that could step on other SSIDs...but why not assign one client to one channel, another client to another channel and so on. You dedicate a radio (or more) per client up to the number of radios in that client and then you don't have any Tx or Rx worries at all.

I mean, 802.11 already does frequency hopping, so I don't see where channel hopping (with the appropriate driver updates are concerned) would be that much of a stretch. Just dynamically assign out the channels and radios and as the clients' usage patterns changed to maximize throughput, you can balance what the best assignment is. I am sure it would work better on 5GHz than on 2.4GHz as there are many more channel options and channel widths there, but this is an 11ac spec we are looking at.

 

[L&LD]

I think Receive (Rx) is going to be almost impossible because that would mean all clients would need to broadcast at the same instant. Hard because that would mean each client would need to communicate with every other client...

 

[thiggins]

MU MIMO also depends on explicit beamforming, so must be supported on router and STA. MU-MIMO is also limited to four STAs.

I highly recommend Matt Gast's 802.11ac Survival Guide if you want to get a better understanding of how hellishly complex beamforming is. Chapter 4 on beamforming can be read for free.

My simple experiments with beamforming so far say your mileage will definitely vary on performance improvement. I recently witnessed a beamforming demonstration that showed significant throughput improvement. I then had the demonstrator rotate the client device a bit and throughput dropped significantly.

 

[connorm]

Isn't the up coming Asus router 4:4 ALSO MU:MIMO? If so that is dramatically better (if it works well) than any other setup on the market.

IMO MU-MIMO will not be a factor for about 2 years untill its regularly supported in clients like laptops, smartphones and tablets. which probably account for a solid 50-75% of the wifi chips sold nowadays

 

[L&LD]

I don't think it will be two years for my specific situation.

The only support needed by client's is they be AC class compliant.

The router then has the full burden of implementing MU-MIMO effectively. (The client's don't know or care if it will be used or not).

 

[azazel1024]

I think Receive (Rx) is going to be almost impossible because that would mean all clients would need to broadcast at the same instant. Hard because that would mean each client would need to communicate with every other client...

Yes, but if the router/access point assigned clients to different channels, that wouldn't necessarily be the case.

If you were talking a router with a whole slew of radios, it could broadcast an SSID using a standard beacon. Then when connecting to the router, the router could tell the client to connect on a specific channel. As use of that channel became saturated with multiple clients, it could dynamically tell some of the clients to switch to different channels to move clear of current heavy user clients. With mulitple radios on the router, it shouldn't be that difficult to have some of the radios running on one channel and some on another and dynamically assign them around to the various channels as the client load changed.

I do get it would be incredibly complex to manage all of it, but I don't see anything in hardware that would prevent it. Just standards that would need to be come up with and drivers/firmware to manage it all.

I do also realize in a world with limited free spectrum, such a hypothetical router could hog a heck of a lot. We have 160MHz of usage in the 11ac spec though, even if current devices only are capable of 80MHz maximum. Just build in to the specs channel intolerance and make it mandatory (which doesn't mean some won't side step it, I know). If the router sees other networks at or above a certain signal strength, it will revert to single channel usage, instead of spreading clients across multiple channels and radios, or it can be set to just avoid the channels that are in use (supposing that is a possibility).

In the end though, even if its only ever workable as Tx (base station) MU:MIMO I am very, very excited about it. It should go a long way toward helping with Wifi congestion. Especially since Beamforming is required to get it to work, it would also likely mean at least some reduced congestion between neighboring networks as the received signal strength from the base stations would likely be a lot lower (baring being in line with a client and the base station from the neighboring network).

 

[connorm]

I don't think it will be two years for my specific situation.

The only support needed by client's is they be AC class compliant.

The router then has the full burden of implementing MU-MIMO effectively. (The client's don't know or care if it will be used or not).

?? no. it DOES NOT work on any ac client, not even close. the client and the router MUST BOTH support MU-MIMO for it to work. it doesnt just work on all ac clients, not even close.

the ONLY way for anyone to use this tech once its released is to have a client that specifically supports it (probably won't be a while untill vendors start using these chips as the standard 3x3, 2x2, and 1x1 ac chips are finaly just starting to get used. its not likely there gonna spend more money on MU-MIMO chips when no one supports it yet.

even 2 ac87u will not use MU-MIMO if i am correct. because all 4 antennas will tx/rx which defeats the entire purpose. it will however be possible to use beamforming, whether or not this will see a large benefit is still a question

 

[L&LD]

connorm,

have to disagree with almost everything you state. From what I've read, my statement is accurate.

 

[kenrok1]

Here is a concise explanation - it seems in a nutshell that a MU-MIMO router will be able to service multiple legacy AC clients more efficiently than a traditional MIMO router, and pairing MU-MIMO clients and routers net an even bigger gain

http://www.anandtech.com/show/7921/...increasing-the-efficiency-of-80211ac-networks

 

[connorm]

link doesnt work for me

 

[L&LD]

link doesnt work for me

Anandtech was down for maintenance or something yesterday - try it now.

 

[connorm]

im confused. cause reading that article only confirmed my previous claims you argued with.

"With MU-MIMO however it is possible to form groups of multiple devices that can be served at the same time. Through beam forming a MU-MIMO access point can group together multiple MU-MIMO aware devices into a single transmission slot."

"but as long as you have a MU-MIMO AP and some MU-MIMO clients even SU-MIMO clients will benefit from the improved efficiency on the network."

"On the device side, Snapdragon 801 platforms that use Qualcomm's integrated 1-stream 802.11ac WiFi will already support MU-MIMO with nothing more than a software update (the device vendor needs to actually offer that update however)"

all statements imply that in order to use MU-MIMO you need MU-MIMO capable clients. plz explain why you still think MU-MIMO will benefit SU-MIMO clients at all with 0 MU-MIMO clients on the same router.

unless you are atleast connection 2 MU-MIMO capable devices, the tech is useless, proving my previous points made valid.

as i have said before MU-MIMO will be pointless unless using MU-MIMO capable clients, which probably still won't become common for a while, but i could be wrong

 

[sm00thpapa]

MU-MIMO is new to me as it is to others. I'm just wondering why my RT-N66W works so well 24 hours a day with 23 devices connected to it at all times. I always have 5 streams of 1080P video from Netflix, Amazon Prime and Youtube simultaneously with out any issues. Will I even see a difference once I have all MU-MIMO devices in my household? My family and I are very happy with our network. The RT-N66W is a true work horse.

 

[L&LD]

connorm, it's okay. We are interpreting the same thing differently (I still think I'm right ).

And by we, I'm not talking only of you and I; but everyone that still doesn't have access to these future devices yet.

I'm don't want to have a discussion on how to interpret a third parties interpretation, but I can respect your position as possible as long as you can do the same for me.

 

[jalyst]

MU-MIMO is new to me as it is to others. I'm just wondering why my RT-N66W works so well 24 hours a day with 23 devices connected to it at all times. I always have 5 streams of 1080P video from Netflix, Amazon Prime and Youtube simultaneously with out any issues. Will I even see a difference once I have all MU-MIMO devices in my household? My family and I are very happy with our network. The RT-N66W is a true work horse.

The whole debate's moot, you only start to see advantages when you've got lots of clients regularly doing large simultaneous sequential.
And that advantage only kicks-in when there's at least 2x MU-MIMO clients on that network...

If it's a really large network, I imagine 2x wouldn't make much of a dent, you'd need a higher ratio of MU-MIMO to non_MU-MIMO.
I guess you could dedicate 4x MU-MIMO clients (the limit) to regular sequential, & leave the others for other uses, & in that scenario you may notice quite a difference.

Certain enterprise env. will benefit, but most home set-ups aren't likely to notice a difference unless they're real heavy (yours may for e.g.).
It's a nice feature, but it remains to be seen how much of a difference it can make for the typical (or even enthusiast) home network.
TH has said it's fraught with potential inconsistency/unreliability....

 

[jalyst]

I don't think it will be two years for my specific situation.

The only support needed by client's is they be AC class compliant.

The router then has the full burden of implementing MU-MIMO effectively. (The client's don't know or care if it will be used or not).

This directly contradicts what TH posted before you, can you point to some doco which supports this assertion?
It'd be great if we didn't need MU-MIMO clients, but everything I've heard/read so far doesn't suggest that.

 

[connorm]

connorm, it's okay. We are interpreting the same thing differently (I still think I'm right ).

And by we, I'm not talking only of you and I; but everyone that still doesn't have access to these future devices yet.

I'm don't want to have a discussion on how to interpret a third parties interpretation, but I can respect your position as possible as long as you can do the same for me.

im the one who is correct, MU-MIMO will not work with current AC devices with there current respective firmwares.

it could be implemented, but if its anything other then easy i would doubt asus isnt gonna add it to any of there current products, if it costs them money why would they.

it is also most likely that the biggest potential benefit of this will be tablets and smart phones, laptops and desktops are highly unlikely to have there wireless drivers updated for the standard home user.

the reason i say tablets and iphones is because they are commonly updated and it would be easy to include a new wifi driver in the update.

but as i said, it doesnt just work with any AC product, both client and router must support MU-MIMO.


"The only support needed by client's is they be AC class compliant." - quote from your post, incorrect.

"?? no. it DOES NOT work on any ac client, not even close. the client and the router MUST BOTH support MU-MIMO for it to work. it doesnt just work on all ac clients, not even close." - quote from my post.

if your arguing with me, your arguing with that statement, and you or wrong, so either your wrong or your wrong, just because you misunderstood doesnt make u right, but i do admit i didnt think manufactures would be able to add support for it with FW updates, that wasnt what we were argueing about though, and is more logical then your claims saying it will just work with any ac device.