What's new

AC should be in all devices now

  • Thread starter Deleted member 22229
  • Start date
  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

Well having 11ac and MIMO is only part of it - set-tops and mobile phones are getting to that point where we do see SU-MIMO 2-stream in late 2015...

The announced Roku is one - the current crop of iPhones launched earlier this month is another...

Client side is generally settling on 1-stream/2-stream 802.11ac in the SU-MIMO context - perhaps they know something that AP/Router vendors don't - and there's hella more clients than AP's, eh?
 
And when one looks at MU-MIMO and 802.11ac...

a 2-stream client - in the SU-MIMO context it's AC867, in MU-MIMO mode is AC533 as it loses the 2nd stream to sounding, ranging, and MU feedback - so the 2-stream clients are not motivated to implement MU-MIMO..

That being said - keeping things simple - an AP deploying MU-MIMO is likely going to have a better experience across all modes it supports, as in late 2015, and perhaps thru 2016, most clients in 802.11ac are going to be SU-MIMO preferred or SU-MIMO only..
 
Huh? Which manufacturers? And what products?

My point was if a Roku player can have AC and MiMo why would top dollar devices omit this from there WiFi interface.
 
My point was if a Roku player can have AC and MiMo why would top dollar devices omit this from there WiFi interface.

Think you're combining MIMO with MU and SU contexts... nothing suggests that the announced Roku box does anything new... e.g. SU-MIMO, perhaps 2-streams on VHT80 channels...
 
And when one looks at MU-MIMO and 802.11ac...

a 2-stream client - in the SU-MIMO context it's AC867, in MU-MIMO mode is AC533 as it loses the 2nd stream to sounding, ranging, and MU feedback - so the 2-stream clients are not motivated to implement MU-MIMO..

That being said - keeping things simple - an AP deploying MU-MIMO is likely going to have a better experience across all modes it supports, as in late 2015, and perhaps thru 2016, most clients in 802.11ac are going to be SU-MIMO preferred or SU-MIMO only..

Only the AP loses a stream, the client does not, as the sounding and ranging is performed exclusively from the AP end, with frames passed back from the client.

So your 2:2 MU:MIMO client stays 2:2, but a 4:4 AP would become 3:3 in MU mode.

On "top dollar", which ones are you thinking of? The new Apple TV has 11ac. The Roku there has 11ac. Pretty much all top laptops and tablets I can think of have 11ac. About the only products still shipping with 11n are lower tier cell phones, tablets and "generic" connected devices (IE cheap streamers).

In those contexts, the difference between a $8 wifi chipset and a $2 wifi chipset makes a big difference on the profit margins.

Same reason there are still some products shipping with 100Mbps LAN chipsets...because it saves a few cents and gigabit isn't necessarily needed (though, so long as 100Mbps is enough, on the wired end of things the "whole network" impact is minor compared to a shared medium where a slow device impacts everything).
 
My point was if a Roku player can have AC and MiMo why would top dollar devices omit this from there WiFi interface.

Top dollar devices don't.

Certain devices built to a price point do.
 
Only the AP loses a stream, the client does not, as the sounding and ranging is performed exclusively from the AP end, with frames passed back from the client.

So your 2:2 MU:MIMO client stays 2:2, but a 4:4 AP would become 3:3 in MU mode.

Ah, but you're missing an important part of MU - to do MU, you need more than a single MU client - so, if you have 2 MU clients (2*2:2 or whatever), someone may short changed, but most likely both will... depends on the scheduler and decision tree logic.

Sometimes is easier and more efficient to go SU instead, blast the bits and get it done with and serve the next client.

MU is not about speed, it's about capacity...
 
On "top dollar", which ones are you thinking of? The new Apple TV has 11ac. The Roku there has 11ac. Pretty much all top laptops and tablets I can think of have 11ac. About the only products still shipping with 11n are lower tier cell phones, tablets and "generic" connected devices (IE cheap streamers).

In those contexts, the difference between a $8 wifi chipset and a $2 wifi chipset makes a big difference on the profit margins.

Same reason there are still some products shipping with 100Mbps LAN chipsets...because it saves a few cents and gigabit isn't necessarily needed (though, so long as 100Mbps is enough, on the wired end of things the "whole network" impact is minor compared to a shared medium where a slow device impacts everything).

Think you're quoting someone else there ;)
 
does that mean that wireless AC needs to be in all servers used in datacenters, 10G switches and routers used in datacenters, your watch,microwave,coffee machine, monitors (not TVs), clocks, external hard drives, desktop motherboards, wired switches, modems, cars, etc?
 
Ah, but you're missing an important part of MU - to do MU, you need more than a single MU client - so, if you have 2 MU clients (2*2:2 or whatever), someone may short changed, but most likely both will... depends on the scheduler and decision tree logic.

Sometimes is easier and more efficient to go SU instead, blast the bits and get it done with and serve the next client.

MU is not about speed, it's about capacity...

True. I wonder how good the schedulers are at maxing capacity.

You could, I would think, have it schedule so it alternates talking 867 to one client and 433 to the other and swap back and forth to create an effective 650Mbps speed to both clients. Or if it would do 867 to one and 433 to the other. If it went SU mode, you'd only have an effective 433 to each unless it wasn't sharing airtime equally.

This is why we need 8:8 basestations.

:)
 
True. I wonder how good the schedulers are at maxing capacity.

You could, I would think, have it schedule so it alternates talking 867 to one client and 433 to the other and swap back and forth to create an effective 650Mbps speed to both clients. Or if it would do 867 to one and 433 to the other. If it went SU mode, you'd only have an effective 433 to each unless it wasn't sharing airtime equally.

This is why we need 8:8 basestations.

:)

It'll probably be a while for 8*8:8 basestations, but 11ac at least has specs for that...

Going into the thread - the devil is in the details on the scheduler - whether to commit or not in MU - there are a lot of factors here - sounding/ranging across the candidates, the type of traffic across those candidates, and consider the trade off whether do do the MU frame or a couple of SU frames...

It's like relationships on the Book of Faces, it's complicated - if we have 10 SU clients, and two MU clients, and everyone is busy, then perhaps the decision is not to do MU, as there is a computational cost with the sounding/ranging for the MU clients, and we still have to serve up bits to all the SU clients out there.

And then consider what that Station/Client is doing and the application profile - if we have two MU capable clients, and one is doing best effort (email/web) vs. the other on a VOIP session - what should we do? Or another client streaming video?

It's not all cut and dried - MU adds an order of magnitude to the complexity, and that is why the scheduler is so darn important...
 
Very much agreed. As much as I am not a huge fan of more expensive spider routers, I can't help wondering once MU:MIMO is finally a common feature if the sweet spot is going to be 5 stream routers. That could provide two 2:2 clients at the same time, four 1:1 clients or it can serve six 2:2 clients in shared MU mode or 12 1:1 clients in shared MU mode (at least all MU:MIMO chipsets I have read details on, claim they can server 3 clients per multiuser setup, I assume before it has to completely break down and use SU mode).

Also since cell phones and tablets are both actually starting to finally move towards 2:2 setups, it might actually (some day) be that 1:1 clients are going to be the exception and not the rule, so having your base station support multiple 2:2 clients is probably a good thing. In a home environment supporting more than 2 devices at once in a multiuser mode is probably going to see the largest gains.

I suspect that 8:8 (or even 6:6) base stations are going to be a ways off and probably exclusively the province of enterprise access points.
 
I would suggest that there's still a benefit to the current crop of MU-MIMO router/AP's - at least for the ones shipping... and that benefit extends to all 11ac clients - single stream or more...

The AC2350/AC2400 class are all 4*4:4 - so there's measurable benefit to having the extra radio chain on the Tx/Rx side of the house - it's not huge, but it's there - additionally, for SU-MIMO client, the 4th spatial stream provides addition coding hints and repetition - put both of these together, and at a given range, one can probably go one MCS higher than a 3-stream/3-radio AP.

That's a win for all.

And all the efforts for the MU scheduler will benefit all clients, even if all are SU only - and this goes towards fairness and quality of service across different clients and application flows.

They might extend the absolute range a tiny bit, but that's more of a physics problem with RF propagation - both in free space and punching thru walls (attenuation both ways - from AP to client and vice-versa).
 
does that mean that wireless AC needs to be in all servers used in datacenters, 10G switches and routers used in datacenters, your watch,microwave,coffee machine, monitors (not TVs), clocks, external hard drives, desktop motherboards, wired switches, modems, cars, etc?

You forgot toasters ;)

I realize your comment might be slightly tongue in cheek, but you actually make a really good point - in the next few years, probably sooner than that, our homes/cars/offices/bodies are going to be ever more connected... and most of it will be wireless in some form of 802.11..

11ac might not be appropriate for all of those tasks, but if one were to consider 11ah (sub 1GHz) for those "Internet of Things" applications, along with 11ad (60Ghz)...

I do remember talking with someone at Dell at an IEEE plenary session where 11ad could certainly be that wireless docking station, more than enough bandwidth there.

When we go into those IoT applications - 11ah is optimized for that purpose - and that's pretty cool - as that frees up your 11n in 2.4GHz and 11ac in 5GHz for the normal applications that we currently use them for (and future applications)
 

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!

Members online

Top