What's new

MU MIMO VS SU MIMO

  • 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!

ulaganath

Very Senior Member
Its been clear that MU MIMO Cleary wins hand on in theory . But in practical it needs a MU MIMO client.

I dont understand what does the air time fairness of AC1900 Nighhawk does.

Say a R7000 connected with 2 AC devices and 3 N devices can only transmit say less than 40MB/s effectively at a time or /s

I dont think so. I do agree the bandwidth is highly getting affected if n devices connected which is now corrected with air time fairness for low connectivity devices when the AC devices are in the network which is given hight priority so that less priority low power devices get the applicable bandwidth later once AC client finishes the Job.

So Does Linksys e8500 and Netgar 7500 really makes the benefit out of it.
 
Air time fairness has nothing to do with multiuser MIMO. All it does is a better job of scheduling uplink and downlink availability for each client based on the speed of the client. Ideally this would be a tweakable, but I don't know of any WAPs where you can adjust airtime fairness (in a CONSUMER WAP).

It helps alleviate situations where, for example, there are 10 clients hammering a WAP, 1 is getting 10Mbps, 1 8Mbps and the other 8 are only getting 3Mbps each. If all are the same relative performance, it should ideally break it up more evenly so that all are getting roughly 5.3Mbps.

It is kind of like a bastardized TDMA for OFDM/CA. Though not nearly as good as TDMA.

MU:MIMO on the other hand you can have multiple clients receiving and in some schemes transmitting, all at the same time. Generally the limit is the number of radio chains - 1 on the WAP. So if you have a 4:4 router, it would be limited to 3 separate clients of 1:1 each at one time. Or you could have a 2:2 and a 1:1 client. Or you can have a single 3:3 or 4:4 client.

If/when MU:MIMO really gets off the ground, I think we will eventually start seeing actual 6:6 or even 8:8 WAPs, even in the consumer market place. I suspect it'll take a few years before we are there. 802.11ac is, what? Roughly 2 years old at this point or a little older. MU:MIMO capable routers have only really been shipping in the last 6 months and firmware for them that allows MU:MIMO has only been available for a couple of months basically. As for MU:MIMO clients, they are still "coming soon", though I suspect you'll start seeing waves of them over the next few months. Certainly by early next year (I'd be surprised if the new iPhone this fall doesn't have a Broadcom MU:MIMO chipset in it).

It isn't necessarily a panacea, but it definitely helps in highly crowded situations.
 
No, you lose a stream by enabling MU:MIMO, a 4:4 will support 3 1:1 clients.

Thanks for correcting the correction. :)

I was remembering it wrong.
 
No worries. If I counted the number of times I didn't remember something right, my wife might have some REAL ammo to use against me (instead of all these trumped up charges :))
 
OK are we considering only WAN TO LAN or LAN TO LAN. Say I have 3 1:1 ac clients and 5 n clients with mixed 1:1 and 2:2 with these clients how SU MI MO is different from MU MI MO. Say if I am transferring 1gb file across all ac devices and rest of all devices using Internet of 10MB line. With turbo transfer I am currently reaching 25MB/s min 14MB/S . So I could not test all 3 at a time but what would be the difference will I get 75MB with MU MI MO.

Sent from my ASUS_Z00AD using Tapatalk
 
A mixed client scenario is going to get complicated, because the SU:MIMO clients aren't going to able to use MU abilities of the router, limited the MU clients as well. It probably depends HEAVILY on the firmware of the router as well as the hardware ability.

In your scenario, I would GUESS with a 4:4 MU:MIMO router what it would do is it would break up the airtime so that it is doing a 6 chunk time block. So each n client would get a time block and the MU clients would get 1 time block, because it can transmit to all 3 MU clients concurrently. As for how fairly each time block will be structured, no idea. You would likely get slightly better overall throughput because of the MU abilities of the router so long as you have at least 2 MU capable clients on the wireless network that are active compared to only SU clients.

As for your performance question, I am not really sure what you are asking as no one here could tell you until MU:MIMO clients are actually on the market and can test with MU:MIMO base stations. However, if implemented correctly, so long as you have multiple MU clients active, even with SU clients, you SHOULD be getting more bandwidth for each client.
 
OK the current ac router even 3.2gb one may not be able to hope up for future needs. Could also depends on how much we utilize. As gigabyte Internet slowly growing. It doesn't matter whether you transfer locally or through net. So with evolving ac clients can these routers handle the maximum throughput even mu mimo will run out of bandwidth if say 5 2:2 clients connected and actively transferring files say 1 to 2 gigs as more and more getting connect it highly unlikely to deliver maximum throughput each device can handle. More over I was wondering is it that practically only 50% of wifi bandwidth any client can reach.

Sent from my ASUS_Z00AD using Tapatalk
 
I didnt say loose a stream. Its practically previous gen ac wireless till wave 1 all the clients and router works with SU MIMO. Being MU MIMO. at a given time it can serve 3-4 clients based on type ac or N. Even if so i dont think so it is practically feasible to server all at once managing Internet and Locally streaming/backup content.

Its theory and not practical yet.

Recently Netgear launched a revolutionized Router with Active antenna technology which has the booster chip in the tip of the antenna. Not sure if this design really makes better but it was intended to work effectively covering huge area.

Even though we say Beamforming its highly unlikely for 5Ghz to reach farther distance as its is the only frequency with more than 1gb/s bandwidth with reaching practically 600Mb/s when taking 2.4Ghz max 600Mb/s or even 1Gb/s in wave 2 the practical is nearly half or even less

which will rule out 1Gb/s connectivity for wireless clients cant reach or utilize the maximum speed even if no client connected to router. It may not be highly usuable in many occasions but if needed one cant take advantage of it thats weired paying for high speed and not able to enjoy the same.

Hope the new wave 2 set a new record in wireless transfer. I do like the tri band but do we note that Intel ED standard which have actually gigabyte connectivity works under similar 5G frequency and its has many applications.

Smart router are being expected to handle atleast 15-30 devices on average household with Internet speed ranging from 10Mb/s to 1Gb/s

So its keeps evolving as more portable personal computing devices every one holds the demand of internet wirelessly is keeps increasing.

Let hope for better stable evolving technologies which completely remove the word of dead spot or slow speed issues irrespective of Internet or Local content transfer.
 
Can you explain why you would lose a stream on a 4:4 client? I was under the impression that a 4:4 would still serve four clients (1:1) simultaneously.

It is a function of how MU:MIMO works, it is always n-1 streams in MU:MIMO mode. This is because one of the streams is lost in the beamforming that is conducted with HOW multiuser beamforming is done.

So a 4:4 router can serve 3 1:1 clients at the exact same time, or 2 1:1 and a 2:2. Or you can serve a single 3:3 or 4:4 client at one time.

Every single companies MU:MIMO chipset I have seen so far is 4:4 and every single description lists "three 1:1 clients simultaneously". It is related to how the beamforming is performed for MU:MIMO, but I don't know the exact technical details of why that is the case (I assume because of the constructive and destructive interference patterns needed to be formed, it effectively nullifies there being a separate spatial stream, with that 4th spatial stream being relegated to aiding the other 3 spatial streams to beamform them exclusively. I believe this is part of why non-MU:MIMO beamforming in Wifi is rather modest, because they are leveraging it in such a way you only get modest gains, because they want you to still get all of the streams, if you reduce the streams carrying data, you can get higher signal gains. Significantly higher in some cases).
 
Keeping in mind that MU-MIMO is not about speed, it is about Capacity, which is a challenge in and of itself..

Trying to put this at a high level - current MU-MIMO enabled AP's are generally 4*4:4, not a hard requirement for MU-MIMO, but it is what it is... mainly due to product strategies, and ensuring that AC1900 class clients are not impacted due to MU-MIMO...

In current implementations - one stream is used for sounding/ranging - where the clients are in the time and space domains, so that stream isn't used during a MU-MIMO transmission by the AP...

The upside perhaps is that the current (and soon to be launched) MU platforms are generally 4 radios per channel and 4 streams in SU mode - so in the analog domain, the extra radio does help all clients, esp. on the uplink side from client to AP, and in the digitial realm, where a 4-stream baseband will also help all clients thru better pre-coding of the MIMO streams - beamforming or omni...

The unspoken benefit for MU AP's is that because of the requirements for MU, the chipset generally will have significantly more processing horsepower, and client scheduling can also be a benefit - but this is largely implementation - the QC-Atheros chipset is a very good example here.
 
I dont understand what does the air time fairness of AC1900 Nighhawk does.

Air time fairness is about ensuring that slower clients don't hog all the air time.

Consider the situation where attached to a wireless router/WAP (Wireless Access Point) is a mixture of clients:
1 x 802.11a client - but because of walls/distance etc. is only getting a slow data rate of 2 Mbps
5 x 802.11n clients - getting a mixture of speeds between 80-150 Mbps​

If all these clients are say downloading files (lets say large Microsoft updates) the slow 802.11a 2 Mbps client will use up by far and away the most airtime to move the least amount of data.

Air time fairness will slow down further those slow clients thereby freeing up airtime for the faster clients.

It's doesn't at all try and "even out the Mbps" but rather, even out the amount of air time/radio time each client gets.

Because the faster clients are running a much higher data, f these clients get a "fair" chunk of airtime, then, they will have a good wireless experience. The very slow client will go even slower. In this case the needs of many out weigh the needs of the one....

It's generally good option to turn on.

I hope that helps.
 
Keeping in mind that MU-MIMO is not about speed, it is about Capacity, which is a challenge in and of itself..

Yes!

Quite correct. In fact, in some cases SU-MIMO will out perform MU-MIMO.

Why?
Because you cannot do TxBF (Transmit Beam Forming) and MIMO spatial multiplexing at the same time! MU-MIMO uses TxBF

Example:
The WAP is 4x4:4:3
(2 Radios, each radio has 4 antenna (that's the 4x4 bit), it can send 4 simultaneous streams and the last 3 is send 3 streams when doing MU-MIMO)

If the clients were two 3 stream capable laptops (Mabook Pro), and assuming a data rate of 150 Mpbs, then:
SU-MIMO could send 1 x 3 stream at 450 Mbps to client one followed by 1 x 3 stream 450 Mbps to client two
MU-MIMO may only send 1 x 1 stream packet to both clients at the same time (150 Mbps to 2 clients at the same time is overall 300 Mbps delivered)​
 
Last edited:
There is also the trade-off that needs to be considered on even a 1-stream client...

Let's say we have the MU-capable AP, so it's a 4*4:4 (in SU mode), it's always a 4 radio device - so a single stream client will get 4 copies of the same stream... which helps in medium/long range, where erasures start to dominate the link... in MU, it gets just a single stream (as the other streams are dedicated to other MU clients in the frame set), so the probability can be lower to get a good frame - and if that frame is lost, the AP now needs to consider, do we retransmit the MU frame again, or just sent an SU frame to the client that missed the previous MU frame...
 
Because you cannot do TxBF (Transmit Beam Forming) and MIMO spatial multiplexing at the same time! MU-MIMO uses TxBF

in 11n, this is accurate - either TxBF or SM - 11ac changed things to support MU (as MU requires SM, and each stream dedicated to a member of that MU frame set) - upside here is that because of this, TxBF in 11ac can use multiple streams...
 
Considering most if not all current 1x1 MU-MIMO capable client devices like those with the Snapdragon 801 chipset don't actually have it enabled (as far as I know) and newer client devices that are coming out are mostly 2x2, by the time MU-MIMO clients are widespread a 4x4 router probably wont cut it as was previously mentioned they only support 3 1x1 streams, which means with multilple 2x2 client devices you wont see much if any impact with MU-MIMO, till the expected 8x8 routers come along. For example I think even the coming Intel 9260ac is a 3x3 device if I am not mistaken and most new phones and tablets on all the major platforms are 2x2.
 
Last edited:
in MU, it gets just a single stream (as the other streams are dedicated to other MU clients in the frame set), so the probability can be lower to get a good frame...

I'm not so sure, because with TxBF even with MU-MIMO you are still getting the antenna gain of 1-4 dBm - those 2 streams sent out to 2 clients comes simultaneously come out (in a 4x4:4:3 radio from) THREE antenna. To be clear - 2 streams are sent out out of all THREE antenna simultaneously using pretty complicated TxBF maths, so I expect the client will be able to decode the signal nearly as well as SU-MIMO.

Therefore MU-MIMO with TxBF will have an additive signal gain for the client.
 
in 11n, this is accurate - either TxBF or SM - 11ac changed things to support MU (as MU requires SM, and each stream dedicated to a member of that MU frame set) - upside here is that because of this, TxBF in 11ac can use multiple streams...

Yes - you're right!
MU-MIMO will be able with a 4x4:4:3 config to deliver successfully a 2 stream and 1 stream simultaneous packet to a 2x2 and 1x1 client at the same time.
 

Latest threads

Sign Up For SNBForums Daily Digest

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