A5 maximum throughtput

What is the maximum throughtput on the A5 with 20MHz channel with?
What is the maximum per CPE?
The following calculations shows 156Mbps aggregated, is it right?

Mac Tx = PHY_Tx rate * (traffic_split /100) * (TDMA_efficiency / 100)
Mac_Rx = PHY Rx_rate * (traffic_split /100) * (TDMA_efficiency /100)
TDMA efficiency is based on 90% at 8ms, 80% at 4ms and 70% at 2ms window.

MAC Tx=(86.7x2)x0.5x0.9=78.03Mbps
MAC Rx=(86.7x2)x0.5x0.9=78.03Mbps
Total=156Mbps aggregated

The MU-MIMO will duplicate this capacity?
When will be availble MU-MIMO?

So your math is a little goofed up, but just because A5’s in SRS use 10 ms windows. (I have no idea what the efficiency is at 10 ms) I would bet that you don’t need SRS to get the benefits of MU-MIMO

In my experience, a 20 MHz channel can do ~65 Mbps down, if there are not other issues to kill the connection. (interference, bad connections, etc.). Aggregate, I would assume to be double that, but no one uses upload so I don’t have any hard numbers beyond having seen ~100 Mbps aggregate one time, I think. Admittedly, I am running my clients a lot further away from the AP then most and some of them don’t have the best of connections, so others may see higher numbers especially if they have customers who actually use their upload Bandwidth…

As far as MU-MIMO doubling capacity, I have no clue, I think that doubling would have to be only in certain situations. We do see the benefits in town when dealing with noise already, but that is not exactly what you are asking about.

Hi William,
Thank you for your comments.
It would be great if someone of Mimosa confirmed the calculations, informed on the expeted date to be released MU-MIMO and how it will be implemented.

I wouldn’t get my hopes too high up.

I don’t know where you got your TDMA efficiency numbers, but they seem suspiciously round to me to be accurate. Though, they do reflect the reality of TDMA Efficiency. Somewhere in the manual there is a markup of Mimosa’s TDMA speeds in relation to window sizes, but I can’t find it ATM.

You might try going to Mimosa Sales and ask them if they have any information they can share. But, while Mimosa seems to be pretty open about what they have available, they are pretty tight lipped about what is coming next…

The highest obtainable PHY rate in ‘ac’ on a dual stream 20Mhz is 174Mbps.
That means at best 55% makes 105Mbps tcp traffic if no package loss takes place.
This is the maximum data stream that can be transported of the medium in 802.11 ac dual stream in one direction. Meaning with UP/Down traffic spit of 50% maximum speed to one client will be some 50Mbps.

If now two clients are associated, even if one is not sending/receiving date the tdma (SRS) then time slot has to be reserved for this 2nd client and available airtime for client 1 goes down. 3rd client the same etc.
On a 20 client network I wouldn’t expect download speeds to become above 45Mbps even if only one client is receiving.
This is maximum achievable throughput. With other extra setting the link can become more stable but not higher capacity. The only way to improve download speeds at the cost of upload speeds is to set different ratio. (30/70?)

The AP itself could deal with much higher data rated. Set it to 80Mhz channel and suddenly it can have an aggregated throughput of some 350-400Mbps…

MU-MIMO would be the AP can actually ‘talk’ or ‘listen’ to 2 devices at the same time. Meaning (if all wireless data are OK) the aggregated throughput over the AP would double since the airrate between each device and AP is the limiting factor. 2 x 45Mbps make 90Mpbs.

How far this will go in a 80Mhz wide channel setup I don’t know. In ideal situations the air rates could transport 866,7Mbps x 55% x 0,5 = 238 x 2 devices = 477Mbps theoratically. I think the A5 should handle this since the specs says up to 1Gb of real data througput…

Hi all,

At 80MHz, you should see around 700Mbps aggregate
At 40MHz, you should see around 350Mbps aggregate
At 20MHz, you should see around 175Mbps aggregate

Let me give you some real world examples from my own, live, real customer network and from networks I’ve help to deploy over the past two years:

On a 20MHz channel, I can reliably do 65/65 for 11 clients on an AP but have seen 75/75 plenty of times. Depends on traffic

On a 40MHz channel, I can do around 170/170 on an AP with several clients.

On an 80MHz channel, I have seen clients doing 350/325 reliably.

Like Rudy said though, you’ll lose throughput for each additional client you add. I have an AP that has 31 clients, on an 80MHz channel, that reliably sees 150/100 for clients when doing a speed test. Now if everyone was running a speed test at the same time, it certainly wouldn’t look great at all.

Also here’s a chart with PHY rates per Modulation: http://ap.help.mimosa.co/access-point-faq-snr-mcs

With Mimosa being acquired by Airspan, all of our timetables have been and are still being reworked. I certainly don’t have any ETA on when true MU-MIMO will be available.


Hate to jump in late here, but have been researching SRS speeds, and in my lab if I enable SRS - I see about 1/2 or 3/4 of the throughput.

I only have 2 clients connected to my lab - but the results are pretty clear.

I guess I need to know the “break even” number of clients in SRS? IE - is there one?

With SRS off I get say ‘50 megs down’, 1 client at a time testing - With SRS on I get about '25 to 28 Megs down’. PER rates lower than 10% etc…
Through G-2 Seeing like '17 Meg’…

Per the chart that Dustin sent (Super helpful) I should have at 26 or 27 SNR - well lets call it 25:

“8 256-QAM 3/4 25 86.7 180 390”

		2 streams at 390 - that should be 780Megs - 

Just unclear why in SRS mode it is so much less.

Does one need 10 clients say in SRS to see it be helpful? Is there a magic number?

I do plan on doing micro pops - so not too much distance - I like Dustins numbers there - but concerned with my lab tests.

Also, if I am doing more long distance, and I kinda plan on that with one AP - should I NOT do SRS or does it matter?

Thanks for the answers.

When we were running non-SRS we ran into issues frequently because we did not have enough spacing between our sectors. Also, we had customers pretty far out and they sometimes had pretty weak connections (SNR < 20 dB)

When we switched to SRS we have not looked back, but I see lots of people who have claimed much higher throughput numbers in Interop mode then SRS. (probably because most traffic is focused in one direction and SRS is stuck at 50/50)

I would used entirely Mimosa clients and see what works for you, with the C5x you have a pretty competitively priced CPE that can get upto 25 dB of gain and is an entirely metal package. If you need more Gain you could look at the C5c and honestly most of the cost for high gain CPE equipment is stuck in the antenna. If you used entirely Mimosa equipment you can switch between SRS and Interop.

1 Like

Thanks William5 - the key thing for me that I was not taking in to consideration was the 50/50 SRS mode traffic split.

This makes sense, and explains why I would see a difference - We do plan on offering people a fair amount of upload, so all good info. Once I have a larger bandwidth pipe, the 50/50 split should matter less, as I wont be offering a speed profile that would use the whole pipe.

I was mainly trying to decide if we wanted to start with SRS or transition to it - but now from your info, I will start off with SRS and see what we can offer people speed wise.


1 Like