MU-MIMO timeframe


#4

This MU-MIMO wil it work in GPS (tdma) mode only? Or also in ‘legacy’ (3rd party) mode?


#5

MU-MIMO only works with the WiFi Interop (CSMA) protocol. We found that the bigger benefit was in sharing spectrum among overlapping APs in micropop deployments so are focusing on doing that well. You need GPS-Sync (TDMA) to do that, and the T in TDMA means that every client gets a separate time slice that can change based on demand.


#6

hmm. Does this mean MU-MIMO is not soon to be expected?

It also means an A5 on its own in a 3rd party (not my controlled network) crowdes spectrum is not doing so much better then any other AP? (Apart maybe for the 4x2 network setup?)

I was hoping with MU-MIMO we would be able to serve more clients from a single AP against the same client speed limits… but this is still far away future?

Actually found some document stating MU-MIMO would outperform TDMA performance.
(https://mediatum.ub.tum.de/doc/621059/621059.pdf)

Since your GPS mode is also only possible in a full Mimosa (with several small Mimosa cells) setup to make better re-use of the spectrum it might be a better business model to develop MU-MIMO at higher pace.
Not only could it possibly perform better (and later still be embedded in the TDMA protocol as some documents makes me believe) but it would also make it easier for existing WISPs to transite from other brand to Mimosa.
The MU-MIMO would also work with 3rd party client devices (in transition or ultimate) where for the GPS option an all Mimosa deployment is needed…
Many WISPS will be held back in investment if they need to changeout their whole network where a start with the AP’s only would make things more easy.

Wouldn’t it help Mimosa penetrade the market faster?


#7

I think you need some more asterisks on your specifications page then. It comes off as pretty misleading.


#8

So this is a good as it gets? I can either have MU-MIMO in WiFi interop or TDMA without MU-MIMO? If so I feel misled and disappointed. I’ll wait to see but I am glad we did a very very small deployment and if a few CPE’s are all I need to swap out to put the testers back on our production system then it was a lesson learned.


#9

Hi Mario and Rudy,

There is definitely a lot more we plan to do with MU-MIMO. It is just that we have prioritized delivery of a high-quality GPS-Sync (TDMA) release later this quarter that will do the following:

  • Improve collocation with A5c sectors
  • Enable synchronization with adjacent micropops
  • Radically increase total geocapacity

It lays the groundwork for additional innovations, and provides a method to predictably scale network-wide capacity by making effective use of spectrum. This continues to be a primary goal for Mimosa.

Here’s where we stand with MU-MIMO…

We see a 10-20% improvement in CSMA with the antenna matrices working to find compatible transmission nulls dynamically. Larger capacity improvements (at a single AP) typically require more transmit antennas and processing power. So far, the only systems that have delivered significant results employ 64 or 128 antenna arrays to achieve the necessary nulling - a massive cost trade-off.

While we can simulate better results in labs with consistent traffic, real user traffic is less consistent (e.g. occasional text messages versus 4k video streaming). Inconsistent traffic reduces the practical capacity, and limits MU-MIMO systems from scaling linearly regardless of antenna count.

Rather than relying on just MU-MIMO to optimize a single AP, GPS-Sync (TMDA) solves the colocation problem AND provides some very promising opportunities for continued MU-MIMO development.

Here are some approaches we are pursuing:

  1. Deployments with non-overlapping antenna patterns present natural nulls where clients don’t overlap. Rather than introducing cost and complexity with larger antenna arrays, clients are naturally isolated with smaller, more-compact sectors. We have a bunch of patent work in this area in particular, and we have several planned improvements on the roadmap.

  2. The LTE community has seen similar cost/size/performance limitations of MU-MIMO. To augment client capacity scaling, they have pushed toward use of OFDMA, allowing coordination of OFDM tones to each client (spectral diversity rather than channel diversity). We intend to use similar techniques that will allow the AP to communicate with each client using a unique OFDM mask that the client can support, and simultaneously use remaining OFDM tones for other clients. This will translate to client capacity gains, and provide a significant interference management advancement in fixed wireless.


A5C why 4x connector
GPS Sync Firmware
#10

Hi Chris,

The TDMA/GPS feature is welcomed. Any idea how close this is to general release?

The 2 items you mention that you are pursuing - will these appear on the existing A5 roadmap, or is this a different product? And what sort of timescale are we talking - months, year(s)?

Thanks,
Richard


#11

Great ideas, but nothing to buy than a normal 4 chain AP doing normal 802.11ac.


#12

The tecnical explanations all sound very nice too me but the major part I’d like to hear is when? And that seems to be lacking.
I learned about Mimosa early 2014 and was enthousiast from day one about all the tecnical advances presented in the creation of this ultimate (for now) PtMP system available for the 5Ghz band. Ok, I was told “by the end of the year”.
That ment 2014… and more related to the hardware only…
Now this summer the PtMP in Europe finally came available and yeah, the same sheering story on the performance was demostrated in slick presentations and enthousiast stories of sales…
We bought probably one of the first A5’s in Europe to try which learned us that half of all the promises made are just that, promises…
I am also glad we only bought one A5 with three C5’s +G2 combo’s for try outs.
Many software issues, many features not working.

Like 99% of future users of Mimosa we are migrating from another platform (Mikrotik in our case) to pursue a higher quality service level with more speeds but so far the ‘interop mode’ only available is basically throwing us back in time…
We have 30 SXT’s that came from an 400Mhz cpu OmniTik 802.11an tdma system towards now an A5 high cpu 802.11an system.
We are in a congested spectrum and althoug indeed plain ‘legacy’ 802.11 would be much worse on the OmniTiks we now see hardly improvement over the tdma working OmniTiks.

Due the higher processing power and probably some better antenna/stream caracteristisks we can sustain higher combined througput on the AP (we went from 40-45 mb max to 60-70Mb on tests running from several clients silmultanously) and special single stream traffic (important for streaming video) was doubled in capacity.

But we now compare an Mimosa AP with 10 times the cost to an OmniTik!
I`ll bet if we’d fit a 720Mhz Netmetal on an 12dBi dual polarized Omni antenna the difference is gone…

But that is not what we’d hoped for. We’d hope the A5 with this truck load of new tecnologies would blow away the competition in capacity both as stability and latency… what a deception for now…

So yes, what is the time line?
We’d understand things might be coming, but soon? And secondly, for this GPS sync’d deployment we are forced to first buy a whole network of new C5’s. This is going to be a major investment for us and that without any guarantee we will see improvements…

If there is no reasonable foresight I think we are going to fit a Netmetal and put the A5 on standby for a while awaiting those other promises
If the real thing doesn’t arrive within half a year it might well be the unit ends up on e-bay…

We were realy hoping to give a show-off with Mimosa in Europe. But this might well end up in a show down if things don’t move soon…


#13

Hi @Chris,

We’re getting close to the end of the quarter. Any news on the GPS/TDMA?

Richard


#14

Santa is coming soon. Will he be bringin the sync package to us? He’d better hurry up… 10 days to go…
:innocent:


#15

“The last reply to this topic was 224 days ago.”

Any updates?


#16

TDMA GPS-Sync (SRS) was delivered and is available now. As soon SRS was released, we were flooded with requests for other features ahead of MU-MIMO. We tend to prioritize development based on user demand, and we’ve been told by many WISPs that there are more immediate needs - like automated client provisioning.

@Andrewski, Can you tell us why MU-MIMO is your biggest need? We’d like to understand the priority from your perspective.


#17

Higher client counts on an AP? And more throughput simultaneously to several stations in parallel?

Apart from SRS another way of saving spectrum is more stations per AP. In stead of 50 stations per ap 100 per AP could save spectrum. Compared to MIMO the AP in MU-MIMO can serve twice as much stations with the same throughput while not needing more spectrum…


#18

Hey Chris,

@Chris MU-MIMO isn’t my biggest need per se. My biggest concern is the fact that SRS was promised to be rock solid when it was released. Then within minutes of releasing it, traffic shaping for download speed was completely broken. There have been a handful of firmware releases (I chat with support frequently) and none of them have ever indicated that they know what is being tested and when various bugs are going to be fixed. I built a new company using pure Mimosa based on the fact that Mimosa claimed it could offer 200-300Mbps to the end user(see YouTubeVideo). I bought a GB fiber pipe, hooked up an A5 and gave it a go. The reality is that I have a hard time delivering 50Mbps to my subscribers. I was told that SRS was going to be immune to other 5G Wifi devices and yet my network is still struggling. I was forced to stop offering my 100Mbps plan because I simply cannot do it consistently. PER was supposed to be lower on SRS but in the GUI I’ve got clients in the red constantly during peak usage. It is frustrating being blind to what are known issues/bugs and what is being fixed in the next firmware and how the beta testings are going.


#19

Hi @Andrewski,

The differences in throughput between WiFi Interop and SRS are because of the 50/50 traffic split and protocol overhead. We’re working on some additional controls for those, but there is no real difference in the traffic shaping between protocols except for the built-in airtime fairness in SRS.

It looks like the PER on your A5-network is between 1-2% which is far less than WiFi Interop at 10-12%. The PHY rates of your clients range from 100-400 Mbps on a 40 MHz channel. Client signal strengths range from -70 to -44 dBm. The clients with lower signal strength are operating on a single chain, so would have lower throughput, and would cause AGC to use a much lower threshold value. Other than improving the signal strength of the weaker clients and moving to an 80 MHz channel, you may want to go back to WiFi Interop for throughput reasons if there are no other Mimosa Access Points nearby with which to sync.

Our Product Management group manages the beta program, so the Support team doesn’t always have the latest updates to share. If you are interested in joining the beta program, I can put you touch with the right folks. We have dramatically improved the traffic shaping algorithm in the latest 2.3.3 beta. It works really well now. The beta also contains AGC improvements, channel selection in 5 MHz increments in WiFi Interop, DFS support for C5c and more.


#20

Hi Chris, can you add me as well to use that beta firmware? Many thanks.


#21

Hi @J.Enrique

Yes, will do.


#22

@Chris Is there a way to follow the Product Management Group in a forum or something that I am missing? Thank you.


#23

Hi @Andrewski,

There is not one specific Product Marketing area within the Mimosa Community, but we do occasionally discuss upcoming firmware features within public threads. Some of the discussion is limited to closed beta groups in early beta phases, and we share it more broadly as the new firmware features stabilize.

Feature requests are usually directed to related threads under each sections for each product family. For example, A5 feature requests are concentrated in this thread, but there are certainly others about specific topics: Feature Requests - A5 Products

If you want to provide specific feedback or ask related questions here, then please go ahead. You could also start a new thread about a different topic if desired.