Mimosa MU-MIMO ?

[ste]

The rocket prism adds rx and tx filtering. This helps with colocating aps. Reducing co channel noise is a big deal on loaded towers. And they add sync. Putting this to rf elements simpers you might increase tower bandwidth a lot. This airmagic stuff is a good idea, too. If they get things done tower performance increases a lot compared to available 11n gear.
And ... they are cheap. So it does not hurt to replace them once there is something better.

Let's see how mimosa performs once we can buy it and tdma and sync is implemented. I guess with plain 802.11ac it is hard to scale. You have to go back to rts/cts to reduce hidden node problems.

I like both. Let the games start ...

[sirhc]

The rocket prism adds rx and tx filtering. This helps with colocating aps. Reducing co channel noise is a big deal on loaded towers.

I am not having problems with co-location noise on towers with the use of RF Armor so I fail to see where this will improve much?

I am curious to test an airPRISIM sector against an AC Lite sector using shield kits, I am not expecting much difference in performance. - We shall see.

But as you said the airMAGIC thing look neat.

[rebelwireless]

I am curious to test an airPRISIM sector against an AC Lite sector using shield kits, I am not expecting much difference in performance. - We shall see.

But as you said the airMAGIC thing look neat.

WITHOUT shielding, airPRISM *destroys* AC-lite. utterly. It does nothing for on-channel interference, but anything around the edges and beyond is very very clean. And for colocating, the output is just as clean (on PRISM models, not ptmp/ptp models). As soon as we have truly stable operation on the PRISM to M units, we will be swapping out a LOT of RocketM with 5-PRISM units.

Even if Mimosa does look interesting, we have a lot of sectors with <=30Mbps throughput and ~30 subs and taking them from 20Mhz to 10Mhz *AND* fixing the out-of-band, we will be golden. and the PRISM units will do that.

airMAGIC looks neat, but in it's current state, it's nearly useless. I can look at it and see nice, clean gaps while my channel is in some noise. Move the channel and when it comes back up, now my old channel looks awesome and the new one is noisy as can be. It's not getting accurate data from the CPE I guess, or some other issue is causing it to give results that can't be trusted. Hope they get it straightened out, could be an awesome feature.

[sirhc]

30 Mbps sector?

I get 70+ Mbps with Legacy M5 sectors with 20 MHz wide channels and 90+ Mbps on 30 MHz wide DFS channels

Anyone does not believe me I will arrange a call with TeamViewer to prove it as I have many times and posted I do not know how many posts and videos showing speedtest.net results from my house and RF Armor.

Matt Hoppes was here and saw 90 Mbps at a commercial customer on a DFS sector with 20+/- subs on it http://community.ubnt.com/t5/The-Lounge ... rue#M27586

Also UBNT had sent 2 guys out here a few years back and JT and Joey saw first hand as I took them around to a couple customers and even on an installation.

[rebelwireless]

30 Mbps sector?

I get 70+ Mbps with Legacy M5 sectors with 20 MHz wide channels and 90+ Mbps on 30 MHz wide DFS channels

Anyone does not believe me I will arrange a call with TeamViewer to prove it as I have many times and posted I do not know how many posts and videos showing speedtest.net results from my house and RF Armor.

Matt Hoppes was here and saw 90 Mbps at a commercial customer on a DFS sector with 20+/- subs on it http://community.ubnt.com/t5/The-Lounge ... rue#M27586

Also UBNT had sent 2 guys out here a few years back and JT and Joey saw first hand as I took them around to a couple customers and even on an installation.

We are more up against client count on RocketM5 than throughput. We are only pushing about 30Mbps on a lot of sectors because that's what these rural sectors happen you get. Can't get more than about 30 people on a RocketM5 though, else performance really tanks. We are selling 5Mbps plans on these as they are WAY out there.

Jumping up to a Rocket5AC-PRISM doubles (maybe tripples) the clients possible on a sector, and 10Mhz channels can do about 60Mbps.

[lligetfa]

I am curious to test an airPRISIM sector against an AC Lite sector using shield kits...

WITHOUT shielding, airPRISM *destroys* AC-lite...

Damn Ubiquiti for mixing the name of a technology with the name of a radio model. Are we talking now about the airPRISM filter technology, the R5AC-PTP and R5AC-PTMP models with the airPRISM, or the R5AC-PRISM model?

Or are we talking about their AP sector antenna?

[sirhc]

I am at RF Armor warehouse right now and this is not the best time to check speed as kids are home watching Netflix/HULU or porn but hey I ran it for giggles

I am on a Legacy M5 sector with RF Armor on sector and NanoBridge using a 20 MHz wide channel a few miles away with under 20 subs on this AP and not an ideal shot as I am shooting across our 300' + steel building only 5 feet above edge of roof and then through high tension wires.

My tower is Diller Ave water tank which is 5 hops from fiber

[jjonsson]

This is a Speedtest from one of my customers. He's on a Rocket M5 + AC 60 degree sector antenna and a PowerBeam M5-19 as client
40 MHz channel width (could be 30 MHz will yield same result)

Those customers I have on AC and 20 MHz channel width get speeds way over 120 Mbps...

I would be happy if UBNT upgraded so AC could make channel sizes in 5 MHz increments. 15 MHz would be perfect.

With N equipment I only load max 10 customers per AP.

Speedtest kunde.jpg (11.99 KiB) Viewed 22599 times

[sirhc]

This is me at home on a DFS sector 30 MHz wide 3.4 miles to Blue Ball Water Tower.

Rocket M5 Sectors with RF Armor and NanoBeam 400mm with shield kit

AP has 22 clients

I have seen as high as 90+ Mbps before

[rebelwireless]

at this point, its not about throughput, its about clients per sector. 10Mhz doesn't make a lot of sense on RocketM5 because the OOBE is so bad you waste most of the spectrum in between channels. This isn't so bad for just a satellite site, but when these sites are also the primary backhaul (and 'side'haul) path and I really need spectrum for those backhaul links, and DFS doesn't have the distance for these rural sites, it's easy to run up against a client count limitation on the RocketM5.

This is a fairly common complaint, and tons of people have asked for even an M5 radio with more CPU to handle more clients.

Nice clean edges solve a lot of issues. Can run channels right up against each other, allowing for more radios. More CPU allows more clients per radio. A Rocket-5AC-PRISM in 10Mhz can nearly match a RocketM5 in 20Mhz at 20 CPE. Hit 30 CPE and it is more than 2x as good, hit 50 and it's 3x+ as good (according to ClaudeSS, I don't have 50 CPE on one) because a RocketM5 can't really handle 50 subs.



The interesting thing about Mimosa is that they completely circumvent the gap between channels issue. Instead of 4x 20Mhz channels in their 'omni', its just a single channel with MU-MIMO and elements pointing in 4x ~90 degree increments. So you can do an 80Mhz channel and get directional isolation for various substreams as if you were running 4x 20Mhz channels, but also send more subchannels towards clients as needed.

I don't know how this handles specific interference in part of the band, but 802.11ac wave2 is supposed to be able to avoid noisy spectrum on any single or group of elements that have some period (time) isolation. Maybe there are some unannounced ubiquiti products based on wave2 ac chips that have multiple elements...idk...