best MIMO setup with NV2 and TDMA

Hardware MikroTik hardware general
1

Dear All,
since the internet speed has been scaled up 10times in our country, we would like to migrate to a new
wisp system, taking advantage of the mimo, nv2 and tdma!

the areas that we and our clients are implementing is very congested so it is preferable to implement 19dbi 90deg
dual pols sectors.
it will be for approx 5 km not radius. (the land topology is to have directional not omnidirectional)

please note that we are competing with ubnt which offers the following;
http://www.ubnt.com/basestation
also ARCwireless has the new arc flex wich offers the same…
http://www.antennas.com/?page_id=18

since we are all used to mikrotik and we are the distributors,
it would be very good to have and offer a Mikrotik full solution and compete with the air-max and arc-flex
the needed throughput to the clients is up to 10mbits!

our questions will be :
what should be the basestations that replaces there’s enabling the clients to use sxts?..
if we have any rb… with r52nm with dual pol sector,will it serve the sxts with nv2 and tdma??
or it will be preferable to use omnitik along with the sxts … ???

thank you in advance

2

We are using Airmax 5GHz basestation and mikrotik rb411ah+R52Hn to make access point and RB sxt for the clients. We are very satisfied of this MIMO 2x2 solution and TDMA NV2 works fine if you have enough frequency space to separate access points and backhauling on the same tower. To build access point use metal box and shields pigtails to minimize interference.

Giuseppe

3

Well, I am playing with it but so far not really a success.
I use one 120º Airmax 19dBi base antenna and have it run of a rb800 with R52Hn mimo card and 6 mtr LMR-400 cable.
From same tower and same rb800 I have another R52Hn mimo card but working in legacy ‘a’ mode and both chains connected to a 90º 17dBi Elboxrf RSLL antenna (very good stuff!) with both 1,5mtr LMR-400.

The dual Elboxrf config works like a charm. I have 20 clients in a 160º range working on this AP from 1km to 6km distance.

Same clients that can ´see´ the Airmax antenna are receiving its signal with 12-16dBm less than they ´see´ the 17dBi RSLL antenna! Even if I take CPE’s almost dead in front of the Airmax sector!
The difference in cable (5 meters!) is not explaining the difference.
(Both Airmax and Elboxrf AP-antenna’s are 4 meters vertical separated on a completely free tower on the edge of a 50mtr hill and working on different frequencies to eliminate interference.)

From the CPE’s I can now easy test the difference in performance between both setups.
I do ´test´ the Airmax antenna by setting the connected card to ‘a’ mode as well and only enable the V-pol so the clients of the legacy AP can ´see´ this one. (To check I also test with both or only H-pol enabled, just to see if all is done right. No better signals, only worse. Off course…)

Because I am puzzled where such a big difference in power could come from I started to investigate more in the antenna’s.

The Elxboxrf RSLL antenna’s are very well documented and their 3dB drop-off is indeed for their 90º sector and they have a 7,8º elevated beam.

The Airmax 120º is not that good documented. Only some vague hard to read diagrams on the manufacturers website.
But by studying the documentation it works out that these antenna’s are sold with overrated specs.
For their 120º degree working sector they use actually a 6dB drop off! That is 1/4 of the maximum.
Using a 3dB dropp-off the 120º antenna is actually a 90º antenna. (Same counts for all other models.)
Any serious manufacturer uses the 3dB so this is cleary a sales trick to make these Airmaxes look good!

Together with the very small vertical beam of only 3º the Airmax antenna would be more classified as an directional imho!
From a high tower (mountain) the small footprint makes it very limited as base station sector.
(Probably the antenna would work better if 'overshooting´ flat landscape from a relative low tower. This is probably their ´showcase´!.)

Only because the Airmax Rocket’s uses 28dBm (!) dual chain (25dBm each radio) power without any power losses (due short cable) bound to the 19dBi antenna plus the ´n´ protocol it makes them usable. Combined with very high powered (25dBm radio and 25dBi antenna!) nanobridges (recommended usage) they make an concept that is workable.
In theory! I can’t get it to work half as good as my alternative…


I’ll bet with other better, antenna and same power from card and same short cable with MT stuff you can make a much better setup in using MT-ROS and hardware.
The problem is only that on this moment there are not yet a lot of good looking, easy fitting high gain duo polarized sector antenna’s on the market.

Conclusion:
To beat the marketing-tuned product of UBNT some manufacturer needs to come up with better antenna than the Airmax’s.

4

thank you server8,
how many clients are there connected on one sector ?
and what is the actual bandwith for each cpe that your getting?!
what is the relative distance for cpes?
have youtried 711ah-n-m for alternative of 411ah+r52hn?
have you tried omnitik as a replacement ?!
theorically it will replace the base station with a lot lower cost( rather than having sector+rb+wcard+encl+pigtail)!..

5

Well,
A big thank you WirelessRudy for the clear description…
what is the actual throughput your getting between the Elxboxrf RSLL antennas and the clients cpes
(ranging from1 to 6km)?
how many concurrent cpes will this basestation sector handles !?

On the other hand, we are looking forward to beat the “marketing-tuned product of ubnt” using mikrotik;
this is why we originated this post.

the sales guys @ mikrotik, introduced the omnitik as an alternative!
(a very cheap cost to cost one…) and they said it will cover easily a 5km radius…
have you tried omnitik!? do you think it would be a good choice to deploy with sxts!?

thank you,

6

Are you actually allowed to run with those power levels? Your talking about 50w’s EIRP and most of the EU allows 1w tops in some of the 5ghz bands and as little as 250mw in others

7

Yes we are and we are not allowed!(for the time being there is no spectrum regulations back here…)
what about omnitik do u think it would be an alternative ?

8

we plan to set 50/60 clients per antenna actually less 20

we shape clients at 10 mb/s

2 max 4 km with -80 db signal level

no but i am quite sure that works fine too

We are testing but I don’t like omni antenna + nv2 in large enviroment too many interference in every direction :smiley:

Giuseppe

9

No. I am not. As I am not to build illegal towers, not to use sector or directional antenna’s, not to use 5,8Ghz without license, not to use 500kW (!) TV transmission systems on 2,4Khz. etc etc.

I am in Spain. Call out the word “illegal” and half the country disappears…
If the wifi industry in Spain was to stay 100% within the regulations the industry would not exist. Even the ´big´ operators do what they want.
I am trying to stay within the regulations as much as possible (like using good gain antenna’s for perfect listening and not using too much power on the transmissions) and yet am I day by day faced with competition that is doing nothing else than acting like the Daltons.
The only reason I survive is service.

10

I don’t bother about the actual maximum throughput. I use 10Mhz wide channel and set conn. rates fixed to levels double to what each client is allowed to produce in one direction. This way I can give good and stable service to my clients (4 to 6Mb allowed with an occasional 9Mb client) with high CCQ so even Voip works like a charm.
Maximum amount of clients I`d figure an rb800 can handle this way is about 40 to maybe even 50.
I have one rb800 AP serving now 35 clients and the rb on average runs at 20-25% cpu during peak hours.
Off course, if all clients would demand their full capacity at the same moment the router probably wouldn’t be able to handle that. But who builds a network for the unlikely event everybody want the max at the same time? That would be a very unrealistic person spending way too much money and resources than actually needed to run a good business.

Regarding the omnitik? Haven’t felt the need to buy them. They look a bit cheap to me but that can be only a impression from the pictures. In general I use sectors.

11

forgot to mention the SXT’s.

I bought some 30 of these, and am still buying them. For short distance (< 2km im my case) and in an urbanised area where people don’t actually want to see their antenna they do fine.
I tried some in the open field at 4km distance and the signal was very poor. I use now the old RIC units fm Mt and they get almost 10dB more signal from same AP plus the fact they have good metal back for shield.
SXT is all plastic which I found to give problems in heavy interference areas.

I think the SXT is product designed to given an answer to the success story of ubnt and their small airmax units.
But in both cases they are not the best quality products.
The SXT also had the issue of mounting them horizontal for V-pol use in migrating (legacy to ´n´) networks which was another downer.

The Groove’s on the other hand are a top product. I have now some 8 in use (yeah not a lot yet) but they our perform everything else I have. Basically because I can use them without cable directly attached to quality antenna.

12

Regarding the omnitik? Haven’t felt the need to buy them. They look a bit cheap to me but that can be only a impression from the pictures. In general I use sectors

we agree to the cheap (GADGET) looking of omnitiks, but yes it seems only an impression :wink:
we Lab test them yesterday, and they showed up very good results, coupled with the sxts; reaching 240mbits.

the big challenge is to bring them outdoor and test in a congested environment and see how they will react.
this is our mission job for today,(will post some results tonight) …

The SXT also had the issue of mounting them horizontal for V-pol use in migrating (legacy to ´n´) networks which was another downer.

in elevated area the mounting issue becomes a real problem, (needs @least 15deg tilt up and down)
not forgetting that the sxts have a weekness in wheatherproofness,
also the plastic with no reflector in heavy interference areas.
so YES we need a solution…
we think kind of like metal shield coupled with tilt up/down support will do the job.
ex: http://cn24.net/SXT-Shield-50.jpg

The Groove’s on the other hand are a top product. I have now some 8 in use (yeah not a lot yet) but they our perform everything else I have. Basically because I can use them without cable directly attached to quality antenna.

yes the grooves are perfect in outdoor as they are weather proofs unlike the sxts,
but the only thing is that they are not MIMO so not suitable in our case…

since internet price where brought down and speeds scaled up 8times…
we are planning to migrate our existing wisp serving more than 1000 cpes to MIMO

what do u think of omnitik mesh, each 1km radius along with sxts.

13

awaiting your experiences with anxiety.

in elevated area the mounting issue becomes a real problem, (needs @least 15deg tilt up and down)

forgot to mention that. A real pain indeed.

also the plastic with no reflector in heavy interference areas.
so YES we need a solution…
we think kind of like metal shield coupled with tilt up/down support will do the job.
ex: > http://cn24.net/SXT-Shield-50.jpg

I’ve seen that picture before. Although it probably does do the job it looks awful. I don’t think many client will appreciate it if they see me coming up with such a metal structure.
Here was the only big pro of the SXT. They are small and reasonable good looking. If some manufacturer would just make an alu shield in the same form as the SXT that would slide over it from the back with an opening for the cable and a new wall/mast fitting with tilt and 90º turn option with only a small ´horn feed look’ at the front than this would make the SXT into a much bigger success I suppose.
Actually I hope some manufacturer would read this and produce some.
If MT than also produces same SXT but a bit bigger to house some bigger gain antenna (22dBi, like their RIC’s had) we would come close to an almost perfect solution! :smiley:

yes the grooves are perfect in outdoor as they are weather proofs unlike the sxts,
but the only thing is that they are not MIMO so not suitable in our case…

Well, they are mimo, but only one chain. You can work with the mimo protocols and get considerably more throughput than in legacy ‘a’ mode. But only half of full dual chain mimo off course. 150Mb theoretical max. througput is much higher then the 54Mbps of legacy ‘a’. This would ‘suit’ most customers for the years to come?
On the other hand, indeed I already wished they would come in a dual chain version for use in full mimo setups. But I presume that we are talking a complete new unit now…
By the way, to reduce any interference possibility on these grooves we wrap a sheet of metal tape around the top end where the radio is located after installation. For these units a nice looking metal ´jacket´ would also be an useful add-on.

what do u think of omnitik mesh, each 1km radius along with sxts.

I have no experiences in mesh networks. But in static users environment I would setup fixed small cell AP’s and use separate backhauls to interconnect. Mesh networks reduce the overall bandwidth capacity of the complete network a is my believe. But like said, with mimo we have much more bandwidth to share and mesh networks are out of my field of interest so far…

14

http://en.wikipedia.org/wiki/MIMO

They are 802.11n and therefor allow 150MBit/s theoretical
operating with 40MHz Channelsize. But this is not MIMO.

Mimo has the advantage to increase speed without the need
for more spectrum. So using groove is a very good option for small scale
installations. But using 4 Sectors with groove needs 160MHz spectrum

  • additional at least 3x10MHz spacing. Doing the same with mimo
    may half the spectrum usage. So you’ve more chances to avoid interference.
15

hmm, MT still calls it “Mimo” but in fact it should be Siso. I think they use the term because it is an ´n´ mode able radio where ´n´ is short for ´mimo´ or the otherway round streetwise spoken.
But in terminology you are right.

MT mentions 125MBit as top speed with NV2. I don’t know if you really need 40Mhz channel for that. The ´n´ protocol has some modulations that should increase speed and NV2 would help it even further. I haven’t test it but I think the 125Mb is the theoratical top speed for 20Mhz channel. But have no argument to support this shortly.

On the other hand; I don’t understand why with real mimo you would need half the spectrum to get the same bandwidth? Top speeds for dual chain mimo is normally considered to be 300Mbs, for a 40Mhz channel. If you don’t use the extension channel the maximum also falls back to 150Mbt. So for real mimo the max. bandwidth is not so much different than for a siso antenna. Mimo will probably perform better under difficult and multipath circumstances than siso. But max. throughput for both will only be reached under ideal circumstances imho and then there won’t be so much of a difference. I have no doubt you’ll correct me if I am wrong here… :sunglasses:

And who wants to have 4 grooves in a 4 sector setup? I think most of us would make two boxes with two radio’s each to do this. (Or even smarter ways, see some of my other topics.)
4 grooves with their cables and the need for an extra router to combine them sounds a bit messy to me… But ok, it is possible.

What imho is more interesting is how a dual chain (“real”) mimo AP would perform with a mix of siso units like grooves as CPE’s and maybe even mixed with some dual chain CPE’s?

I am planning to do something like that in the next half year. Clients don’t need more than 10, 15Mb throughput so there is no need for full mimo CPE. So I am planning to use groove’s in ‘n’ mode and can than make some clients use the horizontal AP chain, and some the Vertical. I am also replacing the many RIC’s I have. I exchange the slow rb133C’s for rb711-5Hn’s but they are only one chain as well. This way I extend the live cycle of the RIC a lot (and they are good!) whilst upgrading my network against relative little costs and ample waste (the old rb133C’s I dump to someone still finding a use for them..)
I am only not so sure if this mixed single chain network use of a dual chain AP would have dramatic effect on the network performance? Maybe this way you could have even more CPE’s associated to AP? Or maybe less?

16

With mimo 2 chains you get 150mbit in 20MHz without mimo you need 40MHz for 150mbit (both theoretical max). So if you’ve 11x20MHz in 5GHz band you’ve problems to do uplink and 4 Sectors on a single tower using 40MHz. Using 20MHz with 2 chains you should get the same
performance but much more channel separation.

17

this is an old topic, but the problems are actually also in 2015, so maybe we can discuss the experiences since 2011

mikrotik wasn’t made big changes in this plan since NV2 appear in 2011, the ac standard was introduced
i can say that NV2 TDMA isn’t enough for increased client speeds
what do you think?
for ex. i have a tower, somehow i need to upgrade there i have cells nv2 and not nv2

or in the near future will be made some changes on NV2?

18

There are some more Antenna Options now. The carrier class Antennas from RF-Elements perform quite good. The simper models should arrive soon (our Distributor claim to get part of the models soon). So with better signal separation nv2 should perform better.
In another post normis claimed that there have been some improvements regarding stability in nv2 in the last releases.

But still the performance/scaling in ptmp is not at the level we would like to see with .ac.

19

For ex.
I need to decrease power consumption at the tower (winter will come, and I have solar panels there),

Is there a good solution to connect 2 dual sector antenna to a single nv2 a/n/ac card?
I have bad experience with splitters or omni antennas

I don’t know if some antenna manufacturer made 2x dual sectors in different angles?

in the last releases(year 2015) changelog, i can’t find “nv2”:frowning:

20

NV2 in the first years was very unstable. I took a whole lot of software release before NV2 finally became stable.
In the mean time the wifi landscape changed;

  1. I lots of regions more players in the field = more antennas in the same band => interference
  2. Manufacturers came with more powerful radios => more electromagnetic energy => interference
  3. More subscribers went to wifi => more CPE’s in the field => more stations associated to tower => more work in same time frame => more radio energy => more interference
  4. Subscribers want more speed => need for higher connection rates / signal levels => more powerful AP’s needed and they now also work more and longer.

At the same time some improvements came into the market:

  1. Better CPE (and more choice) with better and not seldom more powerful radios. But also higher sensitivity.
  2. Software for CPE’s more matured. Regular radio and data throughput improvements are made.
  3. Newer protocols (‘ac’) and later MU-MIMO will further improve things.

Main issue of today is interference combat and smart use of available spectrum. First can be fight at with better (higher gain and directivity) antennas. Second with more and more proprietary protocols that each on its way trying to get the best out of the spectrum. (Which is nice if you already are using that brand, if not you will be reluctant to just swap your existing for something new…)

In the beginning I indeed used some duo chain cards and fit 2 (or even 3 on one early triple chain ubnt card I got!) sectors on it in the same polarization. It worked fine when there was not a lot of spectrum use and not a lot of traffic demand.
Now, to give clients the higher traffic they want and with the need for higher s/n levels (so higher signals needed) we are replacing all single chain ‘n’ network for at double chain CPE’s.
At the same time, when in the early years I would connect any client that could make up to -80 signal. Then we went to a minimum of -75 but nowadays we try what ever it need to get the client connected to at least -69 or better. I really don’t want to see high 60’s anymore, let alone something with a ‘7’.

‘ac’ is not matured by Mikrotik and probably some other mid segment manufacturers. I hear and see too many issues.
We have some links that work fine, but on long distances triple chains are useless (‘ac’ is also bringing more 3x3 mimo) and if you go for 80, 160 etc MHz bandwidth you exchange more throughput by the bigger channel for lower quality because of the higher interference probability.
I also have some very short ac links where on the same link the protocol ‘n’ seems to be more stable…
So I’ll guess ‘ac’ can still improve a lot.