MikroTik

this must be a FAQ but I could not find anything.

When is support for the proposed 802.11n standard planned?

I know it's still in draft state. But since other vendors already sell
equipment based on the draft I have customer requests.

Thanks
Ekkehard

we are working on it. no specific release date has been announced.

Try a search of the forums and you'll see that this is a hot topic and there are a ton of people asking about this. Mikrotik's official line is that they aren't going to announce a date because "programming doesn't work that way" and they could run into issues or bugs that were unexpected and miss their date. I think it is safe to bet that around 3.10 or so they will have some 802.11n support. The other issue revolving around N is that right now everything is only Draft gear and therefore could be totally different with a new draft or when they finalize the standard, that makes it hard for anyone except the vendor to support the gear. If you need 802.11n right now, check out the Ubiquiti SR71's using their RadioDriver madwifi based driver.

I think 802.11n is more a fashion that makes the industry sell. For indoor it will improve speeds, so customers can now browse their broadband (1Mb?-2Mb-3Mb? whatever!) connection with their laptop even faster!
What a joke! I sell my customers ordinary 802.11b/g wifi routers because my internet connection to them comes with 3Mb only! And in a normal house the speed of such routers towards the clients laptops is good enough to consume the internet speed even into the deepest corners of their house.

There is no gain in very high indoor speeds unless you are a freak multimedia user that needs to transport big files within his own network. So the ´n´ protocol is more a fashion, but sadly enough it will dominate the market, like all fashions do.

But read this article, already two year old: http://80211n.wifinetnews.com/archives/ ... speed.html I wonder what the experts will say about this?

It means that a WISP has probably nothing to gain from the ´n´ protocol, unless he would be dealing with a heavy urbanised ´concrete´ environment.
But since most of us Wisp´s are probably in open spaced areas and always are used to as big as possible LOS space we won´t benefit of the ´n´ standard.

Unless the experts can show us it might benefit in a forested area I see no use for ´n´ in the average ´wisp´ world.


rgds

not really, the indoor laptop users would benefit from local network file transfers, now that remote backup (ie. apple's time capsule) and file storage is becoming more popular, faster wireless in local network is very important.

I do have need for 4-5Gb file transfer from one laptop to another, and usually there is no flash storage around.

Well, we are both right.
The indoor user might benefit if he is working of an outdoor AP, which therefore will be one not to far away (otherwise not enough signal in the house). And off course the benefit in indoor situations is there.

I was more referring to the situation lots on non urban Wisp's face. Point to Point and Point to Multi Point over distances in open environments won´t benefit. And off course backhaul links over distances not at all. There is hardly or absolutely no multipath radio links.

The only case I am wondering how the ´n´ standard would perform is in an relative close proximidity environment like estate with villa's etc and lots of trees. So distances are not too big and multipath connectivity between AP and Client (outdoor CPE-units) might exist due the many reflective surfaces around.
It would be nice if under such conditions even NLOS would give highspeed possibilities.

rgds.

well I hope nobody is looking at N as some kind of cure-all for all wireless issues. Of course, as you say - benefits will be only in some situations, and mostly where there are reflections. N is *not* something that will replace b/g in *any* wireless installation.

will N help in a ptp outdoor scenario to backhall our Aps?

My ptp links are about 10-15km.
will N be able in this case to reach 80mbps fullduplex?

Regards

N isn't meant for P2P links... The thread already spells that out...

N isn't meant for P2P links... The thread already spells that out...

Wounder why Ubiquiti are touting the SR71 5Ghz N card as the card for outdoor PTP links?

Wounder why Ubiquiti are touting the SR71 5Ghz N card as the card for outdoor PTP links?

Fashion sells.....

Wounder why Ubiquiti are touting the SR71 5Ghz N card as the card for outdoor PTP links?

Fashion sells.....

Maybe. Or perhaps they have actually tested what we have all been theorising. My guess is the SR71-5 only has 2x2 format so you can connect to dual pol antennas which would make use of any signals that had relected/spun. If that's the case then these cards could be good for near or none line of sight links.

n is something like N-Stream in mikrotik.
It uses 3 different channels 1,6,13 or 11 and bonds those 3 connections which is approx. 108 or 125 mbps.
So you'll have to use 3 antennas on your system.
On the other hand you may buy 3x R52H and 433 or 600 and make n-stream.

n is something like N-Stream in mikrotik.
It uses 3 different channels 1,6,13 or 11 and bonds those 3 connections which is approx. 108 or 125 mbps.
So you'll have to use 3 antennas on your system.
On the other hand you may buy 3x R52H and 433 or 600 and make n-stream.

You are wrong. Read here:

http://en.wikipedia.org/wiki/802.11n

¨n¨ is, as I understood, developed to make use of the fact that in an environment where the signal between one antenna and another gets deflected and bounced through walls, ceilings, constructions or other surfaces deflecting radio signals, which normally cause noise and interferences and actually declines the throughput of the radio link, the ´n´ protocol makes use of these extra radio signals travelling between the both antenna's. By smart use of this data can be send through several antenna's in the same frequency and the receiver picks it up at even so many receiving antenna's. Every radio link can contain data and all the processing unit now needs to do is to make sure data is pack and arranged in the proper order again. This way much more data can be transported on one frequency in the same time frame as on the simple a,b or g protocols. But, all on the same frequency though, so in that respect the remark of sarpkaya was not right.

The ´n´ protocol therefore turns the negative side effect of deflections into a positive one and uses several data streams on the different signals reaching the receiver.

This same effect makes it therefore a technique that hardly works on long bachhaul links with narrow focussed directional antenna's. Just because of the simple fact there are hardly any bounced or deflected signals that reaches the other end. So the receiver only gets one radio signal which can only carry so much data.

You might now think that by using the same high db directional antenna's on each of the three antenna points of a ´n´ card and thus make sure three parallel radio signals could be used and thus profit from the ´n´ protocol this way.
But now each radio link would arrive at exactly the same time on the other end and because they are al in the same frequency we have a big interference problem. So big we probably have no workable signal left.

The ´n´ protocol works with, and its actually depending on, the fact that each deflected radio signal has a very small time difference compared to the other from the same radio which gives the receiving radio the change to separate the radio wave and then the cpu combines the data to reach the goal of high throughputs.

It´s clear that this technique is hardly possible in long distant backhauls.

Apart from the fact that in respect of higher radio powers needed and redundancy of the link it might be wiser to use dual n-stream to reach high throughputs.

Rudy

well , my point is: 11n is very useable for distance ...

me and my partners have successfully setup links with 802.11n cards , just using single radio, turning off the 2x extra radios of the Ar5008 chipset in this case (there are also 2x2 RX/TX version available soon) ... with xscale based board and some firmware other than mikrotik os on a 1.2 kilometer link

Platform xscale (533Mhz) 2 AR5416 cards
Firmware Version: other firmware than mikrotik
Regulatory Germany
Channel 5650
Fast-Framing
WPA2-PSK-AES
Linux Server <-> WDS-AP <->WDS-Station <->Linux-Server

Packets passed Router vlan52 Tagged


iperf -c 172.27.0.75 -t 10 -i 1 -b 74M
WARNING: option -b implies udp testing
------------------------------------------------------------
Client connecting to 172.27.0.75, UDP port 5001
Sending 1470 byte datagrams
UDP buffer size: 64.0 KByte (default)
------------------------------------------------------------
[ 5] local 172.27.0.3 port 55446 connected with 172.27.0.75 port 5001
[ ID] Interval Transfer Bandwidth
[ 5] -0.0- 1.0 sec 8.57 MBytes 71.9 Mbits/sec
[ 5] 1.0- 2.0 sec 8.89 MBytes 74.6 Mbits/sec
[ 5] 2.0- 3.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 3.0- 4.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 4.0- 5.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 5.0- 6.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 6.0- 7.0 sec 8.61 MBytes 72.2 Mbits/sec
[ 5] 7.0- 8.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 8.0- 9.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 9.0-10.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 0.0-10.0 sec 88.2 MBytes 74.0 Mbits/sec
[ 5] Server Report:
[ 5] 0.0-10.0 sec 88.2 MBytes 73.9 Mbits/sec 0.110 ms 0/62900 (0%)
[ 5] Sent 62900 datagrams

i hope that mikrotik os can do the same someday , because my costumer asking me frequently for 11n cards that are supported by mikrotik os ....

well , my point is: 11n is very useable for distance ...

[ 5] -0.0- 1.0 sec 8.57 MBytes 71.9 Mbits/sec

Am I right in saying that this would be using 40MHz of bandwidth? If so, can't you get better throughput using RouterOS, R52's and turbo (i.e. 40MHz)?

[/quote]

Am I right in saying that this would be using 40MHz of bandwidth? If so, can't you get better throughput using RouterOS, R52's and turbo (i.e. 40MHz)?[/quote]
no you cant do it with a single radio..

its like 150mbit/s over the air rate... in which you cant achieve it with a single radio currently with MT..

It´s clear that this technique is hardly possible in long distant backhauls.

Apart from the fact that in respect of higher radio powers needed and redundancy of the link it might be wiser to use dual n-stream to reach high throughputs.

Rudy

Not sure I completely agree on this.

Both technologies below appear to be part of 'N', and both will help in long-haul.

Channel Bonding is a second technology being incorporated to 802.11n which can simultaneously use two separate non-overlapping channels to transmit data. Channel bonding increases the amount of data that can be transmitted. Payload optimization or packet aggregation is a third technology in 802.11n which means more data can be incorporated to each transmitted data packet.

Where we are, we pay the same for a dual polarity antenna on a big tower as we do for a single polarity, so MIMO is a very useful technology, at least 2x2 is.

George

Am I right in saying that this would be using 40MHz of bandwidth? If so, can't you get better throughput using RouterOS, R52's and turbo (i.e. 40MHz)?

no you cant do it with a single radio..

its like 150mbit/s over the air rate... in which you cant achieve it with a single radio currently with MT..

You missed my point. Hebeda was saying that with a third party piece of software using MIMO they where able to get 71.9 Mbits/sec and I was pointing out that MT+turbo you can get better results whilst still using 40MHz bandwidth.

So has anyone else done any MIMO testing in the great outdoors? It must have some use or it wouldn't be a major part of the upcoming LTE standard would it.

Sure, Orthogon (now Motorola PtP series) works great and uses MIMO 2x2. You can do stuff that you would have a very hard time achieving with conventional radios. Particularly useful on long shots with lots of multi-path fade...

George

no mimo was used, only a single radio on a 11n card , the other 2 radios where turned off ....


but the sad thing is , the current AR9160 chipset reference design (2nd revision already) may be obsolote very soon , its seems atheros has invented another hardware bug into the current AR9160 chipset reference design ...

all cards with that chipset seem to be vulnerable to transmissions around the 70cm band (around 430-450mhz) - the cards just crash ....

but that should not bother us anyways,there is no mikrotik support for 11n available yet and much more important:

the Atheros main factory is working at 100% capacity for months , leadtimes for atheros Chips at the moment 18-20 weeks .... i have received that message from various OEM/ODMs now, some have a good backstock of atheros chips, some dont , like sparklan/GEMTEK.

Hebeda,

What was the channel size of your test? 20 Mhz or 40?

I'm waiting ancxiously for 802.11n support

Am I right in saying that this would be using 40MHz of bandwidth? If so, can't you get better throughput using RouterOS, R52's and turbo (i.e. 40MHz)?

no you cant do it with a single radio..

its like 150mbit/s over the air rate... in which you cant achieve it with a single radio currently with MT..

You missed my point. Hebeda was saying that with a third party piece of software using MIMO they where able to get 71.9 Mbits/sec and I was pointing out that MT+turbo you can get better results whilst still using 40MHz bandwidth.

and i said you can only get 108mbit/s over the air maximum meaning 50mbit/s half duplex throuhput which is less than the stated solution above.

and i said you can only get 108mbit/s over the air maximum meaning 50mbit/s half duplex throuhput which is less than the stated solution above.

"Real World" tests with Mikrotik in Turbo have proved your statement incorrect many times in the past

We get between 60 and 65 Mbits in Nstreme Turbo. I expect the 'n' MAC improvments are going to improve on that somewhat, so just over 70 Mbits sounds about right.

George

I think you are all right, or wrong....

If you have a max air rate of 108Mbps it means data througputts of 60 or 70Mb are possible.... in one direction. In theory even higher....
The moment you put both a down AND a upload data stream to one radio link in both directions it drops below mentioned rates.
After all we are working in simplex mode so the radios are switching between sending and recieving.
It's all a matter of what you want to achieve and how you measure it.

I still can't see a lot of benefits of the 'n' protocol for long distant runs though.
If I would have the need of very high througputs I would just put more parralell links together and bond these pipes. How much speed you want now? 300Mbps? 400Mbps? No problem! Just a couple of more radios and hardware which also increases the redundancy.

Price becomes an issue now? Well, who wants more then 50Mbps in both directions but don't want to spend money on it? Normally if there is a need to transport such amounts of data their is also a will to pay for it.....

well I hope nobody is looking at N as some kind of cure-all for all wireless issues. Of course, as you say - benefits will be only in some situations, and mostly where there are reflections. N is *not* something that will replace b/g in *any* wireless installation.

This says it all...

Patience people - patience....!!!

The ever converging technologies and-- DEMAND.. will determine the outcome.

T

well , my point is: 11n is very useable for distance ...

me and my partners have successfully setup links with 802.11n cards , just using single radio, turning off the 2x extra radios of the Ar5008 chipset in this case (there are also 2x2 RX/TX version available soon) ... with xscale based board and some firmware other than mikrotik os on a 1.2 kilometer link

Platform xscale (533Mhz) 2 AR5416 cards
Firmware Version: other firmware than mikrotik
Regulatory Germany
Channel 5650
Fast-Framing
WPA2-PSK-AES
Linux Server <-> WDS-AP <->WDS-Station <->Linux-Server

Packets passed Router vlan52 Tagged


iperf -c 172.27.0.75 -t 10 -i 1 -b 74M
WARNING: option -b implies udp testing
------------------------------------------------------------
Client connecting to 172.27.0.75, UDP port 5001
Sending 1470 byte datagrams
UDP buffer size: 64.0 KByte (default)
------------------------------------------------------------
[ 5] local 172.27.0.3 port 55446 connected with 172.27.0.75 port 5001
[ ID] Interval Transfer Bandwidth
[ 5] -0.0- 1.0 sec 8.57 MBytes 71.9 Mbits/sec
[ 5] 1.0- 2.0 sec 8.89 MBytes 74.6 Mbits/sec
[ 5] 2.0- 3.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 3.0- 4.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 4.0- 5.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 5.0- 6.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 6.0- 7.0 sec 8.61 MBytes 72.2 Mbits/sec
[ 5] 7.0- 8.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 8.0- 9.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 9.0-10.0 sec 8.87 MBytes 74.4 Mbits/sec
[ 5] 0.0-10.0 sec 88.2 MBytes 74.0 Mbits/sec
[ 5] Server Report:
[ 5] 0.0-10.0 sec 88.2 MBytes 73.9 Mbits/sec 0.110 ms 0/62900 (0%)
[ 5] Sent 62900 datagrams

i hope that mikrotik os can do the same someday , because my costumer asking me frequently for 11n cards that are supported by mikrotik os ....

Can you give us some info on this? What driver did you use? How did you setup 802.11na master mode or is this ng? what is the trick? Adhoc seems not to work for 802.11n on most manufacturers cards and we successfully set up AP mode only in windblows. We are experimenting with antennas, configurations, boards etc I have not come around a driver that works properly with this mode in linux... unless this is the ubiquiti's driver which i have not tried (yet) since i got non of those cards in my hands. Our indoor tests point to 70+Mbit for 20MHz and 90+ for 40MHz.
Can you please give us some extra info on this? I have tried anything under the sun with ndiswrappers, madwifi trunks test versions, ralink drvs, drivers more drivers but nothing. I must be missing something Definitely. We need to get a test bed going as for to develop further other aspects of a working system until MT comes out with a working version and 802.11n.
Also if MT can help us on this I would be grateful!

and i said you can only get 108mbit/s over the air maximum meaning 50mbit/s half duplex throuhput which is less than the stated solution above.

"Real World" tests with Mikrotik in Turbo have proved your statement incorrect many times in the past

but my real world tests does prove my stataments so well...

Moreover I dont need "n" anyways..
BTW, does "n" will be better than nstreme protocol ? thats what i want to know thou..

I am pretty sure n outdoors is a cure for fresnel issues, knife edge distortion, urban reelection (my pop is on a downtown office building). My main feed is only 3 blocks but these issues are always a threat. I do not need 50mbit. I need 20mb reliable.

Sure you can get 50 mbit 25 miles if you have a mountaintop or 200ft towers, but what if you only have a rooftop on one side, you might see the tower (or where you think it is) but there is no way to go 200ft up and clear fresnel zone. mimo to the rescue.

Comparing mimo to nstream or bonding is a non-issue because mimo is a layer1 technology, the physical layer. It uses multiple receive signals before they are converted to bits to better discover what data was sent.

I think, N is something like a rumor. If you hear a rumor from one side, it *maybe* real. If you hear it from a second side, its *probable* real. Mimo does the same: Better physical error correction by using more than one signal way.

Some manufacturers does sell polarization diversity directional antennas with 2 coaxial connectors. Each for one polarization layer. In result, it works a little bit similarly to good old helix antennas. Both signal ways, horizontal and vertical polarization layer, are used.

Some people here sayed, that NStreme turbo is equal to 11n. Okay, if data flows in *one* direction, this maybe right. In most cases, the data flows bidirectional. Here is the bandwidth broke down. The best solution for this is NStreme2. One directional link (and frequency within) for each data direction. Can you visualize, what a bandwidth is possible when you build a 11n-NStreme2-Link with dual polarized antennas? I think, there is a multiple of 100M possible.

The next legitimate question is: What we can do with such fast p2p links? Most small WISPs are using normal ADSL connections to the internet. Here is the internet gateway the bottleneck. But, some WISPs are very good connected (we currently negotiating with a great german carrier for a collocation). How we can use such highspeed internet connection optimal when our wireless p2p links are the bottleneck?

My conclusion: 11n can lift up 5 GHz p2p links to the next level of quality. It will get ahead of wimax: Faster, broader, cheaper - better. And no frequency license fees! 11n can also lift up 2,4 GHz omni installations in areas covered with buildings.