MikroTik

In RouterOS v5.0beta5 there is a new wireless protocol Nv2 which is based on TDMA.

To make the connection using Nv2 on the AP side you need to select the wireless-protocol=nv2 and on the Station side you can also select wireless-protocol 'nv2' or 'any'.
Some features are not ready yet for Nv2: no support WDS,Security,Virtual AP,some details in the registration table and status. We will optimize the Nv2 in the upcoming versions.

We would like to get some feedback from you how it is working on your wireless links. When posting, please include info on the distance of the link and settings that you have and latency.

When testing N link with Nv2 you can get approx 97Mbps UDP traffic and 94Mbs TCP traffic through RB711 in one direction using one chain.

Any documentation ? what does setting "nv2 nstreme 80211" in one row ? combination of this three or what..

more documentation will be available next week.
If you select that option all 3 in one row in the Station it means it will chose any from that list.

i really hope this is compatible with airmax... please please please.

hope this is compatible with airmax

I hope not ! I prefer things that Work.

Is Nv2 going to have any kind of Time Synch, or will it be just for the Outback like Airmax ?

nv2 is not compatible with Airmax.

nv2 is supported for all Atheros cards, except old AR5211 based, so in order to try it out it is not required to have 11n cards - just upgrade, set wireless-protocol=any on station (this will enable station to look for whatever protocol AP that is available with appropriate SSID), set wireless-protocol=nv2 on AP and you are ready to go. If something goes wrong change wireless-protocol on AP to nstreme/802.11 - whatever you were using before.

If "Time Synch" means some synchronization between APs, then no, nv2 does not support such feature yet. You can think of it as of improved nstreme. Its benefits are intended to come from removed polling overhead.

We've just done some testing on a 1.6km link, LOS. RB 433 on each side with R52n Cards. 22db Panel on 1 side and 22db Grid on the other. Only 1 chain

http://spreadsheets.google.com/ccc?key= ... tdFE&hl=en

We were bandwidth testing thru the routers with IP routing, We seem to have run into a issue/bug where we couldnt RX more than 33mbit which we are still looking into but so far NV2 is performing worse than Nstream.

Link signals are -72ish with 90+% CCQ

These results are not meant to show the max it can do but merely show the different between the protocols

hi all, glad to see new feaures in router os...

what is the advantages of using TDMA in router os?

Now tested is horrible .

WDS nstreme work very well with one chain switch to station and MPLS work little slower, but when go to Nv2 is disaster TX/Rx rate don't want to go more that 54/54Mbps and when traffic go link disconnected.
Also without traffic ping is 5-6ms with traffic don't know disconnected, old setting have 0ms ping without traffic.


Go back to nstreme WDS and have TX/RX 150-HT/150-HT and good speed.

Tested on two x86 distance 400m one chain card is both side R52n.

If you select that option all 3 in one row in the Station it means it will chose any from that list.

That is a brilliant feature, maybe the devs could think of a better name so it is self explanatory. Maybe evem putting a comma between them e.g. Nv2, Nstreme, 802.11

My results on testbed, RB800 - 1333MHz, x86 1333MHz, both R52N + 5GHZ a/n,

Can you please post wireless config and link information? Distance, Signal etc?

From your post it seems that NV2 only made a difference in a Dual Chain setup and even then a tiny difference

more documentation will be available next week.
If you select that option all 3 in one row in the Station it means it will chose any from that list.

This is great news. My long time wish of stations autoselecting nstreme on/off
has come true

i test link on 6 km distance , 433ah bords and r52n cards with one chain and 23 grid antenas . so i recive up to 69mb tcp and 94mb udp traffic , latency is leap from 3ms ,4ms ~ to 20~35 , link stability is wery bad ! .. what time is going out -- link stability spoil and spoil and some time latency reach up to ~3000ms.

Let's be scientific and not fan boy

Quite right.

It's Sour Grapes with me:-
I bought a load of Airmax when it was released, and at that time it did not work, and neither could the kit be used for CPE on non-Airmax either.

There was no way to get the money back, or to make it work for about 6 months, which made me slightly unhappy.

Seriously? Whats next? Wanting Motorola Canopy compatibility?

Anyone have tested on x86 on 802.11n link ?
I can't get it work properly have low speed and disconnect, on some link WDS+nstreme work ok.
Also after changing some parameter on wireless interface router restart.

Sorry to hijack the thread, but the comment on compatability is right. Mikrotik working with Ubiquiti is so much stronger than proprietary protocols. The battle isn't between Mikrotik and UBNT but between wireless and AT&T or comcast or whatever else competes with wireless. Standards create markets. Proprietary limits markets. Every proprietary non-wireless protocol was swept away by ethernet regardless of the advantages and the networking world advanced. Standard means more production of the base chipsets and lower prices means more penetration. Please work with UBNT and the other vendors we'll keep using Mikrotik for the great routing and features that are beyond what UBNT is offering and we'll grow faster and by more!

uldis, will we be able to assign time slots to individual clients using this TDMA system?
As in, will I be able to assign more important clients a large % of the air time so they get good performance even when the link is under high load?

Hello

is TDMA support a high number of cpe clients?
Nstreame is good for 60 +/- how bout this new protocol?

Thanks

First tests look promising.

Testsetup 2xRB411ah/R52n/V+H Antenna on desk.
AP/Bridge -> Station, 5GHz only N, datarates=default

UDP Test receive gives 192MBit. Ping while test runs is at 8-14ms.
Stopping bandwith test ping goes *up* and stays up and shows
varying results. Pings differ from AP->Client/Client->AP.
Starting Bandwith test again ping go down to 8-14ms-

Looks like bandwith test influences nv2 parameters which do not
adapt back.

Doing same test with nstreme gives 120MBit and dropping connections.
So nv2 is more stable in this config. Got not a single disconnect while
testing. But there are some tweaks needed to give stable pings.

Using nv2 Tx-Signal Strength is not shown.

Can you do TCP tests? UDP is almost always insanely higher than real world TCP data

can post the right configuration here instead of just announce that's a new protocol for this ?

Hello

is TDMA support a high number of cpe clients?
Nstreame is good for 60 +/- how bout this new protocol?

Thanks

Protocol limit is 511 client.
Yes you can connect a lot of clients to the AP, but note that the throughput and the latency could get worse.

If you select that option all 3 in one row in the Station it means it will chose any from that list.

That is a brilliant feature, maybe the devs could think of a better name so it is self explanatory. Maybe evem putting a comma between them e.g. Nv2, Nstreme, 802.11

Wireless-protocol setting info.
When the card is in the AP mode:
nv2-nstreme-802.11 - AP will use Nv2 as it uses the first what is in the row
any - it will use 802.11 or nstreme depending if you have specified the old setting enable-nstreme=yes

When the card is in the Station mode:
nv2-nstreme-802.11 - first will look for the Nv2 AP, then nstreme AP and then regular 802.11 AP
any - it will choose the AP without any special sequence; the sequence can be altered by the use of connect-list

Sorry to hijack the thread, but the comment on compatability is right. Mikrotik working with Ubiquiti is so much stronger than proprietary protocols. The battle isn't between Mikrotik and UBNT but between wireless and AT&T or comcast or whatever else competes with wireless. Standards create markets. Proprietary limits markets. Every proprietary non-wireless protocol was swept away by ethernet regardless of the advantages and the networking world advanced. Standard means more production of the base chipsets and lower prices means more penetration. Please work with UBNT and the other vendors we'll keep using Mikrotik for the great routing and features that are beyond what UBNT is offering and we'll grow faster and by more!

You are right. Even 802.16d had a horrible death.

Mix and match has to be the priority of the vendors.

I agree, I like mikrotik as well. Cooperation would be splendid.
On different note, I changed over my 802.11n wds ptp link to nv2, and lost remote connectivity. I switch it back and everything goes back to normal. I think this might be due to it not supporting wds in this mode. Does anyone know a quick way around this? I would like to implement tdma and not lose remote connectivity due to wds being down on my link.

you need compatibility only if you will use different brands of products with some specific technology that's not a standard (if you use 802.11b everywhere, no problem with compatibility). why do you use different brands? because other-brand cpe's are cheaper? mikrotik is working on that

I agree, I like mikrotik as well. Cooperation would be splendid.
On different note, I changed over my 802.11n wds ptp link to nv2, and lost remote connectivity. I switch it back and everything goes back to normal. I think this might be due to it not supporting wds in this mode. Does anyone know a quick way around this? I would like to implement tdma and not lose remote connectivity due to wds being down on my link.

unfortunately Nv2 doesn't have support for WDS yet. For now you could make MPLS/VPLS tunnel across the link and then you could easy switch between the wireless protocols.

Most people so far seem to be looking at this from a throughput point of view; I'm more interested in the reliability point of view:

Here is a too brief ping test I performed just now:

NV2:

Ping statistics for 10.23.22.1:
Packets: Sent = 100, Received = 99, Lost = 1 (1% loss),
Approximate round trip times in milli-seconds:
Minimum = 14ms, Maximum = 300ms, Average = 48ms

802.11:

Ping statistics for 10.23.22.1:
Packets: Sent = 100, Received = 96, Lost = 4 (4% loss),
Approximate round trip times in milli-seconds:
Minimum = 13ms, Maximum = 1531ms, Average = 174ms

nstream (via protocol drop-down)

Ping statistics for 10.23.22.1:
Packets: Sent = 100, Received = 100, Lost = 0 (0% loss),
Approximate round trip times in milli-seconds:
Minimum = 10ms, Maximum = 272ms, Average = 43ms

This link is being unusually well behaved right now. In the late evenings it is frequently bad enough to be unusable. I will leave it in nv2 mode for a few days and see how it does.

Personally, I prefer stable links with occasional high latency to ones that drop packets or the entire link all the time; thus hiding packet loss (typically by retransmitting behind the scenes until the packet gets through, and thereby converting packet loss to the appearance of a lower throughput link), than links that pretend they are fast, while having 60% packet loss. I'm hoping this is what nv2, being TDMA, will give me. 802.11 pretends to work this way, but usually just drops the link instead (after a limited number of failed retrys). (Basically, I'm describing Canopy vs nearly everyone else).

you need compatibility only if you will use different brands of products with some specific technology that's not a standard (if you use 802.11b everywhere, no problem with compatibility). why do you use different brands? because other-brand cpe's are cheaper? mikrotik is working on that

No. Because they are cheaper and smarter. You've to update your RIC-line of Products.
Dreams of a Wisp:
- get the cpe from MT, don't touch it give it direct to the Customer. cpe gets all it needs from AP
- Customer can install cpe by looking at LEDs while mounting it
He can see that there is power and/or connection without calling/connected pc
- no crimping, simple slider to tighten Ethernetcable
- no try and error to find wireless parameters, working defaults
- 2 different cpes, small for short range, bigger for long range
- cpe needs to look smart and not cheap
- RF-Engineered (Metal housing, good antennas)


Your products *are* cheaper when we have less work to install/manage them.

I just wish to know if I can use XR5 cards

I just wish to know if I can use XR5 cards

Yes, is also working on those cards and regular R52 cards.

511 is the protocol max ?

what is the other bit for ?

Broadcast ?

all this sounds ok if you build a wirelss network from the scratch
but we allready started with that business more than 10 years ago
and i cant change alle clients ...
an as we do also many competetors have grown networks - and also use UBNT cpes
as they do their job for a fair amount of money - sorry but money is a big argument
and also the offer of a complete cpe inlcuding antenna and alectronics in one part
without any frickeling around ....
i know that there are mikrotik based cps on the market but they are more expensive
so why not giving the compatibility tdma a chance ??
it dont think that it is that complicated...

In my experience UBNT CPE have a higher TCO when you're doing advanced functions. You may get around the install issue by having the client install it themself but in my experience 30-40% of the time you'll need a tech to go back onsite and realign it. If you want simple CPE then you need to run a simple network either pure UBNT or pure 802.11. Inter operation is complex, You'd need UBNT and MT to work together on a standard or combining their TDMA spec and then you as the operator would have to undertake a huge compatibility testing rig to check everything works and even then when you have a fault you still have to question a compatibility issue.

And if your dealing with a legacy network then tough, Work around it. Theres no reason why a pure MT network can be installed around/over the top of a legacy wireless network where new services for existing clients would require a CPE upgrade/change. Its not MT's job to build your network for you. Harsh I know but its painful reading everyone wanting MT to be compatible with their Toaster or that their CPE doesnt have this or that, UBNT are cheaper for CPE because the feature-set is extremely limited when you compare it to MT. Once again MT isn't here to build your network for you, Its upto you to shop around for the CPE option that suits you the best and every option has it's pro's and con's so you have to look at them and figure outs whats best for you

i would rather not use ubnt... but point me to a dual polarity CPE for < $100 that doesnt take 30 mins to assemble, configure, and look all messy when its done. when there is a nice mt cpe then I will switch back to all mt. thats why it would be nice for them to interoperate for the time being (on links > 17km).

Yes. The RIC-series needs a strong update. The integrators do not deliver a pleasant device.

ok, please read the forum rules. this forum is for discussing MikroTik products. It would be much more helpful if you all would stick to the topic, and tell us what's to improve (except compatibility, because that will never happen)

Will be created a nv2-dual like nstreme-dual?

no problem staying to the rules !!
but if you would like to sell more of your products you should also here to your customers !!!!
we have many hundreds of your devices in use !
and we need an affordable client
we have this wish since years now - and where is your answer ??
rb 411 ? - which enclosour - which antenna - powersupply .....
we have no time to build our cpes - we want to by them - but with some quality and a resonable price ...
why dont you build a device like the nanostation -
the ric was ok - but quality is not good - the first ones fall off allready - they dont stand the UV and the weather
and the white cover starts to brake ....

PTP devices ar not the problem - here we use only mt products in antennas with included housing from huber&suhner
not cheap but verry good !

511 is the protocol max ?

what is the other bit for ?

Broadcast ?

The AP itself?

Wow a lot of people moaning about nothing in here.

Low cost Mikrotik CPE, easy.

1x Mikrotik RB711
1x Poynting WLAN-A0056
1x All in one PSU/POE Injector

All designed to fit together, no screws, no hot glue, it all snaps together in less than 10 minutes. And based on estimated pricing this will cost USD $100, which in my opinion is damn cheap for a CPE that has such an extensive feature set.


What Mikrotik are doing with Nv2 is proof that they listen to their customers, they are implementing TDMA, it works on a/b/g cards as well as n, it has adjustable time slots and to top it off they have added a mechanism for the CPE to automagically select the wireless protocol. All are features that have been mentioned on these forums.

so far..
we test a 11n link with MPLs setup...

with only MPLS, the link can carry 6Mbps full duplex TCP traffic when using the bandwidth test tools to test it...
with NV2 + MPLS, the link can carry 15Mbps full duplex TCP traffic when using the bandwidth test tools to test it...

in our here, the frequency is damn congested no matter what frequency range you running on.. i also dunno how come... ^_^

BTW, i going to setup another 4 link and test it out with the 11a card... anywhere... great job Mikrotik..

cheers...

Hi, this is a snapshot of my configuration, is a 44km PTP with rb433 r52hn antenna Kozumi (28db V / H).
was improved in 8mb
disabled csma
protocol= nv2
tdma period size = 10 (is 10 because 2 was unstable and 7 smaller bandwidth)
Extension Channel = HT BELOW CONTROL
distance = 44
adpative noise inmunity = ap&client

sorry for my English, it is my first post.

petro, you are on beta4. this topic is about specific improvements on beta5 and a new protocol called nv2. please stay on topic.

Does nv2 disable CSMA ?

petro, you are on beta4. this topic is about specific improvements on beta5 and a new protocol called nv2. please stay on topic.

Normis...
we tested it with the R5h card...
the result is bad... and also, the V5.0Beta5 is not suitable for the wds mode..

when we downgrade it back to the V4.10, the wds back to the normal...

Vincent, uldis clearly said - WDS IS NOT SUPPORTED, please, there is no use testing nv2 with wds because it is not supposed to work!

Vincent, uldis clearly said - WDS IS NOT SUPPORTED, please, there is no use testing nv2 with wds because it is not supposed to work!

ops.. forgotten to said that, we tested the R5h card with mpls setting and it getting bad result instead of wds setting...
Beside this, the V5.0Beta5 is not suitable for the wds setting also.

Vincent, this topic is only about NV2 feature!

uldis, will we be able to assign time slots to individual clients using this TDMA system?
As in, will I be able to assign more important clients a larger % of the air time so they get good performance even when the link is under high load?

why do you use different brands? because other-brand cpe's are cheaper? mikrotik is working on that

There is no question that Mikrotik's software is far superior to the "other guys." The problem is that you have fallen far behind on the hardware and I think you are seeing that in your sales numbers.

You need an all-in-one, or at least modular, MiMo CPE device with LEDs for basic diagnostics.
"My Internet is off"
"Are any lights on?"
"No"
"Is the electric cable plugged in?"
"Oops, my cat moved it."

After my technicians started simply screwing Bullets to panel antennas, they began referring to the years when we assembled RooTennas with routerboards as the dark ages.

The RB711 is a good start, but it's neither MiMo nor has an antenna.

No, you are wrong

The problem is that you have fallen far behind on the hardware and I think you are seeing that in your sales numbers.

haha, not just WISP's use Mikrotik products.

Im pretty sure any loss in CPE sales was more than made up for in sales of RB450, RB750, RB1000, RB1100 products

No, you are wrong

I love to say: You're far off topic
I would like to see this discussion in another thread at it's worth
to be done.

uldis, will we be able to assign time slots to individual clients using this TDMA system?
As in, will I be able to assign more important clients a larger % of the air time so they get good performance even when the link is under high load?

Currently not, but in future we might add this feature.

uldis, will we be able to assign time slots to individual clients using this TDMA system?
As in, will I be able to assign more important clients a larger % of the air time so they get good performance even when the link is under high load?

Currently not, but in future we might add this feature.

Does the access-list AP Tx Limit/Client Tx Limit work with nv2?

uldis, will we be able to assign time slots to individual clients using this TDMA system?
As in, will I be able to assign more important clients a larger % of the air time so they get good performance even when the link is under high load?

Currently not, but in future we might add this feature.

Does the access-list AP Tx Limit/Client Tx Limit work with nv2?

No, they don't work yet.

Normis..

what's the routerboard should apply for this new feature test to get best performance..
and it should couple with what type of the card..

with such infor, then only we can provide the useful data for your reference...
we have a lot of site can test it and can allow you to remote it for fine tuning...

angboontiong, you can use any RouterBoard - all depends what kind of traffic you want to push through and at what speed. You can use any Atheros N card if you want to use N protocol or you can use regular A/B/G Atheros card.

nice improve! ... is good news. Surely the protocol greatly improved with the advance of time.

regards

I have a testlink in my office with v5.0beta5

RB411 with r2n
RB433 with r52n

with a 2Ghz link an nv2 my whole network goes down
with tons of collisions.

this is an edge link - nothing connected to the RB411
and ethernet only for access to the RB433

when I swiched to 802.11 the problem goes away

what is this traffic? It can't be seen by torch....

nv2 doesn't play nice. it wants the whole playground to itself.

?????????????

So how to avoid it?

How do I incorporate a nv2-link in my network
without it bringing it all down?

RGDS

nv2 doesn't play nice. it wants the whole playground to itself.

Nice.

Any more details ?

You sound like a Canopy salesman working for MikroTik, and i like how that sounds.

yes, this is because we use TDMA, in a similar manner as canopy.

Ah. So you have never worked with Canopy then.

I'm slightly dissapointed, but that should help those who think it will interfere with their 802.11a installations.

Unless it's on the same or adjacent channel, then it won't interfere at all.

Your statement about the 'Whole Playground' was a bit misleading.
TDMA-based will only want the nice bit with the Swings - the Slide and the Roundabout will still be available.

From what you've posted, i can only deduce that it is an 802.11n variant with TDMA overlaid, just like AirMax.

An earlier post suggested that it would be Great to have UBNT and MT co-operate on this.
I said No. Now i think i should have said Yes. Why invent 2 wheels when we use MT as AP kit and UBNT as CPE ?

Then again. maybe i'm as thick as a plank and a bit drunk, and smell.

Ok, been running it for a couple of weeks, and so far I love it, when it works.

It has higher throughput with more stability, However, under heavy traffic the client radio seems to lock up.

Wirelessly, everything appears to be fine, but my bgp peer dies, and no traffic seems to pass. I can see the AP in /ip neighbor, so apparently packets can still be recieved, but attempts to mac-telnet to the AP will fail. If I disable and re-enable the radio, it reassociates but traffic still doesn't pass. If I then reboot the client, everything works normally for a while, until it happens again. It seems more likely to happen if the traffic is heavy, and takes anywhere from 5 minutes to an hour (or several hours or days, if traffic is light).

802.11n 2x1 with the two antennas on the client side.

client side, r52hn:

Flags: X - disabled, R - running eles
0 R ;;; chain 0 rx -75 chain 1 rx -71 both together -69
name="wlan1" mtu=1500 mac-address=00:0C:42:xx:xx:xx arp=enabled
interface-type=Atheros 11N mode=station ssid="xxx" frequency=2432 band=2ghz-onlyn
scan-list=default wireless-protocol=any wds-mode=disabled wds-default-bridge=none
wds-ignore-ssid=no default-authentication=yes default-forwarding=yes
default-ap-tx-limit=0 default-client-tx-limit=0 hide-ssid=no security-profile=xxx
compression=no

0 interface=wlan1 radio-name="000C42xxxxxx" mac-address=00:0C:42:xx:xx:xx ap=yes
rx-rate="26.0Mbps-HT" tx-rate="13.0Mbps-HT" packets=728971,1075510 bytes=92547344,1316880575
frames=542358,1043764 frame-bytes=93289116,1300263050 uptime=13h1m23s last-activity=0ms
signal-strength=-75dBm tdma-timing-offset=67 tdma-tx-size=1008 tdma-rx-size=496 tdma-retx=40
tdma-winfull=0

ap side, r52n:

Flags: X - disabled, R - running
0 R name="wlan1" mtu=1500 mac-address=00:0C:42:xx:xx:xx arp=enabled
interface-type=Atheros 11N mode=ap-bridge ssid="xxx" frequency=2432
band=2ghz-onlyn scan-list=default wireless-protocol=nv2
wds-mode=disabled wds-default-bridge=none wds-ignore-ssid=no
default-authentication=yes default-forwarding=yes default-ap-tx-limit=0
default-client-tx-limit=0 hide-ssid=no security-profile=xxx
compression=no

0 interface=wlan1 radio-name="000C42xxxxxx" mac-address=00:0C:42:xx:xx:xx ap=no
rx-rate="13.0Mbps-HT" tx-rate="26.0Mbps-HT" packets=1075142,728512
bytes=1316846323,92239759 frames=1043407,541944
frame-bytes=1316992303,70794244 uptime=13h20s last-activity=0ms
signal-strength=-84dBm tdma-timing-offset=67 tdma-tx-size=496
tdma-rx-size=1008 tdma-retx=36 tdma-winfull=0

in the next release of the beta version there will be a fix for the Nv2 link lockup problem.

in the next release of the beta version there will be a fix for the Nv2 link lockup problem.

Uldis..
another issue is, in the V5.0Beta5, if we key in following channel scan list into the wireless equipment..
it not able to link up, to resolve it, we have to cut short the number of frequency to be scan in the scan list

"5150,5160,5170,5180,5190,5200,5210,5220,5230,5240,5250,
5260,5270,5280,5290,5300,5310,5320,5330,5340,5350,5360,5370,
5380,5390,5400,5410,5420,5430,5440,5450,5460,5470,5480,5490,
5500,5510,5520,5530,5540,5550,5560,5570,5580,5590,5600,5610,
5620,5630,5640,5650,5660,5670,5680,5690,5700,5710,5720,5730,
5740,5750,5760,5770,5780,5790,5800,5810,5820,5830,5840,5850"

The above scan list is work well in the V4.11 with the MPLS or WDS setting.

Vincent, does this question relate to NV2? If not, post it as a new topic!

Vincent, does this question relate to NV2? If not, post it as a new topic!

Normis...
Yes, it we put the wireless protocol to unspecified then having no issue on this.

Vincent, does this question relate to NV2? If not, post it as a new topic!

Normis...
Yes, it we put the wireless protocol to unspecified then having no issue on this.

So you add that ling scan-list enable the wireless-protocol=nv2 on the client and it can't connect to the AP? What does the log say when you enable wireless,debug logs?
On what frequency is the AP?

Vincent, does this question relate to NV2? If not, post it as a new topic!

Normis...
Yes, it we put the wireless protocol to unspecified then having no issue on this.

So you add that ling scan-list enable the wireless-protocol=nv2 on the client and it can't connect to the AP? What does the log say when you enable wireless,debug logs?
On what frequency is the AP?

Hi Uldis..
i tried to use the 5150Mhz to 5850Mhz with MPLS+NV2 setting, it totally not able to connect..
but when i cut the scan list channel and remain 1/4 of it, then no matter what range of the frequency, it can connect.

in the wds or the MPLS only, i having no issue with this.

Anywhere, 1 thing i have to highlight here, the NV2 is very good feature and it's work great, if the channel scan list issue can solve, then it will wonderful.

angboontiong, please enable the wireless,debug logging on the client and make the support output file after it can't connect. Also make the support output file from the AP. Send both files to support@mikrotik.com
We tested that long scan-list in the one nv2 client and it was able to connect to the AP.

angboontiong, please enable the wireless,debug logging on the client and make the support output file after it can't connect. Also make the support output file from the AP. Send both files to support@mikrotik.com
We tested that long scan-list in the one nv2 client and it was able to connect to the AP.

Hi Uldis..
ok, i will do it and send it to support, btw, what's the maximum number of the channel for the scan list?

There are no explicit limits for scan-list length, but still it is better to keep it shorter - the shorter it is, the faster full scan will complete. Indeed there is a problem with nv2 and long scan-lists that will be fixed in next version. Take into account that scan for nv2 networks takes approximately two times more time and this was not accounted for, therefore the problem with connecting with nv2. For now you can use shorter scan lists.

any idea on v5.0beta 6 will be avalable?

most likely next week

How is achieved synchronisation on this TDMA system ? Is there a high precision clock on Wifi boards ? A master / slave system with a PLL on each station ?

If each station is running free wheel, how can they synchronise correctly with each other ?

most likely next week

Another issue is..
is the MPLS work with NV2, it can up to 80Mbps full duplex (TCP)...

but one thing is, when do the bandtest, it up to the 70Mbps full duplex, we stop the bandtest, the link will drop but you can see it's in the registration table, but not able to MAC telnet, MAC ping, and also telnet such thing..

and we have to wait for may be 1 min, then only the link will come back..

for more information..
is the bandtest, the throughput < 70Mbps, it will having no problem at all, the thing is, you have to keep the signal range no lower than -50 the best is -52dbm ~ -60dbm

By the way, somebody should test NV2 in a Point-to-multipoint scenario, with multiple clients. This is where TDMA should really shine

While i'm impatient to test it in a real enviorment, i've to upgrade all the routers to 5.beta5 which scares me a little..
And the chance to have to go to the customer and reset the CPE is a lot of work..
Just waitin some more betas before testing it..

Just a question.. If i upgrade only one AP (on a site of 4 AP), does it interfere with the other 3?

problem with the Nv2 link locking/stalling will be fixed in v5.0beta6, which will be released this week.

Just a question.. If i upgrade only one AP (on a site of 4 AP), does it interfere with the other 3?

This will be an interesting point to look at. As nv2 is told to be more aggressive it may do more
interference to nearby sectors.
Thanks to the new wireless settings the sector can be switched back to nstreme/plain 802.11 with one setting
at the ap ...

if you use different frequency on other APs on the same tower it will not cause any problems. And even if they are on the same frequency, usually they point to different directions. And as ste wrote, you can easy change from Nv2 to Nstreme/802.11 with one command, as in the client side you can specify the wireless-protocol=any

"By the way, somebody should test NV2 in a Point-to-multipoint scenario, with multiple clients. This is where TDMA should really shine"


that's true. TDMA is absolutly unusefull on duplex point to point links, when a different channel is used for transmitt / receive.

TDMA is interesting really when multiple stations want to go online at the same time. Then each one take a time slot without collision. This does allow a true QOS.

As opposite, CDMA based systems exhibit collisions as soon as the number of clients and consummed bandwith rise. CDMA can be compared to an Ethernet networks with hubs instead of switches.


Nevertheless, i doubt that conventionnal Wifi hardware can do a true TDMA network, because there is no precision master clock like a GPS clock on each station. Something we can find on GSM networks.

I suppose that the pseudo TDMA mode of wifi does use the access point as the master clock, and the clients synchronise on it with a PLL.

In this case, the clients need to have a good reception, to have a clean signal to extract clock from.

Currently NV2 works perfectly for me.
Even the links that were not functioning due to interference now are ok.


Is there any provision for correction of WDS?

Testing NV2 on a troublesome link over sea The link is done with dualpol antennas and R52Hn.
Without nv2 the link was unstable due the evaporation and other shit related oversea links. max rates were 52mbps-HT, dropping frequently.
With nv2 rates jumped to 216mbps-HT and really stable. As already noticed, the link locks up. Waiting for the next release.

Well done MT!

I tested on the following hardware (identical @ both sides)

RB411A
R52N
ARC Dual Pol Antenna/Enclosure (2x2 MIMO)
Link distance ~7miles
ROS V: 5.0beta6

I ran the bandwidth tests FROM the router, so if i had tested through the router I may have been able to achieve higher throughputs.

NSTREME ENABLED

NV2 ENABLED
As you can see there's a definite increase in throughput and QOS - latency is much more consistent even on a PTP link w/ the new TDMA powered NV2. The results, again, might be limited to the RB411A's processor because I ran all tests from the RB411A. However, it served the purpose of showing consistent throughput and latency w/ NV2 enabled.

I'll see if I can find an area to test NV2 in a PTMP environment.

Will NV2 support full duplex links?

I have a NSTREME DUAL link I'd like to try this on... Not sure if NV2 was designed or ready for this type of implementation...

I did some PTMP tests with 2.4GHz-G only mode 20MHz channels...

only 3 CPE's connected to the AP, but the average latency when throwing everything i had at the CPE's was still 16ms...

And I was able to get 2.5Mbps synchronous off an RB133 board... about 6-7Mbps asynchronous... If only I had a higher powered board at this site I bet I could post better throughput results. I'm excited to see this go to stable release!

I have another site that has about 25 clients with some heavy users (3Mbps+) in the 2.4GHz band currently using NSTREME.. if this proves stable enough on the smaller deployment i'll test it out on the larger.

@NORMIS or ULDIS

Found a little bug:

When using NV2 protocol - security profiles are overridden. I use RADIUS for MAC-Authentication and it will not authenticate when mode is set to NV2. ROS only consults the local ACL and if the MAC is not present it rejects authentication from the client.

If I have it set to 802.11 or NSTREME they connect and authenticate via RADIUS just fine...

William, in beta6 we have a new latency optimization system, see newsletter

William, in beta6 we have a new latency optimization system, see newsletter

Yeah the newsletter is what prompted me to test the newer beta release. Any ideas on the RADIUS issue? Do you need a supout?

Also SNMP appears to be broken in 5.0beta6...at least for monitoring interfaces.

Normis...

Nowday, If the last miles latency is higher than 10ms, customer will keep asking on it and they starting to compare it, and this also one of the UAT benchmark for them.

The latency is possible to make it <5ms no matter what throughput running on the wireless bridge? hope can heard the good news from Mikrotik in the next release.

and the other issue is, as what we tested, the max throughput for the RB433 in NV2 is...

1) TCP: 23 Full duplex (up 23Mbps and down 23Mbps concurrent),
2) UDP: 40 Full duplex (up 40Mbps and down 40Mbps concurrent).

Is this due to the processor issue? and Is the RB433AH can double up the throughput?

well this was my first test with the new protocol using only NV2 with a rb 433, R52N card, a distance of 20km, there are tests using a network in MPLS-VPLS-OSPF, test mode 5Ghz N-20Mhz, Channel 5745 . Now also did the test mode 5Ghz only on the same channel 20Mhz 5745.

Hi MT,

some questions regarding NV2:

- How does TDMA Period Size work?

Default is 2 seems to give ping 5-6ms on idle link
Set it to 1 gives ping 3ms on idle link but lowers bandwidth test results.

Seems to be the slice duration until the next station is allowed to send?
So what happens when there are 10 stations registered but idle. Does ping
increase to up to 20? (Station waits at max 10 slots to send ping and at
max 10 slots to receive answer).

- CPU impact compared to nstreme / plain 802.11
Does it make sense to use nv2 or nstreme or 802.11 when there are old 133-Boards
as clients?

- Pinging from station to bridge/ap gives much higher response times than the
other way round. Is there a way to bring this down below 10ms?
Using nstreme pings are at 2ms switching to nv2 its between 18 and 37.
1 AP 2 clients connected, no traffic.

Thanxs,
Stefan

What I realized was that following this first test of the season to ping the ap-bridge the ping varies between 5 and 12 to 15ms, while ap-bridge station to ping below 5ms with traffic is stable pings below 6ms, this still does not know how I can because I am in mellhorar test further believe that without traffic has to be below 3 ms, because without using the NV2 that my test drop below that much of a tip as the other.

Another issue in 5.0beta6 w/ NV2

In the two ptmp environments i've installed it on the users disconnect after about 12hrs and will not reconnect unless the interface is disabled/re-enabled.

I forgot the disassociation message that appeared in the logs - but it was pretty generic.

The point to point link I have setup seems to work fine w/o issue.

Are both sides beta6? can you send us the supout.rif files that you made after the disconnection has happened, before restoring the connection?

Another thing I noticed when the test is in the TCP cpu goes 100% even auditioning for band 5mb in both.

Are both sides beta6? can you send us the supout.rif files that you made after the disconnection has happened, before restoring the connection?

Yes all sides were beta6. I forgot to make the supout.rif -- I'll switch them back to NV2 (standard NSTREME was more stable) and generate it when they disconnect. You should have it in about 24hrs.

Also -- Any idea why RADIUS MAC authentication doesn't work w/ NV2? It seems to work fine with any other protocol, just not w/ NV2.

Another issue in 5.0beta6 w/ NV2

In the two ptmp environments i've installed it on the users disconnect after about 12hrs and will not reconnect unless the interface is disabled/re-enabled.

I seen this also, it seems the more clients the quicker it happens
x86 ap

b6

Just submitted a supout.rif.

Hi guys,
as I did for 802.11n protocol (look at this post), I post below some screenshots of my test link using Nv2.

Compared to 802.11n results in the very same conditions and hardware in use, I got better results with this new Nv2 protocol expecially on fullduplex links (100Mbit/s instead of the previous 65Mbit/s with 802.11n!!!), CPU usage is higher but ping time is much much lower. I got worse results only on TCP speeds, but that was limited by CPU and I would have got better results if I cascaded two RB1000 after my test boards.

Well, so far a very good improvement from Mikrotik, thank you guys!


SCENARIO
AP and STATION: RB433AH and R52n radios inside outdoor metal boxes with u.fl-N pigtails
Antennas: Jirous JRC-24 DuplEX
Distance: 4km with perfect LOS

Mikrotik parameters (see my previous post for details)
Security profile: WPA2-PSK AES
Wireless protocol: Nv2
Hw retries: 15

RESULTS:

Highest UDP half duplex speed: 200Mbit/s stable and simmetric on downlink, 165M avarage on uplink, CPU average load @ 33%, average ping @ 13ms



Highest TCP half duplex: 59Mbit/s stable on downlink and 61Mbit/s on uplink , CPU average load @ 100%, average ping @ 5ms



Highest UDP full duplex speed reached: 100Mbit/s stable on both downlink and uplink, CPU average load @ 71%, average ping @ 8ms


Highest TCP full duplex speed: 40Mbit/s stable on both downlink and uplink, CPU average load @ 100%, average ping @ 5ms


Bye,
mattia

Hey iw4eky! A very good result!
If we could meet during the next MUM I would be very proud to shake your hand!

Stay on the piece!

Bye.

LALLO

Another issue in 5.0beta6 w/ NV2

In the two ptmp environments i've installed it on the users disconnect after about 12hrs and will not reconnect unless the interface is disabled/re-enabled.

I forgot the disassociation message that appeared in the logs - but it was pretty generic.

The point to point link I have setup seems to work fine w/o issue.

I experienced the same problem, but forgot to supout.
Will make it next time it happens

We installed a 16 km link PTP in side A RB433, with a card RB52Hn, 32dbi solid dish, on the side B RB600 with Xr5 card and 32dbi solid dish
signal strength -81 dBm
CCQ% 90
Noise: -103 dBm
protocol: NV2 Nstreme
band: 5ghzA
TxRate: 14mbps
TrRate: 12Mbps

We put this new NV2 to work in 2 different set of links:
1st: I'ts a 16km long link.
2nd: It's our newest darling, 100km long link.

In the 1st one, it has been working flawlessly for 3 days in a row and we got for our complete surprise (since it's a PTP it does not get all the benefits of TDMA) a huge increase not only in bandwidth availability (in order of 30/40 increase) but ping times seems to stay very stable even when running at full speed and not a sign of the usual disconnections when you reach max bandwidth limit.

In the 2nd one, sadly we got no results since the station refuses to stay connected to the AP when setting the wireless-protocol to the NV2 or NV2-nstreme.
When enabled the NV2 option will generate several in a row messages:

03:10:54 wireless,info 00:0C:42:XX:XX:BA@wlanV established connection on 5395, SSID BrAP1
03:10:59 wireless,info 00:0C:42:XX:XX:BA@wlanV: lost connection, ranging timeout
03:10:59 wireless,debug wlanV: must select network
03:10:59 wireless,debug 00:0C:42:3A:XX:XX: on 5395 AP: no SSID BrAP1 caps 0x0 rates 0x0 basic 0x0 MT: no
03:10:59 wireless,debug wlanV: no network that satisfies connect-list, by default choose with strongest signal
03:10:59 wireless,info 00:0C:42:XX:XX:XX@wlanV established connection on 5395, SSID BrAP1
03:11:06 wireless,info 00:0C:42:3A:XX:XX@wlanV: lost connection, synchronization timeout

We got several ranging/sync timeout messages, we tried all kind of settings in both the AP/station.

Link runs smoothly when "old-nstreme" is set.

juliano, make the support output files from both ends after the link is disconnecting and send it to the support@mikrotik.com so we could analyze them why they are disconnecting.

Hi Mikrotik..
another issue is, the hidden ssid feature is not working.

hello I have been using Nv2 in v5.0beta6 for about a week now. I have a PTMP with 3 Clients on it. I loaded Nv2 in v5.0beta6 on the Clents as of the AP and updated the bootloader as well. It dose seems to work alot better when it works after about 30mins or so the AP Locks up and I have to go in there and reset the wlan1 and then everything comes back up for 30mins or so. I have tryed it under Nv2 Nstreme would run for like 5 to 15mins and then under Nv2 Nstreme 802.11 worked alot better. Any ideas why it would do that I do know that it was doing this on the clinets. "just giving you all some feed back" but for now I moved it back to Nstreme normal ver.

AP is a 433AH w/ xr9 at 10mhz (No over Clocking)
Clients are running 411AH w/ xr9 at 10mhz (No over Clocking)

when something "locks up" you should make a supout.rif file and send to us

Uldis, support files sent!

I also sent but had no response (except automated one)

Is there any info or does anyone know the technical differences between the different Nv2 type (Nv2, Nv2 802.11, Nv2 Nstreme)

Any ETA on the next release and for 5.0 release ?

Is there any info or does anyone know the technical differences between the different Nv2 type (Nv2, Nv2 802.11, Nv2 Nstreme)

http://forum.mikrotik.com/viewtopic.php ... 82#p219282

you need compatibility only if you will use different brands of products with some specific technology that's not a standard (if you use 802.11b everywhere, no problem with compatibility). why do you use different brands? because other-brand cpe's are cheaper? mikrotik is working on that

The reason why I use different products is that we have constant problems with the R52HN they got so hot it melt the socket of the 433AH.

you need compatibility only if you will use different brands of products with some specific technology that's not a standard (if you use 802.11b everywhere, no problem with compatibility). why do you use different brands? because other-brand cpe's are cheaper? mikrotik is working on that

The reason why I use different products is that we have constant problems with the R52HN they got so hot it melt the socket of the 433AH.

we have not hear about anything like that. can you send pictures to support ?

you need compatibility only if you will use different brands of products with some specific technology that's not a standard (if you use 802.11b everywhere, no problem with compatibility). why do you use different brands? because other-brand cpe's are cheaper? mikrotik is working on that

The reason why I use different products is that we have constant problems with the R52HN they got so hot it melt the socket of the 433AH.

we have not hear about anything like that. can you send pictures to support ?

Normis,

I returned a couple like that already and the unit was returned via the channels

tell me the serial numbers and I will check the reason for the problem

tell me the serial numbers and I will check the reason for the problem

will get it

Is it possible mikrotik-engineers and programers make package wireless NV2 for ROS 4.11 (or other stable version)?

it's not possible, because it relies on many new things in v5.

Cannot wait for ROS 5.x stable and NV2 WDS support.

which is the better configuration for the parameter tdma-period-size?
you can tell me any example?

Javier
CIBERNEK

Hi,

Just been trying out nv2 in my "lab" (well, ok the corridor of my flat!). It looks very promising - lovely stable pings at very high throughputs.

Does Mikrotik have any idea yet on when security will be possible with nv2? By that I mean, for example, wpa2.

cheers

Peter

which is the better configuration for the parameter tdma-period-size?
you can tell me any example?

Javier
CIBERNEK

here is some small info on Nv2 added, maybe it might help you:
http://wiki.mikrotik.com/wiki/Manual:In ... reless#Nv2

I'm trying to get 110Km link working with nv2. Getting lots of range timeouts. With usual nstreme the link is performing as expected.

Read that nv2-cell-radius might be the setting to tweak. I cannot find it in cmd neither gui.
Access is available for Mikrotik if they want to evaluate long distance performance as this is not a production link yet.

Any input appreciated.

dibatech

you can manipulate with tdma-period size...


see link in the previous post........

Period size is not exactly the tool you would use to fix this. Size was adjusted up 10ms with no improvements.
Thank you in any case.

Period size is not exactly the tool you would use to fix this. Size was adjusted up 10ms with no improvements.
Thank you in any case.

Please write to support, we will provide you with a test version where you could specify the range.

Tested a BETA version of the Version 5.0RC1 package to see if the the nv2-cell setting helps to fix the range-timeouts on a 110km link.
Guess what, it works. Well done guys.

Results:
Setup:
RB411U -> RB433AH
R52Hn Cards both ends.
29 dB Parabolic Dishes single feed.
Signal levels -77dB to -80dB
18Mbps/18Mbps clamped on card rates
Frequency: 5220Mhz
Length: 110 Km
TDMA period size = 4

Working well with nv2, nv2-nstreme and normal wireless protocol.

5GHz-20MHz Band Works
5GHz-10MHz Band Works
5GHz-5MHz Band Works

Bugs:
Cell radius in Winbox is wrong. Always reports 10km, but works via CMD.
Wireless registration quality of signal probably needs more work.
Stability 100% over last 14 Hours.

Further results:
[test@test] > /ping 172.21.200.34
HOST SIZE TTL TIME STATUS
172.21.200.34 56 64 5ms
172.21.200.34 56 64 4ms
172.21.200.34 56 64 7ms
172.21.200.34 56 64 5ms
172.21.200.34 56 64 6ms
172.21.200.34 56 64 6ms
sent=6 received=6 packet-loss=0% min-rtt=4ms avg-rtt=5ms max-rtt=7ms
*(While doing around 1 Meg. Bi-Directional)*

Bandwidth Test Results:
Receive TCP: 5Mbps
Send TCP: 8.9 Mbps
Both TCP 3.2/3.9 Mbps

Any ETA on the release? We have a few sites we want to test the P2MP on (in WDS mode if possible) and are in dire need of more capacity on our AP's. Be happy to test and provide feed back on the previously mentioned wireless update.

Been using RC1 for P2MP works great.
AP did freeze with neer frequency interference which was resolved.

WDS, Routed.... what type of setup are you using.... can you send me a copy?

Thanks

uldis, will we be able to assign time slots to individual clients using this TDMA system?
As in, will I be able to assign more important clients a larger % of the air time so they get good performance even when the link is under high load?

I would like to use ap/client tx rates (access-list).
Users with lower tx-rates get fewer TDMA slots.

This will match better with price-modells.

Hi, please where can I download the RC1?
Thank you, also tests P2MP.

Hi, please where can I download the RC1?
Thank you, also tests P2MP.

Contact support and ask for test version.

Hey, the RC1 is still not functional bridge or WDS? (with NV2)

I tried AP brigde ---> pseudobridge station, but failed.

Torch always has 0 bps for TX.
WatchDog Timer will trigger if an IP address is specified, even though enabled is unchecked.
NV2 still freezes from time to time.

Hey, the RC1 is still not functional bridge or WDS? (with NV2)

I tried AP brigde ---> pseudobridge station, but failed.

pseudobridge isn't working with Nv2. Use station-wds on the client side.

Hey, the RC1 is still not functional bridge or WDS? (with NV2)

I tried AP brigde ---> pseudobridge station, but failed.

pseudobridge isn't working with Nv2. Use station-wds on the client side.

Thanks,

I have RB411AH (Nstreme) (5RC1) ----> RB600 (with RB604) (5RC1)
RB600 has 5 miniPCI.
1st Station - Nstreme
2.AP Bridge - 5 Ghz (802.11) (P2MP)
3 AP Bridge - 2.4 GHz (P2MP)
4 AP Bridge - 5 GHz (Nstreme) (P2MP)
5 AP Bridge - 5 GHz (Nstreme) (P2MP)

If the 5th AP Bridge switch to NV2 (all clients 5rc1) .. so after a few hours there is a wide irregular reboot RB600. If the switch back to 5.AP Nstreme, RB600 currently holds longer uptime 24 hours.

WatchDog Timer will trigger if an IP address is specified, even though enabled is unchecked.

There are actually two watchdogs, which isn't entirely clear from the winbox interface.
One, controlled by the enabled checkbox, which monitors the router itself, and the ping watchdog, which monitors an IP address. Its enabled if an IP address is present in the box, independently of the enabled checkbox. Some text or layout should probably be added to the winbox dialog to make this more clear.

WatchDog Timer will trigger if an IP address is specified, even though enabled is unchecked.

There are actually two watchdogs, which isn't entirely clear from the winbox interface.
One, controlled by the enabled checkbox, which monitors the router itself, and the ping watchdog, which monitors an IP address. Its enabled if an IP address is present in the box, independently of the enabled checkbox. Some text or layout should probably be added to the winbox dialog to make this more clear.

Wow, I never knew that. I always just thought it was a persistent bug that it still rebooted when I unchecked the box but left the IP in there. Good to know!

Hello,
I try 5.0rc1 and got worse results results than with 4.6 firmware and exe config wds+nstreame.

In 4.6 firmware ping on idle link is aprox 0.7 ms
In 5.0rc1 firmware (nv,nv nstreame and others) ping on idle link is aprox 4 ms

In 4.6 firmware connection rate link is stable 270/270
In 5.0rc1 firmware (nv,nv nstreame and others) bouncing from 54 to 160 and so on

Throuput also is 2 time smaller than 4.6

Can you give some advise what we doing wrong.
I you said nv2 have some improvements, but i don't see any advantages on 5.0rc1 only problems.

Thanks for advice
Best regards

mindaugasr,
contact support about this problem - include the support output files from both versions and description about your wireless link.

sorry for my ignorance, but if i set "nv2" protocol active, can i see other stations from Motorola or other in scan or frequency-usage lists?

another question, if its possible that in the RC1 i can not select 40Mhz channel bandwith, only 20Mhz?

Thanks a lot

We are have lockup up problems on RB532 and related boards, including a kernal panic on a 112 (not that anyone cares about a 112). We seem to get 6-12hr before it locks up. We have a large number of 532 ap out there. I do not know at this point if the 400 series are effected in the same way as we have only loaded RC1 on clients and on the remote side of the P2P that has the 532 ap. Problems is present regardless of nv2 mode.

Will send supout.rif later to day once the unit has been rebooted and is accessable.

I can verify that on 433 and 411 Routerboards all appears to be working well so far.
The posted V5rc1 is performing Excellent over the Prerelease V5rc1.

We are using NV2 nstreme on 2.4GHz-10MHz.
We also have a backhaul link we are testing using 5GHz-only-N using NV2-nstreme and wds, and it is performing remarkably well also.

We are have lockup up problems on RB532 and related boards

We tried nv2 on an rb532 and didn't work very well. Wouldn't stay associated for more than a few seconds if I recall properly. I'm not sure if it was the 532 or the wireless card (probably an r52 or r2). Its since been replaced with an 433AH which works much better. The 433 has an 11n card. Looking back at the 5.0Bx changelogs, I suspect it might have been the r52 or r2 that was the problem before.

Don't use ROS v5.x on MIPS-LE boards. Throw those old and slow boards to trash and get a newer ones.

Don't use ROS v5.x on MIPS-LE boards. Throw those old and slow boards to trash and get a newer ones.

We have several 100s of them in the field used by our customers.
It would take some weeks of work to replace them.

They are capable to route 10-15MBit/s of Traffic which is enough for
the average internet user. Why should a newer ROS/Linux make them
unusable? I guess newer SW should be better than older ).

You can use v5 on RouterBOARD 100/500 (MIPS-LE), but remove all unneeded packages, disable as many rules as possible, don't keep Winbox open for long time, and try to optimize configuration as much as possible. These devices have less memory and less storage space.

Testing indoor on PtP link on v5.0rc1, both chains active, bridge mode, throughput is pretty good, stability seems to be much better than older nstream, latency is better but still pretty higher than to be expected from fast link without traffic. It is possible to further improve and lower latency?
I will do some real outdoor testing...

UDP half duplex test: UDP full duplex test: TCP full duplex test using iperf:

We have stripped the RB100 and 532 to the bones. No improvement the 532s have plenty of mem but 100's are tight we will likely replace them. The 532 is still not stable despite the changes and the the 411ap with some rb133 cpe's has locked up a few time and we have had some strange issue with the 133. The 532 to 411 back haul does not last past 12 hr and has minimal traffic.

We have a couple of 411 back hauls we are testing on and they seem much better with no disconnects yet after a couple of days.

On P2MP clients will not attach if any "B" data rates are set when up grading (10mhz nstreme G only). Removing the B data rate allow clients to connect.

Anyone tested on a 900mhz (xr9) setup .... had to back out as nothing would register on the AP (411AH)

Can anyone tell me why i cannot select 40Mhz channel width on the wireless interface in routeros v5 rc1? It doesn't seem to matter what cards I use or what band I set it to, it always shows the error "Couldn't change interface <wlan1> - incompatible band and channel-width (6)" I tried every band on the SR71-15 cards and got this error and I tried every band on the R52N cards and get the same result. Any ideas?

Set it to 20mhz, then go to the HT tab and enable the extension channel.

I have it setup that way but it is only using 20 Mhz Channel from what i can tell. I get about 190 Mbps tcp one way and about 95Mbps tcp bi-directionally but i know it would be twice that if i can utilize 40Mhz channel width. I started trying to test Nstreme dual with N today but I must say that my first attempt is a real flop. How do you get the data rates above 54Mbps in dual Nstreme? Does anyone have a basic config that works well or any ideas? I have the same problem there too with 20Mhz channel being used but on the "HT" tab it let me set it to 40Mhz channel width and I set it to double under "HT Streams". Again any help or ideas are extremely appreciated.

You are running 40Mhz. You can't get 190Mbps on 20Mhz.

here are my results, 2x RB411AH and R5nH

hi all..

Once i've used it just for testing... i was quite satisfied, nv2 has very stable results... 80 mb ful duplex with N cards...
But i couldn't see CCQ status.... it will be good if you add some features for troubleshing porpose..and need more detailed information how it works exactly

this protocol has good future

Anyone tested on a 900mhz (xr9) setup .... had to back out as nothing would register on the AP (411AH)

I have tried it a few times with really bad results. I have two 411A XR9 AP's on one tower running in 5mhz channels with regular nstreme enabled and 411's as clients. When I install 5.0 beta, throughput starts getting a little worse right off the bat. As soon as I turn on NV2, I get lots of disconnects on BOTH AP's regardless of how much channel spacing I have. I can have one set at 2427 and the other at 2442, and as soon as I pass much traffic on either radio, clients from both start disconnecting and the whole tower becomes unusable. I've changed every setting and nothing made much of a difference.

I've since downgraded the AP and all clients back to 4.11 with the old nstreme again. I never tested upgrading both AP's because if it doesn't play nice enough to change one over (even with plenty of RF spacing) NV2 is useless since I can't/won't change all towers simultaneously.

Please fix the watchdog so that it creates a supout file.
This way the NV2 protocol can be debugged.
This would allow me to have the supout available from the client as well as the AP.
It also reboots the client and AP so minimal downtime is experienced.

Win Win.... If you can fix the watchdog.

nv2 doesn't play nice. it wants the whole playground to itself.

What happens if I share a tower with other WISPs?

Are we now in a race to see who can implement NV2 first? If the other guys beat me to it, will my links become unusable?

I'm interested in that comment myself, Does it mean that nv2 will be affected by interference more than normal or that it will create interference with normal 802.11 gear?

I'm interested in that comment myself, Does it mean that nv2 will be affected by interference more than normal or that it will create interference with normal 802.11 gear?

Exactly, all my highsites are pretty noisy so NV2 might be counter productive?? Or have we missed something?

From the documentation:

As NV2 does not use CSMA technology it may disturb any other networks in the same channel. In the same way other networks may disturb NV2 network, because every other signal is considered noise

http://wiki.mikrotik.com/wiki/NV2#Compa ... _protocols

News Flash: The cell radius setting is pretty important.

I couldn't get an NV2 link to work. The client side would register and then drop after only a few seconds. The AP side would never register. I had to go into the command line and set it by hand.

I am getting pretty disappointed in the v5 Winbox. It *still* doesn't have manual MCS settings and doesn't include some key NV2 configuration options.

I couldn't find it on this thread, so i repeat it:
max-station-count isn't working on nv2 pmtp
(i limited to 60 but i reached 72 users)

I couldn't find it on this thread, so i repeat it:
max-station-count isn't working on nv2 pmtp
(i limited to 60 but i reached 72 users)

did you email support?

Sent right now.

Sent right now.

I checked, and it's a known missing feature actually.

I couldn't find it on this thread, so i repeat it:
max-station-count isn't working on nv2 pmtp
(i limited to 60 but i reached 72 users)

Are this 72 users happy? Did you run nstreme before on this AP? How does it compare?

I couldn't find it on this thread, so i repeat it:
max-station-count isn't working on nv2 pmtp
(i limited to 60 but i reached 72 users)

How much aggregated throughput do you get?

TBH they weren't happy, but there were a lot -86 users (cause i put on it users that usually don't connect on that antenna).
Right now i have 2 radios with 50 users both.. 10 mbit aggregated traffic, pings between 15ms and 100ms (a lot of jitter). This is one random user CPE ping.. looks good, 2mbit profile perfectly served.(51 users right now on AP)
As soon as i upgrade more clients, i'll will do further tests.

TBH they weren't happy, but there were a lot -86 users (cause i put on it users that usually don't connect on that antenna).
Right now i have 2 radios with 50 users both.. 10 mbit aggregated traffic, pings between 15ms and 100ms (a lot of jitter).

Code: Select all

ping 10.20.100.1 count=20 interval=500ms
HOST                                    SIZE  TTL TIME  STATUS                   
10.20.100.1                             56    64  22ms 
10.20.100.1                             56    64  19ms 
10.20.100.1                             56    64  40ms 
10.20.100.1                             56    64  13ms 
10.20.100.1                             56    64  46ms 
10.20.100.1                             56    64  40ms 
10.20.100.1                             56    64  41ms 
10.20.100.1                             56    64  19ms 
10.20.100.1                             56    64  46ms 
10.20.100.1                             56    64  23ms 
10.20.100.1                             56    64  31ms 
10.20.100.1                             56    64  19ms 
10.20.100.1                             56    64  36ms 
10.20.100.1                             56    64  31ms 
10.20.100.1                             56    64  13ms 
10.20.100.1                             56    64  16ms 
10.20.100.1                             56    64  5ms  
10.20.100.1                             56    64  14ms 
10.20.100.1                             56    64  28ms 
10.20.100.1                             56    64  17ms 
    sent=20 received=20 packet-loss=0% min-rtt=5ms avg-rtt=25ms max-rtt=46ms 

This is one random user CPE ping.. looks good, 2mbit profile perfectly served.(51 users right now on AP)
As soon as i upgrade more clients, i'll will do further tests.

Would having clients connecting on a -86 affect the other clients? Surely TDMA gives each client a time slot to send/receive which makes it irrelevant if you have -86 clients or not?

Your thoughts look correct.. maybe 70 users were too much... i'll try to load the AP with more -70ish users and check again..

News Flash: The cell radius setting is pretty important.

I couldn't get an NV2 link to work. The client side would register and then drop after only a few seconds. The AP side would never register. I had to go into the command line and set it by hand.

I am getting pretty disappointed in the v5 Winbox. It *still* doesn't have manual MCS settings and doesn't include some key NV2 configuration options.

Cell radius and other Nv2 options will be added to the winbox in RouterOS v5.0rc2.
Yes, you have manual MCS settings in the winbox - go to the Data-rate tab, set the configured radio button and then go to the HT-MCS tab and enable/disable any rates you like.

News Flash: The cell radius setting is pretty important.

I couldn't get an NV2 link to work. The client side would register and then drop after only a few seconds. The AP side would never register. I had to go into the command line and set it by hand.

I am getting pretty disappointed in the v5 Winbox. It *still* doesn't have manual MCS settings and doesn't include some key NV2 configuration options.

Cell radius and other Nv2 options will be added to the winbox in RouterOS v5.0rc2.
Yes, you have manual MCS settings in the winbox - go to the Data-rate tab, set the configured radio button and then go to the HT-MCS tab and enable/disable any rates you like.

Sorry, I meant manual TX POWER settings for individual MCS rates.

You cannot do that in rc1 but you can in v4.11.

nv2 doesn't play nice. it wants the whole playground to itself.

What happens if I share a tower with other WISPs?

Are we now in a race to see who can implement NV2 first? If the other guys beat me to it, will my links become unusable?

I've only tested NV2 twice... Both times using a tower which had two 900 mhz sectors with routerboards and XR9's running in 5 mhz channels using old Nstreme from version 4.11.

Both times were a disaster. I never got around to upgrading both routerboards on either tower. Once I had one upgraded (and even before I touched the NV2 settings) performance got worse. Then once I turned on NV2 on the one sector and associated clients, neither sector was stable. Even if I had one on the lowest supported channel and one on the highest supported channel with two channels between. If I put any traffic on either sector, clients from both sectors would drop continuously and I just couldn't use it at all.

If the protocol messes things up even when I have 10mhz of buffer between channels, it is useless for me. I downgraded as fast as possible and decided I would wait to see if I improved in the future.

nv2 doesn't play nice. it wants the whole playground to itself.

What happens if I share a tower with other WISPs?

Are we now in a race to see who can implement NV2 first? If the other guys beat me to it, will my links become unusable?

I've only tested NV2 twice... Both times using a tower which had two 900 mhz sectors with routerboards and XR9's running in 5 mhz channels using old Nstreme from version 4.11.

Both times were a disaster. I never got around to upgrading both routerboards on either tower. Once I had one upgraded (and even before I touched the NV2 settings) performance got worse. Then once I turned on NV2 on the one sector and associated clients, neither sector was stable. Even if I had one on the lowest supported channel and one on the highest supported channel with two channels between. If I put any traffic on either sector, clients from both sectors would drop continuously and I just couldn't use it at all.

If the protocol messes things up even when I have 10mhz of buffer between channels, it is useless for me. I downgraded as fast as possible and decided I would wait to see if I improved in the future.

Nv2 is only working in v5.x. Maybe you are confusing Nstreme with Nv2?
If you have problems running Nv2 please contact support@mikrotik.com and include the support output files and detailed info on the wireless link.

70 users (with good signals)
Ping from a client to my gateway -headstrong- looks like signals still affect the whole AP performance..
Today i'll try VoIP calls on clients of this AP

Anyone tested on a 900mhz (xr9) setup .... had to back out as nothing would register on the AP (411AH)

I have tried it a few times with really bad results. I have two 411A XR9 AP's on one tower running in 5mhz channels with regular nstreme enabled and 411's as clients. When I install 5.0 beta, throughput starts getting a little worse right off the bat. As soon as I turn on NV2, I get lots of disconnects on BOTH AP's regardless of how much channel spacing I have. I can have one set at 2427 and the other at 2442, and as soon as I pass much traffic on either radio, clients from both start disconnecting and the whole tower becomes unusable. I've changed every setting and nothing made much of a difference.

I've since downgraded the AP and all clients back to 4.11 with the old nstreme again. I never tested upgrading both AP's because if it doesn't play nice enough to change one over (even with plenty of RF spacing) NV2 is useless since I can't/won't change all towers simultaneously.

I have a PtP link with two XR9s, a RB532A, and RB411. I didn't have much success with beta5, but the nv2 protocol seems to be working much better in rc1. My link is NLOS 1.6 miles using a 5MHz wide channel, signal strengths stay around -76 or so. With nv2 enabled I get about 500k more throughput and much stabler pings.

2x RB411AH, R5N, 100mbit/sec tcp from just one machine copying a file across the link.

My only problem now is that I cannot stream video files properly across this system, they seem to get choppy after a while, but yet the link is still transferring the same amount of data. The video transfer rate is anywhere from 10mbit/sec to 25mbit/sec depending on the detail of the video, so its way under the bandwidth that this link can handle. Yet this was never a problem on 3.30 using 802.11a.

Is there any way to debug something like that?

Oh and it crashed if I had the cpu set to 800mhz.

OK. I've woken up now, and am interested.

Will Nv2 have some kind of time synch ?

If not, it's the same as AirMax in the sense that we can't use it effectively.

I checked, and it's a known missing feature actually.

Oooops. I have that Feature. I don't know why i stole it.

Where can i meet you to give it back.

So sorry. I don't know what came over me.

we are testing rc3 with nv2 protocol on n link and an a link
the n link runs fine with nv2
the a link with nv2 keeps on rebooting the rb
is the nv2 protocol only for n links
thanks

70 users (with good signals)
Ping from a client to my gateway

Code: Select all

 ping x.x.x.x count=10
HOST                         SIZE  TTL TIME  STATUS                  
x.x.x.x                         56    63  16ms 
x.x.x.x                         56    63  50ms 
x.x.x.x                         56    63  44ms 
x.x.x.x                         56    63  51ms 
x.x.x.x                         56    63  38ms 
x.x.x.x                         56    63  47ms 
x.x.x.x                         56    63  41ms 
x.x.x.x                         56    63  26ms 
x.x.x.x                         56    63  41ms 
x.x.x.x                         56    63  19ms 
    sent=10 received=10 packet-loss=0% min-rtt=16ms avg-rtt=37ms max-rtt=51ms 

-headstrong- looks like signals still affect the whole AP performance..
Today i'll try VoIP calls on clients of this AP

Hi MediaNet

Did you manage to try voip on the NV2 sector?
How did it perform?

Thanks

If it's TDMA, then it really really needs some time synch.

If it has none, it's like Airmax, and therefore of limited benefit.

I bet Skypilot rue the day they used a nano 2 and tested their TDMA code on it.

Hi there,
Is there a known bug with nv2 when running on x86 where CPU usage will seemingly randomly shoot up, often pounding 100% hard enough to lock up the board so that telnet/winbox are unresponsive and a reboot is required? I'm experiencing this problem with both 4.13 & 5.0rc3 and on two separate x86 boards. Here's some information which might help:

[admin@fragtion-rpc] > system resource print
uptime: 17m44s
version: "4.13"
free-memory: 1914032kB
total-memory: 1945820kB
cpu: "Intel(R)"
cpu-count: 1
cpu-frequency: 3058MHz
cpu-load: 11
free-hdd-space: 416074kB
total-hdd-space: 484630kB
write-sect-since-reboot: 568
write-sect-total: 568
architecture-name: "x86"
board-name: "x86"
platform: "MikroTik"
[admin@fragtion-rpc] > system package print
Flags: X - disabled
# NAME VERSION SCHEDULED
0 X dude 3.6
1 system 4.13
2 routeros-x86 4.13
3 wireless-nv2 4.13
4 X ipv6 4.13
5 X ups 4.13
6 synchronous 4.13
7 routerboard 4.13
8 X wireless 4.13
9 X hotspot 4.13
10 X dhcp 4.13
11 mpls 4.13
12 routing 4.13
13 X ppp 4.13
14 X security 4.13
15 X advanced-tools 4.13
[admin@fragtion-rpc] >

For the most part, nv2 is awesome... some of my link throughputs nearly double with the right nv2 tuning, and cpu usage seems generally lower than nstreme... it's just when this happens that CPU surges for some reason, the log doesn't say much, and the board locks up completely requiring a reboot. Because of this bug, I can't use nv2 at the moment. It only seems to effect x86...
Any feedback would be appreciated, thanks... let me know if I should post any other information or attach anything

have you tried to check if the changing multi-cpu option helps avoiding lockup?

have you tried to check if the changing multi-cpu option helps avoiding lockup?

Thanks for the suggestion, uldis. I tried it: '/system hardware set multi-cpu=no' (it was on 'yes') then manually rebooted, but within minutes it locked up again. In fact it seems a bit more useful when set to 'yes', because then it seems to reboot itself about a minute or two after the lockup... but not the case when set to 'no' - I have to manually powercycle

Edit: the problem (still) exists even in 5.0rc4

OK. I've woken up now, and am interested.

Will Nv2 have some kind of time synch ?

If not, it's the same as AirMax in the sense that we can't use it effectively.

AirMax just announced gps sync coming with their new hardware.

It is time for RouterOS to do the same.

Tom

have you tried to check if the changing multi-cpu option helps avoiding lockup?

Thanks for the suggestion, uldis. I tried it: '/system hardware set multi-cpu=no' (it was on 'yes') then manually rebooted, but within minutes it locked up again. In fact it seems a bit more useful when set to 'yes', because then it seems to reboot itself about a minute or two after the lockup... but not the case when set to 'no' - I have to manually powercycle

Edit: the problem (still) exists even in 5.0rc4

we have the same problem on x86 and it is worse on a links and better on n links

Well said rocnoc - Mikrotik NEED GPS Sync, without it, UBNT will start stealing customers (at least for wireless stuff)

OK. I've woken up now, and am interested.

Will Nv2 have some kind of time synch ?

If not, it's the same as AirMax in the sense that we can't use it effectively.

AirMax just announced gps sync coming with their new hardware.

It is time for RouterOS to do the same.

Tom

Just please release it STABLE

OK. I've woken up now, and am interested.

Will Nv2 have some kind of time synch ?

If not, it's the same as AirMax in the sense that we can't use it effectively.

AirMax just announced gps sync coming with their new hardware.

It is time for RouterOS to do the same.

Tom

It's time to make 11n/nv2 working stable and solid before jumping into the next
adventure. Please dont do the same as these AirMax Guys. I dont want to hear
promises and get months later a product which is alpha quality.
I need products which just do what they promise before I get the next
advertised feature.

Alpha quality? We just popped a pair of bullets in that took us all of an hour to configure and install, 9km with 90-95mbit, Blew MT out of the water

Alpha quality? We just popped a pair of bullets in that took us all of an hour to configure and install, 9km with 90-95mbit, Blew MT out of the water

I tested Rockets some month ago. Does no DFS as promised and therefor was *illegal* in
my country. Signal strength values were false and driving it with allowed power
gave useless connections. Do they still reboot when changing a parameter?
How do they perform when you colocate them with other gear. Saw that some
guys make a hobby out of building metal shields to make them work.

I need some well engineered gear which work under difficult conditions flawless
as it's my business and not my hobby.

So I need MT with stable working NV2 and when it's done I need sync.

Alpha quality? We just popped a pair of bullets in that took us all of an hour to configure and install, 9km with 90-95mbit, Blew MT out of the water

I tested Rockets some month ago. Does no DFS as promised and therefor was *illegal* in
my country. Signal strength values were false and driving it with allowed power
gave useless connections. Do they still reboot when changing a parameter?
How do they perform when you colocate them with other gear. Saw that some
guys make a hobby out of building metal shields to make them work.

I need some well engineered gear which work under difficult conditions flawless
as it's my business and not my hobby.

So I need MT with stable working NV2 and when it's done I need sync.

Yeah they still reboot which is pretty lol but if i'm going to put them into production I'll put n+1 setups together to handle this and any hardware failure, DFS was working for me about 7 months ago when 5.2 just came out, Signal levels are all very close to the math done on the link. THe shields were to improve the back-to-front ratio to allow high density installs, Its and antenna thing now a radio thing.

But yes, MT should release some stable sections. New features are great but there is alot of buggy stuff that needs to be fixed first

Нi everyone!
I have link 4km on two RB411AH+R52Hn work in 2.4Ghz
I have a problem with "disconnected control frame timeout" sometimes.

Configure my device:

First:

/interface wireless> print
Flags: X - disabled, R - running
0 R name="wlan1" mtu=1500 mac-address=XX:XX:XX:XX:12:A4 arp=enabled
interface-type=Atheros 11N mode=ap-bridge ssid="xxxx" frequency=2304
band=2ghz-onlyg channel-width=20mhz scan-list=2264
wireless-protocol=nv2 wds-mode=dynamic wds-default-bridge=bridge1
wds-ignore-ssid=no default-authentication=no default-forwarding=yes
default-ap-tx-limit=0 default-client-tx-limit=0 hide-ssid=yes
security-profile=default compression=no

Second:

/interface wireless> print
Flags: X - disabled, R - running
0 R name="wlan1" mtu=1500 mac-address=XX:XX:XX:XX:12:9A arp=enabled
interface-type=Atheros 11N mode=station-wds ssid="XXX" frequency=2304
band=2ghz-onlyg channel-width=20mhz scan-list=2304
wireless-protocol=nv2-nstreme-802.11 wds-mode=dynamic
wds-default-bridge=bridge1 wds-ignore-ssid=no default-authentication=yes
default-forwarding=yes default-ap-tx-limit=0 default-client-tx-limit=0
hide-ssid=no security-profile=default compression=no

EDIT: Off-Topic.

how did it locked up - were you able to get in from the Ethernet side? Have you checked if the wireless package was enabled?

You can think of it as of improved nstreme.

Does that mean NV2 will also have packet aggregation into frames (framing policy terminology?) as NStreme does?

For PTP link with 2XR52HN cards & dual pol dish antennas, which will be better throughput: NStreme or NV2?

Ekkas

I expected a strange situation. Maybe bug?

I have 2 Sxt5HnD in a dual-chain ptp link (distance ~100m). I had good performance with nv2. For testing changed to nstreme but because a worst performance switch back to nv2. But after switching back the speed was the half of first test. After that protocol changing cannot solve the problem.

I update firmware to a new 5 version. With nv2 got the good speed. Change to nstreme and back and the situation was the same - half speed. Revert firmware to 4.7 - problem solved, good speed.

The link established by this howto: http://wiki.mikrotik.com/wiki/Transpare ... using_MPLS
After that dual chain is configured and some minor tweaks.

On page 4 of this thread I described a lock-up problem when running nv2 on x86 intel multicore board. I would just like to report that since upgrading to ROS 5.2, I've been able to run nv2 on all my links now for days without any problem and it seems to be working much better than before. No CPU lock-ups since then, so far. Thanks!!

Okay, so i know NV2 WDS is not supported. Not disputing it However,

Are there plans ever for it to be?

NV2 + WDS is supported, I am using it now

nz_monkey...how do you find the performance?

It seems to be quite good, getting ~50mbit UDP send/receive performance from RB433AH to SXT using 802.11n 2x2 with 20Mhz channel.

There were a few issues with WDS + Bridging in the 5.x release candidates but it's good in 5.0 onwards.

Is that a single SXT to the AP? Whats the distance? we see around 27mbit MAX from a client with around 8 other clients on the same AP, all less than 1km from the site.

The AP IS only 1x ant though, vertical omni at around 950mW EIRP

Thats a bwtest to a single SXT client that is connected to a 2x2 sector with about 5 other SXT clients on it.

The range of the clients is between 1 and 3km

very very nice What Freq / Antenna size / Type?

5Ghz with Ubiquiti 120degree sector and R52n-M card

nv2 still going strong here on x86 intel SMP... now on ROS 5.3, but it's been good since 5.2! The download page says x86 nv2 not supported, but they're probably leaving that message there while they test stability. Seems stable to me, though!