DFS Nano M5

[TJA]

Hi Guys,


I found out the DFS of the Nano M5 is disable on default. I not sure whether the DFS in Nano M5 is really work in the real time environment. Can anybody give an advice on it ?

Mike.

[CaptainProton]

hey..

ubnt does not care alot about DFS or TPC or other weird regulatory requirements.

officially, there is a DFS option. But what kind of DFS...
at least not suitable for european regulation.

[lncommunications]

hey..

ubnt does not care alot about DFS or TPC or other weird regulatory requirements.

officially, there is a DFS option. But what kind of DFS...
at least not suitable for european regulation.

Confirmation from Mike @ UBNT (see one of my earlier posts) DFS & TPC are implemented in the 'M' gear and is OK for European use.

[UBNT-Mike.Ford]

Confirmation from Mike @ UBNT (see one of my earlier posts) DFS & TPC are implemented in the 'M' gear and is OK for European use.

Correct.

Thanks,

Mike

[CaptainProton]

I am going to check on this...

[CaptainProton]

When M Products were released I ordered a pair of bullet 5MHP.

I tried to configure them but very soon I found, that these units propably do not comply with ETSI 301 893.

I thought that this might have changed with a new firmware. I tried it again with the new firmware but still....

As far as I understand DFS, after booting the master device, it shall check all channels for availability and radar and then select one channel by itself. All client (slave) units have to wait for the master device to tell them which channel has been selected for service.

An Ubnt M-device in AP-Mode does not(!) select the channel by random from the (scanned as) available channels. You can choose the channel on your own.

That is a strong indicator, that maybe DFS is implemented somehow but definetly not the way it is demanded by european regulation.

About TPC I dont know. But until DFS is supported, it does not matter. It is illegal in europe anyway...



ETSI 301 893 about DFS:

Master devices:

a) The master device shall use a Radar Interference Detection function in order to detect radar signals.
b) Before initiating a network on a channel, which has not been identified as an Available Channel, the master device shall perform a Channel Availability Check to ensure that there is no radar operating on the channel.
c) During normal operation, the master device shall monitor the Operating Channel (In Service Monitoring) to ensure that there is no radar operating on the channel.
d) If the master device has detected a radar signal during In Service Monitoring, the Operating Channel is made unavailable. The master device shall instruct all its associated slave devices to stop transmitting on this (to become unavailable) channel.
e) The master device shall not resume any transmissions on this Unavailable Channel during a period of time after a radar signal was detected. This period is referred as the Non Occupancy Period.

Slave devices:

f) A slave device shall not transmit before receiving an appropriate enabling signal from a master device.
g) Before transmitting on an Operating Channel (which is defined by the Master), a slave device which is required to perform radar detection (see table D.3) shall perform its own Channel Availability Check immediately after power-up to ensure that there is no radar operating on that channel.
h) A slave device shall stop all its transmissions whenever instructed by a master device to which it is associated. The device shall not resume any transmissions until it has again received an appropriate enabling signal from a master device.
i) A slave device which is required to perform radar detection (see table D.3), shall stop its own transmissions if it has detected a radar. The Operating Channel is made unavailable for the slave device. It shall not resume any transmissions on this Unavailable Channel for a period of time equal to the Non-Occupancy Period. The slave device shall perform its own Channel Availability Check before re-using the channel.

[CaptainProton]

ok, another nice test:

DFS does not only mean detection of radar pulses and patterns but also distributing wireless devices evenly on the available spectrum.

If one AccessPoint detects another AccessPoint on the same channel, it has to select a different channel.

So take a Ubnt 5M device, select AP Mode, an european country, enable DFS and select a certain channel eg. 140 = 5700 Mhz.

Make the same configuration on another Ubnt 5M device.

Turn both devices on and look what happens. If DFS is implemented correctly, one of the devices should select a different channel then 140 on its own.

I tried it, it does not happen.

Sorry Mike, but I am not convinced!

[lncommunications]

...I hope they have an explanation for this or else all my stock is going back!!!

[rconaway]

Before you have a conniption fit, keep in mind that the unit hasn't been certified in that band yet. At least not in the U.S. I'm sure it got put farther back than the Atheros chipset bug that was locking the unit up.

[CaptainProton]

they dont have, because they dont care for this topic.
but if you ask, hey say that everything is nice and supported.

Ubnt really has nice products, but I cannot use them. I am currently trying some airmax5 units from airlive as CPE. They may be not as fancy as the ubnt devices, but at least they are fully complying with 802.11h.

For the backhaul I always use Lancom and Mikrotik.

I would love to be able to use UBNT, but...

[whitedot]

If the legacy products were compliant then surely it's just a case of bringing the required AirOS features across from 3.x to 5.x?

[lncommunications]

They ain't by the sounds! we never trusted the 5GHz stuff for that reason and have stuck with 2.4GHz with bloody good spectral planning and arrangements.

Some people use the 5GHz legacy and AirMax kit in the UK ask them how they implement ATPC and DFS so they conform to EN specs.

[whitedot]

Pretty sure Bill Lewis (ie Kijoma Broadband) has been using the 5GHz legacy kit and also M stuff here in the UK.

[whitedot]

No comment from UBNT on this?

[kijoma]

hi,

We use it and implement TPC and DFS at the server side, its radio manufacturer agnostic then. we started doing that when we used Tranzeo years ago as they didn't implement it properly and still don't.

All we essentially need is access to the noise floor and SNR data.. anything else is a bonus (such as scan data). PHP scripts do the rest.

It was a worthwhile coding exercise as i don't think any "in radio" implementation of DFS/TPC works properly to IR2007

DFS is a tricky one to implement but if your aerials are surrounded by loads of military radar and primary Band C use especially then you have to play nice else you get a visit :icon_eek:

[kijoma]

They ain't by the sounds! we never trusted the 5GHz stuff for that reason and have stuck with 2.4GHz with bloody good spectral planning and arrangements.


I have yet to see a 2.4ghz provider who complies with the 20dBm EiRP spec.

It is a silly band to use for anything "carrier class" as there are 3 channels, one of those is shared by microwave cookers, and with the rest , shared by bluetooth, video senders and a bucket of unlicensed gadgets.

And then you have the wifi AP's..

Our rule is. 2.4GHz is for in building use, 5 G is for everything else.. Has been since we started (before they officially announced the UK band c plan :icon_mrgreen: ).

I notice a "competitor" has popped up near to us and they are using 5.18 GHz for outdoor use... naughty!!, that is asking for trouble!

cheers

[whitedot]

Thanks for the info Bill. Are they complicated, the methods you implement?

[lncommunications]

I have yet to see a 2.4ghz provider who complies with the 20dBm EiRP spec.

By all means pop in for a cuppa and ill show you one ;) when we have nice big valleys to play in and houses with 3 foot walls not many microwave ovens can blast through them to interfere on the outside....though a few illegal chinese CCTV cameras have started to pop up :@

[Airwip]

All we essentially need is access to the noise floor and SNR data.. anything else is a bonus (such as scan data). PHP scripts do the rest.


How you can detect radar by reading out the SNR and Noisfloor of the radio?
And even if... Firmwares like AirOs3 have removed backroundscannig in the driver they use...

to me it sounds everything else than easy to do it at serverside ..as long you dont know what the radio is doing .

[CaptainProton]

as you can see, no comment from ubnt.

When reading the post of Mike earlier on this thread, saying that the units fully comply with european regulation, I wonder why UBNT isn't putting this important feature to its product descriptions...

[lncommunications]

How you can detect radar by reading out the SNR and Noisfloor of the radio?

Good question!?

When reading the post of Mike earlier on this thread, saying that the units fully comply with european regulation, I wonder why UBNT isn't putting this important feature to its product descriptions...

Mike can you or your software guys step in here and give us some definative answers, surely you cannot ignore this as Ubiquiti are selling this kit in Europe! and from a number of historic posts you confirmed its AOK to use in Europe!

Now if you are working on it brilliant! but if they are never going to be developed and spec'd to European Regs can I return my whole stock of 5M devices (100's of units) to Ubiquiti?

[UBNT-Mike.Ford]

Hey Guys,

I talked with my compliance team and am just relaying the information:

M series 5GHz have been tested to the ETSI EN 301 893 V1.4.1 (2007-07) (Article 3.2 of R&TTE Directive) by a compliance body. This standard does include DFS functionality and radar signature detections. Because a substantial amount of resources went into testing, we prefer not to post publicy and make available to competitors.


So if anyone has serious concerns, email me direclty and we can send a copy of the report off line.

mike@ubnt.com

Thanks,

Mike

[whitedot]

Thanks Mike. Email sent.

[jp498]

If these can get FCC 5.4 approved in the next 6 months, I'd buy $10000 of them at twice the going price if necessary. If my project gets approval, I could use $20000 of them at twice the price. 5.4 is that valuable to me since there are zero affordable options for it in the US.

[wirelesspacman]

Hi Mike,

Sorry, but DFS does not work as it should. It needs a scan list attached to the AP so that it knows which channels to use when assessing the 5G frequency bands. The reason for this is that the different sub-bands (A, B and C) have different regs attached to them. Thus a WISP needs to decide which sub-band a particular link is going to use and then use DFS on that sub-band.

For example, here in the UK Band A is (a) only very low eirp (23dBm) and (b) is not legal for WISP use anyway. Band B has a higher eirp limit (30dBm). Band C has the highest eirp limit (36dBm) but has to have part of it "notched out".

It is great that all of the Band C frequencies are now available for selection in the v5.1 of the firmware, but without a scan list on the AP we are still not able to deploy - which we are dying to do as the devices are otherwise brilliant!

Is it that much of a task to implement a scan list in the AP? I would have thought not, but admit I am speaking from ignorance here! :-)

The relevant ETSI standard for sub-Band C is EN 302 502 v1.2.1:2008

Section 4.6.1 states:

"All devices in a FWA deployment shall employ full Dynamic Frequency Selection (DFS) functionality to detect transmissions from other systems and to avoid co-channel operation with these systems, notably radar systems."

For UK, relevant standard for sub-Band C is IR 2007, which in turn references EN 302 502

http://www.ofcom.org.uk/radiocomms/ifi/tech/interface_req/uk_interface_2007.pdf

The relevant ETSI standard for sub-Bands A and B is EN 301 893 v1.5.1:2008-12

Section 4.7.1 states:

"A RLAN shall employ a Dynamic Frequency Selection (DFS) function to:
• detect interference from radar systems (radar detection) and to avoid co-channel operation with these systems;
• provide on aggregate a near-uniform loading of the spectrum (Uniform Spreading)."

For UK, relevant standard for sub-bands A and B is IR 2006, which in turn references EN 301 893

http://www.ofcom.org.uk/radiocomms/ifi/tech/interface_req/uk2006.pdf

best regards

Peter

[Tbird]

Just to make sure I fully understand you, wirelesspacman - is it something like this (see graphic) you mean..?

Tb

[wirelesspacman]

Hi Tbird,

Yes I guess so. Personally I would prefer something more generic, where you just include the channels you would like included (as with Mikrotik's RouterOS).

[wirelesspacman]

Be nice to have an update from Ubiquiti on the lack of DFS issue. As I explained above, these devices remain non-legal for use in the EU until a suitable implementation of DFS is provided. If Ubiquiti do not intend to provide such functionality, then I for one (and I am sure many others) would appreciate knowing.