Hello,
according to ETSI
https://www.etsi.org/deliver/etsi_en/301800_301899/301893/02.01.01_60/en_301893v020101p.pdf
section 4.2.3.2.2, table 2
Note 3 states:
>Slave devices without a Radar Interference Detection function shall
>comply with the limits for the frequency range 5 250 MHz to 5 350 MHz.
And Tx power limits are defined as following:
5150 to 5350: 20 dbm
5470 to 5725: 27 dbm
Which means that if STA device can not do radar detection, it must use
20dbm Tx powers on all channels (can not use 27 dbm limit).
Looking at regdb
https://git.kernel.org/pub/scm/linux/kernel/git/sforshee/wireless-regdb.git/tree/db.txt,
power limit for frequency range 5470 to 5725 is defined at 27 dbm.
Question is: does wireless core assumes that each device can do radar
detection in slave modes (eg acting as a STA) and it is enabled by
default? I couldn't find any logic in kernel which would limit 27 dbm
power to 20 for STA devices.
Hi Igor,
On Tue, 2018-12-11 at 02:30 +0000, Igor Mitsyanko wrote:
> Hello,
>
> according to ETSI
> https://www.etsi.org/deliver/etsi_en/301800_301899/301893/02.01.01_60/en_301893v020101p.pdf
> section 4.2.3.2.2, table 2
> Note 3 states:
> >Slave devices without a Radar Interference Detection function shall
> >comply with the limits for the frequency range 5 250 MHz to 5 350 MHz.
>
> And Tx power limits are defined as following:
> 5150 to 5350: 20 dbm
> 5470 to 5725: 27 dbm
> Which means that if STA device can not do radar detection, it must use
> 20dbm Tx powers on all channels (can not use 27 dbm limit).
>
>
> Looking at regdb
> https://git.kernel.org/pub/scm/linux/kernel/git/sforshee/wireless-regdb.git/tree/db.txt,
> power limit for frequency range 5470 to 5725 is defined at 27 dbm.
I guess somebody misinterpreted the spec, or some countries are less
strict?
> Question is: does wireless core assumes that each device can do radar
> detection in slave modes (eg acting as a STA) and it is enabled by
> default? I couldn't find any logic in kernel which would limit 27 dbm
> power to 20 for STA devices.
No, we shouldn't assume that it can do radar detection by itself ...
I guess we should have some code? Or just fix the regdb?
johannes
On 1/15/19 5:45 AM, Johannes Berg wrote:
>> Question is: does wireless core assumes that each device can do radar
>> detection in slave modes (eg acting as a STA) and it is enabled by
>> default? I couldn't find any logic in kernel which would limit 27 dbm
>> power to 20 for STA devices.
>
> No, we shouldn't assume that it can do radar detection by itself ...
>
> I guess we should have some code? Or just fix the regdb?
>
> johannes
>
>
Maybe we have to do both, as there are multiple things to consider:
- current regdb values are fine for AP mode
- Tx power values can be 3 dbm higher if TPC is supported. This is
mentioned in a comment in regdb, but not used anywhere.
- if STA detects radar, non-occupancy period must start
- when non-occupancy period elapses, STA must do CAC before returning to
channel. I guess CAC must be triggered by wpa_supplicant?
I'm not sure how to present additional information in regdb while
preserving backwards compatibility. Maybe we can:
1. Have a separate rule marked with NO_RDETECT flag which will advertise
lower Tx power. Linux wireless core will have to select rule with
highest Tx power if possible, for better results.
2. For TPC 3dbm gain, have a flag TPC_GAIN
As an example, AW rules will look like:
country AW: DFS-ETSI
(2402 - 2482 @ 40), (20)
(5170 - 5250 @ 80), (20), AUTO-BW, TPC_GAIN=3
(5250 - 5330 @ 80), (20), DFS, AUTO-BW, TPC_GAIN=3
(5490 - 5710 @ 160), (27), DFS, TPC_GAIN=3
(5490 - 5710 @ 160), (20), DFS, NO_RDETECT, TPC_GAIN=3
Linux wireless core will have to update Tx power values when switching
from AP and STA modes, and somehow notify drivers.
Hello Igor and Johannes,
From my research around TPC and radar detection in the context of the BG
regulatory domain and respectively ETSI, the relevant regulatory rules are more
specific than both what can currently be expressed in the regdb and what will be
possible to be expressed with your suggested modifications.
For example, in [1] it is stated that for the 5470-5725 MHz band:
* The maximum allowed transmission power is 1 W e.i.r.p. with a maximum of 50
mW/MHz spectral density of the average e.i.r.p. for each 1 MHz band.
* The use of TPC that ensures lowering the average e.i.r.p. of the entire
system (as I understand it, this means both the AP and STAs) of at least 3
dbm is required.
* In case TPC (as I understand it -- that exhibits the parameters above) is
not used, both the maximum allowed transmission power and maximum spectral
density of the average e.i.r.p. are lowered by 3 dB.
* The use of methods for limiting radio interference ensuring at least the
described in BDS 301 893 (respectively ETSI 301 893) protection
for providing
coexistance with radio radar systems.
If there is will to extend the regdb format to be able to express accurately and
in their entirety the specifics of the relevant regulations, IMO a wider and
more detailed discussion is in order.
[1] http://www.crc.bg/files/_bg/Spisak_2015.pdf - List of radio equipment that
uses harmonized within the European Union bands and electronic
communications terminal equipment (the List)
Best regards,
Petko
On Wed, Jan 16, 2019 at 6:00 AM Igor Mitsyanko
<[email protected]> wrote:
>
> On 1/15/19 5:45 AM, Johannes Berg wrote:
> >> Question is: does wireless core assumes that each device can do radar
> >> detection in slave modes (eg acting as a STA) and it is enabled by
> >> default? I couldn't find any logic in kernel which would limit 27 dbm
> >> power to 20 for STA devices.
> >
> > No, we shouldn't assume that it can do radar detection by itself ...
> >
> > I guess we should have some code? Or just fix the regdb?
> >
> > johannes
> >
> >
>
> Maybe we have to do both, as there are multiple things to consider:
> - current regdb values are fine for AP mode
> - Tx power values can be 3 dbm higher if TPC is supported. This is
> mentioned in a comment in regdb, but not used anywhere.
> - if STA detects radar, non-occupancy period must start
> - when non-occupancy period elapses, STA must do CAC before returning to
> channel. I guess CAC must be triggered by wpa_supplicant?
>
>
> I'm not sure how to present additional information in regdb while
> preserving backwards compatibility. Maybe we can:
> 1. Have a separate rule marked with NO_RDETECT flag which will advertise
> lower Tx power. Linux wireless core will have to select rule with
> highest Tx power if possible, for better results.
> 2. For TPC 3dbm gain, have a flag TPC_GAIN
> As an example, AW rules will look like:
>
> country AW: DFS-ETSI
> (2402 - 2482 @ 40), (20)
> (5170 - 5250 @ 80), (20), AUTO-BW, TPC_GAIN=3
> (5250 - 5330 @ 80), (20), DFS, AUTO-BW, TPC_GAIN=3
> (5490 - 5710 @ 160), (27), DFS, TPC_GAIN=3
> (5490 - 5710 @ 160), (20), DFS, NO_RDETECT, TPC_GAIN=3
>
> Linux wireless core will have to update Tx power values when switching
> from AP and STA modes, and somehow notify drivers.
> _______________________________________________
> wireless-regdb mailing list
> [email protected]
> http://lists.infradead.org/mailman/listinfo/wireless-regdb
You could simplify the entries for most of the CEPT area by coding in a
common reference to the ECC/DEC/(04)08 instead of trying to express
every detailed restriction from every national regulation.
There is absolutely no reason to make the regulatory database more
complicated than what you find in https://www.erodocdb.dk/document/381
and the associated implemetation status on
https://www.efis.dk/matrixviewer.jsp?annex=18
The CEPT members can and should be expected to keep the ERO database
updated. Duplicating that job is unnecessary and only causes errors.
The national regulations are supposed to be implementations of the
ECC/DEC/(04)08. Any differences should be explained and documented in
the ERO database, or they can be considered translation errors.
So there should not be much difference between most of the CEPT
countries for the 5150-5350 and 5470-5725 MHz bands, according to the
current ECC/DEC/(04)08 implementation status.
Just my .02 €
Bjørn
Petko Bordjukov <[email protected]> writes:
> Hello Igor and Johannes,
>
> From my research around TPC and radar detection in the context of the BG
> regulatory domain and respectively ETSI, the relevant regulatory rules are more
> specific than both what can currently be expressed in the regdb and what will be
> possible to be expressed with your suggested modifications.
>
> For example, in [1] it is stated that for the 5470-5725 MHz band:
>
> * The maximum allowed transmission power is 1 W e.i.r.p. with a maximum of 50
> mW/MHz spectral density of the average e.i.r.p. for each 1 MHz band.
>
> * The use of TPC that ensures lowering the average e.i.r.p. of the entire
> system (as I understand it, this means both the AP and STAs) of at least 3
> dbm is required.
>
> * In case TPC (as I understand it -- that exhibits the parameters above) is
> not used, both the maximum allowed transmission power and maximum spectral
> density of the average e.i.r.p. are lowered by 3 dB.
>
> * The use of methods for limiting radio interference ensuring at least the
> described in BDS 301 893 (respectively ETSI 301 893) protection
> for providing
> coexistance with radio radar systems.
>
> If there is will to extend the regdb format to be able to express accurately and
> in their entirety the specifics of the relevant regulations, IMO a wider and
> more detailed discussion is in order.
>
> [1] http://www.crc.bg/files/_bg/Spisak_2015.pdf - List of radio equipment that
> uses harmonized within the European Union bands and electronic
> communications terminal equipment (the List)
>
> Best regards,
>
> Petko
>
> On Wed, Jan 16, 2019 at 6:00 AM Igor Mitsyanko
> <[email protected]> wrote:
>>
>> On 1/15/19 5:45 AM, Johannes Berg wrote:
>> >> Question is: does wireless core assumes that each device can do radar
>> >> detection in slave modes (eg acting as a STA) and it is enabled by
>> >> default? I couldn't find any logic in kernel which would limit 27 dbm
>> >> power to 20 for STA devices.
>> >
>> > No, we shouldn't assume that it can do radar detection by itself ...
>> >
>> > I guess we should have some code? Or just fix the regdb?
>> >
>> > johannes
>> >
>> >
>>
>> Maybe we have to do both, as there are multiple things to consider:
>> - current regdb values are fine for AP mode
>> - Tx power values can be 3 dbm higher if TPC is supported. This is
>> mentioned in a comment in regdb, but not used anywhere.
>> - if STA detects radar, non-occupancy period must start
>> - when non-occupancy period elapses, STA must do CAC before returning to
>> channel. I guess CAC must be triggered by wpa_supplicant?
>>
>>
>> I'm not sure how to present additional information in regdb while
>> preserving backwards compatibility. Maybe we can:
>> 1. Have a separate rule marked with NO_RDETECT flag which will advertise
>> lower Tx power. Linux wireless core will have to select rule with
>> highest Tx power if possible, for better results.
>> 2. For TPC 3dbm gain, have a flag TPC_GAIN
>> As an example, AW rules will look like:
>>
>> country AW: DFS-ETSI
>> (2402 - 2482 @ 40), (20)
>> (5170 - 5250 @ 80), (20), AUTO-BW, TPC_GAIN=3
>> (5250 - 5330 @ 80), (20), DFS, AUTO-BW, TPC_GAIN=3
>> (5490 - 5710 @ 160), (27), DFS, TPC_GAIN=3
>> (5490 - 5710 @ 160), (20), DFS, NO_RDETECT, TPC_GAIN=3
>>
>> Linux wireless core will have to update Tx power values when switching
>> from AP and STA modes, and somehow notify drivers.
>> _______________________________________________
>> wireless-regdb mailing list
>> [email protected]
>> http://lists.infradead.org/mailman/listinfo/wireless-regdb
>
> _______________________________________________
> wireless-regdb mailing list
> [email protected]
> http://lists.infradead.org/mailman/listinfo/wireless-regdb