5GHz Testing - why UNII-1 and not UNII-3 channels?

[sfx2000]

@tim - noticed in your reviews for dual-band routers, most of your test reports use the lower side of the available channels - specifically in the UNII-1 band.

I assume you're aware that many AP providers deliberately reduce power in the UNII-1 band by 3dB - this is to reduce interference with other users (think weather and military radars).

From a receiver perspective - not a big deal, most are fairly linear across the 5GHz band sensitivity wise, but dropping 3dB due to channel selections can skew the observed results significantly... 3dB is half-power

Check with the vendors... this would be something worth noting for folks planning on deploying 5Ghz on their networks.

Point of reference - http://en.wikipedia.org/wiki/U-NII

 

[stevech]

It's more complex than "reducing power by 3dB". Radiated power is often measured in dBm (dB alone is a dimensionless number).

In some regulatory regions, including the US (FCC), beginning several years ago, new regulations for certain portions of the 5GHz band:

Max radiated power must be below a certain limit in certain channels, and
automatic dynamic frequency selection (DFS) is required - so the radios adapt to other nearby transmitters.
Additionally, regulations require CSMA/CA (listen before transmitting), whereas in 2.4GHz this is a common courtesy practice but not mandatory per the FCC. IEEE 802.11 defines how to do these things, but only the FCC (or equivalent) makes regulations governing radio transmissions. WiFi is just a marketing alliance, with supposed conformance requirements for transmitters and receivers. The FCC does not recognize WiFi's requirements.

 

[sfx2000]

It's more complex than "reducing power by 3dB". Radiated power is often measured in dBm (dB alone is a dimensionless number).

dBm is more relevant, I'll agree - but typically when RF engineers are comparing 17 dBm to 20 dBm, it's common to refer to the delta as 3 dB - I was trying to keep it casual...

The point remains though - the channels used for review are reduced power at both the AP and the STA, and do not reflect potential performance if not looked at across different channels.

 

[stevech]

yeah, IIRC the US FCC regulations allow higher power for point to point service if the antenna beamwidth is reduced. There's a formula to calculate the max effective radiated power (Pout + antenna gain) versus antenna beamwidth, for the two halves of the band - different regulations.

A 14dBi antenna and Pout of about 100mW (20dBm) hits the legal limit for a typical 60 degree patch/panel, in the less restricted EIRP part of the band.

Curious: if the "subscriber/client" end is narrow-beam, pointed at a base station/access point, and if the base station has a wide beamwidth sector or omni, two sets of regulations apply!

But the issue is DFS and dynamic power control - new regulations. If some system owner nearby files a complaint about interference, the FCC is likely to issue a show cause notice to cease and the burden is on you to prove your equipment is "legal". It can be a big hassle. No jail of fines, just a hassle.

This came about due to interference to radar receivers - FAA or DoD, I've forgotten which.

 

[thiggins]

Channels 1 and 36 have been used for SNB tests since the beginning of the charts. No big science went into the selection.

I'm aware of the higher power limits for the higher 5 GHz channels. At NETGEAR's request, I tried rerunning a few tests for the WNDR3800 at the weaker signal locations and didn't see a significant difference in throughput.

Could be that clients aren't able to boost power significantly in the higher channels.

 

[rhombus]

In the US, the FCC maximum power limits for the UNII bands are:

UNII-1 50mW/17dBm
UNII-2 250mW/24dBm
UNII-2e 250mW/24dBm
UNII-3 1000mw/30dBm

In addition to their lower power limits: UNII-1 is for indoor use only to prevent interference with mobile satellites, and UNII-2 and UNII-2 extended (for indoor or outdoor use) must use Dynamic Frequency Selection to prevent interference with military or weather radars.

You can see that, at lower Tx MCS values especially, the rise in Tx power by switching to a higher band can therefore be considerably more than 3dB. This is not just potential or theory - a lot of general wireless routers reviewed here will actually do this, even if they lack the "high-power" moniker. Therefore, so long as there is a reliable connection at the cell periphery, this should in theory improve downlink results, irrespective of the client's power.

 

[rhombus]

yeah, IIRC the US FCC regulations allow higher power for point to point service if the antenna beamwidth is reduced. There's a formula to calculate the max effective radiated power (Pout + antenna gain) versus antenna beamwidth, for the two halves of the band - different regulations.

A 14dBi antenna and Pout of about 100mW (20dBm) hits the legal limit for a typical 60 degree patch/panel, in the less restricted EIRP part of the band.

Curious: if the "subscriber/client" end is narrow-beam, pointed at a base station/access point, and if the base station has a wide beamwidth sector or omni, two sets of regulations apply!

The quoted power limits are for peak power or conducted power, before antenna gain. They are not EIRP. In the 2.4Ghz band the peak power limit is 1W.

If a device is not for point-to-point operation, then for all bands, this power limit must be reduced by y dB for y dB of antenna gain above 6dBi.

If a device is for fixed, point-to-point, non omni-directional, operation then: in the 2.4Ghz band, the power limit must be reduced by y/3 dB for y dB of antenna gain above 6dBi, and in the UNII-3 band the power limit must be reduced by y dB for y dB of antenna gain above 23dBi.

There is no formula or consideration of specific antenna beamwidths.

So in the UNII-3 band, your 14/15dBi 30-degree panel can run at the full 30dBm/1W and be within legal limits.

 

[stevech]

Yes, I should have said gain instead of beamwidth - but of course, with narrowing beamwidth we get gain (on H or V).


"If a device is not for point-to-point operation, then for all bands, this power limit must be reduced by y dB for y dB of antenna gain above 6dB"

This is what I was referring to by the curious condition of a narrow-beam antenna (high gain) on the subscriber/client end linking to a base station/AP with a low gain sector/omni. This is point-to-point from the client's viewpoint, but point to multipoint from the base station's viewpoint. The regulations seem fuzzy.

 

[rhombus]

This is what I was referring to by the curious condition of a narrow-beam antenna (high gain) on the subscriber/client end linking to a base station/AP with a low gain sector/omni. This is point-to-point from the client's viewpoint, but point to multipoint from the base station's viewpoint. The regulations seem fuzzy.

What's curious or fuzzy about it?

Approval is for one device at a time, not in pairs. If we are doing point-to-point, one of the devices isn't an access point, and it isn't really a base station either. If you're silly enough to design or configure a point-to-point device with a 15dBi omni antenna, then you rightly pay the penalty of a 21dBm power limit on the radio because otherwise you would cause widespread interference. If however you use a 9dBi sector antenna, you can have the full 30dBm power limit because this is not an omni.

 

[sfx2000]

Channels 1 and 36 have been used for SNB tests since the beginning of the charts. No big science went into the selection.

I'm aware of the higher power limits for the higher 5 GHz channels. At NETGEAR's request, I tried rerunning a few tests for the WNDR3800 at the weaker signal locations and didn't see a significant difference in throughput.

Could be that clients aren't able to boost power significantly in the higher channels.

Netgear does have a point though about trying the upper UNII-3 channels, as 3 dB can make a difference between connecting and not.

 

[sfx2000]

yeah, IIRC the US FCC regulations allow higher power for point to point service if the antenna beamwidth is reduced. There's a formula to calculate the max effective radiated power (Pout + antenna gain) versus antenna beamwidth, for the two halves of the band - different regulations.

Actually the FCC power limits are EIRP... and EIRP is a theoretical value based on a single point radiating in a sphere (adding this for the non-engineers of the group)

A 14dBi antenna and Pout of about 100mW (20dBm) hits the legal limit for a typical 60 degree patch/panel, in the less restricted EIRP part of the band.

Curious: if the "subscriber/client" end is narrow-beam, pointed at a base station/access point, and if the base station has a wide beamwidth sector or omni, two sets of regulations apply!

Interesting point - esp. if/when beamforming comes into common practice - I think from a regulatory perspective though, EIRP is a better number to work with.

I think beamforming will be more relevant for the base station, and not necessarily for the client. Not to say that the client couldn't do that though...

But the issue is DFS and dynamic power control - new regulations. If some system owner nearby files a complaint about interference, the FCC is likely to issue a show cause notice to cease and the burden is on you to prove your equipment is "legal". It can be a big hassle. No jail of fines, just a hassle.

DFS and Dynamic Power control are only needed for certain regulatory zones and channels - UNII-3 for example, if legal, is free of those mandates.

This came about due to interference to radar receivers - FAA or DoD, I've forgotten which.

Military and Weather Radars were the primary driver of that requirement, IIRC, but it's been a while since I've looked at that.

 

[rhombus]