802.11ac - Did the FCC kick the chair from underneath Tx Beamforming?

[sfx2000]

Transmit Beamforming (TxBF) adds about 3dB of gain when the AP is lighting up a specific client that is capable... think of it as a searchlight as opposed to a single lightbulb...

Appears that FCC''s stance here with TxBF is that it may go beyond permitted unlicensed band power limits, so they're requiring that the total Tx power is similar to what would be done with omni-directional radiated power.

Weasel words follow:

"For devices that have a single transmitter output driving an antenna, the directional gain is simply that of the antenna. However, devices employing multiple input multiple output (MIMO) technology can achieve additional directional gain, called array gain, beyond the gains of the individual antennas by transmitting signals that are mutually correlated. In some cases this additional gain is intentionally created to improve performance, as is the case when transmit beamforming is employed."

additionally;

"Based on the modeled array gain results as a function of transmit signal bandwidth, the new approach requires the continued use of the 10log(N(sub)ant) formula when computing narrowband array gain for determining the required reductions in power spectral density of transmissions. However, a new formula computes broadband array gain as either 0 dB or 3 dB depending on transmit signal bandwidth and number of transmit antennas. The broadband array gain value is used to compute array gain for determining required reductions in transmit power."

The takeaway here is that TxBF may not have any net benefit for clients that support implicit/explicit beamforming in 802.11 space, and may have a negative impact to those that do not if implemented (dependent on RF/Chipset implementation)

More details here -- https://apps.fcc.gov/oetcf/kdb/forms/FTSSearchResultPage.cfm?switch=P&id=49466

and some good, objective commentary here -- http://blogs.aerohive.com/blog/the-wi-fi-security-blog/did-the-fcc-really-limit-80211ac-beamforming

So for 802.11ac - we might be more looking toward MU-MIMO vs. BeamForming to get the best performance...

 

[thiggins]

Thanks for the post. This isn't new as you can see from the 2013 date on the document. Antenna gain has always been factored into some FCC test limits. Beamforming is just another way to increase gain.

The rule doc also refers to "smart antennas". So that puts a crimp in the plans of some startups who were hoping to get a piece of the high end MIMO router market.

MU-MIMO also uses beamforming. What is your expectation of "best performance" from MU-MIMO?

 

[sfx2000]

Thanks for the post. This isn't new as you can see from the 2013 date on the document. Antenna gain has always been factored into some FCC test limits. Beamforming is just another way to increase gain.

The rule doc also refers to "smart antennas". So that puts a crimp in the plans of some startups who were hoping to get a piece of the high end MIMO router market.

MU-MIMO also uses beamforming. What is your expectation of "best performance" from MU-MIMO?

Tim,

This is a really good question...

I've taken a view that MU-MIMO and TxBF are not mutually inclusive - they aid and enable each other in different ways...

SU-MIMO, or traditional MIMO in pre-AC land - in simplistic terms, SU-MIMO sends redundant information across the 2, 3, or 4 spatial streams in 802.11, with just enough redundancy that you get not only encoding gain (up to 3dB typically), but also more data bits across all spatial streams in a given transmission for the user.

NOTE on MIMO - this is a good place to mention that a 3 stream AP is always three streams - so when a user only supports a single stream (N150 for example) with an N450 Access Point - we just encode the three streams the same - benefit here is that you still get Tx/Rx diversity and additional Rx gain from the extra receivers. Same would apply to 3 stream AP, 2 stream receiver, and so forth - it's still MIMO - terms of art go into SIMO, MISO, and the like.

MU-MIMO - it takes those same spatial streams - and encodes things in a way that one can assign streams, within a single transmission, to individual users - for example, Streams 1 and 2 for User A, Stream 3 for User B, and Stream 4 for User C - and then combine them in the baseband for transmission at the Radio.

In other words - SU-MIMO - all streams one user, MU-MIMO, assigned streams per user.

What this also means that since this is all happening in one transmission - not all users get the same bandwidth - all depends on how many spatial streams are allocated to that user for that transmission - which is ok, as you still gain more capacity for users, although, at a lower rate per user.

Tx Beamforming - as we all know that each spatial stream in practice uses a dedicated RF chain - so a 3*3 MIMO has three transmitters (and up to 3 receivers)

Beamforming on the Transmit Side - we use phase and time delays within the RF portion to "space" out the actual emitted waveform - since each RF chain operates independently, we can change the actual pattern - steer it per se in both Time and Phase relationships - and again, this is the same amount of radiated power, but the pattern instead of being omni-directional, is now, erm, lop-sided (I know, trying to keep this simple, please) in such a way that the received energy at the user is higher, perhaps up to 3 dB or more - and 3dB is quite a bit...

MU-MIMO and TxBF really do complement each other - massively in some cases, as MU-MIMO encodes the streams at a UserLevel, and TxBF can then point the energy to those users.

But you can have one without the other - either way - and this is why in 11ac, we have different settings for these features.

IMHO - we'll probably see MU-MIMO as a standalone feature more often than not once the silicon and drivers are all sorted out - and we'll also see SU-MIMO implementations where TxBF is used without MU-MIMO, as we now see with Wave1.5 and upcoming Wave 2 chipsets.

Hopefully I've answered your very thoughtful question - to the collective forum, feel free to correct me.

sfx

 

[sfx2000]

I'd like to add an additional comment about MU-MIMO and some of the challenges...

Use Case - we've got Alice and Bobby - Alice is on her 3 stream Laptop with great signal close to the AP - Bobby on the other hand, he's on his single stream tablet out in the living room watching YouTube or NetFlix or whatever.

Do we combine both Alice and Bobby into a single four stream MU-MIMO transmisssion - likely not, as the transmission needs to be a modulation scheme that both Alice and Bobby's devices can demodulate - which might be 64QAM, 16QAM or worse, starving Alice at the cost of Bobby - rather, we might want to do two SU-MIMO transmissions instead, each at the appropriate MCS/Modulation schemes, and schedule the transmissions as needed.

So the devil's in the details with implementation of MU-MIMO - as resource scheduling is the challenge - done right, it's transparent to the end users, done wrong, and performance will definitely suffer.

sfx

 

[stevech]

Startup Vivato, many years ago, spend $$$ of R&D to realize one of the first WiFi beam-forming (steered beam) access points/routers. It was to address the high density coverage needs such as arenas. They got the FCC to change the rules a bit, or reinterpret. And that led to some nice patent assets.

Alas, they didn't find much demand for the product - they were ahead of their time. But seemingly still operating at a small scale.

 

[speedingcheetah]

Until the most recent firmware for my Asus AC56u, having beamforming enabled gave better range and performance....now, its the opposite.

Probably just something they changed in the newer wifi driver...but still makes me wonder in the future implementation of this beamfourming tech.

 

[thiggins]

Sfx,

Looks like you know far more about MIMO than I. All I can say is that even thinking about the complexity of MU-MIMO makes my head hurt.

I think it is gonna take awhile to get MU-MIMO right. Even single user beamforming is a highly YMMV technology. All that complexity for a little bitty 3dB gain. Yeesh.

 

[sfx2000]

Tim - I know how you feel - sometimes my head hurts as well...

TxBF - it's been around for some time - it works in a WAN environment on a macro-cellular level - and WAN technologies like WiMax, LTE, and iBurst, where transmissions are scheduled, there is benefit

MU-MIMO - pretty much the same, but this is newer stuff - doing SU-MIMO has tremendous benefit to the overall channel capacity, but like TxBF above, it's very dependent on resource scheduling...

Makes things very challenging when trying to build test cases - which I'm sure you understand - as close in, there's no benefit, and long range there is little benefit, so it's mostly in the middle - and with these features, not only test cases present the challenge, but also for those who have to deploy.

There's a reason why in LTE Space, that folks have instead looked more closely at Small Cells vs. TxBF/MU-MIMO in dense environments.

All tools in the kit...

sfx

 

[azazel1024]

Lets also remember that most consumer and business/enterprise routers already fall well below the maximum EIRP that the FCC caps 2.4/5GHz at. So this should have no impact except maybe on some "high power" routers/access points.

Generally where you start seeing EIRP being pushed are on bridges, which are already resonably directional and don't employ beamforming (nor would they likely do so, as the bridge they are pointing at is stationary).

 

[sfx2000]

Lets also remember that most consumer and business/enterprise routers already fall well below the maximum EIRP that the FCC caps 2.4/5GHz at. So this should have no impact except maybe on some "high power" routers/access points.

Generally where you start seeing EIRP being pushed are on bridges, which are already resonably directional and don't employ beamforming (nor would they likely do so, as the bridge they are pointing at is stationary).

I'm starting to see where consumer grade HW is juiced up enough that they're starting to hit the caps with current gen HW, even without the "John Holmes" style antennae...

Example - Apple's Airport Extreme... Apple actually publishes this...

* Radio output power: 32.5 dBm maximum (varies by country)

That's huge - actually, it's a shedload for an AP, that's about twice of what a client would have had, and converting dBm to watts - it's 1.8 watts approx...

I would think that others in current HW are pretty close to that...

Arms race here - closely skirting the EIRP that is permitted in unlicensed space..

sfx

 

[speedingcheetah]

I'm starting to see where consumer grade HW is juiced up enough that they're starting to hit the caps with current gen HW, even without the "John Holmes" style antennae...

Example - Apple's Airport Extreme... Apple actually publishes this...

* Radio output power: 32.5 dBm maximum (varies by country)

That's huge - actually, it's a shedload for an AP, that's about twice of what a client would have had, and converting dBm to watts - it's 1.8 watts approx...

I would think that others in current HW are pretty close to that...

Arms race here - closely skirting the EIRP that is permitted in unlicensed space..

sfx

They always put the highest number in marketing. And "Varies by country"...to which in most countries, the max limit is much lower than that. Would love to know what country allows nearly 1.8 watts....max I have have seen in firmwares is 200mw.

 

[stevech]

32dBm in a low cost consumer product would (a) cost way too much and (b) likely fail the FCC Part 15 tests.

Must have been a typo. Or marketing liberties with TX power + overstated antenna gain and they quited radiated power, not transmitter power as is commonly done.

Or, more likely, for Apple, it was 32mW, not dBm.

 

[azazel1024]

Yeah, that doesn't sound right. 32.5dBm would put it way over FCC limits. Maximum limits for the FCC are 30dBm radio power. You can hit 36dBm EIRP as the limit with a 6dBi antenna as the maximum EIRP with something fairly omnidirectional.

At best that MIGHT be EIRP and not radio power they are advertising, but that would still be pushing the envelope, especially if it is using something like 2-3dBi antennas internally.

 

[sfx2000]

The thing with EIRP and FCC...

It's over time - in the time domain, squirting over a hot micro-TX Burst can actually still meet overall Tx limits...

it's like formula 1 - it's not what they can do, it's what they cannot do... and I'd like to think that other vendors do the same thing at some level...

 

[stevech]

Yeah, that doesn't sound right. 32.5dBm would put it way over FCC limits. Maximum limits for the FCC are 30dBm radio power. You can hit 36dBm EIRP as the limit with a 6dBi antenna as the maximum EIRP with something fairly omnidirectional.

At best that MIGHT be EIRP and not radio power they are advertising, but that would still be pushing the envelope, especially if it is using something like 2-3dBi antennas internally.

OFDM as in 802.11n and onward, has about 6dB peak to average power ratio. Older was much less. So we have RMS vs. Peak. And the FCC rules for max EIRP vs. antenna beamwidth (as in steered beam) are fuzzy about peak/RMS.