HARRIS, WILTSHIRE & GRANNIS LLP | 1919 M STREET | EIGHTH FLOOR | WASHINGTON DC 20036 | T 202 730 1300 | F 202 730 1301

December 16, 2019

Ex Parte Marlene Dortch, Secretary Federal Communications Commission 445 12th Street SW Washington, DC 20554 Re: Unlicensed Use of the 6 GHz Band, ET Docket No. 18-295; Expanding Flexible Use in

Mid-Band Spectrum Between 3.7 and 24 GHz, GN Docket No. 17-183

Dear Ms. Dortch:

Apple Inc., Broadcom Inc., Cisco Systems, Inc., Facebook, Inc., Google LLC, Hewlett Packard Enterprise, Intel Corporation, Microsoft Corporation, NXP Semiconductors N.V., and Qualcomm Incorporated submit the attached analysis in response to AT&T’s filing of November 12, 2019.1

Sincerely,

Paul Margie Counsel to Apple Inc., Broadcom Inc., Cisco Systems, Inc., Facebook, Inc. Google LLC, Hewlett Packard Enterprise, and Microsoft Corporation

Encl.

1 See Letter from Michael P. Goggin, AT&T Services, to Marlene H. Dortch, Secretary, Federal Communications Commission, ET Docket No. 18-295 (filed Nov. 12, 2019).

ATTACHMENT

Correcting the Record on RLAN-FS Interactions

1

AT&T Includes a Series of Errors in Its Recent Analysis That Render Its Filing Unreliable

1. Two incorrect assumptions exaggerate interference in all AT&T examples by more than 1000%.

2. Again and again, AT&T’s examples include mistakes in geometry and FS antenna discrimination that render its conclusions faulty.

3. AT&T fails to disclose significant clutter and other propagation losses in multiple examples.

4. Even ignoring all of these errors, the real-world availability of AT&T’s links would be unaffected by RLANs.

5. AT&T does not consider near-field effects, even when it asserts an RLAN device would be near an FS receiver.

2

Just Two Assumptions Cause AT&T to Exaggerate All of Its Interference Predictions by More Than 1000%

Two incorrect assumptions caused AT&T to exaggerate the risk of interference by 9.27 dB in every example.

RLAN antenna gain — AT&T’s analysis assumes maximum EIRP in the direction of the FS receiver, even though the record clearly shows that RLAN antennas do not exhibit significant gain towards the horizon. ◦ The diagram AT&T provides—which is taken out of context from a Broadcom filing—is of a single dipole antenna, not

a complete RLAN system.◦ Even AT&T’s own graphic, however, shows no more than 1 dBi in antenna gain directed towards the horizon, resulting

in a total EIRP of 25 dBm, not the 30 dBm that AT&T used—a 5 dB reduction.◦ Real-world antenna pattern mismatch from a multi-antenna system would likely result in even less gain in the

direction of FS receivers.

Bandwidth mismatch — AT&T assumed 80 MHz RLAN bandwidth for all RLAN devices thereby inflating PSD.◦ 160 MHz channels will be typical in the 6 GHz band. (In fact, making 160 MHz channels available for unlicensed

applications is a fundamental goal of the 6 GHz proceeding.) Accounting for this reduces the predicted interferenceby 4.27 dB.*

3*4.27 reflects the difference between the 3 dB bandwidth mismatch value included in AT&T’s analysis, and 7.27 dB, the correct value for bandwidth mismatch between a 30-MHz wide FSchannel and a 160-MHz wide RLAN channel.

Many AT&T Examples Use Incorrect FS Antenna Discrimination ValuesMany AT&T examples also misstated FS antenna discrimination either due to inaccurate latitude and longitude, inaccurate antenna or terrain heights, or the selection of inappropriate antenna patterns, which resulted in using lower antenna discrimination, thus overestimating potential RLAN interference.

In an extreme case, Example 5 (WQWA497), a site visit revealed that the antenna indicated by AT&T is incorrect, the height used in AT&T’s calculation is incorrect, or both. Because of these clear errors, we are not able to analyze the example further.

4

AT&T’s Across-the-Board Assumption of Free-Space Loss Is Unsupportable

Our analysis conservatively assumed free-space propagation for our analyses involving high-rise buildings.

However, AT&T improperly extended the use of free-space propagation to any case where there is line of sight from a building to an FS receiver, ignoring the effects of clutter.

AT&T even does so where the RLAN-to-FS path appears to be completely blocked by foliage.

In addition, in some examples the path from FS receiver to low -rise building passes near to the ground, increasing the likelihood of RLAN attenuation due to multipath interference.*

* See, e.g., V. Erceg, et al., An Empirically Based Path Loss Model for Wireless Channels in Suburban Environments, IEEE Journal on Selected Areas in Communication, vol. 17 no. 7 (July 1999) (reporting results by AT&T researchers showing attenuation significantly greater than free-space in suburban areas, even with line of sight).

5

AT&T Example 1A: KPV20AT&T Value

(dB)Corrected Value

(dB)Change

(dB)RLAN/FS Antenna

Discrimination 0 -5.00 -5.00

Antenna Discrimination -36 -37.22 -1.22

Clutter 0 0.00 0

Propagation -96.45 -96.14 0.31

Bandwidth Mismatch (160 MHz) -3 -7.27 -4.27

Noise Figure -3.2 -5.00 -1.80

Feeder Loss 2.00 2.00 0

BEL (50% ile) -32.75*-32.75 --

I/N -4.00 -16.18 -12.18

6

AT&T’s Predicted I/N is 12.18 dB too high.

Records indicate the target building was recently renovated using energy efficient construction.

In addition to across-the-board errors in RLAN/FS antenna discrimination and bandwidth mismatch, this example also has miscellaneous errors in noise figure, FS receiver antenna discrimination, and propagation loss (which was slightly too high).

AT&T also overlooks cabling loss in its 3.2 dB noise figure.

* AT&T did not include specific BEL values. However, the value above reflects the 50th percentile BEL for a thermally efficient building at this angle, consistentwith the distribution of BEL values in AT&T’s presentation.

AT&T Example 1B: KPV20AT&T Value

(dB)Corrected Value

(dB)Change

(dB)RLAN/FS Antenna

Discrimination 0 -5.00 -5.00

Antenna Discrimination -38 -48.00 -10.00

Clutter 0 0.00 0.00

Propagation -93.35 -92.86 0.49

Bandwidth Mismatch (160 MHz) -3 -7.27 -4.27

Noise Figure -3.2 -5.00 -1.80

Feeder Loss 2.00 2.00 0

BEL (50% ile) -17.00* -17.00 --

I/N 12.85 -7.73 -20.58

7

AT&T’s Predicted I/N is 20.58 dB too high.

Errors are similar to example 1A. Here, however, we assume traditional (i.e., not thermally efficient) building construction.

In addition to other across-the-board errors, this example was affected by a significant error in FS receiver antenna discrimination/interfering geometry.

* AT&T did not include specific BEL values. However, the value above reflects the 50th percentile BEL for a traditional building at this angle, consistent withthe distribution of BEL values in AT&T’s presentation.

AT&T Example 2: WQPJ679AT&T Value

(dB)Corrected Value

(dB)Change

(dB)RLAN/FS Antenna

Discrimination 0 -5.00 -5.00

Antenna Discrimination -1.5 -2.538 -1.04

Clutter 0 -25.00 -25.00

Propagation -118.96 -118.92 0.04

Bandwidth Mismatch (160 MHz) -3 -7.27 -4.27

Noise Figure -3.2 -3.20 0.00

Feeder Loss 0 0 0

BEL (50% ile) -17.00* -17.00 --

I/N 20.44 -14.83 -35.27

8

AT&T’s Predicted I/N is 35.27 dB too high.

AT&T overlooks the fact that the purportedly interfering path passes through stands of trees (25 dB clutter loss assumes branches but no leaves).

The target building appears to be made of brick, with very small windows, making 17 dB BEL conservative.

Note that in private homes, RLAN BEL will be greater than average due to typical placement of RLAN APs.

* AT&T did not include specific BEL values. However, the value above reflects the 50th percentile BEL for a traditional building at this angle, consistent withthe distribution of BEL values in AT&T’s presentation.

9

Tree cover obstructing interfering path in AT&T example 2.

AT&T Example 3:WQXC429AT&T Value

(dB)Corrected Value

(dB)Change

(dB)RLAN/FS Antenna

Discrimination 0 -5.00 -5.00

Antenna Discrimination -0.9 -0.678 0.222

Clutter 0 0.00 0.00

Propagation -121.88 -122.34 -0.46

Bandwidth Mismatch (160 MHz) -3 -7.27 -4.27

Noise Figure -3.2 -3.2 0.00

Feeder Loss 0 0 0

BEL (50% ile) -32.00* -32.00 --

I/N 4.02 -5.49 -9.51

10

AT&T’s Predicted I/N is 9.51 dB too high.

In this case AT&T’s result is almost totally driven by their erroneous bandwidth and antenna mismatch assumptions.

Structure is a late model (c. 2002) manufactured home. HUD has required all such structures to meet energy efficiency standards since the 1990s.

Given the extremely low look angle, and the clutter around the home, we expect more attenuation than FSPL.

* AT&T did not include specific BEL values. However, the value above reflects the 50th percentile BEL for a thermally efficient building at this angle, consistentwith the distribution of BEL values in AT&T’s presentation.

AT&T Example 4: WLU230AT&T Value

(dB)Corrected Value

(dB)Change

(dB)RLAN/FS Antenna

Discrimination 0 -5.00 -5.00

Antenna Discrimination -38 -47.165 -9.165

Clutter 0 0 0.000

Propagation -92.84 -92.90 -0.060

Bandwidth Mismatch (160 MHz) -3 -7.27 -4.270

Noise Figure -3.2 -5.00 -1.800

Feeder Loss 0 -2.00 -2.00

BEL (50% ile) -18.46*-18.46 --

I/N 12.00 -10.3 -22.3

11

AT&T’s Predicted I/N is 22.3 dB too high.

This site has multiple frequencies so feeder loss should apply. AT&T also overlooks cabling loss in its 3.2 dB noise figure.*

In addition to other across-the-board errors, this example was affected by a significant error in FS receiver antenna discrimination.*AT&T has said that cabling/feeder loss should be included for asystem capable of operating on multiple frequencies. Because thissite is capable of operating on multiple frequencies, we havetherefore applied a conservative 2 dB value for feeder loss.Likewise, for systems with cabling loss, 5 dB is a more appropriatenoise figure.

* AT&T did not include specific BEL values. However, the value above reflects the 50th percentile BEL for a traditional building at this angle, consistent withthe distribution of BEL values in AT&T’s presentation.

AT&T’s Examples Demonstrate the Robustness of These Links, Even in Extreme Corner Cases

AT&T’s analysis demonstrates that even if there were an unusual corner case where interference could be possible, real-world FS links would not be affected.

Accepting all of AT&T’s incorrect assumptions and using a conservative mean BEL value (assuming non-energy efficient construction in all cases), the predicted level of interference would not meaningfully affect availability or performance, as explained on the next slide.

All of the links AT&T presents have extremely robust fade margins—and the one exception, Example 4, has a diversity antenna, making it even more robust.

12

Even with AT&T’s Incorrect Assumptions, and Ignoring Energy Efficient Construction, Availability Remains Greater than Five-9s for Each Link.

13* Assumes 50 percentile BEL and non-energy efficient construction.† Analysis assumes interference affects both diversity antennas simultaneously.

Example AT&T I/N (dB)*

Diversity† Availability Without Interference (%)

Availability With Interference (%)

Throughput Reduction with Interference (%)

1A 11 Yes 99.999+ 99.999+ 3 ∙ 10-13

1B 13 Yes 99.999+ 99.999+ 5 ∙ 10-13

2 21 No 99.999+ 99.999+ 0.000006

3 19 No 99.999+ 99.999+ 0.000133

4 12 Yes 99.999+ 99.999+ 0.0000003

5 17 No 99.999+ 99.999+ 0.0000003

AT&T Does Not Consider Near-Field EffectsAT&T asserts that some examples involve hypothetical RLAN transmitters very close to the FS receiver. At such short distances, larger antennas, such as those used in AT&T’s examples, exhibit near-field rejection.

This is caused by antenna designs that are optimized to receive signals transmitted far away receiving nearby signals less efficiently.*

Although exact near-field rejection values vary depending on the antenna design and precise interference geometry, these effects are likely to increase attenuation.

However, neither we nor AT&T have included this effect in our analyses.

14* See, George Kizer, Digital Microwave Communications at 296-307 (2013).