HARRIS, WILTSHIRE & GRANNIS LLP | 1919 M STREET NW | 8TH FLOOR | WASHINGTON DC 20036 | T 202 730 1300 | F 202 730 1301

June 29, 2020

BY ELECTRONIC FILING Marlene H. 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:

Enclosed, please find comments in response to the Commission’s Further Notice of Proposed Rulemaking in the above referenced proceedings.1

Sincerely,

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

1 See Unlicensed Use of the 6 GHz Band, Expanding Flexible Use in Mid-Band Spectrum

Between 3.7 and 24 GHz, Report and Order and Further Notice of Proposed Rulemaking, 35 FCC Rcd. 3852 (2020).

Before the FEDERAL COMMUNICATIONS COMMISSION

Washington, D.C. 20554

________________________________ )

In the Matter of ) )

Unlicensed Use of the 6 GHz Band ) ET Docket No. 18-295 )

Expanding Flexible Use in Mid-Band ) GN Docket No. 17-183 Spectrum Between 3.7 and 24 GHz )

) _________________________________ )

COMMENTS OF APPLE INC., BROADCOM INC., CISCO SYSTEMS, INC., FACEBOOK, INC., GOOGLE LLC,

HEWLETT PACKARD ENTERPRISE, INTEL CORPORATION, MICROSOFT CORPORATION, NXP SEMICONDUCTORS, QUALCOMM INCORPORATED, AND

RUCKUS NETWORKS, A BUSINESS SEGMENT OF COMMSCOPE

June 29, 2020

i

Table of Contents

I. INTRODUCTION AND SUMMARY ............................................................................................1 II. THE COMMISSION SHOULD PERMIT PORTABLE 14 dBm VERY LOW POWER

OPERATIONS IN EACH OF THE 6 GHZ U-NII SUB-BANDS. ....................................................4 A. Very Low Power Portable Devices Are Central to the Success of the 6 GHz

Band and U.S. Leadership in Wireless Technologies. .............................................4 B. VLP Devices Authorized to Operate at No Less Than 14 dBm EIRP and

1 dBm/MHz PSD Will Support Reliable Consumer Use Cases and Fully Protect Incumbents from Harmful Interference. ..........................................10 1. VLP devices require the ability to operate at a minimum of 14 dBm

EIRP in a range of channel bandwidths to meet expected consumer use cases, overcome on-body loss, and permit minimum throughput, latency, and power efficiency requirements. ...................................................10

2. In considering permitted power levels, the Commission should recognize that key aspects of portable VLP device operation fundamentally differ from fixed LPI operation. ..............................................16

3. The Commission can authorize VLP operation at 14 dBm EIRP, with a 1 dBm/MHz PSD limit, without increasing the risk of harmful interference to incumbent services. ..................................................................23

III. THE COMMISSION SHOULD PERMIT MOBILE STANDARD-POWER ACCESS POINT OPERATIONS GOVERNED BY AUTOMATED FREQUENCY COORDINATION. ...........................32 A. Innovators Need 6 GHz Mobile Standard-Power APs to Meet Growing

Mobile Wi-Fi Demand and Enable Future Advances. ...........................................33 1. Internet access on public transportation ...........................................................34 2. Industry-specific unlicensed mobile connectivity applications .......................36 3. AP-to-AP mobile mesh applications ................................................................38

B. Reasonable Operating Rules for Mobile Standard-Power APs in the U-NII-5 and U-NII-7 Bands Will Protect Incumbents and Promote Investment. ...............44 1. AFC systems can accommodate a wide range of mobile and portable

operations. ........................................................................................................45 2. The Commission should model the mobile standard-power AFC rules

on its existing personal/portable white space device rules, with certain important modifications. ..................................................................................49

C. Mobile Standard-Power Access Point Operations Will Not Negatively Impact AFC Systems. ........................................................................................................52 1. The Commission can structure mobile standard-power AFC rules as an

optional supplement to the existing fixed AFC rules. ......................................52

ii

2. Mobile standard-power operations will not cause congestion to AFC systems. ...................................................................................................54

IV. THE COMMISSION SHOULD PERMIT LOW-POWER INDOOR ACCESS POINTS TO OPERATE UP TO 8 dBm/MHZ PSD. ....................................................................................56 A. Operation at 8 dBm/MHz PSD is Important to Meet Clear Consumer Needs. .....57 B. The Record Demonstrates that LPI Access Points Operating at 8 dBm/MHz

PSD Will Not Cause Harmful Interference to FS Operations. ..............................59 C. The Record Demonstrates that LPI Access Points Operating at 8 dBm/MHz

PSD Will Not Cause Harmful Interference to BAS, ENG, or CARS Operations. .............................................................................................................61

V. CONCLUSION .......................................................................................................................63

1

I. INTRODUCTION AND SUMMARY

The Commission’s recent Report & Order (“R&O”) opening the 6 GHz band to

unlicensed use lays a strong foundation for wireless innovation in the United States.1 In addition

to creating the nation’s most important new unlicensed band, it establishes two fixed device

classes that will bring consumers stronger and better unlicensed service. However, the R&O

stops short of enabling access to the 6 GHz band for portable and mobile access point (“AP”)

technologies—which are critical to the consumer wireless experience. While the R&O therefore

represents a historic forward step for Wi-Fi and other unlicensed technologies, without

portability and mobility, the 6 GHz band will not be able to reach its potential to facilitate the

Commission’s goal of supporting “ubiquitous connectivity to a full range of services regardless

of location” and “secur[ing] U.S. leadership in the next generation of wireless services.”2

Fortunately, the Further Notice of Proposed Rulemaking (“FNPRM”) recognizes the

importance of opening the 6 GHz band to mobile and portable technologies. Our companies

support the steps outlined in the FNPRM to allow (1) portable very-low-power (“VLP”)

unlicensed operations in each of the 6 GHz U-NII sub-bands and (2) mobile standard-power AP

operation subject to automated frequency coordination (“AFC”) in the U-NII-5 and -7 bands.3

We also support (3) increased power spectral density (“PSD”) for low-power indoor (“LPI”)

operations throughout the 6 GHz band.

1 Unlicensed Use of the 6 GHz Band, Report & Order and Further Notice of Proposed

Rulemaking, 35 FCC Rcd. 3852 (2020) (“6 GHz R&O and FNPRM”). 2 Id. ¶ 1. 3 In these comments, “portable” refers to VLP devices not subject to AFC control and “mobile

standard-power” refers to APs in motion that are controlled by an AFC.

2

Permitting portable VLP operations at 14 dBm EIRP will support a host of immersive,

real-time applications in healthcare and education, dynamic and innovative gaming experiences,

and augmented-reality/virtual-reality (“AR/VR”) devices, among other uses. As Chairman Pai

recently explained, this device class will facilitate “a new and innovative generation of personal

area network technologies with low latency, high capacity, and all-day battery life,” and will

enable “accessibility technology for Americans with disabilities, virtual reality gaming,

augmented reality glasses, in-vehicle systems, and other emerging technologies.”4 This power

level represents the lowest level at which manufacturers can design VLP devices that can reliably

provide consumers with the minimum throughput and latency requirements needed for expected

applications in the range of likely operating environments.

These comments describe these applications and explain why VLP devices require a

minimum 14 dBm EIRP power level. In support of the VLP recommendation, we commissioned

testing to provide comprehensive on-body over the air measurement results and analysis of the

associated body loss distributions applicable to VLP. These comments also provide a new

statistical analysis provided by RKF Engineering Solutions, LLC (“RKF”) specifically tailored to

VLP operation in the 6 GHz band showing that the risk of causing harmful interference to

licensed services is exceedingly small, and that any real-world corner cases where this could

even hypothetically occur are exceedingly unlikely. Importantly, RKF’s technical analysis

contains sensitivity analysis across multiple factors, including a showing that VLP devices

operating at up to 21 dBm EIRP will not cause harmful interference to licensed incumbent

4 Ajit Pai, Chairman, FCC, Remarks to the Broadband India Forum Webinar Celebrating

World Wi-Fi Day, 3 (June 19, 2020), https://docs.fcc.gov/public/attachments/DOC-365033A1.pdf.

3

services. Furthermore, we demonstrate that the Commission could allow VLP devices to operate

with a PSD based on a 20-megahertz bandwidth.

Mobile standard-power operation will be equally important to the success of the 6 GHz

band. The Commission should enable mobile standard-power APs under AFC control with

operating rules that will facilitate innovation, while not increasing the risk of harmful

interference. As is the case for fixed outdoor APs, AFCs can ensure that mobile APs do not

transmit co-channel in any exclusion zone. Although the AFC-authorized operating frequencies

may in general be fewer in number for mobile standard-power APs compared to indoor APs, this

tradeoff is justified by the importance of mobility to the success of the band. Furthermore, the

process of governing mobile standard-power APs will not cause congestion to fixed AFC

systems or delay their rollout. To effectuate this change, the personal/portable white space

devices framework should provide the basis for the mobile AFC rules, with implementations

tailored for mobile AFC use in the 6 GHz band.

Finally, to ensure that fixed (non-mobile) LPI unlicensed operations in the 6 GHz band

are able to satisfy consumer and enterprise use cases, the Commission should adopt a PSD limit

of 8 dBm/MHz for LPI devices to allow 6 GHz Wi-Fi to achieve coverage areas and

performance levels comparable with today’s Wi-Fi solutions. This increased PSD limit will not

cause harmful interference to Fixed Service (“FS”) links, as ample evidence in the record

demonstrates. LPI operations at 8 dBm/MHz will also protect Broadcast Auxiliary Services

(“BAS”), Electronic News Gathering (“ENG”), and Cable Television Relay Service (“CARS”),

as demonstrated by technical analyses from multiple parties.

4

II. THE COMMISSION SHOULD PERMIT PORTABLE 14 dBm VERY LOW POWER OPERATIONS IN EACH OF THE 6 GHZ U-NII SUB-BANDS.

The Commission’s 6 GHz R&O opened 1200 MHz of spectrum for fixed unlicensed

devices. The FNPRM recognizes that the Commission should now consider permitting portable

operations and proposes to create a new VLP device class in the band. The use cases enabled by

VLP operation will facilitate important portable applications that consumers demand and will

produce substantial economic benefits for the country. In authorizing this new device class, the

Commission should establish a power limit of no less than 14 dBm EIRP and 1 dBm/MHz PSD,

which is the lowest possible limit for manufacturers to produce devices that reliably can deliver

minimum throughput and latency levels needed to support anticipated use cases. As

demonstrated by RKF’s new VLP-specific statistical analysis, and new engineering studies on

device operation and expected body loss, the Commission can certainly authorize 14 dBm EIRP

VLP devices while maintaining robust protection from harmful interference for incumbent

licensees.

A. Very Low Power Portable Devices Are Central to the Success of the 6 GHz Band and U.S. Leadership in Wireless Technologies.

As the Commission recognizes, portable 6 GHz devices will produce significant

advantages for consumers. The FNPRM explains that “[t]hese devices can usher in new ways

that Americans work, play, and live by enabling applications that can provide large quantities of

information in near real-time.”5 Wireless VLP connectivity will power a wide range of

5 See 6 GHz R&O and FNPRM ¶ 235; see also Ajit Pai, Chairman, FCC, Remarks at the Wi-Fi

Alliance Virtual Membership Meeting, 2 (June 2, 2020), https://docs.fcc.gov/public/attachments/DOC-364693A1.pdf (“Chairman Pai Remarks to Wi-Fi Alliance”) (“Very-low-power devices could enable a new and innovative generation of personal area network technologies with low latency, high capacity, and all-day battery life. These very-low-power devices could include accessibility technology for Americans with

5

applications, including displays for AR/VR, automotive applications, screen mirroring, wearable

and on-body uses, short-range hotspots, indoor location and navigation, and automation. Further,

non-traditional portable device use is growing across a breadth of settings, including in

education, industry, offices, and homes. Digitally immersive experiences across these settings

require tetherless VLP connectivity, and the multiple 160- and even 320-megahertz wide

channels available for the first time in the 6 GHz band will enable such capabilities. Further, the

benefits of portable VLP devices are driven primarily by their lightweight and low-cost form

factors.

Economic benefits. The use cases facilitated by a VLP device class will produce

enormous economic benefits. The most recent economic study to consider the effect of opening

the 6 GHz band to unlicensed use concluded that a VLP device category “will enable the

deployment of a new generation of AR/VR solutions yielding an overall producer surplus of

$13.74 billion for US firms selling hardware, software, and content in the US market between

2020 and 2025.”6 “More importantly,” the report explains, “the diffusion of AR/VR solutions

among US enterprises will yield a spillover contribution to the GDP equivalent to $25.78 billion

between 2020 and 2025.”7 The portability of AR/VR uses and applications is central to these

economic benefits. The study explained that AR/VR technologies will have spillover effects on

productivity ranging from “improved training to the acceleration of product design and

delivery,” and can, for example, “help warehouse workers provide parts information for

disabilities, virtual reality gaming, augmented reality glasses, in-vehicle systems, and other emerging technologies.”).

6 Raul Katz, Assessing the Economic Value of Unlicensed Use in the 5.9 GHz & 6 GHz Bands, 6 (Apr. 2020), http://wififorward.org/wp-content/uploads/2020/04/5.9-6.0-FINAL-for-distribution.pdf (“Katz Study”).

7 Id.

6

engineers and technicians in the field.”8 Further, recent market research indicates that

functionalities like in-building navigation and asset tracking in warehouses and manufacturing

facilities will result in an indoor location services market valued at $17 billion by 2025.9

Form factors. In the near term, VLP devices using 6 GHz spectrum will primarily rely on

the portable form factors in use today as semiconductor and equipment makers introduce more

functionality into existing types of devices.10 In the longer term, we expect the innovation

spurred by VLP’s additional wireless capabilities to create novel portable device

implementations. 6 GHz portable form factors will include smartphones, watches, headphones,

glasses, goggles and headsets, keyboards, game controllers, medical devices, and tablets.

Portable 6 GHz operational capability will allow form factors in use today to provide the

untethered, real-time capabilities that consumers increasingly expect from all of their wireless

devices. Additionally, we expect that original equipment manufacturers (“OEMs”) will add value

and flexibility for consumers by leveraging devices originally designed for client use with fixed

devices by adding portable hotspot use, peer-to-peer connectivity, ranging, and other capabilities.

Thus, the first 6 GHz portable VLP devices will use today’s existing wireless and portable form

factors but will add high-bandwidth, low-latency capability and open up more use cases at a

higher level of quality. As Commissioner Carr noted, although some assume the next big thing

8 Id. at 54. 9 Markets and Markets, Indoor Location Market by Component (Hardware, Solutions, and

Services), Deployment Mode, Organization Size, Technology, Application, Vertical (Retail, Transportation and Logistics, Entertainment), and Region - Global Forecast to 2025 (May 2020), https://www.marketsandmarkets.com/Market-Reports/indoor-location-market-989.html?gclid=EAIaIQobChMIsvakj9T35gIVGoiGCh1GoQPwEAAYASAAEgKNPfD_BwE.

10 See 6 GHz R&O and FNPRM ¶ 235 (asking “[w]hat form factors will be most useful for performing everyday activities?” and “[w]ill very low power functionality be built into existing devices such as cell phones or will they be standalone devices?”).

7

may “just be a faster version of what we have today,” we cannot always comprehend “the more

visionary applications that are right around the corner.”11

Use cases. We expect a wide range of VLP use cases if the Commission permits

operation at a minimum of 14 dBm EIRP. For example, portable devices have nearly limitless

healthcare use cases, including in-room connections between a handheld ultrasound device and a

tablet or viewer, or a peripheral patient measurement device such as a pulse oximeter and a

patient monitor. In many cases, hospitals and other healthcare facilities could use portable

systems that they can bring into a facility or room, irrespective of the presence of fixed

infrastructure. VLP devices therefore are well positioned to assist doctors by facilitating access

to patient histories or reference materials during treatments, projecting high resolution displays

for real-time analysis, and enabling hands-free IV medication delivery.

Automotive applications and location-based services will also rely on VLP devices. In

automotive settings, we expect VLP portable devices to be used for device-to-device streaming,

mobile AR, immersive location, vehicle settings, and infotainment. All of these subsystems can

be connected wirelessly, reducing cabling inside the vehicle.12 For location-based services, the

160-megahertz wide channels in the 6 GHz band will enable a 50% improvement in location

accuracy compared to ranging over 80-megahertz channels, enabling a host of proximity-based

services such as indoor location and navigation, onboarding, unlocking, and exhibits.13

11 Id. at Statement of Commissioner Brendan Carr. 12 A typical U.S.-built car contains approximately one mile of cabling and more than 50 pounds

of copper. See John Sprovieri, Wire Harness Recycling, Assembly Magazine (July 1, 2014), https://www.assemblymag.com/articles/92263-wire-harness-recycling. Reducing wiring in a vehicle can help reduce both weight and environmental impact.

13 See Sinan Gezici, et al., Localization via Ultra-Wideband Radios, IEEE Signal Processing Magazine, July 2005, at 72 (describing the relationship between channel bandwidth and location accuracy in equation 2). Lightweight 6 GHz-enabled accessories that support high data rates and more accurate location can be ideal companions for narration and explanation

8

Finally, these devices also will be central to provisioning digitally immersive services

such as AR/VR, ultra-HD audio, and gaming. Delivering these real-time, immersive services will

become even more important due to the ongoing 5G transition and will allow device users to

experience seamless connectivity regardless of network type. AR/VR devices will be used to

enhance collaboration, training, and productivity for U.S. companies using AR/VR business

applications.14 Consumers will use AR glasses as mobile peripherals to access “head-up”

displays and alerts in real time. The applications for VR technologies continue to expand, and

include emergency rescue simulations, memory care, and human resources training.15

Additionally, the demand for high-definition (“HD”) VR gaming is growing rapidly, along with

the technology and VR experiences available to consumers, but requires wider channels and

additional spectrum resources to succeed. 6 GHz VLP devices would therefore be at the core of

enabling wireless, real-time AR/VR capabilities.

U.S. wireless leadership and international harmonization. Authorizing portable devices

at 14 dBm EIRP will allow the U.S. to maintain its leadership in the 6 GHz band as portable

device authorization progresses in jurisdictions around the world. Although the EU is behind the

U.S. in authorizing the 6 GHz band for unlicensed use, it is clearly set to authorize portable VLP

in a museum or art exhibit. Location accuracy enables the service to begin at the appropriate time based on proximity to the points of interest.

14 See Cisco, Cisco Annual Internet Report (2018-2023) White Paper, 33 (Mar. 9, 2020), https://www.cisco.com/c/en/us/solutions/collateral/executive-perspectives/annual-internet-report/white-paper-c11-741490.pdf.

15 Oculus, Mine Rescue Teams Discover a New Tool for Training, https://www.oculus.com/vr-for-good/stories/mine-rescue-teams-discover-a-new-tool-for-training/ (last visited June 28, 2020); Dougal Shaw, How Virtual Reality Is Helping People with Dementia, BBC News (Sept. 13, 2019), https://www.bbc.com/news/av/business-49654052/how-virtual-reality-is-helping-people-with-dementia; Emma Kennedy, How VR Is Transforming HR, CNN Business (Feb. 26, 2019), https://www.cnn.com/2019/02/26/tech/vr-transforming-hr-intl-biz-evolved/index.html.

9

6 GHz operations at 14 dBm EIRP, with a PSD of either 10 dBm/MHz or 1 dBm/MHz 16—

making portable VLP devices the core of the European 6 GHz framework. Based on a multi-year

engineering analysis, EU stakeholders have determined that VLP will not cause harmful

interference to incumbent services in the 6 GHz band, because these devices operate at such low

powers and people spend the vast majority of their time indoors.17 The EU’s sharing analysis

relies on a comprehensive Monte Carlo simulation to assess the probability of exceeding long-

term and short-term interference criteria assessed in terms of an interference-to-noise (“I/N”)

threshold. The simulation assumes a maximum power level of 14 dBm for VLP devices, a fixed

body loss value of 4 dB, and that VLP devices transmitting outdoors would comprise 1% of total

expected radio local area network (“RLAN”) usage in the 6 GHz band.18 The study concludes

that authorizing 14 dBm VLP would not meaningfully increase the risk of harmful interference,

as compared to LPI devices. Additionally, South Korea has also recently announced that it plans

to authorize 6 GHz portable operations at 14 dBm EIRP power levels.19 The Administrative

Notice issued by the South Korean Regulator intends to authorize VLP operations at 14 dBm

EIRP with a maximum power spectral density of 1 dBm/MHz.20

16 See CEPT Elec. Commc’ns Comm., ECC Report 316: Sharing studies assessing short-term

interference from Wireless Access Systems including Radio Local Area Networks (WAS/RLAN) into Fixed Service in the frequency band 5925-6425 MHz (May 2020)(“ECC Report 316”), https://www.ecodocdb.dk/download/8951af9e-1932/ECC%20Report%20316.pdf.

17 See id. at 7, 11, 29. 18 See id. at 7–8, Tbl. 1, 11, Tbl. 6. 19 See South Korean Ministry of Science and ICT, Administrative notice of amendment of

technical standards for radio equipment for radio stations that can be opened without reporting (June 26, 2020), https://msit.go.kr/web/msipContents/contentsView.do?cateId=_law4&artId=2942268.

20 See id.

10

B. VLP Devices Authorized to Operate at No Less Than 14 dBm EIRP and 1 dBm/MHz PSD Will Support Reliable Consumer Use Cases and Fully Protect Incumbents from Harmful Interference.

The Commission should authorize a power limit of no less than 14 dBm EIRP and 1

dBm/MHz PSD to allow the VLP device class to achieve the throughput and latency

requirements necessary to support immersive, real-time use cases across expected user

configurations. In considering the appropriate power level for VLP devices, the Commission

should recognize that VLP devices differ in several key ways from LPI devices. This is not only

because VLP devices can operate outdoors, but also because of the dynamic variability in clutter

losses encountered, the impact of itinerancy, body loss variations, dynamic transmit power

control, and battery life optimization.

1. VLP devices require the ability to operate at a minimum of 14 dBm EIRP in a range of channel bandwidths to meet expected consumer use cases, overcome on-body loss, and permit minimum throughput, latency, and power efficiency requirements.

Operation at up to 14 dBm EIRP and 1 dBm/MHz PSD is essential for VLP devices to

reliably accomplish the use cases and applications that consumers require, due to the relationship

between allowed EIRP, power consumption, on-body and other losses, data rate, latency, duty

cycle, device form factors, and cost.

A particularly important factor for portable devices is optimizing the tradeoff between

desired user experience and power consumption. In the presence of on-body loss and other

expected losses, a lower maximum power level will result in lower data rates. Lower data rates—

which can occur on existing Wi-Fi bands that tend to have congestion and small channel

bandwidth—translate into increased latency and higher duty cycles. Because power consumption

increases with duty cycle, these higher duty cycles undermine the ability to achieve low power

consumption, which is critical for the small-form factor battery-power-limited devices

envisioned to operate in the VLP class. Latency is also a critical factor for anticipated VLP

11

applications such as AR/VR, screen mirroring, gaming and other time-sensitive (i.e., display-

based) applications: higher data rates reduce the on-air time required for transmission and thus

reduce total latency.

In the context of this discussion, it is critical to distinguish data rate—referring to the

instantaneous rate of communication (i.e., the physical layer data rate)—from throughput—

which is the time-average of the data rate (i.e., rate at which information bits are conveyed over

periods of many seconds or longer). Even for applications that require relatively low throughput,

such as highly compressed standard definition video, high data rates allow such applications to

operate with low latency, low duty cycle, and low power consumption, because each burst of

data (for example, 30 frames per second) can be transmitted almost instantaneously. The same

advantages are also afforded to higher throughput applications and use cases, including latency-

sensitive, immersive applications that cannot afford the additional latency of high-compression

ratio data formats.

A power limit of 14 dBm EIRP will allow VLP links to achieve the high data rates

necessary for low latency and low power consumption. Depending on the use case, immersive

and interactive applications require roundtrip latencies as low as 5-20 milliseconds. The precise

throughput and latency requirements vary based on the nature of the application. For example, a

non-immersive application, such as peer-to-peer HD video calling, requires a download

throughput of approximately 1.5 Mbps to achieve a latency target of 50 milliseconds or less with

1080p video quality at 30 frames per second.21 High data rates allow this type of use case to be

21 See GSMA, GSMA Cloud AR/VR Whitepaper, § 3.1.2.3 (last updated Apr. 26, 2019),

https://www.gsma.com/futurenetworks/wiki/cloud-ar-vr-whitepaper/; Microsoft, Prepare your organization’s network for Microsoft Teams, at “Bandwidth requirements” (last updated June 24, 2020), https://docs.microsoft.com/en-us/microsoftteams/prepare-network.

12

supported with an extremely low duty cycle and level of power consumption. However, a higher

interactivity application, such as HD AR/VR, requires latency of 20 milliseconds or less, a frame

rate of 60 frames per second, and 10-bit pixel depth, resulting in required throughput of

approximately 400 Mbps.22 This type of application is essentially infeasible on existing Wi-Fi

platforms, but is realistic with the multi-Gbps data rates that would be made possible by the VLP

device class.

Our companies, which include leading product experts and engineers on Wi-Fi and

portable devices and applications, agree that a minimum of 14 dBm EIRP is critical to balance

the tradeoffs between latency, data rate, power consumption, and other essential factors required

to enable useful consumer products and cutting-edge innovative applications envisioned for the 6

GHz band.

On-body loss. The range of potential on-body loss scenarios is a central factor driving the

required power for VLP devices.23 In configurations without significant on-body loss (e.g., when

a VLP device is placed on a table or in handheld mode at close distances to the peer device to

which it is connected), the VLP device would employ dynamic transmit power control (“DTPC”)

to operate at far less than 14 dBm. However, in configurations involving significant on-body loss

between the VLP AP and client device (e.g., when a VLP device is inside a purse or backpack,

and a wearable peripheral is located on the opposite side of a user’s body), a VLP device will

require 14 dBm to achieve minimum link performance requirements. Over-the-air measurements

22 Simone Mangiante et al., VR is on the Edge: How to Deliver 360° Videos in Mobile

Networks, § 3 Tbl. 1 (Aug 25, 2017), available at https://www.researchgate.net/publication/319049968_VR_is_on_the_Edge_How_to_Deliver_360_Videos_in_Mobile_Networks.

23 On-body loss refers to the impact of body proximity to the VLP device antenna gain patterns, i.e., the radiation pattern of the signal as it penetrates into the body, radiates around the body, and travels on the surface of the body.

13

confirm that on-body loss between the VLP AP and client device, and therefore total signal loss,

can vary significantly depending on the use case device positions, static versus dynamic

condition, orientations, and the physical characteristics of the human body. Measurements

conducted by the North Carolina Wireless Research Center to test the links between wireless

glasses worn on a user’s face and a body-worn VLP AP device located at various positions

(handheld, waist, back pocket, and backpack) and device orientations on multiple human users

show that total dynamic position path loss ranges from 26 to 96 dB.24 The nomadic nature of

VLP devices—i.e., the ability to move around freely and therefore change the configuration of

the VLP link—results in highly variable path loss. VLP devices need not contend with maximum

path losses at all times, and will therefore use DTPC to preserve battery life in many scenarios,

while still successfully closing the link for a particular application. However, device

manufacturers must design devices that will function successfully in the worst-case operating

scenarios. The North Carolina Wireless Research Center’s on-body loss measurements show that

even with 14 dBm EIRP, VLP devices may not be able to function for a small percentage of loss

scenarios.

Comparing the effects of different amounts of on-body loss and antenna mismatch loss

demonstrates the relationship between transmit power and throughput. In a scenario with, for

example, only 7 dB of on-body loss and antenna mismatch (approximately 59 dB of total path

loss25), a 14 dBm VLP device will be able to achieve an expected throughput of approximately

24 See Attachment B, Wireless Research Center of North Carolina, On-Body Channel Model

and Interference Estimation at 5.9 GHz to 7.1 GHz Band at 5, Fig. 15 (June 2020) (“Wireless Research Center of North Carolina Report”). “Dynamic position path loss” is a function of polarization loss, antenna pattern mismatch, distance, and on-body loss.

25 Total path loss includes free space path loss at 1 meter, 3 dB polarization mismatch loss, antenna pattern mismatch loss, and on-body loss.

14

1840 Mbps in a 160-megahertz channel (approximately 1080 Mbps in an 80-megahertz channel).

However, when the on-body and antenna mismatch losses increase to 20 dB (approximately 72

dB of total path loss), throughput is reduced by 37% to approximately 1160 Mbps in a 160-

megahertz channel and 650 Mbps (a 40% reduction) in an 80-megahertz channel (assuming a

power limit of 14 dBm EIRP and 1 dBm/MHz PSD). The ability to achieve a given modulation

and coding scheme (“MCS”)—which determines the number of spatial streams, modulation type,

and coding rate possible for a link—determines whether a VLP link can maintain the necessary

data rates for a given application in the face of variable losses. VLP devices will require at least

14 dBm EIRP levels to achieve the MCS levels that will support core use cases in a variety of

common user configurations.

Power Spectral Density. Importantly, the Commission should adopt a PSD limit of 1

dBm/MHz to avoid unnecessarily constraining power (and therefore decreasing throughput and

increasing latency) in narrower channel sizes, particularly in higher on-body loss scenarios. This

will allow VLP devices to achieve a 14 dBm EIRP power level across 20-, 40-, 80-, 160-, and

320-megahertz wide channels. The effect of restricting PSD to a lower limit of, for example, -5

dBm/MHz or -8 dBm/MHz PSD will negatively affect the ability of VLP devices to use a

smaller bandwidth while still achieving the required throughput levels. For example, limiting the

PSD to -8 dBm/MHz means that a VLP link experiencing 16 dB of combined on-body loss and

antenna mismatch loss (still a relatively low amount), could achieve a throughput of

approximately 1280 Mbps in a 160-megahertz channel, but only approximately 150 Mbps in a

20-megahertz channel. This throughput would be insufficient to support all but the most

rudimentary VLP applications and would result in nearly continuous transmission, which would

drain the device battery. Using a smaller bandwidth can be necessary to maximize channel

15

availability in user-dense and high path loss environments and to achieve additional range where

required. As explained in Section II.B.3, below, limiting PSD for VLP devices beyond 1

dBm/MHz is unnecessary to protect incumbent services from the risk of increased harmful

interference.

Transmission distances. The FCC additionally asks about the distances over which

“transmissions to very low power devices [will] be necessary.”26 VLP transmission distances are

a function of the capability of the link itself; thus, the distances over which transmission will be

practical depend on the achievable throughput given the power limit and the desired application.

With a power limit of 14 dBm, VLP transmissions will occur over distances between one and

three meters, depending on the application. Shorter-range applications over approximately one

meter will include wearables, short-range hotspots, and automation applications such as

unlocking and provisioning. Although these applications are very short-range, as noted above,

they will periodically be subject to very high losses and will require sufficient power to function,

even at short distances. Slightly longer-range applications over distances up to two meters will

include file and video sharing and AR/VR, automotive, and screen mirroring displays, while

indoor location services will operate between three and six meters, depending on the channel

bandwidth and power level.

2. In considering permitted power levels, the Commission should recognize that key aspects of portable VLP device operation fundamentally differ from fixed LPI operation.

The FNPRM requests comment on the interference potential of VLP devices operating

outdoors, stating that one difference between VLP and LPI operations will be the absence of

building entry loss (“BEL”), which was a major factor in the Commission’s analysis of LPI

26 See 6 GHz R&O and FNPRM ¶ 235.

16

operations.27 As discussed below, VLP operations are materially different from LPI operations

and it is not advisable to determine power limits by simply subtracting BEL from the

Commission’s LPI power level.

Preponderance of indoor operation. While some VLP devices will operate outdoors,

most VLP devices will operate indoors at most times. Studies of human activity patterns show

that people spend only an average of approximately 6% of their time outdoors and approximately

4% of their time in cars.28 Furthermore, many of the highest data rate VLP applications will be

used indoors for a substantial portion of the time, because users are unlikely to be outdoors on,

for example, sidewalks or walkways, while immersed in a VR experience. Indoor VLP devices

do experience BEL and will operate at lower power levels than LPI devices, and thus will pose

an immaterial interference risk. For this reason, it makes sense to focus VLP analyses only on the

relatively small number of VLP devices that are outdoors. Furthermore, the Commission should

take into account that outdoor applications are generally less data-intensive than indoor

applications, further reducing the interference risk posed by VLP devices.

Clutter. The FCC adopted propagation models for AFC operation in the 6 GHz band in

the R&O. The Commission’s propagation model uses ITU-R P.2108 for clutter for distances

greater than 1 kilometer, and clutter is included as part of the WINNER-II model for distances

27 See id. ¶ 238. 28 See Environmental Protection Agency, Report on the Environment: Indoor Air Quality,

https://www.epa.gov/report-environment/indoor-air-quality#note1 (last visited June 28, 2020) (“Americans, on average, spend approximately 90 percent of their time indoors . . .”). Of the 10 percent spent outdoors, approximately 4 percent is spent in vehicles. See Federal Highway Administration, 2017 National Household Travel Survey: Summary of Travel Trends, Department of Transportation, 54 (2017), https://nhts.ornl.gov/assets/2017_nhts_summary_travel_trends.pdf.

17

less than 1 kilometer. These models are applicable to VLP devices operating in the 6 GHz band,

and RKF used these models in its study.29

VLP devices operating outdoors will be subject to substantial signal losses in the

direction of incumbent receivers caused by environmental clutter, and VLP devices operating

indoors will additionally be subject to BEL.30 The Commission asked how to account for clutter

losses, including the appropriate statistical distributions.31 Outdoor VLP devices will be carried

by, or will operate in close proximity to, a person. The vast majority of people’s movements are

constrained to pathways adjacent to natural or man-made structures (e.g. sidewalks, roads, and

yards) and VLP device emissions will encounter the persistent presence of some form of

surrounding clutter from these objects.

The effects of clutter on radio-wave propagation have been extensively studied, with

ongoing investigations into additional environments and frequency ranges. Recommendation

ITU-R P.2108 contains a statistical model for clutter loss of terrestrial paths that characterizes

clutter loss near the endpoints of the FS link, beyond 250 meters from the antenna.32 At 6 GHz,

as shown in Figure 1 below, the model yields a median clutter loss value of approximately 21 dB

29 See Attachment A, RKF Engineering Solutions, LLC, Frequency Sharing for Very Low

Power (“VLP”) Radio Local Area Networks in the 6 GHz Band at 19–20 (June 29, 2020) (“RKF VLP Report”).

30 Given the lower power levels associated with VLP devices, combined with the effect of BEL, the interference risk presented by indoor VLP devices is miniscule—even smaller than the insignificant risk posed by VLP devices operating outdoors.

31 6 GHz R&O and FNPRM ¶ 238. In the LPI context, the Commission concluded that a standard clutter model, Recommendation ITU-R P.452-16 (July 2015), should provide the clutter value for a specific link budget calculation, and the Commission chose a value of 18.4 dB of clutter loss for one specific LPI-FS interaction based on that model. See 6 GHz R&O and FNPRM ¶ 128 (iii).

32 International Telecommunication Union: Radiocommunication Sector, Recommendation ITU-R P.2108-0: Prediction of Clutter Loss, Section 3.2 (June 2017), https://www.itu.int/dms_pubrec/itu-r/rec/p/R-REC-P.2108-0-201706-I!!PDF-E.pdf.

18

at 250 meters, up to approximately 31 dB at 1 kilometer away from the FS receiver. Figure 1 also

shows the model output for other percentiles (e.g., “p=10%” or “p=1%”), noting that even in the

case of the extremely low probability 1-percentile locations there is approximately 17 dB of

clutter loss at 1 km.

Figure 1: Clutter Loss for Terrestrial Paths at 6 GHz (ITU-R Rec. P.2108-0)

Many of the perceived hypothetical worst-cases for potential VLP-FS interference

assume a VLP device operating in the same general area of the FS receive antenna. The model in

Recommendation ITU-R P.2108 applies to cases where both ends of the link are far enough from

each other that they cannot be affected by the same clutter; thus, the minimum applicable

distance for Recommendation P.2108 is 250 meters. For scenarios where a VLP device is less

than 250 meters from the FS antenna, a different clutter model is appropriate. The FCC has

adopted the WINNER-II model with probabilistic combining of the line-of-sight and non-line-of-

sight path into a single path-loss for AFC calculations for distances from 30 meters to 1 km in

19

the 6 GHz band. This model can be applied to scenarios where a VLP device is close to the FS

antenna and both are located within the same clutter field (i.e., an area characterized by the same

clutter model). Figure 2 shows the difference between the WINNER-II model (urban macro) and

free space path loss (“FSPL”). This difference contains the contribution of clutter to path losses

below 250 meters. The clutter amount is slightly higher than the Recommendation P.2108 model

and increases with the distance between the VLP device and the FS receiver. It is important to

also note that in the rare situations where a VLP device operates within 250 meters of an FS

receiver, there will be extremely high FS off-axis rejection due to the height differential between

the FS receiver and VLP transmitter.

Figure 2: Clutter Contribution for Distances from 30 to 250 Meters at 6 GHz (WINNER-II)

Operation of VLP devices will in almost all cases involve proximity to humans,

structures, foliage, vehicles, and other obstacles that obstruct signal propagation in predictable

ways. These situations are captured by well-understood statistical models, creating excess losses.

These losses, even in extremely low probability cases, are still large enough to create a favorable

sharing condition for VLP devices in the absence of BEL.

40 60 80 100 120 140 160 180 200 220 240

Distance (meters)

0

5

10

15

20

25

30

35

Clu

tter =

Com

bine

d W

II - F

SPL

(dB)

Additional 6 GHz Clutter in FCC Proposed Pathloss Model Below 250 m

p=50%

p=1%

20

Itinerancy. Additionally, while LPI APs are stationary, VLP devices are itinerant. This is

critical from an incumbent protection perspective because it means any unfavorable corner-case

interference geometries—however unlikely—would generally only be experienced for a short

period of time, due to the natural movement of devices and the accompanying changes in

propagation conditions. Body loss, for example, can attenuate the signal by tens of dBs in the

direction of nearby incumbents. While there is a possibility that body loss in the direction of a

nearby FS link may be more limited in some cases, e.g., if a phone is in a user’s pocket and has a

line of sight path towards an FS receiver, the expected motion of a person means such conditions

would generally only be experienced for a very short period of time.33 The average runner, for

example, takes approximately 180 steps per minute, involving very significant motion every

third of a second. Itinerancy also drives device orientation changes that create another significant

source of randomness for VLP devices. This is especially true for devices such as smartphones

that are commonly rotated in a variety of different ways when placed in bags, purses, or pockets.

Because VLP devices with a small form factor have compact internal antenna construction and

are tightly filled with other metal components and obstructions, small changes in orientation can

have a dramatic effect on the antenna pattern.34 This further contributes to the fleeting nature of

worst-case geometries. A static worst-case analysis that does not account for itinerancy and

orientation change, for example, by merely subtracting BEL from LPI assumptions, would not

accurately capture the potential impact of VLP devices on FS systems.

33 The Monte Carlo simulation described below accounts for this dynamic by using a statistical

distribution for body loss that was derived from measurements of VLP devices. 34 The VLP Monte Carlo simulation also accounts for device orientation in the statistical

distribution.

21

On-Body loss and dynamic transmit power control. It is also important to account for the

fact that, unlike for LPI, DTPC on VLP devices will be used to preserve battery power, thus

reducing the transmit power where it is not required to overcome losses. DTPC is expected to

work in a similar manner as dynamic power control for Bluetooth, where the power is reduced to

achieve only the desired throughput. Calculations for expected MCS and data rate based on the

range of VLP on-body loss measurements indicate that, at the 10th percentile of expected VLP

dynamic position path loss (39.8 dB), DTPC will reduce VLP device power by 17 dB.35 In most

implementations, DTPC will be used to reduce battery drain by optimizing the transmit power to

a level just sufficient to achieve the highest possible MCS for a given path loss. Importantly, the

reduction in power caused by DTPC also reduces the amount of power radiated away from the

VLP link, towards an FS link, which is itself subject to losses caused by the VLP user’s body.36

Testing of actual body-worn VLP devices reveals that the 50th percentile average signal loss

away from a user’s body due to body loss and antenna pattern mismatch loss is 14.3 dB, and the

90th percentile is approximately 4 dB.37

Number of devices. Finally, the number of instantaneously transmitting VLP devices will

differ from the number of instantaneously transmitting devices in the LPI context. The vast

majority of outdoor VLP operations are expected between 7 am and 9 pm, and heavy use of

portable VLP devices is not expected outdoors during the Wi-Fi busy hours of 7 pm to 11 pm

when people are indoors (and if VLP devices are used indoors they will be subject to BEL).

Furthermore, given that people spend only approximately 6% of their time outdoors and not in

35 See Wireless Research Center of North Carolina Report at 8, Fig. 24 (showing a CDF of path

loss between the VLP AP and client). 36 See id. at 8–10. 37 See id. at 9, Fig. 26.

22

vehicles, the Commission can safely assume that the number of VLP devices outdoors is less

than 10% of the total population. Additionally, when people are asleep, most VLP devices are

expected to be dormant. These assumptions about the number of transmitting VLP devices, along

with other considerations regarding VLP specific use cases, are reflected in the inputs for the

RKF Monte Carlo analysis, as described further below. Specifically, consistent with ECC Report

316, RKF assumed that, at any given time, 1% of instantaneously transmitting 6 GHz RLAN

devices will be VLP devices operating outdoors.38 In addition, RKF also conducted a sensitivity

analysis modeling 2x, 3x, 6x and 12x the number of active devices.

3. The Commission can authorize VLP operation at 14 dBm EIRP, with a 1 dBm/MHz PSD limit, without increasing the risk of harmful interference to incumbent services.

A combination of statistical and real-world analyses demonstrates that VLP devices

operating at a 14 dBm EIRP and 1 dBm/MHz PSD power limit across all four 6 GHz sub-bands

will not cause harmful interference to FS links in U-NII-5 and -7 or mobile services in U-NII-6

and -8. To demonstrate this, RKF performed a Monte Carlo Analysis, titled Frequency Sharing

for Very Low Power (“VLP”) Radio Local Area Networks in the 6 GHz Band, to investigate the

impact of VLP devices on incumbent services. A statistical Monte Carlo analysis is appropriate

for the VLP context because it accurately captures the range of potential operating scenarios and

makes predictions based on repeated simulations that reflect different possibilities for each input

factor. RKF’s analysis shows that the likelihood that VLP devices will cause FS links in U-NII-5

and -7 to exceed a conservative -6 dB I/N threshold (which the Commission has recognized is a

threshold that does not represent harmful interference) is exceedingly small, even when the

assumed number of VLP devices increases. Furthermore, the analysis shows that, even in worst-

38 RKF VLP Report at 15.

23

case scenarios, the vast majority of FS links suffer no degradation in actual performance, i.e., FS

carrier availability is not materially affected. Furthermore, field tests recently conducted by

Broadcom and Facebook show that VLP operations can coexist with FS, including in worst-case

geometries that would be improbable in the real world. The RKF analysis also demonstrates that

VLP devices can successfully coexist with mobile BAS operations in U-NII-6 and -8 bands,

which would receive further protection through contention-based protocols if the Commission

extends this rule to VLP devices. The analysis also demonstrates that a PSD limit of 1 dBm/MHz

will not result in an increased risk of harmful interference. Finally, RKF’s sensitivity analysis

also considered VLP operations at power levels up to 21 dBm EIRP and found that they will not

create a risk of harmful interference, showing that there is significant safety margin for VLP

operation at 14 dBm.

Monte Carlo analysis. As the Commission explained in the R&O, statistical Monte Carlo

analyses provide the most accurate prediction of the likelihood of RLAN-caused harmful

interference because they capture “the sporadic nature of [AP] transmissions . . . and the

probabilistic nature of co-channel operation,” and take into account all probabilistic factors.39

The Commission also noted the limitations of a “static link budget analysis,” which does not

fully account for the probabilistic nature of a wide range of assumptions and factors.40 As

explained above, VLP devices are fundamentally different from traditional fixed APs. VLP

devices are designed to conserve battery life by limiting transmission time and power, their

positions can change constantly, attenuation sources such as body loss, antenna pattern mismatch

loss, and clutter are dynamic, and each individual combination of factors and geometry often

39 See 6 GHz R&O and FNPRM ¶¶ 109, 116, 127 40 See id. ¶ 124.

24

only exists for brief moments of time. A static link budget analysis can be helpful to understand

interactions between fixed devices with known, unchanging path loss factors, but for VLP

devices, a link budget analysis will only indicate an instantaneous condition, without indicating

the actual risk of any specific condition occurring. A Monte Carlo analysis, by contrast, is based

on statistical ranges for various power levels, path loss factors, and channel assumptions, and is

therefore a much more appropriate tool to model the range of possible interference risks

associated with VLP devices. In a Monte Carlo analysis, each individual “drop” of RLAN

devices, or “simulation iteration,” represents one possible morphology, and to arrive at a

statistical conclusion, the model must run many iterations.

RKF statistical analysis. Therefore, to effectively model the potential for harmful

interference to FS links resulting from VLP operations, RKF performed a very comprehensive

Monte Carlo analysis to simulate the interaction between outdoor VLP devices and the 97,888

registered FS links in the continental U.S.41 RKF found that, over 500,000 simulation iterations

(100,000 iterations per each of five channel bandwidth models), the probability that an FS link

would experience an I/N level exceeding -6 dB was 0.00011%.42 For an I/N level of 0 dB, the

probability was roughly an order of magnitude lower—only 0.00002%. RKF also tested the

41 RKF VLP Report at 11–14. 42 RKF used -6 dB I/N as a baseline value. Importantly, the Commission explained that a higher

I/N value “create[s] a higher risk of harmful interference (although still very low),” not that harmful interference would actually occur where an FS link momentarily experiences an I/N level over -6 dB I/N. 6 GHz R&O and FNPRM ¶ 132 (emphasis added). As the R&O repeatedly emphasizes, the Commission has not determined that any signal received with an I/N greater than -6 dB would constitute harmful interference, and the -6 dB figure is one way to ensure that the potential for harmful interference is minimized. See id. ¶¶ 130 n.337, 71 (“No commenter provides technical justification for using a particular I/N level as the actual level necessary to protect fixed microwave receivers against harmful interference.”).

25

sensitivity of the occurrence probability to the number of active VLP devices.43 RKF repeated

the baseline simulations, but increased the number of active VLP devices over expected levels by

a factor of two, three, six, and twelve respectively. The -6 dB I/N occurrence probabilities across

all 97,888 FS links increased from 0.00011%, to 0.00022% (2x),0.00033% (3x), 0.00066% (6x),

and 0.00133% (12x).44

Availability analysis. Furthermore, even in improbable worst-case scenarios, with

extremely unlikely geometries, VLP operations will not disrupt incumbent FS operations. To

investigate the impact of VLP devices on specific FS links, RKF further analyzed 1,000 FS links

randomly selected from the population of FS links that experienced occurrences of greater than -

6 dB I/N in any of the 500,000 simulation runs and computed the unavailability, if any, for those

links.45 RKF first started by analyzing these links using an ITU model that projected the

minimum necessary fade margin to achieve 99.999% reliability.46 RKF found that the vast

majority of these FS links experienced no increase in link unavailability due to interference from

VLP devices.47 Next, when RKF took into account the actual link margin based on the highest

order modulation listed in the FCC’s Universal Licensing System (“ULS”) licensing database,

100% of the links studied were unaffected even assuming very low probability I/N occurrences.

This result remained true when RKF increased the number of active VLP devices by up to 12x

43 RKF VLP Report at 39. 44 Id. at 34, Tbl. 5-11. 45 Id. at 30. 46 See id. at 30; see also Propagation data and prediction methods required for the design of

terrestrial line-of-sight systems, International Telecommunication Union: Radiocommunication Sector, Recommendation ITU-R P.530: Propagation data and prediction methods required for the design of terrestrial line-of-sight systems (Dec. 2017), https://www.itu.int/dms_pubrec/itu-r/rec/p/R-REC-P.530-17-201712-I!!PDF-E.pdf.

47 RKF VLP Report at 32.

26

and increased the power to 21 dBm EIRP. As with the baseline simulation, a selection of 1,000

links with momentary occurrences above -6 dB I/N showed that none of the links would

experience any increase in unavailability.48

Field testing. Real-world testing also demonstrates that even in unlikely geometries, VLP

devices will not impact FS link performance. For example, Facebook and Broadcom tested the

impact of a 6 GHz VLP RLAN device on a 13.4 km FS link in the field. This test setup was

intentionally designed to produce a very strong potentially interfering RLAN signal, to test a

highly improbable worst-case scenario.49 First, the FS link between the rooftop of the Northridge

Facebook Office and Oat Mountain (call sign WQ9XHF) was aligned to have a representative

received signal level (“RSL”) at the receiver antennas on both sides.50 Next, a VLP device was

placed 190 meters from the FS receiver. In order to ensure the VLP device would be in the FS

link’s boresight, to test the maximum potential impact on the FS link, the VLP device was

mounted on a pole on top of a van at a height of approximately 10 meters in a parking lot

adjacent to the Northridge building, as depicted in additional detail in the “RLAN configuration”

discussion in Attachment C.51 This positioned the VLP antenna approximately 8 meters below

the FS antenna. Given that VLP devices are expected to be predominantly body-worn, combined

with the typical height above ground of FS antennas, this setup represents a highly improbable

scenario in the real-world. The VLP device was battery powered and programmed to transmit

randomly generated traffic consistently at an artificially high 95% duty cycle, for 24

48 RKF VLP Report at 41–47. 49 See Attachment C, Broadcom and Facebook, VLP Testing: Effect of interference of 6 GHz

VLP RLAN device on FS links (May 2020) (“Northridge Field Study”). 50 See id. at slides 2, 3. 51 See id. at slide 5.

27

hours.52 The VLP EIRP was calibrated to produce an I/N level of 4 dB at the FS receiver

(considerably higher than the -6 dB and 0 dB I/N thresholds that RKF simulated), resulting in a

VLP transmit power of 22.8 dBm. The FS link was calibrated to have only the minimum

available fade margin necessary to achieve 99.999% reliability (less than 10 dBm). Yet even in

this carefully constructed worst-case configuration, the results observed in the field during a 24-

hour period showed no degradation in the measured error rate for the FS link.53 This result

further confirms that VLP RLANs will not increase the risk of harmful interference to FS

operations, including in worst-case geometries that would be improbable in the real world.

VLP in U-NII-6 and -8. Neither will VLP devices cause harmful interference to the

mobile services, including BAS, operating in the U-NII-6 and -8 bands. In the VLP Frequency

Sharing Study, RKF also analyzed representative BAS deployments in the U-NII-6 and -8 bands

to assess the potential for sharing between mobile broadcast 6 GHz incumbents and VLP

devices.54 RKF considered locations in Cowles Mountain, San Diego, CA, and a theoretical

location at the Old Post Office in Washington, DC, based on the identification of those sites by

the National Association of Broadcasters in a previous study.55 The simulation analyzed mobile

links between outdoor ENG trucks to the BAS central receive sites in these two locations, for a

total of six links per site. Using a Monte Carlo simulation, RKF determined that, for the Cowles

52 See Northridge Field Study at slide 5. 53 See id. The error rate was quantified using the modem mean squared error (“MSE”). 54 See RKF VLP Report at 11–14. 55 Mark Gowans and Martin Macrae, Alion Science and Technology, Analysis of Interference

to Electronic News Gathering Receivers from Proposed 6 GHz RLAN Transmitters (Oct. 2019), https://ecfsapi.fcc.gov/file/1205735216211/RESED-20-002_v9.pdf, as attached to Letter from Rick Kaplan, General Counsel and Executive Vice President, Legal and Regulatory Affairs, National Association of Broadcasters, to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183 (filed Dec. 5, 2019).

28

Mountain site, all six configurations had a 0% probability of experiencing an I/N level higher

than -6 dB I/N due to VLP devices, and for the theoretical Old Post Office site, five

configurations had a 0% probability and one configuration had a 0.001% probability.56 When

RKF increased the active VLP device assumption over the expected number (by 2x, 3x, 6x, and

12x), the probabilities for the Cowles Mountain location remained at 0%, and increased only

slightly for the Old Post Office site (to 0.0013% for 12x VLP devices).57 Likewise, increasing

the VLP EIRP to 21 dBm resulted in only slightly increased probabilities.58

In addition, VLP devices can successfully coexist with other mobile incumbent

operations in U-NII-6 and -8 such as CARS (including ENG operations), low-power auxiliary

service (“LPAS”), and other short-range broadcast auxiliary operations. A combination of low

VLP power levels, body loss, antenna pattern mismatch loss, device itinerancy, and a contention-

based protocol requirement, and considering real-world operating characteristics, will protect

these U-NII-6 and -8 incumbent services from harmful interference.

VLP devices operating indoors with a contention-based protocol will operate at lower

power levels than either LPI APs or client devices, both of which the Commission has already

concluded can operate indoors without causing interference.59 VLP devices operating outdoors

will likewise not cause harmful interference, because the effective EIRP of the VLP device after

accounting for body loss will be much lower than even the very small amount of indoor LPI

56 RKF VLP Report at 52. 57 Id. at 54, Tbl. 5-26. 58 Id. at 55. 59 See 6 GHz R&O and FNPRM ¶¶ 158, 166 (finding that “low-power indoor operations will

have little potential of causing harmful interference to ENG operations” and that “the risk of harmful interference to outdoor electronic news gathering receivers from indoor unlicensed devices is negligible”).

29

RLAN power that would reach outdoors after BEL, under the FCC’s rationale in the R&O. The

FCC authorized LPI operations at a power limit of 27 dBm EIRP in a 160-megahertz channel,

and endorsed a value of 20.5 dB for building loss.60 In doing so, the Commission implicitly

concluded that approximately 7 dBm of RLAN energy will not cause harmful interference to

ENG operations outside of buildings. The power radiated in the direction of the ENG receiver

from VLP devices will be far lower than 7 dBm due to body and antenna pattern mismatch

losses, which together average 14.3 dB.61 Further, the itinerancy and changing orientation of

VLP devices, as noted above, will ensure that any problematic VLP-to-ENG orientations are

fleeting, if they ever occur. Indeed, in its discussion of LPI operations, the Commission noted

that “the same conditions that cause signal variations in the electronic news gathering signal will

also act upon a signal from an unlicensed device.”62

Additionally, the features that make BAS, ENG, and LPAS links resilient in general

mean that they will not experience harmful interference from VLP devices operating at

comparatively very low power levels. The Commission relied on studies showing that “camera-

back transmitters deliver high quality video to electronic news gathering trucks at signal-to-

interference-plus-noise ratios of 10 dB or greater,” and noted that an ENG truck “can be

positioned to achieve the best possible signal between transmitter and receiver.”63

Further, the use of a contention-based protocol, as discussed in more detail below, will

provide additional protection. In the LPI context, the Commission concluded that “the risk of

harmful interference to indoor electronic news gathering receivers from indoor unlicensed

60 See id. ¶¶ 103, 218. 61 See Wireless Research Center of North Carolina Report at 9, fig. 26. 62 6 GHz R&O and FNPRM ¶ 166. 63 Id. ¶¶ 162, 164.

30

devices is insignificant,” in part because unlicensed devices will use a contention-based

protocol.64 Such a protocol, the Commission explained, will “allow unlicensed devices to sense

the energy from nearby indoor licensed operations and avoid using that channel.”65 The same

interaction is true for VLP devices using a contention-based protocol.

Contention-based protocol. Adopting a contention-based protocol for VLP devices will

allow multiple unlicensed users to share spectrum effectively and will provide another layer of

harmful interference protection for FS links. As the Commission explains, the interference

protection benefit of a contention-based protocol is two-fold. A contention-based protocol avoids

co-frequency interference with other services sharing the band, and also restricts the amount of

time a device can transmit, thus limiting the already very small time periods where harmful

interference could even theoretically occur in a worst-case scenario.66 The two-fold interference

protection function even further lowers the risk of harmful interference to FS links beyond the

results demonstrated in the Monte Carlo analysis.

Unlicensed devices typically use some type of contention-based protocol to account for

the lack of a central controller. Therefore, VLP devices will also automatically employ robust

contention-based protocols in most cases, regardless of whether the Commission requires such

protocols.67 If the Commission chooses to require a contention-based protocol for VLP

64 Id. ¶ 168. 65 Id. 66 Id. ¶ 102. 67 See Letter from Apple Inc., Broadcom Inc., Cisco Systems, Inc., Facebook, Inc., Google

LLC, Hewlett Packard Enterprise, Intel Corporation, Microsoft Corporation, NXP Semiconductors, Qualcomm Incorporated, and Ruckus Networks to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295 (filed Mar. 20, 2020).

31

devices,68 it should follow the approach adopted for indoor APs and decline to specify specific

protocols or dictate burdensome testing requirements, as long as a device manufacturer

accurately describes the operation of the protocol that it implements. Restricting the type of

contention-based protocol that VLP devices can use, by contrast, would unnecessarily limit

manufacturers’ abilities to achieve required throughput in challenging configurations while

complying with the power limit.

Power Spectral Density. Finally, statistical analysis demonstrates that the Commission

can certainly authorize VLP devices at a power limit of 14 dBm EIRP regardless of whether the

bandwidth the device is transmitting at is 20 MHz, 40 MHz, 80 MHz, 160 MHz, or 320 MHz.

The RKF study also measured the sensitivity of potential FS interference to different

distributions of VLP channel bandwidths, including 20 MHz, 40 MHz, 80 MHz, and 160 MHz.

The results indicated almost no correlation between bandwidth and increased I/N probability.69

In other words, the results for an EIRP of 14 dBm over 160 MHz (-8 dBm/MHz) were not

substantially different than the results for an EIRP of 14 dBm concentrated in 20 MHz (1

dBm/MHz). This result was true even with higher numbers of active VLP devices in the

simulation (a two-fold, three-fold, six-fold, and twelve-fold increase in the number of

transmitting devices over expected numbers). This was also true for RKF’s analysis of BAS

central receive sites.70 The Commission should therefore not limit a VLP device from

68 See 6 GHz R&O and FNPRM ¶ 237. 69 RKF VLP Report at 29, Tbl. 5-6 (explaining that “narrower bandwidths correspond to

proportionally higher PSD, but also a proportionally lower chance that the VLP’s bandwidth overlaps with the FS’s channel: i.e., reducing bandwidth from 160 MHz to 20 MHz increases the power density by 8x but also reduces the probability of channel overlap by a factor of eight. These two factors essentially negate each other.”).

70 Id. at 52.

32

transmitting at the maximum authorized power regardless of its operating bandwidth because

doing so will not increase the risk of harmful interference to incumbent services and would only

serve to reduce the capacity in the selected channel.

III. THE COMMISSION SHOULD PERMIT MOBILE STANDARD-POWER ACCESS POINT OPERATIONS GOVERNED BY AUTOMATED FREQUENCY COORDINATION.

The Commission’s decision to permit fixed standard-power APs governed by AFC will

provide consumers with important improvements to existing Wi-Fi networks. But without

mobility for some types of devices, the 6 GHz band will not fully achieve its game-changing

potential. Mobility has become central to consumer expectations, and mobile enterprise data

network access is increasingly critical to a wide range of important applications.

As the Commission has explained, mobile “standard-power [APs], under AFC control …

would expand the area over which unlicensed 6 GHz devices can operate to deliver additional

benefits to the American public.”71 In addition,“[m]obile use at higher power levels … could

also enable new innovative applications.”72 The Commission therefore should enable mobile

standard-power APs under reasonable operating rules that will protect incumbents and promote

innovation. Because AFCs can apply the same incumbent protection criteria already established

for fixed AP operations, doing so will not increase the risk of harmful interference to incumbents

compared to fixed devices. Permitting mobility also will not cause congestion to AFC systems or

delay their rollout.

71 6 GHz R&O and FNPRM ¶ 246. 72 Id.

33

A. Innovators Need 6 GHz Mobile Standard-Power APs to Meet Growing Mobile Wi-Fi Demand and Enable Future Advances.

The Commission seeks comment on potential benefits of permitting mobile standard-

power AP operations.73 Mobility is critical to supporting both significant tangible near-term

benefits and the “visionary applications,” which are difficult to imagine today,74 that have

always characterized unlicensed bands. As the following examples illustrate, opening up the U-

NII-5 and U-NII-7 bands for mobile standard-power AP operations can unleash a wave of near-

term innovation, from next-generation communications for commuters and students while in

transit, to vehicle-area networks that enable sophisticated data gathering and analysis, to

providing connectivity on demand in areas where deploying fixed APs may be impractical.

1. Internet access on public transportation.

Commuters, students, and other users expect Wi-Fi to be available on the go as they

travel in vehicles. According to a survey by Devicescape, “91 percent of passengers expect Wi-

Fi access while traveling.”75 Another survey by Ofcom UK similarly underscores the

“importance of the internet while commuting,” with over a third of respondents deeming it to be

“essential” to completing professional and/or personal tasks.76

73 Id. 74 See, e.g., id. at Statement of Commissioner Brendan Carr. 75 Rob Taylo, Mass Transit & Mobile Wireless Internet: The Challenge of Costs, Mass Transit

(Feb. 19, 2013), https://www.masstransitmag.com/home/blog/10881239/mass-transit-mobile-wireless-internet.

76 Ofcom, Communications Market Report, 16 (Aug. 2, 2019), https://www.ofcom.org.uk/__data/assets/pdf_file/0022/117256/CMR-2018-narrative-report.pdf.

34

Unsurprisingly, therefore, Wi-Fi access is becoming a core feature to attract and retain

riders on virtually all forms of public transportation.77 For example:

• 26% of U.S. public transit buses have already been equipped with Wi-Fi, up from just 1% a decade ago;78

• Wi-Fi is available to 90% of Amtrak customers;79

• Virtually all privately operated intercity bus operators make Wi-Fi available for some—if not all—of their fleets;80

• Atlanta now offers Wi-Fi on all of its buses and trains;81

77 See, e.g., Rob Taylo, Bus Wi-Fi System Considerations for Transit Agencies, Metro

Magazine (Feb. 13, 2014), https://www.metro-magazine.com/blogpost/240249/bus-wi-fi-system-considerations-for-transit-agencies (observing that “transit agencies that successfully install dependable Wi-Fi access, such as Calif.-based Santa Clara VTA — often see ridership increases…”).

78 American Public Transportation Association, 2020 Public Transportation Fact Book, 19, Fig. 19 (Mar. 2020), https://www.apta.com/wp-content/uploads/APTA-2020-Fact-Book.pdf.

79 Amtrak, Wi-Fi? Why Yes! Now Available to 90% of Amtrak Customers (Mar. 2016), http://blog.amtrak.com/2016/03/evenmorewifi/#:~:text=With%20the%20latest%20service%20rollout,most%20out%20of%20the%20technology.

80 See, e.g., Greyhound, Bus features, https://www.greyhound.com/en/discover-greyhound/bus-features-and-virtual-tour#:~:text=Free%20Wi%2DFi,buses%2C%20and%20for%20all%20 passengers (last visited June 28, 2020); Peter Pan Bus, Traveling with Us: What You Need to Know, https://peterpanbus.com/travel-info/traveling-with-us/#:~:text=All%20Peter%20Pan%20buses%20are,with%20headrests%20and%20tinted%20windows.&text=Connect%20to%20Wi%2DFi%20by,WiFi“%20icon%20on%20your%20device (last visited June 28, 2020); Megabus, Traveling on the bus, https://us.megabus.com/help/traveling-on-the-bus (last visited June 28, 2020); RedCoach, Ultimate Comfort, https://www.redcoachusa.com/ultimate-comfort/#:~:text=It's%20not%20just%20 about%20the%20seats&text=Not%20only%20we%20have%20the,to%20them%20and%20complimentary%20WiFi (last visited June 28, 2020).

81 MARTA, Wi-Fi on MARTA, https://itsmarta.com/wifi.aspx#:~:text=The%20MARTA%20Wi%2DFi%20is,on%20all%20buses%20and%20trains (last visited June 28, 2020).

35

• The New York Metropolitan Transit Authority is deploying over 2,000 “new high-tech buses that have Wi-Fi”;82 and

• Houston plans to provide Wi-Fi on all metro buses and trains by mid-2021.83

Wi-Fi plays an increasingly important role in student transportation as well. School

districts across the country have begun using buses to provide Wi-Fi to students, and “districts

that have invested in Wi-Fi on buses are now receiving a lot in return.”84 These benefits include

enabling students to complete homework and communicate with their teachers, providing video

streams to monitor student behavior, delivering real-time information about when students were

picked up and dropped off, and deploying buses to create hotspots in underserved areas.85 Wi-Fi

on school buses is also facilitating remote learning during the COVID-19 pandemic,86 and is

consistent with other efforts to provide additional mobile Wi-Fi hotspot access in response to

COVID-19.87

82 Thomas MacMillan, MTA to Roll Out New Buses With Wi-Fi, Phone-Charging Outlets, Wall

Street Journal (last updated Mar. 8, 2016, 5:55 PM), https://www.wsj.com/articles/mta-to-roll-out-new-buses-with-wi-fi-phone-charging-outlets-1457471735.

83 Dug Begley, All Metro Buses and Trains to Have Wi-Fi by Mid-2021, Agency Says, Houston Chronicle (Mar. 27, 2020), https://www.houstonchronicle.com/news/transportation/ article/All-Metro-buses-and-trains-to-have-Wi-Fi-by-15161288.php.

84 Julie Metea, School Bus Wi-Fi is a Turning Point in Transportation Technology, School Transportation News (Sept. 23, 2019), https://stnonline.com/special-reports/school-bus-wi-fi-a-turning-point-in-transportation-technology/ (“School Transportation News Article”).

85 Id. 86 See, e.g., T. Keung Hui, Pandemic Closed NC Schools. Now Some Buses Will Have Wi-Fi so

Students Can Go Online, Raleigh News & Observer (last updated May 6, 2020, 1:54 PM) https://www.newsobserver.com/news/local/education/article242535261.html.

87 See, e.g., Connie Gentry, NC Senate’s Coronavirus Relief Bill Has $125 Million for Small Businesses, $300 Million for NCDOT, Triangle Business Journal (Apr. 29, 2020), https://www.bizjournals.com/triangle/news/2020/04/29/nc-sentates-coronavirus-relief-bill-has-125.html.

36

Public transportation users will increasingly demand higher bandwidth and lower latency

connections, particularly as mobile APs can take advantage of licensed 5G backhaul in the near

future.88 But this demand cannot be met even once the Commission presumptively permits 6

GHz VLP devices, because mass transportation vehicles have large interior volumes as well as

compartment partitions or obstructions (e.g. luggage storage, restrooms) that require higher AP

power to meet service level requirements. Thus, the Commission must enable mobile standard-

power operations for public transportation users to realize the significant benefits of 6 GHz

spectrum.

2. Industry-specific unlicensed mobile connectivity applications.

Several industry-specific mobile applications that currently rely on mobile Wi-Fi APs to

enable automation and other applications onboard purpose-built vehicles would also benefit

significantly from access to 6 GHz spectrum at higher power levels.

Precision agriculture. Companies such as John Deere offer mobile hotspots in tractors,

combines, and other farm equipment to support a wide range of computing devices in the driver

cabin,89 as well as to provide support for ground personnel with tablets.90 Although many of

these uses are data-intensive, “greater bandwidth [will] support more sophisticated applications

88 See School Transportation News Article (noting that 5G backhaul will enable Wi-Fi

improvements). 89 See, e.g., @AGofTheWorld, Twitter (Apr. 22, 2020, 10:19 AM) (illustrating the wide range

of devices and screens deployed in mobile farm vehicles), https://twitter.com/AGofTheWorld/status/1252965311269273602.

90 See, e.g., Scott Ferguson, John Deere Bets the Farm on AI, IoT, Light Reading (Mar. 12, 2018), https://www.lightreading.com/enterprise-cloud/machine-learning-and-ai/john-deere-bets-the-farm-on-ai-iot/a/d-id/741284.

37

and collect[ing] greater volumes of data—high-definition video, for example—whether it’s from

the machines or the field itself.”91

Container ports. Container ports increasingly make use of Wi-Fi applications to improve

efficiency, including by mobile gantry crane operations.92 However, as with precision

agriculture, “[n]ew applications demand higher bandwidth.”93 For example, Wi-Fi APs mounted

on mobile gantry cranes can provide backhaul connectivity to sensors and cameras attached to

grappling and hoisting equipment. Port cranes similarly have large numbers of cameras to enable

operators to perform activities such as verifying proper twistlock engagement, machine room

status, winch condition, ground conditions while in motion, as well as documenting container

conditions for insurance damage claims. Wireless video backhaul can improve image quality and

permit more rapid upgrades without the constraints of legacy data cabling. Container port

operations can also take advantage of mobile APs on ground vehicles such as small loaders and

trucks to provide opportunistic connectivity to ground personnel who are using mobile data

terminals or tablets while located deep inside metal container stacks in ports.

Construction equipment. Construction site equipment similarly leverages Wi-Fi

connections to enable projects to be completed safer and more efficiently and at a lower cost.

These mobile applications also will benefit from 6 GHz spectrum access. For example, portable

Wi-Fi jobsite applications can include access to “3D models & construction software,” “RFIs,

91 Id. 92 See, e.g., Port Technology, The Internet of Things and the Container Port (Sept. 2, 2015),

https://www.porttechnology.org/news/internet_of_things_and_the_container_port/. 93 Id.

38

change orders & punch lists,” and “[c]ollaboration tools & photo documentation.”94 Other

applications include “high-precision asset and personnel tracking,”95 as well as internet

connectivity to areas such as mines, construction sites, and temporary staging and assembly areas

that can be challenging to serve using fixed infrastructure.

3. AP-to-AP mobile mesh applications.

Mobile Wi-Fi mesh points in which mobile APs communicate with other mesh APs

and/or fixed infrastructure are increasingly common. For example, to minimize backhaul costs,

major rail system operators employ mobile Wi-Fi APs that shift traffic from licensed networks to

Wi-Fi when approaching stations. Applications for moving and tracking rail cars in a railyard or

maintenance depot typically connect to local networks via Wi-Fi as well. Mesh applications are

also heavily used in mining, oil and gas, and other outdoor enterprises spanning extremely large

areas that lack permanent physical infrastructure. These and other mesh connectivity applications

will benefit from access to 6 GHz band spectrum at higher power levels.

A primary benefit of mesh technology in these applications is its ability to perform

“self-healing” by dynamically and opportunistically maximizing all of the available radio

frequency (“RF”) pathways that may exist between a source and destination.96 This is in contrast

to traditional AP-to-client networks where only a single active connection and network pathway

94 DEWALT, Jobsite WiFi System: Built for Construction,

https://wifi.dewalt.com/#_ga=2.203764496.1498671666.1592406131-1977361952.1592406131 (last visited June 28, 2020).

95 See, e.g., CAT, New Products, Cat Detect for Underground: Prevent Unintended Interactions Between Personnel and Assets to Improve Site Safety, https://www.cat.com/en_US/products/new/technology/detect/detect/102360.html (last visited June 28, 2020).

96 See generally, e.g., Yan Zhang, Jijun Luo & Honglin Hu, Wireless Mesh Networking: Architectures, Protocols and Standards (1st ed. 2006).

39

exists at a given time. The wide 80-megahertz and 160-megahertz channels available in the

6 GHz band will enable enterprises to construct high-speed, privately owned wide-area

communications networks without incurring the recurring data costs charged by licensed

network operators.

Passenger and freight rail. Railway mesh backhaul-to-trackside networks are

fundamentally different than traditional Wi-Fi client operations due to the need to maintain at

least two active connections to avoid dropouts, also known as “make before break” connections.

Moreover, trains typically rely on at least two radios, one at each end of the train “consist” (i.e.

the rail vehicles that collectively comprise a train). The figure below illustrates a train at two

different positions relative to a trackside. The green line shows the active RF connection path,

and the red line indicates the backup path. As the train moves along the trackside, the active path

will change to take advantage of the better connection.97

97 See Cisco, Cisco Connected Rail Solution Implementation Guide, 53 (Nov. 2016),

https://www.cisco.com/c/dam/en_us/solutions/industries/docs/cts-ig.pdf.

40

Figure 3: Example Multi-Radio Deployment for Train-to-Trackside Communications98

On-board car-to-car mesh is another important use case for connected rail operations. The

particular organization and arrangement of rail vehicles in a consist can change from day to day

as cars or engines are added and dropped from the consist. Examples of train consists are shown

in the figure below. Although there is a need for high-speed inter-car connectivity, there is no

way to know in advance which cars or engines will be part of which consist. Wi-Fi mesh

technology offers an ideal solution to this problem.

98 Id.

41

Figure 4: Train Consist Examples99

Standard-power Wi-Fi operation is essential in connected rail applications. Passenger

trains have increasing requirements for high speed intra-consist data communications, including

data-intensive video applications such as security video transmissions and providing in-car video

entertainment from a centralized media server. Thus, multiple HD or ultra-HD video streams

could be active across a passenger consist, in addition to basic vehicle telemetry and internet

backhaul via a shared 4G or 5G uplink. In these cases, it is vital to maintain the highest possible

data rates, which require a high signal-to-interference-plus-noise radio. In addition, freight trains

will often require standard power even to accommodate the length of a single consist. The

99 See TrainWeb, USA Rail Guide, Amtrak - Crescent,

http://www.trainweb.org/usarail/crescent.htm; TrainWeb, USA Rail Guide, Amtrak - Lake Shore Limited, http://www.trainweb.org/usarail/lakeshorelimited.htm (last visited Jun. 28, 2020).

42

average length of a freight train in the U.S. is nearly 2 kilometers, and trains can extend to

approximately 6 kilometers with locomotives inserted at midpoints along the consist.100 In this

case, higher power is necessary to enable a single mesh network that can span the entire consist

to provide data communications.

Mining, oil and gas, and other large area operations. Asset productivity is a significant

challenge facing the mining sector today, driven by increased pressure to process material at a

lower cost per ton while working to lower personnel safety risk. The number of sensors and

related devices used to monitor equipment health and fleet performance continues to grow, along

with the number of assets in the fleets themselves. 101 The explosion of the Industrial Internet of

Things (“IIoT”) has enabled mines to track numerous aspects of a machine’s operations.102

Many of these sensor-based applications are bandwidth-intensive and demand low

latency so that data can be rapidly delivered to command centers for real-time evaluation.

Moreover, networks must enable real-time data collection from a variety of equipment that is

broadly dispersed across the mining environment. Many of these assets are continuously moving

and need to maintain nonstop mobile connectivity in order to reliably deliver insights on their

performance and health.

100 See, e.g., Stephen Joiner, Is Bigger Better? 'Monster' Trains vs Freight Trains, Popular

Mechanics (Feb. 11, 2010), https://www.popularmechanics.com/technology/infrastructure/a5314/4345689/; Rich Connell, Safety, Traffic Concerns Raised When 3.5-Mile-Long Freight Train Rolls Through L.A. Basin, Los Angeles Times (Jan. 12, 2010), https://www.latimes.com/archives/la-xpm-2010-jan-12-la-mew-train13-2010jan13-story.html.

101 See, e.g., Alasdair Monk, How IIot Is Changing Mining, Industry Week (Dec. 27, 2018), https://www.industryweek.com/technology-and-iiot/article/22026902/how-iiot-is-changing-mining.

102 See id.

43

Wireless mesh networks can also significantly facilitate autonomous mining operations.

Automated hauling and drilling equipment can reduce operating costs and lower total cost of

ownership by 15% to 40% while enabling production to continue around the clock without

risking worker fatigue.103 With fully autonomous equipment for haulage and drilling available,

and automated blasting and shoveling solutions coming into use, mines have new opportunities

to achieve multi-fold gains in productivity via assets that conduct key mining processes without

direct operator oversight. However, these assets must remain in constant communication with

command centers in order to send and receive the data necessary to remain self-operational. If

connectivity is lost, even briefly, the autonomous asset will shut down as a safety precaution.

Always-on connectivity can be relatively straightforward to achieve if autonomous equipment is

largely stationary, such as an autonomous drill. However, some critical equipment must

constantly move. Autonomous haul trucks, for example, will need to traverse large stretches of

rough, harsh terrain and maneuver around obstacles along the way, and must maintain

continuous connectivity to avoid the asset becoming stranded in place mid-trek. Thus, mesh

networks are a critical tool for network engineers to achieve the promise of autonomy.

Current 5 GHz spectrum availability, however, substantially constrains mobile Wi-Fi

mesh network applications. For example, the 5 GHz band—which has only two unencumbered

higher bandwidth 80 MHz channels and no unencumbered 160 MHz channels—is characterized

by unpredictable spectrum availability that limits investment in these technologies. In addition,

area-scale mesh networks require multiple channels on which to operate, and typically require

wider channels for backhaul links than are employed on radios that communicate with wireless

103 Ajay Lala, et al., Productivity at the Mine Face: Pointing the Way Forward, McKinsey &

Company (Aug. 4, 2016), https://www.mckinsey.com/industries/metals-and-mining/our-insights/productivity-at-the-mine-face-pointing-the-way-forward.

44

clients. There are simply not enough channels in the 5 GHz band to deploy large scale mesh

networks that also enable wireless client communications.

In contrast, mobile standard-power operations in U-NII-5 and U-NII-7 could vastly

increase the utility of mobile mesh applications, because these bands would offer (1) a larger

number of channels; (2) wider channels; (3) operation without constraints such as dynamic

frequency selection; and (4) the ability to take advantage of very high bandwidth licensed 5G

uplinks. The combination of 5 GHz for access, and 6 GHz for mesh backhaul will be

fundamentally transformative to enable these and other use cases.

B. Reasonable Operating Rules for Mobile Standard-Power APs in the U-NII-5 and U-NII-7 Bands Will Protect Incumbents and Promote Investment.

As the Commission has recognized, mobile standard-power APs could operate “under

rules similar to those for personal/portable white space devices.”104 With important

modifications to reflect the difference between 6 GHz and white space operations, we agree with

the Commission that the personal/portable white space rules provide a helpful framework.

The Commission has already authorized personal/portable white space device use not

only on frequencies allocated to broadcast television, but also on frequencies allocated for public

safety and other mobile licensed use based on the Commission’s “high degree of confidence that

the databases can reliably protect [these] operations,”105 because “[p]ersonal/portable devices

104 6 GHz R&O and FNPRM ¶ 246. 105 See Amendment of Part 15 of the Commission's Rules for Unlicensed Operations in the

Television Bands, Repurposed 600 MHz Band, 600 MHz Guard Bands and Duplex Gap, and Channel 37, Notice of Proposed Rulemaking, 29 FCC Rcd. 12,248 ¶ 30 (2014).

45

[that] rely on database access to determine their list of available channels … can protect

[incumbents] in the same manner as fixed devices.”106

This conclusion applies with even greater force in the U-NII-5 and U-NII-7 bands. In

contrast to white space device operations where the precise location of incumbent receivers are

unknown, the ULS already has “extensive technical data for site-based licenses” that includes

receiver locations and other detailed characteristics.107 Thus, the fixed standard-power AFC rules

the Commission adopted in the R&O provide a sound functional basis for successfully

implementing mobile AFC and AP operations.

1. AFC systems can accommodate a wide range of mobile and portable operations.

The FNPRM seeks comment on “operational aspects associated with mobile standard-

power device[s]” that work with AFC systems.108 AFC systems could support multiple use cases

for mobile standard-power device operations depending on application requirements based on

market demand. These can include the following means of defining areas of operation that will

protect incumbent fixed operations, listed in order of the freedom of movement required by the

mobile AP.

Geofencing. Using geofenced areas of operation, which the Commission specifically cites

as an example of how a mobile device could “preload” operating channels,109 is a simple and

static form of area precalculation. Geofencing could be a particularly useful approach in cases

106 Amendment of Part 15 of the Commission’s Rules for Unlicensed Operations in the

Television Bands, Repurposed 600 MHz Band, 600 MHz Guard Bands and Duplex Gap, and Channel 37, Report & Order, 30 FCC Rcd. 9551 ¶ 88 (2015).

107 6 GHz R&O and FNPRM ¶ 30. 108 Id. ¶ 251. 109 See id. ¶ 249.

46

where mobile standard-power AP deployments—perhaps managed and operated by a single

entity—exist within a defined geographic perimeter and are not free to leave the area. For

example, geofenced areas could include an industrial area, university campus, or airport

property. Indeed, several of the industry-specific applications described above are particularly

well-suited to operation within locations such as construction sites, container ports, railyards, oil

fields, mines, and farms,110 which already run their day-to-day operations on a bounded

perimeter basis. Moreover, the geofenced area could accommodate operations on shuttle buses or

light rail vehicles that only traverse well defined areas within the geofence.

One important feature of a geofence for 6 GHz Wi-Fi operation will be the ability to scale

to accommodate a wide range of areas; for example, from several acres for a typical construction

site up to several square miles for a mine, oil field, or farm. However, the method by which a

mobile standard-power AP determines its proximity to the edge of the geofenced area must not

be so conservative as to preclude the ability to operate small protected areas.111 For example, a

square block is a reasonable, practical lower bound for a geofenced area (approximately two

acres).

Precalculation. Mobile AFC systems can determine frequency availability for a defined

geographic area using sizes that are appropriate for localized topographic variability and

expected vehicle velocity based on the existing technical rules for fixed APs. Mobile AFC

systems can then consider AP power levels and make a granular comparison of the frequency

availability variation within the contiguous area of grid squares and determine the optimum

110 See Section III.A, supra. 111 Smaller geofenced areas will require vehicles in motion to operate at lower velocity

compared to larger areas.

47

frequency selections that will maximize frequency availability while minimizing channel

changes as the AP moves within the operating area.

Precalculation can be a particularly beneficial approach for mobile standard-power AP

deployments operating with geographically constrained motion, such as along railways and

roadways. For example, mobile APs deployed in rail cars would be geographically constrained

by the path of the tracks, allowing the mobile AFC system to precalculate frequency availability

along the geographically narrow line of a long-distance route, with minimal uncertainty in side-

to-side location. Similarly, the mobile AFC system can determine frequency availability along

slightly less constrained but still bounded travel paths such as public roads. In each of these cases

the AP could store sufficient precalculation data from the AFC system to travel without

immediate interaction with the AFC.

As another example, a school bus might typically run a certain route in the morning and

another in the afternoon, and might also run alternate—but predictable—routes at certain times

depending on traffic and demand. The geolocation capability of a mobile AP would permit

switching between various stored, pre-authorized channel availability maps based on the current

location without needing to consume mobile AFC resources.

Real-time calculation. For mobile applications with significant freedom of operation,

such as off-road vehicles, an AFC system that performs real-time calculations based on current

and associated future location uncertainty may be the most appropriate solution. For these cases,

a mobile standard-power AP with a limited backhaul connection might choose to accept a

reduction in its available spectrum by providing its geolocation information to the mobile AFC

system with a higher degree of uncertainty than it is capable of in order to reduce the frequency

with which it must check in with the AFC in light of its connectivity situation. Alternatively, the

48

same mobile AP, when it has high speed backhaul connectivity, might elect to contact the mobile

AFC system more frequently and report the much smaller uncertainty it is capable of delivering,

in order to benefit from improved AFC calculation of available spectrum.112

Significantly, each of these three modes is functionally equivalent to the fixed AFC case

the Commission has already approved and presents no greater degree of complexity. Similar to a

still photograph compared to an old-style film movie, operating parameters permitted by the

AFC system for mobile use represent a sequence of “still photos” of availability combined

together to show availability while in motion. Both the approved fixed AFC rules and the

proposed portable AFC rules involve the same queries using a particular geolocation and some

uncertainty. Mobile AFC systems would simply perform more of these queries.

For example, the geofence model is functionally equivalent to a larger uncertainty with

an arbitrary polygonal perimeter in lieu of an ellipse or circular radius. An AFC system will still

be obligated to inspect all FS receivers within a large radius of the geofenced area, just as it

would with a fixed AP using a smaller uncertainty. What differs in each case is the operational

burden on the mobile standard-power AP. In the geofence case, the mobile AP must verify its

position frequently to ensure it is inside the perimeter, but otherwise need only contact the AFC

once a day, as described below. In both the precalculated and real-time cases, the mobile AP

must be capable of transitioning from one area to another, including obtaining spectrum

availability in each area and seamlessly moving its associated clients, if its operating channel is

no longer available.

112 Mobile AP operation in vehicles at highway speeds is unlikely to require real-time

calculation and frequent AFC contact, because sustaining such speeds over long distances is generally only practical within and along the confines of well-maintained roadways, and these operations would generally be more efficient under the pre-calculation model.

49

The Commission should make clear that its rules are flexible enough to accommodate

each of these models, and should also avoid prescriptive rules regarding the implementation of

these or other models.

2. The Commission should model the mobile standard-power AFC rules on its existing personal/portable white space device rules, with certain important modifications.

The FNPRM seeks comment on the personal/portable white space rules that “could be

adopted for 6 GHz standard-power devices,” as well as other changes or requirements that would

be appropriate.113 Several concepts in the personal/portable white space rules could be adapted to

enable mobile standard-power operations in U-NII-5 and U-NII-7.

First, similar to Mode II personal/portable white space devices, the standard-power rules

should create a new class of APs capable of operation while in motion that can obtain allowable

operating parameters in a geographic area.114 The rules should also specify that these APs

operate “pursuant to direction from an [AFC] System,” just as fixed standard-power APs do.115

We recommend that the Commission establish the following definition:

Mobile Standard-Power Access Point. An access point that operates in the U-NII-5 and U-NII-7 bands that uses an internal or connected geo-location capability to select approved operating frequencies in a specific geographic area, pursuant to direction from a mobile Automated Frequency Coordination System.

Second, mobile standard-power APs should include a geolocation capability. This

capability should either be integrated into the device or physically connected (e.g. to leverage a

113 6 GHz R&O and FNRPM ¶ 248. 114 See id. (“Should we define a separate device category for mobile standard-power devices? If

so, how should these differ from fixed standard-power access points? For example, we believe such devices would need an integrated geolocation capability and have an integrated connectorized antenna.”).

115 See 47 C.F.R. § 15.403 (effective July 27, 2020).

50

vehicle’s existing geolocation functionality). APs should check their location upon being

powered on, and e.g. every 60 seconds while in operation (i.e. not in “sleep mode”) to confirm

that they are within the previously authorized area of operation.116 However, consistent with the

white space rules—as well as the Commission’s commitment to technological neutrality—the

Commission should not require a particular geolocation technology.117 Similarly, the

Commission should not set a specific location accuracy standard. Rather, as with the white space

rules, mobile AFC systems can account for the device’s geolocation accuracy when defining the

allowable operating parameters in given areas.118 Moreover, as described above, a mobile

standard-power AP should be permitted to report a larger uncertainty than it is capable of

determining to enable the use of larger or smaller query regions based on the velocity of the

device.

Third, the Commission should enable mobile AFC systems to define areas of operation

by having mobile standard-power APs “preload a list of cleared channels over an area (e.g.,

create a geo-fenced area) and operate without [contacting] the AFC system so long as they stay

within the cleared area.”119 As the Commission explained when it created this capability for

Mode II white space devices, “[a]llowing channel lists to be stored for more than a single

116 Id. § 15.711(d)(1). 117 See id. § 15.703(g); see also Unlicensed Operation in the TV Broadcast Bands, Additional

Spectrum for Unlicensed Devices Below 900 MHz and in the 3 GHz Band, Second Report & Order and Memorandum Opinion & Order, 23 FCC Rcd. 16,807 ¶ 62 (2008) (noting consensus that “the Commission should not require the use of a specific geo-location method …”).

118 See 47 C.F.R. § 15.712; see also id. § 15.407(k)(9)(iv) (effective July 27, 2020) (accounting for geolocation uncertainty for 6 GHz fixed AP operations).

119 See 6 GHz R&O and FNRPM ¶ 249; see also 47 C.F.R. § 15.711(d)(5) (permitting Mode II personal/portable white space devices to load channel availability information to create “a geographic area within which it can operate”).

51

location will allow for more efficient operation of portable devices” because it will (1) “reduc[e]

the number of queries to the database,” and (2) “support mobile operation.”120 We recommend

that the Commission specifically permit mobile AFC systems to define these areas either by a

center point and radius or a geographic polygon.

Fourth, and consistent with both the white space rules and 6 GHz AFC rules, the

Commission should require mobile standard-power APs to communicate with the AFC system at

least once a day.121 However, in some cases mobile standard-power APs may need to contact

AFC systems more frequently. For example, the white space rules contemplate that a device that

has preloaded “channel availability information for multiple locations must contact the database

again if/when it moves beyond the boundary of the area where the channel availability data is

valid.”122

Fifth, the Commission should permit mobile APs to operate using either an integrated or

connectorized antenna.123 For example, roof-mount antennas are required on vehicles such as

buses and rail cars in order to enable several of the use cases described above. These antennas

have special designs and ratings to handle considerations such as water intrusion, vibration, and

wind loading.

Finally, although the white space device rules restrict personal/portable devices to less

power than fixed devices, no such restriction is necessary for mobile standard-power operations.

As noted previously, unlike in TV white spaces, the locations of incumbent 6 GHz receivers are

120 Unlicensed Operation in the TV Broadcast Bands, Additional Spectrum for Unlicensed

Devices Below 900 MHz and in the 3 GHz Band, Second Memorandum Opinion and Order, 25 FCC Rcd. 18,661 ¶ 115 (2010).

121 47 C.F.R. §§ 15.711(d)(4), 15.407 (k)(8)(iv) (effective July 27, 2020). 122 Id. § 15.711(d)(5). 123 See 6 GHz R&O and FNPRM ¶ 249 (seeking comment on device antenna requirements).

52

known to the AFC system, so that there is no difference in interference risk between a fixed

AFC-coordinated AP and a similarly coordinated mobile AP that has previously determined

spectrum availability throughout its operating area. Moreover, the available frequency set for a

given geographic coverage area may be significantly larger at EIRP levels below the regulatory

maximum, which encourages the use of the lowest reliable power for a given application

scenario.

C. Mobile Standard-Power Access Point Operations Will Not Negatively Impact AFC Systems.

The Commission seeks comment on the “effect [that] permitting mobile standard-power

[AP] operation [would] have on the AFC.”124 AFC systems enabling mobile operations can

successfully implement mobile standard-power AP support without any negative impacts.125

Moreover, the Commission can move forward with confidence by permitting individual AFC

systems to decide whether to support mobile APs rather than mandating that all AFC systems

include mobile AP support.

1. The Commission can structure mobile standard-power AFC rules as an optional supplement to the existing fixed AFC rules.

The FNPRM asks whether adding support for mobile APs would “delay the AFC system

development and prevent the American public from reaping the benefits of expanded unlicensed

124 Id. ¶ 250. 125 We also note that most of the work to implement mobile standard power APs rests with the

AP manufacturer, rather than the AFC system operator. Mobile APs must demonstrate compliance with Commission rules regarding geolocation accuracy, updates, and location relative to the boundary of a geofence, pre-calculated area, or real-time calculated uncertainty area. Where channel availability changes between areas, it must also successfully transition all of its associated clients as an operational matter. In contrast, the AFC system performs substantially the same work whether the device making a request is fixed or mobile. In addition, providers that intend to support mobile use cases will be certain to work with AFC systems that have demonstrated their ability to scale to whatever level of queries are expected.

53

use soon.”126 It would not. The Commission already has established a clear path to implement

support for fixed AP operations, and the addition of mobility will not impact fixed standard-

power AP operation or certification.127 The Commission can help ensure that there are no fixed

deployment delays by structuring the mobile standard-power rules as an optional supplement to

the existing rules rather than mandating that all AFC systems develop this functionality.128

Adopting this approach will enable AFC system developers—and the market—to dictate

whether the benefits of supporting mobile standard-power AP operations exceed the costs.129

The Commission should also allow an AFC system operator to initially elect to support only

fixed standard-power APs, and then optionally supplement the scope of their approval over

time.130 This will allow AFC systems to manage performance and scale appropriately. Offering

certainty to AFC systems to enable them to add capabilities and improvements over time by

updating the scope of their authorization will encourage early deployment of fixed AFC

systems—permitting industry to continue to innovate while providing access to additional

unlicensed spectrum in the near term.

126 6 GHz R&O and FNPRM ¶ 250. 127 See id. (seeking comment on whether fixed devices would require updates). 128 Indeed, with respect to fixed-only AFCs, the only additional requirement would be to ensure

that these AFCs do not provide authorization to mobile APs. 129 See 6 GHz R&O and FNPRM ¶ 150 (seeking comment on the costs of permitting mobile

standard-power operations). 130 In a similar fashion, AFC should be allowed to introduce subsequent support for features

such as antenna directionality. See id. ¶ 254 (“[W]e … seek comment on whether the AFC system should be permitted to take the directivity of a standard-power access point’s antenna into account when determining the available frequencies and power levels at a location, rather than assuming an omnidirectional antenna”).

54

2. Mobile standard-power operations will not cause congestion to AFC systems.

The FNPRM asks whether “mobile applications [would] add substantial congestion” to

AFC system connections or introduce other complications that would “potentially degrad[e]”

fixed AP use.131 They would not. Rather, mobile AP manufacturers will partner with AFC

system operators that have the demonstrated scale necessary to support the intended use cases.

First, most of the work necessary to enable a mobile standard-power deployment will be

implemented by the AP manufacturer. This is true both in terms of device functionality—such as

transitioning associated clients between regions with discontinuous channel availability—as well

as compliance certification, which will require more test cases than for fixed devices.

Second, mobile AFC systems would implement support for mobile standard-power APs

using the same core methodology and fixed incumbent data used to establish operating areas for

fixed APs. The exemplary operating models described above (geofencing, precalculation, and

real-time calculation) simply build on the existing fixed AFC spectrum availability determination

methodology by incorporating various degrees of location uncertainty and frequency availability

optimizations over larger areas, as appropriate.

Third, as the Commission has already recognized in the white space device context,

defining areas of operation for mobile devices will substantially reduce the number of queries

needed to support mobile operations.132 Indeed, for many of the use cases described above that

rely on precalculation of routes or operation exclusively within known, established geofenced

areas, the mobile standard-power AP might only query an AFC system once per day, just as a

fixed AP would.

131 Id. ¶ 250. 132 See Section III.B.2, supra.

55

Fourth, there will be other applications that, while mobile, do not necessitate numerous

database queries because they involve “low motion” use cases. These could include, for example,

school districts that deploy Wi-Fi-enabled buses in the evening to areas with limited home

connectivity,133 or libraries that temporarily loan out mobile AP hotspots for household use over

a period of days or weeks.134 Although a subset of these applications theoretically could be

accommodated through placement of “transportable” fixed APs,135 rules for mobile standard-

power APs will significantly facilitate design and deployment.

Finally, even for uses that could potentially involve numerous database queries over the

course of a day—which could be the case for APs with high mobility where longer-term

probabilistic route precalculation is infeasible—AFCs can easily build systems that are

sufficiently robust to handle numerous concurrent queries. One of the primary benefits of

modern cloud-based architectures is the ability to dynamically scale computing resources to

adjust to periods of peak demand.136 Indeed, databases routinely execute transactions at rates far

in excess of those that would be required for a mobile AFC system.137 And because mobile

133 Office of Education Technology, Busing in WiFi, Department of Education,

https://tech.ed.gov/stories/busing-in-wifi/; Sarah Al-Arshani, School Districts Across the Country Are Using School Buses to Deliver WiFi to Students Who Lack Access, Insider (Apr. 1, 2020), https://www.insider.com/wifi-buses-being-used-across-country-to-give-kids-internet-2020-3.

134 See, e.g., Mobile Hotspots: Tech2go, Los Angeles Public Library, https://www.lapl.org/tech2go/mobile-hotspots.

135 See 6 GHz R&O and FNPRM ¶ 211. 136 See, e.g., Ted Simpson & Jason Novak, Hands-On Virtual Computing 451 (2d ed. 2017). 137 For example, Visa alone processes over 65,000 transactions per second or 5.6 billion per day.

Visa, Visa Fact Sheet, https://usa.visa.com/dam/VCOM/download/corporate/media/visanet-technology/aboutvisafactsheet.pdf. The NASDAQ stock market processes 25 million trades per day, or over 64,000 per minute. Daily Market Summary, NasdaqTrader.com, http://www.nasdaqtrader.com/Trader.aspx?id=DailyMarketSummary. And the Internet Domain Name System (“DNS”) is estimated to handle peaks of up to 200,000 transactions per second solely between recursive resolvers and authoritative nameservers. Pawel

56

standard-power AP support should be optional, AFC developers and system operators that do not

want to make the technological investments necessary to effectively support higher numbers of

queries need not do so.138 Thus, permitting mobile standard-power AP operations should not

have any negative impacts on fixed AFC and AP development and operations.

IV. THE COMMISSION SHOULD PERMIT LOW-POWER INDOOR ACCESS POINTS TO OPERATE UP TO 8 dBm/MHZ PSD.

The Commission’s decision in the R&O adopted a very conservative approach to

unlicensed sharing through a careful consideration of power levels that would protect incumbent

services while enabling expanded unlicensed use of the 6 GHz band. The FCC was correct to

focus on PSD levels in doing so, but the FNPRM wisely decided to consider increasing the

maximum permitted PSD. Power levels in a range of channel sizes are important for consumer

utility and the economic benefit of the band, and the Commission and should strike the correct

balance by authorizing a PSD level of 8 dBm/MHz. The record strongly supports the adoption of

an 8 dBm/MHz PSD power limit for LPI operations.139 The Commission is correct that an 8

dBm/MHz PSD limit “would be useful for many indoor devices that require high data rate

Foremski, et al., DNS Observatory: The Big Picture of the DNS (Oct. 2019), https://www.net.in.tum.de/fileadmin/bibtex/publications/papers/foremski2019dns.pdf.

138 In contrast, AFC developers that choose to deploy mobile-only AFC systems could optimize those systems to support high-activity mobile AFC-AP interactions.

139 See, e.g., Letter from Paul Jamieson, Vice President, Government Affairs & Policy, Altice to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183 (filed Apr. 15, 2020); Letter from Rob Alderfer, Vice President of Technology Policy, CableLabs to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183 (filed Mar. 30, 2020) (“CableLabs Mar. 30 Ex Parte”); Letter from David M. Don, Vice President, Regulatory Policy, Comcast to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295 (filed Apr. 10, 2020); Letter from Elizabeth Andrion, Senior Vice President, Regulatory Affairs, Charter Communications to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183 (filed Mar. 16, 2020).

57

transmissions.”140 The Commission can adopt a higher PSD limit for LPI operations because

doing so will not increase the risk of harmful interference to incumbent fixed and mobile

services, as demonstrated by numerous technical studies in the record and the Commission’s own

analysis.

A. Operation at 8 dBm/MHz PSD Is Important to Meet Clear Consumer Needs.

Wi-Fi is the workhorse of broadband internet access. It carries an estimated 75 percent of

all wireless traffic and over half of all internet traffic.141 Wi-Fi access and performance has only

become more critical over the last several months as consumers prioritize “digital resiliency” in

the face of pandemic-driven stay-at-home orders and shifting work and school locations.142 As

Chairman Pai recently explained, “The coronavirus pandemic has really amplified how

indispensable Wi-Fi has become in our lives. Our homes have turned into our offices and our

classrooms, testing the limits of our Wi-Fi networks like never before.”143 Authorizing a PSD

level of 8 dBm/MHz will allow 6 GHz Wi-Fi to better meet the critical need for reliable, fast,

and ubiquitous Wi-Fi coverage.

The key differences between a 5 dBm/MHz and 8 dBm/MHz PSD limit are (1) coverage

area, (2) throughput in the covered area, and (3) a tendency to force traffic onto the widest

channels when narrower channels are sufficient. A PSD limit of 5 dBm/MHz rather than 8

140 See 6 GHz R&O and FNPRM ¶ 244. 141 Claus Hetting, New Numbers: Wi-Fi Share of US Mobile Data Traffic Lingers at Around

75% in Q2, Wi-Fi NOW (Aug. 16, 2018), https://wifinowglobal.com/news-and-blog/new-numbers-wi-fi-share-of-us-mobile-traffic-lingers-at-around-75/.

142 See WifiForward, Wi-Fi at Work, Unlicensed Spectrum: The Economic Boost We Need, YouTube at approx. 28:30 (May 29, 2020), https://www.youtube.com/watch?v=zV8jnPTENFs&feature=youtu.be&t=2.

143 Chairman Pai Remarks to Wi-Fi Alliance.

58

dBm/MHz limits coverage range by 31-43% and throughput by 53-63%, on average.144

Unnecessarily limiting Wi-Fi coverage range means that users either cannot access Wi-Fi in

certain parts of their home, small office, or school, or that they must purchase, install, and

manage additional coverage extenders or access points. Some consumers will not be able to

afford the additional equipment or will not be able to successfully install additional equipment,

therefore foregoing coverage. Thus, users in throughput-challenged environments, particularly

where multiple devices use the same AP—such as schools and small offices—will have no

option other than a lower-throughput experience, reducing access to data-intensive content and

two-way video applications. Furthermore, users that can install additional coverage extenders or

APs will use more of the available spectrum and use this spectrum less efficiently.

Additionally, a PSD limit of 8 dBm/MHz will more closely align 6 GHz Wi-Fi with

existing 5 GHz Wi-Fi operations, better meeting end-user expectations. The rules for Wi-Fi

operations in the nearby 5 GHz band permit operations at radiated PSDs of 23 dBm/MHz (U-

NII-1) and 39 dBm/MHz (U-NII-3), and an EIRP limit of 36 dBm.145 Although the currently

authorized 5 dBm/MHz PSD limit will greatly improve channel availability for indoor Wi-Fi

compared to existing Wi-Fi channels in other bands, a PSD of 8 dBm/MHz will better align with

existing Wi-Fi PSD limits, reducing the risk that consumers will experience inferior Wi-Fi

performance using 6 GHz Wi-Fi than they currently do using 5 GHz Wi-Fi.

Further, a PSD limit of 8 dBm rather than 5 dBm will better allow Wi-Fi networks to

accommodate and adapt to different usage environments. Wi-Fi networks select different channel

bandwidths to best meet the needs of connected clients in difficult coverage scenarios or high-

144 See CableLabs Mar. 30 Ex Parte at 4–5. 145 See 47 C.F.R. § 15.407(a).

59

congestion environments. This allows network operators to ensure compatibility with end-user

devices that do not support the widest channels, maintain the flow of data when part of the

configured bandwidth is being used by another system, and improve link reliability for client

devices at the edge of coverage areas.146 However, a fixed PSD limit means that a reduction in

channel width also requires a reduction in total radiated power. Under the 5 dBm/MHz limit, for

example, a reduction from a 160-megahertz channel to an 80-megahertz channel requires a

reduction from 27 dBm EIRP to 24 dBm EIRP. Thus, a restrictive PSD limit constrains network

operators and prevents them from being able to reduce channel bandwidth to meet the needs of

challenging coverage environments. Increasing the PSD limit from 5 dBm/MHz to 8 dBm/MHz

will significantly improve flexible channel options to compensate for poor coverage

environments.

Relatedly, under the 5 dBm/MHz limit, larger enterprise Wi-Fi users that demand

enterprise-wide coverage are likely to default to 80- or 160-megahertz wide channels with an

effective EIRP of 24 dBm or 27 dBm, respectively, even though they may have IoT applications

that would happy exist on a 20- or 40-megahertz wide channel. An 8 dBm/MHz limit will ensure

a wider range of channel sizes is available at power levels similar to today’s 5 GHz equipment,

capable of supporting a more applications-diverse enterprise network.

B. The Record Demonstrates that LPI Access Points Operating at 8 dBm/MHz PSD Will Not Cause Harmful Interference to FS Operations.

This record already contains compelling demonstrations, including the Commission’s

own analysis, showing that LPI at 8 dBm/MHz PSD will not cause harmful interference to

incumbent FS licensees. CableLabs, for example, has submitted an analysis showing that “LPI

146 CableLabs Mar. 30 Ex Parte at 3.

60

Wi-Fi will not impact FS link availability even when all APs have a peak EIRP of 30 dBm and a

PSD of 8 dBm/MHz in a dense metropolitan environment.”147 Based on a Monte Carlo analysis

modeling LPI-FS interactions in New York City, this analysis tested the sensitivity of the

probability of interference to radiated power levels and conservatively assumed that all LPI APs

would exhibit the peak gain possible for the equipment.148 The Commission expressly noted that

it found the Monte Carlo analysis “persuasive,” explaining that the study showed that the I/N

ratio resulting from RLAN operations was far below even -6 dB,149 a level appropriate as an

AFC threshold but that does not approach harmful interference in the real world. The

Commission also specifically noted that this sensitivity analysis resolved questions about the

power level distribution by assuming all access points operated at 8 dBm/MHz and still showed

that “the I/N was less than -6 dB in all instances.”150 The Commission’s own analysis of the

CableLabs study therefore supports a PSD limit of 8 dBm/MHz for LPI operations.

Even analyses based on a link budget filing submitted by AT&T demonstrate that the

Commission should authorize LPI operations at 8 dBm/MHz PSD. The Commission correctly

recognized the shortcomings of individual link budget analyses, but it explained that AT&T’s

analysis “illustrates that interference is not likely to occur with the proposed power levels” using

147 Letter from Rob Alderfer, Vice President of Technology Policy, CableLabs to Marlene H.

Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183, at 2 (Mar. 19, 2020).

148 See id.; 6 GHz R&O and FNPRM ¶ 117; CableLabs, 6 GHz Low Power Indoor (LPI) Wi-Fi / Fixed Service Coexistence Study (Dec. 2019), as attached to Letter from Rob Alderfer, Vice President of Technology Policy, CableLabs to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183 (filed Dec. 20, 2019).

149 6 GHz R&O and FNPRM ¶ 118. 150 Id. ¶ 119.

61

reasonable assumptions,151 and that “AT&T overstates the potential for harmful interference.”152

We agree, and submitted analysis correcting AT&T’s overly conservative assumptions and

showing that, with more realistic assumptions and a power limit of 30 dBm in a 160-megahertz

channel, equivalent to 8 dBm/MHz PSD, LPI operations will not increase the risk of harmful

interference to FS operations.153 Correcting AT&T’s assumptions regarding antenna

discrimination, propagation losses, and other parameters, we showed that RLANs operating at 8

dBm/MHz PSD will not cause harmful interference to FS links.154

C. The Record Demonstrates that LPI Access Points Operating at 8 dBm/MHz PSD Will Not Cause Harmful Interference to BAS, ENG, or CARS Operations.

The record clearly demonstrates that the FCC can authorize LPI operations with a PSD

limit of 8 dBm/MHz while protecting incumbent mobile operations in the 6 GHz band, such as

BAS, ENG, and CARS, from harmful interference. In describing its assessment that LPI

operations would not cause harmful interference to ENG receivers in the R&O, the Commission

did not rely on a 5 dBm/MHz PSD assumption. Rather, it noted that the 5 dBm/MHz PSD limit

“further reduces” a probability of interference that it concluded was already “negligible.”155

151 Id. ¶ 112, 116. 152 Id. ¶ 124. 153 See Correcting the Record on RLAN-FS Interactions at slide 3 (Dec. 2019) (“Correcting the

Record on RLAN-FS Interactions”), as attached to Letter from Paul Margie, Counsel to Apple Inc., Broadcom Inc., Cisco Systems, Inc., Facebook, Inc. Google LLC, Hewlett Packard Enterprise, and Microsoft Corporation, to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183 (filed Dec. 16, 2019); Letter from Apple Inc., Broadcom Inc., Cisco Systems, Inc., Facebook, Inc., Google LLC, Hewlett Packard Enterprise, Intel Corporation, Microsoft Corporation, NXP Semiconductors, and Qualcomm Incorporated to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183 (filed Feb 12, 2019); see 6 GHz R&O and FNPRM at 3899, Tbl. 4.

154 See Correcting the Record on RLAN-FS Interactions. 155 See 6 GHz R&O and FNPRM ¶ 166.

62

The two technical studies on which the Commission relies to conclude that LPI will not

cause harmful interference to mobile services such as BAS and ENG operations used an 8

dBm/MHz PSD level. First, CableLabs submitted a simulation specific to mobile broadcast

operations—BAS—demonstrating that, even “under aggressive Wi-Fi parameters with higher

Wi-Fi activity and stronger propagation than typical,” BAS/Wi-Fi interactions would maintain

BAS link quality at levels sufficient to deliver high-quality video.156 In the simulation, BAS

signal-to-interference-plus-noise-ratio (“SINR”) remained above 10 dB in 99.9991% of cases.157

Importantly, the simulation assumes an LPI device operating at 30 dBm EIRP in a 160-

megahertz channel—i.e. 8 dBm/MHz PSD.

Second, an additional power sensitivity analysis by CableLabs demonstrates that, even in

a simulation where LPI operations reach a maximum power level more often (at more angles

across the antenna pattern), there remains no material risk of harmful interference to FS

operations.158 Similarly, the study submitted by our companies assumed a PSD of 8 dBm/MHz

and demonstrated that, for three categories of broadcast operations highlighted by the National

Association of Broadcasters—outdoor links from remote trucks to central receive sites, outdoor

156 Letter from Elizabeth Andrion, Senior Vice President of Regulatory Affairs, Charter

Communications, and Rob Alderfer, Vice President of Technology Policy, CableLabs to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183, at 4 (filed Feb. 21, 2020).

157 Id. 158 CableLabs, Wi-Fi Power Sensitivity Analysis Shows No Harmful Interference from Low-

Power Indoor Wi-Fi to FS and BAS in 6 GHz (Mar. 2020), as attached to Letter from Elizabeth Andrion, Senior Vice President of Regulatory Affairs, Charter Communications, Rob Alderfer, Vice President of Technology Policy, CableLabs, David Don, Vice President of Regulatory Policy, Comcast Corporation, and Barry Ohlson, Vice President of Regulatory Affairs, Cox Enterprises, Inc. to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183, at 4 (filed Mar. 9, 2020).

63

links from ENG devices to receivers on trucks, and indoor links between ENG transmitters and

ENG receivers—there is no real-world risk of harmful interference.159

V. CONCLUSION

The Commission took an important leap forward by opening the 6 GHz band to

unlicensed fixed APs. Portability for low power devices and mobility for standard-power APs are

essential to the ways Americans use wireless technologies, and without Commission rules

permitting portable and mobile devices, the country will not realize the full potential of the 6

GHz band. The Commission should adopt rules permitting at least 14 dBm EIRP portable VLP

devices and mobile standard-power AFC-governed devices at 36 dBm in each of the 6 GHz U-

NII sub-bands—and should permit LPI devices to operate at 8 dBm/MHz. This will deliver

important new applications to consumers, support U.S. leadership in wireless innovation,

contribute substantially to the national economy, and will not cause harmful interference to

incumbent operations. We urge the Commission to swiftly adopt a Report & Order implementing

these proposals.

159 See Letter from Apple Inc., Broadcom Inc., Cisco Systems, Inc., Facebook, Inc., Google

LLC, Hewlett Packard Enterprise, Intel Corporation, Microsoft Corporation, NXP Semiconductors, and Qualcomm Incorporated to Marlene H. Dortch, Secretary, FCC, ET Docket No. 18-295, GN Docket No. 17-183 at 2 (filed Feb. 28, 2020).

64

Respectfully submitted,

Apple Inc. Broadcom Inc. Cisco Systems, Inc. Facebook, Inc. Google LLC Hewlett Packard Enterprise Intel Corporation Microsoft Corporation NXP Semiconductors Qualcomm Incorporated Ruckus Networks, a Business Segment of CommScope

June 29, 2020

Attachment A:

RKF Engineering Solutions, LLC, Frequency Sharing for Very Low Power Radio Local Area Networks in the 6 GHz Band

1

Frequency Sharing for Very Low Power (“VLP”) Radio Local Area Networks in the 6 GHz Band

June 29, 2020

Prepared by:

RKF Engineering Solutions, LLC 7500 Old Georgetown Road Bethesda, MD

Prepared for:

Apple Inc., Broadcom Inc., Cisco Systems, Inc., Facebook Inc., Google LLC, Hewlett-Packard Enterprise, Intel Corporation, Microsoft Corporation, NXP Semiconductors, Ruckus Networks, a business segment of CommScope, QUALCOMM Incorporated

2

1 Table of Contents 1 Table of Contents 21.0 Executive Summary 4

1.1 Fixed Service (FS) 51.2 Mobile Service (MS) 6

2 Introduction 82.1 Background 82.2 Incumbent Services 92.3 Very Low Power Device Class 102.4 Approach 11

3 VLP Deployment and Operating Assumptions 153.1 VLP Deployment Assumptions 15

3.1.1 Number of Active VLPs and Deployment Distribution 153.1.2 Population Density 15

3.2 VLP Operating Assumptions 173.2.1 Distribution of Source VLP Power Levels including Body Loss 173.2.2 Bandwidth and Channel Distribution 183.2.3 VLP Height 18

4 Propagation Models 195 Sharing Results 21

5.1 Fixed Service (FS) Sharing 215.1.1 ULS Database Review 225.1.2 Key Modeling Assumptions 23

5.1.2.1 VLP Device Deployment 235.1.2.2 FS Receiver Antenna Performance 245.1.2.3 FS Receiver Noise Figure and Feeder Loss 24

5.1.3 Baseline I/N Occurrence Simulation and Sensitivity to VLP Channel Bandwidth 245.1.4 Selected FS Availability Calculations for the Baseline Model 305.1.5 Ohio Example of Correlation Between Multipath Fading and Interference 365.1.6 Sensitivity Analysis 39

5.1.6.1 Number of Active Devices (2x, 3x, 6x, and 12x) 395.1.6.1.1 Occurrence Probabilities 395.1.6.1.2 Impact on FS link availability resulting from a 2x, 3x, 6x, and 12x increase in the number of active VLP devices 41

5.1.6.2 Higher EIRP Level 445.1.6.2.1 Occurrence Probabilities 44

3

5.1.6.2.2 Impact on FS link availability resulting from 21 dBm EIRP 455.1.7 FS Sharing Conclusions 46

5.2 Mobile Service (MS) Sharing 475.2.1 MS Usage Studied 475.2.2 MS Simulation 525.2.3 Sensitivity Analysis 53

5.2.3.1 Number of Active Devices (2x, 3x, 6x, and 12x) 535.2.3.2 Higher EIRP Level 55

5.2.4 MS Sharing Conclusions 56

4

1.0 Executive Summary

This paper is in response to the Further Notice of Proposed Rulemaking1 (FNPRM) where the Federal Communications Commission (the Commission) seeks comment on expanding unlicensed operation in the 6 GHz band to include Very Low Power (VLP) operation that is not limited to indoor use and does not require an automated frequency coordination (AFC) system.

Portable devices, the subject of this study, will expand innovation even further and will be critical for supporting indoor and outdoor portable use cases such as wearable peripherals, including augmented reality/virtual reality, as well as in-vehicle applications and other personal-area-network applications.

Because VLP portable devices will operate at considerably lower power levels than Low Power Indoor (LPI) devices while indoors, and LPI devices have already been authorized, the focus of this study is only for portable devices operating outdoors.

In the 6 GHz proceeding, VLP devices were discussed as a device class with an EIRP lower than that of an LPI access point. The Commission stated in the 6 GHz Report and Order2 that a “compelling case was made for allowing such use. These devices can usher in new ways that Americans work, play, and live by enabling applications that can provide large quantities of information in near real-time.” The Commission, therefore, proposed “to permit VLP devices to operate across the entire 6 GHz band (5.925-7.125 GHz), both indoors and outdoors, without using an AFC.”

The 6 GHz spectrum is divided into four unlicensed bands (U-NII-5, U-NII-6, U-NII-7, and U-NII-8) that reflect the different incumbent service allocations. This report considers sharing with the FS primarily in the U-NII-5 and U-NII-7 bands and sharing with mobile Broadcast Auxiliary Service (BAS) and Cable Television Relay Service (CARS) in the U-NII-6 and U-NII-8 bands.

This study assumes a number of instantaneously transmitting VLP devices consistent with ECC Report 3163. The sharing studies start with 14 dBm EIRP as a VLP power level because Apple, Broadcom et al. contend it is the minimum EIRP necessary to enable the applications anticipated for these devices. Higher power levels have the potential to provide an expanded and higher quality of service. The studies include a sensitivity analysis on VLP channel size, the number of active VLP devices, and EIRP.

1 In the Matter of Unlicensed Use of the 6 GHz Band, Expanding Flexible Use in Mid-Band Spectrum Between 3.7 and 24 GHz, Report and Order and Further Notice of Proposed Rulemaking, 35 FCC Rcd. 3852 (2020) (“6 GHz Report and Order”). 2 Id. 3 CEPT ECC, Sharing studies assessing short-term interference from Wireless Access Systems including Radio Local Area Networks (WAS/RLAN) into Fixed Service in the frequency band 5925-6425 MHz (May 2020), https://www.ecodocdb.dk/download/8951af9e-1932/ECC%20Report%20316.pdf (“CEPT ECC Report”).

5

1.1 Fixed Service (FS)

Monte-Carlo simulations were performed with random VLP deployments to understand the interference risk to FS operations over CONUS. The baseline simulation consisted of 100,000 VLP deployment iterations to gather stable, long-term interference statistics at each of 97,888 FS sites. The VLP channel deployments included a mix of channel bandwidths (20, 40, 80, 160 MHz) and a fixed EIRP of 14 dBm per VLP device.

Statistics were gathered at each FS, on the occurrence probabilities for both I/N > -6 dB and 0 dB. Because these metrics do not fully describe the interference risk, an additional metric, increased FS unavailability due to VLP interference, was used to assess degradation in FS performance. This analysis assumed a typical FS design target of 99.999% availability (unavailability=0.001% corresponding to 5.3 minutes/year). Results were compared to a target increase in unavailability of less than 10% (availability with interference >99.9989%) sufficient to allow continued robustness of FS links while also allowing the new VLP service. Sensitivity to a 1% increase in unavailability was also considered.

The I/N > -6 dB and 0 dB average occurrence probability of a single FS was 0.00011% and 0.00002% respectively for the Baseline Simulations. Through additional channel bandwidth sensitivity simulations, these results were determined to be independent of the VLP channel size. An EIRP of 14 dBm per 20 MHz channel had the same impact as an EIRP of 14 dBm per 160 MHz because the narrower channel bandwidth had a higher power spectral density but lower probability of overlapping the FS channel. Thus, the analysis demonstrated that VLP devices can operate with the same EIRP, i.e. no constant PSD limitation, in all channel sizes without harmful interference.

For the FS availability analysis, 1,000 out of 97,888 FS were randomly selected from the FS with at least one I/N > -6 dB occurrence. The increase in unavailability due to VLP interference of these FS links was further analyzed in two steps. In the first step, a representative link margin required to meet the target availability was calculated without considering the specific operational parameters of each FS link. This simplified analysis allowed a large number of links to be processed. In the second step, if the simplified analysis indicated an FS link did not meet the target 10% unavailability increase, individual FS operational parameters were analyzed to determine the actual increase in unavailability. This analysis provided a realistic assessment of the long-term impact of the VLP interference on FS stations and showed all 1,000 links met the 10% increase in unavailability target as well as the 1% increase in unavailability sensitivity threshold.

The availability analysis above assumed that fading and interference are independent. However, the Commission reaffirmed TSB-10F guidance4 that fading mostly occurs between midnight and 8 a.m., while VLP activity would be during daylight hours, an inverse correlation. To demonstrate the impact of this inverse correlation, an analysis was performed on an FS link in Ohio, where hourly fade statistics were available. From worst-month statistics, a worst-case daylight hour fading distribution was developed. This worst-hour model was then used to predict the increase in unavailability to the FS link. The result of this analysis exemplified that if hourly fade statistics are considered, the increase in unavailability will further be reduced by an order of magnitude compared to the results derived using the assumption that fading and interference are independent.

4 6 GHz Report and Order ¶ 143.

6

To further understand the sensitivity of the results to parameters used in the analysis, the number of active VLP devices was increased (2x, 3x, 6x, and 12x) and the EIRP was increased to 21 dBm. Looking at the increased number of active devices, even when the number of active devices was increased by 12x, interference was almost always dominated by a single VLP device, indicating interference continued to be independent of the channel size and was no worse if all devices were transmitting 14 dBm in 20 MHz. This further demonstrates that VLP devices can operate with the same EIRP, i.e. no constant PSD limitation, in all channel sizes without harmful interference.

Looking further at FS unavailability impacts, the increased number of active devices showed that the number of I/N > -6 dB and 0 dB occurrences increase roughly linearly with an increase in the number of active VLP devices. However, as in the Baseline Simulation, all 1,000 FS stations selected for further analysis met the 10% unavailability target as well as the 1% sensitivity threshold even with 12x the number of active VLP devices.

When the EIRP was increased to 21 dBm with the baseline number of active VLP devices, the I/N > -6 dB occurrence probability scaled linearly resulting in an increase in the occurrence probability by five times (7 dB difference). However, looking at the FS unavailability impacts, operating at the higher EIRP level = 21 dBm, the links met the 10% increase in unavailability target and the 1% sensitivity threshold as well as 2x, 3x, 6x, and 12x.

In conclusion, our analysis showed that VLP operation with EIRP = 14 dBm per channel and a variety of channel sizes will not cause harmful interference to FS stations. In addition, sensitivity analyses on parameters including bandwidth, number of active devices, and EIRP indicated that in all cases the probability of an I/N > -6 dB occurrence was low and the increase in unavailability was sufficiently low to allow continued robustness of FS links.

1.2 Mobile Service (MS)

The interference from VLP devices to the mobile truck to ENG central receive station mobile BAS/CARS use-case was studied. The analysis is based on Monte-Carlo simulations of two ENG central receive stations, Cowles Mountain in San Diego and DC Old Post Office in Washington DC, that were determined as representative links for this use-case by the National Association of Broadcasters (NAB)5 in the study conducted by Alion Sciences.

Similar to the FS studies, a Monte-Carlo simulation with 100,000 iterations was performed as the Baseline Model, with VLP device fixed EIRP of 14 dBm and the baseline channel distribution. To assess sensitivity of the results to VLP channel bandwidth, four additional Monte-Carlo simulations were performed, assuming a fixed bandwidth (20, 40, 80, or 160 MHz), with 100,000 iterations. Finally, sensitivity analyses on the number of active VLPs (2x, 3x, 6x, 12x) and VLP EIRP (21 dBm) were done.

At the Cowles Mountain receive site, there were no I/N > -6 dB occurrences up to 12x number of active devices with 14 dBm EIRP. At 21 dBm EIRP, the average I/N > -6 dB occurrence probabilities were extremely small, even at 12x the number of active devices (max 0.004%).

5 Letter from Rick Kaplan to Marlene H. Dortch, ET Docket No. 18-295 & GN Docket No. 17-183 (filed Dec. 5, 2019) (“Dec. 5 NAB Letter”).

7

At the theoretical DC Old Post Office receive site, there were no I/N > -6 dB occurrences in the baseline simulations, except for one link in one of the simulations that had one occurrence out of 100,000. When increasing the number of active devices to 2x, 3x, 6x, and 12x and at 21 dBm EIRP, although the average probabilities of occurrence were higher than at Cowles Mountain, they were nonetheless extremely small (max 0.026%).

One of the major differences between the Cowles Mountain and the DC Old Post Office ENG sites is the population in proximity of the receiver antenna. The Cowles Mountain site is on a higher elevation covering a larger operating radius, while the DC Old Post Office is on a lower elevation providing service over a more densely populated area.

In both instances, there was no risk of harmful interference for active VLP devices up to 12x the baseline using an EIRP up to 21 dBm, independent of the channel bandwidth. This is expected for other locations throughout the CONUS, given the high elevation of these ENG Central Receive antennas and the very low power at which the VLP device transmits.

8

2 Introduction

Devices that employ Wi-Fi and other unlicensed standards have become indispensable for providing low-cost wireless connectivity in countless products used by American consumers. On April 23, 2020, the Commission adopted rules6 to make 1200 MHz of spectrum available in the 6 GHz band (5.925-7.125 GHz). These new rules will expand unlicensed broadband operations that promise to bring a wide range of innovative wireless applications to consumers while protecting incumbent users in the band. As has occurred with Wi-Fi in the 2.4 GHz and 5 GHz bands, it is expected that the rules adopted for 6 GHz unlicensed devices will foster the expansion of Wi-Fi hotspot networks to provide consumers access to even higher speed data connections and growth in the Internet-of-things (IoT) industry—connecting appliances, machines, meters, wearables, and other consumer electronics, as well as industrial sensors for manufacturing. This capability will quickly become a part of peoples’ everyday lives.

This study is in response to the FNPRM,7 where the Commission seeks comment on expanding unlicensed operation in the 6 GHz band to include VLP operation that is not limited to indoor use and does not require an AFC system.

2.1 Background

In the 6 GHz rules, the Commission authorized two different types of unlicensed operations—standard-power and LPI operations. The standard-power access points are restricted to operate in portions of the band and can be used anywhere as part of a hotspot network by incorporating an AFC system to protect incumbents. The AFC system determines the frequencies on which standard-power access points operate without causing harmful interference to incumbent microwave receivers and then identifies those frequencies as available for use by the access points. LPI access points and client devices are authorized across the entire 6 GHz band and do not rely on the AFC system for determining the frequencies available for use. These low-power access points will be ideal for connecting devices in homes and businesses, such as smartphones, tablet devices, laptops, and IoT devices, to the Internet. Using these advanced Wi-Fi technologies and wider channels (up to 320 MHz) available in the 6 GHz band, unlicensed devices promise to spur innovations and allow consumers to experience faster internet connections and new applications well beyond what is possible with 2.4 GHz and 5 GHz bands.

Portable devices, the subject of this study, will expand innovation even further and will be critical for supporting indoor and outdoor portable use cases such as wearable peripherals including augmented reality/virtual reality as well as in-vehicle applications and other personal-area-network applications.

Because VLP portable devices will operate at considerably lower power levels than LPI devices while indoors, and LPI devices have already been authorized, the focus of this study is only for portable devices operating outdoors.

6 See 6 GHz Report and Order. 7 Id.

9

2.2 Incumbent Services

Table 2-1 shows the division of the 6 GHz spectrum into four unlicensed bands (U-NII-5, U-NII-6, U-NII-7, and U-NII-8) that reflect the different incumbent service allocations. Per the Commission’s 6 GHz Report & Order:

● Fixed Service (FS) are point-to-point microwave systems that “make significant use of the U-NII-5 and U-NII-7 bands” and “operate in relatively smaller numbers in the U-NII-8.”8

● “The Broadcast Auxiliary Service (BAS) and Cable Television Relay Service (CARS) operate in the U-NII-6 band on a mobile basis, and in the U-NII-8 band on both a fixed and mobile basis.” The BAS and CARS “transmit programming material from special events or remote locations, including electronic news gathering (ENG), back to the studio or other central receive location.”9

● “The Fixed Satellite Service (FSS) Earth-to-space is allocated in all four sub-bands, except for the 7.075-7.125 GHz portion of the U-NII-8 band. FSS operations are heaviest in the U-NII-5 band, which is paired with the 3.7-4.2 GHz space-to-Earth frequency band to comprise the ‘conventional C-band’.”10

Table 2-1 - Predominant Uses of the 6 GHz Band11

Sub-Band Frequency Range (GHz)

Primary Allocation Predominant Licensed Services

U-NII-5 5.925-6.425 Fixed FSS

Fixed Microwave FSS (Uplinks)

U-NII-6 6.425-6.525 Mobile (MS) FSS

Broadcast Auxiliary Service Cable Television Relay Service FSS (Uplinks)

U-NII-7 6.525-6.875 Fixed FSS

Fixed Microwave FSS (Uplinks/Downlinks)

U-NII-8 6.875-7.125 Fixed Mobile FSS

Broadcast Auxiliary Service Fixed Microwave Broadcast Auxiliary Service

8 Id. ¶ 7. 9 Id. ¶ 8. 10 Id. ¶ 9. 11 Id., Table 1.

10

The 6 GHz rules for the standard power and LPI access points were adopted considering incumbent operations in the four different sub-bands. Unlicensed standard-power access points are authorized in the U-NII-5 and U-NII-7 bands through the use of an AFC system. This fosters synergistic use of both the 5 GHz and 6 GHz bands and promotes unlicensed broadband deployment through operations using the same power levels permitted in the 5 GHz U-NII-1 and U-NII-3 bands.12 Second, the 6 GHz rules opened the entire 6 GHz band for unlicensed LPI access points, maximizing future capacity and performance capabilities. Client devices are also unlicensed, and their power levels depend on the type of access point to which they are connected. Table 2-2 summarizes the Commission rules for unlicensed operation in the 6 GHz band.

Table 2-2 - Unlicensed Use of the 6 GHz Band

Device Class Operating Bands Maximum EIRP (dBm)

over 160 MHz13

Maximum Power Spectral Density EIRP (dBm/MHz)

Standard-Power Access Point (AFC-Controlled) U-NII-5, U-NII-7

36 23

Client Connected to Standard Power Access Point

30 17

Low-Power Indoor Access Point U-NII-5, U-NII-6, U-NII-7, U-NII-8

27 5 Client Connected to Low-Power Indoor Access Point

21 -1

2.3 Very Low Power Device Class

In the 6 GHz proceeding, VLP devices were discussed as a device class with an EIRP lower than that of an LPI access point. The Commission stated in the 6 GHz Report and Order that “proponents for very low power unlicensed devices have made a compelling case for allowing such use. These devices can usher in new ways that Americans work, play, and live by enabling applications that can provide large quantities of information in near real-time.”14 The Commission, therefore, proposed “to permit very low power devices to operate across the entirety of the 6 GHz band (5.925-7.125 GHz), both indoors and outdoors, without using an AFC.”

12 U-NII-1 is 5.150-5.250 GHz and U-NII-3 is 5.725-5.850 GHz. 13 The Commission authorized 320 MHz channels, which leads to 3 dB higher power for LPI Access Points and related client devices beyond what is listed in this table (i.e., 30 dBm EIRP for LPI access points and 24 dBm EIRP for LPI clients). The power levels for 160 MHz channels were included for ease in comparing currently authorized power levels against the VLP analysis in this report. 14 6 GHz Report and Order ¶ 235.

11

In the FNPRM, the Commission sought comment on:

● the VLP operating assumptions (e.g., body loss, antenna radiation pattern, projected activity factor, etc.);

● the appropriate way to model the potential interactions between VLP unlicensed devices and the incumbent operations;15 and

● the power level to authorize for VLP unlicensed devices to maximize the utility of the 6 GHz band and protect incumbent services.16

This report responds to the above questions.

The sharing studies start with 14 dBm EIRP as a VLP power level because Apple, Broadcom et al. contend it is the minimum EIRP necessary to enable the applications anticipated for these devices. The studies include a sensitivity analysis using a VLP power level of 21 dBm EIRP that has the potential to provide an expanded and higher quality of service.

This report is intended as an extension to RKF's 2018 Report.17 As such, assumptions and methodologies from RKF's 2018 Report are referenced. Deviations are explained as appropriate.

2.4 Approach

Similar to RKF's 2018 Report, a detailed CONUS-wide Monte-Carlo simulation of the interference environment was performed with changes implemented to focus on outdoor VLP devices. These changes are summarized in Table 2-3 below. The Monte-Carlo simulations were performed over a large number of independent events to establish long-term statistical properties in the environment.

Table 2-3 - Parameter Changes from RKF's 2018 Report

Changes from RKF's 2018 Report

Parameters changed Comments

VLP Device Model Number of Active Devices Barren areas are included with rural areas. Antenna pattern and body loss (i.e., GFarField in Eqn. 2-1) Device height

See Section 3

Propagation Models See Section 4

Outdoor VLP Device No building penetration loss used

15 Id. ¶ 236. 16 Id. ¶ 243. 17 RKF Engineering Solutions, Frequency Sharing for Radio Local Area Networks in the 6 GHz Band (Jan. 2018), https://s3.amazonaws.com/rkfengineering-web/6USC+Report+Release+-+24Jan2018.pdf.

12

Interference Power (I) Polarization loss Receiver feederloss

See Eqn. 2-1 See Section 5.1.2.3 for FS

Noise Power (N) Noise figure See Section 5.1.2.3 for FS

ULS18 Data Effective date Jan 21, 2020 See Section 5.1.1

As in RKF's 2018 Report, the interference power, I, is computed per Eqn. 2-1 below:

I = EIRP + GFarField – LPropagationPath – LSpectralOverlap - LPolarization – Lfeed + GRx-to-VLP (2-1)

where,

● I (dBW) = Interference Power from VLP (aggregate or single-entry (i.e., due to single VLP device))

● EIRP (dBW) = VLP EIRP within VLP channel bandwidth (baseline: 14 dBm, sensitivity analysis: 21 dBm)

● GFarField (dB) = VLP far field gain that includes body loss (see Section 3.2.1)

● LPropagationPath (dB) = Propagation Path loss including Clutter loss per Section 4

● LSpectralOverlap (dB) = 10*log10(spectrum overlap between VLP channel and victim channel / VLP bandwidth), also called frequency-dependent rejection.

● Lpolarization =Polarization Loss of 3 dB19,20

● Lfeed (dB) = Feederloss of victim receiver

● GRx-to-VLP (dBi) = Gain of victim FS Rx towards VLP based on the angle off-boresight

The I/N is the ratio of the interference power and the receiver (Rx) noise power. The receiver noise power is calculated, for each victim Rx, using Eqn. 2-2 below:

𝑁 = 10(𝑘𝑇!𝐵) +𝑁𝐹 (dBW) (2-2)

where,

● 𝑁 = Victim Rx noise power at receiver input (dBW)

18 FCC, Universal Licensing System, https://www.fcc.gov/wireless/systems-utilities/universal-licensing-system (last visited June 18, 2020). 19 International Telecommunication Union, Working Document Towards a Preliminary Draft New Report ITU-R M.[RLAN SHARING 5150-5250 MHZ] - Sharing and Compatibility Studies of WAS/RLAN in the 5 150-5 250 MHz Frequency Range, Appendix 2, Section 5.1.6.7 (Nov. 2017) (noting that with regard to polarization mismatch, a value of 3 dB is considered according to what has been supported by France during TG-5.1), available at https://www.itu.int/md/R15-WP5A-171106-TD-0236/en. 20 VLP on-body device measurements were made with two orthogonal polarized detectors and the combined total gain reported. These antennas are roughly circularly polarized, whereas traditionally FS microwave stations employ linear polarization. Thus, an average polarization loss of 3 dB is reasonable.

13

● 𝑘 = Boltzmann’s constant = 1.38064852 × 10-23 m2 kg s-2 K-1

● 𝑇! = 290 K

● B = Victim Rx Bandwidth (Hz)

● 𝑁𝐹 = Victim Rx Noise Figure (dB)

With these changes, Monte-Carlo simulations were performed to calculate VLP interference to each of the 97,888 FS stations deployed over CONUS. Following the approach used in RKF's 2018 Report, for each iteration, active VLP devices were randomly placed, with their locations weighted according to the population density. The aggregate and single-entry interference power to each of the FS stations were then calculated. Many simulations were then performed to gather statistics on the interference.

A baseline simulation, with 100,000 iterations, was performed where the VLP deployment included a mix of four channel bandwidths (20, 40, 80, 160 MHz) (per Table 3-2). In each case the VLP device transmit EIRP was conservatively set to 14 dBm, independent of channel bandwidth, resulting in a power spectral density (PSD) EIRP with channel bandwidth from 1 to -8 dBm/MHz. This yields consistent sharing independent of channel bandwidth as the smaller channels with higher PSD have a lower probability of overlapping an active FS channel. This is tested through additional simulations with fixed channel bandwidth allocations. Occurrence probabilities for both I/N > -6 dB21 and > 0 dB22 were calculated.

One thousand (1,000) random FS stations were selected from those stations with at least one occurrence of I/N > -6 dB for further analysis. For these selected stations, the resulting increase in FS unavailability was calculated and analyzed.

Following these simulations, sensitivity analyses were performed where individual parameters were varied as shown in Table 2-4.

Table 2-4 - Sensitivity Analysis Variations

VLP Parameters varied Baseline Sensitivity Variations

Bandwidth (MHz) bandwidth distribution

per Table 3-2

Four simulations with 100,000 iterations using fixed channel bandwidth of 20, 40, 80, 160 MHz

Number of Active Devices 1x 2x, 3x, 6x, 12x

EIRP (dBm) 14 21

21 6 GHz Report and Order ¶ 71. 22 See 6 GHz Report and Order ¶ 131 n.339 (indicating that interference protection criteria could be relaxed by 6 dB for interference sources operating at 25% duty cycle. VLP devices are expected to operate at far lower duty cycles than 25% on average).

14

For sharing with MS, representative BAS deployments were analyzed based on a report from Alion Science and Technology23 at locations in Cowles Mountain, San Diego, CA, and a theoretical deployment at the DC Old Post Office in Washington, DC. Monte-Carlo Simulations with 100,000 iterations were performed for each ENG Central Receive station location and three (3) pointing directions to determine the I/N occurrence probabilities.

Simulation results and sharing studies with FS links are covered in Section 5.1, and MS links in Section 5.2.

The report concludes with recommended power levels for VLP devices.

23Mark Gowans & Martin Macrae, Analysis of Interference to Electronic News Gathering Receivers from Proposed 6 GHz RLAN Transmitters, Alion Science and Technology (Oct. 2019), https://ecfsapi.fcc.gov/file/1205735216211/RESED-20-002_v9.pdf (“Alion Report”).

15

3 VLP Deployment and Operating Assumptions

This section describes the analysis and methodology for assigning source quantities to the proposed 6 GHz band VLPs and their operating parameters.

3.1 VLP Deployment Assumptions

3.1.1 Number of Active VLPs and Deployment Distribution

ECC Report 31624 assumed that at any given time 1% of instantaneously transmitting devices are VLP devices operating outdoors. In accordance with this ECC report and starting with over 1 billion total VLP devices in the US, the VLP device population operating outdoors at any given time is over 10 million. With the calculated usage and activity factors, 4,417 outdoor VLP devices (1% of 441,65525 total number of all RLAN devices26 transmitting at each instant of time), referred to in this study as “number of active devices,” are predicted to be transmitting at any instant in time. This value was used as the number of active devices for the baseline simulations and the basis for the active device sensitivity analysis.

In the sensitivity analysis, the models were run representing 2x, 3x, 6x, and 12x of the baseline, whereas ECC Report 316’s sensitivity analysis modeled up to 5x VLP outdoor devices.

3.1.2 Population Density

Sharing analysis for this report used an estimated 2020 population density, based on US Census Bureau (USCB) projections, to randomly distribute the active VLPs estimated in Section 3.1.1.27 Population density thresholds, based on USCB 2010 definitions, were used to divide the country into urban, suburban, and rural28 geo areas.

24 See CEPT ECC Report. 25 As explained in Section 2.4, RKF's 2018 Report did not consider RLAN devices that operate in barren locations. Hence, the number of total active RLAN devices was slightly lower. 26 All RLAN devices refer to all transmitting 6 GHz standard-power, LPI and VLP access points and clients. 27 Socioeconomic Data and Applications Center, Gridded Population of the World (GPW), v4, NASA, http://sedac.ciesin.columbia.edu/data/collection/gpw-v4/maps/gallery/search?facets=theme:population (last visited June 27, 2020). 28 These definitions are consistent with the 2010 Census Bureau classifications (urban clusters, urbanized areas, and rural environments).

16

Figure 3-1 - 60 Arcsecond Resolution of Census Bureau Population Count Map

The resulting population and area percentages shown in Table 3-1 were used in the simulations to randomly distribute the number of VLPs estimated in Section 3.1.1 for sharing analysis with the existing FS and MS services in the 6 GHz band.

As can be seen, approximately 95% of CONUS is rural, which implies that interference will be predominantly concentrated in urban and suburban areas.

Table 3-1 - Population Density29

Population (%) Area (%)

Urban 71.2% 2.8%

Suburban 9.5% 2.2%

Rural 19.3% 95%

29 Id.

17

3.2 VLP Operating Assumptions

3.2.1 Distribution of Source VLP Power Levels including Body Loss

Antenna gain measurements were made in proximity of the human body considering various use case device positioning, static vs. dynamic conditions, device orientations, and the physical characteristics of the human body. The comprehensive on-body over-the-air measurements and analysis of the associated body loss distributions applicable to the VLP device are described in the Wireless Research Center of North Carolina study attached to the RLAN Group Comments, and shown in Figure 3-2.30 In the Monte-Carlo simulations, antenna gain values (GFarField in Eqn. 2-1) are selected randomly from the distribution in Figure 3-2.

Figure 3-2 - Probability of VLP device far-field gain > x-axis: measurements versus simulated distribution

30 Wireless Research Center of North Carolina, On-Body Channel Model and Interference Estimation at 5.9 GHz to 7.1 GHz Band at Fig. 26 (June 2020).

18

3.2.2 Bandwidth and Channel Distribution

As in RKF's 2018 Report, VLPs modeled in this report, such as those to be developed in compliance with IEEE 802.11ax Draft 6.0, are assumed to operate in 20 MHz, 40 MHz, 80 MHz, and 160 MHz bandwidth channels. To determine the number of channels, and how those channels may overlap with FS and MS receivers, the following channel plan outlined in Figure 3-3 was assumed.

Figure 3-3 - IEEE 802.11ax Draft 6.0 Channel Plan

The bandwidth distribution in Table 3-2 is based on the assumption that VLP systems will operate with larger channel sizes to maximize airtime efficiency, resulting in lower latency, higher throughput, and improved battery life. This bandwidth distribution is used in the simulations referred to as “Baseline channel Distribution” in this report.

Table 3-2 - VLP Baseline Channel Distribution

Bandwidth 20 MHz 40 MHz 80 MHz 160 MHz Percentage 10% 10% 50% 30%

3.2.3 VLP Height

VLP devices are worn on mobile users, and a large majority of these use cases are with the VLP device below 1.5 m.

19

4 Propagation Models

The analyses use propagation models adopted per the Commission’s 6 GHz Report & Order.31 As a function of the separation distance between the VLP and victim receiver, these models are as follows:

● “[F]or separation distances of 30 meters or less, the free space pathloss model is the appropriate model.”32

● “Beyond 30 meters and up to one kilometer from an unlicensed device to a microwave receiver, we find that the most appropriate propagation model is the Wireless World Initiative New Radio phase II (WINNER II).”33

● “For separation distances greater than one kilometer . . . the Irregular Terrain Model combined with a clutter model depending on the environment is the most appropriate model.”34

These models are summarized in Table 4-1 below:

Table 4-1 - Summary of Propagation Model

Distance (Slant Range) from VLP to Victim Receiver

Propagation Model

Up to 30 meters Free Space Path Loss (FSPL) 30 meters to 1 km Combined LOS/NLOS Winner II

● Urban VLP: Winner II Scenario C2 ● Suburban VLP: Winner II Scenario C1 ● Rural VLP: Winner II Scenario D1

Above 1 km ITM + Clutter model Clutter model ● Urban/Suburban VLP: ITU-R Rec. P.2108-0

(Section 3.2.2) ● Rural VLP: ITU-R Rec. P.452 Village Center

Clutter

These propagation models are very similar to the models used in RKF's 2018 Report with the exception of using Winner II for Rural VLPs (in place of ITM+P.452 Clutter) for distances < 1 km and using a combined median Winner II path loss model (instead of separate LOS and NLOS models). The combined median path loss model is computed using Eqn. 4-1 for distances between 30 m and 1 Km.

PLCWII (dB) = PLLOS (dB) x ProbLOS + PLNLOS (dB) x {1-ProbLOS} (4-1)

31 6 GHz Report and Order. 32 Id. ¶ 64. 33 See id. ¶ 66 (referencing the urban, suburban, and rural WINNER II channel models as C2, C1, and D1, respectively). See also WINNER & Information Society Technologies, WINNER II Channel Models Part 1, Table 2-1 Propagation scenarios specified in WINNER and Table 4-4 Summary table of the path-loss models, https://www.cept.org/files/8339/winner2%20-%20final%20report.pdf (“WINNER II Channel Models”). 34 See 6 GHz Report and Order ¶ 68 (referencing the Irregular Terrain Model Guide). See also G.A. Hufford et al., A Guide to the Use of the ITS Irregular Terrain Model in the Area Prediction Mode, NTIA Report 82-100 (1982), https://www.ntia.doc.gov/files/ntia/publications/ntia_82-100_20121129145031_555510.pdf.

20

where,

● PLLOS and PLNLOS are the Line-of-Sight (LOS) and NLOS Path Losses per Table 4-4 in WINNER II Report35

● ProbLOS is the LOS Probability per Table 4-7 in WINNER II Report

In RKF's 2018 Report, each deployed VLP was randomly designated as either LOS or NLOS using the LOS probability function.

In addition to the combined median path loss term, the Winner II LOS and NLOS Path Loss components include a random lognormal shadowing term that is included in the simulations.

For distances above 1 km, the methodology is as described in RKF's 2018 Report Section 4.2.2, where ITM with the SRTM 3-arc-seconds Terrain Database is used. The P.452 village center clutter loss of 18.4 dB is used for the 1.5m VLP when the following conditions are met:

● VLP elevation angle towards the victim receiver ≤ 2.86 deg (corresponding to a VLP deployed at an average distance from a village building of average height), AND

VLP distance to victim receiver ≥ 0.7 km

35 See WINNER II Channel Models.

21

5 Sharing Results

5.1 Fixed Service (FS) Sharing

This Section describes analyses performed to investigate the impact of VLP interference on FS links. Table 5-1 lists each section and its contents.

Table 5-1 - VLP to FS Interference Analyses

Section Content Description 5.1.1 ULS Database

Review Analysis of the Jan 21, 2020, ULS Database to ensure valid entries are used in the simulation.

5.1.2 Key Modeling Assumptions

Description of key modeling assumptions such as VLP device deployment, FS receiver antenna pattern, noise figure, and feeder loss.

5.1.3 Baseline I/N Occurrence Simulation

The Baseline model (1x) includes a simulation with 100,000 iterations of VLP deployments to gather I/N interference occurrence statistics at each of the 97,888 FS stations. The Baseline model is outlined in Table 2-3. As described in Section 3.1.1, 1x corresponds to 4,417 simultaneously transmitting VLP devices over CONUS.

5.1.3 Sensitivity-VLP Bandwidth

Additional simulations were performed as a sensitivity analysis with fixed channel bandwidths (20, 40, 80, and 160 MHz) to study the I/N occurrence sensitivity to channel size and power spectral density.

5.1.4 Selected FS Availability Calculations for the Baseline Model

1,000 FS stations were selected, at random, from FS stations that had at least one I/N > -6 dB occurrence in the Baseline Model in Section 5.1.3. The increase in unavailability due to VLP interference was calculated for these links. A detailed analysis was performed to determine the impact on FS link availability due to the VLP interference.

5.1.5 Ohio Example of Correlation Between Multipath Fading and Interference

The impact on the availability of an FS link in Ohio is examined using hourly multipath measurements and an estimate of hourly VLP device activity.

5.1.6.2 Sensitivity-VLP Active Devices

The number of active VLP devices was increased by 2x, 3x, 6x, and 12x to study the impact on I/N > -6 dB occurrence and FS link availability.

5.1.6.4 Sensitivity-VLP EIRP

The VLP EIRP was increased to 21 dBm to study the impact on I/N > -6 dB occurrence and FS link availability.

22

Table 5-2 provides a summary of the simulations and sensitivity analyses performed. The sensitivity analysis on the number of active devices is done for all five simulations. The sensitivity analysis on the EIRP is done for all five simulations and numbers of active devices.

Table 5-2 - Summary of FS Simulations and Sensitivity Analyses

Simulations (100,000 iterations)

Sensitivity on Number of Active Devices

Sensitivity on EIRP

Baseline Model (baseline channel distribution)

1x (baseline) 2x, 3x, 6x, and 12x

14 dBm (baseline) 21 dBm 20-MHz Simulation

40-MHz Simulation 80-MHz Simulation 160-MHz Simulation

5.1.1 ULS Database Review

The FCC’s ULS database was reviewed to determine the number of FS transmitter-receiver links in CONUS with a unique frequency channel within U-NII-5 and U-NII-7.

As shown in Table 5-3, there were 261,161 FS stations as of January 21, 2020. This count included FS links with multiple records that correspond to different EIRP and/or modulation types for each channel. After removing these duplicate links, 109,723 FS stations remained. After subtracting FS stations outside CONUS and the frequency bands of interest and those with invalid data, the total number of valid entries was calculated.

To be as conservative as possible, those FS stations with invalid data were reviewed to determine what data was missing and if that data could be replaced with average parameters based on the radio service. For example, if receive gain, antenna height, or bandwidth were invalid, they were set to average values for the corresponding radio service. If the transmitter location was missing, the receive antenna was pointed in a random azimuth uniformly distributed over 360° and a random elevation angle uniformly distributed over +/- 5°. With these changes, an additional 599 corrected links were included, resulting in a total of 97,888 FS links used in the simulations.

Table 5-3- Fixed Service Simulation ULS Database Summary (as of 01/21/2020)

Total entries in ULS database, including all EIRP/modulation types 261,161 Total entries in ULS database with 1 EIRP/modulation per FS link 109,723 Total entries outside U-NII-5 and U-NII-7 -6,647 Total entries outside of CONUS -5,100 Total entries with invalid data -687 Total valid entries 97,289 Total number of entries with data fields that were updated with assumptions based on representative criteria by RKF

+599

Total entries used in Simulations (Valid + assumed data fields) 97,888

23

Table 5-4 below shows the numbers of FS stations within the population density areas defined by the USCB.

Table 5-4 - GeoArea Type for FS Rx

Geo Area Type Number and Percentage of FS links

Urban 12,412 (12.7%)

Suburban 5,276 (5.4%)

Rural 80,200 (81.9%)

5.1.2 Key Modeling Assumptions

5.1.2.1 VLP Device Deployment

As described in Section 3.1.2, VLPs were randomly distributed throughout CONUS based on population density.

24

5.1.2.2 FS Receiver Antenna Performance

ITU-R Recommendation F.124536 was used to model the FS antenna sidelobe performance. As shown in Figure 5-1, commercial antennas (such as UHX6-59), portrayed by the red line in the figure, significantly outperform F.1245. Based on data provided by Comsearch,37 as of 2011, over 83% of antennas deployed in the 5.925-6.425 GHz band in the United States exceeded FCC Category A requirements, and over 52% of the antennas were classified as high performance or ultra-high performance, which exceed Category A side lobe attenuation requirements by up to 27.5 dB. By using F.1245 this analysis overstates the interference and provides very conservative results.

Figure 5-1 - Comparison of ITU-R 1245, FCC Category A, and Ultra-High-Performance Antenna (UHX6-59) Radiation Patterns

5.1.2.3 FS Receiver Noise Figure and Feeder Loss

An FS receiver noise figure of 5 dB and feeder loss of 2 dB38 was used.

5.1.3 Baseline I/N Occurrence Simulation and Sensitivity to VLP Channel Bandwidth

As described in Table 5-2, the Baseline Model (1x) evaluates interference to FS stations from VLP devices using the baseline channel distribution and an EIRP of 14 dBm. To more comprehensively assess the impact of channel size, four additional Monte-Carlo simulations were performed where all VLP devices were modeled using a single channel size (20, 40, 80, or 160 MHz) with an EIRP of 14 dBm. Metrics were computed for each simulation per FS station including; 1) occurrence probability

36 International Telecommunication Union, F.1245: Mathematical Model of Average and Related Radiation Patterns for Point-to-Point Fixed Wireless System Antennas for Use in Interference Assessment in the Frequency Range From 1 GHz to 86 GHz, Recommendation F.1245 (2019), available at https://www.itu.int/rec/R-REC-F.1245/en. 37 See Letter from Christopher R. Hardy to Marlene H. Dortch, WT Docket Nos. 10-153, 09-106 & 07-121 (filed Apr. 14, 2011). 38 See International Telecommunication Union, F.758: System Parameters and Considerations in the Development of Criteria for Sharing or Compatibility Between Digital Fixed Wireless Systems in the Fixed Service and Systems in Other Services and Other Sources of Interference, ITU-R F.758-6 (2019), available at https://www.itu.int/rec/R-REC-F.758/en.

25

for I/N > -6 and 0 dB; 2) occurrences due to a single VLP device; and 3) occurrences due to an aggregate of multiple VLP devices.

The occurrence probability for each FS (Poc,FSi) from the 100,000 simulation iterations is computed per Eqn. 5-1 below:

Poc,FSi = !"#$%&()*++"&&%,+%-)(&./012!"#$%&()3/%&4/.(,-

(5-1)

where,

● FSi = ith FS Receiver

● Occurrence for ith FS= an iteration where FSi has at least one single-entry I/N > I/Nthreshold

● Number of Iterations = 100,000

These percentages reflect the full range of distributions for all input variables, including scenarios with an unrealistic combination of extreme worst-case values for every input.

Figure 5-2, Figure 5-3, and Table 5-5 show the cumulative CDF of occurrence probabilities across all 97,888 FS stations for the per-FS occurrence probability at I/N > -6 dB. The occurrence of an exceedance for an individual FS station is dependent on factors such as the surrounding terrain and antenna height. As can be seen, the vast majority of the FS stations did not experience an occurrence, and the worst-case FS occurrence probability was only 0.015% (15 occurrences out of 100,000 iterations), a rare occurrence. As discussed in Section 5.1.4, the chance of these rare occurrences causing any significant impact on FS link availability is incredibly small.

With the higher population density in urban and suburban areas, FS links are more likely to see occurrences. For example, the median CONUS FS link distance is 25 Km, whereas FS links with multiple occurrences in the simulations had a median link distance of 17 KM. However, with these shorter distances, these FS links have higher link margins and can generally accept interfering signals 1-10 dB or more above long-haul performance requirements and not affect long-term performance.39

39 National Telecommunications and Information Administration, Interference Protection Criteria Phase 1 - Compilation from Existing Sources, NTIA Report 05-432, 4-8, 4-9 (2005), https://www.ntia.doc.gov/files/ntia/publications/ipc_phase_1_report.pdf (“NTIA Report 05-432”).

26

Figure 5-2 - Probability of “I/N > -6 dB per-FS occurrence probability” exceeding Values on X-axis for 97,888 FS Links in each of the five simulations (different bandwidth models)

Figure 5-3 - Probability (log-scale) of “I/N > -6 dB per-FS occurrence probability” exceeding Values on X-axis for 97,888 FS Links in each of the five simulations (different bandwidth models) - Zoomed-in under 10%

27

Table 5-5 - Percentage of FS (out of 97,888) having an “I/N > -6 dB occurrence probability” in the leftmost column in each of the five 100,000-iteration simulations

per-FS Occurrence Probability (Poc,FSi)

20-MHz 40-MHz 80-MHz 160-MHz Baseline Channel Distribution

0% 90.502% 90.701% 91.008% 91.795% 91.070%

0.001% 8.052% 7.868% 7.417% 6.745% 7.428%

0.002% 1.155% 1.112% 1.184% 1.045% 1.143%

0.003% 0.194% 0.224% 0.245% 0.278% 0.232%

0.004% 0.058% 0.061% 0.074% 0.072% 0.078%

0.005% 0.025% 0.016% 0.036% 0.031% 0.019%

0.006% 0.005% 0.009% 0.017% 0.014% 0.016%

0.007% 0.003% 0.006% 0.008% 0.007% 0.004%

0.008% 0.002% 0% 0.005% 0.005% 0.002%

0.009% 0.002% 0% 0.001% 0.001% 0.005%

0.010% 0% 0.001% 0.001% 0.002% 0.001%

0.011% 0% 0% 0.001% 0.002% 0%

0.012% 0.001% 0% 0.003% 0% 0%

0.013% 0% 0.001% 0% 0.001% 0%

0.014% 0% 0% 0% 0% 0.001%

0.015% 0% 0% 0% 0.002% 0%

28

Similarly, Figure 5-4, Figure 5-5, and Table 5-6 show the Cumulative CDF of “single-entry I/N > 0 dB per-FS occurrence probability.” The results show that in each of the simulations, 98% of the 97,888 FS had no occurrence, and the worst-case FS occurrence probability was only 0.007% (7 occurrences out of 100,000 iterations), a very rare occurrence. As discussed in Section 5.1.4, the chance of these rare occurrences causing any significant impact on FS link availability is incredibly small.

Figure 5-4 - Probability of “I/N > 0 dB per-FS occurrence probability” exceeding Values on X-axis for 97,888 FS Links in each of the

five simulations (different bandwidth models)

Figure 5-5 - Probability (log-scale) of “I/N > 0 dB per-FS occurrence probability” exceeding Values on X-axis for 97,888 FS Links in each of the five simulations (different bandwidth models) - Zoomed-in under 10%

29

Table 5-6 - Percentage of FS (out of 97,888) having an “I/N > 0 dB occurrence probability” in the leftmost column in each of the five 100,000-iteration simulations

per-FS Occurrence Probability (Poc,FSi)

20-MHz 40-MHz 80-MHz 160-MHz Baseline Channel Distribution

0% 97.601% 97.805% 97.881% 98.171% 97.937%

0.001% 2.256% 2.044% 1.942% 1.647% 1.879%

0.002% 0.135% 0.133% 0.150% 0.152% 0.162%

0.003% 0.007% 0.017% 0.019% 0.020% 0.019%

0.004% 0% 0.001% 0.006% 0.006% 0.002%

0.005% 0.001% 0% 0% 0.003% 0%

0.006% 0% 0% 0% 0% 0%

0.007% 0% 0% 0.001% 0% 0%

The 500,000 iterations, with almost 50 billion FS interference assessments of the Baseline Model and the four fixed bandwidth simulations, showed that only one occurrence was caused by an aggregation of multiple VLP devices using a 40 MHz bandwidth. Based on these results, and as recognized by the Commission, the rest of this report assumes that all metrics are based on the single-entry I/N levels.40

The results of the Baseline Model and bandwidth sensitivity simulations are summarized in Table 5-7. The results show that:

- I/N > -6 dB average occurrence probability of a single FS is 0.00011% or 0.11 out of 100,000 iterations.

- The occurrence probability is not sensitive to PSD caused by bandwidth variations but is determined by the device total EIRP.

This is explained by the fact that narrower bandwidths correspond to proportionally higher PSD, but also a proportionally lower chance that the VLP’s bandwidth overlaps with the FS’s channel: i.e., reducing bandwidth from 160 MHz to 20 MHz increases the power density by 8x but also reduces the probability of channel overlap by a factor of eight. These two factors essentially negate each other.

Table 5-7 - Average Interference Statistics from all Independent Simulations (100,000 iterations per bandwidth model) of a CONUS-Wide VLP Deployment

I/N Threshold 20 MHz 40 MHz 80 MHz 160 MHz Baseline Channel Distribution

-6 dB 0.00011% 0.00011% 0.00011% 0.00010% 0.00011% 0 dB 0.00003% 0.00002% 0.00002% 0.00002% 0.00002%

40 6 GHz Report and Order ¶ 72.

30

5.1.4 Selected FS Availability Calculations for the Baseline Model

The availability analysis assumed a typical FS design target of 99.999% availability (unavailability=0.001% corresponding to 5.3 minutes/year). Results are compared to a target increase in unavailability of less than 10%, as established by the ITU,41 that is sufficient to allow continued robustness of FS links.

The increase in unavailability due to VLP interference was further analyzed, using a two-step process, by looking at 1,000 specific FS Stations chosen from those that had at least one occurrence of I/N > -6 dB in the Baseline Model simulation with the baseline channel distribution and looking at the specific impact on unavailability due to VLP devices.

First, a fade margin required to achieve the target availability of 99.999% was determined using ITU-R Rec. P.530-17 (P.530). Then, the increase in unavailability in the presence of interference was assessed.

Second, if an FS link’s unavailability increased more than 10% in Step 1, the actual operating parameters were examined to determine the available fade margin. These links were then reassessed to determine if they would meet the 10% target.

The fade margin probability density function (pdf) is obtained from P.530 (section 2.3.2 Eqn. 18) using FS unavailability and the multipath occurrence factor, 𝑝5. 𝑝5 provides the fade margin required for the average worst month and is computed using P.530 (section 2.3.2, Eqn. 11), with input parameters from the ULS database. The input parameters are the FS Transmitter (Tx) and Receiver (Rx) terrain height, antenna height above ground level, link distance, and center frequency.

Given the fade margin pdf and the pdf of the degradation due to VLP interference for a specific FS (i.e., (I+N)/N from the 100,000-iteration simulation), the impact on FS link unavailability can be determined directly from the combined distribution. The convolution provides the correct answer to this question under the assumption that the two random variables (fading and interference) are independent. This independence is a conservative approximation. In fact, there is an inverse relationship between VLP device activity and when multipath fading occurs. As the Commission reaffirmed TSB-10F guidance42, multipath fading occurs between midnight and 8 am,43 while outdoor VLP usage will primarily be during daylight hours. This inverse correlation, explored in Section 5.1.5, means that the sum of interference and fading is statistically smaller than what is modeled.

The current analysis is for 1,000 FS links selected randomly from stations that had exceedances in the Baseline Model simulation. Choosing 1,000 FS from all FS stations that had exceedances provides a significant statistical representation because calculating the availability of 10,000 or more FS stations would be arduous and would not provide additional insight.

41 International Telecommunication Union, F.1094-2: Maximum Allowable Error Performance and Availability Degradations to Digital Fixed Wireless Systems Arising from Radio Interference from Emissions and Radiations from Other Sources (2007), available at https://www.itu.int/rec/R-REC-F.1094/en. 42 6 GHz Report and Order ¶ 143. 43 See NTIA Report 05-432.

31

Table 5-8 shows the occurrence probabilities for the 1,000 FS links analyzed. As indicated, the 1,000 FS included FS from all ranges of occurrence probabilities, including the worst-case. Table 5-9 shows the geo area (urban, suburban, or rural) identified with each FS Receiver station. Note that all geo areas are represented.

Table 5-8 - I/N > -6 dB Occurrence Probabilities for the 1,000 FS

I/N > -6 dB Occurrence Probability (Poc,FS)

Number of FS (out of 1,000)

0.001% 846 0.002% 122 0.003% 19 0.004% 9 0.005% 2 0.008% 1 0.014% 1

Table 5-9 - FS Rx GeoArea Type (as defined in Section 3.1.2) for the 1,000 FS

FS Rx GeoArea Type % of FS Urban 27.5% Suburban 9.4% Rural 63.1%

Furthermore, for accuracy, the full I/N distribution is used in the analysis including all aggregate interference events.

32

In Step 1, results showed that the 10% unavailability target was met for 989 FS (out of 1,000). The increase in unavailability for these 989 FS is shown in Figure 5-6. As indicated, 92.6% of the 1,000 FS had less than 0.01% increase and the highest increase (among the 989 FS) was 8.19%.

Figure 5-6 - Increase in unavailability for 989 FS that meet the 10% target.

The analysis in Step 1 assumes that each FS link has the exact margin to achieve the target availability. However, given that amplifiers and antennas only come in certain sizes, it is unlikely that these links achieve this margin exactly. In Step 2, the 11 links that failed to meet the 10% unavailability target are examined more closely. As indicated below, after considering the actual FS link operating parameters at highest order modulations in Step 2, they all meet the 10% target.

33

These 11 FS all had very small 𝑝5’s (0.03 to 2.2x10-6) that resulted in very low fade margins (5.3 to 15.9 dB), which made them sensitive to interference. They are also all short-haul links (≤ 20 Km). As noted in Section 5.1.3, the IPC for FS is generally based on long-haul performance, which is usually acceptable on shorter digital paths.44 Table 5-10 shows the link characteristics of these FS stations.

Table 5-10 - Link characteristics of the FS with Increase in Unavailability > 10% using theoretical link characteristics

FS Tx/Rx CallSign FS Tx EIRP (dBm) [ULS]

FS Tx Power45 (Watt)

FS link distance (km) [ULS]

Received C/N (dB) (Eqn. 5-2)

Multipath occurrence factor, p0 (ITU-R P.530)

WRAN774/WQYE338 61.6 0.51 8.85 59.16 0.000004

WQOH921/WQOH921 48.8 0.01 2.56 63.59 0.00001

WPWY772/WQGB861 67.1 0.69 10.75 69.00 0.00008

WQXG421/WQXG421 58.6 0.20 7.07 65.32 0.00006

WQLY872/WPNJ907 53.3 0.02 6.28 69.37 0.00002

WNEH663/WNEH669 37.2 0.00 2.38 57.27 0.00029

WQNF440/WQNF440 58.5 0.10 17.48 56.32 0.002

WPNJ867/WPNJ873 57.9 0.29 8.65 64.93 0.005

WPWT263/WPNM710 56.1 0.03 8.91 70.25 0.00002

WQDD269/WPJD957 42.8 0.00 1.45 69.69 0.000002

WQQW474/WQQW473 68.8 1.00 20.04 61.40 0.030

The ULS database information was used to compute the C/N at the receiver, shown in Table 5-10, using Eqn. 5-2 below:

"#(𝑑𝐵) = 𝐸𝐼𝑅𝑃(𝑑𝐵𝑊) − 𝐹𝑆𝑃𝐿(𝑑𝐵) − 𝐿$%%& (𝑑𝐵) + 𝐺' (𝑑𝐵𝑖) − 𝑁(𝑑𝐵𝑊) (5-2)

where,

● EIPP (dBW) = FS EIRP from the ULS database

● FSPL (dB) = 92.45 + 20*log10(FS link distance [km]) + 20*log10(center frequency [GHz])

● Lfeed = FS Rx Feederloss = 2 dB

● GR = FS Rx Gain (dBi) from the ULS database

● N = Noise Power (dBW) = -228.6 dB(W/K/Hz) + 10*log10(T) + 10*log10(B [Hz])

● T = System temperature = 290 K

● B = FS channel bandwidth (Hz)

44 NTIA Report 05-432 at 4-8, 4-9. 45 FS transmit power is calculated using the EIRP and transmit gain found in the ULS database.

34

The actual FS fade margin, Fa, is then computed as shown in Eqn. (5-3).

Fa (dB) = C/N (dB) – (max) C/Nreq (dB) (5-3)

In the ULS database, as in practice, FS links may operate using more than one modulation. The higher the modulation order, the higher the C/Nreq and the lower the actual fade margin. The analysis that follows is for the worst case and assumes the FS links use their highest-order modulation available. This does not mean that these links cannot adapt to use a lower-order modulation if they are degraded.

The modulations in the ULS database for the 11 FS links varied from 4QAM to 1024QAM, 32TCM, and 128TCM. For each of these, the highest-order modulation was selected.46

Table 5-11 shows C/Nreq values obtained from several manufacturers' datasheets. The 30 MHz channels have a range of values that indicate different coding and receiver performance. For the analysis, the maximum C/Nreq values are used (indicated in bold). This will provide the most conservative answer.

Table 5-11- SNR required used for the 11 FS based on the link’s highest modulation and bandwidth

Modulation Bandwidth (MHz) C/Nreq (dB) Manufacturers 4-QAM 30 5.7-6.2 SAF Integra,

Redline RDL 5000, and

ALFOplus47

16-QAM 30 11.7-12.2 32-QAM 30 16.2-16.7 64-QAM 30 15.7-19.7 128-QAM 30 18.7-23.2 256-QAM 30 22.2-26.2 1024-QAM 30 30.2-31.2 64-QAM 10 19.5 SAF Integra

256-QAM 10 26 1024-QAM 60 31.2 32-TCM 3.75 20.3 Alcatel MDR-

850648 128-TCM 5 26 Alcatel MDR-

670649 128-TCM 30 24.2 Alcatel MDR-

870650

46 One link (Tx/Rx CallSign WNEH663/WNEH669) did not have any modulation listed in the ULS. From the available manufacturer data sheets and for a 10 MHz bandwidth, the highest modulation for this link was determined to be 256-QAM. 47 See SAF Tehnika, SAF Integra Datasheet, https://www.ispsupplies.com/content/datasheets/Integra%20series%20DS%20v1.43.pdf; Redline Communications, RDL-5000 Datasheet, https://rdlcom.com/wp-content/uploads/Redline-DS-RDL-5000.pdf; SIAE Microelettronica, ALFOplus2 Datasheet, available at https://www.siaemic.com/index.php/products-services/telecommunication-systems/microwave-product-portfolio/alfo-plus2. 48 Alcatel-Lucent, Alcatel-Lucent MDR-8000, http://cdn.dreamingcode.com/public/129/MDR-8X06-129-785-1.pdf (“MDR-8X06”). 49 Alcatel-Lucent, MDR-6X06, http://cdn.dreamingcode.com/public/129/MDR-6X06-129-733-1.pdf. 50 MDR-8X06.

35

Table 5-12 summarizes the key performance parameters for each link including the Fade Margin (FM) at the 99.999% availability target, the received C/N (Eqn. 5-2), C/Nreq (from Table 5-11), and Fa (Eqn. 5-3). The actual link fade margin is then compared to the FM at 99.999% availability and the difference is the “Actual Margin Above FM” (column C5). Notice the calculated “Actual Margin above FM” is very high for these links (>14.3 dB).

Next, the additional margin to meet the 10% target is determined and is shown in column C6.

Finally, the “Actual Margin above FM” (C5) is compared against the “Increase in FS link margin to meet the 10% target” (C6). The results show that the actual operating parameters on these 11 links led to more than sufficient margin to meet the 10% target.

To further demonstrate the robustness of this analysis, 1% increase in unavailability was studied as a sensitivity analysis and shown in column C7. As indicated in (C6) and (C7), the overall interference risk from VLP operations is so low that nearly the same margin is necessary to achieve both 10% and 1% increase in unavailability.

This shows that all the 1,000 links meet the 10% increase in unavailability target as well as the sensitivity analysis down to 1% increase in unavailability.

Table 5-12 - FS with Increase in Unavailability > 10% had “Actual Margin beyond FM” (C5) >> “Increase in FS Link Margin to meet 10% target (C6) and 1% sensitivity (C7)”

FS Tx/Rx CallSign FM (dB) @ 99.999%

Received C/N (dB) (Eqn. 5-2)

C/Nreq (dB)

Fa (dB) (Eqn. 5-

3)

Actual Margin

(dB) above FM

Increase in FS Link Margin

(dB), 𝑥, to meet 10%

target

Increase in FS Link Margin

(dB), 𝑥, to meet 1% (sensitivity)

Column C1 C2 C3 C4=C2-C3

C5=C4-C1

C6 C7

WRAN774/WQYE338 5.59 59.16 31.2 27.96 22.37 2.92 2.93 WQOH921/WQOH921 5.90 63.59 26.2 37.39 31.49 1.32 1.32 WPWY772/WQGB861 7.15 69.00 31.2 37.80 30.64 0.94 0.98 WQXG421/WQXG421 7.03 65.32 26 39.32 32.29 2.69 6.36 WQLY872/WPNJ907 6.33 69.37 20.3 49.07 42.73 0.44 0.49 WNEH663/WNEH669 8.19 57.27 26 31.27 23.08 0.43 0.50 WQNF440/WQNF440 10.53 56.32 24.2 32.12 21.59 4.56 4.56 WPNJ867/WPNJ873 11.94 64.93 26.2 38.73 26.79 2.38 2.38 WPWT263/WPNM710 6.20 70.25 26 44.25 38.06 0.30 0.35 WQDD269/WPJD957 5.32 69.69 19.5 50.19 44.87 1.74 3.90 WQQW474/WQQW473 15.91 61.40 31.2 30.20 14.29 0.01 0.22

36

5.1.5 Ohio Example of Correlation Between Multipath Fading and Interference

This section provides further evidence that VLP operations can coexist with Fixed Service operations.

An example is provided to demonstrate that when deep fades and VLP interference are not coincident, the degradation to FS links from VLP interference is negligible. This conclusion is supported by the Commission’s findings:51

“... potential degradation of a microwave link will only occur if a deep atmospheric multipath fade occurs at the same time the microwave receiver receives an excessively high powered transmission from an unlicensed device ... Thus, because the Wi-Fi access point busy hour is not between the 8-hour period after midnight, we conclude that the likelihood of harmful interference to fixed service microwave links from indoor low power Wi-Fi access points is insignificant.”

First, VLP outdoor operations are even less likely to occur during periods of deep microwave fade than other types of RLAN operations. Per the Encyclopedia of Public Health, “The average North American spends approximately 90 percent of the time indoors, 5 percent in cars, and only 5 percent outdoors.52” The vast majority of this time spent outdoors is during the daylight hours.53

To demonstrate the inverse correlation between VLP operations and deep microwave fading, an analysis was performed on an Ohio FS link where hourly fading statistics were available.

Fading measurements were publicly available over a 68-day period in the late summer of 1966 (covering the worst fading month and fading days) in West Unity, Ohio54 over a 28.5 Km path that is nearby the FS with Tx/Rx CallSign of WPNJ867/WPNJ873. The hourly fade measurements from West Unity, Ohio can be applied to this FS Tx/Rx CallSign.

This link is short-haul and requires less than 12 dB of ITU derived fade margin to achieve the availability required to meet five-nines, making this link less resistant to interference compared to the other FS links having more fade margin. This FS had the third-highest increase in unavailability in the analysis of Section 5.1.4 and is shown in Table 5-10 and Table 5-12 above.

51 6 GHz Report and Order ¶ 143. 52 Trevor Hancock, Encyclopedia of Public Health - Built Environment, encyclopedia.com (2002), http://www.encyclopedia.com/doc/1G2-3404000130.html. 53 Neil E. Klepeis, William C. Nelson, Wayne R. Ott, et al., The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants, 11 Journal of Exposure Science and Environmental Epidemiology 231 (2001), available at https://doi.org/10.1038/sj.jea.7500165. 54 W.T. Barnett, Multipath Propagation at 4, 6, and 11 GHz, The Bell System Technical Journal, Vol. 51, No. 2 at Figure 13 (Feb. 1972).

37

Figure 5-7 shows measured hourly fading statistics for the Ohio path for three deep fade ranges:

● Level 1: Fades less than 9.8 dB (not shown on graph)

● Level 2: Fades between 9.8 and 20.4 dB

● Level 3: Fades between 20.4 and 31 dB

● Level 4: Fades between 31 and 40.1 dB

Figure 5-7 shows the fraction of total fade time in a given hour. The Level 2 and above fades only occurred in the eleven-hour period between 10 P.M. and 9 A.M. when VLP activity is minimal. The hashed, shaded region in the figure shows the fading attributed to the worst three measurement days. The large majority of these fades occurred during this short period. As shown in Figure 5-7, for each of Levels 2, 3, and 4, the sum of the hourly probabilities of fade from 10 P.M. to 9 A.M. is equal to 100%. This implies that there are no fades at Level 2 or above from 9 A.M. to 10 P.M.

Figure 5-7 - 6 GHz Hour-of-Day Ranking, 1966 West Unity, Ohio55

55 Id.

38

The expected fade was derived during daylight hours when the VLP activity is highest. The 8 A.M. to 9 A.M. period (“worst hour”) was the overlapping period with the highest probability of fade. The measurements indicate that approximately 8% of daily fades (greater than 9.8 dB - Level 2 and above) during the worst month of fade activity occurred within this worst hour. Figure 5-7 shows that independent of the fade depth (Levels 2, 3, or 4), the percentage of fades was the same (8%).

To model the increase in unavailability due to VLP interference, P.530 was first used to generate the fading distribution for the worst average month for the FS link. From this distribution, only 8% of the deep fades (> 9.8 dB) would be attributed to the period between 8 A.M. and 9 A.M. Thus, the P.530 fade probabilities for fades above 9.8 dB were scaled by 8%.56

Figure 5-8 shows the worst-case average monthly fade distribution (P.530) and the rescaled distribution for the worst daylight hour (P.530 rescaled).

Figure 5-8 - Ohio FS (Tx/Rx CallSign of WPNJ867/WPNJ87) Fade Distribution per P.530 and scaled probability of deep fade above 9.8dB between hours 8 A.M. and 9 A.M. (p0 = 0.005 per Table 5-10).

56 Note that the fade probabilities below 9.8 dB had to be renormalized for the worst hour so that the total probability still integrated to one.

39

The convolution was performed with the rescaled worst hour (8 A.M. to 9 A.M.) fading distribution and the simulated VLP interference distribution (from Baseline Model) for this FS to determine increase in unavailability. In the Section 5.1.4 analysis, this short-haul FS is very sensitive to VLP interference (increase in unavailability >10%) because it requires very low fade margin for the target availability. However, when considering the worst fade hour, when VLPs are active, the predicted increase in unavailability is reduced by a factor of 12.5 to < 10%. Furthermore, as demonstrated in Section 5.1.4, when considering the operational parameters of this link, the increase in unavailability was well below the target. Along with the limited coincidence between the fades and VLP activity during daylight hours and the operational parameters of this link, the risk of harmful interference is negligible.

This exemplifies that the unavailability results considered in Section 5.1.4 if using real-world VLP behavior can be reduced further by at least one order of magnitude when considering hourly fade distribution.

5.1.6 Sensitivity Analysis

This section considers the sensitivity of the Baseline Model simulation and the four bandwidth-sensitivity simulations, referred to collectively as baseline simulations, to the number of active VLP devices and their EIRPs.

5.1.6.1 Number of Active Devices (2x, 3x, 6x, and 12x)

The baseline simulations of Section 5.1.3 were repeated with the number of active VLP devices scaled up by 2x, 3x, 6x, and 12x.57 As in the baseline simulations, there were 97,888 FS stations across CONUS.

5.1.6.1.1 Occurrence Probabilities

Table 5-13 and Table 5-14 show occurrence probabilities for I/N > -6 dB and 0 dB, respectively for the channel bandwidths and numbers of VLP devices simulated. All simulations assume that the VLP devices are transmitting with an EIRP = 14 dBm.

As expected, the average probability of occurrence increases linearly with the number of active devices. As the prior simulations show, interference in the simulations with 12x devices was almost always dominated by a single VLP device. Over all simulation iterations, only 0.5%58 of 97,888 FS stations had an aggregate I/N different from the peak single-entry I/N with a maximum difference of 2 dB.

Note that when increasing the number of VLP devices, the results continue to be independent of the channel size assumed and interference is no worse if all devices are transmitting 14 dBm in 20 MHz. Thus, there is no reason to limit the PSD to -8 dBm/MHz (14 dBm/160 MHz).

57 For computational efficiency, statistics for 6x were based on 50,000 iterations and for 12x were based on 25,000 iterations instead of 100,000. 58 For the remaining 99.5 % of simulated I/N values, the aggregate and peak I/N differed by less than 0.00001 dB.

40

Table 5-13 - Average Interference Statistics from all Independent Simulations for -6 dB I/N of a CONUS-Wide VLP Deployment for 1x, 2x, 3x, 6x, and 12x number of active VLPs

Number of Active VLPs

20 MHz 40 MHz 80 MHz 160 MHz Baseline Channel Distribution

1x 0.00011% 0.00011% 0.00011% 0.00010% 0.00011% 2x 0.00023% 0.00023% 0.00023% 0.00021% 0.00022% 3x 0.00034% 0.00034% 0.00034% 0.00032% 0.00033% 6x 0.00069% 0.00067% 0.00068% 0.00063% 0.00066% 12x 0.00137% 0.00135% 0.00136% 0.00126% 0.00133%

Table 5-14 - Average Interference Statistics from all Independent Simulations for 0 dB I/N of a CONUS-Wide VLP Deployment for 1x, 2x, 3x, 6x, and 12x number of active VLPs

Number of Active VLPs

20 MHz 40 MHz 80 MHz 160 MHz Baseline Channel Distribution

1x 0.00003% 0.00002% 0.00002% 0.00002% 0.00002% 2x 0.00005% 0.00005% 0.00005% 0.00004% 0.00005% 3x 0.00008% 0.00007% 0.00007% 0.00006% 0.00007% 6x 0.00015% 0.00014% 0.00014% 0.00012% 0.00014% 12x 0.00030% 0.00029% 0.00029% 0.00024% 0.00027%

41

Figure 5-9 shows the complementary CDF for baseline channel distribution of the single-entry occurrence probabilities for I/N > -6 dB (left Figure) and 0 dB (right Figure) for active devices from 1x to 12x above the baseline. The Figure shows that the worst-case I/N > -6 dB occurrence probability increases from 0.014% (1x) to 0.124% (12x).

The results show that the number of I/N > -6 dB and 0 dB occurrences increase roughly linearly with increase in the number of active devices.59

Figure 5-9- Comparison of per-FS occurrence probabilities of I/N > -6 dB (left Figure) and 0 dB (right Figure) between 1x, 2x, 3x, 6x, and 12x number of active VLPs from Baseline channel distribution simulation

5.1.6.1.2 Impact on FS link availability resulting from a 2x, 3x, 6x, and 12x increase in the number of active VLP devices

Following the methodology in Section 5.1.4, the impact on FS link availability for the same randomly chosen 1,000 FS with at least one I/N > -6 dB occurrence was analyzed using the aggregate I/N distributions corresponding to 2x, 3x, 6x, and 12x number of active VLP devices.

From Step 1, for 2x, 3x, 6x, and 12x number of active VLPs, there were 15, 16, 17, and 20 FS respectively (out of 1,000 FS) that did not meet the 10% target using the FM (calculated based on the target availability). These corresponded to the 11 FS in Table 5-10 plus the nine FS in Table 5-15. As indicated below, after considering the actual FS link operating parameters at highest order modulations in Step 2, they all meet the 10% target.

59 This is as expected. If the 1x simulation has a probability of occurrence = p, scaling the number of RLAN's by a factor of N (Nx simulation) would have a probability of occurrence of: 1-(1-p)^N. For p<<1, this reduces to N*p.

42

Table 5-15 - Link Characteristics of the additional nine FS with Increase in unavailability > 10% for 2x, 3x, 6x, and 12x number of active devices (per last column)

FS Tx/Rx CallSign FS Tx EIRP (dBm) [ULS]

FS Tx Power (Watt)

FS link distance (km) [ULS]

Received C/N (dB) (Eqn. 5-2)

Multipath occurrence factor, p0 (ITU-R P.530)

Cases for which Increase in Unavailability > 10%

WQUL352/ WPON281 58.9 0.08 13.97 59.63 0.004 2x-12x

WNTZ387/ WNTZ386 44.9 0.004 1.9 61.13 0.00007 2x-12x

WHI696/ WPUH740 70.3 0.66 23.65 75.95 0.002 2x-12x

WNTJ676/ WNTJ921 52.8 0.02 23.69 57.68 0.003 2x-12x

WQEF238/ WQEE778 67.9 0.81 23.50 62.96 0.006 3x-12x

WNTU484/ WNTU484 67 0.65 8.72 71.23 0.002 6x-12x

WNTQ313/ WNTQ313 53.2 0.03 6.69 58.90 0.005 12x

WQNQ502/ WQNQ502 51.2 0.02 12.42 56.98 0.00003 12x

KLG67/ KLG67 63.5 0.63 50.75 57.09 0.019 12x

43

Table 5-16 summarizes the calculations in Step 2 for all 20 links. As indicated, the actual operating parameters on these 20 links led to more than sufficient margin to meet the 10% target. Note that as indicated in the last three columns, some of the FS links met the 10% target for 2x, 3x, or 6x number of active VLPs. In addition, the “Increase in FS Link Margin to meet 10% target” was the same between 2x and 3x scenarios for the top 15 FS links because the maximum aggregate I/N was the same under both scenarios. Therefore, all 1,000 FS met the 10% unavailability target when the number of active VLPs was increased to 2x, 3x, 6x, and even 12x from the baseline.

Table 5-16 - 2x, 3x, 6x, and 12x number of active VLPs: 15, 16, 17, and 20 FS with Increase in Unavailability > 10% had “Actual Margin beyond FM” > “Increase in FS Link Margin to meet 10% target”

FS Tx/Rx CallSign FM (dB) @

99.999%

Received C/N

(dB) (Eqn. 5-2)

C/Nreq (dB)

Fa (dB) (Eqn. 5-3)

Actual Margin

(dB) above FM

Increase in FS Link Margin

(dB), 𝑥, to meet 10%

target (2x,3x)

Increase in FS Link Margin

(dB), 𝑥, to meet 10% target (6x)

Increase in FS Link Margin

(dB), 𝑥, to meet 10%

target (12x)

Column C1 C2 C3 C4=C2-C3

C5=C4-C1

C6 C7 C8

WQXG421/ WQXG421 7.03 65.32 26 39.32 32.29 5.28 8.83 9.18

WRAN774/ WQYE338 5.59 59.16 31.2 27.96 22.37 2.92 3.06 3.36

WPWY772/ WQGB861 7.15 69.00 31.2 37.80 30.64 0.94 1.25 1.54

WQUL352/ WPON281 11.50 59.63 26.2 33.43 21.93 2.72 3.02 3.43

WNTZ387/ WNTZ386 7.09 61.13 26.2 34.93 27.85 0.79 1.08 1.37

WQOH921/ WQOH921 5.90 63.59 26.2 37.39 31.49 1.32 1.49 1.77

WHI696/ WPUH740 9.97 75.95 26 49.95 39.98 1.37 1.74 2.10

WNEH663/ WNEH669 8.19 57.27 26 31.27 23.08 0.43 0.66 0.91

WQLY872/ WPNJ907 6.33 69.37 20.3 49.07 42.73 0.44 0.66 0.90

WPWT263/ WPNM710 6.20 70.25 26 44.25 38.06 0.30 0.48 0.69

WQNF440/ WQNF440 10.53 56.32 24.2 32.12 21.59 4.56 4.74 5.16

WNTJ676/ WNTJ921 10.92 57.68 20.3 37.38 26.45 0.91 1.23 1.57

WPNJ867/ WPNJ873 11.94 64.93 26.2 38.73 26.79 2.38 2.74 3.14

WQDD269/ WPJD957 5.32 69.69 19.5 50.19 44.87 1.74 8.77 9.07

WQQW474/ WQQW473 15.91 61.40 31.2 30.20 14.29 0.01 0.20 0.45

WQEF238/ WQEE778 12.12 62.96 31.2 31.76 19.64 2.04 (3x only)

2.43 2.83

WNTU484/ WNTU484 10.12 71.23 23.2 48.03 37.91 N/A 0.08 0.23

WNTQ313/ WNTQ313 11.90 58.90 23.2 35.70 23.80 N/A N/A 0.07

WQNQ502/ WQNQ502 6.45 56.98 19.5 37.48 31.02 N/A N/A 0.03

KLG67/ KLG67 14.68 57.09 26 31.09 16.41 N/A N/A 0.01

44

Next, as a sensitivity analysis, Table 5-17 shows the number of FS stations not meeting 10% and 1% increase in unavailability in Step 1. The table shows that even when the number of active devices was increased by 12x, at 1% increase in unavailability, only 27 out of the 1,000 links required analysis of their real-world operating parameters. Further analyses of Step 2 indicated that the actual operating parameters on these 27 links led to more than sufficient margin to meet the sensitivity analysis down to 1% increase in unavailability.

Table 5-17 - Number of FS stations with increase in unavailability > 10% and 1%, without consideration of actual margin on the FS link, when the number of active VLP devices is 1x, 2x, 3x, 6x, and 12x

Increase in Unavailability

1x 2x 3x 6x 12x

10% 11 15 16 17 20 1% 15 21 22 25 27

5.1.6.2 Higher EIRP Level

This section studies the impact on FS performance of increasing the VLP EIRP from 14 dBm to 21 dBm. All other assumptions are consistent with the baseline simulations of Section 5.1.3. The I/N distributions from the five simulations with 14 dBm EIRP (Section 5.1.3) were increased by 7 dB to get I/N distributions for 21 dBm EIRP.

5.1.6.2.1 Occurrence Probabilities

Table 5-18 shows average I/N > -6 dB occurrence probabilities (average of Poc,FS over 97,888 FS) for the five different channel models at 21 dBm EIRP. Note that the worst-case occurrence probability of any of the 97,888 FS stations, over all simulation iterations, is 0.050%.

As indicated, the occurrence probability scales linearly with EIRP, resulting in an increase in the occurrence probability by five times (7 dB difference) for 21 dBm EIRP.

Table 5-18 – Average Interference Statistics from all Independent Simulations for -6 dB I/N (100,000 iterations per bandwidth model) of a CONUS Wide VLP Deployment for 14 dBm and 21 dBm EIRP (1x number of active VLPs)

VLP EIRP 20 MHz 40 MHz 80 MHz 160 MHz Channel Distribution

14 dBm (baseline) 0.00011% 0.00011% 0.00011% 0.00010% 0.00011% 21 dBm 0.00056% 0.00058% 0.00061% 0.00060% 0.00059%

45

Figure 5-10 shows a comparison between single-entry occurrence probabilities (Poc,FS) of 14 dBm and 21 dBm EIRP for the Baseline channel distribution simulation.

Figure 5-10 - Comparison of I/N > -6 dB per-FS occurrence probability between 14 dBm (baseline) and 21 dBm EIRP for baseline channel distribution Simulation (1x number of active VLP devices)

Table 5-19 shows the average I/N > -6 dB occurrence probabilities from 21-dBm EIRP VLPs for each of the bandwidth-model simulations when the number of active VLPs is increased to 2x, 3x, 6x, and 12x.

Table 5-19 – Average Interference Statistics from all Independent Simulations for -6 dB I/N of a CONUS-Wide VLP Deployment for 1x, 2x, 3x, 6x, and 12x number of active VLPs (21 dBm EIRP)

Number of Active VLPs

20 MHz 40 MHz 80 MHz 160 MHz Baseline Channel Distribution

1x 0.00056% 0.00058% 0.00061% 0.00060% 0.00059% 2x 0.00111% 0.00115% 0.00122% 0.00121% 0.0012% 3x 0.00167% 0.00173% 0.00183% 0.00181% 0.0018% 6x 0.00334% 0.00346% 0.00365% 0.00363% 0.00359% 12x 0.00667% 0.00692% 0.00731% 0.00726% 0.00718%

5.1.6.2.2 Impact on FS link availability resulting from 21 dBm EIRP

Next, impact on FS availability was analyzed when the device EIRP = 21 dBm for the same 1,000 randomly chosen FS. The increase in unavailability was calculated for the Baseline channel distribution simulation with 1x, 2x, 3x, 6x, and 12x number of active VLPs.

46

Table 5-20 shows the number of FS stations with an increase in unavailability > 10% and 1% in Step 1, and for comparison shows results for the baseline 14 dBm EIRP presented in Section 5.1.6.1.2.

Table 5-20 – Number of FS stations with increase in unavailability > 10% and 1%, without consideration of actual margin on the FS link, when the EIRP is increased to 21 dBm (from 14 dBm) for 1x, 2x, 3x, 6x, and 12x number of active VLP devices

Number of Active Devices

1x 2x 3x 6x 12x

Unavailability

14 dBm

21 dBm

14 dBm

21 dBm

14 dBm

21 dBm

14 dBm

21 dBm

14 dBm

21 dBm

10% 11 47 15 56 16 63 17 68 20 79 1% 15 65 21 76 37 91 25 110 27 128

In Step 2, the link budgets and actual fade margin for the links with increase in unavailability > 10% and 1% was calculated. With the device EIRP=21 dBm and the active number of devices 12x, the maximum number of FS that did not meet the 10% target was 79 (59 new links and 20 links analyzed in Section 5.1.6.2) assuming these links had only the minimum fade margin to achieve 99.999% reliability. To analyze actual FM for this many new FS links, the SNR required for 1024-QAM (highest modulation for conservativeness) of 31.2 dB was assumed.60 For all cases, the “Actual Fade Margin above FM” was found higher than the “Increase in FS Link Margin to meet the 10% target.”

The results show that even operating at the higher EIRP level = 21 dBm, the links meet the 10% increase in unavailability target as well as the sensitivity analysis down to 1% increase in unavailability for the baseline number of active VLP devices. These targets were achieved for 2x, 3x, 6x, and 12x sensitivity analysis as well.

5.1.7 FS Sharing Conclusions

To assess the interference impact from VLP devices to FS stations, five Monte-Carlo baseline simulations (corresponding to different bandwidth models), each with 100,000 iterations, were run for 97,888 FS over CONUS. Additional simulations were run to determine the sensitivity of the interference impact to the number of active VLP devices and EIRP.

Simulation results, even at 12x the number of active VLP devices, confirmed that in almost all cases, a single VLP device dominated the aggregate I/N levels at each FS, indicating that, although this analysis included aggregate effects, analysis of single-entry I/N levels is sufficient.

The simulation results indicated low average I/N > -6 dB and 0 dB occurrence probabilities of 0.00011% and 0.00002% respectively for the baseline simulations (1x number of active devices, using the channel distribution at a 14 dBm fixed EIRP). For 12x the number of active VLP devices transmitting at 21 dBm fixed EIRP for all channel sizes, the average I/N > -6 dB occurrence probability was still low (0.00718%) and the occurrence probabilities showed independence to the channel size.

60 For one of the new FS using 1024-QAM, the actual fade margin was not sufficient. However, after using the SNR required for its modulation (256-QAM), the “Actual Fade Margin above FM” was higher than the “Increase in the FS Link Margin to meet the 10% target”. Another FS was missing “modulation type” and “transmitter manufacturer” in the ULS. Even assuming QPSK with 6.2 dB C/Nreq, this link does not have sufficient margin to operate at 99.999% availability. As such, this link was removed from the analysis.

47

To accurately assess the impact of VLP interference on FS performance, the increase in FS unavailability was computed for 1,000 FS randomly chosen from the FS in Baseline Model simulation (1x using baseline channel distribution) that had at least one I/N > -6 dB occurrence. The increase in FS unavailability analysis showed that using ITU derived fading distributions and considering the operating parameters of the FS, the increase in unavailability did not exceed the 10% target and the 1% sensitivity threshold for all 1,000 FS, even at 12x the number of active VLPs with 21 dBm fixed EIRP.

Finally, analysis of an FS link in Ohio where hourly multipath fading distribution was available exemplifies that if hourly measured fade statistics are considered, the increase in unavailability will be reduced by an order of magnitude compared to the results derived using the study assumption that fading and interference are independent.

In conclusion, VLP devices operating at up to fixed 21 dBm EIRP over 20, 40, 80, or 160 MHz channel bandwidth do not cause harmful interference to an FS station.

5.2 Mobile Service (MS) Sharing

5.2.1 MS Usage Studied

National Association of Broadcasters (NAB)61 retained Alion Sciences to conduct an analysis of the extent of interference to ENG systems from unlicensed operations in U-NII-6 and U-NII-8, specifically non-common carrier mobile operations. The Alion Report62 considers three typical types of ENG deployment use cases:

1. Indoor Camera to Indoor Receiver

2. Outdoor Camera to News Truck

3. Outdoor News Truck to Central Receive Site using two representative locations:

a. Cowles Mountain, San Diego CA

b. DC Old Post Office, Washington DC

61 Dec. 5 NAB Letter. 62 Alion Report.

48

This report will focus on use case 3. The ENG Receivers were modeled using parameters in Table 5-21 and Table 5-22 below.

Table 5-21 - ENG receive sites (Table 7 of Alion Report63)

ENG receiver Latitude Longitude Antenna Height, AGL, m

Cowles Mtn. ENG central receive site

32° 48’ 49.30” N 117° 1’ 56.43” W 50

DC Old Post Office ENG central receive site

38° 53’ 38.86” N 77° 1’ 40.94” W 90

Table 5-22 - ENG central site receiver and antenna characteristics (Table 1 of Alion Report64)

Parameter Data Cable/Feeder Loss (dB) 1.0 Bandwidth (MHz) 20 Noise Figure (dB) 4.0 Receiver Thermal Noise Power (dBW)

-127.0

Antenna Nomenclature Vislink ProScan III Antenna Type Parabolic reflector Antenna Gain (dBi) 36.0 Antenna Pattern Azimuth and Elevation

patterns obtained from Alion (Figure 5-11 and 5-12 below)

63 Id., Table 7. 64 Id., Table 1.

49

Figure 5-11 - Central Receive Station’s ProScan Antenna Azimuth Pattern

Figure 5-12 - Central Receive Station’s ProScan Antenna Elevation Pattern

50

For each central receive (Rx) site, six links were studied (Table 5-23) using the following parameters:

● Two central frequencies for the highest VLP channel overlap: 6437.5 MHz in U-NII-6 and 6987.5 MHz in U-NII-8.

● The same three Rx station boresight azimuths as in the Alion Report (depicted in Figure 5-13 and Figure 5-14).

Table 5-23 - Six links simulated at each of the central receive sites

Link ID Center Frequency (MHz)

Rx Antenna Azimuth angle (deg) (Cowles Mtn. Rx site)

Rx Antenna Azimuth angle (deg) (DC Old Post Office Rx site)

1 6437.5 108 94 2 6437.5 194 315 3 6437.5 227 180 4 6987.5 108 94 5 6987.5 198 315 6 6987.5 227 180

51

Figure 5-13 - Cowles Mtn. ENG Central Rx Site and the three Azimuth Angles simulated (108°, 194°, and 227°)

Figure 5-14 - DC Old Post Office ENG Central Rx Site and the three Azimuth Angles simulated (94°, 180°, and 315°)

52

5.2.2 MS Simulation

Monte-Carlo Simulations with 100,000 iterations were performed for the mobile truck to central Rx station.

As described in Table 5-2, the Baseline Model evaluates interference to the ENG receive stations from VLP devices using the baseline channel distribution and an EIRP of 14 dBm. To more comprehensively assess the impact of channel size, four additional Monte-Carlo simulations were performed where all VLP devices were modeled using a single channel size (20, 40, 80, or 160 MHz) with an EIRP of 14 dBm. Metrics were computed for each simulation per ENG Rx station including: 1) occurrence probability for I/N > -6 and 0 dB; 2) occurrences due to a single VLP device; and 3) occurrences due to an aggregate of multiple VLP devices.

The 500,000 iterations, with 3 million65 MS interference assessments, of the Baseline Model, and the four fixed bandwidth simulations showed that occurrence probabilities from aggregation of multiple VLP devices were the same as single-entry occurrences. As such, the occurrence probabilities in this section correspond to single-entry as well as aggregate interference.

Table 5-24 shows I/N > -6 dB per-link occurrence probabilities (Eqn. 5-166) at the two central Rx sites, where:

● Cowles Mtn: For I/N > -6 dB, all six links had 0% occurrence probability

● DC Old Post Office: For I/N> -6 dB, five links had 0% occurrence probability; one link (ID 1) had 0.001% occurrence probability in the 40 MHz simulation and 0% occurrence probability in the other four simulations. For I/N > 0 dB, there were no occurrences.

Table 5-24 - I/N > -6 dB per-link occurrence probabilities at Cowles Mtn. and DC Old Post Office ENG central Rx sites from independent Monte-Carlo simulations (100,000 iterations per bandwidth model) for each of the six links

Central Rx Site

20 MHz 40 MHz 80 MHz 160 MHz Baseline Channel Distribution

Cowles Mtn. 0% 0% 0% 0% 0% DC Old Post Office

0% 0% for 5 links 0.001% for link ID 1

0% 0% 0%

The results show VLPs with maximum EIRP of 14 dBm do not impact the BAS/CARS ENG central receive stations, independent of the channel bandwidth used.

This result differs from the Alion study because Alion did not consider the VLP devices and Alion’s LOS calculation overestimated LOS paths probability.67

65 500,000 iterations x 6 links per ENG receiver site (= 3 frequencies x 2 azimuth directions). 66 Replace “FS” with “link” in Eqn. 5-1 (link as defined in Table 5-23). 67 Letter from Apple Inc., Broadcom Inc., Cisco Systems, Inc., Facebook, Inc., Google LLC, Hewlett Packard Enterprise, Intel Corporation, Microsoft Corporation, NXP Semiconductors, and Qualcomm Incorporated to Marlene H. Dortch, ET Docket No. 18-295 & GN Docket No. 17-183 (filed Feb 28, 2020).

53

5.2.3 Sensitivity Analysis

5.2.3.1 Number of Active Devices (2x, 3x, 6x, and 12x)

Similar to the methodology described in Section 5.1.6.1, the interference impact on a Mobile ENG Central Receiver is examined when the number of active VLPs is increased from 1x (baseline) to 2x, 3x, 6x, and 12x.

Cowles Mountain

Table 5-25 shows that when the number of active VLP devices is increased from the baseline to 12x, the average (over 6 links) I/N > -6 dB occurrence probabilities remain unchanged at 0%.

Table 5-25 - Cowles Mountain: Average Interference Statistics from all Independent Simulations for -6 dB I/N, for 1x, 2x, 3x, 6x, and 12x number of active VLPs

Number of Active VLPs

20 MHz 40 MHz 80 MHz 160 MHz Baseline Channel Distribution

1x (baseline) 0% 0% 0% 0% 0% 2x 0% 0% 0% 0% 0% 3x 0% 0% 0% 0% 0% 6x 0% 0% 0% 0% 0% 12x 0% 0% 0% 0% 0%

Figure 5-15 shows the distribution of aggregate I/N levels across all 6 links for the Baseline Model simulation (1x, baseline channel distribution) and 3x number of active devices. As indicated, all I/N levels (over 600,000 iterations) are less than -6 dB I/N.

Figure 5-15 - Probability (log-scale) of Aggregate I/N > I/N values on X-axis for the 600,000 MS interference assessments (6 links/iteration x 100,000 iterations) at Cowles Mountain - Zoomed In

54

DC Old Post Office

Table 5-26 shows that when the number of active VLP devices is increased from the baseline to 12x, the average I/N > -6 dB occurrence probabilities show no correlation with channel bandwidth. Occurrence probabilities slightly increased as the number of devices increased, however, the probabilities are so low they are not considered a long-term risk.

Table 5-26 - DC Old Post Office: Average Interference Statistics from all Independent Simulations for -6 dB I/N, for 1x, 2x, 3x, 6x, and 12x number of active VLPs

Number of Active VLPs

20 MHz 40 MHz 80 MHz 160 MHz Baseline Channel Distribution

1x (baseline) 0% 0.0002% 0% 0% 0% 2x 0.0003% 0.0007% 0.0003% 0% 0.0003% 3x 0.0005% 0.0010% 0.0005% 0.0002% 0.0003% 6x 0.0010% 0.0020% 0.0010% 0.0003% 0.0007% 12x 0.0020% 0.0040% 0.0020% 0.0007% 0.0013%

Figure 5-16 shows the distribution of aggregate I/N levels across all 6 links for the Baseline Model simulation (1x, baseline channel distribution) and 3x number of active devices. As indicated, all I/N levels are less than -6 dB I/N except in 2 iterations (out of 600,000 total iterations) of the 3x scenario.

Figure 5-16 - Probability (log-scale) of Aggregate I/N > I/N values on X-axis for the 600,000 MS interference assessments (6 links/iteration x 100,000 iterations) at DC Old Post office - Zoomed In

55

5.2.3.2 Higher EIRP Level

To study the impact of the higher EIRP VLP on a BAS ENG central receive site, the aggregate and single-entry I/N distributions from the five simulations with 14 dBm EIRP were increased by 7 dB to generate the statistics for 21 dBm EIRP levels.

Cowles Mountain

Table 5-27 shows the average I/N > -6 dB occurrence probability at 21 dBm EIRP for all six Cowles Mountain links for each of the five simulations. The sensitivity analysis for 21 dBm EIRP shows an extremely small number of occurrences.

Table 5-27 - Cowles Mountain: Average Interference Statistics from all Independent Simulations for -6 dB I/N, for 1x, 2x, 3x, 6x, and 12x number of active VLPs (14dBm and 21 dBm EIRP)

VLP EIRP (Number of Active Devices)

20 MHz 40 MHz 80 MHz 160 MHz Baseline Channel Distribution

14 dBm (1x to 12x)

0% 0% 0% 0% 0%

21 dBm (1x) 0.00017% 0.00017% 0% 0% 0% 21 dBm (2x) 0.00067% 0.00033% 0% 0% 0% 21 dBm (3x) 0.001% 0.0005% 0% 0% 0% 21 dBm (6x) 0.002% 0.001% 0% 0% 0%

21 dBm (12x) 0.004% 0.002% 0% 0% 0%

DC Old Post Office

Table 5-28 shows the average I/N > -6 dB occurrence probabilities at 21 dBm EIRP for all six DC Old Post Office links for each of the five simulations. Even though the probability of occurrence is higher than at Cowles Mountain, they are nonetheless extremely small.

Table 5-28 - DC Old Post Office: Average Interference Statistics from all Independent Simulations for -6 dB I/N, for 1x, 2x, 3x, 6x, and 12x number of active VLPs (14dBm and 21 dBm EIRP)

VLP EIRP 20 MHz 40 MHz 80 MHz 160 MHz Baseline Channel Distribution

14 dBm 1x: 0% 2x: 0.00033% 3x: 0.00050% 6x: 0.001% 12x: 0.002%

1x: 0.0002% 2x: 0.00067% 3x: 0.001% 6x: 0.002% 12x: 0.004%

1x: 0% 2x: 0.00033% 3x: 0.0005% 6x: 0.001% 12x: 0.002%

1x: 0% 2x: 0% 3x: 0.00017% 6x: 0.00033% 12x: 0.00067%

1x: 0% 2x: 0.00033% 3x: 0.00033% 6x: 0.00067% 12x: 0.00133%

21 dBm 1x: 0.00250% 2x: 0.00417% 3x: 0.00650% 6x: 0.013% 12x: 0.026%

1x: 0.00233% 2x: 0.00467% 3x: 0.007% 6x: 0.014% 12x: 0.028%

1x: 0.00117% 2x: 0.00333% 3x: 0.00483% 6x: 0.00967% 12x: 0.0193%

1x: 0.00117% 2x: 0.00217% 3x: 0.00333% 6x: 0.00667% 12x: 0.0133%

1x: 0.00167% 2x: 0.00333% 3x: 0.005% 6x: 0.010% 12x: 0.020%

56

5.2.4 MS Sharing Conclusions

To assess the interference impact from VLP devices to mobile truck-to-ENG Central receive station links, five Monte-Carlo simulations (corresponding to different bandwidth models), each with 100,000 iterations, were run for the representative ENG central receive stations in Cowles Mountain, San Diego and the theoretical DC Old Post Office receive site. Furthermore, sensitivity analyses on the number of active VLP devices and EIRP were performed.

One of the major differences between the Cowles Mountain and the DC Old Post Office ENG sites is the population in proximity of the receiver antenna. The Cowles Mountain site is on a higher elevation covering a larger operating radius, while the DC Old Post Office is on a lower elevation providing service over a more densely populated area.

In both instances, there was no risk of harmful interference from up to 12x the baseline number of active VLP devices using an EIRP up to 21 dBm, independent of the channel bandwidth used. This is expected for other locations throughout the CONUS, given the high elevation of these ENG Central Receive antennas and the very low power at which the VLP device transmits.

More generally, as mentioned in RKF’s 2018 Report, as is standard practice among MS operations, the MS transmitter operating parameters are optimized on a location-by-location basis (e.g., slightly closer, clearer path to MS receiver). We would expect the introduction of VLPs to require no change to these current practices by MS operators.

Attachment B:

Wireless Research Center of North Carolina Report On-Body Channel Model and Interference Estimation at 5.9 GHz to 7.1 GHz Band

1

On-Body channel model and interference

estimation at 5.9 GHz to 7.1 GHz band

Wireless Research Center of North Carolina1

Abstract—The on-body channel was characterized between mockup AR/VR glasses worn on the head and a mockup handset worn on various points on the human body in an anechoic chamber. Six human test subjects (three male and three female subjects), with various body mass index (BMI) were used in the experiment. Performance of the wireless node in the proposed 6 GHz band (5.9 GHz to 7.1 GHz) is presented. The potential threat to the existing 6 GHz network is also evaluated based on the measured results. We believe this is the first review of 6 GHz propagation on body-worn networks in spectrum shared by existing 6 GHz licensed users.

Index Terms—On-body wireless network, 6 GHz band, link loss model, interference model, wearable antenna.

I. INTRODUCTION

WITH the exponential growth of the personal devices market, the use of on-body wireless devices is becoming more popular. Applications such as Augmented Reality (AR) and Virtual Reality (VR) will require wireless devices to be located on-body and will need an antenna solution that can maintain a consistent signal. The antenna performance and propagation characteristics are dependent on the surrounding environment and the frequency of operation. For on-body applications, this becomes more critical because the body can block the signal between two devices. This paper generates a path loss model that can be used to understand the transmit (Tx) power needed to maintain a communication link for body area networks.

Because the FCC is considering enabling portable device use for applications such as AR/VR in the 5.925 GHz to 7.125 GHz band (6 GHz Band), we consider the potential threat and interference to the existing 6 GHz systems in addition to the loss for the body worn application. The link loss model generated for the 6 GHz band will help define the Tx power needed to maintain a constant communication link on-body. The Tx power requirements determine the potential interference to existing network devices using the 6 GHz band.

This paper evaluates the performance of two body worn devices on multiple subjects using passive testing over the 6 GHz band frequency range. The first device-under-test (DUT1) is a body worn device, similar to a handset phone. DUT1 was

1 The authors are affiliated with the Wireless Research Center of North Carolina, Wake Forest, NC 27587. (email: ko.takamizawa@wrc-nc.org; shruthi.soora@wrc-nc.org; jameelia@wrc-nc.org)

measured on-body in six different test locations on six human subjects with different body types. The second device-under-test (DUT2) is an eyeglass form factor worn on the head mimicking AR glasses. DUT2 maintained a fixed location for all tests.

Compared to other studies [8] that evaluate the path loss at lower frequencies, this paper evaluates the on-body link loss between DUT1 and DUT2 and the full sphere radiation pattern for both devices on-body and in free space in the 6 GHz band.

To differentiate from the path loss which is defined as the propagation loss, we define the link loss as the power loss between the transmit antenna port of DUT1 to the antenna port of DUT2. The link loss includes antenna gains of DUT1 and DUT2, power losses caused by the human body, propagation losses, and any mismatch losses such as antenna pattern and polarization mismatch of DUT1 and DUT2 antennas. Evaluating both the antenna pattern and the link loss in the identical environment provides insights on how the human body affects the RF links between two devices on the body and the signals that radiate away from the body, and how the signals between DUT1 and DUT2 may propagate away from the users and potentially interfere with the existing licensed systems.

This paper is organized as follows: Section II provides descriptions of the experimental setup for the link loss and spherical antenna radiation pattern measurements, including the mockup device descriptions, the test facility, and test subjects and test conditions. The DUT1 and DUT2 free space test results are presented in Section III. The measured results and data analysis of the different configurations are included in Section IV and V. Finally, the interference analysis is described in Section VI, and a short summary is presented in Section VII.

II. EXPERIMENTAL SETUP

A. Description of Devices-under-TestFor the link loss and antenna radiation pattern experiments, two mock up devices that represent typical performance of mobile handset and smart glasses were built. Figure 1 shows DUT1 which represents a body worn handset device. DUT1 utilizes a Pulse W3540 chip antenna mounted on a Pulse evaluation PCB. To represent an average size handset, the PCB ground plane was extended to 73mm x 162mm with the chip antenna located at the top corner of the device which is a typical location for WLAN antenna in a handset. The antenna PCB was enclosed inside a plastic external case for handsets with a clip on one side for body mounting. The clip can rotate 360 degrees which allows the orientation of the DUT1 to be changed on test subjects. Since the DUT1 is missing many components that exist in a typical handset such as display, battery and other electrical circuit components, the antenna performance of the DUT1 is expected to represent the best antenna performance case.

DUT2 represents a smart glasses device shown in Figure 2, which utilizes the Taoglas FXUWB10.01.0100C flex PCB

2

antenna with a SMA coaxial cable as a 6 GHz band antenna. A thin plastic board was attached to the right temple arm of the glasses near the right lens to attach the flex PCB to a stable platform. The antenna was mounted in the orientation such that the peak of the pattern points toward the feet of the person wearing the glasses where the DUT1 is in general expected to be located. The antenna used for the DUT2 is larger than what would be used in final commercial products. Thus, the DUT2 also represents the best antenna performance case, and thus the experimental results represent conservative estimates for the link loss and the interference to existing licensed systems.

The antenna cable of DUT2 was attached to the right temple arm. The glasses included a tightening strap to ensure consistent placement of glasses on the head during measurements.

Figure 1 - Body worn device (DUT1) inside phone case with belt clip

Figure 2 - Smart glasses model (DUT2) from front and right-side view

B. Measurement Facility and SetupAll link loss and full spherical antenna pattern measurements

on human subjects were conducted at Wireless Research Center of North Carolina (WRC) inside a 5m x 5m x 5m anechoic chamber. The chamber is equipped with the MVG SG-64 multi-probe near-field antenna measurement system [4] as shown in Figure 3. SG64 measures complex near-field patterns in amplitude and phase in two orthogonal polarizations. The far-field antenna pattern is mathematically calculated from the measured near-field using near-field to far-field transform [5]. WRC maintains ISO 17025 accreditation for antenna testing according to IEEE Std 149 [6] which includes radiation pattern, gain, input impedance and mutual impedance measurements. For the body worn testing, the WRC Human Walkable Platform (HWP) was utilized to place the human subject in the center of the SG64 probe arch for testing as shown in Figure 4. HWP includes RF absorber material under the platform to minimize the reflections and scattering from the metal mast supporting the platform underneath. During antenna pattern measurements, the height of the platform was adjusted with additional plastic blocks over the HWP to keep the DUT1 and DUT2 locations

within the quiet zone of SG64 in the test frequency. This study measured the antenna patterns at 6000 MHz, 6500 MHz and 7000 MHz.

Figure 3- MVG SG-64 multi-probe near-field antenna measurement system at WRC

Figure 4- WRC Human Walkable Platform

The link loss measurements were made using a 2 port Keysight E5071C vector network analyzer (VNA) with port 1 connected to DUT1 and port 2 connected to DUT2 as shown in Figure 5. The full 2-port S-parameter measurements were performed for both free space and body worn conditions over the frequency range of 5800 MHz to 7200 GHz.

The separation distance was determined and recorded for all conditions using a measuring tape around the shortest path over the subject’s body. The antenna feed points on DUT1 and DUT2 were used as the reference points for the distance measurements. The line-of-sight separation distances between DUT1 and DUT2 through the subject’s body were too difficult to measure accurately, so they were not measured.

Figure 5- VNA connected to DUT1 on port 1 and DUT2 on port 2

C. Human Test SubjectsFor this study, all measurements were performed on six

human subjects; three were male subjects and three were female subjects as shown in Figure 6. The test subjects covered the normal, overweight and obese body mass index (BMI) categories for each gender. The test subjects in the middle three BMI categories were specifically selected since they represent the majority of the population. Their height and weight were measured using a commercial off-the-shelf (COTS) weighing scale on the day of RF testing, and their BMI were calculated based on these measurements. The scale

3

was calibrated with a known 25-pound weight as a reference. The subjects remained anonymous and were assigned a subject ID designator (A, B, C, D, E or F) for data collection.

The subject height, weight and calculated BMI is shown in Table 1. The BMI was calculated using

𝐵𝐵𝐵𝐵𝐵𝐵 = �𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤ℎ𝑡𝑡ℎ𝑤𝑤𝑤𝑤𝑤𝑤ℎ𝑡𝑡2

� (703)

(1)

where weight is in pounds and height is in inches.

Table 1- BMI Chart for the Six Subjects.

Subject ID Designator

Subject Gender

Height (inches)

Weight (lbs)

BMI BMI Type

A Male 76 285.4 34.7 Obese B Male 70.5 200.6 28.4 Overweight C Male 73 187.4 24.7 Normal D Female 64 183.2 31.4 Obese E Female 69.5 205.6 29.9 Overweight F Female 66 137.2 22.1 Normal

Figure 6 - Six subjects used in testing

Figure 7- Distribution of all subjects on the BMI Chart

D. Test positions and device orientation For the on-body measurements, DUT1 and DUT2 were

placed on the 6 subjects for link loss and 3D antenna pattern measurements. DUT1, the body worn device was tested in 6

different positions to evaluate the link loss in line-of-sight (LOS) and non-line-of-sight (NLOS) conditions on all sides of the body. Table 2 and Figure 8 describe and summarize the test positions.

Position 1 represents the handheld device position where the subject holds the device at a comfortable reading position in front of their body with their fingers in a position that does not cover the antenna. Subjects were able to use whichever hand they normally would use to hold a phone. Position 1 creates LOS propagation path between DUT1 and DUT2.

Positions 2 and 3 place the DUT1 on the right and left side of the waist and was clipped to the subject’s clothing. Position 4 and 5 place the device in the right and left back pocket and was clipped to the subject’s clothing. Position 6 places the device on the outside center of a backpack to simulate the handset in a backpack. Figure 8 shows the DUT1 on-body orientation and antenna location.

DUT2 was mounted on each subject’s head and secured using the glasses tightening strap. DUT2 stayed in a fixed position on the test subjects head for all measurements. The antenna is located on the right side of the head on the glasses.

For testing, the subjects wore loose clothing with minimal metallic content representative of normal, everyday clothing. The belt clip and pant zipper were the only metallic features of their clothing and were not near the device under test. The backpack had a metal zipper for one of the compartments.

Table 2 – Position numbers and descriptions of six body positions of

DUT1 on human subject used in the experiment

DUT1 Position No DUT1 Position Description 1 Hand Held 2 Left Waist 3 Right Waist 4 Left Back Pocket 5 Right Back Pocket 6 Backpack

Figure 8 - Visual model of DUT1 and DUT2 positions and

orientation

E. Test conditions For evaluating the link loss over multiple conditions, the

testing was performed in both static and movement conditions. In the static condition the subject remains still in the desired

4

position. In the movement condition, the subject slightly shifted and moved during the link loss test. All antenna patterns were measured only in the static condition.

For the movement condition, Subject A, B, C, D, E and F moved slightly during the link loss measurement. 10 link loss measurements were taken over 2 minutes for each of the 6 positions. Because the majority of test positions are NLOS between DUT1 and DUT2, slight movement can cause fade in the radio link at varying frequencies. Making multiple measurements with slight movement ensures that we capture the statistical variations in the link loss.

Subject A was measured for additional link loss variations for the static condition to understand the polarization impact. For each of the 6 positions, the DUT1 device was rotated in 4 orientations as shown in Figure 9. The descriptions list device position numbers between 1 and 6, the device orientation (horizontal or vertical), and the antenna location (top or bottom). In addition, each test case was measured with the head facing center, head 45 degrees to the right and head 45 degrees to the left as shown in Figure 10. The head rotation was used to evaluate the LOS and NLOS impact on the position of the glasses on the subject’s head.

Together, the static and movement conditions allow for thorough analysis of the polarization impact and LOS/NLOS difference for the 6 different positions.

Figure 9 - DUT1 device in four test orientations. The blue rectangle represents DUT1 outline. The yellow square represents the location of 6 GHz antenna on DUT1. (a) Vertical orientation with antenna on top (V-T), (b) Vertical orientation with antenna on bottom (V-B), (c)

Horizontal orientation with antenna on top (H-T), (d) Horizontal orientation with antenna on bottom (H-B)

Figure 10 – Illustration of three head orientations. (a) face pointing to right of body at 45 degrees (head right), (b) face pointing forward (head forward), (c) face pointing to left of body at 45 degrees (head

left)

III. FREE SPACE TEST RESULTS OF DUT1 AND DUT2 DUT1 and DUT2 were measured in the SG64 chamber at

WRC in the free space condition using a Styrofoam mast to hold DUT1 and DUT2 similar to the setup shown in Figure 3. The polar plots in three primary axis and 3D antenna pattern plots are shown in Figure 11 and Figure 12 for DUT1 and DUT2, respectively. The peak gain over frequency is shown in Figure 13 and the free space return loss is shown in Figure 14.

Figure 11 - 2-D and 3-D antenna radiation pattern of DUT1 handset

Figure 12 - 2-D and 3-D antenna radiation pattern of DUT2

eyeglasses

Figure 13 - Peak Gain of DUT1 & DUT2 in free space

5

Figure 14 - Return Loss of DUT1 & DUT2 in free space

IV. DYNAMIC BODY POSITION LINK LOSS TEST RESULTS

The link loss between DUT1 and DUT2 was measured for all6 subjects with DUT1 in the 6 positions over frequency range between 5800 MHz to 7200 MHz at 1 MHz increments. The spherical antenna patterns were measured at position 4 (left back pocket) and position 6 (backpack) at 6000 MHz, 6500 MHz and 7000 MHz. The DUT2 antenna pattern was also measured on all 6 subjects for the same frequencies.

The measured link loss data was grouped into eight arbitrary 160 MHz channels to represent potential 802.11AX channel bandwidth in the 6GHz band, and average link loss within the 160 MHz channels was calculated. The center frequencies of the assumed channels are indicated in Table 3.

Table 3 - List of center frequencies of assumed eight 160 MHz channels in 6 GHz band for the link loss analysis. The corresponding wavelength is also shown.

Channel No. 1 2 3 4 5 6 7 8

Center Frequency

(MHz) 5940 6100 6260 6420 6580 6740 6900 7060

Wave-length (mm)

50.5 49.2 47.9 46.7 45.6 44.5 43.5 42.5

Figure 15 shows a scatter plot of link loss values for 160 MHz channels and separation distances for the corresponding measurements. The separation distances were normalized to wavelength for each channel center. The scatter plot includes link loss data for all cases of dynamic testing which includes six test subjects, six positions, and ten samples taken at each position. The markers for the scatter plots are grouped in two ways. First, the three shades of blue makers are used for the male subjects, and the three shades of red markers are used for the female subjects. The second grouping is based on the BMI categories. The “plus” markers are used for the obese subjects, the “x” markers are used for the overweight subjects, and the “dot” markers are used for the normal subjects.

Figure 15 - Dynamic Position Link loss for all 6 positions separated by subject

The visual inspection of the scatter plot indicates that there are two small groupings for subject A and subject D at a longer separation distance above 22 wavelengths and a shorter separation distance below 10 wavelengths, respectively. This is due to the differences in height of the test subjects. Subject A is tallest, while subject D is shortest in height among six subjects. Naturally, the DUT1 and DUT2 separation distance becomes longer for subject A, the tallest subject, and shorter for subject D, the shortest subject. For the separation distances between 10 to 20 wavelengths, it is difficult to see any particular grouping by subjects, gender or BMI categories.

Figure 16 shows the same scatter plot as Figure 15 but with markers that are grouped by the test positions. It is clear from the scatter plot that link loss depends on the test positions. Depending on the positions of DUT1 and DUT2, the communication link can be line-of-sight (LOS) or non-line-of-sight (NLOS) which impacts the magnitude of link loss. It is seen that position 1 (hand held) has the lowest link loss, but also has a shorter separation distance and is in LOS to DUT2. Position 4 (left back pocket) has the highest link loss, but is located diagonally through the body away from the DUT2 antenna. The signal has to travel around the body in a NLOS path over a larger separation distance.

One way to evaluate link loss data is to determine the mathematical approximation to the measured data by curve-fitting to a link loss model

𝐿𝐿𝐿𝐿(𝑑𝑑) = 10𝛼𝛼𝑙𝑙𝑙𝑙𝑤𝑤10(𝑑𝑑) + 𝛽𝛽 + 𝑁𝑁(0,𝜎𝜎) (2)

where d is the separation distance between DUT1 and DUT2 in wavelength, α is the power coefficient associated with link loss, β is the link loss constant offset, and N is zero mean gaussian with standard deviation σ. The black line in Figure 16 indicates a fitted curve for the mean link loss for all test positions. The red curve shows a fitted line for the right-side group data (right waist and right back pocket positions). The blue curve shows a fitted line for the left-side group data (left waist and left back pocket positions). The fitted models indicate that link loss to the right of test subjects are 5 to 10 dB less than the link loss to the

6

left of test subjects. This is consistent with the previously demonstrated right-side bias in the antenna pattern coverage of the glasses due to the right-side location of the DUT2 antenna. This also implies that average link loss between DUT1 and DUT2 can be reduced if DUT2 glasses has two antennas, one on each side of the temple arm.

Figure 16 - Dynamic Position Link loss for all 6 subjects separated

by position

V. STATIC BODY POSITION TEST RESULTS In order to further evaluate the link loss impacts, additional

testing was performed on subject A in a static position without dynamic movements during measurements. Because the initial dynamic testing showed low correlation between subjects, gender and BMI, the testing was performed on a single subject.

A. Gain pattern Measurements The pattern measurements evaluated all 6 positions in 4

polarization orientations as described in Section II.D. Measuring four orientations permitted the evaluation of the impact of polarization on link loss. When worn on the head, the DUT2 antenna pattern was more dominant in the theta polarization due to the vertical nature of the antenna installation on the glasses. In general, DUT1 was dominant in the theta polarization when placed in the vertical orientation and was dominant in the phi polarization when placed in the horizontal orientation. This is expected with the rectangular shape of the DUT2 ground plane and antenna location.

Figure 17 shows DUT1’s 3D antenna pattern, in all 4 orientations, represented in a contour plot format for Position 3. The horizontal axis represents azimuth angle from -180 degrees to +180 degrees where 0 degree is the direction of the face. -90 degree in azimuth is toward left of the test subject, +90 degree is toward right of the test subject, and +/-180 degrees is in the back of the test subject. The vertical axis of the contour plots represents the elevation angle. +90 degrees in elevation points toward top of the head, while -90 degrees in elevation points toward the feet. The filled contours are shown in the range of -10 dBi to +10 dBi in antenna gain for clarity. The white region in the contour plots indicates less than -10 dBi in gain.

When looking at the total gain patterns, the general shape is very similar between all 4 orientations.

Figure 17 - 3D Antenna Pattern Contour Plot of Position 3 in 4

DUT1 orientations

Figure 18 shows the contour plots of antenna patterns for DUT2 when the head is rotated in 3 different positions (head facing left, forward, and right). The overall pattern shape is similar between all 3 head orientations, but the pattern peak and nulls rotate in azimuth. Because the DUT2 antenna is located on the right side of the glasses, the antenna pattern is not centered in the facing forward orientation, but biased slightly to the right side. Because the pattern rotates with head orientation, the link loss can be improved or degraded depending on the position of DUT1 since the separation distance and LOS condition is changing.

Figure 18 - 3D Antenna Pattern Contour Plot of DUT2 in the 3 head

tilt positions

B. Static Link loss Measurements The link loss was evaluated for subject A for all 6 positions,

4 DUT1 orientations and 3 head orientation positions. Figure 19 shows that the link loss is still dependent on position, which is similar to what was seen in the dynamic link loss measurements.

The static link loss measurements evaluated DUT1 in 4 orientations where the device was placed in the horizontal (H) and vertical (V) orientation and the antenna was on the top (T) or bottom (B) side. Figure 20 shows that the vertical orientation of DUT1 has lower link loss compared to the horizontal orientation when looking at all 6 positions. This is expected due to the dominant theta polarization of DUT2.

7

Figure 19 - Static Position Link loss for all subject A separated by

position

To look at the link loss on the body between DUT1 and DUT2, a link loss (PL) curve was generated for the vertical vs. horizontal and antenna top vs. bottom cases to look at the link loss on the body between DUT1 and DUT2. For the horizontal case, the location of the antenna at the top or bottom does not have a significant delta as shown in the link loss (PL) curve fit. The vertical orientation with the antenna located at the bottom has a lower link loss because the peak antenna pattern points toward the glasses. This analysis shows that the antenna pattern direction and device orientation can impact the on-body link loss.

Figure 20 - Static Link loss for Subject A separated by DUT1

orientation

When evaluating the same static link loss measurements based on the head orientation (head pointed forward, pointed right 45 degrees and pointed left 45 degrees), the link loss is higher when the head points toward the left. The link loss is similar when the head is pointed forward or toward the right side. This is expected because the antenna is mounted on the right side of the glasses, as a result the 3D pattern is dominant on the right side of the body described by the measured data in the previous section.

These results show that the 3D pattern impacts the overall link loss, and optimizing the overall antenna pattern can

reduce the link loss. One optimization example is to add a diversity or switched antenna on the left side of the glasses, and then select the best DUT2 antenna based on the DUT1 location.

Figure 21 - Static Link loss for Subject A separated by head tilt

orientation

C. Combined link loss model for all NLOS data Figure 22 shows the measured data and calculated mean link

loss for all NLOS cases for the combined static and dynamic movement cases. The NLOS cases are defined as positions 2, 3, 4, 5 and 6 for all 6 subjects (static and dynamic). Figure 23 shows the distribution of the delta between the measured link loss and the mean for the static and dynamic movement cases. Table 4 shows the link loss model coefficients for the dynamic, static and combined movement cases. Figure 24 shows the CDF of the Link loss between DUT1 and DUT2. The static and dynamic coefficients are not identical, but similar where the link loss can be defined as looking at the combined data. The similarity between the static and dynamic movement cases show that the position dependent body blockage is the main contributor to the link loss versus body movement.

Figure 22 – Static and dynamic movement NLOS combined link loss

8

Figure 23 – Distribution of measured link loss from NLOS Combined

Link loss Model

Figure 24 – CDF of the Link loss between DUT1 and DUT2

Table 4 – Link loss Model Coefficients

Test Type Path Type α β σ Dynamic Test NLOS 3.59 30.19 8.43

Static Test NLOS 4.20 25.77 8.28 Combined Tests NLOS 4.27 23.99 8.45

VI. BODY LOSS TO FAR FIELD INTERFERENCE ANALYSIS The testing in this paper evaluates the link between a

wireless device on eyeglasses and a body worn device located at different locations including handheld, abdominal or upper leg portions of the body. The signal between the two devices will have energy that penetrates into the body, radiates around the body and travels on the surface of the body. These effects will impact the overall link loss between the two devices.

When body worn devices, such as a handset and smart glasses that operate in the 6 GHz band, are used by consumers, the communication signals that exist between two devices will predominantly radiate away from the users. If the radiated signals from the handset and glasses are high, they can potentially interfere with the communication systems, currently

using the 6 GHz band, which were developed and deployed without prior knowledge of the shared spectrum use by the unlicensed band user equipment.

Figure 25 - RF penetration into various human body materials versus

frequency

When a handset and glasses are used on individuals, the

radiation pattern from these devices change from the radiation pattern in the free space. The human body is made of multiple organs with frequency dependent electrical properties. At 6 GHz, the human tissues act in general as very lossy conductors [3] with shallow penetration depth as shown in Figure 25. The effects of lossy conductors proximate to DUT1 and DUT2 antennas are twofold: 1) the human body blocks or largely attenuates the signal from propagating in the direction blocked by human tissues, and reduces the radiation efficiency of the antenna; 2) the human body acts like a lossy ground plane which increases the directivity of the antenna pattern. In some cases, the peak gain of the antenna pattern of the handset and AR glasses on particular users can be higher than the peak gain of the devices in free space. In particular Table 5 shows the comparison of DUT1 handset at left back pocket (Position 4) where peak gains are higher for the subjects A, D, E and F, and lower for the subjects B and C compared to free space.

Table 5 – Measured peak gain of DUT1 on six test subjects at

position 4 (left back pocket) and free space

Test Position

Subject Free Space A B C D E F

Position 4 6.7 4.4 4.0 6.9 6.7 8.3 4.6

As demonstrated previously in Section V, the peak gain

direction of body worn devices change as the device position on the body and device orientation on the body changes. The interference from the body worn devices to the legacy system depends on the peak gain of the device on the user and bearing of the existing licensed system from the user. For example, a body worn device that has peak gain to the right side of a user has higher probability to interfere with the licensed system on the right bearing of the user, compared to the licensed system on the left bearing of the user. We need to treat the spherical antenna gain pattern as a component with statistical gain probability.

30 40 50 60 70 80 90 100 110 120

Link Loss (dB)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Cum

ulat

ive

Porb

abilit

y

CDF of Link Loss

All Subjects Cases

9

Figure 26 shows cumulative probability function (CDF) of the relative antenna gain values of DUT1 over the entire sphere for the ensemble of all test cases and all test subjects. Relative antenna gain is defined as the gain of DUT1 on-body as referenced to gain of DUT1 in free space. The increase in antenna gain due to body as a lossy ground is included in the analysis. The DUT1 antenna gain in free space is removed from analysis because it depends on the antenna design implementation on a future, actual device.

Figure 26 - Gain CDF of DUT1 for all 6 Subjects and 6 Positions

From the CDF curve of the gain on all cases and all subjects, mean and standard deviation can be calculated, where the mean gain of DUT1 is determined as -14.3 dBi, and standard deviation of gain is determined as 7.75 dBi.

Path loss from body worn device to existing licensed system in the 6 GHz band is also described by probability. There are many propagation models that were proposed which can be used for the analysis. ITU-R P.1411-10 recommends propagation models for short-range outdoor radio communication systems in 300 MHz to 100 GHz. We will use the ITU model for this analysis. However, any other model can be applied similarly. The recommended ITU site-general propagation model for the short-range outdoor communication is described by

𝑃𝑃𝐿𝐿(𝑑𝑑, 𝑓𝑓) = 10𝛼𝛼𝑙𝑙𝑙𝑙𝑤𝑤10(𝑑𝑑) + 𝛽𝛽 + 10𝛾𝛾𝑙𝑙𝑙𝑙𝑤𝑤10(𝑓𝑓)+ 𝑁𝑁(0,𝜎𝜎) 𝑑𝑑𝐵𝐵 (3)

where d: distance between the transmitting and receiving antennas (m)

f: operating frequency (GHz) α: coefficient associated with the increase of the basic

transmission loss with distance β: coefficient associated with the offset value of the basic

transmission loss γ: coefficient associated with the increase of the basic

transmission loss with frequency N(0,σ): zero mean Gaussian distribution with standard

deviation σ

Table 6 summarizes the recommended coefficients for the short-range LOS and NLOS cases.

Table 6 – Coefficients of ITU-1411 path loss models for LOS and NLOS cases

Path Type α β γ σ

Urban high-rise, Urban low-rise, suburban

LOS 2.12 29.2 2.11 5.06

Urban high-rise NLOS 4.00 10.2 2.36 7.60

Figure 27 shows mean path loss general propagation models for LOS urban and suburban in blue line and NLOS urban high rise in red line. The light blue shades and light red shades indicate the zero mean gaussian random variable N with a standard deviation σ in (3). The LOS model is applicable from 5 to 660 meters in separation distance, and the NLOS model is applicable to the separation distances of 30 to 715 meters.

Figure 27 - Propagation Model for Outdoors

The antenna radiation patterns of the existing licensed system can be also described statistically since the spatial relationships between the licensed system antenna and a person with 6 GHz AR/VR glasses and a handset are random. One of the existing radio systems approved for 6 GHz band are terrestrial point-to-point microwave links. These links typically utilize reflector antennas that are mounted on microwave towers. Figure 1 shows normalized radiation pattern envelopes of CommScope HX6-6W 1.8-meter reflector antenna [12], and CommScope SHP3-6W 0.9-meter reflector antennas [10]. HX6-6W and SHP3-6W has peak gain specifications of 39.1 [11] and 33.6 dBi [9]. From peak gain and normalized pattern envelopes, the average gain and standard deviation of gain over the sphere are calculated and shown in Table 1. Even though HP-6W has higher peak gain than SHP-6W, the average gain of HP-6W is lower than SHP-6W because the main beam width and side lobes near the main beam are narrower in angular extent. Also, 5 dB difference in back lobe envelope has significant impact in the average gain. It is noted that the calculated average gain values are pessimistic because the radiation pattern nulls which will reduce the average gain are not included in this model.

10

We also note that other statistical antenna models for the description of licensed system antenna patterns can be used in the interference analysis.

(a)

(b)

Figure 28 - Examples of normalized radiation pattern envelopes of 6

GHz band terrestrial microwave link antennas from CommScope datasheet. (a) CommScope HX6-6W 1.8 m reflector [12] , (b)

CommScope SHP3-6W 0.9m Reflector Antenna [10]

Table 7 - Calculated average and standard deviation of gain from CommScope HX-6W and SHP-6W reflector antennas

Model Diameter Peak Gain

Average Gain

Standard Deviation

HX-6W 1.8 m 39.1 -19.43 13.99 SHP-6W 0.9 m 33.6 -17.74 16.17 The probability of antenna gains of wearable devices, and

probability of propagation path loss, and probability of reflector antenna gain on the existing 6 GHz band links are statistically independent. Thus, the mean values of antenna gains and path loss adds linearly in dB, and standard deviations of antenna gains and path loss adds as square root of sum of squares. Table 7 shows an example of expected power spectral density (PSD) at the existing 6 GHz microwave link receiver when the body worn device is at 100 meters away from DUT1 in NLOS conditions, and DUT1 is transmitting a 160 MHz wide signal with 14 dBm transmitted channel power. PSD is used in the interference assessment because the channel bandwidth and sensitivity of the incumbent system is unknown.

Table 8 - PSD for NLOS at 100 meters with SHP-6W Antenna on

existing 6 GHz microwave link

Mean Std Unit Tx Power in 160 MHz channel

14.0 dBm

Bandwidth 160.0 MHz Bandwidth 82.0 dBHz Tx PSD -68.0 dBm/Hz DUT1 Antenna Gain -14.3 7.8 dBi NLOS Path Loss at 100 meters

109.4 7.6 dB

Commscope SHP3-6W Antenna Gain

-17.7 16.2 dBi

PSD at 100 meters -209.5 19.5 dBm/Hz Figure 29 shows the calculated PSD from DUT1 as a

function of separation distance between DUT1 and an incumbent system with SHP3-6W 0.9-meter reflector antenna. DUT1 is assumed to transmit at 14 dBm over 160 MHz bandwidth. The solid red and blue lines indicate mean PSD for NLOS and LOS conditions. The darker blue shade and darker red shade indicates one standard deviation, while lighter blue shade and lighter red shade indicates the two times the standard deviation. The black dash line is at -174 dBm/Hz indicating the thermal noise floor at room temperature.

Figure 29- Calculated PSD from DUT1 as a function of separation distance for LOS (blue) and NLOS (red) path for SHX-6W 0.9-meter reflector. Black dashed line indicates thermal noise floor at 300K.

For the NLOS case, 95 percent of signal PSD from DUT1 at 123 meters is -174 dBm/Hz or less. For the LOS case, the ITU path loss model is only available for the ranges up to 660 meters. By extrapolating the light blue shaded area, we expect that 95 percent of signal PSD from DUT1 is below -174 dBm/Hz around 1000 meters.

Figure 30 shows the calculated PSD from DUT1 as a function of separation distance between DUT1 and incumbent system with HP-6W 1.8-meter reflector antenna. For the NLOS case, 95 percent of signal PSD from DUT1 at 91 meters is below thermal noise floor of -174 dBm/Hz or less. For the LOS case, 95 percent of signal PSD from DUT1 is below -174

11

dBm/Hz at 664 meters.

Figure 30 - Calculated PSD from DUT1 as a function of separation distance for LOS (blue) and NLOS (red) path for HP-6W 1.8-meter reflector. Black dashed line indicates thermal noise floor at 300K

Note that this analysis assumes 100% duty cycle on the DUT1 for the incumbent system which is pessimistic. The probability of a wireless protocol utilizing 100% duty cycle is low as most current wireless standards utilize portions of the time frame. The potential for interference is less with a lower duty cycle since the device will not be transmitting at all times.

More detailed analysis is necessary to assess the impact of unlicensed spectrum use in 6 GHz band in the presence of incumbent systems and will depend on the duty cycle, RF characteristics and the antenna pattern of the incumbent system.

VII. SUMMARY

Two types of path loss measurements were conducted and presented in this paper– (a) Device-to-Device which are both on-body and (b) Device-to-Far-Field. An on-body communication path loss model between DUT1 (simulates a body worn mobile handset) and DUT2 (simulates smart glasses) for the 6 GHz band was presented. Propagation measurements were performed between DUT1 and DUT2 on six human test subjects, six on-body test device positions, static and dynamic body movement cases, four DUT1 device orientations, and three test subject head-pointing directions in an anechoic environment where multi-path scattering effects are negligible. 3D antenna radiation patterns were also measured in the same anechoic environment. Path loss models applicable to the NLOS path condition between DUT1 and DUT2 were derived.

Using a statistical description of body worn DUT1 antenna gain and statistical path loss models recommended by ITU for short-range communication, the statistical power spectrum density from DUT1’s transmit signal as a function of separation distance between DUT1 and the statistical description of antennas used in the existing licensed devices was derived.

The work presented in this paper is at the early stage of understanding the impact that unlicensed use of the 6 GHz

band may have to the licensed use of the existing legacy systems in the same frequency band. Additional information on the RF characteristics, antenna pattern and implementation of the incumbent system is needed. Testing and analysis is required to refine the path loss model between two on-body devices, and interference model to the legacy systems in 6 GHz band.

ACKNOWLEDGMENT We would like to thank Pulse Antenna for providing samples and evaluation boards for testing. We would also like to thank the test subjects for their time.

REFERENCES

[1] “Pulse W3540” [Online]. Available:https://productfinder.pulseeng.com/product/W3540

[2] “Taoglas FXUWB10.01.0100C” [Online]. Available: https://www.taoglas.com/product/fxuwb10-01-310ghz-ultra-wideband-uwb-flex-antenna-100mm-1-37mm-smam/

[3] Calculation of the Dielectric Properties of Body Tissues,” Institute for Applied Physics – Italian National Research Council, 2007. [Online]. Available: http://niremf.ifac.cnr.it-/tissprop/htmlclie/htmlclie.htm#atsftag

[4] “MVG SG-64” [Online] Available https://www.mvg-world.com/en/products/antenna-measurement/multi-probe-systems/sg-64

[5] J. E. Hansen, “Spherical Near-Field Antenna Measurements”[6] “IEEE Standard Test Procedures for Antennas”, IEEE Std 149-1979[7] “Propagation data and prediction methods for the planning of short-

range outdoor radiocommunication systems and radio local areanetworks in the frequency range 300 MHz to 100 GHz”, ITU-R P.1411-10, (08/2019)

[8] David B. Smith, Dino Miniutti, Tharaka A. Lamahewa, and Leif W.Hanlen, “Propagation Models for Body-Area Networks: A Survey and New Outlook”, IEEE Antennas and Propagation Magazine, Vol. 55, No.5, October 2013, pp. 97-117.

[9] “CommScope SHP3-6W” [Online]. Available:https://www.commscope.com/globalassets/digizuite/263660-p360-shp3-6w-6wh-a-external.pdf

[10] “CommScope SHP3-6W Radiation Pattern Envelope” [Online].Available: https://www.commscope.com/globalassets/digizuite/46913-7290a-9-24-19-pdf.pdf

[11] “CommScope HX6-6W” [Online]. Available: https://www.commscope.com/globalassets/digizuite/262695-p360-hx6-6w-6gr-external.pdf

[12] “CommScope HX-6W Radiation Pattern Envelope” [Online]. Available: https://www.commscope.com/globalassets/digizuite/46963-7376-pdf.pdf

Attachment C:

Broadcom Inc. and Facebook, Inc. VLP Testing: Effect of interference of 6 GHz VLP RLAN device on FS links

VLP TestingB roa d c om In c . & Fa c eb ook , In c .

M a y 20 20

E ffec t of Interferenc e from 6 G H z VLP R LAN D evic e on Fixed S ervic e P erform anc e

● RLAN interference testing on a 13 km FS link (30 MHz c h annel) from P orter R anc h /O at M ountain to N orth rid ge C aliforn ia

● M easurem ents○ A V LP R LA N d evic e w as m ounted 10 m ab ove ground level in

ord er to p osition in FS b oresigh t■ V LP E IR P w as c alib rated to p rod uc e I/N = +4d B at th e FS rec eiver

○ FS S IN R w as 30 d B for 256 Q A M inc lud ing 10 d B of fad e m argin■ C orresp on d s to five-n in es availab ility at th e lin k loc ation

○ Test w as exec uted c ontinuously for 24 h ours

● N o FS p erform anc e d egrad ation w as ob served○ In ord er to ob serve interferenc e, FS link m argin h ad to b e

red uc ed to insuffic ient levels

Interference Testing Summary

2

FS Link Information

● FCC Call Sign: WQ9XHF● Porter Ranch/Oat Mountain to Northridge

office (NRG) building● Alignment ensured maximum possible

RSL on Rx antennas at both sites

Oat Mt NRG

3

Link Setup Parameters

Parameter Oat-NRG

NRG-Oat

Tx power (dBm) 26 26

Tx antenna gain (dBi)

35.4 35.4

Rx antenna gain (dBi)

35.4 35.4

Frequency (MHz) 6200 6000

BW (MHz) 30 30

Modulation 256-QAM 256-QAM

RSL (dBm) -41 -42

Link parameters

NRGCeragon IP-20Tx: 6000 MHzRx: 6200 MHz

Oat MtCeragon IP-20Tx: 6200 MHzRx: 6000 MHz

FB corp network

FB corp network

13 km

BRCM VLP device

190m

NRG RooftopLAT/LONG: 34.224574° -118.499133°Elevation: 8’ above roof level (roof height: 10m)TX Power: 26 dBm, 256-QAM, BW 30 MHzLink margin: 10 dB

Oat MtLAT/LONG: 34.324044° -118.581545°Elevation : 25’ AGLTX Power: 26 dBm, 256-QAM, BW 30 MHzLink margin: 10 dB

ACM/ATPC not enabled on either side of link

4

RLAN Configuration

● A VLP device was placed 190m from the FS receiver, along boresight

● Mounted on a van ~10m height● The device was battery powered● Programmed to transmit randomly

generated traffic continuously at 95% duty cycle

● The VLP device power was calibrated to obtain I/N at FS Rx = +4dB (Tx = 22.8 dBm)

● Test was run over 24 hours

10m elevation is not typical for RLAN.

We are demonstrating an improbable scenario, where RLAN antenna is only ~8m below the FS receive antenna

5

Expected FS Receiver Interference Performance

• The MSE of the FS receiver was measured in a shielded lab

• In the presence of a strong interferer, MSE is expected to degrade as shown here

Lab characterized MSE

6

Field Results

● No degradation in MSE was observed during 24 hours of testing

7