Monday, October 23, 2017

Google Loon - Floating Nest of Radio Technologies

Google is deploying a High Altitude Platform System (HAPS) in the form of balloons that float around 20 km in altitude above a service area.  Representative link budgets for both the LTE service and the air-ground unlicensed feeder link show that one balloon can deliver about 10 Mbps uplink and 20-25 Mbps downlink, on average, to LTE subscribers in an 80 km coverage area.

While there has been much written about Loon, patents filed, and lots of marketing (including videos), precious little is really known about the radio package. I have taken as a personal task to consider the technology and the system; to understand better the challenges and the possibilities for curiosity sake.  I do not profess to be proficient in Loon technology, as that is entirely proprietary.  I fully expect that Loon have resolved many issues and performs beyond what my cursory analysis projects.  The public information I have access to, and my understanding and methods, may be seriously flawed; yet I have faithfully proceeded with my best judgement.

Each Loon balloon includes a radio package comprising a variety of services.  Google is operating without a meaningful dedicated spectral capacity, and thus has reaped the low-hanging fruit of unlicensed spectrum coupled with borrowed spectrum.
1.  Unlicensed, point-point U-NII 3 high gain feeder link (S1)
2. Unknown tracking, telemetry and control link (Iridium?)
3. Unlicensed, free-space-optical/RF inter-balloon mesh network (X2)
4. Licensed, point-to-multipoint UHF LTE service link (eNodeB)
5. GNSS (GPS) navigation receiver

In the context of LTE, Loon is a floating eNodeB (eNB).
LTE links between eNB are called X2.
LTE Links to the Evolved Packet Core (EPC) are called S1.

Loon has received authorization to operate four LTE 5 MHz channels that operate on two LTE bands.  Each band operates in FDD (frequency division duplexing), where two 5 MHz channels are paired, one for uplink (User Equipment, UE) to Base Station (BS or eNB) and the other for downlink (BS to UE).


Channels (MHz)
Band 28           722-728     (UE transmit)       and      782-788 (BS transmit)
Band 8             896-901.3  (UE transmit)       and      937-942 (BS transmit)

Loon has authority to operate both a 3 dBi and an 8 dBi antenna, each with 5 Watts, on each band.

It is not apparent what utility the transmit authority on the two channels assigned for UE transmissions (722, 896). The service is LTE. The bands operate only FDD (as per the user equipment, notably Apple iPhone). Loon transmissions would likely overwhelm the integrated receiver, further complicating their utility.  No further consideration is made for the Loon transmitting on them.

Loon relies on a sophisticated flight planning tool that is expected to align the balloons optimally for covering the zone.  Here is their position for over five days (Wed-Mon Oct 25-30).  The balloons proximity to Puerto Rico appears to be excessive for most of the time.

flightradar24.com

The effectiveness of altitude change to speed/direction control is evident with the arrival of BALL209 from the mainland.  A slight climb and the balloon slowed to loiter nearby Puerto Rico.

flightradar24.com

Loon coverage has been quoted to 5000 km2 or 80 km in diameter. In the representation shown below, the largest coverage zone shown is about 80 km in diameter, equating to the coverage of a 3 dBi antenna with a maximum path length of 44 km.

The other two coverage zones are different in size, with the smaller about 55 km in diameter.

The maximum path length from a Loon at 20 km altitude with a 40 km coverage zone is about
28 km. Link Budgets for the 8 dBi antenna were run out to 28 km; the 3 dBi to 44 km.

8 dBi Antenna
2 x 5 MHz
Uplink (Mbps)
Downlink (Mbps)
Band 28
3.1
7.2
Band 8
2.4
6.1

8 dBi Antenna
2 x 3 MHz
Uplink (Mbps)
Downlink (Mbps)
Band 28
2.2
4.7
Band 8
1.8
4.3

3 dBi Antenna
2 x 5 MHz
Uplink (Mbps)
Downlink (Mbps)
Band 28
1.1
4.3
Band 8
0.9
3.6
Band 28 maximum path length 38 km (uplink limited)
Band 8 maximum path length 30 km (uplink limited)

3 dBi Antenna
2 x 1.4 MHz
Uplink (Mbps)
Downlink (Mbps)
Band 28
0.7
1.6
Band 8
0.5
1.3

Puerto Rico is 170 km x 60 km.

Two, Three, or even six balloons could provide coverage to an island with over 3 million persons. For the perfect case of two balloons, there are three scenarios considered with different arrangement of antenna and bandwidth.

Running Band 28 on the 3 dBi antenna and Band 8 on one 8 dBi antenna would yield about 3.5 Mbps uplink and 10.4 Mbps downlink from one Loon. This would offer overlapping service, with the 3 dBi antenna directed to edge users (may not reach 80 km diameter coverage). 

Two balloons would yield about 7/21 Mbps (up/down) for the coverage area if optimally positioned. With more 3 dBi overlay than shown, there would not be gaps in the coverage area.


Two 2x5 MHz 8 dBi
Two 2x5 MHz 3 dBi

The expected 8 dBi data rates are about 11 Mbps uplink and 26 Mbps downlink from one Loon if using four 8 dBi antennas.  

Two balloons would yield about 22/52 Mbps (up/down) for the coverage area if optimally positioned. Without the 3 dBi overlay, there would be gaps in the coverage area.


Splitting each 2x5 MHz band into 2x3 MHz and 2x1.4 MHz offers the ability to overlay the 3 dBi antenna over the four 8 dBi antennas, filling any gap. This would assign the 3 MHz channels to the 8 dBi antenna and the 1.4 MHz channels to the 3 dBi antenna. The capacity is 9/20 Mbps (up/down).

Two balloons would yield about 18/40 Mbps for the coverage area if optimally positioned.


Two balloons can be seen in service on Tuesday Oct 24.

Two balloons can be seen in good position on Tuesday Oct 24.
flightradar24.com
Sizing the S1 and the inter-balloon link X2 can be based on peak connectivity, where most usage is in favorable locations.  One balloon peak demand (2x average) could be about 20/50 Mbps (uplink/downlink).

The X2 link serving a single slave eNB would need 20/50 Mbps data rate.
The X2 link serving a six slave eNBs would need 120/300 Mbps data rate.

Google professes a free-space optical link for the X2 link.  That would serve only one balloon-balloon link (or 20/50 Mbps).  A second optical link is needed to connect a daisy-chain network.  RF solutions may provide a lower cost solution for the inter-balloon X2 link.

One S1 air-ground link serving a host eNB and one slave eNBs would be sized about 40/100 Mbps. One S1 link serving a host eNB and six slave eNBs would be sized about 140/350 Mbps.

A 5.8 GHz Wi-Fi point-to-point bridge, using one 40 MHz channels, can support about 113 Mbps from the edge of an 80 km coverage area, about 151 Mbps for the beam center. 

SES O3b is providing the backhaul connectivity from the Loon ground station (cellular switching and Internet access).

The backup tracking/telemetry/control link is not documented.  I would use Iridium as the backup radio link.  The GPS receiver is not considered further.

LTE – Loon provides floating eNB



Cell Phone support for Band 8 and Band 28 (FDD-LTE)

Cell phones have embraced emerging LTE bands for many years now.  Both bands are provisioned as of the Apple iPhone 6 (per Apple published specs).

FDD-LTE Band 8 support began with iPhone 5S

FDD-LTE Band 28 support began with iPhone 6

FCC Experimental Special Temporary Authorization

Recently, the FCC issued an experimental license to Loon Inc. (their first) which authorized LTE services over Puerto Rico.

Authorized Transmissions

The most current authorization shows the following 5 MHz LTE carrier assignments.


To free the bands in question, operators have voluntarily released their services for the duration.  The contributions are piece-meal, and were not organized around the LTE service in question.  The point is to shake the 5 Mhz spectrum for each channel to allow Loon to occupy them without interference.


The FCC lists the rules for each band in question.  The bottom line is the FCC granted two 5 MHz channels out of the 700 MHz commercial pool that appear to be vacant for Band 28, and has adopted band 8 FDD-LTE rules (to which precedent or legacy is unclear).






  

LTE Antenna

Based on the FCC STA, there appear to be two transmit antenna (Classes): 8 dBi and 3 dBi

Polarization and Orientation

Cellular communications are normally vertically polarized.  The user terminal antenna may lose about 3 dB of the signal if cross-polarized when communicating with Loon.  It does not appear worthwhile to try to orient the antenna such that vertical polarization is presented to each subscriber, and take the small hit for cross-polarization. MIMO using cross-polarization is an option if the client can support, while noting that overhead emissions to end-on antennas may minimize cross-polarization discrimination.

The base station and user normally communicate along a horizontal plane following the surface of the Earth. User antennas are purposely omni-directional to allow their use without regard to orientation.  For the case of Loon, the user antenna to Loon orientation is vertical (when overhead), to about 30 degrees from vertical at edge of coverage.

It is likely that the user terminal antenna will lose “one bar”, or up to 10 dB of signal, due to mis-orientation towards Loon than towards a terrestrial base station.  It is modeled as -3 dBi. 

Aircell Showed Immunity to Overhead Cell Phones

User terminal antennas are optimized for radiating towards the horizon, not towards the sky.  Aircell famously took advantage of these features when launching their overlay service, reusing the cellular frequencies through immunity.  Note that Aircell has the User Equipment in flight, whereas Loon has the Base Station in flight.  Furthermore, Loon is operating in a cooperative manner.  This was just an early example of re-use.



In the Aircell situation, Aircell had to ensure no interference to co-frequency base stations or their subscribers when operating in close proximity.

For Loon, frequencies are coordinated such that the Loon carriers are presented exclusively in the regions they are overflying. 

8 dBi Antenna

While the STA includes reference to other antennas, I was left with searching on my own for a representative aperture.  The 8 dBi patch resembles the Loon payload.

A patch antenna provides a useful 8 dBi coverage pattern, looking downwards.



3 dBi Antenna

I have not found what I had hoped in a  3 dBi antenna - a circular coverage zone roughly 140 degrees in extent.  A perfect 3 dBi antenna would extend to 180 degrees.  The analysis assumes that coverage would be a perfect hemisphere, in this case the lower one.

A monopole “stub” antenna is not ideal.  As shown here, about half the energy is directed upwards, and the gain falls off by 3 dB down 30 degrees from the horizon, leaving a big gap directly underneath. In many ways, this antenna covers everything but the desired zone.


A compact, 3 dBi Yagi provides a much more useful coverage pattern that can be directed looking downwards. It's 60 x 35 degree coverage is lacking from the desired 70 x 70 degree zone.



http://www.l-com.com/multimedia/datasheets/DS_HG903YE.PDF

No patch antenna jumped out at me in searching.  In conclusion, while no example was forth-coming, I am confident that a method to cover the 70 degree cone with a 3 dBi peak antenna is feasible.

MISO and SIMO

MIMO (multiple antennas – spatial streaming). Free-space MIMO may have benefits if the antennas are sufficiently separated. The extended (20 km or greater) path length creates some challenges for diversity. The modeling will assume it is beneficial without regard to the user equipment capability. In this context: assume two antennas on the Loon balloon and a single antenna user equipment. The benefit from cross-polarization is not certain.

LTE Link Budgets

Hat tip for a nice link budget starting point.  I modified to account for free space path loss, and to reflect specific conditions.


The following link budgets are a crude estimate of performance.  The User Equipment gain is set to -3 dBi to reflect the upwards radiation pattern.  A 3 dB diversity gain is applied to both uplink and downlink (assuming two antennas).  No cable losses are applied. Only free-space loss is accounted for. 8 dB fade margin and 5 dB penetration loss.

Iridium famously advertised their users outside, never in a building.  Loon may suffer a similar issue with mobile handsets brought inside.


Link budget for 8 dBi antenna reveals the limitations in having the Loon balloon hover a minimum of 20 km above the user equipment. 

The 8 dBi antenna “70 degree coverage” maximum path length is 24 km; minimum path length is 20 km when directly overhead. Coverage area diameter would be about 30 km (-3 dB edge).

The 3 dBi antenna “120 degree coverage” maximum path length is 40 km. Coverage area diameter would be about 70 km (-3 dB edge).


A 40 km coverage diameter would require a 45 degree beam and a maximum LOS path of 28 km. This is a bit outside of the 8 dBi beamwidth.  Ignoring antenna gain rolloff, model from 20 km to 28 km path length (8 dBi).

An 80 km coverage diameter would require about 63 degree beam and a maximum LOS path of 44 km.  This is within the 3 dBi beamwidth.  Model from 20 km to 44 km path length (3 dBi).

8 dBi link budget

Downlink

37 dBm conducted power
8 dBi transmit gain
8 dB Fade Margin
5 dB Penetration loss
Free-space path loss
-3 dBi receive antenna gain
3 dB diversity gain 

Uplink

23 dBm conducted power
-3 dBi transmit gain
8 dB Fade Margin
5 dB Penetration loss
Free-space path loss
8 dBi receive antenna gain
3 dB diversity gain

2 x 5 MHz FDD (8 dBi)


8 dBi Antenna
2x5 MHz
Uplink (Mbps)
Downlink (Mbps)
Band 28
3.1
7.2
Band 8
2.4
6.1

A 2600 MHz service would be unable to establish an uplink.

2 x 3 MHz FDD (8 dBi)


8 dBi Antenna
2x3 MHz
Uplink (Mbps)
Downlink (Mbps)
Band 28
2.2
4.7
Band 8
1.8
4.3

The smaller carrier punches through better, notably the 2600 band uplink only suffers on the edge of coverage.

3 dBi link budget

Downlink

37 dBm conducted power
3 dBi transmit gain
8 dB Fade Margin
5 dB Penetration loss
Free-space path loss
-3 dBi receive antenna gain
3 dB diversity gain

Uplink

23 dBm conducted power
-3 dBi transmit gain
8 dB Fade Margin
5 dB Penetration loss
Free-space path loss
3 dBi receive antenna gain
3 dB diversity gain

2 x 5 MHz FDD (3 dBi)

3 dBi Antenna
2 x 5 MHz
Uplink (Mbps)
Downlink (Mbps)
Band 28
1.1
4.3
Band 8
0.9
3.6
Band 28 maximum path length 38 km (uplink limited)
Band 8 maximum path length 30 km (uplink limited)

Of note, 1700 MHz service and 2600 MHz service would be unable to establish an uplink.

2 x 1.4 MHz FDD (3 dBi)

3 dBi Antenna
2 x 1.4 MHz
Uplink (Mbps)
Downlink (Mbps)
Band 28
0.7
1.6
Band 8
0.5
1.3

Both Band 8 and Band 28 operate throughout the coverage area. 1700 MHz service uplink suffers on band edge.  2600 MHz service would be unable to establish an uplink.

Unlicensed 5.8 GHz Point-Point

Google video has revealed what appears to be a Ubiquiti network RocketDish RD-5G30, providing about 30 dBi gain.





FCC rules are relaxed for fixed, point-to-point connections, allowing for unlimited EIRP based on a maximum 1 Watt conducted power.

Characterizing the Loon as a fixed station is not correct.  The Loon ground station could be fixed, but the beam will be moving to track the balloon. One could argue that mis-steering towards a Loon has no impact, as the energy is directed far above any terrestrial applications (hence, no interference if managed within some constraints that prevent low-elevation steering).  Given that there are pictures of Loon equipment that is known to have high gain, I am assuming the FCC has permitted the use of a high gain antenna on the ground within the boundaries of a fixed, point-to-point connection.


Running a simple link budget at 5850 MHz for path lengths from 20  km (beam center), to 28 km (edge of a 40 km coverage area), to 44 km (edge of an 80 km coverage area), assuming free space path loss and no other factors, results in a received signal between -68 to -75 dBm.


The Ubiquiti airLink Outdoor Wireless Link Calculator can be configured for the same link budget.

Airlink Link Budget Tool

40 MHz channel -  5 GHz band

EIRP reduced from 60 dBm to account for 4 dB extra receive gain at Loon.

Antenna gain of 10 dB (should be 6) accommodated by reducing EIRP by 4 dB.



20 km (minimum) Path:   40 MHz = 151 Mbps


24 km Path:   40 MHz = 151 Mbps



28 km Path:  40 MHz = 151 Mbps



44 km Path (maximum)    40 MHz = 113 Mbps

Based on one Loon balloon with a ground link relaying to six neighboring balloons would require 140/350 Mbps (uplink/downlink), or about 500 Mbps.

airLink spectral efficiency is about 2.8 - 3.8, or 130-180 Mhz needed.

One Loon balloon needs 20/50 Mbps (peak service) or about 70 Mbps.  This is easily accommodated from one 40 MHz channel.

While it appears there is a high gain dish that might be useful for 5.8 GHz, I would expect that the FCC would have to issue some form of assessment to show how the mobile high gain antenna would qualify for fixed point-to-point service.  It is reasonable to note that the 20 km altitude offers a bubble, where there may be no other users to interfere with; that the high gain antenna normalizes the airborne transmissions; by the time they reach the surface of the Earth their power spectral density has been reduced significantly.


Stay tuned!


Peter Lemme

peter @ satcom.guru
Follow me on twitter: @Satcom_Guru
Copyright 2017 satcom.guru All Rights Reserved


Peter Lemme has been a leader in avionics engineering for 35 years. He offers independent consulting services largely focused on avionics and L, Ku, and Ka band satellite communications to aircraft. Peter chairs the SAE-ITC AEEC Ku/Ka-band satcom subcommittee, developing ARINC 791 and 792 characteristics and contributes to the Network Infrastructure and Interfaces (NIS) subcommittee developing Project Paper 848, standard for Secure Broadband IP Air/Ground Interface.


Peter was Boeing avionics supervisor for 767 and 747-400 data link recording, data link reporting, and satellite communications. He was an FAA designated engineering representative (DER) for ACARS, satellite communications, DFDAU, DFDR, ACMS and printers. Peter was lead engineer for Thrust Management System (757, 767, 747-400), also supervisor for satellite communications for 777, and was manager of terminal-area projects (GLS, MLS, enhanced vision).

An instrument-rated private pilot, single engine land and sea, Peter has enjoyed perspectives from both operating and designing airplanes. Hundreds of hours of flight test analysis and thousands of hours in simulators have given him an appreciation for the many aspects that drive aviation; whether tandem complexity, policy, human, or technical; and the difficulties and challenges to achieving success.

3 comments:

  1. Thank you for this thoughtful and expansive post, it was most educational. You taught me to look for flightradar24 for balloon positions that I have been following since. I note that most of the time, the balloons are rarely able to provide much service. (See my nightly postings at www.twitter.com/sawilkus) I only take evening snapshots, do you think this is missing something or do you agree that Loons are not working well in the Caribbean?

    ReplyDelete
  2. I would love to see a response to Steve's question. From what I see in viewing your snapshots, Steve; I'd say this is not a reliable solution. Drone technology to keep lighter than air craft in a fixed position makes a lot more sense to me.

    ReplyDelete