Starlink direct to cell (was SpaceX & T-mobile event 25 Aug 2022)

[Danderman]

I mentioned Starlink support for handhelds in Alaska in an earlier post. After some minutes on the Google, I found this:

https://www.fiercewireless.com/operators/t-mobile-touts-5g-roaming-deal-gci-alaska

T-Mobile uses a rental network that is deployed in UHF. Starlink will use T-Mobile’s 1.9GHz in the rest of the USA. So, unless Starlink can change frequencies when over Alaska, it looks like at least some of Alaska will not be covered by Starlink for handhelds.

For Canada, T-Mobile has multiple roaming agreements, some roaming partners cover 1.9GHz. So Canada is a definite maybe for Starlink handheld service. There are 3 big carriers in Canada. These carriers may seek a direct agreement with Starlink for their subscribers, except that 1 of the 3 does not support 1.9GHz.

[Danderman]

Elon seems to be rather dismissive of the v2 mini concept, not sure he would want to spend the resources to add cell tower antenna to mini. I mean if mini is just a temporary stopgap, any additional work on it would be wasted once it's phased out in a few (hopefully short) years.

I mentioned this in an earlier post.

What makes this more critical is that Shell #1 will have to support handhelds, and is also the heart of the standard Starlink internet service. So if Starship is delayed, and mini-sats go up on F9, handheld service will be delayed. And the mini-sats will have to be replaced quickly.

[MP99]

T-Mobile’s 1.9MHz

Sorry, but it's 1,900 MHz / 1.9 GHz.

[Danderman]

T-Mobile’s 1.9MHz

Sorry, but it's 1,900 MHz / 1.9 GHz.

You are correct.

I should drink more coffee before I post.

[TheRadicalModerate]

The first gen constellation is ~4k sats, not ~12k.

The first license is 4K but the first generation is the first two licenses, which do add up to 12K.

[macpacheco]

I am hearing that Starlink handheld service is only available at 1.9 - 2.0 MHz, which is  T-Mobile spectrum in the US. I am not sure why Starlink chose this frequency (apart from the T-Mobile partnership), since AFAIK, they aren’t licensed for this frequency elsewhere.

It would be an odd design choice if that was a hardware limitation.  Are you sure that's not just the frequencies they're choosing to use in the US?

First item: isn't the frequencies 1.9 - 2.0 GHz? Which gives T-Mobile 100Mz of bandwidth. Else Mhz would give T-Mobile 100KHz of bandwidth.

In the initial 12,400 sat constellation to be fully operational on or before Nov 2027 (the FCC licensing date). Would give T-Mobile excellent coverage. For the more advanced constellation of 30,000 sats which would be fully operational on or before 2030. Could have total of 3 cell phone providers each have 10,000 dedicated sats each. which would result in at least 3 cell phone providers from sat just by Starlink plus the other sat constellations that also implement cell services. Such that by 2030 you could have more than a dozen cell phone providers operating from orbit.

Additionally what SpaceX and Starlink design may be going for is to be like a cell phone tower operator that receives a fixed fee for each operational tower. As more "towers" (sats) are added the more revenue that SpaceX would get and the more advantage that T-Mobile would have in capabilities to have a solid connection space that covers all of US and territories. Which should bring in to T-Mobile more subscribers since they can cover areas and customers that no one else can.

Back in PCS (2G) times 100MHz would not be allocated to a single carrier. It would be allocated in 20 or 25 MHz for each carrier.
And there's more.

A cell tower typically uses 6 sector antennas. Each sector gets a separate frequency (macro channel).
This is done to avoid interference between the sectors.
I think each GSM macro channel is just 5MHz. This is divided into one control channel and several bearer channels.
SMS send/receive can only use the control channel, its also used for call setup, to register a phone, ...
So supporting voice is pretty much a given, otherwise you're wasting most of the channels.
Unless Starlink can manage multiple macro channels, it can't use an entire 10 or 20 MHz frequency range.
I think Starlink would get just 5MHz to use. That's why its just 2-4Mbps. Even using 2G (GSM) 100MHz would be a much closer to 100Mbps than 4Mbps.
I don't know how and if Starlink phased array technology could come into play here. Starlink can transmit/receive simultaneously on the same frequency into different directions (each called a separate beam, as long as the beams have a lot of angular distance). This might simply not apply to cell phone service.

[gongora]

The first gen constellation is ~4k sats, not ~12k.

The first license is 4K but the first generation is the first two licenses, which do add up to 12K.

There hasn't been any recent indication that the V-band sats will actually happen.  The gen 1 constellation is the 4k Ku/Ka-band sats.

[TheRadicalModerate]

The first gen constellation is ~4k sats, not ~12k.

The first license is 4K but the first generation is the first two licenses, which do add up to 12K.

There hasn't been any recent indication that the V-band sats will actually happen.  The gen 1 constellation is the 4k Ku/Ka-band sats.

Is there any reason why the v2 birds can't have a version that supports V-band?  That license is approved and the clock is ticking on it.  The Gen2 license isn't.  (Again, "v2" vs. "Gen2" is horribly confusing.  They might mean the same thing, or they might not.)

[Danderman]

I am hearing that Starlink handheld service is only available at 1.9 - 2.0 MHz, which is  T-Mobile spectrum in the US. I am not sure why Starlink chose this frequency (apart from the T-Mobile partnership), since AFAIK, they aren’t licensed for this frequency elsewhere.

It would be an odd design choice if that was a hardware limitation.  Are you sure that's not just the frequencies they're choosing to use in the US?

First item: isn't the frequencies 1.9 - 2.0 GHz? Which gives T-Mobile 100Mz of bandwidth. Else Mhz would give T-Mobile 100KHz of bandwidth.

In the initial 12,400 sat constellation to be fully operational on or before Nov 2027 (the FCC licensing date). Would give T-Mobile excellent coverage. For the more advanced constellation of 30,000 sats which would be fully operational on or before 2030. Could have total of 3 cell phone providers each have 10,000 dedicated sats each. which would result in at least 3 cell phone providers from sat just by Starlink plus the other sat constellations that also implement cell services. Such that by 2030 you could have more than a dozen cell phone providers operating from orbit.

Additionally what SpaceX and Starlink design may be going for is to be like a cell phone tower operator that receives a fixed fee for each operational tower. As more "towers" (sats) are added the more revenue that SpaceX would get and the more advantage that T-Mobile would have in capabilities to have a solid connection space that covers all of US and territories. Which should bring in to T-Mobile more subscribers since they can cover areas and customers that no one else can.

Back in PCS (2G) times 100MHz would not be allocated to a single carrier. It would be allocated in 20 or 25 MHz for each carrier.
And there's more.

A cell tower typically uses 6 sector antennas. Each sector gets a separate frequency (macro channel).
This is done to avoid interference between the sectors.
I think each GSM macro channel is just 5MHz. This is divided into one control channel and several bearer channels.
SMS send/receive can only use the control channel, its also used for call setup, to register a phone, ...
So supporting voice is pretty much a given, otherwise you're wasting most of the channels.
Unless Starlink can manage multiple macro channels, it can't use an entire 10 or 20 MHz frequency range.
I think Starlink would get just 5MHz to use. That's why its just 2-4Mbps. Even using 2G (GSM) 100MHz would be a much closer to 100Mbps than 4Mbps.
I don't know how and if Starlink phased array technology could come into play here. Starlink can transmit/receive simultaneously on the same frequency into different directions (each called a separate beam, as long as the beams have a lot of angular distance). This might simply not apply to cell phone service.

This definitely isn’t my field, but a phased array antenna allows for a directional signal, which is useful for shaping the effective footprint of the satellite transmission  on the ground. So, as an example, coverage of Florida doesn’t impinge on Cuba.

[Danderman]

This is probably the best topic to point out that large launch vehicles are not optimal for deploying telecommunications payloads. Because LEO comsats are deployed in planes. The planes are separated by degrees in orbit, but if they are all injected into the same orbit after launch, it’s not easy to have them organized into separate planes. Typically, Starlink deploys 66 sats into 3 adjacent planes of ~20 sats apiece.

How a LV with 10 times the capacity of F9 can support deployment into multiple planes is a bit of a mystery to me. One explanation is the V2 sats will be 10 times bigger than Gen 1.

[JayWee]

This is probably the best topic to point out that large launch vehicles are not optimal for deploying telecommunications payloads. Because LEO comsats are deployed in planes. The planes are separated by degrees in orbit, but if they are all injected into the same orbit after launch, it’s not easy to have them organized into separate planes. Typically, Starlink deploys 66 sats into 3 adjacent planes of ~20 sats apiece.

How a LV with 10 times the capacity of F9 can support deployment into multiple planes is a bit of a mystery to me. One explanation is the V2 sats will be 10 times bigger than Gen 1.

V2 sat is 7x2.5m 1.250t. At least that was before the cell antenna added. Therefore will be bigger and heavier.
There are pics available of the current version.

[gongora]

Is there any reason why the v2 birds can't have a version that supports V-band?  That license is approved and the clock is ticking on it.  The Gen2 license isn't.  (Again, "v2" vs. "Gen2" is horribly confusing.  They might mean the same thing, or they might not.)

Is there any reason SpaceX would actually deploy satellites that only had V-band?  The Gen 2 Starlinks are Ku-/Ka-/E-band.  SpaceX was one of many companies that basically filed V-band constellations just in case they wanted one after Boeing triggered an FCC processing round.  Most of those constellations aren't going to be built.

[DanClemmensen]

This is probably the best topic to point out that large launch vehicles are not optimal for deploying telecommunications payloads. Because LEO comsats are deployed in planes. The planes are separated by degrees in orbit, but if they are all injected into the same orbit after launch, it’s not easy to have them organized into separate planes. Typically, Starlink deploys 66 sats into 3 adjacent planes of ~20 sats apiece.

How a LV with 10 times the capacity of F9 can support deployment into multiple planes is a bit of a mystery to me. One explanation is the V2 sats will be 10 times bigger than Gen 1.

A vehicle does not need to be optimal. It just needs to be priced lower than any alternative. If Starship meets its goals it will be much cheaper than any near-term competitor. A "more optimal" alternative is not optimal until it actually exists. to compete, and alternative must have a lower absolute cost per launch than Starship, but Starship and its ground system are designed to minimize operations costs: full reuse, rapid reuse, RTLS, and low cost of manufacture. About the only advantage of a smaller vehicle is lower fuel cost.

[Asteroza]

This is probably the best topic to point out that large launch vehicles are not optimal for deploying telecommunications payloads. Because LEO comsats are deployed in planes. The planes are separated by degrees in orbit, but if they are all injected into the same orbit after launch, it’s not easy to have them organized into separate planes. Typically, Starlink deploys 66 sats into 3 adjacent planes of ~20 sats apiece.

How a LV with 10 times the capacity of F9 can support deployment into multiple planes is a bit of a mystery to me. One explanation is the V2 sats will be 10 times bigger than Gen 1.

Careful there. Starlink has few inclinations, but for a given inclination has multiple RAAN shifted planes. A given launch will have many sats doing RAAN drift during orbit raising to shift their plane to the desired one. That trades mass for time, as it takes some time (1-3 months) to drift, but avoids using more propellant to brute force it.

The trade space there is rate of mass on orbit and time to first revenue operation. If you can burn mass to get faster revenue, the numbers might swing that way. That's why GEO bound sats that do electric orbit raises are trading sat mass (more accurately initial deltaV after rocket separation) to first revenue.

Starship changes the metrics such that it may make more sense for Starship itself to do the plane shifts and/or increase tankage on the sats to accelerate their RAAN changes on their own. There is the third option of a tug/OTV to haul a planar set of sats from Starship release (which implies less tankage for a given sat and/or potentially allowing a thruster rated only for station keeping rather than orbit raise, thus a lighter sat), but SpaceX isn't electing that route.

[Asteroza]

I think each GSM macro channel is just 5MHz. This is divided into one control channel and several bearer channels.
SMS send/receive can only use the control channel, its also used for call setup, to register a phone, ...
So supporting voice is pretty much a given, otherwise you're wasting most of the channels.

If it's a pure 4G VoLTE deployment (which is not an unreasonable requirement), there are no voice channels, just control (and SMS within control) channel and data channels. Considering any messaging beyond SMS (especially the IoT stuff) is data, that implies at a minimum a control and one data channel for now (which was implied by what Elon was saying). As bandwidth improves, more data channels, and possibly PoTS voice on a data channel via VoLTE.

[macpacheco]

I am hearing that Starlink handheld service is only available at 1.9 - 2.0 MHz, which is  T-Mobile spectrum in the US. I am not sure why Starlink chose this frequency (apart from the T-Mobile partnership), since AFAIK, they aren’t licensed for this frequency elsewhere.

It would be an odd design choice if that was a hardware limitation.  Are you sure that's not just the frequencies they're choosing to use in the US?

First item: isn't the frequencies 1.9 - 2.0 GHz? Which gives T-Mobile 100Mz of bandwidth. Else Mhz would give T-Mobile 100KHz of bandwidth.

In the initial 12,400 sat constellation to be fully operational on or before Nov 2027 (the FCC licensing date). Would give T-Mobile excellent coverage. For the more advanced constellation of 30,000 sats which would be fully operational on or before 2030. Could have total of 3 cell phone providers each have 10,000 dedicated sats each. which would result in at least 3 cell phone providers from sat just by Starlink plus the other sat constellations that also implement cell services. Such that by 2030 you could have more than a dozen cell phone providers operating from orbit.

Additionally what SpaceX and Starlink design may be going for is to be like a cell phone tower operator that receives a fixed fee for each operational tower. As more "towers" (sats) are added the more revenue that SpaceX would get and the more advantage that T-Mobile would have in capabilities to have a solid connection space that covers all of US and territories. Which should bring in to T-Mobile more subscribers since they can cover areas and customers that no one else can.

Back in PCS (2G) times 100MHz would not be allocated to a single carrier. It would be allocated in 20 or 25 MHz for each carrier.
And there's more.

A cell tower typically uses 6 sector antennas. Each sector gets a separate frequency (macro channel).
This is done to avoid interference between the sectors.
I think each GSM macro channel is just 5MHz. This is divided into one control channel and several bearer channels.
SMS send/receive can only use the control channel, its also used for call setup, to register a phone, ...
So supporting voice is pretty much a given, otherwise you're wasting most of the channels.
Unless Starlink can manage multiple macro channels, it can't use an entire 10 or 20 MHz frequency range.
I think Starlink would get just 5MHz to use. That's why its just 2-4Mbps. Even using 2G (GSM) 100MHz would be a much closer to 100Mbps than 4Mbps.
I don't know how and if Starlink phased array technology could come into play here. Starlink can transmit/receive simultaneously on the same frequency into different directions (each called a separate beam, as long as the beams have a lot of angular distance). This might simply not apply to cell phone service.

This definitely isn’t my field, but a phased array antenna allows for a directional signal, which is useful for shaping the effective footprint of the satellite transmission  on the ground. So, as an example, coverage of Florida doesn’t impinge on Cuba.

The fact that mobile handsets must work as is (with low power omni antennas and existing protocols) severely limits how much beam forming (the end product of phased array tech) will be able be used. Yes, they will certainly use beam forming to allow the signal to be produced in border areas without interference. Its mind boggling they claim the ability to send a signal with a tiny gain antenna in the handset to be received hundreds of kms away.
Lastly there's a question of how this will be able to be used at all between multiple nearby satellites at the same time due to the consequence of multiple satellites receiving signals from a single handset.
I'm also not an expert on the field but I have some friends that are. Just trying to delve a little into the topic.

[TheRadicalModerate]

I think each GSM macro channel is just 5MHz. This is divided into one control channel and several bearer channels.
SMS send/receive can only use the control channel, its also used for call setup, to register a phone, ...
So supporting voice is pretty much a given, otherwise you're wasting most of the channels.

If it's a pure 4G VoLTE deployment (which is not an unreasonable requirement), there are no voice channels, just control (and SMS within control) channel and data channels. Considering any messaging beyond SMS (especially the IoT stuff) is data, that implies at a minimum a control and one data channel for now (which was implied by what Elon was saying). As bandwidth improves, more data channels, and possibly PoTS voice on a data channel via VoLTE.

VoLTE is just a slightly funny VoIP profile.  It's all IP under the hood, although I think that the voice packets do some header compression for the sake of efficiency, and they use codecs that have been blessed by the IMS (IP multimedia services) people.  All the signaling is SIP (Session Initiation Protocol).

I'm not sure how they're jamming text messages through.  There are profiles that allow SIP to establish TCP-based message services, instead of using the RTP (real-time protocol) transport that voice and video use.

Bottom line:  Everything is TCP/IP.  It's just that the IP headers and their underlying datalink layers get a little incestuous in the name of LTE bandwidth efficiency.  I'm sure that whatever weird thing that Starlink decided to do with its IP headers and datalinks is not the same as what the LTE guys decided to do with theirs, but it's all software.  The big differences will all be in the radio layers.

Do we know if Starlink is going to support UMTS (3G wireless) phones?  3G used completely different signaling, and the bearer channels weren't IP.  If Starlink has to support that, it'll be kind of a pain.

[TheRadicalModerate]

The fact that mobile handsets must work as is (with low power omni antennas and existing protocols) severely limits how much beam forming (the end product of phased array tech) will be able be used. Yes, they will certainly use beam forming to allow the signal to be produced in border areas without interference. Its mind boggling they claim the ability to send a signal with a tiny gain antenna in the handset to be received hundreds of kms away.
Lastly there's a question of how this will be able to be used at all between multiple nearby satellites at the same time due to the consequence of multiple satellites receiving signals from a single handset.
I'm also not an expert on the field but I have some friends that are. Just trying to delve a little into the topic.

Even if the handsets are omnidirectional, the satellites can still use phased-array antennas.  You can use phasing to make your antenna sensitive to inbound signals from a particular direction just as much as you can use beam-forming to boost your output in a particular direction.  Note that this is exactly what happens in 5G MIMO systems, and it's all backward compatible with dumber handsets.

As for satellite handoff:  It's no different than mobility handoffs in the LTE or UMTS networks.  It just happens more often.

[Asteroza]

Do we know if Starlink is going to support UMTS (3G wireless) phones?  3G used completely different signaling, and the bearer channels weren't IP.  If Starlink has to support that, it'll be kind of a pain.

I was under the impression 3G phaseouts for some MNO's had begun, thus the headless chicken routine for vehicle telematics modems (early Teslas are a victim of this and need a modem swap)

[raptorx2]

What is the business case for TMobile and other potential international carriers to forgo let's say as a wild guess, 2.5Mhz (2Mhz plus guard bands)  of their extremely valuable and in many cases, extremely expensive L/S Band spectrum to dedicate to this endeavor?  Especially for FREE on some plans?  What is the business case for SpaceX to undergo development and production costs, let alone potential regulatory headaches to add an additional L/S band payload to V2.0?  I mean there has to be a business case here.  This whole satellite connectivity for consumer handsets isn't much of a  business plan. 

The money here is in Satellite IoT, and TMobile and Starlink understand this.  Satellite IoT is expected to grow from a 775MM market in 2021 to a 6.7 Billion dollar market by 2031.  There are a number of Satellite IoT operators like Globalstar, Irdium, Orbcomm, and Swarm.  All use "dedicated spectrum", satellites, and custom user terminals for either Simplex or Duplex services.  This "dedicated spectrum" creates an environment of expensive development and low quantities on a manufacturing basis.  So the UT's tend to be "moderately expensive" $100's of $ due to lack of Economics of Scale in the development of components. Likewise, Service Plans tend to be a bit elevated for the tiny amount of data being transmitted/processed. Other issues are integration issues of bringing data from the Satellite IoT specific network into the corporation's other Terrestrial IoT networks. Some operators like Iridium are providing gateways via AWS.  So now you say you can move this Satellite IoT into global CMRS spectrum.  The Economics of Scale of the Satellite IoT User Terminal just explodes. Tiny parts/antennas/GPS/modems developed by hundreds/thousands of companies are now available to manufacture CMRS Satellite IoT devices that will be frequency agile to cover several CMRS bands for tens of dollars.. Not hundreds of dollars.  In addition, the pricing plans would likely be "Highly Disruptive" to these traditional Satellite IoT providers.  This is the 800lb Gorilla in the room.  Companies like Iridium and Globalstar have largely dismissed these "Mega-Constellations" as any threat of competition to their futures.  They have claimed in some cases that constellations like Starlink and OneWeb are actually "complementary" to their business models.  Well, with this move, these Satellite IoT companies have surely modified their "viewpoints", but perhaps not publically.   Especially those Simplex-centric Satellite IoT companies. After all, you go from needing specific globally harmonized Satellite spectrum to compete in this market to a point that most CMRS Spectrum can now be used as Simplex Satellite IoT.  This business model will propel other Telecom Operators from around the globe to integrate this Satellite IoT featrure into their own spectrum allowing these Satellite IoT devices to roam from country to country, yet still deliver the data on a unified network that is already in place. The really expensive part of the equation is already in place.  The global 4G/5G Terrestrial Core Networks.  Heck, while they are at it. Starlink could add 1090Mhz ADSB receiver and give companies like Aireon a run for their money.