Starlink ground control station and gateway

[kartofell]

Hello guys, I'm new on the forum and just wanted to share some ideas...

Considering the last 1B$ development fund Musk managed to obtain, it seems very likely that it would be enough to finance manufacturing sats + launch for about 15-20 launches, that should covers the US easily as he said and starts to show revenues rapidly. However nothing is said about gateways and ground control facilities he would need to operate his constellation.

I believe it is rather expensive to operate a constellation. I've read from Globalsat that "managing high fixed-cost infrastructures is an important challenge" for them and they spend around $200 million each year in their ground facilities. Does it apply to SpaceX ? Maybe they have so many sats that the cost of building ground facilities will be easily absorbed...
I am curious to know how it works, how many ground stations and gateway do you need to operate such a constellation ? If you want to cover the whole world, you'll probably need to build facilities at every corner of the planet... political issues may arise if we consider the potential military use of Starlink.
Anyway, Musk talked a lot about manufacturing its satellites and deploy them but stayed rather quiet about the ground facilities it would requiere. 

[Tulse]

Once there is intercommunication between satellites (which this initial group doesn't have, if I understand correctly), then the need for a huge number of ground control stations presumably is greatly mitigated.  You wouldn't need line-of-sight to every satellite in order to control it, or to act as an internet gateway.

[Brovane]

I work in the Telecommunications industry for a major ISP and these are just some of my personal observations.

Normally a ISP would carry the traffic and then have a arrangement with a peering provider.  For example a Cable Company in the US might have a peering relationship with another company to carry the traffic between continents.  Because it would be horrendously expensive to lay your own fiber across the world. 

For Starlink, SpaceX can send traffic between satellites so essentially it can be it's own provider to carry that traffic across oceans etc.  Actually because of the characteristics of laser's in vacuum versus fiber optics, SpaceX would essentially have lower latency than a fiber link across the Pacific Ocean for example.  Video below talks about this. 



However at some-point traffic will have to traverse out of SpaceX's network and into ground based networks to reach servers in data-centers.  The more ground stations that SpaceX has for these peering relationship, the closer that this traffic can be dropped off to it's destination.  This drives lower latency and makes SpaceX's less dependent on 3rd party fiber. 

For example you are in Europe and you are trying to get stock trading information from a server located in New York.  If the closest ground station to this server in New York is in Virginia, then that imposes additional latency.  If a ground station was in New York then you have lower latency.  For ping responses, milliseconds do matter. 

[kartofell]

I agree that once there is intercommunication with laser technology, the need to use ground control facilities is drastically reduced. But nothing is said about when the next generation of satellites will be deployed. I thought laser tech was supposed to be implemented for the second batch of VLEO sats so I guess not before 2024. If spaceX wants to be operational in 2020 it will need to build a large number of ground facilities I suppose.

So maybe starlink first generation will just pay to use existing stations as many operators do instead of building their own.

[gongora]

Starlink will have their own ground stations (although I wouldn't be surprised if they colocate with other ground stations in some areas). The number they need will depend on both geography and how much bandwidth they need.

[catdlr]

Hello guys, I'm new on the forum and just wanted to share some ideas...

Excellent initial post. Welcome to the Forum.

[kartofell]

Hello guys, I'm new on the forum and just wanted to share some ideas...

Excellent initial post. Welcome to the Forum.

Thank you

Starlink will have their own ground stations (although I wouldn't be surprised if they colocate with other ground stations in some areas). The number they need will depend on both geography and how much bandwidth they need.

For geography, it is hard to imagine the terrestrial infrastructures necessary beyound US, because SpaceX has not the authorization to operate beyond its own frontier right now... I believe it will get access to worldwide market but by then, laser intercommunication technology will probably do the job and ground stations won't really be a matter as we discussed.

However, it is clear that bandwidth is an important factor but I don't believe Musk will provide the 100mbps he talked about to everyone on the planet... especially in high traffic periods. I do believe that somewhere close to 10mbps would be clearly more accurate (this is pure speculation of course), for two reasons: First, it is actually 10 times better from what you can actually have with internet broadband services in the US (ISP sell you subscription saying they reach 50-100mbps but everyone know that's not the reality, they rather offer something close to 500kbps - 1mbps on average. Second, if you offer a high bandwidth to every one, you limit the number of users per sats, you have to trade off between number of users and bandwidth, and SpaceX will maximise number of users/sats such that it offers a decent connectivity (10mbps seems to be decent for satellite service). If I recall, 10mbps is somewhere close to a good 3G right?

[vipeout]

If they can close the business case in US and rural areas only, then I think you can assume such low bandwith. Otherwise 10Mbps is nowhere near good enough in many countries.

I'm a huge fan, I have great ISP and don't complain. I want to support Starlink and I'll gladly swap, but if they can't give at least 50 then it's a no-go. 100 is a standard of a good connection in my country and every major company (plus many smaller ones) provides that, at least in bigger cities. I did not research into this, but I believe it's safe to assume more European countries have such bandwidth offers.

Therefore I believe he's serious on that number. Needs to be competitive on performance, if we assume the often used here monthly cost used in US (40$). I pay ~15$ for 100Mbps, I don't think starlink can compete on cost with that.

I'm just jumping in to help verify assumptions. After all, this is supposed to be global.

[gongora]

I do believe that somewhere close to 10mbps would be clearly more accurate (this is pure speculation of course), for two reasons: First, it is actually 10 times better from what you can actually have with internet broadband services in the US (ISP sell you subscription saying they reach 50-100mbps but everyone know that's not the reality, they rather offer something close to 500kbps - 1mbps on average.

This is simply not true.  Some internet providers are better than others and some can have slowdowns from their max rates but they don't drop to 1% of the stated values.  None of them are 1Mbps on average.

If SpaceX wants to provide 100Mbps then they'll limit the number of subscribers in an area to allow that level of service.  They've already said they're not targeting densely populated areas.

[kartofell]

If they can close the business case in US and rural areas only, then I think you can assume such low bandwith. Otherwise 10Mbps is nowhere near good enough in many countries.

US shouldn't be the only market but I assume it will be in the short run, at least for the next coming years up to 2023-2024. And yes rural areas is definitely the only market. There is no way internet satellite can compete with terrestrial internet, especially with the technology progress of fibre.

I do believe that somewhere close to 10mbps would be clearly more accurate (this is pure speculation of course), for two reasons: First, it is actually 10 times better from what you can actually have with internet broadband services in the US (ISP sell you subscription saying they reach 50-100mbps but everyone know that's not the reality, they rather offer something close to 500kbps - 1mbps on average.

This is simply not true.  Some internet providers are better than others and some can have slowdowns from their max rates but they don't drop to 1% of the stated values.  None of them are 1Mbps on average.

If SpaceX wants to provide 100Mbps then they'll limit the number of subscribers in an area to allow that level of service.  They've already said they're not targeting densely populated areas.

Well, I'm not US citizen nor internet satellite user but from what I'm reading on different forum, most users of internet satellite (in US rural area), are often complaining about the poor access they get. If they could get those 100mbps all the time, I'm not sure there would be such a huge market for Starlink in the US as Elon said. From what I know, most of the US is already covered, so SpaceX will have to make people switch from their ISP to Starlink. If US citizen living in rural area had already the opportunity to get the 100mbps (or even 10mbps) at 50$-100$ a month, why so much people would be complaining and willing to switch to Starlink?
My intuition is that most people get less than 10mbps at a very high price and that's why Starlink is such a golden opportunity. I may be wrong as I said it's pure speculation

[Unrulycow]

Most rural homes in the US have the option of DSL that is massively overpriced and pathetically slow, or GEO satellite internet (HughesNet) that has high latency and low data caps.  In bigger towns there might be the option of cable internet (expensive and terrible customer service) or 3G/4G from cell carriers (data caps).  Starlink won't be able to compete with fiber, but that's not available in most of the country.  If it doesn't have data caps it will be very competitive outside major cities. If it's close to the same price and speed as cable, it will be a huge success as it would be competing with the most hated companies in America.

[joek]

...
Well, I'm not US citizen nor internet satellite user but from what I'm reading on different forum, most users of internet satellite (in US rural area), are often complaining about the poor access they get. ...

Current consumer satellite broadband (*cough*) is served by GEO satellites.  Not a good comparison with Starlink, which uses LEO satellites and eliminates the latency and bandwidth issues inherent in current (and likely for the foreseeable future) GEO-based services.

My intuition is that most people get less than 10mbps at a very high price and that's why Starlink is such a golden opportunity. I may be wrong as I said it's pure speculation

That is generally incorrect.[1]  In any case, no need to speculate; see FCC 2019 Broadband Deployment Report.

[1] edit: unless by "most people" you mean US rural populations.  From the FCC report note that the primary underserved markets are rural, which generally have DSL, GEO satellite, or cellular at relatively high $/bit (and relatively slow), or which do not have any broadband access.  Underserved markets are a good place to start.

[lonestriker]

Elon and Gwynne have always highlighted the under-served populations with respect to Internet service, so it's a fair bet to say that is one of their primary target markets.  Even without going into the densely-populated urban areas, there are plenty of opportunities to market Starlink services: ships, airplanes, military (including secure communications to bases), remote sites (including scientific sites like observatories out in the middle of nowhere).

ISPs could use Starlink to solve their "last mile" problem to provide connectivity to remote sites without having to lay cable or fiber to every last subdivision or remote town.

[joek]

Elon and Gwynne have always highlighted the under-served populations with respect to Internet service, so it's a fair bet to say that is one of their primary target markets.  Even without going into the densely-populated urban areas, there are plenty of opportunities to market Starlink services: ships, airplanes, military (including secure communications to bases), remote sites (including scientific sites like observatories out in the middle of nowhere).

ISPs could use Starlink to solve their "last mile" problem to provide connectivity to remote sites without having to lay cable or fiber to every last subdivision or remote town.

Yup. FCC's recent Rural Digital Opportunity Fund has up to $20.4B for providing broadband for up to 4M US rural communities-residents.[1] Starlink could bid on that directly, or partner with incumbent ISPs to help address.  (Think similar to Tesla's selling offsets to incumbent auto makers.)  And that does not include the other markets you mention.

[1] Note FCC's current definition of "broadband" is 25Mbps down and 3Mbps up; looks like they may change that in the future to 10Mbps down and 1Mbps up(?).

p.s. I prefer to phrase it as the "first mile" problem.  Telco's referring to it as the "last mile" instead of the "first mile" is indicative of their less-than-customer-centric attitude.

[lonestriker]

p.s. I prefer to phrase it as the "first mile" problem.  Telco's referring to it as the "last mile" instead of the "first mile" is indicative of their less-than-customer-centric attitude.

  Even though I dislike Comcast, Ma Bell and the Baby Bells as much as the next disgruntled consumer in the US, calling it the "last mile" can also be translated to being Internet-centric. i.e. the Internet itself is the first mile, and getting into homes and businesses is the last leg of the network.

But agreed completely, there is no customer focus for many of the cable companies and telcos because there's little to no competition in the US.  Anything that adds competition is a good thing to keep vendors honest (and not abuse their monopoly position with ridiculous random fees, caps and pricing.)

[joek]

Even though I dislike Comcast, Ma Bell and the Baby Bells as much as the next disgruntled consumer in the US, calling it the "last mile" can also be translated to being Internet-centric. i.e. the Internet itself is the first mile, and getting into homes and businesses is the last leg of the network.

But agreed completely, there is no customer focus for many of the cable companies and telcos because there's little to no competition in the US.  Anything that adds competition is a good thing to keep vendors honest (and not abuse their monopoly position with ridiculous random fees, caps and pricing.)

Oooh... Have not heard reference to the ghost of "Ma Bell" recently (except when I play The President's Analyst for my nieces and nephews).  Agree from an Internet-centric view it may be the last mile, or more properly, the last hop.  But IMHO it is the first, as we don't have an Internet without those people on the first hop.
 
But to the telco Bell-heads, it was always something they wished would go away, along with asymmetric bandwidth, datagram-based protocols (they will have to pry virtual circuits from our cold dead hands), and ISDN (which was going to Save Everyone, and which they had to rip out for DSL, b*tched about it to their graves, and f*cked up completely).

Sorry, OT, but brings back memories; apologies if any readers was a Bell-head.

[Keldor]

Initially, they're not going to have a lot of collective bandwidth with their satellites since they just won't have most of them up.  During this time, using satellites as relays between two ground stations puts less stress on the constellation's resources.

In this service mode, SpaceX sells antennas to various carriers, who are responsible for integrating them with their own network, whether in cell towers, data centers, or any number of other possible places.  On the other end, the service provider installs the antennas directly with customer's houses or perhaps in a box shared between several houses in a neighborhood.  Remember, more antennas in a small area won't make things faster since they're still sharing airspace.  A single shared antenna is not only cheaper, but also faster since it's better able to schedule things rather than wrangling for bandwidth with other antennas.

Eventually there will be enough satellites up to form a backbone for the internet.  In the meanwhile, they will simply form "arm bones" connecting remote connection points with main hubs.

It's not clear whether Starlink will ever provide "last 100 meters" infrastructure, or if service providers will simply use it to avoid having to install and maintain millions of miles of fiber to every corner of the world.

[speedevil]

Initially, they're not going to have a lot of collective bandwidth with their satellites since they just won't have most of them up.  During this time, using satellites as relays between two ground stations puts less stress on the constellation's resources.

Caching may also help a lot, on several levels.

A whole netflix library these days is several kilograms of flash. This could plausibly sit on each sat.
Several tens of gigabytes of storage in the end-user device along with multicast would enable a severalfold reduction of traffic for denser regions. (your neighbour starts watching the latest episode of whatever when it's released, and it's downloaded en-block to everyone in that cell).
Multicast for live events, similarly.

[ThomasGadd]

Initially, they're not going to have a lot of collective bandwidth with their satellites since they just won't have most of them up.  During this time, using satellites as relays between two ground stations puts less stress on the constellation's resources.

Caching may also help a lot, on several levels.

A whole netflix library these days is several kilograms of flash. This could plausibly sit on each sat.
Several tens of gigabytes of storage in the end-user device along with multicast would enable a severalfold reduction of traffic for denser regions. (your neighbour starts watching the latest episode of whatever when it's released, and it's downloaded en-block to everyone in that cell).
Multicast for live events, similarly.

Look at how content serving is done now, on large specialized server farms.  Adding that level of complexity to all the satellites would be a lot of overhead. 

[speedevil]

Initially, they're not going to have a lot of collective bandwidth with their satellites since they just won't have most of them up.  During this time, using satellites as relays between two ground stations puts less stress on the constellation's resources.

Caching may also help a lot, on several levels.

A whole netflix library these days is several kilograms of flash. This could plausibly sit on each sat.
Several tens of gigabytes of storage in the end-user device along with multicast would enable a severalfold reduction of traffic for denser regions. (your neighbour starts watching the latest episode of whatever when it's released, and it's downloaded en-block to everyone in that cell).
Multicast for live events, similarly.

Look at how content serving is done now, on large specialized server farms.  Adding that level of complexity to all the satellites would be a lot of overhead.

Adding some limited capability to the satellite may significantly increase its effective capacity without interlinks.