Satellite internet constellations

  • Thread starter mfb
  • Start date
  • Featured
Fast, reliable high-bandwidth internet everywhere in the world - that is the promise of the upcoming satellite constellations. To get a good bandwidth with many customers this needs many satellites; to get a good latency they have to be in low Earth orbit (LEO). Existing satellites providing internet access are mainly in geostationary orbits and there are just a few of them, leading to high latency and low bandwidth. Reduced satellite launch prices and advancements in mass production can make large LEO constellations possible.
Sounds like an advertisement, but three companies are working on very large constellations:
  • OneWeb plans to launch 650 satellites and up to 2000 more later. 6 have been launched in February, more launches are planned towards the end of the year, carrying about 30 satellites each, mainly with Arianespace/Soyuz. They expect initial service in 2021, the full 2650 satellites might be in space by 2027.
  • Starlink by SpaceX is a plan to operate up to 12,000 satellites. This is not a typo. The satellites will be launched with rockets from the same company, obviously - Falcon 9 for now, maybe Starship later. Two test satellites have been launched early 2018. The first large batch of 60 satellites will be launched May 16, 2:30-4:00 UTC (evening of the 15th in the US) - spaceflightnow will cover it and show the livestream as well. Picture of the stacked satellites - car for size comparison. 2-6 more launches are expected for this year, carrying 60 satellites each. They don't have the satellite-to-satellite laser links but will test everything else. Initial service is planned for the end of 2020 with about 800 satellites. They need 2200 satellites by March 2024 and 4400 satellites by March 2027 to keep the rights for their frequency spectrum.
  • Project Kuiper is Amazon's planned satellite constellation. Not much is known about it - about 3000 satellites, but no timeline, status of tests or anything. Amazon has the best funding situation, if they want to they will make that happen, but they seem to be significantly behind in terms of development.
  • Boeing had some plans but they seem to be stalled.
There are also Telesat and LeoSat working on constellations with about 100 satellites each.

For comparison: In all the history of spaceflight about 8000 satellites have been launched, 5000 are still orbiting Earth, 2000 of them are operational. Currently the Iridium satellites form the largest constellation with 82 operational satellites. The upcoming Starlink launch will make it the second-largest satellite constellation. Within less than 10 years OneWeb wants to operate more satellites than everyone else in the world combined at the moment while SpaceX wants to more than double the overall satellites launched in the history of spaceflight.

The satellite constellations share quite a lot of their features. To connect to a geostationary satellite you can put up your satellite dish and aim at the satellite. That doesn't work with LEO satellites - they move across the sky. To get a high bandwidth you have to track them. Moving an antenna is impractical, therefore both OneWeb and Starlink (and probably Kuiper as well) work on phased-array antennas. They are about the size of a pizza box with many individual antennas in it - by controlling the phase between them you can produce a narrow beam or receive data from a narrow range in the sky. The direction is adjustable by software, no moving parts. You also have to jump from satellite to satellite smoothly every few minutes, an interesting challenge for the data transfer protocol.

OneWeb will use the satellites only for "end user <-> satellite <-> ground station" communication. Starlink, Telesat and LeoSat work on laser links between the satellites. They can connect end users directly. For longer distances they can also provide lower latency than ground-based internet as light travels faster in vacuum than in glass fibers. This would be very interesting for high-speed trading.

Space debris is a major concern with so many satellites. After the end of their useful life the satellites have to be deorbited in a controlled way, ideally with redundant systems. Starlink's satellites are in a relatively low orbit and they are large and light - even if the controlled deorbiting fails they will deorbit relatively soon on their own. OneWeb's satellites are in a higher orbit, if one of them fails it will just keep orbiting Earth and be a potential hazard for other satellites.

These constellations can potentially provide tens of billions of dollars revenue per year. We'll probably see OneWeb and Starlink compete over customers in the next years (while also competing with ground-based internet providers), with Amazon entering a bit later. The race to the first operational large constellation is in full swing.


Summary: Three companies plan to launch more than 1000 satellites for global internet access in the next 10 years. Starlink's first launch will be May 16 (May 15 in the US).

Fast, reliable high-bandwidth internet everywhere in the world - that is the promise of the upcoming satellite constellations. To get a good bandwidth with many customers this needs many satellites; to get a good latency they have to be in low Earth orbit (LEO)
What are the receiving stations on Earth like? They need to have some tracking capability, no?


Science Advisor
Homework Helper
Insights Author
Gold Member
What are the receiving stations on Earth like? They need to have some tracking capability, no?
Actually, a phased array will receive signals from all angles. The signals are encoded so that it can "piece together" the signals arriving at each array element and reconstruct which signals belong together. That is a simplistic description. There is a lot of signal processing going on to reconstruct the signal.
More info:

"The gist of it is as follows. Starlink centers around using low earth orbit (LEO) satellites. Basically, once enough satellites are in orbit, they link to ground terminals on Earth that are roughly the size of a pizza box. Compared to existing internet satellite systems, LEO satellite constellations are much closer to Earth—99 to 1,200 milesinstead of the 22,000 miles used by traditional geostationary satellites. That means LEO satellites can transfer information more quickly, with speeds comparable to wired broadband and fiber-optic internet"

With Musk being involved, I wonder what kind of changes we should expect.

Net neutrality for one.

Going to keep an eye on this one.


Gold Member
Why wouldn't the big telecoms be interested/involved?
The telecom model is probably to utilise the investment in the extensive optical fibre already in the ground in the local network. It is also more spectrally efficient to use 5G rather than satellite, as frequencies are re-used geographically more closely. 5G base stations can utilise existing fibre and provide very high capacity. The satellite constellations seem to have a potential for spoiling the night sky.
Why wouldn't the big telecoms be interested/involved?
So far their main interest seems to be to slow these constellations down.

The overlap in potential customers is mainly coming from rural areas with poor broadband connection and low population densities. The satellites can provide a roughly equal bandwidth per area: A lower population density means more bandwidth per person. At the same time cities have better terrestrial connections already.

The static fire of the Falcon 9 has been successful, the target launch time is still the same (20.5 hours from now), the forecast is 80% chance of acceptable weather. A backup launch date is a day later.
SpaceflightNow started its live coverage

Want to reply to this thread?

"Satellite internet constellations" You must log in or register to reply here.

Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving