After SpaceX Starlink upgrading to terabit space satellite internet

It’s going to cost the mobile-phone companies, chipmakers, device manufacturers and software developers about $200 billion a year in research and capital spending to get 5G fully deployed. Next generation space-based satellites could entirely replace the ground-based fiber and cell towers as a lower cost and more maintainable infrastructure.

There are some estimates that mass production of small low earth orbit internet satellites could drop to $100,000 each. SpaceX low-cost launch could make building, replacing and upgrading internet satellites at lower-cost than cell towers. Cell towers have land leasing costs.

Instead of networks of 4000 to 12,000 internet satellites there could be 500,000. An entire space network of 500,000 could be swapped out for $60 billion.

They would not need fiber because they would send multi-terabit lasers through space. Space based satellites could replace cell tower on a one for one basis.

The US Army is already testing laser communication with cubesats and could make minor modifications to get to 2.5 gigabit per second speeds.

Lasers clearly can have terabit per second communication speed. There would be many issues to achieve this in terms of targeting. Low Earth Satellites with hyperefficient propulsion could maintain lower orbits of 150 kilometers. This would reduce latency from 30 milliseconds to a little less than 10 milliseconds.

Arxiv – Design Considerations for a 5G Network Architecture

According to, there are currently about 120,000 cell towers in the US.

Other estimates are 307,626 cell towers in 2016.
Average cost of building a cell phone tower – $175,000+
Average yearly cell phone tower lease rate – $45,000+
Lowest annual cost to lease a cell tower – $100

5G network architects predict the need to support a 1000-fold increase in traffic compared to 2010.

Assuming only one additional small cell per current tower for 5G that doubles the current number of towers. These new cells may not need full-size towers, however, because they may be placed on buildings and will be located in more densely populated areas. But that does not negate the fact that the available real estate must be found, then purchased or leased, properly zoned, and power and backhaul utilities run to the locations.

5G may need an estimated 300,000 to 400,000 NEW cell sites on top of existing towers.

Optical satellite backbones in space are already 1.5 times faster than backbones on the ground

16 thoughts on “After SpaceX Starlink upgrading to terabit space satellite internet”

  1. We can’t simply replace all fiber connections with satellite links. Even though Musks Starlink constellation will provide significantly lower latency and higher bandwidth thant most satellite solutions today, it will never be able to meet 5G demands of down to a 1ms latency. It is simply impossible for radio signals to travel all the way from the earth station – satellite – user that fast. Satellites are to far away. But it could function as a backup for macrocell sites in case fiber connection is somehow disrupted.

  2. Elon Musk knows how to make new markets and disrupt existing ones successfully. Electric cars, once am expensive curiosity, are well on their way to the economic mainstream because of Tesla. Transportation to Earth orbit via the more economical SpaceX’s boosters allows Starlink and other possibilities that didn’t exist when this mode of transportation was restricted to large governments and multinational communications companies. The dream of economically feasible renewable energy is coming true because of Solar City and Tesla’s home battery business. The Boring Company will make space under our feet which is presently unused into vital pathways for mass transit.

    Starlink and SpaceX are nowhere near their full potential as businesses. Just as the scope of aviation was difficult to gauge in the early twentieth century, the economic scope of spaceflight is something few of us can imagine.

    If nothing else, we will need resources now in the asteroid belt to make a civilization in which everyone has the blessings of advanced technology – one large nickel-iron asteroid could replace countless open-pit mines on Earth with none of their ecological damage. Solar energy in space could be beamed down much more efficiently than it can be harnessed from solar farms, with no carbon signature to speak of per kilowatt.

    We can make the Earth back into a garden – if we do all the “dirty” activities we now use to make energy and produce infrastructure in orbit, and not on Earth. It’s worth working for.

  3. Since this constellation is non-GEO, the system isn’t using a single satellite. It can’t. It uses so many satellites to assure one’s always in the sky over a customer.

  4. I think I see what you’re thinking – the Clarke, or geosynchronous Earth orbital belt, in which satellites can remain on station over one spot on Earth with a limited amount of station-keeping burns, is so limited in the number of satellites it can host that litigation over the rights to occupy a particular spot in GEO is now a “thing”. So 500,000 satelites (which is a technical possibility, not the 12,000 satellites Starlink’s actually filed applications to launch) does sound like an impassable mess. It’s not.

    Starlink is non-GEO – while the lower orbits its satellites occupy are smaller in circumference than the Clarke Belt, they aren’t confined to a narrow belt in that space. They are distributed over a much wider space. And because they must be in KNOWN orbits to work as comsats, their location is much more closely tracked than anything in controlled airspace in Earth’s atmosphere – it has to be, if gigabit lasers are going to communicate with it.

  5. What happens when all of our communications is via satellite and Earth gets hit with a CME (coronal mass ejection)? I hope we keep an emergency backup fiber backbone. Maybe someone in the DOD can black budget that … just in case.

  6. Because these are communications satellites, their exact location is – must be – known every microsecond. If it were not, those multi-terabit lasers could not engage in data transfer with them. The risk of collision between a launch vehicle passing through that orbit and one of those satellites is many orders of magnitude lower than between airliners touching down in a major airport’s terminal airspace.

  7. You still don’t understand, it doesn’t matter if the satellites are using terabit optical links. The bottleneck is the downlink/uplink not the communication between the satellites.

    State of the art satellite technology can give you around 1Gbps per channel at those frequencies. If you are going to have a big constellation, channels must be allocated in a manner that adjacent satellites don’t use the same frequencies and so, they don’t interfere with other, so each satellite can only use a limited set of channels, similar to how cell phone towers work.

    If you have 80 thousand users connected to a single satellite (that’s a optimistic assumption, this also assume all the other satellites will also be loaded as well), and each user wants 1Gbps and they use their connection only 10% of time, that gives you a combined throughput between satellite and ground of around 8 terabit per second.

    So I would assume a throughput from each satellite to the ground not higher than 200Gbps, compare that with the 1500Gbps from the each inter-satellite optical link, a 10 times mismatch. On this scenario, this would allow only 25Mbps per user, which is a good bit higher than what is available now.

    And no, you can’t cramp much more data on those bands, you must use geographic division to allow for channel reuse, that is really hard to do at those low orbital altitudes, 200Gbps is a crazy assumption, I’m assuming 200 channels per satellite.

    Starlink could have its uses, like offering service to remote locations, connecting distant cities cheaper to the internet than laying optical cables, but assuming it could replace 5G is fantasy.
    A single 5G tower will be capable to surpass the user throughput of a Starlink satellite, now imagine tens of those towers in a single big city like New York.

  8. So at worst SpaceX need merely improve their protocols 26 fold over what exists?

    Seems quite doable in a constellation where any most local concentration node has hundreds of destinations to which to route it’s packets, and phased array transceivers for noise and location selectivity.

  9. Imagine that Starlink is going to supply service to around 1billion of users, and with 12 thousand satellites, if you could balance equally among all satellites, you would end up with more than 80 thousand users for every single satellite.

    The problem is not the maximum throughput of the satellite, they can take the load easily, but how to you allow a single antenna in the satellite to take 80000 users without significant interference. The optical links can indeed take terabits per second, but each channel in the Ku band, can only take a few gigabits at best, and those need to be shared across thousands of users.

    I wouldn’t like a LTE tower to take more than 3000 users in a small town, and that is using three antenna pairs, so you can physically divide the cell in three regions to reduce the load over the radio medium.

    Now those satellites service a spot under the satellite of over 1000km of diameter. From the point of view of the satellite it doesn’t matter if the all the users are equally distributed across that spot or if the are all crowned together.

    How do you design the protocol to keep them from interfering with themselves?
    Normal satellite internet services use medium control protocols that really on the fact the there aren’t many users on each satellite beam, and those users have limited speed, mostly less than 10Mbps. And beyond the lag because of the geostationary high, the protocol itself also adds lots of lag because of medium control.

    Now we are talking about a system that is going to squeeze thousands of users, each user with speeds of over 100Mbps, over a single satellite that is at least 600km high and over that it needs to deliver low lag and need to be cheaper than a ground system.

    I’m very skeptical they will be doing that, when doing just the radio magic, for the same thing with a few hundreds of users, connecting with an antenna 1km away, in a small town is already a nightmare.

  10. I thought about Mike McCulloch’s LEMdrives too. Even a few milli Newtons of thrust would suffice to keep then at currently quickly decaying orbits.

    But if those or any other of these thrusters actually work, then their application for station keeping of ultra-low orbit satellites is just a niche of their very many potential applications.

  11. A good thing about very low orbit satellites is the implicit incentive to build propellantless propulsion engines.
    This should cause more money to be given to the Mach Propulsion people.
    A propellantless engine could extend the life of a very low orbiting satellite a lot. Reducing costs and increasing profits.
    And if it gives profits people will continue to develop it, invest in it and use it.

  12. Does anyone ever actually read what they propose? Like, sit back and actually look at it? “Instead of networks of 4000 to 12,000 internet satellites there could be 500,000.” and “They would not need fiber because they would send multi-terabit lasers through space”. IF these were the only pieces of hardware up there it might, just maybe, be a workable idea but throw up another half a million things to avoid while trying to get into deeper space is absolutely nuts.

  13. A really big part of the technical difficulties with 5G is how to provide those really high speeds for more users than 4G(LTE). Thus the use of higher frequencies and new exploration of possibilities with more aggressive space division multiple access(SDMA), more towers, more antennas(mostly phased arrays), new digital signal processing and so on.

    Now, I’m extremely skeptical Spacex has magically solved the biggest problem with wireless communications, that is, how to allow a really big amount of users to share the same radio transmission medium with good quality of service and minimal interference. The backbones connecting the cellphone towers never were really a bottleneck for that kind of technology.

    It doesn’t matter if a satellite can communicate with another satellites at terabits speeds using laser signals whose have very tight apertures. If the ground users are going to use microwave signals, transmitted and received probably using, again phased array antennas, we return to the same traditional problems faced by ground systems, where a good amount of effort is being but designing 5G.The bottleneck is connecting the final users(a really big amount of users btw) to the wireless node, not connecting the node with the Internet.

    Sending the towers to space could actually make the problems worse when one take in consideration the aperture of the users microwave beams and how potentially way more users are going to connect with a single satellite than what would happen with a 5G basestation tower.

  14. “Instead of networks of 4000 to 12,000 internet satellites there could be 500,000. An entire space network of 500,000 could be swapped out for $60 billion…maintain lower orbits of 150 kilometers. This would reduce latency from 30 milliseconds to a little less than 10 milliseconds.”

    Take it low enough that the swap-out occurs as satellites deorbit from atmospheric friction and that will not only reduce latency even more (not by much I admit), but guarantee a self-cleaning constellation, avoiding a permanent Kessler Syndrome. (Vuukle won’t let me post a link to Wikipedia’s article without classing it as “spam”.)

  15. A hyper-dense giga/terabit communication mesh around the world?

    The speeds and complexity of such a thing boggle the mind, but it’s certainly not above the capabilities of computers to track and control.

    And now imagine more modest but equally omnipresent equivalents to this around every major celestial body of interest in the Solar System, Mars, the Moon, on several asteroids, providing remote logistics and communications with manned locations and automated infrastructure.

    A true backbone of the upcoming space economy. This could really be a multi trillion dollar business, ready for SpaceX to take. The potential size of this economy and the riches of its companies also boggles the mind.

Comments are closed.