# Is it possible to replace Internet with LEO satellites?

Bulk information regarding anything in this world is available in a single touch through Internet. My question is whether it is possible to replace this Internet with a constellation of Cubesats in low Earth orbit for transferring data all over the world more securely than over Internet?

• Elon Musk seems to think so. SpaceX is working on basically what you're saying, but I don't think with CubeSats. washingtonpost.com/business/economy/…
– Dan
Jul 28, 2015 at 16:49
• Replace, no. Augment, yes. There are actually devices that have some internet connectivity out there already, although it's very limited. Will it be worthwhile to use for general purpose, that's the question. Jul 28, 2015 at 17:59
• There are already a number of satellite internet providers; a Google search for "satellite internet" will show them. Jul 28, 2015 at 19:20
• I wouldn't replace the internet. IP can be used over any communication medium, so if such satellites were to be used, we would likely use them for IP traffic. In principle it could even reduce latency compared to what we have now, because light travels faster in vacuum than in optical fibers. However the drawbacks of using satellites would likely outweigh the advantages for most usages. Jul 29, 2015 at 8:28

You can't replace the Internet like that since Internet is also a collection of protocols for routing and addressing, and transmitting users' data. You can try swapping fiber optic links for LEO sat comm ones, but throughput will invariably suffer and connection will be less reliable.

• Satellites' transponders are the bottleneck - instead of an hierarchy of ISPs and various telecom companies with widening link capacities as you go up the "food chain" you have to put maximum imaginable downlink/uplink capacity into each satellite - while now a satellite may be servicing Boise, Idaho, an orbit later it will fly over Los Angeles and will be swamped with traffic.

• Expanding capacity is much more costly and risky. You have to plan a year or two in advance, book launches, pay hefty sums to sat manufacturers, launch providers, insurance companies, sweat about some doofus dropping your sat from a table. That doesn't mean land links are cheap - rights-of-way, and actual construction cost money, but much less than satellites. One must say a large scale satellite constellation has its advantages, especially if you can lower launch costs and create reliable satellite-to-satellite laser interconnections. If a huge company like Google goes into a price war with ISPs to win over part of the mobile user base, it may well establish a beachhead, drive some ISPs to bankruptcy and lock in many users.

• Comms will be patchy - heavy rain and magnetic storms will bite into your link budget and sometimes block communications over whole continents. While sat-to-sat optical links will be unaffected, customers will suffer.

• Alaska, Canada, Norway, Sweden, Greenland, Iceland, Finland, Russia, and the Antarctic penguins may be worse off after the switch - first because high-latitude connections need high-inclination satellites in a separate constellation (which is not profitable), and second because of aurora borealis/magnetic storms. (Please bear in mind this point is about small and mid-sized townships, not nomadic settlements/geological parties).

Security will be actually much worse - fiber optic cables are intercepted by advanced teams and state-level agencies, while satcomms are much more easily intercepted, jammed, and triangulated. Much like the mobile vs. fixed phones.

This said, here are some afterthoughts/nice tricks (that haven't been covered elsewhere) to improve link budgets and make a LEO satellite comms constellation more competitive:

• There are usually many smartphones in a given urban locality. One can exploit that to do some heavy preprocessing (stealth bomber-style) to correct for multipath propagation and raise downlink speeds. The number crunching will have to be done in the satellite or, less likely, in the ground control center.

• Again, many nearby emitters (smart phones) may kind of be made to coordinate with each other to act as a distributed phased array to get higher uplink speeds. This would make use of the fact that statistically not everyone is uploading a Youtube video (in normal situations).

I'm specifically not discussing latency - a variable well covered in other answers and comments. Nor do I touch upon the mass/size/energy budget of satellites in a global constellation.

• most of canada's population is actually around the same latitude as most of europe. (Edmonton, which is the most northern of Canada's large cities, is lower than the north-most part of Germany). Jul 28, 2015 at 14:47
• @njzk2 - what about the Inuits and the First Nations? Jul 28, 2015 at 14:49
• most of canada's population. Also, the antartic penguins are already using satelite internet connection. Jul 28, 2015 at 15:34

Cubesats won't work. Because they're so small, you need powerful transmitters and receivers on the ground, plus large and complicated tracking dish antennas. If you want to use a small antenna on the ground, you need a large satellite.

Satellites have far less bandwidth (=how much data can it send per second) than a fiber cable. And you can upgrade a fiber cable to have more bandwidth easily (just replace the equipment at both ends, no new cable necessary).

• Missed the cubesat bit in my own A. +1 Jul 28, 2015 at 7:28
• And even if a new cable is necessary, it stands to reason that putting that into existing tubing is less complex than to possibly design, and definitely build and launch, a sattellite.
– user
Jul 28, 2015 at 10:44
• As well as low bandwidth, the latency to a satellite is awful compared to fibre. Jul 28, 2015 at 11:37
• @DavidRicherby - not necessarily. We're talking about 400-500 km AGL, not geostationary satellites. Fiber has speed of light at $2/3 c$, and cables aren't laid often along shortest routes. Jul 28, 2015 at 11:50
• @DeerHunter: On the other side of that argument, many major web presences (like google) actively engage in geolocation, decreasing response times for their servers, so there's precedent for companies already optimizing that issue. Jul 28, 2015 at 22:41

A significant drawback to satellite based internet connectivity has nothing to do with the satellites but connection hardware, and the restrictions you'd have using the technology. Satellite communications requires line of sight, so you'd only be able to use a service like this if you had access to unobstructed sky. High-speed satellite communications requires a dish, and significant power requirements.

Any mobile devices using a satellite based internet service would be very limited in access and bandwidth - you'd have to be in clear space outside to use it and as you'd be stuck with directionless antennas your access speed would be very, very slow.

• Yeah, the cubesats certainly have to communicate omnidirectionally because of their LEO speed and the distribution of ground based counterparts. And that's bad. So is Elon Musk and his Google investors stupid or what? I hesitate drawing that conclusion because of their tremendous success. @Hobbes My comment is as much relevant to your answer. Jul 28, 2015 at 10:13
• @LocalFluff - please see the tricks section at the end of my answer. Jul 28, 2015 at 10:19
• @DeerHunter Collective coordination of pocket antennas? That's surprisingly clever. That kind of concept sounds much like how biology does communication (unradio). Biology is btw hardwired, I've learned. Life loves threads. That's how the right thing gets to the right place all the time. Of course impractical for spaceflight, so these long distance photons are an interesting challenge to what has happened on Earth during the last billion years of engineering. Jul 28, 2015 at 10:32
• @LocalFluff: A good many people live, or would like to live, in areas that are long distances from the nearest population center. Would it be economic to run say 10 miles of fiber to serve a single house, or a village of a few dozen people? Jul 28, 2015 at 19:04
• @LocalFluff: Musk's plan doesn't involve cubesats, they will use full-scale satellites which mitigates the problem somewhat. Satellite phones can get away with using a small rod antenna these days. Jul 28, 2015 at 19:40

One thing that many previous answers aren't taking into account are ways that a LEO satellite constellation could significantly enhance the internet for people who live in rural areas.

Roughly 45% of the world is in a rural area (20% of the US), and for them internet currently kind of sucks. It would be prohibitively expensive to connect rural clients to fiber. Even for those who do, there is usually only one broadband provider, so prices are often completely unaffordable for many people. This means that for a large number of customers, LEO satellite internet could be significantly faster than status-quo -- even if the only thing it did was bounce the signal directly up and down to a nearby building with fiber.

The other big advantage of such a system is it would add a competitor to a market that has a lot of monopolies. If everyone has an alternative choice, it could go a long way to keeping the telcos in line.

• These are good points!
– uhoh
May 25, 2020 at 23:43

Wikipedia defines LEO as 100-1200 miles above the earth's surface. Let's assume 1,000 miles, since it's a nice round number, and we want to stay well clear of the atmosphere for fuel minimization.

Your minimum round trip distance is now 2,000 miles. If you previously had a fiber connection to a server 100 miles away, your latency has just degraded by an order of magnitude! And this assumes assumes a satellite directly overhead.

It is common for major websites to be hosted at multiple physical servers to cut down on latency - when I check BBC News from the East Coast of the US, I am not waiting for my packets to travel to Britain and back. The BBC has set up a mirror in the States - probably within a few hundred miles of my location.

Additionally, industrial grade routing equipment is big. A Google search for "industrial router" turned up this site: http://www.juniper.net/us/en/products-services/routing/ptx-series/ptx5000/

It lists the router as over five feet tall and weighing over 1,200 lbs - and it is plugged into grid power, and kept in a climate controlled room. If you start adding solar panels, altitude management, thermal control, RF comms equipment, radiation hardening - this is a multi-ton comms sat, not a cube sat. The market for routers this size is large enough to support several companies, each with several different models offered. There are likely many hundreds to thousands of routers this size currently in use.

You could leave the router on the ground, and turn the satellite into a "dumb" repeater - passing the packets on to other satellites and finally a ground station in a pre-set path, determined by the router on the ground. This exacerbates the latency issues. Now, any time a bit error or lost packet occurs, the satellite must send a message back through the chain all the way to the ground site to ask for a re-transmission - round trip minimum is now 4,000 miles.

I don't have stats to address power requirements, but I wouldn't be surprised if RF transmissions were an order of magnitude (or more) more energy expensive than fiber.

In the end, I think you've massively underestimated the amount of traffic that the Internet carries, the size of the hardware required to handle it, and the speed at which packets are required to move.

• From your own example, the latency would only have increased by a factor of ~ 5, so not an order of magnitude. Why? Simply because RF travels at the speed of light, and light in fiber-optical cable on average about half that. When you consider longer distance relays, this shifts substantially in favor of LEO based systems, assuming that switching at each node costs the same amount of time. Ground laid cable will also have to be substantially longer than a direct line, but even discounting that and more relaying, you'd already get a lot better ping time at the distance between NY and London. Jul 28, 2015 at 17:16
• True, but with smart routing, you can even substantially reduce the total distance that a packet would have to travel, than from your own example where the satellite is straight up, if you select a satellite between the two points. Say, if we're fine with max range to satellite of 4,000 km, then your ping time between NY and London decreases from ~ 74 ms for a direct fiber-optical cable to ~ 57 ms for a satellite based system. Of course, you can always select a lower constellation's altitude and further decrease that. 1,000 miles seems a bit much. I think they're aiming for about half that. Jul 28, 2015 at 17:46
• So if each of those satellites have to be capable of routing all Internet traffic between the East Coast of the US and Western Europe, how many tons will the total constellation weigh? Jul 28, 2015 at 19:21
• @codeMonkey Elon Musk was talking about this in some talk that I don't have the time to find now (I think it was at the MIT centennial aero-astro symposium, or maybe during the constellation announcement in Seattle) and if I remember correctly, said they'd be looking at roughly 9-12 satellites deployed per launch. That seems like 1 ton per satellite then, if he had Falcon 9 in mind (range seems to correlate nicely with Falcon 9 v1.0 to v1.1 payload to LEO capability). So about 80 tons? Let's round it up to 100. Not that bad, that's just shy of the mass of a single loaded Space Shuttle Orbiter. Aug 18, 2015 at 3:50
• @uhoh - OP asked about replacing the internet with LEO sats. I can see a business case for supplementing the internet with LEO sats - for very rural areas, disaster relief, ocean going vessels, etc. But I don't see putting the whole net in orbit any time soon. Guess we'll see! May 27, 2020 at 23:51

Possible, yes, feasible no. The number of satellites required, and the cost, would far exceed any possible benefit.

LEO statellites, at best, are just very expensive 4G towers with fewer dead-spots. However they're vastly inefficient because they each only cover 3% of the internet market at any time, which puts the full bandwidth on 3% of the satellites, wasting 97% of the network. And on top of that, it only serves the areas not covered by other ISP's, while costing more. Likewise, the LEO satellites can't intercommunicate like internet hubs.

So the LEO networks like Starlink and Oneweb, can connect to the internet, but can't replace it.

• Since answers are not supposed to be opinion based, including IMHO in the answer is problematic. Suggest pruning it back to maths on coverage per base station and why that is a problem. It is also not clear what the link to your GEO proposal is about other than as a place to put you sales pitch? Having it 'as discussed by me here' would look less spam like, even better would be linking to the sources you worked from to make the video where relevant to the facts of the answer. May 24, 2020 at 8:19
• I calculated the figures myself, they're pretty basic. I was showing a GEO internet vs. an LEO. May 24, 2020 at 11:49
• As you are an invester in a competing service, your post comes across very much like trying to smear your competition. Please re-read our tour and How to Answer pages - specifically around self-promotional behaviours! May 24, 2020 at 12:32
• I'm answering the question with facts. Both LEO and GEO satellites have comparative advantages, but the current market does not offer significant demand for LEO satellites, due to the availability of ground-coverage for less cost. Meanwhile GEO satellites have problems pertaining to distance; however there are methods of circumventing this problem. May 25, 2020 at 15:25
• You're going to have to define what you mean by 'very expensive,' and please provide sources for your claim about 3 percent of the Internet market. Even if that number is accurate, once Starlink introduces laser interlinks that will no longer be valid. May 26, 2020 at 3:28