Hacker News Reader

Comment by petschge | Hacker News Reader

I do wonder a bit why why they need THAT many satellites. Assuming an orbit at 200 miles altitude, we have a horizon distance of about 2000 miles. Comparing the area of that 2000 mile radius ground patches with the surface of the Earth, we need about 10 such patches. Filling in the gaps between patches might require 10 more satellites. Naively I would have assumed that orbital mechanics costs us another factor of two, so fourtyish satellites required. But they plan to launch more than 4400 satellites. More than 100 times the naive estimate. And at the cost involved they don't do it for fun. Where does that discrepancy come from?

View on HN

Other people mentioned the issues with antenna obscura— buildings, walls, trees etc can all get in the way.

If you connect to a satellite at an oblique angle, it is also much further away. A satellite in a 500km orbit is 500km away from you when it's directly above you. But when it's on the horizon, it can be more like 2500km away.

This adds a little bit of latency (maybe 10-20 ms). It also hurts the signal, since going through atmosphere attenuates signals more than going through vacuum. Especially for Ka band signals, which particularly weakened by water vapor.

On top of that, SpaceX needs to have redundant coverage over an area. You don't want your customers' internet to stop working because a bit got flipped on one of your satellites or a thruster blew up unexpectedly. Also, a single satellite may be able to handle Alaska or Siberia but have bandwidth issues handling the state of California.

Is your horizon distance calculation based on no obstructions? Most people don't live on the plains ;)

I am genuinely curious how you arrived at that number, FWIW – I don't know how to go about calculating something like that.

That does indeed assume a sphere. If you are curious about the geometry, maybe https://en.wikipedia.org/wiki/Horizon#Distance_to_the_horizo... can help.

If you run the calculation the other way around, every satellite has to service a ground patch with about 380 miles radius. And your antenna has to swivel to a point that might be more than 60 degrees away from zenith. So every obstruction of more than 30 degrees above the horizon will lead to at least intermittent disconnects.

More bandwidth and better resilience against loss of individual satellites? Similar rational to how when you install a wifi network in a high density usage environment you want to use a lot of lower power access points each covering a small area to get better throughput for a large number of users rather than trying to cover the whole area with a small number of higher power AP that each cover a wider area.

I have thought about bandwidth, but that should only matter once you have a lot of customers.

This is just a guess, but maybe to simplify the tracking requirements on the ground?

With a minimal satellite system, your tracking needs to follow satellites all the way across the sky, and probably also needs to be able to rapidly make large changes in where it is aimed.

With a very large number of satellites, a ground station might be able to get by with only having to be able to track within a small part of the sky.

If you have enough that there are many of these small areas that always have a satellite in them distributed all across the sky, then even places that have a lot of buildings or trees blocking much of the sky should still have some patch they can view that works.

Two reasons: more bandwidth, and more coverage in a specific region. One of the biggest downside with a LEO constellation is that only a few satellites are covering a given area at a time. Most of them, by far, are over water, where there's likely little to no usage. So because of that design, do you have to launch a lot of satellites to make up for the lack of throughput in the single areas.

Area of earth is 170 million square miles. 2000 mile radius patches have 12.5 million square miles. Even assuming perfect tiling, 10 wouldn't be enough.

Throw in realistic angle limitations (not horizon) and coverage per satellite rapidly decreases

Ok, make it 14 instead of 10 then. That's a small correction. Filling the gaps on the other hand does not require a factor of two, but only 2 \pi / sqrt(27) or about 1.21. Multiplying the 14 with that gives us 17 instead of the 20 I estimated. I don't want to squabble about 20% here, but ask about the factor 100 discrepancy between naive estimate and projected number.

That is because you’re expecting ground access points to magically communicate with a satellite at 0 degree elevation. With a realistic elevation to deal with obstacles, diffraction, and electronic steering limitations, you’ll end up with a radius of coverage closer to 200 miles, which would give you that factor of 100. This shorter radius also limits losses due to distance attenuation. Your naive calculation assumes the limiting factor is the curvature of the only planet, which it is not; the earth could be a flat slab and you’d still need the same number of satellites.

You launch that many you expect a lot of them fall down I suspect. Probably some go into a higher orbit and serve as spares until they decay into some more appropriate orbit.

Better to get approval for too many than not enough?

Reply on news.ycombinator.com