Sort of related question: is it possible to get more precise global positioning from a lot more satellites? I was wondering if a swarm of satellites is capable of GPS.
In most cases, though, faster/more reliable fix will probably be a bigger advantage than precision. When you have a partial sky view, more satellites means a better chance of enough satellites falling in that window for a fix.
Most GNSS are largely similar, though, and there are certainly enough papers describing how to perform fixing across satellites of independent constellations so I'm sure it's been implemented in at least higher-end devices. Still, I suspect inexpensive consumer receivers are probably calculating a fix per constellation and then combining those. Fixing across constellations requires combining ephemera, time base, etc. data that varies between them, but all of these are known with high precision, you just have to pull them together.
Since most GNSS are mostly the same, new GNSS constellations are usually launched either to improve coverage in a region, or these days with several global constellations, more often for sovereignty reasons. If you have weapons systems that rely on GNSS, it's better to have your own constellation. GPS for example no longer has a wartime selective availability capability (that would deny use by anyone other than US forces) but it did in the past and, if WW3 broke out, it would probably be implemented again. For much this reason different GNSS constellations tend to operate in different bands (not always, though) so that a country can selectively jam adversarial GNSS systems without affecting their own. GNSS has its origins in the military and is still largely a military capability today, with Galileo being the odd one out that focuses more on commercial applications (Galileo's selective availability capability is used to charge a subscription for higher-precision uses).
And some GNSS do have "accessory" capabilities, probably the most prominent being Galileo's two-way search and rescue messaging capability that will compliment COSPAS-SARSAT. But that's always been the way with satellites. For example, the GPS constellation carries detonation sensing equipment for nuclear counter-proliferation. It just saves money to combine payloads.
One of the reasons to not put GNSS constellations in separate bands is that it simplifies the construction of multi-constellation receivers. Of course, it also makes it more difficult to selectively jam. Some combination of these interests mean that GPS, GLONASS, and Galileo are very close siblings in the radio spectrum. It gets surprisingly political.
https://en.wikipedia.org/wiki/Galileo_(satellite_navigation)...
I actually have a receiver that does exactly this! It is a Trimble Resolution T, available for very cheap (I believe they're from old cell towers), and it's intended exclusively as a time reference. You first run a 24 hour data collection, and it precisely determines your location before storing it. From then on, the receiver only calculates the time. Very cool stuff, I'm planning on building a GPS-disciplined rubidium clock and stratum 1 NTP server with it.
>Still, I suspect inexpensive consumer receivers are probably calculating a fix per constellation and then combining those.
This sounds like a reasonable solution now that you mention it, for an inexpensive device. Part of my confusion is based in the fact that the constellations actually use different reference datums [0]. Wouldn't the different constellations give meaningfully different fixes even if the solution was "perfect"?
I mentioned in another comment that I pretty much forgot that this could be a political thing. That's a very good point and I don't know why I didn't think of it. Especially with selective availability, which as far as I understand, the US government could just turn on again at a moments notice.
[0] https://gssc.esa.int/navipedia/index.php/Reference_Frames_in...
Obviously if you just mix up the bytes willy nilly then nothing good will come out of it.
How familiar are you with the terminology of mathematical optimisation? If you are then the intuition is that each satelite provides a different error term in one big function you are trying to optimise. The “knowns” you plug in to the terms are your observations (pseudorange, phase, etc) and the orbital parameters of the satelite broadcast by the satelites. The unknowns you are trying to solve are the x,y,z,t coordinates of the receiver you have. You are trying to find the unknowns which minimise the sum of the error terms. The error terms of the various constellations might have slightly different ”shapes”, but you can still run your optimisation them the same. As you add more satelite the number of unknowns remain the same (4), but you are adding more and more equations to your overdetermined system. When things are not pathological that leads to a better, more precise fit.
> aren't they all doin the same thing
Even if they were, sometimes it is valuable to have your own toy. If Russia is having a war in some region of the world and the US decides to turn off the GPS signals there is not much Russia can do about that. It is a US satelite and the US adjusts it as they see it fit.
If Russia has their own satelites and the US would like to deny them this the US have to attack Russian assets directly. That is a very different ball game.
Regarding the political aspect - you're right, and it's easy to forget about. I could say something cynical about that but I'll leave it.
[0] https://gssc.esa.int/navipedia/index.php/Reference_Frames_in...
It might be possible to do a different technique and use a swarm of internet satellites: something like how the TRANSIT (aka NAVSAT) system could, in theory, provide navigation from satellite internet machines. In practice you'd need to know each member of the swarm's orbits very precisely to have sufficient accuracy with their Doppler shift, which means that you prefer station-keeping burns to be infrequent[2], and because they are in low orbit- lots of drag, so they need to regularly burn- that's harder to do (this is why the existing systems use medium orbits). Also, a TRANSIT style operation can't really give you turn-by-turn directions on your cellphone, it was about updating a very powerful submarine inertial guidance system with big honking mechanical gyroscopes every few days, the INS on your phone isn't nearly good enough to handle turn-by-turn directions with TRANSIT update rates.
1: Japan has their own Just Japan system, QZSS, which I have vague memories is different, but I don't remember enough to explain it.
2: The way these systems work, you can never really have more accuracy than you know the satellites location in space. In general, you want ground based systems to track and update orbital parameters after every satellite burn to get the necessary accuracy.
Japan's QZSS is complementary to America's GPS and also covers Australia and everywhere in-between due to its orbit.
For generally practical purposes, it can be considered part of GPS. Any GPS receiver in range can receive its signals.
On the satellite side, I had vague memories they were weird when I wrote that. From some quick googling I think they are doing a trick with clock synchronization from ground stations to avoid needing an atomic clock in space, since over Japan and Australia they will always have LOS to a ground-station (which they can use in place of the atomic clock as the master time source for each satellite)[1]. At least, that's what the 2008 Ph.D thesis "Remote Synchronization Method for the Quasi-Zenith Satellite System: study of a novel satellite timekeeping system which does not require on-board atomic clocks"[2] suggests that they were doing. That was published before the QZSS system actually flew, so I still am not sure if they actually went that way or stuck with the conventional atomic-clocks-in-space approach. So maybe they are unique in not needing so much hardware in space, perhaps?
1: GPS, designed to deal with at least the opening stages of nuclear war, had as a design goal the ability to operate without ground stations at least for a while, and there are large swathes of the world that don't have suitable ground stations, so they had to go with the atomic-clocks-in-space route to provide global coverage. By focusing on just a small area and ignoring the 'must operate for several days during a nuclear war' case, QZSS maybe could get away with something cheaper.
2: https://unsworks.unsw.edu.au/server/api/core/bitstreams/9386...
In practice, there's a floor to the noise which is a reasonable lower bound on your uncertainty.
There's also factors that can't really be overcome from more sats, like multipath effects.
However, this would require different hardware on both the sats and the ground stations.
You'd prolly need much better coordination algo's too.
Under ideal conditions GPS is already accurate to 30cm, which is frankly astonishing on a planet of 200 million square miles.
You're half-way there, keep it up !
If you meant a single LEO swarm operated by a single entity, yes that's possible too, and there are proposals to do it, because LEO birds can provide much stronger signals, you'd have a decent shot of it working indoors and stuff.
But, the ground-segment necessary to keep track of the clocks on a large number of satellites would be formidable. They'd also be out of view of ground stations for a larger portion of their orbits (since they're so much lower), which means some new work in tracking the clock drift, which is one of the main limitations on accuracy.