It’s from FR24’s KML output, which in turn comes from the plane’s transponder. I think it has one of those ADS-B thingies that replace one of the lights. I don’t know whether they’re WAAS or not.
Certified ADS-B OUT needs a WAAS GPS input.
If it is indeed true that there is no way to get a 3D velocity out of either NMEA or from the internal GPS API, that would explain why e.g. ski apps have a reputation for showing funny speeds. They would have to use the altitude data to generate a vertical velocity vector, to add to the horizontal velocity vector they already have, and altitude is never going to be all that reliable. And they don’t have WAAS.
AF wrote:
Doppler is far more precise than using 3D discrete positions…
For example, height error in 3D autonomous positions (uncorrected) is typically 5-20m whereas horizontal error is typically 1-3m.
Thanks, I would have expect the design accuracy to be the same in all axis (they sit in 20,000km orbits which are about the same as earth size) but they are not stationary, so will have more chance of going really far away than that in X,Y axis
The issue, if I have got it right, is that both NMEA and the phone OS API gives you only horizontal (2D) velocity. If for some reason you want to generate the 3D velocity you have limited options. It seems that using the altitude data is the only way.
I am sure there are android/ios programmers (“artisans”, in current terminology) here 
The other way is the build your own GPS, or buy a GPS module which outputs the whole lot.
@Ibra gladly, the increased vertical error was a conscious tradeoff for better horizontal accuracy, as the original intended application didn’t include the concept of differential corrections and therefore they chose to design the constellation to favor horizontal position over vertical.
(factor of about 3x better xy over z)
One last cool thing: there are codes embedded in the signal which are encrypted and enable military receivers to achieve WAAS-like accuracy without any differential corrections (e.g. WAAS or EGNOS)
AF wrote:
One last cool thing: there are codes embedded in the signal which are encrypted and enable military receivers to achieve WAAS-like accuracy without any differential corrections (e.g. WAAS or EGNOS)
Are you thinking about Selective Availability? That was turned off during the Gulf War and permanently disabled in 2000 on president Clinton’s orders. The increased accuracy of military receivers is because they use both the L1 and the L2 signals — and the L2 signal is indeed encrypted. By using two signals on different frequencies the receiver can compensate for the unpredictable delay in the ionosphere which is the major source of GPS inaccuracy.
Airborne_Again wrote:
By using two signals on different frequencies the receiver can compensate for the unpredictable delay in the ionosphere which is the major source of GPS inaccuracy.
Maybe with two analog signals on different frequencies you can have lot of noise cancelling?
Also (just guessing) you can do a lot of corrections with binary signals stored into an analog one (delta-modulation), this technique is partially embedded in VLT telescope adaptive optics where binary signals (e.g. pulsing stars) are used to correct for the same random atmosphere noise but in visible/infrared images, the only difference to an “enhanced GPS” is that in VLT corrections are done in real-time by physically twisting small pixels in a large deformable mirror and one has zero control on binary transmissions, I guess one cannot scale “VLT like” to consumer GPS unless you put few ground receivers/transmitters of the correction, which amount to same infrastructure as WAAS (but without having to know ground station coordinates)
Ibra wrote:
Maybe with two analog signals on different frequencies you can have lot of noise cancelling?
Noise isn’t a problem from an accuracy point of view. It could be from an availability point of view, of course.
I recall reading something about the encrypted signal still having recoverable information which, while not being usable for absolute positioning, it was usable for accurate position delta and thus velocity. The technique used complex signal processing.
The last I’ve heard on this, from one app developer, is that Android gives you x y z velocity vectors accessible via the API but IOS doesn’t give you z. So while with Android you may have unreliable z (due to constellation geometry etc) with IOS you have nothing and have to try to generate z from deltas in the altitude value, which is a very noisy way to do it.
What aviation units do I have no idea and there is no easy way to test it because there is no reference.
@Airborne_Again
No, selective availability was a modulation scheme which introduced noise into the L1 and L2 signal codes.
The military still has access to the P code which enables them to do ionospheric correction without any additional differential corrections (e.g. WAAS).
Standalone, they achieve decimeter accuracy worldwide.
Which is part of how they dropped bombs through windows in the Gulf War way back in 1990.
(not a pleasant topic, but it’s where we get the tech which enables us to do so much of the flying we do so easily)
