The place for technical discussions about GA and flying.
Forum rules: Technical discussions about GA only, please.
User avatar
By PeteSpencer
FLYER Club Member  FLYER Club Member
#1896265
flyer5 wrote:
The impression I get is RAIM takes calculations from several combinations and draws conclusions about which single satellite might be suspect. If it can see 6 satellites it can determine a correct fix, and potentially exclude the rogue satellite from further calculations.

However if the GPS is taking an average of all the fixes then I don't see why you need RAIM. With high numbers of satellites in view, the plain average would surely be pretty good even with the duff one in the calculation.

I wonder if anyone knows how this works in practice. Does SkyDemon just take a simple position output from the device or does it have access to individual fixes? Do typical devices take averages of all available satellites? Do typical devices do RAIM of any sort?


We're at risk of comparing apples with pears here: If you are on an IFR flighplan, planning to carry out a PBN (formerly RNAV) approach at your destination, you are obliged to check RAIM availability at your destination during a small window either side of your ETA: This can be done online, but most navigation systems (EG Garmin 430W - the W is crucial) will do it for you:
By tailbob
#1896398
If you are receiving a WAAS signal then that is all that is required to monitor GPS accuracy. If no WAAS signal is received or able to be received then RAIM is required to be checked prior to a LNAV approach. At present the EGNOS signal is not available for “Safety of Life” use in UK but hopefully that will soon be replaced by the Augmentation signal from the Inmarsat-3 F5 satellite which has been arranged for UKNOS.
PeteSpencer liked this
User avatar
By Flyin'Dutch'
FLYER Club Member  FLYER Club Member
#1896438
non-Aviation GPS receivers do not use RAIM.

But why worry about RAIM? We all know that it is perfectly possible for experienced pilots with decent kit to end up as a smoking hole, and indeed that experienced professional pilots en up in trouble after a home-baked approach into St. Elsewhere so what is stopping any pilot, flying a home-baked approach in their jalopy without any decent instruments.

Fill your boots.
Dusty_B, A4 Pacific liked this
By riverrock
FLYER Club Member  FLYER Club Member
#1896530
RAIM is just an integrity check on the GNSS (essentially, checks the signal from the various received satellites against each other, looking for dodgy signals).
The availability check (as I understand it) just predicts whether you will have enough satellites in view to do the calculations.

Doesn't stop a home baked approach, other than it's hard to load them into certified equipment!

Phone location services is a bit of a black box, albeit most of the code is open source if you know where to look. I believe SkyDemon adds a Geoid correction, depending on the SBNS source - although info may be proprietary.

A side note - any mapping data from China (such as BeiDou) is intentionally shifted for security reasons. They use GCJ-02 which intentionally warps coordinates. No idea how they get around that for aviation use - but I doubt their Satellite constellation is being used outside of China for this reason! https://en.wikipedia.org/wiki/Restricti ... a_in_China
By Rjk983
FLYER Club Member  FLYER Club Member
#1898784
flyer5 wrote:
However if the GPS is taking an average of all the fixes then I don't see why you need RAIM. With high numbers of satellites in view, the plain average would surely be pretty good even with the duff one in the calculation.



Mean, Median or Mode?

Slightly facetious point, but it is difficult to know what happens if a cluster of positions are calculated, do you pick the centre of all points, do you delete any outliers then calculate the centre of the rest, if you get two clusters of points which do you take as correct or do you pick a mid point between the two clusters?
User avatar
By flyer5
#1900819
Rjk983 wrote:Mean, Median or Mode?

Median should do nicely.

I'll do a bit more research on how accurate/reliable a SkyDemon android tablet is.

I would like to show - purely for interest - that it is accurate enough to permit an IMC approach based on flying into the overhead, joining the circuit and then descending to x00 feet on a normal final approach. (That would be a default pattern based on the premise it must be safe because that's what everybody does in VMC ).
User avatar
By Tim Dawson
SkyDemon developer
#1900889
If everything is going well, the altitude reading on SkyDemon will be super accurate, and most of the time this will be the case. It's easy to cross-check with your pressure altimeter set to local QNH and it should be within 100ft, possibly much better than that.

The trouble is, sometimes, this reading can be out by quite a long way, with no good way of knowing that it's out. In my experience this is due to your tablet having a poor view of the sky. Perhaps it's only receiving GPS from a few satellites, or from more but which are set out in a flat geometry that's poor for calculating altitude very well.

Typically SkyDemon gets its altitude directly from the device's location services with very little additional information. Certainly nothing about constellations and satellites in view.

Receiving from an external well-sited GPS receiver ought to be better, and best of all if you've got an antenna outside the aircraft.

You could use SkyDemon to guide you all the way down a GPS-based glideslope in theory, but to be safe you would have to keep cross-checking with your pressure altimeter to ensure your GPS altitude remained accurate. Who wants to do that? That's why we cut it off at 500ft.
User avatar
By flyer5
#1901124
From a cursory look at Google Play, there are apps that show individual satellites, their position and signal strength, so there must be an easy way of getting this detail from the operating system.

This has now changed my vision. As you begin your descent, SkyDemon would review the number of satellites visible/the number of acceptable calculations. Secondly it would attempt to exclude any outlier fixes, Thirdly it would assess the geometries. From all of this it would calculate a decision height in real time. If you had lots of satellites in good positions it might say 150 feet whereas if the actual or theoretical deviations were not so favourable, you would have to do with 450 feet.

This all may seem along way off, but 50 years ago I dreamt of having a doppler inertial navigator on my plane. Now I have SkyDemon which is a hundred times better and less that a thousandth of the price. :D

I'll get one of these apps and have a look at how good the signals are inside a 172/PA-28.