GPS relative accuracy for two nearby units

[chrisN]

2 questions about GPS, when read by two different gps units:

First, as posed by Peter H on the AF447 thread: [What is] the difference in readings between two bog-standard GPS receivers at the wing-tips. Bog-standard, without superfast processor and GPS chips.

The issue being, whether they would have the same error as each other in positional accuracy and hence would be very accurate in relative position to each other (e.g. to measure bank angle?)

The second issue, which I am interested in for reasons totally unconnected with AF447 – in fact, when two gliders using Flarm are in close proximity: As well as whether two GPS units have the same GPS position error, when using the same satellites because they are close to each other: if two adjacent gliders are differently banked, or say on opposite sides of a steeply turning circle, could they be calculating positions by reference to different, and continuously changing, satellites; and if so, could this lead to quite different errors and so not have accurate relative positions and heights?

(Flarm works by comparing the position and projected track broadcast by one unit with the position and projected track known to the other, and hence projected whether potentially colliding or not). In the case of two different gliders each with its own Flarm device, they may be made by different manufacturers – albeit using the same software and/or algorithms – but if they might use different makes of GPS engine, I wonder if they would anyway be susceptible to picking up different satellites and getting different local errors in position.

Chris N.


[edit: posted before I was able to see EmBee’s helpful answer to the first question above on the other thread.

And, by the way, I think the second issue also has implications for ADS-B as a proximity warning, particularly for light aircraft, if that ever comes about, as ADS-B seems to be a sort of Flarm-like concept AIUI.]

[Urs]

The GPS position error between nearby antennas is almost the same if:
1) the same GPS receivers are used
2) the GPS receivers use the same filters (kalman filters etc)
3) the GPS receivers have the same configuration (e.g. SBAS, WAAS, EGNOS on/off)
4) the same satellites are used for calculting the position
5) some additional considerations (e.g. multipath...)

FLARM fulfills 1-3 and is working on 4 and 5 (all units, independent of manufacturer, use the same GPS receiver, algorithms and core software)

ADS-B fulfils none of the above.

Urs - FLARM

[chrisN]

Urs, many thanks. Regards - Chris N.

[ATCast]

The issue being, whether they would have the same error as each other in positional accuracy and hence would be very accurate in relative position to each other (e.g. to measure bank angle?)Some research has been done to see if this is feasible. I remember NLR (http://www.nlr.nl/smartsite.dws?id=2614) did something in this area. To get accurate results you need to have the antenna's on separate locations (wingtips), but you probably will have to do the data processing in one unit, making sure that you use the same satellites for calculating the positions of both antennas. And by looking at phase shifts, you can probably derive the roll rate.

And, by the way, I think the second issue also has implications for ADS-B as a proximity warning, particularly for light aircraft, if that ever comes about, as ADS-B seems to be a sort of Flarm-like concept AIUI.]It has indeed. Since ADS-B does not restrict the algorithms used inside the GPS unit, different errors exist between different receivers. But even if a same GPS unit would be used in the other aircraft, the position errors would differ. As Urs explains, this can happen due to multipath (reflection of the signal), or the use of different satellites (i.e. satellites shielded by the wing in a turn).
Among other things, ADS-B sends quality indicators which indicate the accuracy and the integrity of the position report. They can be combined into a containment radius. This containment radius is usually governed by the integrity bound. That is the size that an error can grow undetected by the RAIM algorithm due to a single faulty satellite. A typical position error is in the order of 10 feet, but the size of the undetectable error due to a faulty satellite can grow to much more than that. A typical containment bound is 0.2 NM. Note that the RAIM algorithm is not standardized either, so different GPS receivers will give different containment bounds.
For proximity warning the typical containment bound is good enough, for collision avoidance it might be better to look at the accuracy instead of the integrity, and assume that the satellites are performing well.

Regards,

ATCast

[chrisN]

Atcast, thanks for that.

For ease of writing, I use “Flarm” and “PCAS” as generic words, like Hoover for vacuum cleaners or Biro for ball point pens – there are more than one model and more than one manufacturer of each sort of device, but no simple phrase that I know to encompass them and convey a clear meaning.

I had been led to believe by some commentators that gliding use of Flarm could be seen as nothing more than an interim step, until ADS-B comes in as a proximity warning and collision avoidance system for everything from the smallest to the largest aircraft. This interchange is convincing me that that on the contrary, Flarm is here to stay and cannot (on the basis of what I have seen so far) have its utility for gliding replaced by ADS-B.

[For those who don’t know why, it is in the nature of gliding that we go to the same places and get very close to each other as a normal practice. For example, thermals are energy sources – the best remote indicator of a thermal some distance away is another glider circling in it. So any other glider will head straight for it, and try to get into the same circle. Similarly with wave lift and ridges – we go to the same parts which seem to be working best. A proximity warning that another glider is approaching is useful but not necessarily a matter of urgency. A collision alert, that its predicted path is to intersect our own predicted path, is a potential life-saver. That is what Flarm does, and ADS-B on the basis of the above clarifications cannot do.]

My campaign in UK gliding, for what it is worth, will remain for the time being:

1. To encourage Flarm for glider-glider collision alerts – our most frequent collision source, by an order of magnitude;

2. PCAS for gliders that can accommodate that too, to help avoid transponder-equipped aircraft, though glider-power collision are far fewer;

3. And transponders, for those who can also fit them. (I have written elsewhere and do not repeat it now the issues, and also the difficulties, including those made by EASA.)

[My interest is in trying to improve aviation safety, particularly for gliding, as a member of the BGA safety Committee among other things. I have no commercial connection with Flarm or PCAS etc. except as a satisfied customer.]

Chris N.
[edited to clarify use of proprietary names etc.]

[ATCast]

... until ADS-B comes in as a proximity warning and collision avoidance system for everything from the smallest to the largest aircraft.This interchange is convincing me that that on the contrary, Flarm is here to stay and cannot (on the basis of what I have seen so far) have its utility for gliding replaced by ADS-B.

ADS-B is essentially the means to send the position of an aircraft to anyone who is interested. On it's own it is not a proximity warning or collision avoidance system, but it can be used to gather data to feed those systems. If a manufacturer can make his case that for collision avoidance only the accuracy and not the integrity matters, he might be able to have his system accepted.
As far as I can see, that same approach is used in FLARM, since there is no guarantee that both FLARM units use the same set of satellites. Also the integrity bounds for FLARM are similar to those of ADS-B, since both rely on the same GPS system.
The changes of a satellite being at fault are extremely small, so it might be an acceptable approach to use accuracy and not integrity for collision avoidance.

An advantage of ADS-B is that in the longer term many more aircraft will be equipped, and that it provides radar like surveillance capabilities in areas not covered by radar. The planning is to have ADS-B mandatory in Europe by 2015, for aircraft heavier than 5700 kg . For lighter aircraft there are no plans yet to mandate ADS-B implementation.

Regards,

ATCast

[chrisN]

ATCast, where I think ADS-B would be useful to gliders too would be a combined unit for gliders that does the Flarm alert thing for another glider, and adds any proximity for an aircraft with ADS-B, with a display that shows all.

I would accept that the accuracy of an ADS-B indication might be less than one I need for Flarm – I don’t want a powered aircraft anywhere near me, whereas a glider I tolerate in close proximity provided that it is not actually colliding, as we are both doing the same thing and (usually!) both know what we are doing that close. Anything else, if it is near, I want to know where and can then decide whether to take avoiding action in case the other doesn’t.

There are already some Flarm-type units which have a graphic display of all Flarm threats and nearby transmitting entities. I think at least one can also display ADS-B signals received from nearby aircraft.

[At present, I would have to engage in a significant modification to use such devices, which might be worthwhile when ADS-B take-up is more widespread, and suitable units are widely available that can be easily installed in my limited instrument panel space. My present Flarm is the cheapest and simplest, however, and only indicates the nearest or the most threatening other Flarm unit. I was able to add it without an EASA modification, as it is carry-on personal equipment. The more elaborate ones need an instrument panel re-engineering job for many gliders, including mine. The costs are likely to be disproportionate, particularly for many older, low-value but quite airworthy gliders.

The EASA thing is much more of a problem in the UK than in at least some countries in Europe - the CAA is taking the usual line that the Government here does with all EU regulations: sticking to the absolute letter of the most restrictive possible interpretation of the rules, if not adding to them. So even if there are approved EASA modifications valid in the rest of Europe, they don’t help much, if they don’t cover the necessary wiring or instrument panel etc. changes in the application and approval. I am advised that I need another application and approval before I can have a transponder, for which I have neither expertise, nor resources, as well as being reluctant to spend whatever it takes to get there. Every unique glider/wiring/battery/instrument panel combination looks like it would need another unique EASA approval. The BGA is working with the CAA to try to get some sense into it, but at present that’s how things are. Idiotic. Changing a non-ADS-B unit for one which also displays ADS-B would be yet another EASA modification, it seems.]


Chris N.

[swh]

It is possible to use various GPS antennas located around a vehicle to measure roll, pitch, and yaw. I know of at least one autonomous weapons system that uses it as the technology is cheaper than an inertial system and requires no alignment. I do not know if the technology is available in the civil sector.

[chrisN]

I have had two+ solid state attitude indicators in my glider, and I would want a lot of reassurance that any new one would be worth an upgrade. One was designed for power, assuming that continuous turns would not be required, and when steadily turning it “assumed” that a steady state was straight and level and stopped indicating a turn. Useless. It had to be reprogrammed.

Another over-read the bank angle by a factor of about 2. It was reprogrammed, and now only over-reads a bit. Knowing about it, I make allowances. The worst thing is that it lags real time. Not a problem in a steady state, or slowly changing, but if it all goes to worms or one manoeuvres rapidly, it lags and then jumps to a new position (where it was about ½ second before). Only my mechanical turn and slip keeps reading right.

Any new device needs rapid and accurate response to be of use in a glider, and I would want proof before upgrading. Having said that, if it works in a military application, it probably does the job – but at what cost?

I will watch with interest to see if a GPS-based attitude indicator comes on the market, but another risk is loss of satellites, particularly when turning tightly with large bank angles.

That might even be an issue in the circumstances that led to the original question by another enquirer on the AF447 thread – if all else on the airliner fails, could such a device save the day?

Chris N.

[00nix]

To come back on measuring bank angle: NLR together with Delft University of Technology have equipped their aircraft with GPS receivers on a wing tip, fuselage and nose. It has indeed been proven that GPS could be used for attitude measurement, although GPS accuracy comes into play there. Determination of roll and pitch rates however are easy and suprisingly accurate. The doppler effects are unaffected by the accuracy effect (you are not actually reading the content of the signal, only phase).

Another, possibly military, application of far spaced receivers (i.e. wingtips) is that it becomes impossible to spoof a GPS as you can only broadcast one set of coherent but fake satellite signals. One then only has to check whether both receivers indicate the same position to detect a spoofing.

As for using GPS for position reporting, the algorithm or the satelites in the solution do not really matter as long as the accuracy of the position solution can be either guaranteed or reported (RAIM).

[Lex19]

GPS relative accuracy for two units depends of GPS time accuracy in each of units.
Time difference between two units 1 nanosec created position error approx 1 feet multiple to number of units.
Regular GPS have 100 nsec time error, which correspodent to 200 ft position error. What do you think???
Lex19