Is speed measured with Doppler by consumer handhelds or not?
Vincent van der Laan
I would like a conclusive answer, please!
My feeling is that Doppler is used (as soon as I move my GPS around it
immediatly gives a speed reading, it can't have made position measurements
yet, because it does that periodically like once per second (I don't know
this period exactly). I have read a lot of articles that say yes: the
Doppler effect is used.
But Mattheww Weber says no:
in other words the GPS uses the locations measured, looks at the location
difference, looks at the time difference and thus calculates speed.
See the following discussion:
mattheww weber <matth...@cox.net> wrote:
----- Original Message -----
From: "Vincent van der Laan" <vin...@allalin.nl>
To: "mattheww weber" <matth...@cox.net>
Sent: Saturday, September 14, 2002 5:37 AM
Subject: Re: Using GPS as speedometer
> I got this on private email, but I'm putting it back in the newsgroup
where
> it belongs:
>
> mattheww weber <matth...@cox.net> wrote:
>
> [Remark about doppler to measure speed snipped]
>
> > In theory possible, in practice, extremely impractical. Consider for a
> > moment how fast the sat is moving (roughly 8000mph). Depending upon your
> > exact orientation relative to the sat it produces a Doppler shift from 0
to
> > 8000mph if you are standing still. You are quite literally trying to
> > measure the height of a molehill on top of a mountain, from survey
> > instruments located in the valley.
>
> This happens allready literally: the height of mount Everest is known to 1
> cm (depending on what you consider sea). We know that K2 grows with a
speed
> in the order of 1 cm per 10 years (or was it 100 years) (which is very
fast
> on a geological timescale) so it seems we're pretty good at measuring
> anthills (not molehills) on mountains.
Different problem.
To make that measurement, and get that accuracy, you rely on carrier phase
measurements and position relative to known object (true Differential
operation) as well as two GPS's. Carrier phase measurements can NOT be made
from moving objects. If the points of reference are fixed (as they are in
survey measurements), and you wait long enough (typical carrier phase
measurement takes about 20 minutes), and take enough measurements random
errors tend to cancel nicely. There is a wonderful demonstration of that.
Do you know how to calculate Pi with a random number generator (it actually
works)?
The reality is Doppler is not used in any handheld or non-survey grade GPS.
If it was, you wouldn't see the wander when you stopped moving. That is one
of the reasons why a handheld can sell for $100, and an Ashtech or survey
grade trimble costs $5000.
A 66 mhz general purpose CPU costs about $50 still. No way you can affort
to put it in a $100 instrument, and no way it will run for long on 4 AA
batteries. The $100 product has to have a raw parts costs for everything
including the processor, rom, ram, receiver and display, on the order of
$40.
>
> 1 cm compared to roughly 8800 m is in the order of 1 in a million.
>
> > A Doppler correction is made to retrieve
> > the data from the sat, but no effort is made to measure it, and it is
part
> > of any transmitter/receiver frequency drift. Doppler and drift are
all
> > bundled together to get a lock on the sat.
> >
> > It is a ugly spherical geometry problem, and even a 1% error is 80mph!
> >
> > As I said, in theory it can be done, in practice, it is far messier, and
far
> > more complex then simply calculating the difference between the
points.It
> > is not clear that the processors used inside most handhelds have the raw
cpu
> > performance to even do the calculation in real time.
>
> That is (if I understand things correctly) what the Kalman filter does
very
> elegantly, with only 66 MHz cpu's. (It seems most processing power is used
> to drive the display (on a map GPS).)
>
> After doing a Google search on: GPS speed doppler, I found many articles
> that confirm Doppler use by GPSr.
>
> Here is just one of them: http://www.aprs.net/vm/gps_cs.htm
>
> regards, Vincent
>
Mark Topham
individual says it isn't.
And (no offence to the individual) you take the individual over the group
because it SOUNDS BETTER? (Not because the individual is more authoritative)
As for the argument relating to the processor, it's a red herring. So, even
IF the hard part can't be done in a typical general purpose cpu you can buy,
from Garmin, an OEM GPS receiver set that is small enough to install into a
handheld unit. It is all custom chips to do the hard part, all you need is a
serial interface to it.
The general purpose CPU only has to handle the serial data and allow for
user interaction. It doesn't even have to do all the work.
(None of this implies how a typical garmin unit actually works, they have a
ton of options for their own units...)
"Vincent van der Laan" <vin...@allalin.nl> wrote in message
news:B9A985AC.5208%vin...@allalin.nl...
Sam Wormley
>
> Could someone from Garmin or Magellan comment on this?
>
> I would like a conclusive answer, please!
>
> My feeling is that Doppler is used (as soon as I move my GPS around it
> immediatly gives a speed reading, it can't have made position measurements
> yet, because it does that periodically like once per second (I don't know
> this period exactly). I have read a lot of articles that say yes: the
> Doppler effect is used.
>
Try: http://groups.google.com/groups?q=Doppler+group:sci.geo.satellite-nav+author:wormley
Vincent van der Laan
> http://groups.google.com/groups?q=Doppler+group:sci.geo.satellite-nav+author:w
> ormley
I read your articles and I'm totally convinced about using the Doppler
effect!
while reading I stumbled across this excellent reference you made:
http://groups.google.com/groups?q=velocity+doppler+kalman
Which convinced me even more!
:) Now how do we convince Matthew Weber? :)
Thanks Sam!
Regards, Vincent
Sam Wormley
>
We should be able to come of with an experiment that would tell us
something about the design/velocity algorithms employed in a particular
GPS receiver. There are a lot of good brains in the newsgroup... who's
up to the challenge.
An experiment [well conducted that tells us something new] is worth a
thousand expert opinions.
Hans-Georg Michna
>We should be able to come of with an experiment that would tell us
>something about the design/velocity algorithms employed in a particular
>GPS receiver. There are a lot of good brains in the newsgroup... who's
>up to the challenge.
>
>An experiment [well conducted that tells us something new] is worth a
>thousand expert opinions.
Sam,
wouldn't the following already do?
Look at a track record. Calculate the velocity from position and
distance, then compare it to the velocity the device logged.
If there is no difference, the device obviously did the same. If
there is, the device must have gotten the speed information
elsewhere, presumably from the doppler effect.
(I can't do it easily, because I stored my tracks in Garmin's
MPS format. Apparently a stupid mistake of mine. I'll change
that. Would anybody please write an MPS to CSV converter?)
Hans-Georg
--
No mail, please.
George Tyrebyter
>
> We should be able to come of with an experiment that would tell us
> something
I wondered about it. I went outside and stood for a while in the yard,
the GPS receiver happily calculating its position once per second, with
the positions bouncing around a bit, but 0.0 for speed.
Then I swung the GPS receiver in a circle at arm's length and it
registered speeds well over 10 miles per hour. Its position never
varied more than about a meter, so it could never have calculated
successive positions indicating a speed more than about 1 meter per
second.
This is repeatable at will for as long as you want,-- hold the GPS
receiver still, and it shows 0 speed. Swing it vigorously in an arm's
length circle, and it shows speed in double digits miles per hour.
--
George Tyrebyter
Hans-Georg Michna
>Sam Wormley wrote:
George,
very good idea! This already proves it, more or less. Unless we
want to assume that the GPS receiver takes several position
points while you swing it around, this proves the direct speed
sensing.