Since my last message, I've done a more thorough accuracy test on
the Motorola i730. Median error of my 20 fixes was 5.6 meters. (That
is, half the fixes were within that accuracy.)
I collected the observations during a 2 week period. Since the
satellites return to almost identical positions over the site every 24
hours, I was careful to scatter the time of day of the fixes as widely
as possible in order to get a random sample of satellite geometries.
Of the 20 fixes, 7 were after sunset, from twilight to almost
midnight. The others were during the day.
The test point is obstructed to the east by a wood building spanning
nearly 180 degrees of azimuth and about 30 degrees high at the closest
point. There is some signal penetration through the building; I've
gotten reasonable fixes inside with the phone. To the west and
northwest of the site are two trees about 30 degrees high. At this
time of year the foliage is thin. I believe the trees don't seriously
block the satellites.
For the correct coordinates of the test point I used the mean of
several postprocessed pseudoranging sessions with a Magellan Promark.
I'm confident the coords are accurate to better than 1 meter (1-sigma
uncertainty). Latitude of the site is about 35 degrees.
It helps a GPS receiver if you give it time to acquire all available
satellites. But I simulated a user who doesn't know any tricks to
optimize performance. In most cases my cell phone had been off for
several hours and I turned it on a few seconds before arriving at the
test point. A few times the phone was already on. However, in those
cases the GPS had not been used recently. As soon as possible after
turn-on, I commanded a GPS position refresh. (The GPS in the i730
doesn't run continuously.) Then I recorded the fix in my spreadsheet.
Immediately afterward, I also took a fix with a Magellan 315 for
comparison. The Magellan had an advantage over the cell phone: I gave
it time to lock on to all the satellites it could find. However, I
didn't use averaging. When I was satisfied the 315 had acquired all
available sats, I moved it a little to clear the average, then
immediately took a fix.
Here are the statistics for the 20 data points from the Motorola i730.
(The Magellan 315 values are in parentheses.)
Median actual horizontal accuracy: 5.6 meters (3.5 meters), or 18 feet
(11 feet).
Median estimated accuracy: 4.6 m (3.0 m), or 15 feet (10 feet).
The accuracy estimate you see on phone's display has only a weak
relationship with the actual accuracy of the fix. I discovered that by
dividing the 20 observations from the i730 into two sets of 10. The
"accurate" set contained the observations with estimated accuracy (the
value reported by the receiver) better than median, and vice versa for
the "inaccurate" set. Then I tabulated the median error (actual and
estimated) of both sets:
"accurate" set "inaccurate" set
actual error est. error actual error est. error
i730 5.2 m 3.0 m 7.0 m 14.8 m
The "accurate" set is only slightly better than the "inaccurate" set,
though there's a large difference in its estimated accuracy.
There's no altitude on the GPS location screen, but you can read
altitude in meters by entering Trace Mode and going to the GPS menu. I
didn't collect statistics for that value.
I have not yet found a place with no cell service but a good view of
the sky, so I can't say if the GPS is independent of the cell towers.
Simple GPS navigation is possible with the Telenav application. This
is a Java program I found on the Internet. After I downloaded it to a
floppy, a wireless service center put it on my phone for a few
dollars. Most of the functions are disabled if you're not a subscriber
to the Telenav service, but you can save your location and navigate
back to it later. I used it one time to find my car on a parking lot
and missed by 10 feet.
The bearing of the destination is shown as a dot on the
edge of a
circular "compass", and distance in feet or miles is displayed in
digital form. Unfortunately, the compass doesn't rotate to your
direction of movement. North is always up.
When the Telenav app is executing, the phone's GPS runs continuously.
If you need course and speed over ground, the GPS menu in Trace Mode
can display that data. It's even possible to run Trace Mode and
Telenav simultaneously, so the course/speed screen periodically
overlays the Telenav compass screen, though I don't think that trick
is worth the trouble.
--
Paul Hirose <ewwb4-khh25@earINVALIDthlink.net>
To reply by email delete INVALID from address.
Linear Mode
