GPS unit is not exactly accurate. Is this normal?

December 12, 2006 3:17 PM

GPS unit sometimes is sometimes a bit inaccurate... Is this normal?

I am new to GPS and am having problems and I was wondering if any of these are normal (using the mio C310x)?

Sometimes when I arrive at my destination it will tell me to make a left in the opposite direction from my destination another 60ft (the destination is on my right). Is it supposed to be 100% accurate? Or do GPS units tend to be a bit off? (It also tells me that my apartment is 4 or 5 buildings down from where it really is.)
Is this normal or is the unit broken?

Another problem I had is that I went downtown and the GPS kept losing my signal and rerouting me over and over again. There were tall buildings surrounding me if that matters. It kept telling me I was on the wrong street (ie. it would tell me I was on 5th st. when I was on center st.) Also, it sometimes would tell me to go left and a split second later tell me to go right (when there were no streets in between. It seemed a bit confused).
Someone told me that this is called bounce. Do all GPS units have trouble in that kind of environment? Is there anything I can do about it? Will getting an external antenna help?

posted by D Wiz to technology (19 comments total)

About your first question. Yes, this is basically normal. This is not a shortcoming of the GPS per se, but of the maps that it's using and where on the map it has recorded that certain addresses are located.

The maps that the GPS systems don't (typically, to my knowledge) have every building on them, instead they just have numbers recorded every so often on the street and then the machine interpolates where a particular address ought to be based on known intersections. E.g., 150 Foo Street ought to somewhere around the middle of the 100 block of Foo St. on the "even" side of the street. So this is probably why it thinks your apartment is slightly further up the street than it is; when people build things, we often don't allocate exactly the same amount of frontage to each street number, hence the offsets.

Not so sure on the second question, because I've never noticed behavior like that on mine (mostly the ones I've used just lose signal lock when between tall buildings) but if the machine could only 'see' a few satellites, I suppose it could get its position inaccurately, and think you're on a different street or give you bad route corrections based on this.
posted by Kadin2048 at 3:30 PM on December 12, 2006

When taken together, autonomous civilian GPS horizontal position fixes are typically accurate to about 15 meters (50 ft).

GPS units do typically have trouble in high-density urban areas.

GPS is good enough to get you close to where you want to be. You still have to verify street names and your destination address.
posted by b1tr0t at 3:30 PM on December 12, 2006

This is not a shortcoming of the GPS per se, but of the maps that it's using and where on the map it has recorded that certain addresses are located.

Actually, it is a combination of the two. GPS signals are accurate to +/- 20 meters or so, and logical addresses don't always match up with physical addresses. Usually GPS will get you to the correct block, but you can't reasonably expect it to point you to the right driveway every time.
posted by b1tr0t at 3:33 PM on December 12, 2006

GPS units can be inaccurate - I don't know about your in-car navigation unit, but my hand-held unit I use for survey work gives me an "accuracy" reading in metres, and tells me the number of satellites it is connecting to. And if I'm in an area with lots of tall trees, for instance, the accuracy goes down.

Directional accuracy can also be a problem, because the GPS unit doesn't have a "compass" - it calculates what direction your travelling in based on the last few locations you've been at. If you're moving very slowly (or are stopped), or the locational accuracy is low, then it won't determine your direction very well.

For example, imagine you're at point X. Then, a few seconds later, you're 5 metres north of point X. In a perfect world, the GPS unit should work out you've moved north. But, if your locational accuracy is down to, say 20 metres, the GPS might inaccurately read your second point as being south of that first point, and hence it will tell you you're going south instead. Now if you're travelling fast enough, it won't matter, because that second point is likely to be north of the first point no matter what the accuracy, but if you're travelling slow, the points it's calculating your direction from might be within the range of uncertainty. I don't know if I explained that very clearly...

But yes, an external antenna can help.
posted by Jimbob at 3:33 PM on December 12, 2006

Yes, it's normal.
Usually routing software compensates somewhat for the error rate, but it's normal for the fix to wander.
The fix depends on the how many satellites (and how strongly) the gps unit can "see". Buildings, clouds, moisture in the air, trees, etc can all have an effect on the signal strength.

An external antenna can help somewhat(especially in a vehicle) but at the end of the day, you are at the mercy of the environment.
posted by madajb at 3:33 PM on December 12, 2006

The inaccuracy (50-60 ft off) is even when not surrounded by tall buildings. It is anywhere. The other symptom is only around high buildings.
posted by D Wiz at 3:33 PM on December 12, 2006

Jimbob... You explained it well. Thanks. Do external antennas go on top of your car or can you put them on the dash?
posted by D Wiz at 3:36 PM on December 12, 2006

Take a look at the wikipedia GPS link I provided, it details the typical sources of error in a GPS signal.

With an open view of the sky, expect +/- 50 feet of accuracy. With tall buildings nearby, expect to lose the signal entirely.
posted by b1tr0t at 3:36 PM on December 12, 2006

Another question: Is there a significant accuracy (or antenna strength) difference between different models? Is it worth it to spend another $100 to get another model? Or is there no significant difference?
posted by D Wiz at 3:42 PM on December 12, 2006

You can get ~2-5 meter accuracy if your GPS is picking up WAAS augmentation.

However, from what I understand that only truly applies to North America because it uses a combination of ground stations.
posted by jkaczor at 3:48 PM on December 12, 2006

If you have a favorable constellation (most GPS units can show you the constellation as a sky map) you can get accuracy to +/- 10m or so, uncorrected. A favorable constellation has at least six SVs between 15 deg and 45 deg above the horizon spaced close to symmetrically. Some regions of the USA (famously, near Ames, Iowa) just have rotten constellations at predictable times.

If the constellation is very unfavorable owing to multipath (from buildings), occultation of the SVs (buildings or trees), or poor reception the accuracy degrades a lot, sometimes to tens or hundreds of meters.

It would help if you counted the number of SVs your GPS sees above the 15 degree elevation number. If you don't live near Ames, Iowa, and you see fewer than six SVs when you have a good view of the sky, you will probably benefit from a better antenna.

Your GPS unit might have a screen to show you the SNR of each SV, and you can compare those to see how many SVs are usable by the unit.
posted by jet_silver at 4:16 PM on December 12, 2006

Jimbob's compass explanation is the reason many handheld "hiker's" GPS units include magnetic compasses, too. When you're moving slowly and dealing in small distances, the extra info really helps. Geocaching got a lot easier when I picked up a compass-equipped unit. :)

Most of the problem here is related to the base map you're using, not the accuracy of the fix. Try this: Set a waypoint in a known location, say "the center of the driveway", and then return to that point a few hours or days later. See how far the GPS says you are from it. Or, mark a waypoint in the same place every day for a week, and see how tight the grouping is.

If the group is pretty tight but all on the wrong side of the displayed street, blame the map. If the group is more than 50 feet wide and some are on the right side of the street but some are on the wrong side too, blame the receiver.

You can also find a USGS survey marker (check the geocaching forums for info) and set your receiver right on top of it. Give it a minute to settle, then compare the displayed reading with the coordinates stamped into the brass marker. There's your accuracy figure, right there.
posted by Myself at 4:41 PM on December 12, 2006

The accuracy is not really a function of signal strength beyond a certain trivial level, and anyway because of the frequencies involved a "bigger" antenna isn't actually a "better" antenna. (GPS uses a series of frequencies ranging from about 1200 MHz to about 1800 MHz, so the ideal half-wavelength antenna is 8-12 centimeters, or something near 4 inches.)

The main reason for the inaccuracy is that GPS is a US military system, and all civilian receivers are deliberately made slightly inaccurate.

The satellites transmit a slightly erroneous signal in clear, plus a enciphered correction. Civilian units don't have the cipher key and can't use the correction information. Military units do have the cipher key, and they're accurate to something like two meters.

That was done so that civilian units couldn't be used by enemy militaries against us the way our military uses them (e.g. in the JDAM).
posted by Steven C. Den Beste at 5:53 PM on December 12, 2006 [1 favorite]

Steven, that was correct a few years ago, but is no longer the case. Selective Availability (SA) for the was turned off in 2000. Reference here.
posted by hackwolf at 6:33 PM on December 12, 2006

If you get a unit with a Sirf chip, they are dramatically more accurate.
posted by unSane at 7:53 PM on December 12, 2006

Hackwolf, they turned it back on again for the Iraq area in 2003 and the Afghanistan area in 2001.
posted by Steven C. Den Beste at 10:27 PM on December 12, 2006

You may have to recalibrate the unit. This is often labeled "Track Up". Get that started, then set it some place out in the open and wait for it to finish. This is usually between 5 and 15 minutes. Don't move it during this time.
posted by kc0dxh at 7:30 AM on December 13, 2006

unSane- The Mio C310X (the unit I am using) has Sirf.

Thanks everyone for all the great answers!
posted by D Wiz at 1:20 PM on December 13, 2006

You may have to recalibrate the unit. This is often labeled "Track Up". Get that started, then set it some place out in the open and wait for it to finish. This is usually between 5 and 15 minutes. Don't move it during this time.

'Track Up' has nothing to do with recalibrating the unit. It simply means that the map display rotates so that your current heading is 'up'. The opposite of this is 'North Up', which is like a regular map with North always being the top edge of the map display.

To 'recalibrate' as you put it, what you actually do (on a Garmin unit anyway) is hit 'New Location' and do what you describe. This informs the unit that its current satellite information is invalid and it should search and acquire the satellites all over again.
posted by unSane at 1:24 PM on December 14, 2006