Ok, How Does Gps Work?

[DataBoy]

A friend and I were discussing the relative accuracy of the GPS Systems, and came up with some interesting observations.

 

First, the GeoSynchronous orbiting Satallites CAN'T possibly be in an EXACT 3D location. I know NASA is good, but the government contracts typically allow a 50 ft radius to be considered as "Accurate" targeting. So every Satalite that our GPS units lock onto can be plus or minus 50 feet from the exact location the satalite is broadcasting.

 

Now, I am assuming that based on the location the signal is being broadcast from, that it is a simple matter of triganometry to get the 3 readings required to locate the unit in 3D Space. any additional readings are than averaged into the total positioning logic to allow for a greater resolution of the exact point in space the GPS Unit is. That foes explain some of the in accuracies we see when geocaching, and why a 12 channel unit (Garmin) may not have the accuracy of a 14 channel unit (Megellin) But I am still not convinced I have all the information correct.

 

I would love to have anyone that really knows whats happening either explain it here, or direct me to a white paper that will let me know the nuts and bolts of the technology.

 

THANKS and happy geocaching!

[+tozainamboku]

Google

[+StarBrand]

Hyper accurate clocks measure the time it takes signal to leave sat and arrive at your reciever. Then Trilateration to find location (like triangulation only 3d). In addtion, lots of radar and tracking of sats update the position information frequently. Over simplified.

 

Lots of web sites can explain it. Trimble used to have a nice powerpoint you could download.

[+Renegade Knight]

Step back a bit and see the bigger picture.

 

It's true that each GPS satalite can't be placed in an exact orbit, but they don't need to be. What they need to know though is exactly where they are relative to a defined model of the earth and that can be calculated using various high tech tricks and updated continously.

 

Because they know where they are, your GPS can then use the signal the satalites send, to know where it's at.

[+Stunod]

How Stuff Works - GPS

[+Markwell]

First, they're not GeoSynchronous orbiting satellites. Those can only orbit at a set altitude along a band surrounding the equator. The original 24 are in an orbit that allows for the possibility of 4-12 to be overhead at any given time at any spot on the planet. Of course, they've augmented them with newer satellites and increased the fleet, but the idea is still the same.

 

Simply put it's triangulation.

 

The satellite has an atomic clock, and sends out a pseudo non-repeating signal. The satellite knows where it's supposed to be. Based on the time lag between the satellite producing the signal and when the receiver receives the signal, the GPS can calculate the distance from the satellite. Your GPS doesn't have an atomic clock, but if it gets a time stamp from four or five satellites, it will correct itself.

 

In a 2D landscape if you know you're distance from two locations, you can be at one of two possible points:

 

Add a third point and you know your location:

 

But in 3D, just knowing your distance from two points gives you the intersection of two spheres, which is a circle (think of two bubbles colliding):

 

Add a third distance, and that circle now can intersect the new sphere in at most 2 points:

 

Normally, that second possible point can be discarding as being ludicrous or moving at an extreme velcocity, so the GPS can give you a location with just 3 satellites, but the fourth satellite narrows it down to only one possible point:

 

============

But what if the satellite is wrong about where IT is?

 

The answer is that there's usually more than 4 satellites overhead. The more satellites from which you receive your signal, the more accurate your location. So your GPS should still be able to give a good location in most normal situations.

 

NASA also has methods in place that they can correct the satellite's impression about where it is as well. So if it gives consistently bad data, they go in and nudge the computer to say "Hey, you're not there, you're here."

Edited May 4, 2006 by Markwell

[wmas1960]

There seems to be a lot of technical discription that will take some time for me to read. I am not that, specifically, technical minded but in simple terms, as I understand it, Please correct me if I am wrong,

 

It is basically using a navigation process called triangulation. Back when I took map reading during JROTC classes we would use highly detailed charts or maps of an area. That chart would contain landmarks like water towers, tall buildings, lakes, rivers, railroad tracks, bridges and what have you. The area where I was consisted of a lake, There was an astronomical observatory on one shore that stood out very clear. Then there were a couple points along the shore that stuck out some. Add water towers and even the schools flag pole which was very tall and considered an official landmark and you had the information that you would need. You would seek a high point or look around you and see if you can find any of these land marks. Standing in one point, you would use a compass to determine the location of each landmark. Subtracting 180 degrees from your compass to the object that you located on a map, you would then draw a line from that object in that direction. Then you would repeat the process for your second object. The point where the two lines crossed is your location.

 

With GPS it is doing the same thing but using sattelites. the satellites are launched and tracked in relation to the earths location. The GPS is a receiver is reading data from satellites as to, I guess, things like the time, orbit, location relative to the earth... Using all kinds of math, I would guess it could get quite accurate. Your receiver picks up the signals from as many of these satellites as it can and collects the data. The more satellites it receives, I think the more precise it's location can be. I think it needs, at least, 3 satellites to determine latitude, longitude and elevation. Calculating the coordiates, and I guess distance from where you are to where the satellites are, the GPS can then tell you where you are, where you are headed and track your movement. I believe, the stronger the signals and the more satellites that your receiver can continuously detect, the more precise your coordiates and tracking will be. My eTrex sometimes veries from about 70 feet down to 12, I think. Military versions are supposedly far more precise or maybe it is that the military has access to more of the satellites, allowing them more precise or complete sets of data. I seem to recall an issue some time ago where the civilian market was only given access to so many of these satellites to allow us some sort of navigation. That being why our machines are only so good. But then, we aren't trying to drop a missile down a chimney.

 

Edited to correct for what I had just read. Didn't turn out to be as complex as I had anticipated. Graphics and charts can sometimes be helpful or scary. I guess this time they were more helpful.

Edited May 4, 2006 by wmas1960

[+Sputnik 57]

In reality, the 4 (or 12 or whatever) spheres don't all intersect at a single point, so the GPS has to guess at its exact location. Manufactures put the variance into a black box and report something called "estimated position error" to give you some relative feel for how much guessing is going on at any given time.

 

From what I understand, the SirF III chip set inside some GPSrs does a better job of studying these position errors. It says to itself, "What if the signal from satellite 12 is being bent by the ionosphere by x amount? What if the signal from satellite 17 is bouncing off of a canyon wall?") It does these "what ifs" millions of times a second to come up with a better educated guess, thereby providing better accuracy.

[+Renegade Knight]

... It is basically using a navigation process called triangulation....

 

Exactly. It's just 3 Dimensional triangulation.

 

The satalites know where they are.

The GPS can calculate it's distance from the overhead satalites and use that to determin it's position.

[+StarBrand]

... It is basically using a navigation process called triangulation....

 

Exactly. It's just 3 Dimensional triangulation.

 

The satalites know where they are.

The GPS can calculate it's distance from the overhead satalites and use that to determin it's position.

 

Trilateration

 

See: http://www.google.com/search?q=define:tril...ie=utf8&oe=utf8

[+Markwell]

but the fourth satellite narrows it down to only one possible point

 

In reality, the 4 (or 12 or whatever) spheres don't all intersect at a single point, so the GPS has to guess at its exact location.

 

When I do this demonstration with full Powerpoint (which I will again in a couple of weeks for some scouts), I use the 2D example with the screen bean and then indicate that the distance from the known point is an approximation.

 

I tell them you don't get this:

 

You get this:

 

and that brown curved hexagon is the area in which you could be...

 

But for the idea behind HOW it works, the "point" was if everything were perfect and in a vacuum.

[+Sputnik 57]

Nice graphics, Markwell! It looks like a great Powerpoint presentation.

[+Woodbutcher68]

All this time I thought it was magic!

[peter]

Add a third distance, and that circle now can intersect the new sphere in at most 2 points:

 

Normally, that second possible point can be discarding as being ludicrous or moving at an extreme velcocity, so the GPS can give you a location with just 3 satellites, but the fourth satellite narrows it down to only one possible point

Just a little nit with the above. Since our receivers *don't* have atomic clocks, they don't actually know the radii of the circles you've shown, only the differences between the radii. I.e. the receiver knows that it got a signal from Sat. A precisely so many microseconds before it got a signal from Sat. B, so it can calculate how much closer it is to A than to B but it doesn't directly know the distance to either one.

 

Therefore just getting signals from 3 satellites isn't enough to narrow things down to 2 points. In your drawing think of changing the radii of all three circles by the same amount. In that case they may still intersect at two points, but not the same ones as before. The receiver would have no way of knowing which set of radii to use so with only three satellite signals there would be an infinite possible number of intersection points.

 

An additional signal from a 4th satellite is required so the receiver can solve for its lat., long., elevation, and time. Alternatively, the receiver can make the assumption that it already knows the altitude from a previous measurement (usually not a bad assumption since we don't readily change our altitude rapidly). In that case instead of using a sphere from a 4th satellite it can use the surface at the assumed elevation over the earth as the 4th surface and determine the intersection point of it with the three spheres centered on the satellites.

That's why the receiver can give a "2D" solution based on only 3 satellites but needs 4 for a "3D" solution.

Edited May 5, 2006 by peter

[+Tikibirds]

Some technical insight:

- Actually your hand held does know the distances to the satellites and the points of origination. All of the satellite positions are found using an "Almanac" (which is really math equation stuff) that is downloaded to your GPS every 12 minutes. The nature of the pseudo-random signal (which is a different signal from each satellite) is such that the exact transmision time and thus distance (at the speed of light) can be determined. The Atomic clocks are needed in the satellites as the time of transmision is what is so critical. Needing 4 satellites is really more a matter of geometry, and cross checking the different satellites against each other.

- As for the initialization assumptions.. The display we're used to of Lat/Lon and Altitude are really conversions based on an internal system of earth/centered, earth/fixed vector representations. The conventional trigonometry in doing lat/lon breaks down in polar areas and around (E/W 180 degrees). If you've ever used UTM's you'll notice the GPS switches in less than a second.

- As for 2D lock, guessing altitude does nothing to support a vehicle driving up/down a hill, or an airplane. The GPS really provides this (2D) indication as more of a user display of "I'm loosing accuracy". The entire global navigation system requires 3D + Time (which is really 4D) for anything to work. The hald held doesn't ever switch to a flat earth model. Altitude suffers first when signal is lost, especially while we are on the ground, because most of the satellites near the horizon are blocked by ground clutter. These satellites are geometrically needed for a vertical measurement (keep in mind everything is moving and vertical position is a measurement difference from acceleration due to gravity).

- As for the civil vs military signals... There are actually two separate frequencies that are transmitted from each satellite. Right now, atmospheric distortion is the largest contributor to our handheld inaccuracy. Civil systems can only use one of these frequencies. Military systems use both. By measuring the same transmission on two different frequencies, the military receivers can far more accurately determine (and correct for) the effects of atmospheric distortion. The Civil receivers have to rely on a very complicated, yet not as accuracte model. There is however a new (third) signal that was just activated last december (L2C). There is only one GPS satellite currently on orbit that supports it, (along with the two brand new WAAS satellites), but it is intended for civil use, it'll be a year or two until the hand helds will be able to support it reliably, but it promises to greatly improve civil accuracy. The current military signal (P-code) is encrypted so you need the special dept. of defense secret decoder ring to decrypt it.

- Right now, WAAS does a very nice job of measuring atmospheric distortion and sending the corrections to your handheld, so for now, use it, if you'd like improved accuracy.

[nobby.nobbs]

All this time I thought it was magic!

 

no not magic just a little imp using his demonic power to figure out where he is. hence if he's in a good mood the gps straight to the cache and other time he gets you walking in ever wideneing circles!

[+fizzymagic]

Some technical insight:

There's another thread active right now about "forum rudeness." One common reason for what is perceived as rudeness is when somebody posts information as if they are an expert, but the information is incorrect and misleading. I often respond to those posts with a rather sarcastic tone, but, in light of the other thread, I won't do that here.

 

All of the information Peter posted is correct. Much of the information Tikibirds posted is not, although some is. Separating the two is left as an exercise to the reader.

Edited May 5, 2006 by fizzymagic

[peter]

Some technical insight:

- Actually your hand held does know the distances to the satellites and the points of origination. All of the satellite positions are found using an "Almanac" (which is really math equation stuff) that is downloaded to your GPS every 12 minutes. The nature of the pseudo-random signal (which is a different signal from each satellite) is such that the exact transmision time and thus distance (at the speed of light) can be determined. The Atomic clocks are needed in the satellites as the time of transmision is what is so critical. Needing 4 satellites is really more a matter of geometry, and cross checking the different satellites against each other.

It's actually the ephemeris data broadcast by the satellites which is used for the position calculation (the almanac is the less precise data that's just used so the receiver knows which satellites are in view). But you're missing the key point that this still doesn't let the receiver know the distances directly. Sure it knows the exact time the signal was sent, but since it doesn't have its own atomic clock it doesn't know exactly when it was received. That's why GPS documentation always talks about 'pseudo-ranges' as the receiver doesn't know the real ranges until it has solved for all four variables: lat, long, elevation, and time (technically the three space variables are based on the ECEF [Earth-Centered, Earth-Fixed] coordinate system but the principle is the same). To do that absolutely does require that the receiver either get signals from four separate satellites or make some assumption about one of those variables; and the assumption that's used (in 2D mode) is that the altitude hasn't changed since the last position fix. To do any cross-checking and validity checking requires that you have *more* than four satellite signals.

- As for 2D lock, guessing altitude does nothing to support a vehicle driving up/down a hill, or an airplane. The GPS really provides this (2D) indication as more of a user display of "I'm loosing accuracy".

No, what it's "losing" is reception of the fourth satellite. With only three satellites (or sometimes even with four but a poor geometry), there is no way for the receiver to continue to calculate all three space dimensions plus time. The 2D mode works pretty well in most cases, but in situations where you actually are changing altitude rather rapidly you'll see your calculated horizontal position start to drift away from reality. The receiver is still trying to calculate your position based on the satellite signals and *assuming* a fixed altitude. If the assumption is wrong then the results of the horizontal position calculation will also be wrong.

The entire global navigation system requires 3D + Time (which is really 4D) for anything to work.

That would be ideal, but that requires always having four satellite signals so you can solve for those '4D' unknowns. The designers recognized that often only three satellite signals might be receivable. Rather than giving up altogether under those circumstances it's better to give a reasonable estimate of the horizontal position using a best guess for the altitude. That eliminates one of the variables and therefore lets you calculate the remaining three using data from only three satellite signals.

Edited May 5, 2006 by peter

[+Tikibirds]

First, I must apologize if you intrepreted my post as rude as that was not the intention. My job requires me to work with this stuff everyday so sometimes my descriptions get too technical, and sometimes too general as I try to leave out technical information. Perhaps the technical and or perhaps lack of technical information is somewhat missleading, as it appears I misintrepreted some of Peters information.

Agreeing with Peter's additional information:

Some of the key points I missed:

- there is a subtile difference between establishing an initial fix, and maintaining one, I was referring to the initial condition.

 

- Time is the most critical component of the solution. This is why 4 are required for an initial "lock-on". (The hand-held internal oscillators are usually accurate enough to "coast" for a second or two if signal is lost without any significat loss of accuracy.)

- The GPS receivers have many "tricks" that they use when signal degrades (once the initial fix is established) - freezing altitude is one - I've also seen receivers "coast" based on present speed/heading. - among many others. This area is really the only difference between vendors (Garmin vs Magellan etc..)

- I'm curious as to what parts of my description you are finding incorrect.. As I am certainly not the end all - be all of GPS experts, perhaps there's someting I'm missing or haven't explained enough...

[jamieb520]

A friend and I were discussing the relative accuracy of the GPS Systems, and came up with some interesting observations.

 

First, the GeoSynchronous orbiting Satallites CAN'T possibly be in an EXACT 3D location. I know NASA is good, but the government contracts typically allow a 50 ft radius to be considered as "Accurate" targeting. So every Satalite that our GPS units lock onto can be plus or minus 50 feet from the exact location the satalite is broadcasting.

 

Now, I am assuming that based on the location the signal is being broadcast from, that it is a simple matter of triganometry to get the 3 readings required to locate the unit in 3D Space. any additional readings are than averaged into the total positioning logic to allow for a greater resolution of the exact point in space the GPS Unit is. That foes explain some of the in accuracies we see when geocaching, and why a 12 channel unit (Garmin) may not have the accuracy of a 14 channel unit (Megellin) But I am still not convinced I have all the information correct.

 

I would love to have anyone that really knows whats happening either explain it here, or direct me to a white paper that will let me know the nuts and bolts of the technology.

 

THANKS and happy geocaching!

 

Read Geocaching for Dummies, it explains it all. The same satalites that our GPS's use are the same that the military uses for guided missles, so I'm guessing that depending on what kind and how much you want to spend on a GPS will get you closer to your mark, and depending of how clear on a view of the sky you have.

[+TEAM 360]

Here I always thought the GPSr could 'detect' the cache whenever you start to get close, by using the forward-mounted electronic 'sniffer' (like a freon detector)...for best results, you have to stand downwind of the cache to pick up its scent...

[rynd]

Will we need new GPSr to utilize the new signals (L2C)? I assume they are on a new frequency and that the current batch of receivers can't hear it.

[+Tikibirds]

Will we need new GPSr to utilize the new signals (L2C)? I assume they are on a new frequency and that the current batch of receivers can't hear it.

 

Yes, you will need a new one. I know that Trimble is selling the chips that support it, but don't have the hand-held packaging we like. I haven't seen anything yet from Garmin or Magellan. Garmin indicated they do not yet have a product for this (actualy their tech hadn't heard of it) and Magellan had no reply. I'd guess it'll be a year or so til something is available - however - the signal itself is not guaranteed yet and is only available on 1 GPS satellite right now.