GPS performance variations

[kgokal]

So a co-worker had a good question regarding GPS performance.
And I took an educated guess to the reasons.

The comparison was between dash-mount GPS's vs Cell-phone gps.
Why one acquires signals so much quicker then cellphone versions.
He was wondering if a GPS chip is just a chip, how can can its performance vary.

My guess was antenna size and processing power available to chip.
Are those the only variables that would affect performance under the same conditions.

 

[TuxDave]

It could be algorithmic in why one converges faster. For the car case, when the GPS turns on, the last known position is most likely going to still be the current position. And while the GPS starts to converge to "double check" it can mimic a GPS as you start driving since the direction and speed are known from the car's computer so it can intelligently guess where your new position is until it gets a lock.

Some of the newer hand held GPS systems converge very quickly because as opposed to starting with a blank slate and searching for any and all signals, some keep an internal table (with the assumption that you haven't travelled very far from where you were last) to track what signals it should expect to find to converge faster.

 

[KIAman]

Also, cellphone GPS use cell towers for triangulation and can lose signal while dash mounted uses the GPS satellites.

 

[Aberforth]

It depends on the antenna positioning, weather and the GPS protocol. The standard protocol is NMEA- which I prefer, but garmin devices use their own proprietary translation - so this might add up more processing time.

 

[CycloWizard]

I have a couple of related questions while we're on the subject...

What information do the satellites actually transmit? Naively, I would expect them to transmit latitude, longitude, and elevation.
Are the satellites geostationary? It would certainly simplify things if they were, but the system could work if the satellites were moving.

 

[Aberforth]

Most satellites are geostationary unless they are designed for a specific purpose such as mineralogy mapping, spying, cartography etc. A GPS receiver locates 3-4 satellites to deduce the exact position of the ground source and it's altitude by a method called trilateration. This is what a GPS Processor does but there are different translation methods. NMEA is very popular, it has latitude, longitude and altitude. If your device supports serial port, IrDA interface you could write a simple program to log the gps data flow.

NMEA format:
$GPGGA,205541.797,4305.1766,N,00603.3850,E,0,0,,101.0,M,-
49.0,M,,*4F
$GPGSA,A,1,,,,,,,,,,,,,,,*1E
$GPGSV,1,1,01,02,,,15*7E
$GPRMC,205541.797,V,4305.1766,N,00603.3850,E,0.00,114.92-
,041208,,,N*77
$GPVTG,114.92,T,,M,0.00,N,0.00,K,N*3D
$GPGGA,205541.997,4305.1766,N,00603.3850,E,0,0,,101.0,M,-
49.0,M,,*41
$GPGSA,A,1,,,,,,,,,,,,,,,*1E
$GPGSV,1,1,01,02,,,15*7E
$GPRMC,205541.997,V,4305.1766,N,00603.3850,E,0.00,114.92-
,041208,,,N*79
$GPVTG,114.92,T,,M,0.00,N,0.00,K,N*3D
$GPGGA,205542.197,4305.1766,N,00603.3850,E,0,0,,101.0,M,-
49.0,M,,*4A
$GPGSA,A,1,,,,,,,,,,,,,,,*1E
$GPGSV,1,1,01,02,,,15*7E
$GPRMC,205542.197,V,4305.1766,N,00603.3850,E,0.00,114.92-
,041208,,,N*72
$GPVTG,114.92,T,,M,0.00,N,0.00,K,N*3D

 

[Biftheunderstudy]

Also, importantly time. Actually, this is probably the most important. To get sub-metre resolution you need to account for a host of things including Special and General Relativity as well as atmospheric effects.

 

[bobsmith1492]

GPS satellites transmit a signal at a particular frequency with a particular phase.

A GPS receiver chip compares the phase shift between the different satellites; the difference in phase, given the location of the satellites, shows the position of the receiver device.

A receiver chip with a better phase discriminator will deduce your location more accurately; slight variations in the device manufacturing or differences between manufacturers could account for differences in convergence time and accuracy.

 

[Texun]

Please pardon the interruption but I've been curious about GPS accuracy and precision and have a question:

What is the highest level of accuracy of the GPS system? I know some consumer units claim to be accurate to within ~3 meters but what about commercial units? I have a friend who works for a utility company and he says they have GPS units that are accurate to within one centimeter. Does this sound possible?

I think I read somewhere that the government placed a timing offset in consumer units to prevent them from being used for commercial use as well as flight and terrorist attacks, although I'm not sure what difference 3 meters of 3 feet would make to a terrorist.

 

[Aberforth]

Originally posted by: Texun
Please pardon the interruption but I've been curious about GPS accuracy and precision and have a question:

What is the highest level of accuracy of the GPS system? I know some consumer units claim to be accurate to within ~3 meters but what about commercial units? I have a friend who works for a utility company and he says they have GPS units that are accurate to within one centimeter. Does this sound possible?

I think I read somewhere that the government placed a timing offset in consumer units to prevent them from being used for commercial use as well as flight and terrorist attacks, although I'm not sure what difference 3 meters of 3 feet would make to a terrorist.

The accuracy depends on the signal response time, errors range from 3 to 20 meters but most of the time there is 1 to 10cm errors- weather highly affects the signal strength like moisture,wind,rain and snow. Armed forces have their own sophisticated navigation system ( http://en.wikipedia.org/wiki/D..._Advanced_GPS_Receiver )

I think civilian GPS devices won't work in commercial airplanes, they will have a hard time finding satellites.

 

[Mark R]

Originally posted by: CycloWizard

What information do the satellites actually transmit? Naively, I would expect them to transmit latitude, longitude, and elevation.
Are the satellites geostationary? It would certainly simplify things if they were, but the system could work if the satellites were moving.

1. The satellites transmit a high precision timing signal. The signals are all transmitted on the same frequency but each satellite uses 1 of 36 orthogonal PRN spread spectrum encodings.

Embedded in the timing signal is the satellite's 'ephemeris' - a precise description of its orbit; angle of inclination, right ascension, velocities, drifts, etc. The encoding rate is very low, something like 50 baud, so it takes 15 minutes to download a full almanac (each satellite transmits ephemeris for all the other satellites so that as soon as a new satellite comes into view, it can be used). Ground stations track the satellites, and periodically upload corrected ephemeris.

From this, the receiver is able to calculate the satellite's position at the time it transmitted the signal. By receiving multiple signals from 4 or more satellites, the recevier can then calculate the 'partial ranges' to each satellite (it can't calculate the actual range, as the receiver doesn't have it's own reference time source), from which the receiver can triangulate its position.

The satellites are medium earth orbit, and orbit with a 12 hour orbital period - they use heavily inclined polar orbits - I think there are 6 orbits, 60 degrees of longitude apart, each with 4 satellites. The result is that the satellites are clustered over temperate lattitudes. In Arctic/Antarctic lattitudes, there is very poor view of the satellites, giving very erratic GPS performance.

In addition to the MEO satellites, there are several geostationary satellites - 2 over the US, 2 over Europe and 2 over Asia. These form 3 seperate 'Satellite based augmentation systems' (SBAS) - the US SBAS, is sometimes called WAAS. These satellites transmit a modified signal - they transmit an atmospheric correction signal. There is a source of error in the ionosphere, as the speed of propogation of the signal may vary due to electrical effects. Ground stations monitor this due to errors in location of fixed receivers, and they calculate a correction factor, which is retransmitted by the geostationary satellites.

Getting a satellite 'lock' is a huge problem for a receiver. The satellite signals are very weak at the receiver - something like -166 dBmW - which is way below the noise floor. As a result the receiver must perform extensive signal averaging, of the repetitive timing signal, using a Kalman filter in order to even verify the existence of a singal. In order to find the signal, the receiver must know the correct PRN, PRN phase, and the Doppler shift on the satellite signal (both carrier and data rate shifts), and then run the filter for each combination until it scores a hit.

There are various optimisations that can be done, like estimating Doppler shifts, and only searching for satellites that are likely to be in view based upon cached ephemeris data. However, top-end GPS chips do it with brute force and ignorance - by simply having a signal processor with 250,000 digital filter cores, each searching on a different parameter combination. Such high-end chipsets can typically regain lock within about 10 seconds if they have a reasonably up-to-date ephemeris, or under a minute, if they are cold booted.

GPS enabled cell phones have other sources to accelerate locks. E.g. they can download ephemeris data over high-bandwidth terrestrial links, avoiding the need for a 15 minute download period, or they can supplement satellite data by measuring partial ranges to nearby base stations. A lot of cell phones with integrated GPS use this approach as it's cheaper than using a DSP with 250k correlator cores.

 

[Mark R]

Originally posted by: Texun
Please pardon the interruption but I've been curious about GPS accuracy and precision and have a question:

What is the highest level of accuracy of the GPS system? I know some consumer units claim to be accurate to within ~3 meters but what about commercial units? I have a friend who works for a utility company and he says they have GPS units that are accurate to within one centimeter. Does this sound possible?

I think I read somewhere that the government placed a timing offset in consumer units to prevent them from being used for commercial use as well as flight and terrorist attacks, although I'm not sure what difference 3 meters of 3 feet would make to a terrorist.

GPS precision is stated to be between about 3-20 meters under normal circumstances for a civilian receiver.

The military use a second signal for higher precision location - this is transmitted on a different frequency, and has a higher data rate, although the contents of the signals are encrypted. Different frequencies experience different propogation delays through the upper atmosphere, and so if you measure the 2 signals transmitted at different frequencies, you can solve the simultaneous equations and obtain the partial range without atmospheric effects. This is the main difference between civilian and military receivers - and is supposed to improve precision to 1-5 meters. Where cost is of limited object, there are other incremental improvements that could be made - e.g. upgrading the receiver's clock to an atomic clock, as this can allow you to calculate true ranges, rather than partial ranges - and serves as an additional data point from which to solve the location problem.

There are special units available for precision work - however, they work in a slightly different way. By having a reference unit at a fixed position, and by measuring additional parameters of the satellite signal (the commonest way is to measure the phase of the carrier signal, with respect to a different receiver) it is possible to extract additional data from the 2 GPS signals (although the carrier phase of the military signal can be measured, the signal can't be decoded - so it can be used for correction, given a reference receiver, even though it can't give you a location on its own), which when combined with long averaging (e.g. 10 minutes) then you can get precision in the centimeter range. However, this requires more sophisticated receivers with highly stable time sources and a method of communication between the multiple receivers (or simply recording the raw data from all the receivers, and reconstructing it back at base once the reference data is available).

The reason that civilian receivers don't work in aircraft, is simply one of signal strength - the GPS signal is very weak, and the military signal at least 6 dB weaker. The metal in the fuselage will attenuate the signal to the point that it is unuable. The aircraft's navigation unit will need an external antenna, or a hand portable unit with the antenna placed near the largest possible window.

As for the time offset - when GPS was first opened for civilian use, there were security concerns. As a result the satellites deliberately corrupted the civilian timing signal by changing the offsets periodically. This was called 'selective availability'. A military receiver with the correct key, would be know the offsets and correct them - but a civilian receiver would miscalculate its location. With SA on, the error was typically 50-100 meters. SA was switched off permanently by presidential order in about 2000 - it is reported that the newest satellites no longer include the SA function, so it could not be switched back on again, even if ordered.

 

[Texun]

Originally posted by: Mark R

Originally posted by: Texun
Please pardon the interruption but I've been curious about GPS accuracy and precision and have a question:

What is the highest level of accuracy of the GPS system? I know some consumer units claim to be accurate to within ~3 meters but what about commercial units? I have a friend who works for a utility company and he says they have GPS units that are accurate to within one centimeter. Does this sound possible?

I think I read somewhere that the government placed a timing offset in consumer units to prevent them from being used for commercial use as well as flight and terrorist attacks, although I'm not sure what difference 3 meters of 3 feet would make to a terrorist.

GPS precision is stated to be between about 3-20 meters under normal circumstances for a civilian receiver.

The military use a second signal for higher precision location - this is transmitted on a different frequency, and has a higher data rate, although the contents of the signals are encrypted. Different frequencies experience different propogation delays through the upper atmosphere, and so if you measure the 2 signals transmitted at different frequencies, you can solve the simultaneous equations and obtain the partial range without atmospheric effects. This is the main difference between civilian and military receivers - and is supposed to improve precision to 1-5 meters. Where cost is of limited object, there are other incremental improvements that could be made - e.g. upgrading the receiver's clock to an atomic clock, as this can allow you to calculate true ranges, rather than partial ranges - and serves as an additional data point from which to solve the location problem.

There are special units available for precision work - however, they work in a slightly different way. By having a reference unit at a fixed position, and by measuring additional parameters of the satellite signal (the commonest way is to measure the phase of the carrier signal, with respect to a different receiver) it is possible to extract additional data from the 2 GPS signals (although the carrier phase of the military signal can be measured, the signal can't be decoded - so it can be used for correction, given a reference receiver, even though it can't give you a location on its own), which when combined with long averaging (e.g. 10 minutes) then you can get precision in the centimeter range. However, this requires more sophisticated receivers with highly stable time sources and a method of communication between the multiple receivers (or simply recording the raw data from all the receivers, and reconstructing it back at base once the reference data is available).

The reason that civilian receivers don't work in aircraft, is simply one of signal strength - the GPS signal is very weak, and the military signal at least 6 dB weaker. The metal in the fuselage will attenuate the signal to the point that it is unuable. The aircraft's navigation unit will need an external antenna, or a hand portable unit with the antenna placed near the largest possible window.

As for the time offset - when GPS was first opened for civilian use, there were security concerns. As a result the satellites deliberately corrupted the civilian timing signal by changing the offsets periodically. This was called 'selective availability'. A military receiver with the correct key, would be know the offsets and correct them - but a civilian receiver would miscalculate its location. With SA on, the error was typically 50-100 meters. SA was switched off permanently by presidential order in about 2000 - it is reported that the newest satellites no longer include the SA function, so it could not be switched back on again, even if ordered.

That was an excellent response and the best one I've seen. I did knot know the technical details, but I knew timing offsets and atmospheric conditions played a large role in the accuracy of the system, and therefore assumed the military must be using a technology that is more robust or advanced than civilian units.

Much appreciated!

 

[Sukhoi]

The reason that civilian receivers don't work in aircraft

Not exactly true. I have an old Garmin i2, and it works great on planes. Twice I've had my GPS with me when I've had window seats, and I just suction it to the window. Pretty funny to watch it telling me my ground speed is 550 MPH.