How GPS Works

GPS is an acronym for Global Positioning System. It is comprised of 3 main parts those being the User, Satellite, and Control segments. The User segment is the GPS receiver at its simplest and this is the most obvious aspect of the system from the users perspective. Although the satellites are not visible we can view their location on most GPS receivers using a skyplot. The control segment comprises the least obvious component and essentially makes sure that the system is running well. These segments are explained further in the module notes and lecture set but our main aim here is to explain the process by which a position is calculated with the GPS. How GPS calculates a position can be summed up in one word triangulation. Triangulation is a process where an unknown position is determined by measuring angles and distances between known objects. In the context of GPS triangulation this can be further broken down into; 1. GPS measures distance using the travel time of microwave energy. 2. Time measurements need very accurate clocks to do this 3. Once the distance to a satellite is known you need to know where the satellite is in space to triangulate. 4. You also need to account for any timing errors to get accurate positions These key aspects of the system are explained below. Satellite Ranging or Pseudo-ranges We will cover this aspect of the process first and fill in the blanks later. Satellite ranging involves knowing the distance between ourselves and at least a single satellite. If we do this for one satellite then we can locate ourselves somewhere on a sphere centred about the satellite. If we determine the distance to a second satellite this also places us on a second sphere centred on the second satellite. The two spheres must intersect and that intersection forms a circle. This is the only location where we are the correct distance between both satellites. If we measure our distance to a third satellite and follow the same principles then Source: Trimble Navigation1989 the intersection becomes just two possible locations. One of these will be on or about the surface of the planet and the other will be off the planet or travelling at an exceedingly high speed. To unambiguously determine our position we could

measure our distance from a fourth satellite but in theory we need just three satellite to determine our position. In reality we do use four satellites to determine a position but it is not for the reason of ambiguity and will be explained later. Obviously this is a simplistic representation of the process and in reality the problem is solved with mathematics. Linear algebra allows the solution of our location on the planet expressed as a 3D Cartesian Source: Trimble Navigation1989 coordinate (x,y,z) where the following characteristics are valid. The origin (0,0,0) of the coordinate system is on the earths centre of mass and the actual location of the three satellites is known. Linear algebra allows for the solution of the location as 3 unknowns (x,y,z) according to the association between the distance from the satellites and their location in space. Measuring your Distance form a Satellite Since the principle of positioning relies on knowing how far we are from a satellite then it is critical that we have an accurate method of doing this. As the satellites are in an earth orbit of approximately 20 000km even light and radio waves take time, and luckily a measurable amount of time, to travel that distance. From high school physics we should know that velocity=change in dist Source: Trimble Navigation1989 divied by change in time. So, for a satellite, if we know the time taken for the signal to reach us, and the velocity of the signal (300 000m/s) then we can determine the distance. For a GPS satellite that is orbiting overhead a radio signal takes only 6/100ths of second to travel the 20 000km distance. How do we determine the time taken? This key to this aspect of GPS lies in what is transmitted by the satellites. Each satellite has a unique code that repeats every 1msec. The code is very special and is termed a pseudorandom code. It looks random but is not and as the signal level from the satellites are very small it pays to have a repeatable code that can be detected from all of the other radio noise that we have. If the receiver and the GPS satellite are time synchronised so that they each repeat the code at the exact same time then there will be a slight delay between when the receiver generates the code and when the same code, as a signal from the satellite, is received. This slight difference in time is how the distance between the satellite and the receiver is determined. The key to accuracy of the system is to synchronise the satellites and receivers to generate the code at exactly the same time.

Accurate Timing If timing is the key then it will have to be very good. If we are out by 1/100th of a sec the distance measurement from a satellite will be out by 3000m! So accurate timing is critical and the GPS satellites have a number of onboard atomic clocks that accurately measure time to 10-12 seconds (thats 0.000000000001s). For the atomic clock to show an error of 1 sec (drift) it would need to run for over 32 000 years. Our most accurate clocks measures to 10-15 seconds and has a drift rate of 1 sec in 30 million years. The GPS satellites each have 4 atomic clocks onboard so you can say the problem of timing is solved as long as all of the satellites are time synced together. That is, they know the correct time and they all produce a time signal and code at the same time. From the users viewpoint it is not feasible to supply each GPS receiver with one or more atomic clocks, cost and size would limit the availability of the system dramatically. In fact the cheapest atomic clock is about US$50000 for a caesium (10-12 sec) desktop version. In the GPS receiver we cope with garden variety quartz clocks which have an accuracy of approximately +/-0.5 sec per day (thats 10-6 sec). Clearly this is nowhere near as accurate as the atomic clocks of the GPS satellites and really this timing error just introduces a new variable into our position fix problem so that we now have to solve for x,y,z and t (time). To solve for this fourth unknown we just add in another measurement to a fourth satellite and consequently any GPS receiver requires a minimum of 4 satellites in view to get a position fix. Satellite Position To get the algebra to work correctly we need to know where in space the 4 satellites are. Each satellite orbits the planet at an altitude of 20 200km. At this altitude the orbit path is essentially immune to significant positional variation. The orbit path information is transmitted constantly by all satellites over a period of about 30sec as a part of the ephemeris data. Ephemeris means a table which gives the location of celestial bodies at a particular time and over a certain period. The ephemeris data is the precise locational information required for accurate position fixes and each satellite transmits information about itself and all other satellites. The satellites also transmit another set of important data called an almanac that contains the status of the satellite and the time and date. The time and to a lesser degree the date are important in our calculation of position. It would be naive to assume that once the satellites are in orbit nothing changes when in fact there are many minor changes made to and/or monitored by the system. These changes are uploaded to the satellites and transmitted back down to the receivers to maintain accurate position fixes. Errors which degrade position fixes There are a number of errors that can degrade the positional accuracy of GPS. These errors in decreasing size are atmospheric and ionospheric effects, receiver errors, ephemeris errors and clock errors.

The ionosphere forms an important part of the earths atmosphere and has a significant influence on the speed at which radiowaves travel through it. The change varies according to the angle at which the signal is passing through the atmosphere and more significantly it varies in concert with variations in the Suns energy output. Clearly this is not an easy thing to account for and adds an error of +/-5 metres to the position fix of a single handheld GPS. The GPS has a partial solution to this error which involves some heavy maths and physics. But in simple terms the GPS satellites transmit signals at different frequencies. The ionosphere will delay these frequencies by slightly different amounts, the physics part, and with some maths the amount of error can be estimated. Most modern GPS receivers incorporate some ability to account for this error. An indication of the magnitude of the errors is given in Table 1.

Table 1: Sources of User Equivalent Range Errors (UERE) Source Ionospheric effects Ephemeris errors Satellite clock errors Multipath distortion Tropospheric effects Numerical errors Effect 5 meter 2.5 meter 2 meter 1 meter 0.5 meter 1 meter

Source: Wikipedia 2008

Another error attributed to the atmosphere is caused by water vapour in the troposphere. The troposhere comprises the lower 10km portion of the earths atmosphere. We know how varied this part of the atmosphere is and it is virtually impossible to estimate the contribution of the tropospheric error and remove it. The atomic clocks on board the GPS satellites are very accurate but even they are prone to slight variations. Affects based on Einsteins theories of special and general relativity cause these errors. Time slows down when you travel really fast (-7usec/day for a GPS satellite), but being a long way from the Earths mass causes clocks to tick faster (45.9usec/day). Errors from these phenomena, if not corrected, would accumulate at a rate of 10km/day. Back at the user segment the receivers can also make mistakes in calculating the time differences in the codes. Sometimes these errors are big and

therefore obvious but the smaller biases are much harder to detect and, therefore, account for a small component of the overall error. Multipath errors where the satellite signal has not taken a direct line of sight path between the satellite and the receiver comprise another small error. Most multipath errors are due to large objects reflecting the GPS signal such that the receiver gets multiple signals or if a direct path to a satellite is blocked a further delayed signal from a satellite. Multipath errors are most common in built up areas. Geometry of the satellites The geometry of the satellites also influences the accuracy of a position fix but is significantly less of an issue with current receivers which have 12 or 16 channels, high sensitivity and a full complement of satellites in the GPS constellation to access. Conditions where geometry errors arise and become significant are when large areas of the sky are occluded by Source: Trimble Navigation1989 significant tree or vegetation cover or tall buildings. Most GPS receivers will give you an indication of how well they are finding a position based on the geometry of the satellites. The indicator can be called a number of different things by each GPS manufacturer but is technically called the Geometric Dilution of Precision or GDOP. GDOP can be further broken down into its component parts of horizontal, vertical and time and unless you Source: Trimble Navigation1989 are particularly interested in one component we generally stick with GDOP as an overall measure of the positional accuracy. GDOP is generally given as a unitless integer value ranging from 1 to about 20 and these values are explained in Table 2. You can estimate the accuracy of a position fix by multiplying the GDOP value by the average RMS value of 5 metres. Some GPS units have RMS values lower than 5 metres, and some higher than 5 metres. Magnitude of position fix errors Typical handheld GPS receivers have 12 or more channels and can track all GPS satellites that are visible in the sky. Still there are certain areas and

certain times during a 24hr period where an insufficient number or a poor constellation of satellites will result in poor position fixes. These areas and times are outlined in Fig 1. An indication of the size and extent of poor GPS constellation areas around the world given a set number of GPS satellites. Table 2 DOP values and their influence on position

DOP Value 1

Rating

Description This is the highest possible confidence level to be used for applications demanding the highest possible precision at all times At this confidence level, positional measurements are considered accurate enough to meet all but the most sensitive applications Represents a level that marks the minimum appropriate for making business decisions. Positional measurements could be used to make reliable in-route navigation suggestions to the user Positional measurements could be used for calculations, but the fix quality could still be improved. A more open view of the sky is recommended Represents a low confidence level. Positional measurements should be discarded or used only to indicate a very rough estimate of the current location At this level, measurements are inaccurate by as much as half a football field[citation needed] and should be discarded

Ideal

2-3

Excellent

4-6

Good

7-8

Moderate

9-20

Fair

21-50

Poor

Source: Wikipedia 2008

Characteristics of GPS position fixes The size, extent and location of areas on the Earth that have poor GPS constellations is shown in Figure 1 below. The satellites are continuously moving so there will always be certain areas of the earth that have poor satellite geometries and therefore poor position fixes. The cumulative time given in Figure 1 shows the areas and length of time of poor position fixes over a 24 hour period. These poor geometry periods ie times may or may not be contiguous. You should also note that as the number of satellites in the constellation is increased these areas diminish in size and time but are predominantly located in the polar regions. If a number of satellites are lost from the constellation then there will be certain time periods and areas where position fixes are poor.

Fig 1. GPS constellation number with location and extent of signal dropouts per day
(Source: Aerospace Corporation Magazine http://www.aero.org/publications/crosslink/summer2002/03.html)

A position fix from a GPS receiver is given once every second by default. Even if the user is stationary the position fix given by the GPS receiver will wander. How much the position wanders is primarily determined by the satellite constellation and ionospheric affects. Figure 2, below, shows the position variation in +/- X and Y for a stationary GPS over a 24 hour period. The red (larger extent) indicates Selective Availability (SA) on and blue (smaller extent) with SA off. We are only interested in the blue (smaller) area. Note that most position fixes cluster around the (0,0) location but there are times when the position fix is up to 50m from the true location. An instantaneous position fix from a GPS receiver will generally be within 10 metres of the known location but could be up to 50 metres out. Our PDOP gives us an indication of how good our position fix is but this is not an accurate indication in metres of the error.

Fig 2. Variation in position fix location for a static receiver over a 24hr period
(Source: Aerospace Corporation Magazine http://www.aero.org/publications/crosslink/summer2002/03.html)

Figure 3 shows a histogram plot of GPS position fixes over a 20 day period for a stationary receiver. Note that that bulk of the position fixes lie within the 0-5 metre range and the number of position fixes decrease rapidly with increasing distance. The data is highly skewed but in this case it is a good characteristic. Using statistics we can define confidence intervals to the data and give an indication of how well the GPS receiver performs. Not surprisingly there are many statistical measures that can be applied to GPS position data to indicate position fix accuracy particularly by manufacturers. In reality there will be minimal difference in the accuracy of one receiver over another given the performance of GPS modules used in most receivers today.

Fig 3. Histogram showing the distribution of horizontal position fix errors

(Source: Aerospace Corporation Magazine http://www.aero.org/publications/crosslink/summer2002/03.html)

Improving position fix accuracy There are a number of methods used to improve the accuracy of a GPS receiver. The two main types are outlined below. Averaging: If the location given by a GPS receiver wanders over time and we are only interested in stationary positions then we can log position fixes over time and average them. The longer you log the more accurate the position fix. There are no set time limits here but if your PDOP is good then a couple of minutes logging will improve the accuracy of your position fix considerably. Differential: There are many different types of differential GPS and only the simplest example is presented here, but in all cases differential GPS requires two receivers. One receiver must be on a known location and must be stationary. The second receiver can either be stationary or moving. For simplicity in this example the second receiver is stationary but the extension to a moving GPS receiver is trivial. If the first receiver is on a known location and that location is entered into the GPS receiver, then for every calculated position fix the receiver can determine the difference in x (delta x) and y (delta y) between the currently determined position and the entered known position. These values can be stored in a simple file with date and time stamps and as long as they are collected at the same or a greater sampling rate when compared with the second receiver the delta x and delta y values can be used to differentially correct the second receiver. The general assumption in differential GPS is that the deviation/error in position fixes between two spatially separated receivers will not differ by very much as long as the two receivers are seeing the same constellation of

satellites. Thus, this type of correction can be applied to receivers up to about 500km away from one another, although errors increase as the distance between the receivers increases. So, if the stationary receiver has a set of delta x and delta y measurements then these deltas can be directly applied to the second receiver to improve the accuracy of the position fix. Depending on the accuracy of the known location the improvement can reduce the error to less than 1 metre. Differential GPS can be applied in two ways, either real-time or post processed. Real time differential relies on a radio link (or telephone or satellite link) between the two receivers such that the delta x and delta y corrections plus a time stamp are transmitted to the second receiver instantly. The second receiver uses these delta measurements to correct the positions it calculates and stores the corrected data in a file. Post processing requires that both receivers collect and store the calculated positions in files during the collection process. The files from both receivers are downloaded to a computer and computer software performs the processing of the data to create a differentially corrected file of position fixes for the second receiver.