1
Basics of the GPS Technique: Observation Equations§
Geoffrey Blewitt
Department of Geomatics, University of Newcastle
Newcastle upon Tyne, NE1 7RU, United Kingdom
geoffrey.blewitt@ncl.ac.uk
Table of Contents
1. INTRODUCTION.....................................................................................................................................................2
2. GPS DESCRIPTION................................................................................................................................................2
2.1 THE BASIC IDEA ........................................................................................................................................................2
2.2 THE GPS SEGMENTS..................................................................................................................................................3
2.3 THE GPS SIGNALS .....................................................................................................................................................6
3. THE PSEUDORANGE OBSERVABLE ................................................................................................................8
3.1 CODE GENERATION....................................................................................................................................................9
3.2 AUTOCORRELATION TECHNIQUE .............................................................................................................................12
3.3 PSEUDORANGE OBSERVATION EQUATIONS..............................................................................................................13
4. POINT POSITIONING USING PSEUDORANGE .............................................................................................15
4.1 LEAST SQUARES ESTIMATION ..................................................................................................................................15
4.2 ERROR COMPUTATION.............................................................................................................................................18
5. THE CARRIER PHASE OBSERVABLE ............................................................................................................22
5.1 CONCEPTS................................................................................................................................................................22
5.2 CARRIER PHASE OBSERVATION MODEL...................................................................................................................27
5.3 DIFFERENCING TECHNIQUES....................................................................................................................................32
6. RELATIVE POSITIONING USING CARRIER PHASE...................................................................................36
6.1 SELECTION OF OBSERVATIONS.................................................................................................................................36
6.2 BASELINE SOLUTION USING DOUBLE DIFFERENCES .................................................................................................39
6.3 STOCHASTIC MODEL................................................................................................................................................42
7. INTRODUCING HIGH PRECISION GPS GEODESY......................................................................................44
7.1 HIGH PRECISION SOFTWARE ....................................................................................................................................44
7.2 SOURCES OF DATA AND INFORMATION ....................................................................................................................45
8. CONCLUSIONS .....................................................................................................................................................46
§ Copyright ? 1997 by the author. All rights reserved.
Appears in the textbook “Geodetic Applications of GPS,” published by the Swedish Land Survey.
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
2
1. INTRODUCTION
The purpose of this paper is to introduce the principles of GPS theory, and to provide a
background for more advanced material. With that in mind, some of the theoretical treatment
has been simplified to provide a starting point for a mathematically literate user of GPS who
wishes to understand how GPS works, and to get a basic grasp of GPS theory and
terminology. It is therefore not intended to serve as a reference for experienced researchers;
however, my hope is that it might also prove interesting to the more advanced reader, who
might appreciate some “easy reading” of a familiar story in a relatively short text (and no
doubt, from a slightly different angle).
2. GPS DESCRIPTION
In this section we introduce the basic idea behind GPS, and provide some facts and statistics
to describe various aspects of the Global Positionining System.
2.1 THE BASIC IDEA
GPS positioning is based on trilateration, which is the method of determining position by
measuring distances to points at known coordinates. At a minimum, trilateration requires 3
ranges to 3 known points. GPS point positioning, on the other hand, requires 4
“pseudoranges” to 4 satellites.
This raises two questions: (a) “What are pseudoranges?”, and (b) “How do we know the
position of the satellites?” Without getting into too much detail at this point, we address the
second question first.
2.1.1 How do we know position of satellites?
A signal is transmitted from each satellite in the direction of the Earth. This signal is encoded
with the “Navigation Message,” which can be read by the user’s GPS receivers. The
Navigation Message includes orbit parameters (often called the “broadcast ephemeris”), from
which the receiver can compute satellite coordinates (X,Y,Z). These are Cartesian coordinates
in a geocentric system, known as WGS-84, which has its origin at the Earth centre of mass, Z
axis pointing towards the North Pole, X pointing towards the Prime Meridian (which crosses
Greenwich), and Y at right angles to X and Z to form a right-handed orthogonal coordinate
system. The algorithm which transforms the orbit parameters into WGS-84 satellite
coordinates at any specified time is called the “Ephemeris Algorithm,” which is defined in
GPS textbooks [e.g., Leick, 1991]. We discuss the Navigation Message in more detail later
on. For now, we move on to “pseudoranges.”
2.1.2 What are pseudoranges?
Time that the signal is transmitted from the satellite is encoded on the signal, using the time
according to an atomic clock onboard the satellite. Time of signal reception is recorded by
receiver using an atomic clock. A receiver measures difference in these times:
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
3
pseudorange = (time difference) × (speed of light)
Note that pseudorange is almost like range, except that it includes clock errors because the
receiver clocks are far from perfect. How do we correct for clock errors?
2.1.3 How do we correct for clock errors?
Satellite clock error is given in Navigation Message, in the form of a polynomial. The
unknown receiver clock error can be estimated by the user along with unknown station
coordinates. There are 4 unknowns; hence we need a minimum of 4 pseudorange
measurements.
2.2 THE GPS SEGMENTS
There are four GPS segments:
? the Space Segment, which includes the constellation of GPS satellites, which
transmit the signals to the user;
? the Control Segment, which is responsible for the monitoring and operation of the
Space Segment,
? the User Segment, which includes user hardware and processing software for
positioning, navigation, and timing applications;
? the Ground Segment, which includes civilian tracking networks that provide the
User Segment with reference control, precise ephemerides, and real time services
(DGPS) which mitigate the effects of “selective availability” (a topic to be
discussed later).
Before getting into the details of the GPS signal, observation models, and position
computations, we first provide more information on the Space Segment and the Control
Segment.
2.2.1 Orbit Design
The satellite constellation is designed to have at least 4 satellites in view anywhere, anytime,
to a user on the ground. For this purpose, there are nominally 24 GPS satellites distributed in
6 orbital planes. So that we may discuss the orbit design and the implications of that design,
we must digress for a short while to explain the geometry of the GPS constellation.
According to Kepler’s laws of orbital motion, each orbit takes the approximate shape of an
ellipse, with the Earth’s centre of mass at the focus of the ellipse. For a GPS orbit, the
eccentricity of the ellipse is so small (0.02) that it is almost circular. The semi-major axis
(largest radius) of the ellipse is approximately 26,600 km, or approximately 4 Earth radii.
The 6 orbital planes rise over the equator at an inclination angle of 55o to the equator. The
point at which they rise from the Southern to Northern Hemisphere across the equator is
called the “Right Ascension of the ascending node”. Since the orbital planes are evenly
distributed, the angle between the six ascending nodes is 60o.
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
4
Each orbital plane nominally contains 4 satellites, which are generally not spaced evenly
around the ellipse. Therefore, the angle of the satellite within its own orbital plane, the “true
anomaly”, is only approximately spaced by 90o. The true anomaly is measured from the point
of closest approach to the Earth (the perigee). (We note here that there are other types of
“anomaly” in GPS terminology, which are angles that are useful for calculating the satellite
coordinates within its orbital plane). Note that instead of specifying the satellite’s anomaly at
every relevant time, we could equivalently specify the time that the satellite had passed
perigee, and then compute the satellites future position based on the known laws of motion of
the satellite around an ellipse.
Finally, the argument of perigee is the angle between the equator and perigee. Since the orbit
is nearly circular, this orbital parameter is not well defined, and alternative parameterisation
schemes are often used.
Taken together (the eccentricity, semi-major axis, inclination, Right Ascension of the
ascending node, the time of perigee passing, and the argument of perigee), these six
parameters define the satellite orbit. These parameters are known as Keplerian elements.
Given the Keplerian elements and the current time, it is possible to calculate the coordinates
of the satellite.
GPS satellites do not move in perfect ellipses, so additional parameters are necessary.
Nevertheless, GPS does use Kepler’s laws to its advantage, and the orbits are described in
parameters which are Keplerian in appearance. Additional parameters must be added to
account for non-Keplerian behaviour. Even this set of parameters has to be updated by the
Control Segment every hour for them to remain sufficiently valid.
2.2.2 Orbit design consequences
Several consequences of the orbit design can be deduced from the above orbital parameters,
and Kepler’s laws of motion. First of all, the satellite speed can be easily calculated to be
approximately 4 km/s relative to Earth’s centre. All the GPS satellites orbits are prograde,
which means the satellites move in the direction of Earth’s rotation. Therefore, the relative
motion between the satellite and a user on the ground must be less than 4 km/s. Typical
values around 1 km/s can be expected for the relative speed along the line of sight (range
rate).
The second consequence is the phenomena of “repeating ground tracks” every day. It is
straightforward to calculate the time it takes for the satellite to complete one orbital
revolution. The orbital period is approximately T = 11 hr 58 min. Therefore a GPS satellite
completes 2 revolutions in 23 hr 56 min. This is intentional, since it equals the sidereal day,
which is the time it takes for the Earth to rotate 360o. (Note that the solar day of 24 hr is not
360o, because during the day, the position of the Sun in the sky has changed by 1/365.25 of a
day, or 4 min, due to the Earth’s orbit around the Sun).
Therefore, every day (minus 4 minutes), the satellite appears over the same geographical
location on the Earth’s surface. The “ground track” is the locus of points on the Earth’s
surface that is traced out by a line connecting the satellite to the centre of the Earth. The
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
5
ground track is said to repeat. From the user’s point of view, the same satellite appears in the
same direction in the sky every day minus 4 minutes. Likewise, the “sky tracks” repeat. In
general, we can say that the entire satellite geometry repeats every sidereal day (from the point
of view of a ground user).
As a corollary, any errors correlated with satellite geometry will repeat from one day to the
next. An example of an error tied to satellite geometry is “multipath,” which is due to the
antenna also sensing signals from the satellite which reflect and refract from nearby objects.
In fact, it can be verified that, because of multipath, observation residuals do have a pattern
that repeats every sidereal day. As a consequence, such errors will not significantly affect the
precision, or repeatability, of coordinates estimated each day. However, the accuracy can be
significantly worse than the apparent precision for this reason.
Another consequence of this is that the same subset of the 24 satellites will be observed every
day by someone at a fixed geographical location. Generally, not all 24 satellites will be seen
by a user at a fixed location. This is one reason why there needs to be a global distribution of
receivers around the globe to be sure that every satellite is tracked sufficiently well.
We now turn our attention to the consequences of the inclination angle of 55o. Note that a
satellite with an inclination angle of 90o would orbit directly over the poles. Any other
inclination angle would result in the satellite never passing over the poles. From the user’s
point of view, the satellite’s sky track would never cross over the position of the celestial pole
in the sky. In fact, there would be a “hole” in the sky around the celestial pole where the
satellite could never pass. For a satellite constellation with an inclination angle of 55o, there
would therefore be a circle of radius at least 35o around the celestial pole, through which the
sky tracks would never cross. Another way of looking at this, is that a satellite can never rise
more than 55o elevation above the celestial equator.
This has a big effect on the satellite geometry as viewed from different latitudes. An observer
at the pole would never see a GPS satellite rise above 55o elevation. Most of the satellites
would hover close to the horizon. Therefore vertical positioning is slightly degraded near the
poles. An observer at the equator would see some of the satellites passing overhead, but
would tend to deviate from away from points on the horizon directly to the north and south.
Due to a combination of Earth rotation, and the fact that the GPS satellites are moving faster
than the Earth rotates, the satellites actually appear to move approximately north-south or
south-north to an oberver at the equator, with very little east-west motion. The north
component of relative positions are therefore better determined than the east component the
closer the observer is to the equator. An observer at mid-latitudes in the Northern Hemisphere
would see satellites anywhere in the sky to the south, but there would be a large void towards
the north. This has consequences for site selection, where a good view is desirable to the
south, and the view to the north is less critical. For example, one might want to select a site in
the Northern Hemisphere which is on a south-facing slope (and visa versa for an observer in
the Southern Hemisphere).
2.2.3 Satellite Hardware
There are nominally 24 GPS satellites, but this number can vary within a few satellites at any
given time, due to old satellites being decommissioned, and new satellites being launched to
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
6
replace them. All the prototype satellites, known as Block I, have been decommissioned.
Between 1989 and 1994, 24 Block II (1989-1994) were placed in orbit. From 1995 onwards,
these have started to be replaced by a new design known as Block IIR. The nominal
specifications of the GPS satellites are as follows:
? Life goal: 7.5 years
? Mass: ~1 tonne (Block IIR: ~2 tonnes)
? Size: 5 metres
? Power: solar panels 7.5 m2 + Ni-Cd batteries
? Atomic clocks: 2 rubidium and 2 cesium
The orientation of the satellites is always changing, such that the solar panels face the sun, and
the antennas face the centre of the Earth. Signals are transmitted and received by the satellite
using microwaves. Signals are transmitted to the User Segment at frequencies L1 = 1575.42
MHz, and L2 = 1227.60 MHz. We discuss the signals in further detail later on. Signals are
received from the Control Segment at frequency 1783.74 Mhz. The flow of information is a
follows: the satellites transmit L1 and L2 signals to the user, which are encoded with
information on their clock times and their positions. The Control Segment then tracks these
signals using receivers at special monitoring stations. This information is used to improve the
satellite positions and predict where the satellites will be in the near future. This orbit
information is then uplinked at 1783.74 Mhz to the GPS satellites, which in turn transmit this
new information down to the users, and so on. The orbit information on board the satellite is
updated every hour.
2.2.4 The Control Segment
The Control Segment, run by the US Air Force, is responsible for operating GPS. The main
Control Centre is at Falcon Air Force Base, Colorado Springs, USA. Several ground stations
monitor the satellites L1 and L2 signals, and assess the “health” of the satellites. As outlined
previously, the Control Segment then uses these signals to estimate and predict the satellite
orbits and clock errors, and this information is uploaded to the satellites. In addition, the
Control Segment can control the satellites; for example, the satellites can be maneuvered into
a different orbit when necessary. This might be done to optimise satellite geometry when a
new satellite is launched, or when an old satellite fails. It is also done to keep the satellites to
within a certain tolerance of their nominal orbital parameters (e.g., the semi-major axis may
need adjustment from time to time). As another example, the Control Segment might switch
between the several on-board clocks available, should the current clock appear to be
malfunctioning.
2.3 THE GPS SIGNALS
We now briefly summarise the characteristics of the GPS signals, the types of information that
is digitally encoded on the signals, and how the U.S. Department of Defense implements
denial of accuracy to civilian users. Further details on how the codes are constructed will be
presented in Section 3.
2.3.1 Signal Description
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
7
The signals from a GPS satellite are fundamentally driven by an atomic clocks (usually
cesium, which has the best long-term stability). The fundamental frequency is 10.23 Mhz.
Two carrier signals, which can be thought of as sine waves, are created from this signal by
multiplying the frequency by 154 for the L1 channel (frequency = 1575.42 Mhz; wavelength =
19.0 cm), and 120 for the L2 channel (frequency = 1227.60 Mhz; wavelength = 24.4 cm). The
reason for the second signal is for self-calibration of the delay of the signal in the Earth’s
ionosphere.
Information is encoded in the form of binary bits on the carrier signals by a process known as
phase modulation. (This is to be compared with signals from radio stations, which are
typically encoded using either frequency modulation, FM, or amplitude modulation, AM).
The binary digits 0 and 1 are actually represented by multiplying the electrical signals by
either +1 or ?1, which is equivalent to leaving the signal unchanged, or flipping the phase of
the signal by 180 o. We come back later to the meaning of phase and the generation of the
binary code.
There are three types of code on the carrier signals:
? The C/A code
? The P code
? The Navigation Message
The C/A (“course acquisition”) code can be found on the L1 channel. As will be described
later, this is a code sequence which repeats every 1 ms. It is a pseudo-random code, which
appears to be random, but is in fact generated by a known algorithm. The carrier can transmit
the C/A code at 1.023 Mbps (million bits per second). The “chip length”, or physical distance
between binary transitions (between digits +1 and ?1), is 293 metres. The basic information
that the C/A code contains is the time according to the satellite clock when the signal was
transmitted (with an ambiguity of 1 ms, which is easily resolved, since this corresponds to 293
km). Each satellite has a different C/A code, so that they can be uniquely identified.
The P (“precise”) code is identical on both the L1 and L2 channel. Whereas C/A is a courser
code appropriate for initially locking onto the signal, the P code is better for more precise
positioning. The P code repeats every 267 days. In practice, this code is divided into 7 day
segments; each weekly segment is designated a “PRN” number, and is designated to one of
the GPS satellites. The carrier can transmit the P code at 10.23 Mbps, with a chip length of
29.3 metres. Again, the basic information is the satellite clock time or transmission, which is
identical to the C/A information, except that it has ten times the resolution. Unlike the C/A
code, the P code can be encrypted by a process known as “anti-spoofing” , or “A/S” (see
below).
The Navigation Message can be found on the L1 channel, being transmitted at a very slow rate
of 50 bps. It is a 1500 bit sequence, and therefore takes 30 seconds to transmit. The
Navigation Message includes information on the Broadcast Ephemeris (satellite orbital
parameters), satellite clock corrections, almanac data (a crude ephemeris for all satellites),
ionosphere information, and satellite health status.
2.3.2 Denial of Accuracy
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
8
The U.S. Department of Defense implements two types of denial of accuracy to civilian
users: Selective Availability (S/A), and Anti-Spoofing (A/S). S/A can be thought of as
intentional errors imposed on the GPS signal. A/S can be thought of as encryption of the P
code.
There are two types of S/A: epsilon, and dither. Under conditions of S/A, the user should be
able to count on the position error not being any worse than 100 metres. Most of the time, the
induced position errors do not exceed 50 metres.
Epsilon is implemented by including errors in the satellite orbit encoded in the Navigation
Message. Apparently, this is an option not used, according to daily comparisons made
between the real-time broadcast orbits, and precise orbits generated after the fact, by the
International GPS Service for Geodynamics (IGS). For precise geodetic work, precise orbits
are recommended in any case, and therefore epsilon would have minimal impact on precise
users. It would, however, directly impact single receiver, low-precision users. Even then, the
effects can be mitigated to some extent by using technology known as “differential GPS”,
where errors in the GPS signal are computed at a reference station at known coordinates, and
are transmitted to the user who has appropriate radio receiving equipment.
Dither is intentional rapid variation in the satellite clock frequency (10.23 MHz). Dither,
therefore, looks exactly like a satellite clock error, and therefore maps directly into
pseudorange errors. Dither is switched on at the moment (1997), but recent U.S. policy
statements indicate that it may be phased out within the next decade. As is the case for
epsilon, dither can be mitigated using differential GPS. The high precision user is minimally
effected by S/A, since relative positioning techniques effectively eliminate satellite clock error
(as we shall see later).
Anti-Spoofing (A/S) is encryption of the P-code. The main purpose of A/S is prevent “the
enemy” from imitating a GPS signal, and therefore it is unlikely to be switched off in the
foreseeable future. A/S does not pose a signficant problem to the precise user, since precise
GPS techniques rely on measuring the phase of the carrier signal itself, rather than the
pseudoranges derived from the P code. However, the pseudoranges are very useful for various
algorithms, particularly in the rapid position fixes required by moving vehicles and kinematic
surveys. Modern geodetic receivers can, in any case, form 2 precise pseudorange observables
on the L1 and L2 channels, even if A/S is switched on. (We briefly touch on how this is done
in the next section). As a consequence of not having full access to the P code, the phase
noise on measuring the L2 carrier phase can be increased from the level of 1 mm to the level
of 1 cm for some types of receivers. This has negligible impact on long sessions for static
positioning, but can have noticeable effect on short sessions, or on kinematic positioning.
Larger degradation in the signal can be expected at low elevations (up to 2 cm) where signal
strength is at a minimum.
3. THE PSEUDORANGE OBSERVABLE
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
9
In this section, we go deeper into the description of the pseudorange observable, an give some
details on how the codes are generated. We develop a model of the pseudorange observation,
and then use this model to derive a least-squares estimator for positioning. We discuss formal
errors in position, and the notion of “Dilution of Precision”, which can be used to assess the
effect of satellite geometry on positioning precision.
3.1 CODE GENERATION
It helps to understand the pseudorange measurement if we first take a look at the actual
generation of the codes. The carrier signal is multiplied by a series of either +1 or -1, which
are seperated by the chip length (293 m for C/A code, and 29.3 m for P code). This series of
+1 and -1 multipliers can be interpreted as a stream of binary digits (0 and 1).
How is this stream of binary digits decided? They are determined by an algorithm, known as
a linear feedback register. To understand a linear feedback register, we must first introduce the
XOR binary function.
3.1.1 XOR: The “Exclusive OR” Binary Function
A binary function takes two input binary digits, and outputs one binary digit (0 or 1). More
familiar binary functions might be the “AND” and “OR” functions. For example, the AND
function gives a value of 1 if the two input digits are identical, that is (0,0), or (1,1). If the
input digits are different, the AND function gives a value of 0. The OR function gives a value
of 1 if either of the two input digits equals 1, that is (0,1), (1,0), or (1,1).
The XOR function gives a value of 1 if the two inputs are different, that is (1,0) or (0,1). If
the two inputs are the same, (0,0) or (0,1), then the value is 0.
What is XOR(A,B)? Remember this: Is A different to B? If so, the answer is 1.
? If A ≠ B, then XOR(A,B) = 1
? If A = B, then XOR(A,B) = 0
The XOR function can be represented by the “truth table” shown in Table 1.
Input
A
Input
B
Output
XOR(A,B)
0 0 0
0 1 1
1 0 1
1 1 0
Table 1. Truth table for the XOR function.
3.1.2 Linear Feedback Registers
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
10
Linear feedback registers are used to generate a pseudorandom number sequence. The
sequence is pseudorandom, since the sequence repeats after a certain number of digits (which,
as we shall see, depends on the size of the register). However, the statistical properties of the
sequence are very good, in that the sequence appears to be white noise. We return to these
properties later, since they are important for understanding the measurement process. For
now, we look at how the register works.
Cycle, N AN = XOR(AN-1,C N-1) BN = AN-1 CN = BN-1
1 initialise: 1 1 1
2 XOR(1,1) = 0 1 1
3 XOR(0,1) = 1 0 1
4 XOR(1,1) = 0 1 0
5 XOR(0,0) = 0 0 1
6 XOR(0,1) = 1 0 0
7 XOR(1,0) = 1 1 0
8 (=1) XOR(1,0) = 1 1 1
(pattern repeats)
Table 2. A 3 stage linear feedback register. The output is in column C.
Table 2 illustrates a simple example: the “3 stage linear feedback register.” The “state” of the
register is defined by three binary numbers (A, B, C). The state changes after a specific time
interval. To start the whole process, the intial state of a feedback register is always filled with
1; that is, for the 3 stage register, the initial state is (1, 1, 1). The digits in this state are now
shifted to the right, giving (blank, 1, 1). The digit (1) that is pushed off the right side is the
output from the register. The blank is replaced by taking the XOR of the other two digits
(1,1). The value, in this case, equals 0. The new state is therefore (0, 1, 1). This process is
then repeated, so that the new output is (1), and the next state is (1, 0, 1). The next output is
(1) and the next state is (1, 1, 0). The next output is (0), and the next state is (0, 1, 1), and so
on.
In the above example, the outputs can be written (1, 1, 1, 0, ....). This stream of digits is
known as the “linear feedback register sequence.” This sequence will start to repeat after a
while. It turns out that during a complete cycle, the feedback register will produce every
possible combination of binary numbers, except for (0, 0, 0). We can therefore easily
calculate the length of the sequence before it starts to repeat again. For a 3 stage register,
there are 8 possible combinations of binary digits. This means that the sequence will repeat
after 7 cycles. The sequence length is therefore 7 bits. More generally, the sequence length
is:
L(N) = 2N?1
where N is the size of the register (number of digits in the state). For example, a 4 state linear
feedback register will have a sequence length of 15 bits.
3.1.3 C/A Code
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
11
The C/A code is based on the 10 stage linear feedback register sequence, for which the
sequence length is L(10) = 210?1 = 1023 bits. The C/A code really has a repeating sequence
of 1023 bits; however the design is slightly more complicated than presented above. The C/A
code is actually a “Gold code”, which is derived by taking the XOR of the output from 2
linear feedback registers. Unique C/A codes can be generated for each satellite by selecting
different pairs of cells from each register to define the output.
In summary, the C/A code is a unique Gold code on each satellite, which is a pseudorandom
sequence of bits with a repeating sequence length of 1023. C/A bit transitions occur at 1.023
Mhz. Note that the fundamental frequency in the satellite is 10.23 Mhz, so this represents one
transition every 10 cycles. At this rate of bit transitions, the full sequence of 1023 bits is
transmitted in 1 ms. Therefore, the sequence repeats 1000 times per second. The chip length
(distance between bit transitions) is 293 m. Therefore, the sequence repeats every 300 km.
3.1.4 P Code
The P code is also generated from a combination of two different registers, in such a way that
it repeats every 266.4 days. Each 7 day section is assigned a “PRN code.” Satellites are often
identified by their PRN number; however, the user should beware that any given satellite can
have its PRN code changed. Therefore, PRN codes should not be used in place of Satellite
Vehicle Numbers (SVN) when talking about particular satellites. (For example, if someone
writes software which identifies satellites using PRN numbers, there might be a problem in
orbit modelling, for example, PRN 2 is assigned to a Block II satellite now, but to a Block IIR
satellite next year). There are 38 possible PRN codes; given that there are 24 nominal
satellites, some PRN codes are left unused. The PRN sequence is reset at Saturday midnight,
defining the start of “GPS week.”
3.1.5 GPS signal transmission and reception
Let us now summarise how the GPS signal is transmitted from space, and then received on the
ground. The GPS signal starts in the satellite as a voltage which oscillates at the fundamental
clock frequency of 10.23 Mhz. (If selective availability is on, this signal is then “dithered” so
that the frequency varies unpredictably). This signal is then separately multiplied in frequency
by the integers 154 and 120, to create the L1 and L2 carrier signals. The signals are then
multiplied by +1 and ?1 according the algorithms described above to generate the C/A code
(on L1) and the P code (on both L1 and L2). These codes are unique to each satellite. Finally,
the Navigation Message is encoded onto the signal. The signals are boosted by an amplifier,
and then sent to transmitting antennas, which point towards the Earth. These antennas are
little more than exposed electrical conductors which radiate the signal into space in the form
of electromagnetic waves.
These electromagnetic waves pass through space and the Earth’s atmosphere, at very close to
the speed of light in a vacuum, until they reach the receiver’s antenna. The waves create a
minute signal in the antenna, in the form of an oscillating voltage. The signal is now pre-
amplified at the antenna, to boost the signal strength, so that it is not overcome by noise by the
time it gets to the other end of the antenna cable. The signal then enters the receiver, which
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
12
then measures it using a process known as “autocorrelation.” It is beyond the scope of this
paper to go into the details of receiver design, so our description will be kept at the level
required to understand how the observable model can be developed.
3.2 AUTOCORRELATION TECHNIQUE
We have described how the GPS satellites construct the GPS signals. Actually, the receiver
also generate GPS-like signals internally. The receiver knows precisely what the transmitted
GPS signal is supposed to look like at any given time, and it generates an electronic replica, in
synchronisation with the receiver’s own clock. The receiver then compares the replica signal
with the actual signal. Since the GPS signal was actually created in the satellite some time
previously (about 0.07 seconds ago, due to the speed of light), the receiver’s replica signal
must be delayed in to match up the incoming signal with the replica signal. This time delay is
actually what the receiver is fundamentally measuring. Clearly, this represents the time taken
for the signal to pass from the satellite to the receiver, but it also includes any error in the
satellite clock, and any error in the receiver clock. One can see that the time delay is therefore
related to the range to the satellite. We return to this model later, and now turn our attention
to how the receiver matches the two signals.
The time difference is computed by autocorrelation. The first bit from signal one is
multiplied by the first bit of signal two. For example, if the first bits from the two signals
both have values ?1, then the result is (?1) × (?1) = +1. Similarly, if both bits have values
+1, then the result is +1. On the other hand, if the two bits disagree, the result is (+1) × (?1) =
?1. This process is repeated for the second pair of bits, and so on. The result can be written
as a sequence of +1 (where the bits agree) and -1 (where the bits disagree). This sequence is
then summed, and divided by the total number of bits in each signal. For example, if signal
A can be written (+1, ?1, ?1, +1, ?1), and signal B can be written (+1, +1, ?1, ?1, +1), then
multiplication gives (+1, ?1, +1, ?1, ?1); the sum of which gives ?1; then dividing by the
number of bits (5) gives ?0.2. Note that if the two signals matched perfectly, the result would
be +1. If the two signals were completely random, we should expect a result close to zero.
This is why the GPS signals are designed to look random. When the two signals are not
properly matched in time, the result of autocorrelation gives an answer close to zero; if the
signals are matched in time, the result is close to +1 (but not exactly, since a real signal also
has noise, so some bits are incorrect). One can see that the larger the number of bits that are
compared, the better the resolution. This is because the random bits will average to zero
better, the more bits we compare.
The Gold codes have the property that the autocorrelation is constant until we get to within
one chip of the correct answer. Within that window of ±1 chip, the autocorrelation function
looks like an equilateral triangle, with a value of 1 at its peak (assuming no noise). We can
therefore use the known triangular shape as a model to help us find the time displacement that
maximises the autocorrelation. (More sophisticated receivers account for the fact that
multipath distorts the shape of this triangle, and can thus reduce the effect of multipath).
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
13
Now that we have found the peak autocorrelation, the inferred time displacement between the
two signals is multiplied by the speed of light. This observation is called the pseudorange.
The pseudorange measurement is described schematically in Figure 1.
3.3 PSEUDORANGE OBSERVATION EQUATIONS
3.3.1 Simplified Pseudorange Model
Receivers record data at regular, specified intervals (say, every 30 seconds, as instructed by
the receiver user). It is the reading of the receiver clock time T, which is used to say exactly
when the measurement is sampled. Therefore, the value of T at a measurement epoch is
known exactly, and is written to the data file along with the observation. (What is not known,
is the true time of measurement). The actual observation to satellite s can be writted:
PS = (T? TS) c
where T is the known reading of the receiver clock when signal is received, TS is the reading
of the satellite clock when the signal was transmitted, and c is the speed of light (in a vacuum)
= 299792458 m/s.
(T?Ts)
Received signal, driven by satellite clock Ts
Replica signal, driven by receiver clock T
Antenna
Satellite
clock, Ts
Transmitted signal
Receiver
clock T
Figure 1: A schematic diagram showing how the GPS pseudorange observation
is related to the satellite and receiver clocks.
The modelled observation can be developed by setting the clock time T equal to the true
receive time t plus a clock bias τ, for both the receiver and satellite clocks:
T t
T tS S S
= +
= +
τ
τ
Substitution gives the pseudorange as a function of the true time the signal was received:
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
14
PS(t) = ((t + τ) ? (t S + τ S ))c
= (t ? t S )c + cτ ? cτ S
= ρ S (t, t S ) + cτ ? cτ S
where ρ S (t, t S ) is the range from receiver (at receive time) to the satellite (at transmit time).
This model is simplified; for example, it assumes the speed of light in the atmosphere is c, and
it ignores the theory of relativity; but this simplified model is useful to gain insight into the
principles of GPS. From Pythagoras Theorem, we can write:
( ) ( ) ( )ρ S S S S S S S St t x t x t y t y t y t y t( , ) ( ) ( ) ( ) ( ) ( ) ( )= ? + ? + ?2 2 2
The Navigation message allows us to compute the satellite position (x S , y S , z S ) and the
satellite clock bias τ S . Therefore we are left with 4 unknowns, the receiver position (x, y, z)
and the receiver clock bias τ.
We note here one complication: that the satellite position must be calculated at transmission
time, t S . This is important, because the satellite range can change as much as 60 metres from
the time the signal was transmitted, to the time the signal was received, approximately 0.07
seconds later. If the receive time were used instead, the error in computed range could be tens
of metres. Starting with the receive time, t, the transmit time can be computed by an iterative
algorithm known as “the light time equation,” which can be written as follows:
( )t t T
t t t tc
t t t tc
S
S
S S
S
S S
( )
( ) ( , ( ))
( ) ( , ( ))
0
1 0
2 1
= = ?
= ?
= ?
τ
ρ
ρ
where the satellite position (and hence the range ρ S (t, t S )) is calculated at each step using the
Keplerian-type elements from the Navigation Message, and the algorithm is stopped once the
computed range converges (i.e., don’t change by more than a negligible amount). Although
more rapidly converging methods have been implemented, the above method is probably the
easiest to understand.
Note that the above algorithm starts with the true receive time, which requires the receiver
clock bias. We usually don’t know in advance what the bias is, but for most receivers it never
gets larger than a few milliseconds (beyond which, the receiver will reset its clock). If we
assume it is zero in the above computation, the error produced is a few metres, which is much
smaller than typical point positioning precision of approximately 50 metres with S/A switched
on. We can therefore safely ignore this effect for now, and return to it later when we discuss
the more precise carrier phase observable.
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
15
We now look at our system of simplified observation equations from 4 satellites in view of the
receiver. Using the above notation, we can write the pseudoranges to each satellite as:
P1 = ( (x1 ? x)2 + (y1 ? y)2 + (z1 ? z)2 )? + cτ ? cτ1
P2 = ( (x2 ? x)2 + (y2? y)2 + (z2 ? z)2 )? + cτ ? cτ2
P3 = ( (x3 ? x)2 + (y3 ? y)2 + (z3 ? z)2 )? + cτ ? cτ3
P4 = ( (x4 ? x)2 + (y4 ? y)2 + (z4 ? z)2 )? + cτ ? cτ4
(Note that in this and subsequent equations, the superscripts next to the satellite coordinates
are meant to identify the satellite, and should not be confused with exponents). In the
following section, we proceed to solve this system of equations for the 4 unknowns, (x, y, z, τ)
using familiar least squares methods. Although this is not strictly necessary for 4 unknowns
with 4 parameters, it does generalise the solution to the case where we have m ≥ 4 satellites
in view.
4. POINT POSITIONING USING PSEUDORANGE
4.1 LEAST SQUARES ESTIMATION
4.1.1 Linearised Model
We solve the point positioning problem by first linearising the pseudorange observation
equations, and then using the familiar methods of least-squares analysis. For completeness, we
summarise the linearisation procedure and the development of the least squares method
specifically for the GPS point positioning problem. First, we assume we can write the actual
observation to be the sum of a modelled observation, plus an error term:
P P
P x y z v
observed el noise= +
= +
mod
( , , , )τ
Next, we apply Taylor’s theorem, where we expand about the model computed using
provisional parameter values (x0, y0, z0, τ0), and ignore second and higher order terms.
( ) ( ) ( ) ( )P x y z P x y z x x Px y y Py z z Pz P
P Px x Py y Pz z P
( , , , ) ( , , , )τ τ τ τ τ
τ
? + ? + ? + ? + ?
= + + + +
0 0 0 0 0 0 0 0
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?computed ? ? ? ?τ
Note that the partial derivatives in the above expression are also computed using provisional
values (x0, y0, z0, τ0). The residual observation is defined to be the difference between the
actual observation and the observation computed using the provisional parameter values:
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
16
?
? ? ? ?τ
P P P
P
x x
P
y y
P
z z
P v
≡ ?
= + + + +
observed computed
?
?
?
?
?
?
?
?τ
This can be written in matrix form:
( )? ???
?τ
P Px Py Pz P
x
y
z v= +
?
?
?
?
?
?
?
?τ
We get such an equation for each satellite in view. In general, for m satellites, we can write
this system of m equations in matrix form:
?
?
?
?
?
?
?
?τ
P
P
P
P
P
x
P
y
P
z
P
P
x
P
y
P
z
P
P
x
P
y
P
z
P
P
x
P
y
P
z
P
x
y
z
v
v
v
m
m m m m
1
2
3
1 1 1 1
2 2 2 2
3 3 3 3
1
2
3
=
+
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
?
τ
τ
τ
τ
The equation is often written using matrix symbols as:
b Ax v= +
which expresses a linear relationship between the residual observations b (i.e., observed
minus computed observations) and the unknown correction to the parameters x. The column
matrix v contains all the noise terms, which are also unknown at this point. We call the above
matrix equation the “linearised observation equations”.
4.1.2 The Design Matrix
The linear coefficients, contained in the “design matrix” A, are actually the partial derivatives
of each observation with respect to each parameter, computed using the provisional parameter
values. Note that A has the same number of columns as there are parameters, n = 4 , and has
the same number of rows as there are data, m ≥ 4 . We can derive the coefficients of A by
partial differentiation of the observation equations, producing the following expression:
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
17
A =
? ? ?
? ? ?
? ? ?
? ? ?
x x y y z z c
x x y y z z c
x x y y z z c
x x y y z z
c
m m m
0
1
0
1
0
1
0
2
0
2
0
2
0
3
0
3
0
3
0 0 0
ρ ρ ρ
ρ ρ ρ
ρ ρ ρ
ρ ρ ρ
Note that A is shown to be purely a function of the direction to each of the satellites as
observed from the receiver.
4.1.3 The Least Squares Solution
Let us consider a solution for the linearised observation equations, denoted x . The
“estimated residuals” are defined as the difference between the actual observations and the
new, estimated model for the observations. Using the linearised form of the observation
equations, we can write the estimated residuals as:
v b Ax= ?
The “least squares” solution can be found by varying the value of x until the following
functional is minimised:
( ) ( )J vi
i
m
( )x v v b Ax b AxT T≡ = = ? ?
=
2
1
.
That is, we are minimising the sum of squares of the estimated residuals. If we vary x by a
small amount, then J(x) should also vary, except at the desired solution where it is stationary
(since the slope of a function is zero at a minimum point). The following illustrates the
application of this method to derive the least squares solution:
( ) ( ){ }
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( )
( )( )
( )
δ
δ
δ δ
δ δ
δ
δ
δ
J()
x
b Ax b Ax
b Ax b Ax b Ax b Ax
A x b Ax b Ax A x
A x b Ax
x A b Ax
x A b A Ax
A Ax A b
T
T T
T T
T
T T
T T T
T T
=
? ? =
? ? + ? ? =
? ? + ? ? =
? ? =
? =
? =
=
0
0
0
0
2 0
0
0
.
The last line is called the system of “normal equations”. The solution to the normal equations
is therefore:
( )x A A A bT 1 T= ?
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
18
This assumes that the inverse to ATA exists. For example, m ≥ 4 is a necessary (but not
sufficient) condition. Problems can exist if, for example, a pair of satellites lie in the same
line of sight, or if the satellites are all in the same orbital plane. In almost all practical
situations, m ≥ 5 is sufficient. Alternatively, one parameter could be left unestimated (e.g.,
the height could be fixed to sea-level for a boat).
4.2 ERROR COMPUTATION
4.2.1 The Covariance and Cofactor Matrices
If the observations b had no errors, and the model were perfect, then the estimates x given by
the above expression would be perfect. Any errors v in the original observations b will
obviously map into errors vx in the estimates x . It is also clear that this mapping will take
exactly the same linear form as the above formula:
( )v A A A vx T 1 T= ?
If we have (a priori) an expected value for the error in the data, σ, then we can compute the
expected error in the parameters. We discuss the interpretation of the “covariance matrix”
later, but for now, we define it as the (square) matrix of expected values of one error
multiplied by another error; that is, Cij ≡ E(vivj). A diagonal element Cii is called a “variance,”
and is often written as the square of the standard deviation, Cii ≡ E(vi2) = σi2. We can
concisely define the covariance matrix by the following matrix equation:
C ≡ E(vvT).
Let us for now assume we can characterise the error in the observations by one number, the
variance σ2 = E(ν2), which is assumed to apply to all m observations. Let us also assume that
all observations are uncorrelated, E(vivj) = 0 (for i ≠ j). We can therefore write the covariance
matrix of observations as the diagonal matrix, Cσ = σ2I, where I is the m m× identity matrix:
Cσ
σ
σ
σ
=
×
2
2
2
0 0
0
0
0 0
m m
Under these assumptions, the expected covariance in the parameters for the least squares
solution takes on a simple form:
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
19
( )
( )( ) ( )( )
( ) ( )( )
( ) ( ) ( )
( ) ( )
( ) ( ) ( )
( )
C v v
A A A v A A A v
A A A vv A A A
A A A vv A A A
A A A A A A
A A A I A A A
A A
x x x
T
T 1 T T 1 T
T
T 1 T T T 1
T 1 T T T 1
T 1 T T 1
T 1 T T 1
T 1
=
=
=
=
=
=
=
= ×
? ?
? ?
? ?
? ?
? ?
?
E
E
E
E
Cσ
σ
σ
σ
2
2
2 cofactor matrix
Note that the “cofactor matrix” (ATA)?1 also appears in the formula for the least squares
estimate, x . The “cofactor matrix” is also sometimes called the “covariance matrix,” where it
is implicitly understood that it should be scaled by the variance of the input observation errors.
Since GPS observation errors are a strong function of the particular situation (e.g., due to
environmental factors), it is common to focus on the cofactor matrix, which, like A, is purely
a function of the satellite-receiver geometry at the times of the observations. The cofactor
matrix can therefore be used to assess the relative strength of the observing geometry, and to
quantify how the level of errors in the measurements can be related to the expected level of
errors in the position estimates.
It should therefore be clear why A is called the “design matrix”; we can in fact compute the
cofactor matrix in advance of a surveying session if we know where the satellites will be
(which we do, from the almanac in the Navigation Message). We can therefore “design” our
survey (in this specific case, select the time of day) to ensure that the position precision will
not be limited by poor satellite geometry.
4.2.2 Interpreting the Covariance Matrix
The covariance matrix for the estimated parameters can be written in terms of its components:
( )C A Ax T 1=
=
?σ
σ
σ σ σ σ
σ σ σ σ
σ σ σ σ
σ σ σ σ
2
2
2
2
2
2
x xy xz x
yx y yz y
zx zy z z
x y z
τ
τ
τ
τ τ τ τ
As an example of how to interpret these components, if the observation errors were at the
level of σ = 1 metre, the error in y coordinate would be at the level of σy metres; if the
observation errors were σ = 2 metres, the error in y would be 2σy metres, and so on.
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
20
The off-diagonal elements indicate the degree of correlation between parameters. If σyz were
negative, this means that a positive error in y will probably be accompanied by a negative
error in z, and visa versa. A useful measure of correlation is the “correlation coefficient,”
which is defined as
ρ σ
σ σij
ij
i j
=
2 2
The correlation coefficient is only a function of the cofactor matrix, and is independent of the
observation variance, σ2. Its value can range between ?1 to +1, where 0 indicates no
correlation, and +1 indicates perfect correlation (i.e., the two parameters are effectively
identical). Several textbooks show that the “error ellipse” in the plane defined by the (z, y)
coordinates (for example) can be computed using the elements σz2, σy2, and ρzy .
4.2.3 Local Coordinate Errors
Applications tend to focus on horizontal and vertical position. Also, height, h, tends to have
largest error than horizontal coordinates. It is therefore more convenient to look at errors in
local geodetic coordinates; that is to transform geocentric coordinates (u,v,w) to local
topocentric coordinates (n, e, h). For this, we have to transform the covariance matrix, using
the laws of error propogation. Consider the rotation matrix G which takes us from small
relative vector in geocentric system into the local system at latitude ? and longitude λ:
? ?
?
?
?
?
?
?
L G X=
=
? ?
?
n
e
h
x
y
z
sin cos sin sin cos
sin cos
cos cos cos sin sin
? λ ? λ ?
λ λ
? λ ? λ ?
0
Obviously, matrix G would also transform the errors in ?X into errors in ?L:
v GvL X=
We now derive how to transform the covariance matrix of coordinates from geocentric system
to the local system. This procedure is sometimes referred to as the “law of propogation of
errors”:
( )
( )( )( )
( )
( )
C v v
Gv Gv
Gv v G
G v v G
GC G
L L L
T
X X
T
X X
T T
X X
T T
X
T
=
=
=
=
=
E
E
E
E
For future reference, the general form of the resulting equation C GC GL X T= is applicable to
any problem involving an affine transformation (i.e., multiplication of a column vector by any
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
21
rectangular matrix, G). Note that for this particular problem, CX is really the 3×3 submatrix
taken from the original 4×4 matrix (which also included coefficients for the clock parameter
τ). The covariance matrix in the local system CL can be written in terms of its components:
CL
n ne nh
en e eh
hn he h
=
σ
σ σ σ
σ σ σ
σ σ σ
2
2
2
2
We could then use this covariance, for example, to plot error ellipses in the horizontal plane.
4.2.4 Dilution of Precision
We can now define the various types of “dilution of precision” (DOP) as a function of
diagonal elements of the covariance matrix in the local system:
VDOP
HDOP
PDOP
TDOP
GDOP c
h
n e
n e h
n e h
≡
≡ +
≡ + +
≡
≡ + + +
σ
σ σ
σ σ σ
σ
σ σ σ σ
τ
τ
2 2
2 2 2
2 2 2 2 2
where, for example, VDOP stands for “vertical dilution of precision,” H stands for horizontal,
P for position, T for time, and G for geometric. As an example of how to interpret DOP, a
standard deviation of 1 metre in observations would give a standard deviation in horizontal
position of HDOP metres, and a standard deviation in the receiver clock bias of TDOP
seconds. If VDOP had a value of 5, we could expect pseudorange errors of 1 metre to map
into vertical position errors of 5 metres, and so on. As we have seen, the cofactor matrix and
therefore the DOP values are purely a function of satellite geometry as observed by the
receiver. A “good geometry” therefore gives low DOP values. A “bad geometry” can give
very high DOP values. As a general rule, PDOP values larger than 5 are considered poor. If
there are fewer than a sufficient number of satellites to produce a solution, or if 2 out of 4
satellites lie in approximately the same direction in the sky, then the cofactor matrix becomes
singular, and the DOP values go to infinity. The above formulas assume that all 4 parameters
(x, y, z, τ) are being estimated. Of course, if fewer than these are estimated, for example if
height is not estimated, then the modified DOP values would get smaller, and they would no
longer be generally infinity for only 3 satellites in view.
4.2.5 Mission Planning
Mission planning is the term used to describe the pre-analysis of the satellite geometry in
advance of a survey. Essentially, it typically involves using commercial software to plot the
DOP values as a function of time at a given geographical location. Since most applications
involve local to regional distances, it is not too important which station’s location is used for
this analysis, since the satellites will appear approximately in the same position in the sky for
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
22
all stations. One thing that can vary a lot from station to station is the “elevation mask”. Most
software allow the user to specify which parts of the sky obstruct the view of the satellites
(e.g., due to trees, buildings, or mountains). The elevation mask can substantially change the
DOP values, so careful attention should be paid to this. Even if the elevation mask went
down to the horizon, the user may wish to set it to 15 degrees all around, as research shows
that data below 15 degrees is usually plagued by multipathing errors and other problems, such
as cycle slips, and a low signal to noise ratio. As mentioned previously, the user might only
be interested in horizontal position, where the height is known adequately in advance (e.g., for
a boat at sea). Most software allow for DOP values to be computed under the assumption that
height is fixed.
5. THE CARRIER PHASE OBSERVABLE
5.1 CONCEPTS
We now introduce the carrier phase observable, which is used for high precision applications.
We start with the basic concepts, starting with the meaning of “phase”, the principles of
interferometry, and the Doppler effect. We then go on to describe the process of observing
the carrier phase, and develop an observation model. Fortunately, most of the model can be
reduced to what we have learned so far for the pseudorange. Unlike most textbooks, we take
the approach of presenting the model in the “range formulism”, where the carrier phase is
expressed in units of metres, rather than cycles. However, there are some fundamental
differences between the carrier phase and the pseudorange observables, as we shall see when
we discuss “phase ambiguity” and the infamous problem of “cycle slips”.
5.1.1 The Meaning of “Phase,” “Frequency” and “Clock Time”
“Phase” is simply “angle of rotation,” which is conventionally in units of “cycles” for GPS
analysis. Consider a point moving anti-clockwise around the edge of a circle, and draw a line
from the centre of the circle to the point. As illustrated in Figure 2, the “phase” ?(t) at any
given time t can be defined as the angle through which this line has rotated.
Phase is intimately connected with our concept of time, which is always based on some form
of periodic motion, such as the rotation of the Earth, the orbit of the Earth around the Sun
(“dynamic time”), or the oscillation of a quartz crystal in a wristwatch (“atomic time”). Even
our reprentation of time is often based on rotation, such as the angle of the hands on the face
of a clock. Angles of rotation give us our measure of “time.” In this way, phase can be
thought of as a measure of time (after conversion into appropriate units). We can write this
formally as:
( )T t k t( ) ( )= ?? ? 0
where T(t) is the time according to our clock at time t (whatever the clock may be), ?0 =
?(0) is so that the clock reads zero when t = 0, and k is a calibration constant, converting the
units of cycles into units of seconds. Indeed, we can take the above equation as the definition
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
23
of clock time. Whether of not this clock time is useful depends on the constancy of rate of
change of phase. This brings us to the concept of frequency.
The “frequency,” expressed in units of “cycles per second,” is the number of times the line
completes a full 360o rotation in one second (which of course, is generally a fractional
number). This definition is somewhat lacking, since it seems to assume that the rotation is
steady over the course of one second. One can better define frequency instantaneously as the
first derivative of phase with respect to time; that is, the angular speed.
f d tdt≡ ?( )
We chose to treat phase as a fundamental quantity, and frequency as a derived quantity. For
example, we can say that frequency is a constant, if we observe the phase as changing linearly
in time. Constant frequency is the basis of an ideal clock. If the frequency can be written as a
constant, f0, then we can write the phase of an ideal clock as:
? ?ideal = +f t0 0
therefore
T kf tideal = 0
Since we want our a clock second to equal a conventional second (Tideal=t), we see that an
appropriate choice for the calibration constant is k f= 1 0 , where f0 is the nominal frequency
of the oscillator. Going back to our original equation for clock time, we can now define clock
time as:
T t t f( ) ( )= ?? ? 0
0
φ A
A0
A
A0
Rotates with
frequency, f
Time, t
?A0
φ Phase
(cycles)
Signal
1.00.5
Figure 2: The meaning of phase.
5.1.2 How phase is related to a periodic signal
At time t, the height of point A(t) above the centre of the circle in figure 2 is given by:
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
24
[ ]A t A t( ) sin ( )= 0 2pi?
where A0 is the radius of the circle. Since the concept of phase is often applied to periodic
signals, we can call A(t) the “signal” and A0 the “amplitude of the signal”. For example, in the
case of radio waves, A(t) would be the strength of the electric field, which oscillates in time as
the wave passes by. Inverting the above formula, we can therefore determine the phase ?(t) if
we measure the signal A(t) (and similarly, we could infer the clock time).
Note that, for an ideal clock, the signal would be a pure sinusoidal function of time:
( )
( ) ( )
A A
A f t
A f t A f t
A t A tS C
ideal ideal=
= +
= +
= +
0
0 0 0
0 0 0 0 0 0
0 0 0 0
2
2 2
2 2 2 2
sin
sin
cos sin sin cos
sin cos
pi?
pi pi?
pi? pi pi? pi
ω ω
where the “angular frequency” ω pi0 02≡ f has units of radians per second. For a real clock, the
signal would the same sinusoidal function of its own “clock time,” (but would generally be a
complicated function of true time):
A T A T A TS C( ) sin cos= +0 0 0 0ω ω
We note that the nominal GPS signal takes on the above form, except that the signal is
modulated by “chips”, formed by multiplying the amplitudes A0S (for C/A code) and A0C (for
P code) by a pseudorandom sequence of +1 or ?1. The underlying sinusoidal signal is called
the “carrier signal.” It is the phase of the carrier signal that gives us precise access to the
satellite clock time; therefore we can use this phase for precise positioning.
5.1.3 Carrier Beat Signal
The GPS carrier signal G(t) from the satellite is “mixed” (multiplied) with the receiver’s own
replica carrier signal R(t). The result of this mixing is shown in Figure 3.
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
25
Reference signal
GPS signal
Reference x GPS
Beat signal
Figure 3: Producing a beat signal by mixing the carrier and replica signals
Mathematically, one can show that one would expect the result to be the difference between a
low frequency signal and a high frequency signal:
( ) ( )[ ]
R t G t G t R t
G R t t t t
G R
R G R G
( ) ( ) sin ( ) sin ( )
cos ( ) ( ) cos ( ) ( )
? = ×
= ? ? +
0 0
0 0
2 2
2 2 2
pi? pi?
pi ? ? pi ? ?
The high frequency component can be easily filtered out by the receiver electronics, leaving
only the carrier beat signal.
{ }
( )
( )
B t R t G t
G R t t
B t
R G
B
( ) ( ) ( )
cos ( ) ( )
cos ( )
= ?
= ?
≡
Filter
0 0
0
2 2
2
pi ? ?
pi ?
where we have introduced the carrier beat phase ?B(t), which is defined to be equal to the
difference in phase between the replica signal and the GPS signal.
? ? ?B R Gt t t( ) ( ) ( )≡ ?
By differentiating the above equation with respect to time, we find that the “beat frequency” is
equal to the difference in frequencies of the two input signals.
f ddt f fB B R G≡ = ??
We note that the above formulas apply even when the carrier phase is modulated with codes,
provided the replica signal is also modulated (because the values of ?1 will cancel when
multiplying the two signals). If the codes are not known, it is possible to square both the
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
26
incoming signal and the replica signal prior to mixing. The problem with this is that squaring
amplifies the noise, thus introducing larger random measurement errors.
5.1.4 Origin of the Phase Ambiguity
Our model of carrier beat phase not a complete picture, since we can arbitrarily add an integer
number of cycles to the carrier beat phase, and produce exactly the same observed beat signal.
Suppose we only record the fractional phase of the first measurement. We would have no way
of telling which integer N to add to this recorded phase so that it really did equal the
difference in phase between the replica signal and the GPS signal. This is fundamentally
because we have no direct measure of the total phase of the incoming GPS signal. We can
express this as follows:
Φ + = ?N R G? ?
where we use a capital Greek symbol Φ to emphasise that it represents the phase value
actually recorded by the receiver. Provided the receiver does keep track of how many
complete signal oscillations there have been since the first measurement, it can attach this
number of cycles to the integer portion of the recorded beat phase. However, there will still
be an overall ambiguity N that applies to all measurements. That is, we can model N as being
the same (unknown) constant for all measurements. If the receiver looses count of the
oscillations (e.g., because the signal is obstructed, or because of excessive noise), then a new
integer parameter must be introduced to the model, starting at that time. This integer
discontinuity in phase data is called a “cycle slip.”
5.1.5 Interpretation of the Phase Ambiguity
The reader might also be wondering if there is some kind of geometrical interpretation for N.
It turns out that there is, but it does require some oversimplified assumptions. By definition,
the unknown value of N can be written as:
N = (integer portion of ?R??G) ? (integer portion of Φ)
The second term is completely arbitrary, and depends on the receiver firmware. For example,
some receivers set this value to zero for the first measurement. Let us assume this is true, and
drop this term. For the sake of interpretation, let us now assume that the receiver and satellite
clocks keep perfect time. Under these circumstances, the first term would equal the integer
portion of the number of signal oscillations that occur in the receiver from the time the signal
was transmitted to the time the signal was received. We can therefore interpret N as equal to
the number of carrier wavelengths between the receiver (at the time it makes the first
observation), and the satellite (at the time it transmitted the signal). Of course, we made
assumptions about perfect clocks and the particular nature of the firmware; so we must beware
not to take this interpretation too literally.
5.1.6 Intuitive Model: The Doppler Effect
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
27
How can phase be used to measure distance? One way hinted at above is that the phase
essentially tells you the clock time. As we shall see in the next section, we can develop phase
in almost the same way as the pseudorange model. Another intuitive way of looking at it is to
consider the Doppler effect. We are all familiar with the acoustic version of the Doppler
effect, as we hear a vehicle’s at a higher pitch when it is approaching, and a lower pitch when
receding. Can we use the Doppler effect to design a distance measuring device?
Imagine two perfect clocks; one is at a fixed point, the other is approaching in a vehicle. Let
both clocks be generating a sinusoidal signal. The frequency difference between the reference
signal, and the approaching signal, increases with the vehicle’s speed of approach. Let us
build a receiver to mix the two signals and measure the beat signal. The beat frequency would
be a measure of the speed.
Let us count the cycles of the beat signal; or better yet, let us measure the phase (cycles plus
fractional cycles) of the beat signal. Clearly, the beat phase would be measures the change in
distance to vehicle. We can therefore (after appropriate unit conversion) write the intuitive
equation:
Beat phase = distance to vehicle + constant
This demonstrates that, although beat phase can be used to precisely measure change in
distance from one time to another, there is an unknown constant which prevents us from
knowing the full distance. This can be seen by considering moving the reference observer 10
metres away from the original position, and then repeating the experiment. The Doppler
effect is clearly exactly the same, and the number of cycles passing by would not change. The
very first value of the measured beat phase will indeed be different, but this single
measurement cannot be used to infer distance. For example, we have already discussed that
don’t know what integer number of cycles to attribute to the first beat phase measurement.
5.2 CARRIER PHASE OBSERVATION MODEL
5.2.1 Carrier Beat Phase Model
We now move towards a more rigorous treatment of the carrier beat phase observable,
building on our concepts of phase and signal mixing. Our notation will change slightly in
preparation for further development.
To summarise what we know already, the satellite carrier signal (from antenna) is mixed with
reference signal generated by receiver’s clock. The result, after high pass filtering, is a
“beating” signal. The phase of this beating signal equals the reference phase minus the
incoming GPS carrier phase from a satellite; however, it is ambiguous by an integer number
of cycles. From this point on, “carrier beat phase” will be simply called “carrier phase” (but it
should not be confused with the phase of the incoming signal!).
Observation of satellite S produces the carrier phase observable Φ S :
Φ S S ST T T N( ) ( ) ( )= ? ?? ?
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
28
where ? is the replica phase generated by the receiver clock, and ? S is the incoming signal
phase received from GPS satellite S. The measurement is made when the receiver clock time
is T.
Now take the point of view that the phase of the incoming signal received at receiver clock
time T is identical to the phase that was transmitted from the satellite at satellite clock time TS.
? ?S S S S Sx y z T x y z T( , , , ) ( , , , )= transmitS transmit
Of course, if we adopt this point of view, then we shall eventually have to consider the model
of how long it takes a wavefront of constant phase to propagate from the satellite to the
receiver, so that we may model the appropriate satellite clock time at the time of signal
transmission, TS. We return to that later.
As discussed previously, we can write clock time as a function of phase and nominal
frequency:
T t t f( ) ( )= ?? ? 0
0
We can therefore substitute all the phase terms with clock times:
? ?
? ?
( )
( )
T f T
T f TS S S
= +
= +
0 0
0 0transmit
S
transmit
Therefore, the carrier phase observable becomes:
( )
ΦS S S S
S S S
T f T f T N
f T T N
( ) = + ? ? ?
= ? + ? ?
0 0 0 0
0 0 0
? ?
? ?
where we implicitly understand that the clock times refer to different events (reception and
transmission, respectively).
We note that any term containing the superscript S are different for each satellites, but all
other terms are identical. Receivers are designed and calibrated so that the phase constant ?0
is identical for all satellites; that is, there should be no interchannel biases. Receivers should
also sample the carrier phase measurements from all satellites at exactly the same time. (If the
receivers have multiplexing electronics to save on cost, then the output should have been
interpolated to the same epoch for all satellites). The time TS will vary slightly from satellite to
satellite, since the satellite transmission time must have been different for all signals to arrive
at the same time. We also note that the last three terms are constant, and cannot be separated
from each other. We can collectively call these terms the “carrier phase bias,” which is clearly
not an integer.
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
29
In preparation for multi-receiver and multi-satellite analysis, we now introduce the subscripts
A, B, C, etc. to indicate quantities specific to receivers,, and we introduce superscripts j, k, l,
etc. to identify satellite-specific quantities. We write the carrier phase observed by receiver A
from satellite j:
( )Φ Aj A A j A j AjT f T T N( ) = ? + ? ?0 0 0? ?
Note that data should be sampled at exactly the same values of clock time (called “epochs”)
for all receivers, so all values of TA are identical at a given epoch. However receivers clocks
do not all run at exactly the same rate, therefore the true time of measurement will differ
slightly from receiver to receiver. Also, note that each receiver-satellite pair has a different
carrier phase ambiguity.
5.2.2 Range Formulation
It is convenient to convert the carrier phase model into units of range. This simplifies
concepts, models, and software. In the range formulation, we multiply the carrier phase
equation by the nominal wavelength.
( ) ( )
( ) ( )
( )
L T T
f T T N
c T T N
c T T B
A
j
A A
j
A
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
( ) ( )≡
= ? + ? ?
= ? + ? ?
≡ ? +
λ
λ λ ? ?
λ ? ?
0
0 0 0 0 0
0 0 0
Φ
where we still retain the name “carrier phase” for L TAj A( ) , which is in units of metres. We see
immediately that this equation is identical to that for the pseudorange, with the exception of
the “carrier phase bias,” BAj which can be written (in units of metres):
( )B NAj A j Aj≡ ? ?λ ? ?0 0 0
Note that the carrier phase bias for (undifferenced) data is not an integer number of
wavelengths, but also includes unknown instrumental phase offsets in the satellite and
receiver.
We have not mentioned yet about any differences between carrier phase on the L1 and L2
channel. Although they have different frequencies, in units of range the above equations take
on the same form. Actually, the clock bias parameters would be identical for both L1 and L2
phases, but the carrier phase bias would be different. The main difference comes when we
develop the model in terms of the propagation delay, which is a function of frequency in the
Earth’s ionosphere.
5.2.3 Observation Model
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
30
We note that the first term in the carrier phase model is simply the pseudorange, and the
second term is a constant. We have already developed a simplified model for pseudorange, so
we can therefore write a model for carrier phase as follows:
( )L T c T T B
t t c c I B
A
j
A A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
( )
( , )
= ? +
= + ? + ? +ρ τ τ Ζ
In the above expression, we have explicitly included the delay on the signal due to the
troposphere ZAj and the ionosphere ?I Aj (the minus sign indicating that the phase velocity
actually increases). Models for the atmospheric delay terms are beyond the scope of this text.
The model for pseudorange can be similarly improved, with the small difference that the
ionospheric delay has a positive sign.
( )P T c T T
t t c c I
A
j
A A
j
A
j
A
j
A
j
A
j
A
j
( )
( , )
= ?
= + ? + +ρ τ τ Ζ
This is because, from physics theory, any information, such as the +1 and ?1 “chips” which
are modulated onto the carrier wave, must travel with the “group velocity” rather than “phase
velocity”. According to the theory of relativity, information can not be transmitted faster than
c. From the physics of wave propagation in the ionosphere, it can be shown that the group
delay is (to a very good first order approximation) precisely the same magnitude, but opposite
sign of the phase delay (which is really a phase “advance”).
5.2.4 Accounting for Time-Tag Bias
Before proceeding, we return to the problem posed in our discussion of the pseudorange
model, that we typically do not know the true time of signal reception tA which we need to
calculate the satellite-receiver range term ρ Aj A jt t( , ) precisely. From section 3.3.1, the true
time of reception can be written:
t TA A A= ? τ
where the epoch TA is known exactly, as it is the receiver clock time written into the data file
with the observation (and hence called the “time-tag”). However, the receiver clock bias τ A
is not known initially, but could be as large as milliseconds. The problem is that, due to
satellite motion and Earth rotation, the range will change by several metres over the period of
a few milliseconds, so we must be careful to account for this for precision work (especially
when using the carrier phase observable). For precision work (1 mm), we should use a value
τ A that is accurate to 1 μs.
There are various approaches to dealing with this in GPS geodetic software, which typically
use some combination of the following methods:
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
31
? use values of the receiver clock bias computed in a first step using a pseudorange point
position solution at each epoch;
? iterate the least-squares procedure, processing both carrier phase and pseudorange data
simultaneously, and using estimates of the clock bias to compute the true receive time, and
therefore the new range model;
? use an estimate t j of the true transmit time t j to compute the satellite position.
( )
t T
T P c
j j j
A A
j j
= ?
= ? +
τ
τ
where the satellite clock bias τ j is obtained from the Navigation Message. One can then
directly compute the range term and true receive time with sufficient precision, provided
the approximate station coordinates are known to within 300 m (corresponding to the 1 μs
timing requirement). Interestingly, this is the basis for “time transfer,” since it allows one
to compute the receiver clock bias using pseudorange data from only one GPS satellite.
(For precise time transfer, two GPS satellites are always in operation with no S/A switched
on.) As a method for computing range for precise positioning, this is not often used,
perhaps for the reason that it is not a pure model, as it depends on pseudorange data and
approximate positions.
? one can take a modelling “short cut” to avoid iteration by expanding the range model as a
first order Taylor series. Since this method often appears in the development of the
observation equation in textbooks, we discuss it in more detail here.
5.2.5 A Note on the Range-Rate Term
The observation equation can be approximated as follows:
( )
L T t t c c I B
T t c c I B
T t c I B
T t c c I B
A
j
A A
j
A
j
A
j
A
j
A
j
A
j
A
j
A A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
A
j
( ) ( , )
( , )
( , )
( , )
= + ? + ? +
= ? + ? + ? +
≈ ′ ? ? + ? +
= ′ + ? ? + ? +
ρ τ τ
ρ τ τ τ
ρ ρ τ τ
ρ ρ τ τ
Ζ
Ζ
Ζ
Ζ
where we see that the effect can be accounted for by introducing the modelled range rate (i.e.,
the relative speed of the satellite in the direction of view). The “prime” for the satellite
transmit time ′t j (which is used to compute the satellite coordinates) is to indicate that it is
not the true transmit time, but the time computed using the nominal receive time TA. A first
order Taylor expansion has been used. The higher order terms will only become significant
error sources if the receiver clock bias is greater than about 10 ms, which does not usually
happen with modern receivers. In any case, clock biases greater than this amount would result
in a worse error in relative position due to the effect of S/A (see section 5.3.1).
Textbooks sometimes include a “range rate” term in the development of the phase observation
model, even though, strictly speaking, it is unnecessary. After all, the first line line of the
above equation is correct, and the lack of a priori knowledge of the receiver clock bias can
easily be dealt with by least-squares iteration, or prior point positioning using the
pseudorange. On the other hand, it is nevertheless instructional to show the above set of
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
32
equations, since it does illustrate that it is more correct to use ( )c Aj? ρ as the partial
derivatives with respect to the receiver clock in the design matrix, rather than simply using c
(section 4.1.2). This is crucial if one is not initialising clocks using point position solutions
or iteration (as is typical, for example, with the GIPSY OASIS II software). It is not important
if initialisation of τ A is achieved with 1 μs accuracy.
In the expressions to follow, we shall not explicitly include the range rate term on the
assumption that time-tag bias has been handled one way or another.
5.3 DIFFERENCING TECHNIQUES
5.3.1 Single Differencing
ρAj
ρBj
Station A
records LAj and LAk
at epochs 1, 2, 3, ...
Station B
records LBj and LBk
at epochs 1, 2, 3, ...
Satellite j
Figure 4: Single differencing geometry
The purpose of “single differencing” is to eliminate satellite clock bias. Consider the
observation equations for two receivers, A and B observing same satellite, j:
L c c Z I B
L c c Z I B
A
j
A
j
A
j
A
j
A
j
A
j
B
j
B
j
B
j
B
j
B
j
B
j
= + ? + ? +
= + ? + ? +
ρ τ τ
ρ τ τ
The single difference phase is defined as the difference between these two:
( ) ( )
( ) ( ) ( ) ( ) ( ) ( )
?
? ? ? ? ?
L L L
c c Z I B c c Z I B
c c c c Z Z I I B B
c Z I B
AB
j
A
j
B
j
A
j
A
j
A
j
A
j
A
j
B
j
B
j
B
j
B
j
B
j
A
j
B
j
A B
j j
A
j
B
j
A
j
B
j
A
j
B
j
AB
j
AB AB
j
AB
j
AB
j
≡ ?
= + ? + ? + ? + ? + ? +
= ? + ? ? ? + ? ? ? ? ?
= + + ? +
ρ τ τ ρ τ τ
ρ ρ τ τ τ τ
ρ τ
where we use the double-subscript to denote quantities identified with two receivers, and the
triangular symbol as a mnemonic device, to emphasise that the difference is made between
two points on the ground. The geometry of single differencing is illustrated in Figure 4.
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
33
An assumption has been made, that the satellite clock bias τ j is effectively identical at the
slightly different times that the signal was transmitted to A and to B. The difference in
transmission time could be as much as a few milliseconds, either because the imperfect
receiver clocks have drifted away from GPS time by that amount, or because the stations
might be separated by 1,000 km or more. Since selective availability is typically at the level
of 10?9 (variation in frequency divided by nominal frequency), over a millisecond (10?3s) the
satellite clock error will differ by 10?12s. This translates into a distance error of 10?12c, or 0.3
mm, a negligible amount under typical S/A conditions (however, it may not be negligible if
the level of S/A were increased; but this effect could in principle be corrected if we used
reference receivers to monitor S/A). Another point worth mentioning, is that the coordinates
of the satellite at transmission time can easily be significantly different for receivers A and B,
and this should be remembered when computing the term ?ρ ABj .
The atmospheric delay terms are now considerably reduced, and vanish in the limit that the
receivers are standing side by side. The differential troposphere can usually be ignored for
horizontal separations less than approximately 30 km, however differences in height should be
modelled. The differential ionosphere can usually be ignored for separations of 1 to 30 km,
depending on ionospheric conditions. Due to ionospheric uncertainty, it is wise to calibrate
for the ionosphere using dual-frequency receivers for distances greater than a few km.
Although the single difference has the advantage that many error sources are eliminated or
reduced, the disadvantage is that only relative position can be estimated (unless the network is
global-scale). Moreover, the receiver clock bias is still unknown, and very unpredictable. This
takes us to “double differencing”.
5.3.2 Double Differencing
ρAj
ρBjρAk
ρBk
Station A
records LAj and LAk
at epochs 1, 2, 3, ...
Station B
records LBj and LBk
at epochs 1, 2, 3, ...
Satellite j Satellite k
Figure 5: Double differencing geometry.
The purpose of “double differencing” is to eliminate receiver clock bias. Consider the single
differenced observation equations for two receivers A and B observing satellites j and k:
? ? ? ? ? ?
? ? ? ? ? ?
L c Z I B
L c Z I B
AB
j
AB
j
AB AB
j
AB
j
AB
j
AB
k
AB
k
AB AB
k
AB
k
AB
k
= + + ? +
= + + ? +
ρ τ
ρ τ
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
34
The double difference phase is defined as the difference between these two:
( ) ( )
( ) ( ) ( ) ( ) ( )
?? ≡ ?
= + + ? + ? + + ? +
= ? + ? + ? ? ? ? ?
= ?? + ?? ? ?? + ??
L L L
c Z I B c Z I B
c c Z Z I I B B
Z I B
AB
jk
AB
j
AB
k
AB
j
AB AB
j
AB
j
AB
j
AB
k
AB AB
k
AB
k
AB
k
AB
j
AB
k
AB AB AB
j
AB
k
AB
j
AB
k
AB
j
AB
k
AB
jk
AB
jk
AB
jk
AB
jk
? ?
? ? ? ? ? ? ? ? ? ?
? ? ? ? ? ? ? ? ? ?
ρ τ ρ τ
ρ ρ τ τ
ρ
where we use the double-superscript to denote quantities identified with two satellites, and the
upside-down triangular symbol as a mnemonic device, to emphasise that the difference is
made between two points in the sky. Figure 5 illustrates the geometry of double differencing.
A point worth mentioning, is that although the receiver clock error has been eliminated to first
order, the residual effect due “time tag bias” on the computation of the range term (section
5.2.4) does not completely cancel, and still needs to be dealt with if the receiver separation is
large.
Any systematic effects due to unmodelled atmospheric errors are generally increased slightly
by approximately 40% by double differencing as compared to single differencing. Similarly,
random errors due to measurement noise and multipath are increased. Overall, random errors
are effectively doubled as compared with the undifferenced observation equation. On the
other hand, the motivation for double differencing is to remove clock bias, which would
create much larger errors.
One could process undifferenced or single differenced data, and estimate clock biases. In the
limit that clock biases are estimated at every epoch (the “white noise clock model”), these
methods become almost identical, provided a proper treatment is made of the data covariance
(to be described later). It is almost, but not quite identical, because differencing schemes
almost always involve pre-selection of baselines in a network to form single differences, and
data can be lost by lack of complete overlap of the observations to each satellite. (This
problem can be minimised by selecting the shortest baselines in the network to process, and
by assuring that no more than one baseline be drawn to a receiver with a significant loss of
data).
5.3.3 Double Differenced Ambiguity
The double difference combination has an additional advantage, in that the ambiguity is an
integer:
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
35
( ) ( )
( ) ( ) ( ) ( )
( )
?? = ?
= ? ? ?
= ? ? ? ? ? ? ? ? + ? ?
= ? ? ? +
= ? ??
B B B
B B B B
N N N N
N N N N
N
AB
jk
AB
j
AB
k
A
j
B
j
A
k
B
k
A
j
A
j
B
j
B
j
A
k
A
k
B
k
B
k
A
j
B
j
A
k
B
k
AB
jk
? ?
λ ? ? λ ? ? λ ? ? λ ? ?
λ
λ
0 0 0 0 0 0 0 0 0 0 0 0
0
0
Hence we can write the double differenced phase observation equation:
?? = ?? + ?? ? ?? ? ??L Z I NABjk ABjk ABjk ABjk ABjkρ λ 0
From the point of view of estimation, it makes no difference whether we use a minus or plus
sign for N, so long as the partial derivative has a consistent sign (which, for the above
equation, would be ? λ 0 ).
5.3.4 Triple Differencing
ρAj
ρBjρAk
ρBk
Station A
records LAj and LAk
at epochs 1, 2, 3, ...
Station B
records LBj and LBk
at epochs 1, 2, 3, ...
Satellite j, epoch i
Satellite k, epoch i+1
Satellite k, epoch i
Satellite j, epoch i+1
Figure 6: Triple differencing geometry
The purpose of “triple differencing” is to eliminate the integer ambiguity. Consider two
successive epochs (i, i+1) of double differenced data from receivers A and B observing
satellites j and k:
?? = ?? + ?? ? ?? ? ??
?? + = ?? + + ?? + ? ?? + ? ??
L i i Z i I i N
L i i Z i I i N
AB
jk
AB
jk
AB
jk
AB
jk
AB
jk
AB
jk
AB
jk
AB
jk
AB
jk
AB
jk
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
ρ λ
ρ λ
0
01 1 1 1
The triple difference phase is defined as the difference between these two:
δ
δ ρ δ δ
( , ) ( ) ( )
( , ) ( ) ( , ) ( ) ( , ) ( )
i i L L i L i
i i i i i Z i i i I i
AB
jk
AB
jk
AB
jk
AB
jk
AB
jk
AB
jk
+ ?? ≡ ?? + ? ??
= + ?? + + ?? ? + ??
1 1
1 1 1
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
36
where we use the delta symbol to indicate the operator that differences data between epochs.
Figure 6 illustrates triple differencing geometry.
The triple difference only removes the ambiguity if it has not changed during the time interval
between epochs. Any cycle slips will appear as outliers, and can easily be removed by
conventional techniques. This is unlike the situation with double differencing, where cycle
slips appear as step functions in the time series of data.
The disadvantage of the triple difference is that it introduces correlations between
observations in time. Generally, increasing correlations in data has the property of decreasing
the data weights. With triple differencing, the degradation in precision is substantial; so triple
differenced data are inappropriate for precise surveys. On the other hand, it is a very useful
method for obtaining better nominal parameters for double differencing (to ensure linearity),
and it is a robust method, due to the ease with which cycle slips can be identified and
removed.
It can be shown that triple difference solution is identical to the double differenced solution,
provided just one epoch double differenced equation is included for the first point in a data
arc, along with the triple differences, and provided the full data covariance matrix is used to
compute the weight matrix. This special approach can provide tremendous savings in
computation time over straightforward double differencing, while retaining robustness.
6. RELATIVE POSITIONING USING CARRIER PHASE
6.1 SELECTION OF OBSERVATIONS
6.1.1 Linear Dependence of Observations
We can usually form many more possible combinations of double differenced observations
than there are original data. This poses a paradox, since we cannot create more information
than we started with. The paradox is resolved if we realise that some double differences can
be formed by differencing pairs of other double differences. It then becomes obvious that we
should not process such observations, otherwise we would be processing the same data more
than once. This would clearly be incorrect.
Station A
records LAj , LAk, LAl
at epochs 1, 2, 3, ...
ρAj
ρBj ρAk ρBk
Station B
records LBj , LBk, LBl
at epochs 1, 2, 3, ...
Satellite j Satellite k Satellite l
ρAl ρBl
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
37
Figure 7: Double difference geometry with 3 satellites.
Figure 7 illustrates the simplest example of the problem. In this example, we have 3 satellites
j, k and l, observed by two receivers A and B. If we ignore trivial examples (e.g.,
L LABjk ABkj= ? ), there are 3 possible double differences that can be formed:
( ) ( )
( ) ( )
( ) ( )
L L L L L
L L L L L
L L L L L
AB
jk
A
j
B
j
A
k
B
k
AB
jl
A
j
B
j
A
l
B
l
AB
lk
A
l
B
l
A
k
B
k
= ? ? ?
= ? ? ?
= ? ? ?
Note that we can form any one of these observations as a linear combination of the others:
L L L
L L L
L L L
AB
jk
AB
jl
AB
lk
AB
jl
AB
jk
AB
lk
AB
lk
AB
jk
AB
jl
= +
= ?
= ?
The data set { }L L LABjk ABjl ABlk, , is therefore said to be linearly dependent. A linearly
independent set must be used for least squares. Examples of appropriate linearly independent
sets in this example are:
{ } { }
{ } { }
{ } { }
L L L a j b j
L L L a k b k
L L L a l b l
AB
jk
AB
jl j
AB
ab
AB
kj
AB
kl k
AB
ab
AB
lj
AB
lk l
AB
ab
, ;
, ;
, ;
= ≡ = ≠
= ≡ = ≠
= ≡ = ≠
Λ
Λ
Λ
6.1.2 The Reference Satellite Concept
The “reference satellite concept” involves using either set Λ Λj k, or Λl throughout the data
set. For example, double differences in set Λl all involve the satellite l. Any set is equally as
valid, and will produce identical solutions provided the data covariance is properly
constructed (see the next section). Obviously, the reference satellite itself has to have data at
every epoch, otherwise data will be lost. This can cause problems for less sophisticated
software. Typically, a reference satellite should be picked which has the longest period in
view. A better algorithm is to select a reference satellite epoch by epoch.
Our simple example can be easily extended to more than 3 satellites. For example consider
satellites 1, 2, 3, 4 and 5 in view. We can pick satellite 4 as the reference satellite; therefore
our linearly independent set is:
{ }
{ }
Λ4
41 42 43 45
4 4≡ = ≠
=
L a b
L L L L
AB
ab
AB AB AB AB
;
, , ,
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
38
Note that for a single baseline (i.e. 2 receivers), the number of linearly independent double
differenced observations is s?1, where s is the number of satellites being tracked.
6.1.3 The Reference Station Concept
However, if we have a network of more than 2 receivers, we must account for the fact that
double differenced data from the set of all baselines are linearly dependent. We therefore
introduce the “reference station” concept, where our set of double differences all include a
common reference station. This guarantees linear independence. For example, consider
satellites 1, 2, 3 and 4 being tracked by stations A, B, and C. If we pick our reference satellite
to be 3, and reference station to be B, then our chosen set is:
{ }
{ }
Λ B cdab
BA BA BA BC BC BC
L a b c B d B
L L L L L L
3
31 32 34 31 32 34
3 3≡ = ≠ = ≠
=
; ; ,
, , , , ,
Note that the number of linearly independent double differenced observations is (s?1)(r?1),
where s is the number of satellites being tracked, and r is the number of receivers. So, in our
previous example, 3 receivers and 4 satellites gives 6 observations. This assumes that s
satellites are observed by all stations. This may not be the case, either due to obstructions,
receiver problems, or because the receivers are separated by such a large distance that the
satellite is not above the horizon for some receivers.
If using the reference station concept, it is therefore best to choose a receiver close to the
middle of a large network, with few obstructions, and no hardware problems, otherwise the set
of double differences may not be as complete as it could be. The reference station concept is
obviously not optimal, and is seriously problematic for large networks. A better strategy for
large networks is to select short baselines that connect together throughout the entire network,
being careful not to introduce linear dependent observations, by not including any closed
polygons (such as triangles) in the network. In principle, there must be only one possible path
between pairs of stations. An even better strategy would be to optimise this choice for every
epoch.
6.1.4 Solution Uniqueness
It should be stressed that, if all stations were tracking the same set of satellites at all epochs,
then the selection of reference station and reference satellite will not matter, since an identical
solution will be produced whatever the selection. This assumes that the data weight matrix is
properly constructed (as described below) and that no data outliers are removed.
The problem of linear dependence usually introduces a level of arbitrariness into the solutions
due to violation of the above assumptions. This problem is also true even if the previously
suggested improvements are made to the reference station concept, since the user typically has
to make decisions on which baselines to process (even for more sophisticated software). This
is somewhat unsatisfactory, since it is there generally no unique solution, However,
experience shows that any reasonable selection will only produce small differences in the final
solutions.
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
39
There is a way to produce a unique solution, and that is to process undifferenced observations,
estimating clock parameters at each epoch. As stated previously, this will produce a solution
identical to double differencing under ideal conditions. This class of software is not typically
available commercially; however, it is should be stressed that double differencing software
does not produce significantly inferior results for most situations. What is far more important
is the quality of the observable models, the range of appropriate estimation options, and the
ability to detect and deal with cycle slips and outliers.
6.2 BASELINE SOLUTION USING DOUBLE DIFFERENCES
6.2.1 Simplified Observation Equations
We now show how relative coordinates can be estimated between two receivers using the
double differenced carrier phase data. We start by simplifying the observation equation,
assuming that the relative atmospheric delay is negligible for short distances between
receivers. We also drop the symbols “ ?? ” of the previous section to simplify the notation.
We shall therefore use the following simplified observation equation:
L NABjk ABjk ABjk= ?ρ λ 0
6.2.2 General Procedure
Processing double differenced data from two receivers results in a “baseline solution.” The
estimated parameters include the vector between the two receivers, in Cartesian coordinates
( )? ? ?x y z, , and may include parameters to model the tropospheric delay. In addition, the
ambiguity parameters N ABjk for each set of double differences to specific satellite pairs (j, k)
must be estimated.
The observation equations therefore require linearisation in terms of all these parameters
(according to the process explained in section 4.1). Typically, one station is held fixed at
good nominal coordinates, which quite often come from an initial pseudorange point position
solution. We should mention, however, that due to S/A, point position solutions can have
substantial errors (100 m) which may create significant errors in the double differenced
observation model, and in the design matrix.
If we call the fixed station A, then estimating the baseline vector is equivalent to estimating
the coordinates of station B. It is convenient to formulate the problem to estimate parameters
( )x y zB B B, , . For example, consider a GPS survey between stations A and B, which observe
satellites 1, 2, 3 and 4 for every epoch in the session, where we arbitrarily pick satellite 2 as
the reference satellite. For every epoch i, we have the following linearly independent set of 3
double differenced observations:
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
40
{ }
{ }
Λ2
21 23 24
2 2( ) ( ) ;
( ), ( ), ( )
i L i a b
L i L i L i
AB
ab
AB AB AB
≡ = ≠
=
We therefore have the parameter set { }x y z N N NB B B AB AB AB, , , , ,21 23 24 . If any cycle slips had
occured and could not be corrected, then additional ambiguity parameters must be added to
the list.
As in Section 3.4.1, the linearised observation equations can be expressed in the form
b Ax v= +
where the residual observations are listed in the b matrix, which has dimensions d × 1, where
d is the number of linearly independent double differenced data. The design matrix A has
dimensions d p× where p is the number of parameters, and the parameter corrections are
contained in the x matrix, which has dimensions p × 1. The observation errors are
represented by the v matrix, which has the same dimensionality as b. We shall discuss the
design matrix later on.
It is important to use a “weighted least squares” approach, because of correlations in the
double differenced data. We shall not not derive the weighted least squares estimator here,
but for completeness, the solution is given here:
( )x A WA A WbT 1 T= ?
where W is the data weight matrix, to be derived later on, and b is a vector containing the
double-differenced residual observations.
The covariance matrix for the estimated parameters is given by:
( )C A WAx T 1= ?
The covariance matrix can be used to judge whether the theoretically expected precision from
the observation scenario is sufficient to allow ambiguities to be fixed to integer values. If
ambiguity parameters can be fixed in the model, a theoretically more precise solution can be
generated from the same data, but without estimating the ambiguities. This process will
necessarily reduce the covariance matrix, lowering the expected errors in the station
coordinates. This does not necessarily mean that the solution is better, but that it statistically
ought to be better, assuming the integers were correctly fixed. The assessment of solution
accuracy goes beyond the scope of this discussion, but basically one can compare results with
previous results (using GPS, or even some other technique). In addition, how well the data
are fit by the model is reflected in the standard deviation of the post-fit residuals.
6.2.3 The Design Matrix
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
41
The coefficients of the design matrix can be illustrated by looking at a single row, for
example, corresponding to observation L iAB24 ( ) :
A i L ix L iy L iz L iN L iN L iN
i
x
i
y
i
z
AB
AB
B
AB
B
AB
B
AB
AB
AB
AB
AB
AB
AB
B
AB
B
AB
B
24
24 24 24 24
21
24
23
24
24
24 24 24
00 0
( ) ( ) ( ) ( ) ( ) ( ) ( )
( ) ( ) ( )
=
= ?
?
?
?
?
?
?
?
?
?
?
?
?
?ρ
?
?ρ
?
?ρ
? λ
As an example of one of the partial derivatives for one of the coordinates:
( )?ρ? ?? ρ ρ ρ ρ
?ρ
?
?ρ
?
?ρ
?
?ρ
?
?ρ
?
?ρ
?
ρ ρ
AB
B B
A B A B
A
B
B
B
A
B
B
B
B
B
B
B
B
B
B
B
i
x x i i i i
i
x
i
x
i
x
i
x
i
x
i
x
x x i
i
x x i
i
24
2 2 4 4
2 2 4 4
4 2
0
4
4
0
2
2
( ) ( ) ( ) ( ) ( )
( ) ( ) ( ) ( )
( ) ( )
( )
( )
( )
( )
= ? ? +
= ? ? +
= ?
= ? ? ?
6.2.4 Minimum Data Requirements for Least Squares
For a least squares solution, a necessary condition is that the number of data exceed the
number of estimated parameters
d p≥
where we allow for the “perfect fit solution” ( d p= ). Under the assumption that all receivers
track the same satellites for every epoch, the number of linearly independent double
differences is
( )( )d q r s= ? ?1 1
where q is the number of epochs, r the number of receivers, and s is the number of satellites.
Assuming no cycle slip parameters:
( )( )p r s= + ? ?3 1 1
where there are ( )( )r s? ?1 1 ambiguity parameters. Therefore,
( )( ) ( )( )
( )( )( )
q r s r s
q r s
? ? ≥ + ? ?
? ? ? ≥
1 1 3 1 1
1 1 1 3
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
42
Now, we know that s ≥ 2 and r ≥ 2 for us form double differences. Therefore, we can deduce
that q ≥ 4 if we have the minimal geometry of 2 receivers and 2 satellites (only one double
difference per epoch!). Obviously, this minimal configuration is very poor geometrically, and
would not be recommended as a method of precise positioning.
Note that no matter how many receivers or satellites we have, q is an integer, and therefore
under any circumstance, we must have at least q ≥ 2 . That is, we cannot do single epoch
relative positioning, if we are estimating integer ambiguities. If we can find out the
ambiguities by some other means, then single epoch relative positioning is possible.
Otherwise, we have to wait for the satellite geometry to change sufficiently in order to
produce a precise solution.
For a single baseline r = 2 with 2 epochs of data q = 2 (which we should assume are
significantly separated in time), the minimum number of satellites to produce a solution is
condition s ≥ 4 . Interestingly, this corresponds to the minimum number of satellites for point
positioning. If a tropospheric parameter were also being estimated, the condition would be
s ≥ 5. Of course, these conditions can be relaxed if we have more than 2 epochs, however it
is the end-points of a data arc which are most significant, since they usually represent the
maximum geometrical change which we require for a good solution. In summary, one can
achieve very good results over short distances with only 4 satellites, but over longer distances
where the troposphere must be estimated, a minimum of 5 satellites is recommended (at least
some time during the session).
6.3 STOCHASTIC MODEL
6.3.1 Statistical Dependence of Double Differences
We have seen how double differences can be linearly dependent. The problem we now
address is that double differenced observations that involve a common receiver and common
satellite are statistically dependent. For example, at a given epoch, double differences LAB21 ,
LAB23 and LAB24 are correlated due to the single differenced data in common, LAB2 . Any
measurement error in this single difference will contribute exactly the same error to each of
the double differences. Therefore, a positive error in LAB21 is statistically more likely to be
accompanied by a positive error in LAB23 . As another example, if we are processing a network
using a reference satellite j and reference receiver A, all double differences in the linearly
independent set will be statistically dependent because of the data in common, LAj .
6.3.2 Data Weight Matrix for Double Differences
In a situation where data are correlated, weighted least squares is appropriate. To complete
our description of how to compute a relative position estimate, we therefore need to explain
how to compute the appropriate data weight matrix, W. The construction of W can be
generally called the “stochastic model,” which describes the statistical nature of our data (as
opposed to the “functional model” described so far, from which the observables can be
computed deterministically.)
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
43
(As an aside for more advanced readers, some software process undifferenced observations,
estimating clock biases as “stochastic parameters” at every epoch. It should be emphasised
that there is a equivalence between explicit estimation of “stochastic parameters,” and the use
of an appropriate “stochastic model” which, in effect, accounts for the missing parameters
through the introduction of correlations in the data. In principle, any parameter can either be
estimated as part of the functional model, or equivalently removed using an appropriate
stochastic model. To go more into this would be beyond the scope of this text.)
The weight matrix is the inverse of the covariance matrix for the double differenced data:
W C= ???1
which has dimensions ( )( ) ( )( )q r s q r s? ? × ? ?1 1 1 1 .
We start by assuming a covariance matrix for undifferenced data (i.e., the actually recorded
data), which has dimensions qrs qrs× . Typically, this is assumed to be diagonal, since the
receiver independently measures the signals from each satellite separately. We shall,
however, keep the form general. So the problem is, given a covariance matrix for
undifferenced data, how do we compute the covariance matrix for double-differenced data?
This is achieved using the rule of propagation of errors, which we have already seen in section
4.2.3, where geocentric coordinates were mapped into topocentric coordinates using an affine
transformation. By analogy, we can deduce that the covariance of double-differenced data can
be written:
C DCDT?? =
where D is the matrix which transforms a column vector of the recorded data into a column
vector of double differenced data:
?? =L DL
Clearly, D is a rectangular matrix with the number of rows equal to the number of linearly
independent double-differenced data, and the number of columns equal to the number of
recorded data. Using our previous assumptions, D has dimensions ( )( )q r s qrs? ? ×1 1 . The
components of D must have values of either +1, ?1, or 0, arranged such that we produce a
linearly independent set of double differences (see section 6.1.1). To complete this
discussion, the double differenced data weight matrix can be written:
( )W DCDT= ?1
6.3.3 Covariance Matrix for Estimated Parameters
As we have already seen, for weighted least squares we can write the computed covariance
matrix for estimated parameters as:
( )C A WAx T 1= ?
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
44
We can now write down the full expression for the computed covariance matrix, by
substituting for the double differenced data weight matrix W:
( )( )C A DCD AX T T= ? ?1 1
As mentioned above, for the (undifferenced) data covariance C we often use a diagonal
matrix, assuming a value for the standard deviation of an observation. Typical realistic values
for this are several mm. Although the receiver can usually measure the phase with better
precision than a mm, the post-fit residuals typically show several mm standard deviations, due
to unmodelled errors such as multipath.
Even using such an inflated value for measurement precision might not produce a realistic
covariance matrix for station coordinates. This is partly due to two effects: (i) unmodelled
errors can be correlated with the parameters being estimated (an “aliasing effect”), and (ii)
post-fit almost always show some degree of time-correlation (e.g., due to multipath). A
simple, and often surprisingly effective way to deal with this problem, is to multiply the final
coordinate covariance matrix by an empirical scaling factor, inferred “by experience,”
according to the brand of software being used, the observation scenario, and the estimation
strategy used.
7. INTRODUCING HIGH PRECISION GPS GEODESY
7.1 HIGH PRECISION SOFTWARE
The observable model discussed so far has been very basic, as it glosses over advanced
features that are important for high precision software. Several software packages have been
developed since the 1980’s that are capable of delivering high precision geodetic estimates
over long baselines. This software is a result of intensive geodetic research, mainly by
universities and government research laboratories.
Typical features of such software include:
? orbit integration with appropriate force models;
? accurate observation model (Earth model, media delay...) with rigorous treatment of
celestial and terrestrial reference systems;
? reliable data editing (cycle-slips, outliers);
? estimation of all coordinates, orbits, tropospheric bias, receiver clock bias, polar
motion, and Earth spin rate;
? ambiguity resolution algorithms applicable to long baselines;
? estimation of reference frame transformation parameters and kinematic modelling
of station positions to account for plate tectonics and co-seismic displacements.
We can summarise the typical quality of geodetic results from 24 hours of data:
? relative positioning at the level of few parts per billion of baseline length;
? absolute (global) positioning at the level of 1 cm in the IERS Terrestrial Reference
Frame (ITRF);
? tropospheric delay estimated to a few mm;
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
45
? GPS orbits determined to 10 cm;
? Earth pole position determined to 1 cm;
? clock synchronisation (relative bias estimation) to 100 ps.
Two features of commercial software are sometimes conspicuously absent from more
advanced packages: (i) sometimes double differencing is not implemented, but instead,
undifferenced data are processed, and clock biases are estimated; (ii) network adjustment
using baseline solutions is unnecessary, since advanced packages do a rigorous, one-step,
simultaneous adjustment of station coordinates directly from all available GPS observations.
Some precise software packages incorporate a Kalman filter (or an equivalent formulism).
This allows for certain selected parameters to vary in time, according to a statistical
(“stochastic”) model. Typically this is used for the tropospheric bias, which can vary as a
random walk in time. A filter can also be used to estimate clock biases, where “white noise”
estimation of clock bias approaches the theoretical equivalent of double differencing.
Although many more packages have been developed, there are 3 ultra high-precision software
packages which are widely used around the world by researchers and are commonly
referenced in the scientific literature:
? BERNESE software, developed by the Astronomical Institute, University of Berne,
Switzerland;
? GAMIT software, developed by the Massachussets Institute of Technology, USA;
? GIPSY software, developed by the Jet Propulsion Laboratory, California Institute of
Technology, USA
There are several other packages, but they tend to be limited to the institutions that wrote
them. It should be noted that, unlike commercial software packages, use of the above
software can require a considerable investment in time to understand the software and how
best to use it under various circumstances. Expert training is often recommended by the
distributors.
7.2 SOURCES OF DATA AND INFORMATION
For high precision work, it is important to abide by international reference system standards
and use the best available sources of data and ancilliary information. We therefore summarise
two especially important international sources of data information for the covenience of the
interested reader:
? IERS: International Earth Rotation Service
? Central Bureau located at the Paris Observatory, France
? Documented IERS Conventions for observation models and reference systems
? IERS Annual Reports
? IERS Terrestrial Reference Frame for reference station coordinates
? Routine publication of Earth rotation parameters
? IGS: International GPS Service for Geodynamics
? Central Bureau located at the Jet Propulsion Laboratory, USA
GEOFFREY BLEWITT: BASICS OF THE GPS TECHNIQUE
46
? Documented IGS Standards for permanent GPS stations
? Oversees operation of global GPS network (~100 stations)
? Distributes tracking data and precise ephemerides
? Maintains on-line database with Internet access
8. CONCLUSIONS
Having read and understood this text, you should now understand the basics of GPS
positioning observation models and parameter estimation. You should also have an
appreciation of the difference between basic positioning, and the more advanced positioning
using high precision software packages. If all has gone well, and you think the above
statements are true, then you should now have a good background knowledge and an
appropriate context to prepare you for more advanced material.
西南交通大学：卫星定位技术与方法（中英文）：200510301347
| 课件名称： | 西南交通大学：卫星定位技术与方法（中英文） |
| 课件分类： | 测绘 |
| 课件类型： | 电子教案 |
| 文件大小： | 22.29MB |
| 下载次数： | 2 |
| 评论次数： | 2 |
| 用户评分： | 7.5 |
- 11. 西南交通大学：卫星定位技术与方法（中英文）：200510301202[20253]
- 12. 西南交通大学：卫星定位技术与方法（中英文）：200510301203
- 13. 西南交通大学：卫星定位技术与方法（中英文）：200510301203[30014]
- 14. 西南交通大学：卫星定位技术与方法（中英文）：200510301205
- 15. 西南交通大学：卫星定位技术与方法（中英文）：200510301205[10224]
- 16. 西南交通大学：卫星定位技术与方法（中英文）：200510301205[11543]
- 17. 西南交通大学：卫星定位技术与方法（中英文）：200510301205[12056]
- 18. 西南交通大学：卫星定位技术与方法（中英文）：200510301205[12907]
- 19. 西南交通大学：卫星定位技术与方法（中英文）：200510301205[26843]
- 20. 西南交通大学：卫星定位技术与方法（中英文）：200510301347
- 21. 西南交通大学：卫星定位技术与方法（中英文）：200510301350
- 22. 西南交通大学：卫星定位技术与方法（中英文）：200510301350[14871]
- 23. 西南交通大学：卫星定位技术与方法（中英文）：200510301350[25421]
- 24. 西南交通大学：卫星定位技术与方法（中英文）：200510301351
- 25. 西南交通大学：卫星定位技术与方法（中英文）：200510301351[11249]
- 26. 西南交通大学：卫星定位技术与方法（中英文）：200510301351[14302]
- 27. 西南交通大学：卫星定位技术与方法（中英文）：200510301351[17112]
- 28. 西南交通大学：卫星定位技术与方法（中英文）：200510301351[2101]
- 29. 西南交通大学：卫星定位技术与方法（中英文）：200510301351[22119]
