Professional Documents
Culture Documents
a)
cess of capturing the image - they can de-
vote all of their talents to the composition of
the picture.
Something of the challenge of photogra-
phy has clearly been lost in this evolution of
the machine/operator relationship. Neverthe-
less, thinking people will continue to be in-
trigued by the photographic process and will
"Innovation" is a regular column in GPS want to have at least a basic understanding of
World featuring discussions on recent how a camera is able to capture an image on b)
advances in GPS technology and its a piece of plastic coated with silver halide
applications as well as on the fundamentals crystals - and how that process can be con-
of GPS positioning. This month we look at trolled. So it is with GPS.
some of the mathematics involved in The GPS receiver is the latest in a length-
determining a position using GPS ening line of smart machines. In the January
pseudorange measurements. We also 1991 issue of GPS World, we looked at how
examine some of the ways of gauging the a GPS receiver works but glossed over how
accuracy of GPS positions. the measurements are converted into posi-
This column is coordinated by Richard tions by the microprocessor. Let's take a
Langley and Alfred Kleusberg of the look at that now.
Department of Surveying Engineering at the
University of New Brunswick. We very much DETERMINING POSITIONS FROM c)
welcome your comments and suggestions of PSEUDORANGES
topics for future columns. The basic measurement made by a GPS re-
ceiver is the time required for a signal to
Smart engines, smart ovens, smart cameras. propagate from a GPS satellite to the re-
We live in a world of microprocessor-con- ceiver. Because the signal travels at the
trolled devices that function with a high de- speed of light, c, this time interval can be con-
gree of reliability with little control or verted to a distance simply by multiplying it
thought on the part of the consumer. by c.
Not so many years ago, amateur photog- Let's assume that the clock in thereceiver
raphers had to know something about film is synchronized with the clock in the satel-
speeds, f-stop and shutter-speed reciprocity, lite, and that the ionosphere and troposphere,
and the absorption and reflection of light in which slightly delay the arrival of the signal,
order to take consistently good pictures. do not exist. Furthermore, let's assume there
They had to manually adjust the controls on is no measurement noise, that is, no random Figure 1. With synchronized clocks, si-
the camera to match the lighting conditions perturbation to the measurement, something multaneous range measurements to
and, of course, focus the lens. With the cam- that invariably affects all measurements to a three GPS satellites produce a determina-
eras available today, they just have to point greater or lesser degree. With a single such tion of a receiver's position. Each range
and shoot. measurement of the distance or range to the measurement can be portrayed as the
If a good picture isn't possible under the satellite, we can determine something about radius, p, of a sphere centered on a par-
existing lighting conditions, the camera will the position of the receiver: it must lie some- ticular satellite, with the intersections of
complain with a beep and may even refuse where on a sphere centered on the satellite additional spheres producing ever fewer
to take the picture unless switched over to with a radius equal to the measured range, as possible points of receiver location. The
manual operation. Photographers no longer illustrated in the first frame of Figure 1. line of position is represented in b) and c)
must think carefully about the physical pro- Let's call that distance p 1• by the perimeter of the shaded area.
INNOVATION
makes are biased by the receiver and satel- represent the position of the receiver. The
lite clock errors and therefore are referred to term c dT is the contribution to the pseu-
as pseudoranges. dorange from the receiver clock offset, dT.
A timing error of a millisecond would re- The set of four equations must be solved
sult in an error in position of about 300 simultaneously to obtain the values for X, Y,Z
kilometers, clearly an intolerable amount. Sys- together with the clock offset, dT. Although
tem operators conceivably could better syn- the equations are written in terms of geocen-
chronize the satellite clocks by frequently tric Cartesian coordinates, the resulting
sending them adjustment commands from X,Y,Z values can easily be converted to lati-
the ground, but atomic clocks have been tude, longitude, and height in any geodetic da-
found to actually keep better time if they are tum or into map grid coordinates.
left alone and their readings are corrected. Linearization of the pseudorange equations. Be-
The United States Naval Observatory moni- cause of the squares and square roots in the
tors the GPS satellite clocks and determines equations, the pseudorange measurements
the offsets and drifts with respect to GPS are dependent on the receiver coordinates in
Figure 2. Determination of receiver time. These parameters subsequently are a nonlinear way. Consequently, the equa-
clock offset (dT) and true user position uploaded to the satellites and transmitted as tions cannot be solved in the usual fashion
(intersection of shaded lines) from the in- part of the navigation message broadcast by we all learned in high school. Instead, a pro-
tersection of spheres centered on the sat- the satellites. A GPS receiver uses these sat- cedure known as Newton-Raphson iteration
ellites; pseudoranges are shown by arcs ellite clock offset values to correct the mea- is used. In this procedure, each of the equa-
of solid lines sured pseudoranges. tions is expanded into an infinitely long poly-
Nonetheless, we still have the receiver nomial based on a set of trial values or
clock error to deal with. Because of this er- guesses for X, Y,Z and dT. Then each series
ror, the three spheres with radii equal to the is truncated after the first degree term, result-
measured pseudoranges corrected by the sat- ing in an equation that is linear in incre-
ellite clock offsets will not intersect at a com- mental corrections to the trial values. The
mon point. However, if the receiver clock four linearized equations can then be solved
error, dT, can be determined, then the pseu- simultaneously to determine the values of
doranges can be corrected and the position of these increments and the trial values adjusted
the receiver determined. The situation, com- accordingly.
pressed into two dimensions, is illustrated in Because the linearized equations are an ap-
Figure 2. proximation of the nonlinear ones, this pro-
So, we actually have four unknown quan- cess, in general, must be iterated, with sub-
tities or parameters that we must determine: sequent iterations yielding smaller and
the three coordinates of our position (say, lati- smaller increments. The final solution is the
tude, longitude, and height) and the receiver one that satisfies the original nonlinear equa-
Figure 3. The set of basic equations clock offset. Now, it is mathematically im- tions to within an acceptable tolerance. Sev-
for determining user position and re- possible to uniquely determine the values of eral iterations may be required to converge to
ceiver clock offset from four pseudorange four parameters given only three measure- the final solution. However, if the initial po-
measurements ments. The way out of this conundrum is to sition estimate is close to the actual position,
measure simultaneously an additional pseu- the GPS pseudorange equations may be
lustrated in the third frame of Figure 1. One dorange to a fourth satellite. solved in just one iteration.
of these points can be dismissed immediately But just how does the GPS receiver actu- If one or more of the receiver coordinates
as being the location of our receiver because ally extract the position coordinates and the is already accurately known, then the remain-
it will lie far out in space. So, the simulta- clock offset from the measurements? In the ing coordinates and the receiver clock offset
neous measurement of the ranges to three software embedded in the GPS receiver is an can be determined using fewer than four pseu-
satellites provides sufficient information to de- algebraic model that describes the geometri- doranges. For example, say that the height
termine a position fix in three dimensions - cal arrangement we've just looked at. For of the GPS receiver is known. Then pseu-
at least in principle. each pseudorange measurement, an equation doranges to three sate IIi tes will suffice to de-
When we started our analysis, we as- can be written that relates the measurement termine the two horizontal coordinates and
sumed that the clock in the GPS receiver was to the unknown quantities. The four equa- the clock offset. To use GPS to synchronize
synchronized with the clocks in the satellites. tions are shown in Figure 3. a clock at a site with known coordinates,
This assumption, however, is fallacious. The pseudorange measurement made by only one pseudorange measurement to a sin-
When a GPS receiver is switched on, its the receiver, in units of distance, is on the gle satellite is actually required.
clock will, in general, be mis-synchronized left-hand side of each of the equations. The Inconsistent equations. What happens when
by an unknown amount with respect to the sat- expression under the square root sign is the more than four satellites are above the GPS
ellite clocks. Furthermore, the atomic clocks true range to the satellite. It is actually a rep- user's horizon? If the user's receiver can
in the satellites are synchronized with each resentation of the sphere centered on coordi- only track four satellites at a time, then the
other and to a master time scale - called nates x,y,z, the position of the satellite. The receiver will have to choose which four sat-
GPS time - only to within about a millisec- satellite coordinates are obtained from the ellites to track. We'll have something to say
ond. The range measurements the receiver navigation message. The coordinates X, Y,Z about a possible selection method a little
INNOVATION
!!ll ~~
~~1u
time, there are some small residual clocker- the horizontal axis.
rors remaining in the pseudoranges. The po- The shape of the probability density curve
sitions of the satellites as computed from the depends on the particular parameter being
~ 011 predicted ephemerides in the navigation mes-
sages are also slightly in error. Other errors
measured. However, in science and engineer-
ing the probability density curve is often of
~ o~.~~~----~~--~~~
-4 -3 -2 -1 0 2 3 4 in the pseudoranges include unmodeled ef- a particular shape known as a Gaussian or nor-
Error
fects of the ionosphere and troposphere, mul- mal distribution. To quantify such a distribu-
tipath, receiver measurement error, and, for tion or dispersion of possible errors with a sin-
the civilian user, additional clock and orbiter- gle number, we use the standard deviation
Figure 4. The Gaussian probability dis- rors due to selective availability (SA) when and usually represent it by the Greek letter
tribution function; the shaded area indi- it is in effect. sigma (a). We can determine a experimen-
cates a 68 percent probability of an error User equivalent range error. Each of these er- tally by making a large number of observa-
having a value between -1 a and + 1a rors, regardless of its origin, can be ex- tions and calculating the square root of the
INNOVATION
Other accuracy measures. In general, the three If the standard deviations for the three coor- zontal dimensions. As with the error ellip-
coordinates of a three-dimensional position dinate directions are identical, the ellipsoid de- soid, an error ellipse has a certain probabil-
fix will have different error probability dis- generates into a sphere. The radius of such ity that the true horizontal coordinates of a
tributions and, hence, different standard de- a sphere, inside of which there is a 50 per- position lie within it. For the standard error
viations. Also, the errors between any two co- cent probability of the true position fix being ellipse, this probability is 39 percent. As
ordinates may be mutually correlated; that is, located, is called spherical error probable with the error ellipsoid, the semi-axes of the
an error in one coordinate will have an effect (SEP). error ellipse, in general, are not equal to the
on the other. The term SEP is used even when the ac- standard deviations. However, given the
If we trace out a contour of equal error tual error figure is an ellipsoid. If we make standard deviations in the horizontal coordi-
probability density in all three coordinates, a large number of position fixes at a given lo- nates and their correlation, the semimajor
we get an ellipsoid centered on our position cation, we can say that the SEP is the radius and semiminor axes of the ellipse can be
fix. The shape of the ellipsoid is determined of the sphere containing 50 percent of the in- calculated.
by the standard deviations of the coordinates dividual fixes. The Department of Defense's The two-dimensional analogue of SEP is
and their correlations. (See Figure 6.) Note accuracy goal for GPS is to have a world- circular error probable (CEP). As shown in
that, in general, due to the correlations, the wide SEP of 15 meters. Figure 7, CEP is the radius of the circle in-
ellipsoid axes are not oriented in the same di- If we forget about the height coordinate side of which the true horizontal coordinates
rections as the coordinate axes. There is a cer- for the time being and consider just the hori- of a position have a 50 percent probability of
tain probability that the true position lies zontal coordinates, we can construct the two- being located.
within the ellipsoid. If this probability is 20 dimensional analogue to the error ellipsoid: Another accuracy measure frequently used
percent, then this ellipsoid is referred to as the error ellipse. It is defined as the contour in navigation is twice the root-mean-square
the standard error ellipsoid. of equal probability density in the two hori- of the horizontal distance error, or 2 drms for
short. It is equal to twice the square root of
the sum of the squares of the lengths of the
semimajor and semiminor axes of the error
ellipse. A circle of radius 2 drms will con-
I 6o tain the true horizontal position with a certain
Q)
""0
probability. Unfortunately, a drawback of 2
.a
·o, drms as a measure of error is that it does not
c correspond to a fixed value of probability for
s a given value of error. The probability var-
ies with the eccentricity of the error ellipse,
ranging from 95.4 percent (the ellipse col-
• •
.
• -20• 60
Latitude (m)
horizontal positioning accuracy for the GPS
Standard Positioning Service (SPS) and Pre-
cise Positioning Service (PPS). The latest is-
• • • sue of the Federal Radionavigation Plan
• • •
-20 (FRP) states that when GPS is declared op-
•• • erational, the horizontal accuracy for SPS is
planned to be 100 meters 2 drms at 95 per-
cent probability. This means that 95 percent
• of all horizontal position fixes should be
within 100 meters of the true position.
But what about the other 5 percent? Theo-
retically, if the position errors due to the vari-
ous UEREs including SA are from a Gauss-
-60 ian distribution, we could occasionally get
extremely large errors. However, the Depart-
Figure 7. A sample of 100 normally distributed horizontal position fix errors. These ment of Defense will control SA so that ex-
artificially generated errors have standard deviations in latitude and longitude of 15.8 cursions will not exceed 300 meters 99.99 per-
meters and 13.6 meters, respectively, with a correlation of 0.64. Shown are the error cent of the time. The corresponding designed
ellipse, the circle with 2 drms (41..8 meters) radius, and the circle with a radius equal 2 drms (95 percent) horizontal accuracy for
to the CEP (16.4 meters) corresponding to this sample. PPS is 17. 8 meters.
INNOVATION
The FRP describes the designed accuracy It should be pointed out that the stated and statistics involved in determining posi-
of the vertical component of a GPS-derived PPS position and time accuracies are de- tions using GPS. Although a GPS receiver
position at the 2u level. As this corresponds signed estimates of GPS capabilities, and su- can be operated without the user knowing
to a 95 percent probability level, it is consis- perior results have already been obtained in any of this math, a basic understanding of
tent with the accuracy quoted for the horizon- practice. Significantly greater accuracies can how the receiver determines a position is im-
tal position. For SPS, the designed vertical be obtained by both PPS and SPS users by portant for assessing the accuracy and reliabil-
2u is 156 meters; for PPS, it is 27.7 meters. operating in a differential mode with two or ity of the numbers presented on a receiver's
The FRP also gives the designed accuracy more receivers used simultaneously. In fact, display screen. •
of receiver clock synchronization at the 1u almost all of the effects of SA can be re-
level. For SPS, accuracy is planned to be moved when operating in this mode.
167 nanoseconds; for PPS, it is given conser-
vatively as 100 nanoseconds. CONCLUSION
In this column, we've taken a brief introduc-
tory look at the geometry, algebra, calculus,