You are on page 1of 10

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

Propagation Effects on the Global Positioning


System (GPS)
Ushe Chipengo, Student Member, IEEE, Shanka Wijesundara, Student Member, IEEE, Stephen J. Watt, Student
Member, IEEE Ahmed Balakhder, Student Member, IEEE

AbstractGlobal Positioning System (GPS) satellites transmit


modulated radio signals that propagate to receivers on or near
the Earths surface. These signals are subject to several electromagnetic propagation effects including: ionosopheric effects,
troposheric effects, and multipath effects. Signal processing in
GPS receivers must account for these effects for accurate performance. Here we detail numerical models, empirical models, and
typical delays associated with the various propagation effects on
GPS signals.
KeywordsGlobal Positioning System, propagation effects, ionosphere, troposphere, multipath

I. I NTRODUCTION
HE Global Positioning System was designed and built by
the U.S. Department of Defense, with the first satellite
being launched in 1978. The system became fully
operational in the mid-1990s, and is operated and maintained
by the U.S. Air Force. GPS is a medium earth orbit (MEO)
satellite constellation used in many navigation and scientific
applications. The constellation orbits at an altitude of approximately 20,200 km (12,550 mi) and includes 31 operational
satellites. The satellites radiate modulated radio signals that
propagate to receivers on or near the Earths surface. Each
GPS signal uses spread spectrum techniques to make it less
susceptible to interference and jamming [1].
Radio signals from the satellites contain important information such as the satellite clock readings, health status, orbit,
and correction data. The key to the systems accuracy is
the fact that all signal components are precisely controlled
by atomic clocks. Synchronized GPS receivers leverage this
precise timing along with digital processing techniques to
determine the receivers three dimensional position (latitude,
longitude, altitude). In addition to geolocation, GPS satellites
can be used for scientific studies of weather, climate, crustal
deformation, plate tectonics, sea level, and ice dynamics [2].

Fig. 1: GPS trilateration is used to determine GPS three


dimensional receiver position [3].
In the case of free-space medium and error free scenario,
the distance between the satellite and receiver is equal to the
geometric distance:
(tr , te ) =

p
(xs xr )2 + (ys yr )2 + (zs zr )2 ,

(1)

where the satellite coordinates (xs , ys , zs ) are a function of


the GPS signal emission time te , and the receiver coordinates
(xr , yr , zr ) are a function of the GPS signal reception time tr
[4].
To accurately determine distance between the GPS receiver
and satellites, realistic propagation affects must be considered.
Significant errors come from the ionosphere (ion ), troposphere
(tro ), Earths tide (tide ), multipath (mul ), relativistic effects
(rel ), receiver clock error (tr ), and satellite clock error (ts )
[4]. The geometric distance can be modified with these errors
as
R(tr , te ) = (tr , te ) (tr ts )c
+ion + tro + tide + mul + rel + 

(2)

A. GPS Trilateration and Propagation Errors


At least four satellites are required by the GPS receiver
to calculate three dimensional position data. The receiver
calculates the distances to each satellite and uses these distances to determine the receiver position in a process called
trilateration, as shown in Fig. 1. The distance to each satellite
is dependent upon the signal travel time, which is subject to
several propagation effects.

to determine the pseudorange, R(tr , te ), between the satellite and receiver. Here c denotes the speed of light in freespace and  denotes any residual measurement errors. In this
paper, we focus on the propagation errors introduced by the
ionosphere, troposphere, and multipath.

U. Chipengo, S. Wijesundara, S. Watt, and A. Balakhder are with The


Ohio State University Department of Electrical and Computer Engineering,
Columbus, OH, 43210.
Manuscript received December 9, 2015

When considering atmospheric propagation effects, we divide the atmosphere into the neutral atmosphere (up to about
100 km) and the ionosphere (about 60 - 2000 km):

B. Atmospheric Overview

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

vg =

df 2
c
=
.
d
ng

(5)

Following the derivation described in [2], the relationship


between phase and group velocity, known as the Rayleigh
equation, can be derived as
vp
.
(6)
d
Performing additional differentiation and substitution results
in a modified Rayleigh equation that relates the phase and
group refractive indices as
vg = vp

Fig. 2: Ionization of an oxygen molecule due to incoming solar


radiation [6].
1) Neutral Atmosphere: The lowest layer of the neutral atmosphere, known as the troposphere, has the most atmospheric
mass. Specifically, most of the atmospheric water vapor resides
in the troposphere, which is usually the largest source of
variable atmospheric delay [5]. In addition to water vapor,
gases and other particles in the neutral atmosphere affect GPS
signal propagation. Typical tropospheric delay in the zenith
direction is about 2 meters [4].
2) Ionosphere: Within the ionosphere, intense solar radiation ranging from ultraviolet (UV) to X-rays hits the atoms
and molecules. This solar radiation ionizes neutral molecules
such as oxygen, creating free electrons (shown in Fig. 2).
These free electrons and ions affect GPS signal propagation
in a mechanism known as ionospheric refraction. Generally,
the ionospheric effects vary greatly depending on time of day,
solar activities, and Earths magnetic field [4].
C. Electromagnetic Background
In the ionosphere, the total electron content (TEC) represents the total amount of free electrons along the signal path.
The TEC in the zenith direction can be defined using the
electron density, Ne , as
Z
T EC =
Ne dz.
(3)
zenith

Usually TEC is given in TEC units (TECU), where 1 TECU


is 1016 electrons per m2 . For satellites at zenith, total vertical
electron content (TVEC), which can be considered the total
overhead electron content, is used. The TEC directly affects the
observed signal error, ion , and consequently is an important
factor when determining GPS signal propagation effects. To
determine the TEC along a slanted signal path a mapping
function must be used [2]. The mapping function is discussed
further in Section II.
Propagation errors result from both refractive delays and
scattering delays. To examine refractive delays in the ionosphere, consider an electromagnetic wave propagating in space
with a frequency f , wavelength , and corresponding refractive
index n. The phase velocity is given as
c
vp = f =
.
(4)
np
For a group of waves, the propagation speed for the resulting
energy is defined as the group velocity:

dnp
.
(7)
df
The phase and group indices are directly related to the
electron density, Ne . Due to the different velocities, GPS signal
code measurements are delay and signal carrier phases are
advanced. Therefore, measured pseudoranges using the signal
codes are too long and measured pseudoranges using phase are
too short compared to the geometric range (Eqn. 1) between
the satellite and receiver [2].
In the neutral atmosphere, refractive delays are introduced
by gases and other particles depending on their permittivity
and concentration. In dry air, refractivity is proportional to air
mass in the signal path and can be predicted from atmospheric
pressure. These cases will be discussed further in Section III.
Atmospheric-induced propagation delays contribute significantly to GPS measurement errors [7]. The zenith path delay
(in mm) is expressed as
Z
ZP D =
N dz
(8)
ng = np + f

where N = 106 (n 1) is refractivity, n is the index of


refraction, and z is the vertical signal path in kilometers. This
zenith path delay can contribute range error to GPS receivers,
and can be modeled using measurements or stochastic methods
[8]. In the atmosphere, water vapor is typically the largest
source of variable atmospheric delay. Water vapor distribution
depends on changes in clouds, convection, and storms [5].
D. Dealing with Propagation Errors
Three carrier frequencies are generated by GPS satellites:
L1 (1575.42 MHz), L2 (1227.60 MHz), L5 (1176.45 MHz).
Refraction in the ionosphere is dispersive, and therefore it can
be corrected using dual-frequency GPS measurements. Dualfrequency processing using both L1 and L2 bands will allow
the TEC, and subsequently the refractive index along the signal
path to be calculated.
In non-dispersive mediums, such as the troposphere, corrections can be made using numerical and empirical propagation models (Section III-B, III-C). Multipath effects can
be accounted for using in-receiver signal processing known
as maximum-likelihood (ML) estimation techniques [9][10].
Spatial processing, using relatively directive antennas or corrugated ground planes, can also be used to mitigate multipath
effects (Section IV-C) [11][12].

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

II. I ONOSPHERIC P ROPAGATION


A. Introduction
Ionosphere is the ionized region in space generally defined
to be between 100 km and 1000 km in altitude. If not
understood and accounted for properly, it acts as one of the
major sources of error for earth-space communication links. Incoming photons from ultraviolet (UV) radiation split molecules
resulting in positively charged ions and negatively charged free
electrons. Even though the free electron generation process is
primarily driven by UV radiation, it is strongly affected by
factors such as space weather, time of day and solar cycle.
This results in fluctuations of the total electron content (TEC),
which is the total number of electrons in a vertical (zenith)
path of 1 m2 cross sectional area as defined earlier in 3. Spatial
fluctuation scales of the ionosphere can range from thousands
of kilometers to less than a meter, whereas temporal fluctuation
scales range from multiple years to less than an hour [13][14].
In fact, more than doubling of relative TEC values with respect
to background values have been observed over polar regions
in less than 10-15 minute time scales [15].
Since communication paths dont necessarily follow a vertical trajectory, a mapping function is defined to convert between
zenith and line-of-sight (LOS) TEC values as follows,
map(el) = p

R0
R02

Re2 cos2 (el)

(9)

where, el is defined as the elevation angle, and Re is defined


as the nominal earth radius at the ground station ( 6375
km). Assuming that the nominal height of the F-layer (where
maximum electron density can be observed) is approximately
350 km, R0 is defined as the height of the maximum electron
density ( 6375 km + 350 km) [13]. As a result, the zenith TEC
value can be obtained from LOS TEC (T ECLO S ) as follows
T EC =

T ECLO S
map(el)

(10)

Fig. 3: Observed TEC variations during a solar storm in April


2011 [16]

Figure 3, generated by the NASA/Goddard Space Flight


Center Scientific Visualization Studio [16], shows the variations in TEC over North America measured over a 4 hour
period during a solar storm in April 2001. Blue shades indicate
low TEC, red shades indicate high TEC, and gray indicates
no data. Variations in TEC over the signal path through the
ionosphere introduce errors such as scintillation, dispersion,
refraction and attenuation, to earth-space GPS communication
links.
B. Scintillations
Scintillation can be defined as rapid amplitude fading and
phase fluctuations in received microwave signals due to small
scale time varying irregularities in the ionosphere. Additionally, the frequency of the GPS signal compared to plasma
frequency strongly influences the degree to which scintillations
are present. As shown in figure 4 scintillation effects can be
frequently observed over equatorial (one hour after local sunset
to midnight) and auroral regions, and are mainly driven by solar tides, solar wind, and geomagnetic field lines [13][14][18].

Fig. 4: Global occurrence characteristics of scintillation [17] .

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

Approximately 11 year solar cycle and seasonal changes


are two of the key factors that strongly affect equatorial
scintillation activity. Figure 5 shows the variations in monthly
scintillation activity over Ancension Islands in the Southern
Atlantic Ocean between 1991 and 2000 and corresponding
sun spot number, which indicates the solar activity level.
Occurrences of deepest amplitude fades (greater than 20dB)
coincide with solar maxima. In addition, yearly maximum and
minimum fading levels approximately correspond to months
of February and July respectively. This is expected from an
observations station that is located in the southern hemisphere
[19][17].

GPS data recorded at Thule, Greenland reveals a high degree of


correlation between L1 and L2 links for individual amplitude
fades. Correlation coefficients can especially exceed 0.7 during
high scintillation time periods, which implies that a loss of
signal on one link could affect both L1 and L2 channels
[15]. Furthermore, [21] shows how the scintillation index
(S4 ) is affected by the irregularities introduced by ionospheric
plasma bubbles. Plasma bubbles are characterized by large
scale regions of plasma depletions that are approximately
aligned in the north-south direction. They are formed shortly
after local sunset, and can drift upwards to altitudes as high
as 1500 km at the geomagnetic equator [21].
C. Range Errors
In addition to loss of signal and phase lock, dispersive nature
of the ionosphere introduces range and range rate errors. As
described earlier, the group delay introduced by varying TEC
conditions over the signal path can result in errors ranging from
sub-meter to more than 100 meters. Furthermore, Rapid phase
changes in excess of receiver bandwidth can produce Doppler
shifts in excess of 1 Hz per second [14]. The range delay (R)
resulting from ionospheric effects can be approximated by,
R =

Fig. 5: Frequency of occurrence of scintillation fading depths


at Ascension Island versus season and solar activity levels [17]
.
Strong scintillations introduce a multitude of errors including signal loss, loss of lock and range and range rate errors.
Amplitude scintillations can be quantified using the S4 index,
which is defined as,
s
hI 2 i hIi2
S4 =
(11)
hIi2
where I is the square of the amplitude, and brackets h.i denote ensemble averaging. Phase scintillations are characterized
by the sigma-phi ( ) index, which is the standard deviation
of phase in radians. GPS phase signal can be approximated by
=

40.3
T EC
fc

(12)

where c is the speed of light in m/s ,and f is the frequency


of the signal in Hz[19].
According to [20], there is a high degree of correlation
between strong fading events and the rate of change in phase
during phase lock and signal loss events. Modeling simulations
conducted under S4 = 0.61, and r.m.s phase fluctuations
of 5.8 radians (corresponds to strong scintillations) show a
clear cross correlation minimum at 0 s, whereas simulations
conducted under S4 = 0.125, and r.m.s phase fluctuations
of 1.3 radians (corresponds to weak scintillations) show no
correlation. In addition, an analysis conducted using dual band

40.3
T EC
f2

(13)

where T EC is the total electron content, and f is the GPS


link frequency in Hz. Based on experiments conducted at MIT
Haystack Observatory, a single TEC unit (TECU) corresponds
a 0.162 meters of range delay on primary L1 link. Therefore,
if not properly corrected, an average daytime TECU value of
30 over Massachusetts could introduce range delay of over
4 meters [13]. This underscores the importance of mapping
and characterizing the TEC distribution. Dual band receivers at
various locations over the globe utilize nowcasting techniques
to produce high spatially and temporally resolved real time
TEC maps over the globe by cross correlating the received
GPS signal on L1 and L2 bands. One such example is shown
in figure 3. Since most low-end civilian GPS devices use single
frequency receivers (frequently, L1 band), ionospheric models
and numerical methods are used to minimize range errors [14].
Range delays have a strong dependence on the elevation
angle at which the GPS link is established. Smaller elevation angles result in longer path lengths through ionosphere
compared to larger elevation angles, which results in more
group delay effects affecting the GPS link. According to
simulation results shown in [22], a 10 degree elevation angle
could introduce as much as 70 mm of additional range error
compared to a 60 degree elevation angle over an equatorial
region.
D. Refraction
Ionospheric refraction is the bending of rays due to index
of refraction changes over the path length of the signal within
the atmosphere. Refraction errors introduce discrepancies in
apparent angle of arrival of the GPS signal, which translates to
errors in elevation angle. Even though tropospheric refraction

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

dominates ionospheric refraction, it typically introduces a


daytime error of 0.3 degrees over VHF frequencies between
0-5 degree elevation angles [13]. For low elevation angles, the
angular error (E in radians) can be expressed as
E =

cos(E0 )
R
2hi

(14)

where R is the range error calculated from 13, E0 is


the apparent elevation angle in radians, and hi is the centroid
height of the TEC distribution [13].
There are other error sources such as ionospheric attenuation
of signals, and Faraday rotation (as derived in [23]). Circularly
polarized radio frequency links are used to counter the effects
of Faraday rotation. Advancements in mapping the TEC distribution using dual frequency receivers, and improved modeling
and simulation techniques can be used very effectively to
remove majority of the errors introduced by the ionosphere.
III.

T ROPOSPHERIC P ROPAGATION

The troposphere is the lowest layer of the earths atmosphere. Strictly speaking, the troposphere extends from the
earths surface to a height of 9 km and 16 km at the poles
and equator respectively [24]. The stratosphere also contains
neutral atoms and molecules, however, its effects on signals
is less pronounced because 75% of atmospheric mass resides
in the troposphere. Therefore, the region of the atmosphere
extending from the earths surface to the ionosphere is loosely
referred to as the troposphere [25][24].
While the ionosphere contains charged ions and electrons, the troposphere mainly consists of neutral atoms and
molecules. These neutral particles affect electromagnetic wave
propagation by effectively delaying the signal[26][27][28].
Specifically, the troposphere presents itself as a non-dispersive
medium with a refractive index n > 1 at the GPS frequency.
The speed of the GPS signal in the troposphere is therefore
less than the speed of light in a vacuum (n = 1). Therefore,
the signal received from the GPS satellite is a delayed version
of the actual signal in the absence of the troposphere. This
delay can also be interpreted as a result of wave refraction
as the signal traverses the troposphere. The resulting path
due to refraction is longer than the straight path that the
wave would travel in a vacuum (see Fig. 6). The associated
delay which also depends on temperature, humidity, pressure
and location can range from 6 to 80 ns which corresponds
to a pseudo range error of 2 to 25 m.[23]. This error can
be significant or insignificant depending on the application.
Civilian applications have relaxed accuracy constraints while
25 m errors can prove detrimental for military applications.
As previously mentioned, dual carrier techniques can be
used to mitigate ionospheric delay. However, this technique
cannot be used to mitigate tropospheric delay since the troposphere behaves as a non dispersive medium for frequencies
below 15 GHz [23]. To mitigate tropospheric delays, predictive
models based on numerical analysis and empirical results are
used correct for the anticipated delay.

Fig. 6: Tropospheric refraction of GPS signals leading to delay


of signals [25] .

A. Zenith Tropospheric Delay


As previously mentioned, the excess delay experienced by
the signal is due to the refractive index n > 1 of the
troposphere. Therefore the tropospheric delay can be expressed
as [23]
Z
1
ta =
n(s)ds
(15)
c S
Where c is the speed of light in a vacuum. The integration is
done along the actual path S followed by the signal as shown
in Fig. 6. In the absence of the troposphere, the wave would
travel as though it were in a vacuum. Therefore, the delay that
the signal would experience is
Z
1
ds
(16)
ts =
c Ss
Here, the integration path is the geometric straight path
that the signal would travel in the absence of refraction. The
difference between these two delays gives rise to the excess
tropospheric delay. Therefore, the excess tropospheric delay
can be described as
te = ta ts

(17)

Using the equations 15 16 and 17 yields the following


equation [23]
Z
S
1
[n(s) 1] +
ds
(18)
te =
c S
c
Where S is the difference in path length between the
geometric path and the actual refracted path. The term from
equation (18) (n(s) 1) is described as the refractive index
anomaly. This leads to the tropospheric refractivity Nt =
106 (n(s) 1). This tropospheric refractivity can be split into
wet and dry refractivity. That is, Nt = Nw + Nd . This is
because 90% of the refractivity can be attributed to the dry
atmosphere, this corresponds to excess path lengths of 2.3 m
to 2.6 m at sea level. On the other hand, water vapour in the
atmosphere accounts for 10% of the total excess path length
corresponding to approximately 0.4 m [25].
While the dry refractivity is relatively easy to model due to
the predictable nature of atmospheric pressure, water vapour
does not follow the same trend. Specifically, it is much more
difficult to model the wet refractivity due to the unpredictable

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

temporal and spatial distribution of water vapour in the troposphere. The severity of this situation is mitigated by the
fact that the wet refractivity only accounts for a small amount
of the total refractivity. However, water vapour radiometers
(WVRs) are still used to measure the water vapour content in
the atmosphere for calibration purposes [24].
B. Modified Saastamoinen Model
If a quasi geometric assumption is made, the difference in
path length can be approximated to be zero in (18), S 0.
The excess delay can now be modeled as an excess distance
by using the following equation
= d + w = cte .

(19)

The modified Saastamoinen model (Saastamoinen 1972,


1973) uses the atmospheric pressure, temperature, humidity
and a reference height to predict the excess path length due to
the troposphere through the equation [4]


0.002277
1255
2
=
P +(
+ 0.05)e Btan + R (20)
cos
T
where is the zenith angle, P is the atmospheric pressure
(mb), T is the temperature in Kelvin. The partial pressure of
water vapour is given by e while B and R are correction terms
that depend on the height H and zenith angle . Variable e is
given by
e = Rh exp(37.2465 + 0.213166T 0.000256908T 2 ) (21)
where Rh is the relative humidity in %. For this model,
height dependence can be used to determine the pressure P ,
temperature T and humidity Rh as follows [4]
P = P0 [1 000226(H H0 )]5.225

(22)

T = T0 0.0065(H H0 )

(23)

Rh = Rh0 exp[0.0006396(H H0 )].

(24)

The standard height H0 = 0m, standard pressure P0 =


1013.25 mbar, standard temperature T0 = 291.16 K, and
standard humidity Rh0 = 50% are used in 22 to 24. Other
models have been used and are still implemented in different
cases.
One common model is the modified Hopfield model (Hopfield 1969, 1970, 1972). The modified Hopfield model takes
the approach from equation (19) in determining the total excess
path length by calculating the dry and wet components separately. Although this model is slightly more computationally
intensive than the Saastamoinen model, both models exhibit
small differences in their predictions for zenith angles less
than 75 degrees. Fig. 7 shows the excess path length against
the zenith angle obtained using the modified Hopfield model.
As seen, the excess delay increases as the zenith angle is
increased. This is because as the zenith angle is increased,
the physical distance that the wave has to travel under the
influence of the troposphere increases as well.

Fig. 7: Excess tropospheric path lengths for various Zenith


angles [4] .

C. Simplified Tropospheric Delay Models


The previously outlined models are quite accurate for most
applications. However, their tedious computations make them
unattractive for simple first principle calculations. A simple
model that provides a mapping function for the delay experienced at different zenith angles is given in [23]. Fig. 7 shows
that the minimum delay is experienced when the zenith angle
is 0 degrees i.e. when the satellite is directly overhead. A
mapping function is therefore needed to determine the actual
delay for a specific zenith angle. Using the standard U.S.
atmospheric conditions, the predicted zenith delay for both
wet and dry refractivity is given by [23]
= d + w = cte = 2.4405m.

(25)

For a specific zenith angle , the mapping function is given


by
1.0121
(26)
sin + 0.0121
Finally, for a receiver at sea level (h = 0), the total delay
for varying zenith angles is given by
m() =

2.47
m.
(27)
sin + 0.0121
For stations that are above sea level (h > 0), the total delay
is less than that observed at sea level since the tropospheric
distance traversed by the signal is significantly reduced at high
altitudes. Therefore, for a station at an altitude of h km , the
excess path is given by
()(h = 0) =

(h) = (h = 0)e0.133h .

(28)

D. Other Troposphere Related Propagation Phenomenon


In addition to tropospheric delay, when the refractivity
gradient dN/dh becomes less than -157 N-units/km, ducting
of the GPS signal occurs [29][30]. Specifically, the troposphere
acts as a waveguide through which the GPS signal propagates.
Ducting can lead to significant errors in range tracking by

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

or more different paths due to reflections and refractions. If


the link between the satellite and the GPS receiver is only
through direct path, the effect of multipath will be ignored
and the position of the receiver will be determined accurately.
However, due to the presence of huge objects, such as buildings
in urban environments, the received signal is exposed to
significant reflections and refractions. This section discusses
multipath that occurs near the receiver, which is different
from multipath that could happen in the media between the
satellite and the receiver, such as ionospheric and tropospheric
multipath.

Fig. 9: Multipath generated by the environment [31].

Fig. 8: Refractivity plots and propagation plots of GPS signals


in the a) absence and b) and presence of tropospheric ducts
[30].

changing the apparent elevation angle of a satellite. Specifically, a low elevation satellite can send a signal that gets
trapped in a tropospheric duct. Since the signal power in a
duct is significantly high, the ground receiver can interpret
the duct signal as a signal originating from another satellite
at a higher elevation angle than the actual satellite. This then
leads to significant range errors. Ducting of GPS signals can be
studied using ray tracing techniques (RT), parabolic equation
(PE) methods, and split step parabolic equations (SSPE). Fig.
8 shows how low level ducting occurs.
IV.

M ULTIPATH P ROPAGATION

GPS signals are like any other electromagnetic signals,


which experience multipath phenomenon in different environments. Multipath occurs when the signal is received from two

A. Effect on Pseudorange
The path of the reflected signals is always longer than the
path of the direct signal as shown in Fig. 9. The reflected signal
path depends on the position of the satellite and the structure
of the reflected object. This difference in the path length will
lead to change in the amplitude and phase for the same signal.
Therefore, receiving both direct and reflected signals can cause
increase/decrease in measured pseudorange [32], which is the
time difference of the encoded signal between the satellite and
the receiver.
P seudorange = (T ime Dif f erence) (Speed of Light).
(29)
The maximum pseudorange error could happen for P-code
measurements, and can be calculated theoretically as about
15 meters when the amplitude of the direct and the reflected
signal are equal. Moreover, because of the coded transmitted
signals, GPS receivers eliminate the delay signals caused
by the multipath. However, empirical experiments show that
in highly reflective environments, the error caused due to
multipath is in the range of 4 to 5 meters. For low reflection
environments it is in the range of 1.3 meters [32].
B. Effect on Carrier Phase
Multipath also affects the carrier phase, which produces a
phase shift on the received signal. This phase shift introduces
an error in the actual range in order of several centimeters
[33]. The direct and reflected signals can be represented in a
simple form as:

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

AD = A cos D .

(30)

AR = A cos(D + ).

(31)

where A is the amplitude of the direct signal, A is the


amplitude of the reflected signal, is an attenuation factor
(0 1). Here = 0 means no reflection signal is
present, and = 1 means the amplitude of the reflected signal
is as strong as direct signal. The phase of the direct signal is
represented as D , and the phase of the reflected signal with
respect to the direct signal is denoted as .
Adding both the direct and the reflected signals gives the
following:
A = AD + AR = A cos(D ) + A cos(D + ).

Fig. 10: Received and absorbed signal by Choke-Ring Antenna


[34].

(32)

with AD,max = A, AR,max = A.


The error in the carrier phase measurement due to the
multipath is:


sin
= arctan
.
(33)
1 + cos
The signal amplitude is:
B = A = A

1 + 2 + 2cos.

(34)

For strong reflection signals, = 1, the maximum value of


is 90 . For L1 signals ( = 19.05 cm), the maximum error
due to the change in the carrier phase is in the range of 5 cm
[33].
C. Multipath Mitigation Techniques
The simple way to reduce the effect of multipath of the
received signal is to locate the receiver in the free-multipath
environment. For example, if the receiver is located in the top
of a huge building, the received signal will be only through
the direct path. However, this is not always the case. Usually
the GPS receiver is located in an environment which has many
reflections. To reduce the effect of the multipath, there are two
main approaches: 1) improve the signal processing inside the
GPS receiver, and 2) multipath mitigation, which improves the
effect of multipath outside the receiver. This section discusses
some of the previous work done of the second approach, which
is multipath mitigation techniques. These techniques are primly
used to improve the receive of the direct signal and reduce the
effect of multipath.
1) Choke-Ring Antenna: Choke-Ring antenna is a special
type of an omnidirectional antenna. This type of antenna has
the ability to reduce the multipath error by placing radiofrequency absorbing material underneath the antenna. The
reflected signal from the ground or the surface under the
antenna is rejected by Choke-Ring antenna, Fig. 10. However,
this type of antenna cannot reject the multipath signals which
are reflected from high elevation angles. Therefore, it improves
the GPS accuracy if most of the reflected signals are ground
bounces [31].

Fig. 11: Geometry of multipath changes for different positions


of the satellite [33].

2) Long Term Signal Observation: As the satellite is moving, the geometry of multipath is changing as shown in Fig. 11.
This change also affects the carrier phase for typical periods
between 15 to 30 minutes, depending on the environmental
conditions [33]. This effect has a cyclic behavior as illustrated
in Fig. 12. Therefore, it is important to locate the multipath
positions by observing the measurements for a long time
period, more than an effective cycle, in order to identify the
signature of the multipath error and minimize it[35].
3) Using Signal-to-Noise Ratio: Since there is a relationship
between the carrier phase and the signal-to-noise ratio (SNR),
in terms of the location and the orientation of the reflector, it is
possible to estimate the phase multipath errors and identify the
geometry of the reflector using SNR. This mitigation technique
is based on empirical mapping between the angle and the
SNR measurement of the GPS signal[36].
V.

C ONCLUSIONS

The geolocation application of GPS depends on the accurate


determination of signal delay from satellite to user. Various

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

[8]

D. Tralli and S. Lichten, Stochastic estimation of tropospheric path


delays in global positioning system geodetic measurements, Bulletin
godsique, vol. 64, no. 2, pp. 127159, 1990. [Online]. Available:
http://dx.doi.org/10.1007/BF02520642
[9] M. Sahmoudi and M. Amin, A maximum-likelihood synchronization
scheme for gps positioning in multipath, interference, and weak signal
environments, in Vehicular Technology Conference, 2006. VTC-2006
Fall. 2006 IEEE 64th, Sept 2006, pp. 15.

Fig. 12: Calculated (Solid line) and observed multipath effect


[33].

[10]

J. Soubielle, I. Fijalkow, P. Duvaut, and A. Bibaut, Gps positioning


in a multipath environment, Signal Processing, IEEE Transactions on,
vol. 50, no. 1, pp. 141150, Jan 2002.

[11]

C. C. Counselman, Multipath-rejecting gps antennas, Proceedings of


the IEEE, vol. 87, no. 1, pp. 8691, Jan 1999.
F. Scire-Scappuzzo and S. Makarov, A low-multipath wideband gps
antenna with cutoff or non-cutoff corrugated ground plane, Antennas
and Propagation, IEEE Transactions on, vol. 57, no. 1, pp. 3346, Jan
2009.
A. Coster, Mitigation of ionospheric propagation errors with gps, in
Radar Conference, 2007 IEEE, April 2007, pp. 922926.
J. A. Klobuchar. (1991, April) Ionospheric effects on gps. [Online].
Available: http://goo.gl/EQlUBL
M. B. El-Arini, J. Secan, J. A. Klobuchar, P. H. Doherty, G. Bishop,
and K. Groves, Ionospheric effects on gps signals in the arctic
region using early gps data from thule, greenland, Radio Science,
vol. 44, no. 1, pp. n/an/a, 2009, rS0A05. [Online]. Available:
http://dx.doi.org/10.1029/2008RS004031
(2005, December) Ionosphere total electron content - april 2001.
[Online]. Available: http://goo.gl/ceiyWE

[12]

[13]
[14]
[15]

[16]

Fig. 13: Line of Site GPS and angle [36].


[17]

propagation phenomenon affecting the electromagnetic signal


can cause inaccuracies in the determination of this delay
and subsequent range. In this paper we have identified the
main propagation effects experienced by earth-space GPS
links. The ionosphere, troposphere and multi-path propagation
were identified as the main causes of error in pseudorange
determination in GPS . Various techniques for mitigating these
effects were discussed. Use of dual frequencies was applied
to mitigate ionospheric effects. Tropospheric effects were
mitigated using empirical and numerical predictive models
that anticipate the delay. Long term signal observance and
choke ring antennas were discussed as possible solutions for
multipath mitigation. Therefore, understanding and accounting
for propagation effects on GPS signals remains an integral part
in ensuring accurate GPS performance.
R EFERENCES
[1]
[2]
[3]
[4]
[5]

[6]
[7]

(2015)
Gps.gov:
Gps
overview.
[Online].
Available:
http://www.gps.gov/systems/gps/
J. C. B. Hofmann-Wellenhof, H. Lichtenegger, Global Positioning
System Theory and Practice, 5th ed. Springer, 2001.
M. Schmandt. Gis commons: An introductory textbook on geographic
information systems. [Online]. Available: http://www.giscommons.org
G. Xu, GPS Theory, Algorithms and Applications. Springer, 2007.
F. Solheim, Propagation delays induced in gps signal by dry air, water
vapor, hydrometeors and other particulates, Journal of Geophysical
Research, vol. 104, 1999.
J. Bohm and S. H, Atmospheric Effects in Space Geodesy. Springer,
2013.
C. Meertens, Antenna Type, Mount, Height, Mixing, and Snow effects
in High-Accuracy GPS Observations. National Academy Press, 1997.

(2012, April) How irregularities in electron density perturb satellite


navigation systems. [Online]. Available: http://gpsworld.com/gnsssystemsignal-processinginnovation-ionospheric-scintillations-12809/

[18]

H. J. Strangeways, Determining scintillation effects on gps receivers,


Radio Science, vol. 44, no. 1, pp. n/an/a, 2009, rS0A36. [Online].
Available: http://dx.doi.org/10.1029/2008RS004076
[19] P. M. Kintner, B. M. Ledvina, and E. R. de Paula, Gps and ionospheric
scintillations, Space Weather, vol. 5, no. 9, pp. n/an/a, 2007, s09003.
[Online]. Available: http://dx.doi.org/10.1029/2006SW000260
[20] V. Gherm, N. Zernov, and H. Strangeways, Rate of phase change and
fade depth on gps paths due to ionospheric scintillation, in Antennas
and Propagation, 2003. (ICAP 2003). Twelfth International Conference
on (Conf. Publ. No. 491), vol. 1, March 2003, pp. 413416 vol.1.
[21]

H. J. Strangeways, V. Gherm, and N. Zernov, Modeling and mitigation


of the effect of scintillations on gps, in ELMAR, 2007, Sept 2007, pp.
6973.

[22]

K. Nagarajoo and C. S. Fah, Significant of earths magnetic field and


ionospheric horizontal gradient to gps signals, in Space Science and
Communication (IconSpace), 2013 IEEE International Conference on,
July 2013, pp. 110114.
[23] C. Levis, J. Johnson, and F. Teixeira, Radiowave Propagation Physics
and Applications. John Wiley and Sons, 2010.
[24] R. B. Langley, GPS for Geodesy. Springer, 1998.
[25]

(2012, Feb) Tropospheric delay and gnss signals tutorial. [Online].


Available: http://goo.gl/WbcNQL
[26] G. Peng, M. Torre-Juarez, R. Farley, and J. Wessel, Impacts of
upper tropospheric clouds on gps radio refractivity, IEEE Aerospace
Conference, Jan 2006.
[27] R. Norman, J. Le Marshall, W. Rohm, B. Carter, G. Kirchengast,
S. Alexander, C. Liu, and K. Zhang, Simulating the impact of refractive
transverse gradients resulting from a severe troposphere weather event
on gps signal propagation, Journal of Selected Topics in Applied Earth
Observations and Remote Sensing, IEEE, 2015.
[28] J. Collins and R. Langley, Mitigating tropospheric propagation delay
errors in precise airborne gps navigation, Position Location and Navigation Symposium, IEEE 1996, 1996.

WIRELESS PROPAGATION AND REMOTE SENSING, NOVEMBER 2015

[29]

[30]
[31]

[32]
[33]
[34]

[35]
[36]

G. cheng Li, L. xin Guo, and Y. Liang, Simulation modelling the gps
signal propagation in the tropospheric ducts, Antennas Propagation
and EM Theory (ISAPE), 2010 9th International Symposium on, pp.
592595, 2010.
G. Balvedi and F. Walter, Gps signal propagation in tropospheric
ducts, International Union of Radio Science (URSI), 2008.
S. H. Byun, G. A. Hajj, and L. E. Young, Development and application
of gps signal multipath simulator, Radio science, vol. 37, no. 6, pp.
101, 2002.
P. J. Teunissen and A. Kleusberg, GPS for Geodesy. Springer Science
& Business Media, 2012.
G. Seeber, Satellite geodesy: foundations, methods, and applications.
Walter de Gruyter, 2003.
(2015)
Choke-ring-antenna.
[Online].
Available:
http://www.trimble.com/Infrastructure/Trimble-Choke-RingAntenna.aspx
L. Wanninger and M. May, Carrier-phase multipath calibration of gps
reference stations, Navigation, vol. 48, no. 2, pp. 112124, 2001.
P. Axelrad and C. P. Behre, Satellite attitude determination based on
gps signal-to-noise ratio, Proceedings of the IEEE, vol. 87, no. 1, pp.
133144, 1999.

10

You might also like