International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 2
Number 3
September 2008
235
An Effect of Urban Development on the
Accuracy of the GPS/EGNOS System
R. Bober, T. Szewczuk & A. Wolski
Maritime University of Szczecin, Szczecin, Poland
ABSTRACT: This article presents the results of research into the accuracy of position determination by a
GPS/EGNOS system in a densely built-up residential area. These results are compared with the relevant
results obtained for a position in an open area.
1 INTRODUCTION
The EGNOS system makes up a superimposition for
the GPS and GLONASS systems which allows to
enhance the accuracy, availability, reliability and
continuity of position determination.
The accuracy of position determination in the
GPS/EGNOSS system is about 3-5 m (p=0.95) and
depends on a number of factors. The number and
configuration of observable satellites is one of the
factors significantly affecting the accuracy. These
depend on the position of the antenna and its
environment. The antenna location has an influence
on the number of tracked satellites, while the
environment may cause the effect of multitracking.
This article attempts at assessing the accuracy of
position determination by a GPS/EGNOS system in
an urban densely built up area and at comparing it
with positions obtained in an open area.
2 RESEARCH AREA
Three characteristic measurement locations were
chosen within Szczecin area (Fig. 1) for the
assessment of position determination accuracy
obtained with the use of a GPS/EGNOS system:
− measuring point Glinki(A), located in an open
area without buildings, with grassy ground,
offering good conditions of signal reception;
− measuring point ul. Bolesława Śmiałego(B),
located in a narrow street with rows of adjacent
buildings about 25 metres high. The antenna was
installed on a car roof (at height of 2 metres): the
celestial sphere was considerably covered by
buildings, so reflections from walls could be
expected;
− measuring point AM Szczecin(C), located on an
antenna platform, about 35 metres above the
ground, overlooking the roofs of neighbouring
buildings; the antenna could see the whole
celestial sphere, although there were chances for
reflections from neighbouring roofs and slight
shades from radio antennas at a distance of 2 to 4
metres.
236
Fig. 1. Szczecin – measuring points: A – Glinki, B –
ul. Bolesława Śmiałego, C – AM Szczecin
The three different measuring points are located
in conditions characteristic of most measurements
performed by GPS/EGNOS receivers. The
measuring point Glinki is very good measuring
location, typical of field measurements; the location
at AM Szczecin (Maritime University) well
represents conditions found on board ships, while
the point at Bolesław Śmiały Street is a typical
location for measurements carried out in a city
traffic.
a)
b)
c)
Fig. 2. Dispersion for average position in measuring points: a)
Glinki, b) ul. Bolesława Śmiałego c) AM Szczecin
(date: 2006-10-26)
a)
237
b)
c)
Fig. 3. Deviation for average position in measuring points:
Glinki, b) ul. Bolesława Śmiałego, c) AM Szczecin
(date: 2006-10-26)
a)
b)
c)
Fig. 4. HDOP factor in measuring points:
a) Glinki, b) ul. Bolesława Śmiałego c) AM Szczecin
(date: 2006-10-26)
3 RESEARCH METHODS
In order to find out the position determination
accuracy obtained by the GPS/EGNOS system, and
the relation of the accuracy with measurement
location, parallel measurements were performed
using three identical CSI-made MiniMAX receivers.
The use of the same type receivers (with the same
software) for investigations was supposed to
eliminate possible errors resulting from the use of
various type receivers for measurements. The
receivers were handled using PocketMax
PC_Ver.2.0 software, enabling the choice of
recorded data from MiniMAX receivers by PC
computers. Three measurement sessions were
executed on 10 October, 18 October and 26 October
2006.
The following parameters were measured and
logged at 1 Hz frequency:
238
GPGGA – moment, latitude and longitude,
height, HDOP, number of satellites tracked,
correction age;
GPGGL – latitude and longitude, moment,
status;
GPGSA – position mode (2D/3D), (M/A),
PDOP, VDOP.
4 THE RESULTS
The calculations for each session shown in Table 1
were made on the basis of recorded data from three
field locations:
− mean latitude (φ),
− mean longitude (λ),
− mean height (H),
− radius of error circle (M) (p=0.95),
− parameters of error ellipsoid (a, b, α) ( p=0.95).
The measurement data, shown in Table 1, are
additionally presented graphically:
− dispersion for average position in each measuring
point (Fig.2),
− deviation for average position in each measuring
point (Fig.3),
− values of the HDOP coefficient (Fig.4).
An analysis of data in Figures 2 and 3 shows that
at the locations with similar reception conditions
(Glinki, AM Szczecin) deviations from the mean
position are small and do not exceed 2.5 metres.
On the other hand, in a densely built up area
(location ul.Bolesława Śmiałego) values of
maximum deviations reach as much as 380 m.
An assessment of the HDOP (Fig.4) has shown
that at ul.Bolesława Śmiałego this coefficient at
short time intervals has very large values reaching
1000, which considerably decreased position
accuracy.
5 CONCLUSIONS
The research has clearly proved that the accuracy of
position obtained from a GPS/EGNOS system
depends on the antenna location, as this strictly
relates to the number of observed satellites. If the
receiver antenna observed the whole celestial sphere
(Glinki, AM Szczecin), position determination
accuracy was M = 0.72-2.57 m. When the receiver
worked in a densely built up area (ul.Bolesława
Śmiałego), then the accuracy was visibly lower
(M=33.26-58.72 m). The data recordings featured
very short but significant oscillations of the HDOP
coefficient that could not be explained by satellite
constellation. Besides, signals from EGNOS
satellites were disappearing for significant periods of
time during the measurements.
Table 1. Measurement results
ul. B. Śmiałego
AM
Glinki
ϕ [Ν]
53° 25' 59.74"
53° 25' 44,93"
53° 30' 16,66"
λ [Ε]
014° 32' 13,51"
014° 33' 49,19"
014° 36' 14,86"
H [m]
44,29
44,95
81,86
M(95%) [m]
33,26
1,12
2,57
a [m]
18,89
0,53
1,47
b [m]
3,51
0,36
0,13
α [ ° ]
31,3
16,7
22,2
ϕ [Ν]
53° 26' 00,18"
53° 25' 44,93"
53° 30' 16,60"
λ [Ε]
014° 32' 13,97"
014° 33' 49,15"
014° 36' 14,81"
H [m]
8,32
44,88
82,54
M (95%) [m]
58,72
0,72
1,16
a [m]
33,84
0,37
0,61
b [m]
2,39
0,18
0,27
α [ ° ]
19,7
60,3
11,6
ϕ [Ν]
53° 25' 59,86"
53° 25' 44,87"
53° 30' 16,60"
λ [Ε]
014° 32' 13,76"
014° 33' 49,15"
014° 36' 14,81"
H [m]
24,91
44,87
82,30
M (95%) [m]
33,37
1,45
1,16
a [m]
18,88
0,76
0,59
b [m]
3,91
0,35
0,34
α [ ° ]
20,5
33,4
4,3
REFERENCES
Banachowicz A., Bober R., Banachowicz G., Wolski A., 2005,
Spatial Accuracy of GPS/DGPS Position in a Fairway,
Reports on Geodesy No. 2 (73), Warsaw 2005.
Januszewski J., 2004, GPS and other satellite systems in
marine navigation (in Polish), Akademia Morska, Gdynia
2004.
Felski A., Specht C.,1995, Some aspects of optimized use of
DGPS systems in the Southern Baltic Sea, (in Polish)
Zeszyty Naukowe AMW no 3, Gdynia 1995.
Specht C., 1997, Polish DGPS systems as an element of
radionavigational shield of the southern Baltic (in Polish),
Zeszyty Naukowe AMW no 2, Gdynia 1997.