International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 2
Number 1
March 2008
37
Accuracy Analysis of the EGNOS System During
Mobile Testing
E. D. Aguilar, L. Jaworski & M. Kolodziejczak
Space Research Center, Warsaw, Poland
ABSTRACT: The European Geostationary Navigation Overlay Service or EGNOS has been operational and
broadcasting signals on PRN120 and PRN126 as of July 2006. EGNOS is designed to broadcast embedded
correction signals in Europe which will provide improved performance with GPS. There are three EGNOS
geostationary satellites PRN codes 120,124 and 126. The three satellites have positions 15.5 degrees west fro
PRN 120, 25 degrees east and 21.5 degrees east for PRN’s 126 and 124. Satellite PRN124 is still in the test
phase. The Space Research Centre has performed a mobile testing to demonstrate how EGNOS improves GPS
accuracy. A mobile GPS laboratory was used to collect GPS data in the rural outskirts of Warsaw. Two
receivers and one antenna were used to collect kinematic and navigation data. Post-processed GPS position
and height deviations are compared to the solution given when GPS is augmented with EGNOS. The
summarized results in this paper show that GPS augmented with EGNOS greatly improves position accuracy.
1 INTRODUCTION
The EGNOS overlay satellite system provides
improved GPS performance in the areas of accuracy,
availability, continuity and integrity. The objective
of the research was to collect GPS and EGNOS data
over short distances in clear surroundings. Three
GPS monitors were mounted within the mobile test
van. GPS and GPS with EGNOS data were
collected. Static EGNOS and GPS data was also
collected and used for post processing test data.
This paper details the preliminary results and
analyzes the horizontal and vertical accuracies from
each data set.
2 TEST SETUP
2.1 Equipment
The equipment used in the mobile GPS laboratory is
detailed in table 1.
Table 1. Test Equipment
__________________________________________
GPS Receiver Antenna
__________________________________________
Trimble 5700 Trimble Zephyr
Septentrio PolaRX2 Aero AT 2775
CSI N/A (navigation)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
Dell and HP Notebook
__________________________________________
2.2 Setup
The test equipment consisted of three GPS receivers,
two rooftop antennas and two notebooks for data
acquisition. The Trimble Zephyr Antenna was
magnetically mounted on the van rooftop and cable
attached to the Trimble 5700 receiver. The Trimble
receiver processed and stored data in RINEX format.
The Aero antenna was connected to the Septentrio
receiver. The GPS data from the Septentrio receiver
was collected on the HP notebook in RINEX and
NMEA format. The program RXControl 2.6 was
used for data acquisition on the HP. The third
navigation receiver, CSI, downloaded GPS and
38
EGNOS data to the Dell notebook in NMEA format.
GPS data was collected at one minute intervals
throughout the test route. The three receivers were
mounted on a platform towards the center of the van.
3 TEST RUN
The test run route was formulated to include a
mostly rural route with a few kilometres of urban
areas. The rural route gave clear GPS conditions
and minimized the time of unavailability. The 12
kilometre route took approximately ten minutes.
The route was repeated six times for the various
configurations.
An attempt was made to keep a constant velocity
of about 50 km/hr. However, various outside factors
prevented a constant velocity throughout the full
test.
4 ANALYSIS OF RESULTS
4.1 Post-processing
GPS kinematic data from the Trimble and Septentrio
receivers was post processed with Trimble
Geomatics Office. The DGPS post-processed data
was then used as a baseline against the navigation
data from the Septentrio and CSI receivers.
4.2 Satellite Availability
Satellite availability for each test run is graphically
represented above each deviation plot. During times
of satellite availability the number of satellites
averaged between six and nine. Data was masked at
0 and 15 degrees. Satellite availability was slightly
reduced by about two satellites at a 15 degree mask.
The gaps in data represent the times when it was not
possible to compare the kinematic solution to
navigation data.
4.3 Coordinate Deviation Percentages
The vertical and horizontal accuracy from the
Septentrio navigation data was analyzed using the
kinematic solution as the zero baseline. The
summary of Septentrio accuracies with a 15 degree
mask is listed in table 2. The Septentrio GPS data
was augmented with EGNOS PRN 126.
The GPS only x-coordinate data had nearly 16%
of the data within one meter accuracy. When
augmented with PRN126 over 40% of the data was
within one meter accuracy. Nearly 100% of the
EGNOS augmented data was within 1.50 meter
accuracy.
The y-coordinate also improved in accuracy when
augmented with PRN126. Nearly 100% of the data
was within 0.50 meter accuracy when augmented.
The height measurements also improve with
PRN126. The accuracy improves from 50% to 85%
for data within 0.50 meters. The accuracy of data
within the range of 0.00 to 0.25 meters is more than
double when augmenting GPS with PRN126.
The summary of Septentrio and CSI receiver
coordinate accuracies with a zero degree mask is
listed in table 3. A bigger improvement in data
accuracy was seen by the CSI receiver. The x-
coordinate deviation improved from 5% to 60% for
data within 0.50 meter accuracy. The y-coordinate
improved from 30% to nearly 90% for data within
0.50 meter accuracy. The Septentrio receiver had
almost the same performance when comparing GPS
only and GPS + PRN126 deviations. A major
improvement can be seen in the x-coordinate in the
0.00 to 0.25 meter range.
Table 2. Accuracy of Septentrio Receiver
Receiver Septentrio PolaRx2, Mask: 15 degree
X-coord (m)
Only GPS
GPS + PRN 126
<2.25; 2.00)
13.90%
0.00%
<2.00; 1.75)
14.95%
0.00%
<1.75; 1.50)
3.16%
0.19%
<1.50; 1.25)
21.05%
2.64%
<1.25; 1.00)
31.16%
54.34%
<1.00; 0.75)
12.84%
37.55%
<0.75; 0.50)
2.95%
4.53%
<0.50; 0.25)
0.00%
0.76%
< 1.00
15.79%
42.83%
Y-coord (m)
<1.50; 1.25)
0.84%
0.00%
<1.25; 1.00)
13.90%
0.00%
<1.00; 0.75)
32.21%
0.57%
<0.75; 0.50)
25.26%
0.76%
<0.50; 0.25)
17.47%
39.25%
<0.25; 0.00)
10.31%
59.44%
<0.50
27.79%
98.69%
Height
<2.25; 2.00)
0.21%
0.00%
<2.00; 1.75)
4.42%
0.00%
<1.75; 1.50)
5.05%
0.00%
<1.50; 1.25)
7.37%
0.00%
<1.25; 1.00)
13.05%
0.76%
<1.00; 0.75)
6.95%
3.02%
<0.75; 0.50)
13.48%
10.57%
<0.50; 0.25)
25.68%
26.42%
<0.25; 0.00)
23.79%
59.25%
<0.50
49.47%
85.66%
39
Table 3. Accuracy of Septentrio and CSI Receivers
Receiver Accuracies, Mask: 0 degree
X-coord (m)
Only GPS
GPS + PRN 126
Septentrio
CSI
Septentrio
CSI
<2.25; 2.00)
0%
1.01%
0%
0.17%
<2.00; 1.75)
0%
1.34%
0%
2.40%
<1.75; 1.50)
0%
1.68%
0%
3.94%
<1.50; 1.25)
0%
4.20%
0%
3.94%
<1.25; 1.00)
5.09%
30.37%
1.55%
5.14%
<1.00; 0.75)
29.83%
41.11%
9.14%
5.48%
<0.75; 0.50)
39.83%
15.77%
28.28%
20.03%
<0.50; 0.25)
22.71%
4.20%
30.35%
17.47%
<0.25; 0.00)
2.54%
0.34%
30.69%
41.44%
< 0.50
25.25%
4.53%
61.03%
58.91%
<1.00
94.92%
61.41%
98.45%
84.42%
Y-coord (m)
Only GPS
GPS + PRN 126
Septentrio
CSI
Septentrio
CSI
<1.50; 1.25)
0%
1.68%
0%
0%
<1.25; 1.00)
0%
0.34%
0%
0.34%
<1.00; 0.75)
0.34%
17.95%
0.52%
4.62%
<0.75; 0.50)
3.39%
47.32%
3.45%
6.51%
<0.50; -0.25)
21.02%
29.53%
43.79%
28.25%
<0.25; 0.00)
75.25%
3.19%
52.24%
60.27%
<0.50
96.27%
32.72%
96.03%
88.53%
4.4 Deviation Plots
The following deviation plots visually exemplify the
percentage results from the previous section. The
top bar graph has two sections. Each section shows
the number of satellites available for that specific
configuration. Data is only shown when the
kinematic solution was available.
Figures one and two are the x and y coordinate
deviations of the Septentrio receiver. The black line
is GPS data augmented with PRN 126. The
deviations from the kinematic solution improve
when GPS is augmented. A clear improvement can
be seen in the y-coordinate deviation plot.
Figure three is the height plot deviation for
the Septentrio receiver at a 15 degree mask.
Improvements can be seen throughout the plot,
specifically from seven minutes till the end.
Figures four and five show the x and y coordinate
deviations for the CSI receiver configured to a zero
degree mask. Both graphs show improved
performance when using GPS and PRN126 data.
Also, a small spike is prevalent in both graphs at the
five minute mark when using GPS and GPS with
PRN126. At this point the GPS only data improves
from 0.50 meter deviation to nearly zero. The GPS
with PRN126 data jumps from about 0.25 meter to
a one meter deviation. Overall, accuracy is greatly
improved as shown in table three.
Figures 6 and 7 are the x and y coordinate
deviations for the Septentrio receiver configured to
a zero degree mask. Overall, as seen in table 3,
deviations are nearly the same when comparing GPS
only and the GPS with PRN126 data. From seven
minutes to the end, Figure seven shows better
accuracy for the GPS only data and Figure six shows
improved accuracy for GPS with PRN126 data. The
difference between the two is within 0.50 meters and
is not a major concern.
Fig. 1. Septentrio X Coordinate Deviation, 15 Deg. Mask
Fig. 2. Septentrio Y Coordinate Deviation, 15 Deg. Mask
Fig. 3. Septentrio Height Deviation, 15 Deg. Mask
40
Fig. 4. CSI X Coordinate Deviation, 0 Deg. Mask
Fig. 5. CSI Y Coordinate Deviation, 0 Deg. Mask
Fig. 6. Septentrio X Coordinate Deviation, 0 Deg. Mask
Fig. 7. Septentrio Y Coordinate Deviation, 0 Deg. Mask
4.5 EGNOS and GPS Static Comparisons
Static data used for post-processing was collected
using two Septentrio receivers. The vector between
the two receivers was seven meters; therefore, the
constellation seen by the two receivers was identical.
The next six figures directly compare the deviations
of navigation and post-processed data between GPS
vs. EGNOS augmented and PRN120 vs. PRN126.
Figures 8-10 compare GPS results with EGNOS
PRN126. When using GPS augmented with
EGNOS, fewer satellites were available for data
collection. The number of GPS satellites ranged
from seven to fourteen and only four to ten GPS
satellites with EGNOS corrections were available.
The x and y coordinate data were within two
meters of accuracy. There were a few data spikes in
the y-coordinate for PRN126 that can be attributed
to satellite availability. Overall, the GPS data
augmented with EGNOS provided a slight
improvement in accuracy.
4.6 EGNOS PRN120 vs. PRN126
Figures 11-13 compare the static data for the two S-
bus EGNOS satellites 120 and 126. Data plots from
both satellites gave near mirror-image results in the
horizontal and vertical directions. The gaps in the
charts that show the times when EGNOS corrections
were not available.
Fig. 8. GPS vs. PRN126, X-Coordinate
Fig. 9. GPS vs. PRN126, Y-Coordinate
41
Fig. 10. GPS vs. PRN126, Height
Fig. 11. PRN 120 vs. 126, X-Coordinate
Fig. 12. PRN 120 vs. 126, Y-Coordinate
Fig. 13. PRN 120 vs. 126, Height
5 CONCLUSIONS
Results from the test run were as expected with the
GPS/EGNOS data clearly improving horizontal and
vertical accuracy. At a 15 degree mask the Septentrio
deviations improved significantly when GPS was
augmented with PRN126. The x-coordinate deviation
from one meter improved from 15% to 43%. In the
y-coordinate accuracy improved from having 30% to
nearly 100% of the data within a 0.50 meter
deviation. For the height, the number of data points
within 0.50 meters of zero deviation improved from
50% to 86%.
The next configuration analyzed was with each
receiver having a zero degree mask. The biggest
improvement was seen in the CSI receiver when
augmented with EGNOS PRN126. The x-coordinate
deviation improved from having 5% to 60% of the
data within 0.50 meters. For the y-coordinate, it
improved from 32% to 88%. The Septentrio receiver
had mixed results. In the x-coordinate, data
improved from 25% to 60% of the data within 0.50
meter deviation. In the y-coordinate there was no
overall improvement, the two data sets were equally
within 0.50 meters throughout the test.
EGNOS has been shown to improve horizontal
and vertical GPS accuracy during periods of
availability. EGNOS PRN 120 and 126 have similar
accuracy performance in the horizontal and vertical
coordinates. In order to get an accurate representation
of GPS/EGNOS performance more test runs are
planned over the next few months. Continued
observations throughout the following months will
be compared to these preliminary results in order to
better understand deviations in performance.
REFERENCES
Januszewski J., 2006, Systemy Satelitarne GPS Galileo i inne.
Wydawnictwo PWN Warszawa.
Parkinson W. & Spilker J., 1996, Global Positioning System:
Theory and Applications, Vol. I & II
Seeber G., 1993, Satellite Geodesy.
