ARTIFICIALSATELLITES,Vol.47,No.

1–2012
DOI:10.2478/v1001801200101


GNSS RECEIVER ZERO BASELINE TEST USING

GPS SIGNAL GENERATOR

Magdalena Dród, Ryszard Szpunar

Department of Geodesy and Carography, Warsaw University of Technology
Pl. Politechniki 1, 00-661 Warsaw
[email protected], [email protected]

ABSTRACT. Validation of the performance of GPS receivers is crucial for many

applications. This paper presents testing GPS receivers with the use of a GPS signal
simulator. Unlike live testing, testing with simulators provides full control of simulated
satellite signals and simulation environmental conditions. Moreover, generating repeatable
signals makes possible to extend typical procedures. It allows to determine the absolute value
of the position precision for each receiver without relation to any other one.
Keywords: baseline test, GPS signal generator, validation, GNSS receivers
1. INTRODUCTION
It is common known that satellite technologies dominate the geodesy market. Moreover, the
GNSS technologies start to play a crucial role in many other application fields such as:
navigation, aviation, logistics, localization, for both civilian and military purposes, Leick
(1995). That creates the need for proper validation of GNSS receivers and systems. Testing of
GNSS receivers in regard to a specific application should be an initial stage of each project.
From economical prospective of companies it is more convenient to evaluate the receiver’s
performance under certain conditions before the complete utilization of the system. There is
also the need to control the quality of the geodetic measurements. Accomplishment of that is
impossible without proper validation and certification of receivers. Moreover, testing starts to
become a crucial operation considering a growing supply of cheap GNSS receivers which are
often provided without any detailed specification or form unreliable sources.
Unfortunately common tests using live signals can no longer be satisfactory. These tests
can only provide limited information because the live signals are highly variable and non-
repeatable for a receiver. It is impossible to evaluate an absolute receiver’s performance; the
precision/accuracy of a particular instruments is always obtained with respect to the other
receiver. Moreover, testing a receiver in different locations or under specific conditions is
very unpractical and expensive. These limitations can solve a signal simulation. GPS signal
simulator generates the same signals as those transmitted by GPS satellites, so the receiver
processes them in exactly the same way. The advantage of this approach is that the specific
set of signals can be presented to the receiver an unlimited number of times. Additionally, it is
possible to control the signal processing environment and to determine the receiver’s ability
to work under various conditions, locations, times and movements. It makes possible to
evaluate the receiver’s performance in real conditions, its precision and accuracy, and also to
determine its sensitivity to certain error sources and its software quality. The tests presented

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC

14


in this paper
p were carried
c out using
u live signals
s and those
t generrated by the GSG-54 Peendulum
GPS sattellite simullator, and appplied to diffferent typees of receiveers.
2. THE
E GPS SIGN
NAL SIMU
ULATOR
The GS SG-54 Penduulum is an 8-chanell GPS G constelllation simuulator (L1, C C/A), so it provides
p
an 8 sattellites simuulation. Mooreover, it iss possible to
o modify naavigation data, antennaa model,
atmosphheric modell, multipathh and more.. The user can c configuure the pre--defined trajjectories
(static, circles, recctangular) byb changinng parameteers such ass user posittion and tim me, and
upload own
o trajectoory in the NMEA
N standard formatt. The simulator is alsoo suited for accurate
testing of timing GPSG receivvers, there is
i an input for externaal synchronnization from m e.g. a
Cesium m or Rubidiuum clock. It can also woork as a pseeudo-satellitte.
3. MET
THODOLO
OGY
The fielld tests alloow to exammine the recceiver’s quaality in partticular spann of time an nd under
certain conditions. Unfortunately they do not prrovide infoormation abbout the reeceiver’s
mance (softw
perform ware and hardware
h performancee). An obtaained error can be caaused by
softwaree but also by
b influencees of atmospphere, multipath or sattellite consteellation. Thhe partial
solutionn is the zeroo baseline teest. In the zero
z baselin
ne test, Szpuunar et al. (2004), two or more
receiverrs are conneected to the same antennna/a signal simulator (Figure
( 1).

Figg.1. The connfiguration of receivers in the zeroo baseline teest.

The test gives an a impressiion of the observations
o s, noise chaaracteristicss, since all common
c
errors, like
l those duue to multippath, atmospphere, satelllite orbits and a clocks aare eliminateed in the
GPS baaseline processing. But still the deerived resultts can contaain the impaact of the geometry
g
of the saatellite consstellation orr the locatioon of receiv
ver, Parkinsoon et al. (19996). Moreo over, the
impresssion of the tested instrrument’s peerformance is always relative, inn other worrds, with
respect to the refereence receiver. To overrcome thesee disadvantaages, laboraatory tests sh hould be
conductted with hellp of the sattellite signall simulator.
In proposed
p appproach all receivers are
a connecteed to the saame signal simulator (the( zero
baselinee mode), so all of themm process the same set of
o signals. The
T mentionned set of signals
s is
presenteed to the recceivers a feew times (thhe measurem
ment is repeeated). In thhis way we obtain a

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC

 15


few sessions of theoretically exactly the same observations (considering one receiver).
Differencing observations between receivers but within the same session we can obtain the
relative accuracy and precision. Analyzing observations of the one particular receiver but
from all sessions allows to obtain the absolute accuracy and precision.
To track the receiver’s performance with the time, the processing of observations is done
in the kinematic mode. Due to the fact that the position is known in each epoch, the time
series are created. The useful tool to generate the time series is the Trimble Total Control
software (TTC). The kinematic processor in TTC performs the carrier phase reduction in two
steps; the first step is called double difference float solution and derives baseline components
and float phase ambiguities. The following step is ambiguity search and statistical tests like
Fisher and Chi-square tests are used to verify the ambiguity resolution. In the GPS software
Trimble Total Control is used the OTF mode in order to process the resolution, allowing the
determination of ambiguities without any static initialization. The ambiguities are propagated
throughout the whole tracking time of a satellite without loss of lock. In the last step the final
carrier phase positions are computed, Trimble Total Control (2002).
4. EXPERIMENTS
Two Trimble 4700 receivers and two Leica GPS GRX1230GG receivers were used in
experiments; for the sake of convenience, they are simply called: Trimble09, Trimble14,
Leica06, Leica07. All measurements were performed in the zero-baseline scheme with the use
of live and generated signals. All raw observations were processed with the help of the
software Trimble Total Control in order to calculate coordinates of baselines in relation to a
reference receiver. Obtained time series were filtered (the low pass filter – the moving
average, the window 15). The receivers have been tested on two aspect – precision and
accuracy.
x Precision
Firstly, the aim was to obtain results in terms of position repeatability, Szpunar et al.
(2007). The measurements in a static scenario were carried out with the use of the GPS
simulator. Five sessions were performed, all of them with the same ephemerides file on the
15th January 2011 and with the same conditions applied to the simulator. In order to obtain an
absolute position precision, for each receiver separately, four time series of zero-baselines
were generated (sessions 2-5 were related to the first session). Figure 2 presents obtained
results. To give the impression of satellite constellation’s influence, PDOP coefficients versus
time were also plotted there.

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC

16


Fig. 2. Time series of zero baselines (color plots) and PDOP versus time; observations of each
session related to observations of first session, separately for each receiver.

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC

 17


To analyse the consilience between sessions and further to obtain conclusions about
differences between receivers, correlation coefficients were calculated (Table 1). Firstly,
correlation coefficients between all series (all possible 2-element permutations) and then an
average coefficient for the pair of receivers were generated.
Table 1. Correlation coefficients for time series - precision.
L07 T09 T14
L07 0,85 0,72 0,63
T09 0,72 0,86 0,75
T14 0,63 0,75 0,83
PDOP 0,66 0,62 0,58

The auto-correlation coefficients of receivers are very similar. The lower consilience
presents the Trimble14 receiver what is also seen in Figure 2. The highest cross-correlation
coefficient have two Trimble receivers what is quite predictable. Values of cross-correlation
coefficients between different receivers are very similar to cross-correlation coefficients
between PDOP and receivers. This fact lets to conclude that the systematic part in coefficients
is caused mainly by the geometry of satellite constellation. The another part is related to the
repeatability of performance of each receiver.
Table 2. Statistical summary of results - precision.
Leica 07 Trimble 09 Trimble 14
serie serie serie serie serie serie serie serie serie serie serie serie
1 2 3 4 1 2 3 4 1 2 3 4
mean 0,8 0,8 0,8 0,8 2,8 2,7 3,3 3,5 3,0 2,8 4,0 4,0
[mm] 0,8 3,0 3,4
0,30 0,27 0,28 0,29 0,86 0,74 1,38 1,83 0,87 0,90 1,26 1,16
std [mm]
0,28 1,20 1,05

Table 2 contains the statistical summary of results. The mean values reflect the absolute
position precision of each receiver. The newer Leica07 receiver presents significantly better
quality than Trimble receivers.
x Accuracy
The aim of the second approach was to obtain some remarks about a relative accuracy of
receivers. The quality of the Trimble05, the Trimble14 and the Lecica06 receivers was
assessed in the zero-baseline test in comparison with the Leica07 receiver. Time series of
zero-baselines were generated separately for two sessions with the ephemerides file on the
15th January 2011 (observations of the Leica07 receiver subtracted from observations of the
other receivers, separately for each session). Results are shown in Fig. 3. Trimble receivers
have pointed out more significant “sensitivity” to the bad geometry of the satellite
constellation.

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC

18


Fig. 3. Time series of zero baselines (color plots) and PDOP versus time, observations of each
receiver related to the Leica07 receiver; separately for each session.

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC

 19


Table 3. Statistical summary of results - accuracy.

Leica 06 Trimble 09 Trimble 14

serie 1 serie 2 serie 1 serie 2 serie 1 serie 2
0,4 0,4 2,8 2,1 3,3 2,3
mean [mm]
0,4 2,5 2,8
0,18 0,18 1,34 0,63 1,16 0,73
std [mm]
0,2 1,0 0,9

It appears clearly (Table 3), that there is significant difference in the relative accuracy
between Leica06 and two Trimble receivers. The evaluation was made under the assumption
that the Leica07 receiver is the best quality. So it seems natural that the Leica06 presents
better results with the relation to the Leica07. It can lead to an incorrect impression that any
other reference receiver would give different results. But in fact, the obtained above cross-
correlation coefficients between different types of receivers are very similar. So there is no
significant difference in the way of performance between receivers. The type of receiver
should not have an influence on results. The choice of the reference receiver was based only
on the quality of receiver meaning mainly random errors. Within the tests basing on position
repeatability the Leica receivers have proved to maintain the highest precision comparing to
the Trimble receivers. Due to that fact, one of the Leica receivers – the Leica07 receiver has
been chosen as the reference receiver for analyzing the accuracy.
All experiments presented above were performed with the help of the signal simulator. To
prove reliability of tests based only on simulated signals, the same calculations were executed
but applying live and simulated signals. Within two sessions the live signal and the simulated
signal were applied, both of them with exactly the same ephemeris file on the 8th February
2011. Again, the quality of the Trimble05, the Trimble14 and the Lecica06 receivers was
assessed in the zero-baseline test in comparison with the Leica07 receiver. Results are
presented in Figure 4.

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC

20


Fig. 4. Time series of zero baselines (color plots) and PDOP versus time, observations of all
receivers related to the Leica07 receiver; separately for two series of live and simulated
signal.

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC

 21


Table 4. Statistical summary of results; live versus simulated signal.

Leica 06 Trimble 09 Trimble 14

real sim. real sim. real sim.
0,7 0,4 1,7 2,5 1,7 2,3
mean [mm]
0,5 2,1 2,0
0,17 0,12 0,28 0,56 0,27 0,46
std [mm]
0,1 0,4 0,4

The statistical summary presented in Table 4 shows that a way of signal generation does
not have significant impact. Also in this approach, the Leica06 receiver has better relative
accuracy. In case of the Trimble receivers, all values calculated with the help of the live signal
are lower than the other based on the simulated signal. Probably it is caused by the bad
signalto noise ratio (SNR) during the simulation. The variable which has to be implemented
before simulation is the power level of the signal, the proper value of it plays a crucial role.
During the experiment the choice of the power level was made basing on tests of a receiver’s
performance when applying different values of the power level. This is quite intuitive way,
but the best possible solution at that moment, could lead to too noisy signal. Furthermore
using splitter could have negative influence on survey by lowering the SNR ratio. But in case
of the splitter used during these all experiments this impact is mitigated thanks to the
construction of this splitter. The lowering the power level due to the splitting of signal is
compensated by an amplifier.
In case of the Leica06 receiver, the results form simulation are better. The Leica06 is
modern receiver, probably with better filters. But in fact, it is difficult to state if the receiver
Leica06 was working properly. Obtained time series are very short (Figure 4); the time of
experiment was more longer). So the results should not be trusted to make the final
conclusions. More tests ought to be carried out.
To sum up, the Leica receivers have shown significantly better quality of performance
than the Trimble receivers. Both analysis: precision and accuracy have proven the difference
between these two types of receivers on level 2-3 mm. That is probably mainly due to better
filters and algorithms in case newer Leica receivers. This result seems reasonable in case of
the kinematic solution wchich was applied within conducted experiments. But it should be
pointed out that considering static solution a user should obtain values of accuracy in size of
tenths of mm.
5. CONCLUSIONS
The results derived from conducted experiments generate the new direction for the testing of
receiver’s performance. The results obtained when applying simulated signals are almost the
same as these generated from live signals. So, the satellite signal simulator can be used for
evaluation of the receiver’s quality and performance. Moreover, applying the same signal a
few times makes possible to extend typical procedures. It allows to determine the absolute

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC

22


value of position precision for each receiver without relation to any other one. This is not
achievable when tracking only live signals.
The main advantage of the complementary approach (precision and accuracy) is that it
makes possible to separate systematic and random errors. Comparing sessions of the same
receiver gives the drawback about the random error and bugs in the receiver’s
software/hardware. Relating receivers within the one session allows to determine systematic
errors in performance, but only under the assumption that the reference receiver works
properly. Putting that all together provides full and more detailed information of the receiver’s
performance.
The disadvantage of the proposed methodology for observations processing is that the
additional software (TTC) is used for calculating of the position in the kinematic mode. So, it
must be taken into account that obtained results can be influenced by possible errors of this
software. It suggests the need to process raw observations directly.
Our tests show that the modern Leica receivers present better quality than the Trimble
receivers. They have better statistical summaries and also are less “sensitive” to the bad
geometry of the satellite constellation and noisy signals.

REFERENCES
Leick A. (1995) GPS Satelite Surveying, Second Edition, A Wiley-Interscience Publication,
New York.
Parkinson B.W., Spilker Jr. J.J. (1996) Global Positioning System: Theory and Applications,
American Institute of Aeronautics and Astronautics, Washington.
Szpunar R., Walo J., Szelachowski W. (2004) Analysis of the permanent GPS observations
applied for tracking the periodical changes of engineering objects, Reports on Geodesy,
Vol. 69, No. 2, 329-334.
Szpunar R., Walo J. (2007) Displacement monitoring of engineering objects using GPS-RTK
technique; EGU Congress “Geodetic and Geodynamic Programmes of the CEI” Vienna,
Austria 15-20 April 2007, Reports on Geodesy, Vol. 83, No. 2, 83-87.
Trimble Total Control, User Guide, version 2.7, (2002).

Received: 2011-10-11,
Reviewed: 2012-02-09, by M. Figurski,
Accepted: 2012-02-16.

Unauthentifiziert | Heruntergeladen 07.11.19 05:49 UTC