International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 1
Number 2
June 2007
137
Software Navigation Receivers for GNSS and
DVB
F. Vejrazka, P. Kovar, M. Eska & P. Puricer
Czech Technical University in Prague, Prague, Czech Republic
ABSTRACT: We describe the software GNSS receiver, its schema, implementation into a computer, results
of tests and application for railway, municipal transportation and for shipping of dangerous matters. The
receiver, originally for the Galileo system, is on a printed board which is the size of a Euro Card (160?100
mm). Because the Galileo signal is not in the air, it was modified for the GPS and GLONASS systems.
Experimental GNSS receiver (EGR) was used as a tool for its development and it is also described. Even if
we use the receiver which is able to process signals of all three systems, it is impossible to ensure reception of
GNSS signals in adverse conditions (under leaves canopy, in urban canyons, in hollow tracks, etc.). Therefore
we have studied the possibilities of communication systems which will use modern signals known from
satellite navigation and we have obtained very interesting results when we used DVB-T transmitters as
beacons.
1 EXPERIMENTAL GNSS SOFTWARE
RECEIVER (EGR)
Requirements of the Czech Ministry of Transport
have led us to development of a Galileo navigation
receiver. After analysis of state of Galileo system
and other GNSS navigation systems we formulated
our own requirements on the receiver which have
been as follows:
1 The processing of all GPS, GLONASS, SBAS
and Galileo signals
2 High flexibility and rapid implementation of the
new signal processing
3 Enough performance for the very complex signal
processing
Those requirements can be satisfied by the
software defined radio (SDR) architecture. The
principal schema of the software defined radio is
shown on Fig. 1.
RF front
end
Analog
to
Digital
Convertor
Programable
logic
Computer
Fig. 1. Block diagram of the SDR receiver solutions
Requirements mentioned above have led to
receiver concept which will allow to process signals
with wide frequency bandwidth. The signal samples
are firstly processed in the programmable logic,
where the bandwidth of the signal is reduced and
then the signal is processed in a computer. Both
programmable logic and the computer can be parts
of a FPGA.
Demand of flexible development of hardware and
software of the receiver has tended to an experimental
hardware platform which would allow very simple
replacement of particular components and blocks
(radio frequency filters, etc.) and allow very easy
creation and realization of signal processing
algorithms.
138
The block diagram of the experimental GNSS
receiver is in Fig. 2. The receiver consists of three
parts: radiofrequency unit, DSP unit and PC
workstation.
The radio frequency unit consists of four
independent radio channels which can operate at any
frequency in range 1 2 GHz. The bandwidth of
each channel is 24 MHz; the RF unit supports active
and passive GNSS antennas. The intermediate
frequency is 140 MHz, gain of the receiver can be
controlled either by AGC (>40 dB) loop, either via
external input signal by DSP. The output signal of
radio frequency unit is digitalized by 8 bit A/D
converter with sampling frequency 80 MHz and the
remaining signal processing is performed in
programmable digital hardware. The advanced FPGA
Virtex II Pro by Xilinx with integrated PowerPC
processors is applied and placed in prototyping board
which has reduced technological demands for FPGA
board development and construction.
Main task of connected PC is to be user interface.
It also translates programs in Simulink language,
runs them, loads them into FPGA and serves
as display and control unit for their verification in
EGR.
LNA
Channel 1
Cannel 2
LNA
A/D
Virtex II Pro
Prototyping
Board
DSP Xilinx
DSP Unit
Radio Frequency Unit
GNSS antenna
Synthetiser
PC workstation
Cannel 3
LNA
Cannel 4
LNA
Fig. 2. Block diagram of the Experimental GNSS Receiver
(EGR)
The receiver is equipped with switched power
supplies and high precision frequency reference with
stability 0.03 ppm. The complete receiver mounted
into the 19-inch rack is in the Fig. 3. The receiver is
capable to process the all known navigation signals
except of the Galileo E5 signal. The concept of the
modernized version of the experimental receiver
with higher performance and capable to process
Galileo E5 signal was prepared; receiver is currently
in prototype realization state.
EGR has served as a development tool and has
been used for development of the GNSS receiver for
railway applications.
Fig. 3. Experimental GNSS Receiver in 19-inch rack
2 GNSS RECEIVER FOR RAILWAY
APPLICATIONS
The Czech Republic is characteristic by its dense
railway network because of the long tradition of this
kind of transport. Besides of the primary network
(railway corridors), which is equipped with the
railway signalling interoperable with the systems
used by surrounding countries and EU, the
secondary railway network is usually safeguarded
with the national non compatible systems. The use of
GNSS in safety railway system would bring many
savings.
Since the position information is to be used by
railway station equipment for traffic control and
guaranty of safety in appropriate area, the GNSS
receiver as a source of information has to meet
reliability, integrity and safety requirements based on
common standards for signalling systems (EN
50129).
The main problem of the use of commercial
GNSS receiver in railway operations is verification
of proper function of its software. Therefore the
Ministry of Transport decided to order design of the
receiver which algorithms could be documented and
proved by a simple way.
The hardware of the designed receiver supports
reception and processing of signals of all three
systems (GPS L1 C/A, GPS L2C, GLONASS,
Galileo E1 and WAAS/EGNOS). Receiver is built
on PCB of Euro Card size (160×100 mm, Fig. 4),
designed for industry temperature range -40 ÷
+80 ºC and the mechanical construction meets the
standards for rail signalling systems. The circuit
design utilizes solely modern 3.3V technology.
Consumption of the receiver is approximately 4W.
Receiver cooling is ensured by the passive cooler
with no rotating parts. The receiver was successfully
tested as the GPS receiver by the Spirent simulator.
139
Fig. 4. PCB of the GNSS receiver for railway application
3 DVB T AS A NAVIGATION MEAN
There are problems with the use of GNSS (signal
availability and signal power level) in so called hard
conditions: under leaves canopy, in urban
environment, in hollowed tracks and indoor. Even
various methods have been developed to process
weak GNSS signals the reliable reception is not
satisfied. One method - Assisted GPS (AGPS) is
closely integrated with communication infrastructure.
It can imply an idea to use it for navigation; there are
a lot of mobile phones providers that use their
networks for user position determination. They use
Signal Strains (SS), Time of Arrival (ToA) or
Differential Time of Arrival (DToA) navigation
methods. But geometry of mobile phones base
stations and their signal properties don’t satisfy
required precision of position.
Properties of signals of DVB-T transmitters are
similar to ranging signals of GNSS systems. We will
deal with the exploitation of the DVB-T signal for
position determination using ToA or DToA
navigation methods. The similar system was
described in (Do 2006, Rabinowitz 2003) where
American ATSC Digital Television (DTV) signal
has been tested for ranging. This DTV system used
8-ary Vestigial Sideband Modulation (8VSB) with
symbol rate 10.76223 MHz. The signal delay was
derived from the symbol synchronization.
3.1 DVB-T signal characteristics
The European standard DVB-T (ETSI EN 2004) is
based on OFDM multiplex with OFDM symbol rate
of approximately 1 4 kHz depending on the mode
(2K, 4K or 8K - number of carriers) and signal
bandwidth (channel spacing). The OFDM
synchronization methods are described in (Keller et
al. 2006). The acceptable synchronization error for
successful data demodulation is equal to the fraction
of symbol duration T
U,
i.e. tens or hundreds of
microseconds. This synchronization error
corresponds with the range error in the order of
kilometers or tens of kilometers, which is not
practically usable for navigation.
On the other hand, the minimum accessible
deviation (
σ
2
r
)
MIN
of the range measurement in
additive white Gaussian noise (AWGN) channel
(Cook 1969) is determined by (1) and depends on
the signal effective bandwidth
2
β
, signal energy E,
and single side spectral power density of the noise
N
0
( )
2
0
2
.
2
2
r
MIN
N
c
E
σ
β
=
(1)
where c is speed of light,
( )
2
2
22
4
2
f S f df
E
π
β
−∞
=
(2)
and |S(f)| is spectral energy density of the processed
signal complex envelope. The optimal signal
processing is based on correlation reception.
3.2 Correlation reception and simulation results
The correlation reception is based on the
computation of the cross correlation function of the
processed signal and its known replica. The signal
delay is determined by position of the maximum of
this function. Thanks to the good signal to noise
ratio, the DVB-T signal has not to be processed
continuously like GNSS signal to reach reasonable
measurement error. It is sufficient to work with
snapshots of the signal. The following analysis
focuses on processing of one OFDM symbol
duration snapshots in 8K mode, which is typical
mode for the DVB-T service in the Czech Republic.
The DVB-T signal (ETSI EN 2004) consists of
the transmitted data s
d
(t), scattered pilot cells s
sp
(t),
continual pilot carriers s
cp
(t), and TPS carriers s
TSP
(t)
. (3)
The OFDM symbol with duration T
S
consists of
symbol part T
U
, which is extended by the guard
interval
. All components of the OFDM symbol
are orthogonal for the symbol part duration T
U
.
The method of replica generation uses a part of
the DVB-T signal only, for example continual pilot
carriers, and the rest part of the signal is assumed as
a signal modulated by the random data. The benefit
of this procedure is the fact that such proper part of
signal is identical in every symbol and is orthogonal
140
to the rest of signal components in vicinity of the
correlation maximum. It means that those random
signals do not affect the range determination. The
weakness of this procedure can be seen in the fact
that only small part of the signal energy is used for
ranging.
The simulated range measurement errors obtained
by the application of the method described above are
in Fig. 5. The theoretical values of standards
deviation of range error (1) and simulated ones for
various SNR are compared in Table 1.
Fig. 5. PDF of range measurement error of the DVB-T signal
The ranging performance (see Table 1) of the
DVB-T signal is very good despite of the fact that
only continual carrier part of the DVB-T signal was
used for range estimation. This is thanks to wide
bandwidth of the DVB-T and good signal to noise
ratio in the service area.
Table 1. DVB-T 8K Signal Range Measurement Error
_________________________________________________
SNR [db] σ
r
[m] σ
r
[m]
theory simulation
_________________________________________________
5 0.487 0.422
10 0.274 0.224
15 0.144 0.134
20 0.086 0.079
25 0.048 0.046
30 0.028 0.031
_________________________________________________
The comparable range measurement error in
GNSS receiver can be reached by long averaging
process (Kaplan 2003). The computational
requirements to the DVB-T signal processing for
obtaining comparable error are therefore much lower
than for the case of the GNSS signal processing.
3.3 Multipath sensitivity
Multipath is the main part of errors of the navigation
systems based on ToA or DToA navigation method.
Error can be expressed by so called multipath error
envelope, which shows the worst positive and
negative measurement errors for the channel with
one straight and one reflected delayed signal (Misra
& Enge 2001). Power level of the reflected signal is
expressed by the Multipath-to-Direct Ratio (MDR).
The range error caused by the multipath is lower
for the DVB-T then for the GPS C/A because of the
wider signal bandwidth (Fig. 6). However the
multipath error in DVB-T signal ranging is
negatively influenced by the relatively high adjacent
correlation peaks.
0 100 200 300 400 500
-40
-30
-20
-10
0
10
20
30
40
Differential path length [m]
Range error [m]
0 100 200 300 400 500
-40
-30
-20
-10
0
10
20
30
40
Differential path length [m]
Range error [m]
DVB-T
GPS L1 C/A (d=1)
GPS L1 C/A, narrow correlator (d=0.1)
MDR -6dB
Fig. 6. Multipath error envelope of DVB-T and GPS
(d is GPS correlator spacing)
3.4 Software receiver for DVB-T ranging (Fig. 7)
Precision of the position determination of the ToA or
DToA systems depends on the signal arrival time
measurement error, transmitters–receiver geometry
and propagation effects, e.g. multipath. The software
DVB-T receiver was developed for further
investigation of ranging performance, propagation
phenomena and positioning tests. The receiver
captures snapshots of the DVB-T signals on various
frequencies and sends them via USB interface to the
PC workstation. Because of the low computational
complexity of the signal processing, the real time
signal processing has been programmed in Matlab
environment.
4 CONCLUSIONS
The Experimental Software GNSS Receiver is the
tool for experiments with real GNSS signals which
cannot be carried out with the commercial receivers.
141
Its main advantage is versatility. The utilization of
the modern development tools for FPGA, signal
processing and embedded computers supports rapid
implementation of the investigated algorithms to the
receiver.
Fig. 7. Experimental DVB-T receiver for navigation
Processing of the wideband Galileo E5 signals
requires more powerful algorithms and therefore
next generation of EGR will be based on Virtex 4
FPGA with higher performance. It will be able to
process all Galileo signals in space.
First analysis and experiments indicate that DVB-
T signal can be used for mobile navigation based on
ToA or DToA positioning principles. The range
measurement errors are comparable with the
measurement errors of the GPS. Moreover, the
DVB-T signal is available indoor at sufficient level.
The DVB-T ranging can be used in stand alone
navigation system or it can be integrated with other
navigation systems.
The next research will be focused on study of
errors caused by the propagation of the DVB-T
signal and transmitters’ geometry. The other
technical problem to be solved is determination of
the OFDM symbol timing or symbol synchronization
to the UTC or GPS time with precision of one
nanosecond and better.
ACKNOWLEDGEMENT
Research described in the paper was financially
supported by the research program MSM
6840770014 of the Czech Ministry of Education.
REFERENCES
Bertoncini B., Luise M., Principe F. & Terzi M., 2004. SOFT-
REC: An Innovative GPS Receiver following the Software
Radio Paradigm. Proceedings of NAVITEC 2004, 2nd ESA
Workshop on Satellite Navigation User Equipment Techno-
logies. Noordwijk: ESA, pp.25-26.
Borre K. et al. 2007a., A software-Defined GPS and Galileo
Receiver, A Single-Frequency Approach. Boston:
Birkhauser.
Borre K., 2007b. The GPS Easy SuiteMatlab code for the
GPS newcomer. GPS Solutions, no 7, pp 47-51.
Cook Ch. C. & Bernfeld M., 1969. Radar Signals, An Intro-
duction to Theory and Application. Academic Press.
Do J., Rabinowitz M. & Enge P., 2006. Performance of Hybrid
Positioning System Combining GPS and Television Signals.
Proceedings of the Position Location and Navigation
Symposium - PLANS 2006, pp. 556-564.
ETSI EN, 2004. Digital Video Broadcasting (DVB); Framing
structure, channel coding and modulation for digital
terrestrial television. ETSI EN 300 744 V1.5.1
Kaplan D., 2003. Understanding GPS Principles and Applica-
tions. Artech House Inc.: London
Keller T., Piazzo L. & Mandarini P., 2006. Ortogonal
Freguency Division Multiplex Synchronization Techniques
for Frequency-Selective Fading Channels. IEEE Journal on
Selected Areas in Communication, vol. 19, no. 6, pp. 999-
1007.
Kovář P., Vejražka F., Seidl L., & Kačmařík P., 2003.
Experimental Software Receiver of Signals of Satellite
Navigation Systems. 11th IAIN World Congress on Smart
Navigation - Systems and Services. Berlin.
Kovář, P., Vejražka, F & Špaček, J. 2004. Availability of the
EGNOS System for a Land Mobile User. The International
Symposium on GPS/GNSS. Sydney.
Kovář P., Seidl L. & Vejražka F., 2006. Availability of the
EGNOS Service for a Land Mobile User. Proceedings of
the Position Location and Navigation Symposium - PLANS
2006 [CD-ROM]. Piscataway: IEEE, pp. 524-530.
Kovář, P., Puričer, P. & Špaček, J. 2006. Design Methods for
RF Part of the GNSS Software Receivers. Coordinates, Vol.
2, No. 4, pp 34-37.
Misra P. & Enge P., 2001, Global Positioning Systems, Signals,
Measurements, and Performance. Ganga-Jamuna Press.
Rabinowitz M. & Spilker J., 2003 “A new positioning system
using television synchronization signals. IEEE Transactions
on Broadcasting, vol.51, no.1 , pp 51-61.