Integration of gnss and Loran-C
Yüklə 126,99 Kb. Pdf görüntüsü
|
| Paper-ID: VGI 200308 Integration of GNSS and Loran-C Johannes Vallant 1 , G ¨unther Abwerzger 2 , Bernhard Hofmann-Wellenhof 3 , Klaus Legat 4 1 TeleConsult Austria Bernhard Hofmann-Wellenhof GmbH, Schwarzbauerweg 43, A-8043 Graz 2 TeleConsult Austria Bernhard Hofmann-Wellenhof GmbH, Schwarzbauerweg 43, A-8043 Graz 3 Technische Universit ¨at Graz, Abteilung f ¨ur Positionierung und Navigation, Steyrergasse 30, A-8010 Graz 4 Technische Universit ¨at Graz, Abteilung f ¨ur Positionierung und Navigation, Steyrergasse 30, A-8010 Graz VGI – ¨
Osterreichische Zeitschrift f ¨ur Vermessung und Geoinformation 91 (1), S. 54–60 2003
BibTEX: @ARTICLE
{Vallant_VGI_200308, Title =
{Integration of GNSS and Loran-C}, Author =
{Vallant, Johannes and Abwerzger, G{\"u}nther and Hofmann-Wellenhof, Bernhard and Legat, Klaus}, Journal = {VGI -- {\"O}sterreichische Zeitschrift f{\"u}r Vermessung und Geoinformation}, Pages =
{54--60}, Number =
{1}, Year =
{2003}, Volume =
{91} }
Integration of GNSS and Loran-C Johannes Vallant, Günther Abwerzger, Bernhard Hofmann-Wellenhof and Klaus Legat, Graz Abstract For many navigation applications, e.g. in urban or mountainous areas, insufficient satellite visibility of the Global Navigation Satellite System (GNSS) is an issue. This problem can be reduced by integrating GNSS with other dis similar systems, where the drawbacks of the individual systems compensate each other. An attractive option is the terrestrial radio navigation system Loran-C. lntegrating GNSS and Loran-C improves the reliability and availability of the positioning information significantly. Within the GLORIA (GNSS and Loran-C in Road and Rail Applications) project, funded by the European Community (EC), the development and evaluation of a hybrid navigation receiver is demonstrated [1 ]. This innovative approach opens the door to new applications and to major improvements in existing application designs for land transport. Zusammenfassung Viele Landanwendungen der Navigation, vor allem im städtischen bebauten Bereich und im alpinen Raum, leiden unter der schlechten Satellitensichtbarkeit von Global Navigation Satellite Systems (GNSS). Diese Schwierigkeiten können durch eine Integration von GNSS mit anderen, verschiedenen Systemen kompensiert werden. Die Idee der Integration beruht darauf, dass ein Sensor die Nachteile der anderen Sensoren kompensiert. Neben der Verwen dung von autonomen Navigationstechniken wie Koppelnavigation (dead reckoning) oder lnertialnavigation sind terrestrische Radionavigationssysteme wie z.B. Loran-C eine gute Ergänzung zu GNSS. Die Integration von GNSS und Loran-C erhöht die Zuverlässigkeit und Verfügbarkeit von Positionslösungen signifikant. Dieser Bericht präsentiert die Entwicklung und Evaluierung einer neuen hybriden Empfängergeneration im Rah men des von der EU finanzierten Projektes GLORIA (GNSS and Loran-C in Road and Rail Applications). Die In novation beruht auf einer Integration von Rohdaten. Dieser neuartige Zugang verbessert bestehende Anwen dungsmöglichkeiten und öffnet Türen für neue Betätigungsfelder auch außerhalb der Landnavigation. 1 . lntroduction Envisaged are navigation solutions for land ap plication focused on road and rail transport, but not exclusively limited to these domains. In the future, also pedestrian navigation will be fa voured by reduced receiver size and weight. In land navigation, specific situations with var ious environments are given: On the one hand, there are urban areas with high buildings and ty pical infrastructure of a city. On the other hand, no artificial infrastructure is present in rural areas, but we have to cope with limitations in signal reception due to difficult topography. Due to the required direct line-of-sight between satel lites and receiver, today's satellite navigation systems most trequently used are not able to de liver position information every time in every en vironment. The U.S. Global Positioning System (GPS), is the best-known GNSS system [2]. For navigation applications, primarily C/A-code measurements 54
are used to derive the position information. Also carrier phase smoothing of code measurements is sometimes applied. For improving the accu racy and reliability of satellite-based positioning, Space Based Augmentation Systems (SBAS), like the American WAAS (Wide Area Augmenta tion System), the European EGNOS (European Geostationary Navigation Overlay Service), and the Japanese MSAS (Multi Satellite Augmenta tion System) can be used. Also terrestrial techni ques (Eurofix in Europe) support high perfor mance requirements in navigation by broadcast ing augmentation information. lntegrity informa tion and enhanced accuracy of position solutions are the most important benefits resulting from these augmentation systems. However, it is not possible to overcome all insufficiencies of GPS. Beside the required direct line-of-sight to the sa tellites, the most critical problem is the accidental or deliberate jamming of satellite signals [3]. But GPS is not the only satellite-based naviga tion system, which is open for civil use: GLO NASS (Global Navigation Satellite System) is the VGi 1/2003 Russian pendant to GPS. At present, only 7 of 24 necessary satellites are operational, but there exist plans of the Russian government to moder nise the GLONASS system within the next years and to regain Full Operational Capability (FOC). However, GLONASS has the same problems and insufficiencies like GPS due the similar sys tem properties. Galileo is Europe's future satellite navigation system and its FOC is scheduled for 2008. This new civilian system belongs to the enhanced GNSS-2 level. [4] There exist various approaches to overcome the shortcomings connected to satellite naviga tion. One promising candidate is to supplement satellite-based navigation by data of other sen sors. The concept of this sensor fusion techni que will be outlined in the subsequent sections. 2. Sensor fusion Sensor fusion means a combination of differ ent sensors to compensate drawbacks of one sensor by another. Some candidate sensors and systems are listed below. 2.1 . Autonomous Navigation Techniques Autonomous positioning techniques, i.e., tech niques without support by an external system like terrestrial transmitters or satellites, are rela tive methods. This means that they determine positioning information relative with respect to a given reference station. To get the absolute po sition information, e.g. an initialisation of the autonomous navigation system has to be carried out. Due to propagation and accumulation of various measurement errors, autonomous tech niques frequently suffer from accuracy degrada tion over time. Electronic compasses for direction determina tion and odometers for relative distance mea surements may be regarded as autonomous sensors. Differential odometers deliver distance and direction changes at once. lnertial naviga tion systems (INS) can also be used for measur ing relative distance and direction variations. However, high quality INS are very expensive. 2.2. Radio navigation Terrestrial Radio navigation uses on the one hand existing radio networks, like mobile phone or digital TV transmitter, and on the other hand VGi 1 /2003 dedicated navigation systems, like Loran-C. Po sition solutions derived from measurements in cellular networks and digital TV networks are characterised by a good availability, but cur rently cannot meet the performance require ments of land navigation. Loran-C, on the other hand, fulfils the minimal position accuracy in the test areas. Other features of the Loran-C signal are the high signal strength and a wavelength in the low frequency range. Due to these character istics, a good signal penetration in all outdoor environments is given. In addition to the naviga tion signal, an augmentation signal (Eurofix) can simultaneously be transmitted via the Loran-C carrier. Main disadvantages of Loran-C are: • Hardly predictable propagation effects of the carrier, if the signal travels over landmasses. This unknown signal propagation delay is re ferred to as ASF (Additional Secondary phase Factor) and can cause degradations of the absolute positioning accuracy down to some kilometres. • Due to organisational uncertainties, the future of the system is unknown [6,7]. • Long signal integration times of measure ments limit the use of current receivers for dy namic applications. 2.3. Mathematical methods and options of fu sion
A very common mathematical method for combining different sensors and systems is to use a Kaiman filter. The Kaiman filter is a recur sive and linear algorithm for the optimal estima tion of various navigation parameters, which is based on a dynamic model of the vehicle mo tion. For more information about Kaiman filtering in navigation applications see [6]. Also an epoch per-epoch adjustment algorithm can be applied for the fusion of various sensors. Generally, sensors can be loosely or tightly coupled. The integration of data can be done on the position data level (loose coupling) or on raw data level (tight coupling). The integration on the position data level offers the advantage that each system works independently and out puts an individually computed position. These individual positions are then integrated within the filter algorithm. When integrating the sensors on raw data level, all raw measurements are used together to deliver a common position so lution. The advantage is that the integrated sys tem provides useful navigation information even if one of the individual systems fails to compute an individual position. 55
When integrating different sensors and sys tem, various aspects have to be discussed: Most sensors deliver measurements or position data based on different coordinate reference frames and time reference systems. GPS, e.g., uses the coordinate reference frame WGS-84 (World Geodetic System 1 984). The position of the Northwest European Loran-C System (NELS) stations are also given in WGS-84 coor dinates. lt necessary, coordinates of different Cartesian reference frames can be harmonised by performing a seven-parameter coordinate transformation also called Helmert transforma tion. The time reference systems in use for the GPS/Loran-C integration are GPS system time and UTC (Universal Time Coordinated) as rea lized by the N ELS control centre at Brest (France). The time synchronisation can be achieved by considering an additional unknown parameter to be solved within the estimation process. Note that there is a variable time offset between GPS time and UTC of 1 3 seconds (May, 2003). Finally, it has to be mentioned that different systems deliver measurement or position data on different accuracy and availability levels. An other task in sensor fusion is the calibration of the integrated system which can be done in la boratory tests and setups. 3. Realization The choice of sensors for an integrated navi gation system depends on the characteristics of the various candidate systems. E.g., the terres trial system Loran-C has a dissimilar signal char acteristic compared to the space based GPS. Common vulnerabilities are almost not present. Another advantage for this combination is the good signal penetration of Loran-C in GPS hos tile environments, like dense canopy in rural areas, or between high buildings in cities. Although Loran-C has low absolute accuracy, the relative (repeatable) accuracy of the system is very high and even comparable to GPS. The idea when combining the two systems is there fore to calibrate the absolute accuracy of Loran C (i.e. to remove the influence of ASFs) during phases of good GPS availability and to use the calibrated Loran-C signal to continue the posi tion computation during phases of limited GPS availability or also during complete GPS outages. The combination is also suitable for additional sensors, although the mentioned combination 56
can already fulfil the user requirements for many land transportation applications. In the GLORIA project, GPS and Loran-C have been combined within one receiver type called DURAN (Dual Radio Navigation Receiver). This receiver integrates three major components, i.e. the LORADD prototype (innovative Loran-C re ceiver), a commercial GPS receiver, and a micro processor for carrying out all computations. 3.1 . DURAN Components The LORADD is a fully digital, multi-chain, all in-view Loran-C (and Chakya) receiver. lt in cludes two 1 6-bit analog-to-digital (AD) conver ters, which operate at a sampling frequency of 400 kHz, and a high-end digital signal processor (DSP). The range-measurement loops track the incremental phase of all received and selected Loran-C signals. Since the phase of a signal may only be measured within one cycle, the re ceiver tracks each station from an unknown starting point that is "arbitrarily" fixed when turn ing on the receiver. The initial unknown number of full carrier cycles between the receiver and each transmitter station is obtained by tightly coupling the LORADD to the GPS component of the DURAN (this procedure is denoted as , Loran-C calibration). Apart from its Loran-C navi gation functionality, the LORADD also supports Eurofix, i.e., the GNSS augmentation service re lying on Loran-C as a data link. Along with the receiver, Reelektronil type of an omni-directional magnetic field (H field) antenna for the LORADD. According to ear
lier investigations, the H-field of the Loran-C sig nals provides a better penetration into urban
areas than the electric field (E-field) (8]. The GPS part of the DURAN is realised by a
commercial GPS receiver board developed by NovAtel (i.e., the GPS component is a common
off-the-shelf (COTS) product) [9]. Although the OEM4 also provides carrier phase tracking on
both GPS carrier frequencies (L 1 and L2), only L 1 C/A-code pseudorange measurements are
processed within the DURAN. The OEM4 was chosen since it is a high-qual
ity instrument, which avoids introducing errors into the DURAN navigation solution that are typi
cal of low-cost GPS receivers. The receiver is used together with its corresponding geodetic
GPS antenna. The heart of the DURAN is the integrated navi
gation software {lntNav) designed by TeleCon sult-Austria (www.teleconsult-austria.at). In the
VGi 1 /2003 present version of the prototype, the lntNav soft ware runs on a high-pertormance Pentium-type processor. For a future miniaturisation, this pro cessor will be replaced by a more suitable DSP. 3.2. lntNav - The GPS and Loran-C integration software As it becomes clear from the previous sec tions, the fusion of GPS and Loran-C should al low to compensate the main disadvantages of the individual systems. E.g., the low absolute po sitioning accuracy of Loran-C can be compen sated by GPS, whereas Loran-C can bridge GPS outages caused by limited satellite visibility. Thus, pertorming a GPS-aided calibration of the Loran-C measurements seems to be the most convenient way to realise the integration. The calibration is carried out during phases of good GPS satellite visibility by a continuous computa tion of the theoretical Time of Arrival (TOA) of Loran-C signals and a comparison of that value with the corresponding measured TOA. The dif ference between computed and measured TOA yields a calibration value for the respective Loran-C station. During limited GPS availability, these calibration values are used to compensate the low absolute accuracy of Loran-C and, thus, to continue the position computation with near GPS-accuracy. The computation of Loran-C ca libration values uses a simplified propagation model of Loran-C signals and bases on the cur rent estimated receiver position. To ensure the reliability of the calibration, it is further essential to apply an automated decision-making proce dure for deciding whether the GPS pertormance is qualified for calibrating Loran-C. To realise these theoretical considerations, TeleConsult Austria has developed the GPS - Loran-C integration software lntNav. Although Programming language C/C++
GPS and Loran-C snuggle together in this inte gration software, it has to be distinguished be tween individual pre-computational parts and a common integration part: GPS pre-computations Positions of GPS satellites: As GPS satellites move around the earth, the position of satellites has to be computed for each measurement epoch. The corresponding information can be derived from the broadcast ephemerides, which are transmitted by the GPS system itself. Compensation of GPS signal propagation ef fects: The propagation of GPS signals is mainly affected by the ionosphere and the troposphere. Therefore, the integration software is able to ap ply models for compensating these propagation effects: The influence of the ionosphere is re duced by the broadcast 8-coefficients iono spheric model, whereas the influence of the tro posphere is reduced by the modified Hopfield model. Applying these models to the measure ments is optional. Computation of measurement weights: To ob tain individual weights for GPS measurements, mainly the elevation of the corresponding satel lite is considered. Loran-C pre-computations Compensation of Loran-C propagation effects: The propagation of Loran-C signals over land masses is mainly affected by the conductivity of the underground. The propagation delay, which is therefore introduced, is called Additional Sec ondary Factor (ASF). The pre-computation part of lntNav corrects Loran-C TOAs by applying ca libration values, which have been derived during phases of good GPS availability. Smoothing of Loran-C TOAs: The software has a built-in low-pass filter, which can optionally be applied to smooth Loran-C measurements. Currently supported OS Win 9x, Win 2k, Linux Application environment Console
Currently supported navigation systems GPS, GLONASS, Loran-C Measurement value GNSS pseudoranges, Loran-C TOAs Operation modus Selectable: Real-time - Post-processing Calibration of Loran-C GPS-aided, real-time Data pre-filtering Elliptic low-pass filter Quality check of solution RAIM, DOP computation Position computation algorithm Selectable: Epoch-wise Adjustment - Kaiman Filter Overall structure Modular, expandable Currently supported devices Ashtech GG24, NovAtel OEM4-31 51 R, Locus SatMate 1 000 & 1 020.
Table 1: Features of lntNav Software VGi 1/2003 57
Computation of measurements weights: The
quality of Loran-C measurements is in a first ap proximation indirectly proportional to the dis tance between Loran-C transmitter and receiver. This is also the main input for individual Loran-C weight computations. GPS
and Loran-C integration part The core of the l ntNav software is an adjust ment algorithm, which performs the integrated position computation. Besides the current posi tion of the receiver, also some quality informa tion, i.e., an estimation for the position accuracy, is obtained. The common integration part further consists of various consistency checks of the measurements, as weil as a Receiver Autono mous lntegrity Monitoring (RAIM) algorithm. Concluding, Table 1 summarises the main fea tures of the lntNav software. The software is currently subject to further de velopment and modification. Because of its modular structure, it can easily be adapted to various platforms and operating systems. Also the list of supported navigation devices can ea sily be expanded. 4. Receiver test and evaluation The major aim of the tests was to compare the results of the DURAN with stand-alone GPS (re- 411475
'.§: € 411450 0 z 411425 +-25 m--?
r , f ,r .............. „�:·;·„·'· ·�tt;,\ . . „ . . . . . 1 1 21775 Positioning mode LORAN-C Calibration 21800 East [m) 21825 ference case) and to investigate the potential benefits of the new receiver for road applica tions. Rail applications were also treated within the GLOR !A
project but will not be presented in this report. For all tests, the two chains of the Northwest European Loran-C System (NELS) with the master stations at Lessay (France) and Sylt (Germany) have been used. These Loran-C chains provide the best reception quality within the testing areas in the Netherlands, France, and Belgium for the major environment types rural, urban, highway. In these areas 1 4 static and kinematic tests were performed. The test vehicles were not specifically modi fied for the tests. All measurement data were recorded in the internal memory of the recei vers. The investigation of the DURAN perfor mance was mainly done in post processing. The required offline version of the lntNav soft ware used in this case operates completely analogous to the real-time version installed in the DURAN. The main advantage of post pro cessing analysis is, that certain parameters of the algorithms can be modified to achieve bet ter overall performance of the receiver. This op timization phase was carried out in parallel to the investigation of the receiver performance. Afterwards, the resulting final settings of the software were stored in the receiver for online position computing. 411475
'.§: € 411450 � 411425
1 +-25
m--? 21775
GPS Satellites $ 6 $ 8 s 1 1 s 1 4 1111111 > 1 4 21800 East [m) 21825 Figure 1: Loran-C Position (Jett) and GPS Position so/ution (right) 58 VGi 1 /2003 Method applied Error ellipse (95%) axes [m) orient. [0) GPS stand-alone 3.5 / 2.2 59 (reference case) DURAN (fully integrated) 3.5 / 2.2 60 Calibrated LORAN-C 1 6 / 1 3 87
Table 2: Numerical resu/ts of Static test Eiffel tower T eleConsult employed some additional mea surement equipment for determining alternative GPS reference trajectories. The additional equip ment comprised two geodetic Ashtech GPS re ceivers (GG24, Z1 2) and the corresponding an tennas. In case of GPS outages, no reference data are available. Still, the quality of the DURAN results may be derived from the tests: the result ing trajectories should continue smoothly during GPS outages. The data can be interpreted vi sually and the consistency of the DURAN results with the nominal trajectories can be verified using map information. 623000 '.[
.r:::. 622000 t 0 z 621 000
• • • „ • • • • • • „ • ! „ • • • • • • • • „ • • • • • • • • . . . . . . . � . . . · : · . . . . . . . . . . . . . . . . . . · · · · · · · · · · · · : · · · · · · · · · · · · · · · · . . Bias [m] Avail . w.r.t. north east
time [%) Max. out. [s) - -
2 0.0
0.0 1 00
0 2.3
-1 .3 93.2
1 5 l t was clear from the beginning, that even under optimised conditions the variance of the Loran-C position solution would be larger than for GNSS. The test near the Eiffel tower in Paris (F) was per formed to investigate the static positioning quality of the Loran-C component of the DU RAN and un derlines our assumption [Figure 1 , Table 2]. Kinematic tests are represented by a test on a motorway south east of Brussels (B). Figure 2 shows the details of the test track during a GPS outage. The DURAN is able to bridge the outage even though there are some short Loran-C outages as weil [Figure 2, Table 3]. Drive . Directipn Receiver B GPS
DU RAN 620000
• • • • • • • • • • • „ • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • i • • • • • • • • • • • . • • . • • • • i . . . . . • . . . .
29000 30000
31000 East
[m] 32000
Figure 2: Kinematic test motorway VGi 1/2003 59
Receivers Availability Maximum outage w.r.t. time [%] duration (s] GPS stand-alone 91 .0 76
(reference case) DURAN (fully in- 99.7 37
tegrated solution) Table 3: Numerical results of kinematic test motorway 5. Summary and outlook The approach to integrate GNSS and Loran-C is feasible and is verified by good test results. The DURAN increases temporal and local avail ability of the positioning solution in case of par tial GPS outages. Also the overall continuity of the integrated navigation system can be in creased. Further, most of the requirements of navigation in road applications, like raute gui dance of private vehicles, tracing of vehicles for floating car data, and monitoring of dangerous goods on road can be met. However, there are still some subjects - espe cially concerning Loran-C - to be investigated in more detail than they are known today. The ma jor drawback of the current DURAN prototype is its limited resistance against electromagnetic disturbances. Furthermore, the improved reac quisition time of the receiver cannot be exploited at the moment because of the required recalibra tion of the Loran-C data after a Loran-C outage. lt such an outage occurs simultaneously with a GPS outage, the new calibration of Loran-C can only be performed if GPS is "back at service". Further improvements are possible in the issue on when to use GPS to calibrate the Loran-C ranges. Especially for automotive applications, the miniaturisation of the receiver is crucial. DURAN has a !arge potential in this domain that needs to be exploited in the close future. The currently limited accuracy of Loran-C in Europe is also caused by the sparse network of transmitter stations available. Further, it was em- 60
phasised that a significantly increased perfor mance could be achieved if the network was ex tended only by a few additional transmitter sta tions.
References [1]
Abwerzger G., Beyer J„ Legat K., Maurer M„ Meinhard D., Pfister J., van Willingen 0. (2001):
GLORIA - lntegrating GNSS and Loran-C for high requirement applications. Ga Jileo's World, 2(3): 1 0-1 6. [2] Department of Defense (2001 ): Global Positioning System standard positioning service performance standard. U.S. Assistant for GPS, Positioning, and Navigation, Defense Pentagon, Washington, DC 20301 -6000. [3] Department of Defense and Department of Transportation {2002a): 2001 Federal Radionavigation Plan. U.S. National Technical Information Service, Springfield, Virginia, DOT VNTSC-RSPA-01 -3/DOD-4650.5. [4] Department of Defense and Department of Transportation (2002b): 2001 Federal Radionavigation Systems. U.S. Na tional Technical Information Service, Springfield, Virginia, DOT-VNTSC-RSPA-01 -3.1 /DOD-4650.5. [5] European Commission (2002): Galileo - The programme for global navigation services. ESA Publications Division, Noordwijk. [6] Hofmann-Wellenhof 8, Legat K, Wieser M (2003): Naviga tion - Principles of positioning and guidance. Springer, Wien New York. (In print.) [7] NovAtel lnc. (2002): „NovAtel OEM4 Specification Sheet"; available under www.novatel.com [B]
Pfister J., Kraft M.,
Legat K.,
Abwerzger G. (2002): Assess
ment Report; GLORIA D1 0 WP5. [9]
Reelektronika b. v. (2002): „Receiver system prototype: DURAN"; GLORIA DB-B WP4. [10] Volpe National Transportation Systems Center (2001): Vul nerability assessment of the transportation intrastructure relying an the Global Positioning System. Final Report, August 29. Contact
Dipl.-Ing. Johannes Vallant, Dipl.-Ing. Günther Abwerz ger: TeleCorisult Austria Bernhard Hofmann-Wellenhof GmbH, Schwarzbauerweg 3, A-8043 Graz. email: jvallant@teleconsult-austria.at, abwerzge@teleconsult-austria.at Univ. Prof. Dr. Bernhard Hofmann-Wellenhof, Univ.Ass. Dr. Klaus Legat: Technische Universität Graz, Abteilung für Positionierung und Navigation, Steyrergasse 30, A-801 O Graz. email: hofmann-wellenhof@tugraz.at, legat@tugraz.at VGi 1 /2003 Yüklə 126,99 Kb. Dostları ilə paylaş:
|