433
1 INTRODUCTION
It has been recognized that the increasing global
transportation of goods is not only pushing the
rapidly growing vessel dimensions, but also the
amountofvesselssharingthesameroutesatthesame
time. As a consequence the requirements for the
provision of reliable and accurate navigational
information increase, in order to minimiz
e the
probability of situations that could compromise the
safety of the ship, crew and the environment. The
capability to provide onboard Position, Navigation
andTiming(PNT)datacompliantwiththeaccuracy,
integrity,continuity andavailability requirements in
accordance to the different phases of vessel
navigation is one core element of the International
MaritimeOrganizations(IMO) e‐Navigationstrategy
[1].TheGl
obalNavigationSatelliteSystems (GNSS),
in particular the Global Positioning System (GPS) is
the key component in maritime navigation for the
provision of absolute positioning, navigation and
timing information. In order to fulfil these
requirements, the Germa
n Aerospace Center (DLR)
hasdevelopedaPNT(dataprocessing)Unitconcept
[2][3](seeFigure1)).
In this concept different processing channels are
used in parallel in order to provide PNT data
compliantwiththeperformancerequirementsforthe
differentphasesofvesselnavigation.
Inthisresearchstudythefocusliesonnavigation
insidetheport,wherethehighestrequirementswith
respecttoaccuracyandint
egrityneedstobefulfilled.
In[4]maritimeuserrequirementswithanaccuracy<
1m for port operations and < 10cm for automatic
docking are stated. These requirements cannot be
Long Term Validation of High Precision RTK
Positioning Onboard a Ferry Vessel Using the MGBAS in
the Research Port of Rostock
R.Ziebold&S.Gewies
GermanAerospaceCentre(DLR),Neustrelitz,Germany
ABSTRACT: In order to enable port operations, which require an accuracy of about 10cm, the German
AerospaceCenter(DLR)operatestheMaritimeGroundBasedAugmentationService(MGBAS)intheResearch
Port of Rostock. The MGBAS reference station provides GPS dual frequency code + phase correction data,
whicharecontinuouslytra
nsmittedviaanultra‐highfrequency(UHF)modem.Uptonowthevalidationofthe
MGBASwasratherlimited.Eitherasecondshorebasedstationwasusedasanartificialuser,ormeasurement
campaignsonavesselwithdurationofafewhourshavebeenconducted.Inordertoovercomethi
s,wehave
installed three separate dual frequency antennas and receivers and a UHF modem on the Stena Line ferry
vesselMecklenburg‐VorpommernwhichisplyingbetweenRostockandTrelleborg.Thispaperconcentrateson
theanalysisofthehighlyaccuratephasebasedpositioningwithaRealTimeKinematic(RTK)algorit
hm,using
correctiondatareceivedbytheUHFmodemonboardthevessel.Weanalyzedtheavailabilityandaccuracyof
RTKfixsolutionsforseveraldays,whenevertheferryvesselwasinsidetheserviceareaoftheMGBAS.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 11
Number 3
September 2017
DOI:10.12716/1001.11.03.06
434
fulfilledusingstandardGNSScodebasedpositioning
techniques. Therefore GNSS phase based techniques
needstobeapplied.Inthisanalysiswefocusonthe
applicationofRealTimeKinematics(RTK)[5].RTKis
a differential technique based on carrier phase
ambiguity fixing using the data provided by a
reference station.
While RTK based positioning is a
standard approach for applications like surveying,
hydrography [6] and precise farming, its usage for
safety of life critical applications like vessel
navigationisstillunderdevelopment.
Inthispaperwewillpresentlongtermvalidation
results of RTK positioning onboard a ferry vessel
using
theMGBASintheresearchportofRostock.The
restofthepaperisorganizedasfollows.Insection2
therealizationofthe MGBASin theportof Rostock
will be described. In section 3 the setup of the
measurement campaign will be given, before in
section4thevalidation
resultswillbepresented.
Figure1.ConceptofaPNTdataprocessingUnitaspartofa
futureIntegratedNavigationSystem(INS)
2 MGBASREALIZATION
The Maritime Ground Based Augmentation System
(MGBAS) was developed in 2009 [7] following the
principleofIALABeaconserviceformulatedinIALA
recommendation [8]. In difference to the well
acceptedanddistributedIALABeaconservice,which
providesaccuraciesofbetterthan10mtogetherwith
integrityinformationwithin
theserviceareaofabout
300 km around service stations the MGBAS concept
addresses the nautical tasks with a demand on
accuracy of better than 1m or even 10cm together
withintegritywhichisnecessaryforportoperations
and automatic docking. In contrast to IALA Beacon
service, which
provides only code corrections for
GNSS measurements, the MGBAS transmits more
precisephasemeasurementswhichcanbeutilizedfor
RTK positioning. This approach enables higher
accuraciesonthecostofareducedsizeofservicearea
offew10kmaroundthereferencestation.
The concept of MGBAS supports all available
globalandregionalnavigationsatellitesystemssuch
asGPS,Galileo,GLONASS,Beidou,andQZSS[7].It
consists of a reference station, with exactly known
coordinates, which receives GNSS signals and
evaluates them with respect to predefined
performancekeyidentifiers(PKI).Withintheservice
area of the reference station one or
more integrity
monitoring station(s) are installed. They serve as
artificialusers andcalculate their position, basedon
the correction information broadcasted by reference
station. The positioning results are checked against
thesurveyedfixedposition.Thisallowsanevaluation
oftheservicequality.AdditionallyfurtherPKIwillbe
calculated. After the evaluation
of PKI of both
stations, Radio Technical Commission for Maritime
Services SC‐104 Version 3.x (RTCM3) messages are
generated for the maritime users which comprise a
consistentsetofGNSSphasecorrectionsdatatogether
withintegrityinformation.
Currentlythereexistsonly oneMGBASreference
implementation within the German research port
Rostock[9](seeFigure6).Thereferencestationinthe
centerofresearchportisequippedwithaLeicaGNSS
choke ring antenna, which suppresses the impact of
GNSS signal reflections on the water surface and
constructionsinthevicinity,ageodeticGNSSreceiver
connected to a rubidium atomic clock (reduces
receiverclockjumps),astationcomputerconnectedto
theinternetandaradiomodemwithanantennafor
data broadcasting. TheMGBAS integrity monitoring
station is set about 4 km away from the reference
station near the port entrance (see Figure 6). It is
equipped with a geodetic receiver a
GNSS antenna
fromnavXperienceandadataprocessingunitwhich
isconnectedtotheinternet.
An ultra‐high frequency (UHF) radio
communication link is used for data transmission.
The reference station, equipped with a SATELLINE
3ASd radio modem, broadcasts MGBAS generated
corrections and integrity information using an
extended RTCM3 protocol. The
broadcast power at
the frequency of 447.95 MHz is 1W. A data rate of
38400bpsisachieved.
All necessary data processing can be performed
directly at the reference and integrity monitoring
station.Additionallytheentiremeasurementdatacan
be received via internet at the central processing
facilityin
Neustrelitzand the entire data processing
couldbedoneonhighperformancecomputersthere.
Theevaluation resultsare sentback tothe reference
station,whereitwillbebroadcastedtothemaritime
usersoftheMGBAS.
Figure2. MGBAS Monitor for visualization of MGBAS
relevantintegritystatements
435
This approach provides the opportunity for an
extensivemonitoringofMGBAS,asshowninFigure
2. On the left hand side the multifunctional display
shows, which of the services (different processing
chains) meet (green) or fail to meet (red) the
requirements of the selected performance class (e.g.
port or
automatic docking). The operator can make
conclusions about the performance stability of the
servicesbasedonthetimebehaviorofpositionerror
per process (Fig 2, bottom right). In addition, the
upper right part of the screen shows, which GNSS
signalsareidentifiedasuseableforserviceprovision.
Figure3. Vessel Mecklenburg‐Vorpommern with positions
oftheGNSSantennas(redcircles)andlocationoftheboard
withreceivers(greencircle)
3 SETUPMEASUREMENTCAMPAIGN
The original sensor measurements were recorded
using the ferry vessel Mecklenburg‐Vorpommern
from Stena Lines, which is plying continuously
betweenRostockandTrelleborg.Themeasurements,
analyzed in this paper, were taken in 2015. In that
timeframe the vessel was traveling a single route
between Rostock and Trelleborg
3 times a day (see
Figure5).OnSundaysthevesselstayedforalonger
periodoftime(~4h)intheportofRostockandwas
travelling between Rostock and Trelleborg just 2
times.Thatresultsinthefactthatthevesselisatleast
once perday
for ~3h within theservice area in the
portofRostock.
The vessel was equipped with three dual
frequencyGNSSantennasandreceivers(JavadDelta
and Sigma). The antennas were placed on the
compass deck (red circles in Figure 2). The chosen
geometry of the antenna placement enabled an
accurate
determination of the vessel 3D attitude by
applyingGNSSCompassalgorithms.AUHFmodem
was used for the reception of RTK corrections data
fromtheMGBASstation.
All relevantsensor measurements were provided
either directly via Ethernet or via serial to Ethernet
adapter to a Box PC (see Figure 4)
where the
observationsareprocessedinreal‐timeandstoredin
aSQLite3databasealongwiththecorrespondingtime
stamps. The described setup enables a record and
replay functionality for further processing of the
originalsensordata.The system consistsofahighly
modularhardwareplatform(hererealizedwithaBox
PC) and a Real‐Time software framework
implementedinANSI‐C++.
Figure4:Left:InstallationofstarboardGNSSantennawithUHFantenna(right)forthereceptionofMGBAScorrectiondata
Right:BoardwithGNSSreceivers,UHFmodem,IALABeaconreceiver,serialtoEthernetadapterandBoxPCfordata
processingandstorage
436
Figure5.Overviewof24hvesseltrajectorytopleft:latitude
vs.timegraphforthatday
Figure6.VesseltrajectorywithintheportofRostock(green:
starboard antenna, red: portside antenna), white dashed
circlesmarkthedistancetotheMGBASReferenceStation
4 RESULTS
4.1 Communicationlink
TheMGBASreferencestation provides2HzGPS L1
and L2 code + phase correction in RTCM3 data
format.Onboardthevessel,thesedataarereceivedby
aSATELLINE3ASdradiomodem.TheUHFantenna
forthatmodemisplacedbesidesthestarboardGNSS
antenna on
the compass deck (see Figure 3). A
prerequisite for RTK based positioning is the
receptionoftimelycorrectiondatafromthereference
station.Thereforewefirstcheckedtheperformanceof
thecommunicationchannel.
For the purpose of this evaluation, we were not
interested in the details of the communication link,
but just in the final performance for this specific
application. For this reason we have only evaluated
the number of RTCM3 messages which have been
successfully transmitted at the MGBAS station and
decodedonboardthevessel.InFigure7(left)thedata
linkavailabilityasafunctionofthedistancebetween
vessel and reference station is plotted for the
timeframe of 21 days. Here for each day and each
distanceintervalthepercentageofreceivedmessages
compared to the send messages is calculated. In
Figure7(left)themeanvalue(averagedover21days)
andtheminimumandmaximum valuesare
plotted.
Starting with nearly 100% availability for the
moorageposition,theavailabilitydrops significantly
atadistanceof~2kmto~92%whileincreasingagain
at4kmdistance.Finallythemeanavailabilitydrops
below90%at~7kmandbelow50%at15km.
InFigure7(centerandright)thetrajectoryofthe
vessel is plotted for one arrival and departure in
Rostock for day of year (DOY) 120. The green dots
markpoints,whereadirectcommunicationwiththe
UHF link could be established successfully and the
RTCM3 data could be decoded
and used for RTK
positioning. The timespan for the generation of
correctiondatainthereferencestation,theencoding,
transmission, reception anddecoding of that data is
typically<0.5s. Together withthe applieddata rate
of 2Hz, with a stable UHF communication channel,
theageofappliedcorrections
fortheRTKpositioning
algorithmistypically=0.5s.Withincreasingageof
appliedcorrectionsthechancetogetahighaccurate
solution with fixing the integer ambiguities is
decreasing.InFigure7(center)significantdifferences
between the incoming and outgoing vessel can be
found.Comparedtothe
outgoingvesseltheincoming
vessel shows for a longer distance a good
communication channel. A possible route cause for
thatbehaviorcouldbethecombinationofthevessel
track/orientationandtheconcreteinstallationofthe
UHFantennaonthevessel.
In order to analyze the drop of the data
link
availabilityat ~2kmdistance inFigure 7(right) the
age of applied correction for DOY 120 only for the
port area is shown. Here several short data link
outages (< 5s) can be identified, which are possibly
causedbyobstructionsinthelineofsightby
theport
infrastructurelikecranes.
Focusing onlyon the service area of the MGBAS
within 5km distance from the reference station, the
implemented communication channel is not perfect,
butfortheprovisionofcorrectiondataforRTKbased
positioning reasonable good. Looking at the age of
appliedcorrectionsasthe
relevantparameterforRTK
positioning (seeFigure 8) one sees, that forall days
the availability of timely correction data is higher
than98%andfornearly100%ofthetimetheageof
correctiondataisbelow5s.OnlyforDOY 140 fora
shortperiod
oftimetheageofappliedcorrectionsis
between5sand10s.
437
data link availability [%]
Figure7.left‐MGBASdatalinkavailabilitywithvaryingdistancetothereferencestation;
Ageofappliedcorrectionswithindatareceptionrange(center)andwithintheserviceareaofMGBASintheportofRostock
(right)
Figure8. Age of applied correction for RTK positioning
withinserviceareaoftheMGBASintheportofRostock
4.2 RTKPositioningperformanceintheportofRostock
InordertovalidatetheRTKpositioningperformance,
while using the MGBAS in the port of Rostock in a
reproducibleandreceivermanufacturerindependent
way, we have decided to use the open source code
RTKLib (http://www.rtklib.com) version 2.4.2. We
haveembeddedthe
CcodeoftheRTKlibinourC++
real time (RT) Framework, so that the real time
positioning performance can be evaluated directly.
Due to the modular structure of the RT‐Framework
[10] several RTK positioning solvers could run in
parallel.Thisenables independentandsimultaneous
RTK positioning for the
three different antennas
onboard the Stena Line ferry vessel Mecklenburg‐
Vorpommernatones.
The high accuracy of RTK based positioning can
onlybeachieved,ifthedoubledifferenceambiguities
(the unknown number of wave cycles between
receiverandsatellite)canbefixedsuccessfully.Asthe
measure of trust in the ambiguity
fixing, the
ambiguity validation threshold is used [5]. For this
evaluationathresholdof2.5isapplied.
Inordertoestimatethe positioningaccuracyand
tochecktheoccurrenceofwrongambiguityfixes,the
three antenna setup onboard the vessel can be
exploited. Therefore for all three antennas the RTK
based positions using the MGBAS correction data
were calculated independently. In Figure 9 the
histogram of the resulting distances between two
antennasisplottedforthecase,thatforbothantennas
the RTK solver stated a successful fix of the
ambiguities.Fortheappliedconfigurationnodistance
errorlarger than5
cmwasfound, whichmeans that
theambiguitieshavebeenfixedcorrectly.
The resulting distances between the antennas
show nearly Gaussian distributions with a standard
deviation below 1cm. The significant smaller
standard deviation for the distance between the
portside and starboard antenna compared to those
including the mid ship
antenna could be caused by
theconcreteinstallationoftheantennasonboardthe
ferryvesselMecklenburg‐Vorpommern.Themidship
antenna is placed closer to the mean mast and
therefore suffers stronger from resulting multipath
andobscurationeffects(seeFigure3).
Anaccuratedeterminationofthepositionaccuracy
ofthe
RTKbasedpositioningwouldrequireaGNSS
independenthighlyaccuratepositioningsourceandis
not within the scope of that paper. The calculated
errorinthedistancebetweentheantennascanonlybe
alowerboundoftheabsolutepositionaccuracy.An
upper bound of the position accuracy can be
determined
bycomparingtheRTKbasedpositioning
results with a post processed reference trajectory
based on the Precise Point Positioning (PPP)
approach, which delivers dmup to cm accuracy for
continuous observations [11]. The PPP reference
trajectories, which we have calculated for several
days,areconsistentwiththeRTKbasedpositioning.
438
Figure9.HistogramsofcalculateddistancesbetweentheantennasbasedonRTKpositionresults
Figure10.RTKFixrateintheportofRostock,left:usingalldata;right:usingonlyrecentcorrectiondata
Figure11.RTKfixrateformultipleantennasintheportofRostock
439
4.3 RTKFixrateofasingleantennaintheportof
Rostock
FortheusageofRTKbasedpositioningforsafetyof
life critical applications, the availability of position
solutionswithfixedintegerambiguitiesisoneofthe
mostcriticalpoints.InFigure10theRTKfixrate
for
thestarboardantennafor21days in2015 is plotted.
Here the RTK fix rate is the percentage of epochs
(times),wheretheintegerambiguitieshavebeenfixed
successfully,divided by thetotal number of epochs,
where the vessel was inside the service area of the
MGBASinthe port
of Rostock.The servicearea has
beendefinedbyacirclewith5kmradiusaroundthe
MGBASreferencestation.
For 17 out of the 21 days a fix rate larger than
99.9% can be found (see Figure 10 left side) and a
minimumfixrateof~96
%forDOY128.Thisfixrates
areastonishinghigh,takingmultipathandnon‐lineof
sighteffectsfromcranes,buildingsandothervessels
intoaccount.
In a next step the influence of the imperfect
communicationchannel(see4.1)ontheRTKfixrate
has been evaluated. For this analysis those
epochs,
wheretheMGBAScorrectiondataareolderthan1s,
havebeenfilteredout.Theresultingfixrateisshown
inontherightsideofFigure10.Herenosignificant
differencestotheleftsidecanbeobserved.Thisleads
totheconclusionthatfortheevaluatedtimeframe,
the
interruptionsinthecommunication channelhaveno
measurable influence on the RTK fix rate and
thereforeonthepositioningperformance.
4.4 RTKFixrateofusingmultipleantennasintheportof
Rostock
For our measurement campaign we have chosen a
setup with three separated GNSS antennas and
receivers onboard
a vessel (see Figure 3). This setup
enablesa)anextremelyaccuratedeterminationofthe
attitude(roll,pitch,heading)ofthevesselbyusinga
GNSS compassing algorithm and b) a redundant
determination of ships position (and velocity) by
usingeachantennaseparately.In[12]wehaveshown
that, using
the GNSS Compass setup on the vessel
Mecklenburg‐Vorpommern, we can determine the
attitude with an accuracy of better than 0.01°.
FurthermorethecombinationofthatGNSSCompass
and a low cost tactical grade Inertial Measurement
Unit (IMU) ensuresnot only this high accuracy, but
also 100% availability of the
determined attitude
information under normal conditions. Note that
duringalongerGNSSoutagetheIMUbasedattitude
informationwillslowlydriftaway(ca.1deg/hfora
tacticalgradeIMU).
Within the scope of this paper we have now
evaluatedtheadvantage ofusingthe threeantennas
separately for the redundant
RTK based
determinationoftheposition.Thisredundancycould
ingeneralbeusedforintegritymonitoringpurposes,
bychecking theconsistency betweenthe positioning
results. Here we focus just on the increase of
availability of highly accurate positioning results by
using thethree antennas separately. For an accurate
positioningit
issufficient,thatonlyforoneoutofthe
three antennas the integer ambiguities have been
fixedsuccessfully.InFigure 11theRTK fix ratefor
the threeantennas forthe 21days is plotted. For12
outof21daystheRTKfixrateforallthreeantennasis
higher
than99.9%.Itisremarkable,thattheRTKfix
rate varies significantly for the different antennas.
There is no antenna, which performs always better
thantheothers.OnlywhenevaluatingthemeanRTK
fixratefor thewhole21 days(seeTable1), themid
ship antenna clearly shows a
slightly worse
performancecomparedtotheportsideandstarboard
antennas.Thisfindingisconsistentwiththeresultsof
section4.2andcouldbecausedbytheinstallationof
themidshipantennaclosetothemainmast.
TheblackbarsinFigure11indicatethefixratefor
themultipleantenna
approach,whereatleastoneout
of the three antennas needs to have a fixed integer
ambiguity. For the evaluated timeframe of these 21
days in 2015, the resulting fix rate is always larger
than 99.96 % and a mean fix rate of 99.998% is
achieved. So that the multiple
antenna approach
substantially outperforms the single antenna
approach.
Table1.RTKfixrateforthe21dayperiod
_______________________________________________
starboard portside midshipmultiple
_______________________________________________
mean 99,71% 99.73% 99.23% 99.998%
min. 96.3%98.3%90.5%99.97%
max. 100%100% 100% 100%
_______________________________________________
The big improvement of the multiple antenna
approachcouldbeexplainedbythefact,thatatleast
for the evaluated period of time, local effects like
multipathandobscurationsarethedominatingerror
sourcesforRTKbasedpositioningofavesselinaport
environment.
5 SUMMARY
TheMGBAS
intheResearchPortofRostockprovides
continuously GNSS correction data enabling highly
accurate phase based positioning using RTK
positioning algorithms.In this paper we presentthe
results of along termmeasurement campaignusing
the MGBAS onboard the Stena Line ferry vessel
Mecklenburg‐Vorpommern, which is continuously
plyingbetweenRostock
andTrelleborg.
Inafirststep,theperformanceoftheUHF(447.95
MHz)radiocommunicationchannel,usingamodem
with 1W output power was evaluated. The
established communication link enables a 99%
availability of timely correction data onboard the
ferry vessel in the port area. Although short
interruptionsof
thecommunicationlinkaredetected,
no negative influence of those on the RTK based
positioninghasbeendetermined.
The crucial point of RTK based positioning for
safety oflife criticalapplicationis the availabilityof
thehighlyaccurate(~cmaccuracy)fixedsolutions.By
usingasingleantennaonlyanavailabilityof
99%for
the analyzed 21 days (2015, DOY 120‐140) has been
observed within the service area of the MGBAS. A
dramaticincreaseofthatavailabilityto99.997%could
440
be realized, by using a multiple antenna setup with
threeseparateantennasandreceivers.
These encouraging results open the door for the
developmentof real applicationsof the phase based
MGBASforhighlyaccurateportoperation.
ACKNOWLEDGEMENTS
The authors would like to thank Stena Lines and
especiallythecrewof
theMecklenburg‐Vorpommern
with theircaptains Mr. Watsack and Mr. Franke for
theirsupportduringthemeasurementactivities.The
authors are also grateful for the assistance of their
colleaguesCarstenBecker,AnjaHesselbarthandUwe
Netzband.
REFERENCES
[1]IMO, “NAV 54/25 Annex 12 Draft strategy for the
development and implementation of e‐Navigation,”
2008.
[2]R.Ziebold,Z.Dai,T.Noack,andE.Engler,“Conceptfor
an Integrated PNT‐Unit for Maritime Applications,”
Proc. 5th ESA Workshop Satell. Navig. Technol. Navitec
20108‐10Dec2010NoordwijkNeth.
,2010.
[3]GermanyIMO,“NAV58/6/1IMOinputpaper:Proposed
architecture for the provision of resilient PNT data
SubmittedbyGermany.”30‐Mar‐2012.
[4]IMO, “Resolution A.915(22): Revised Maritime Policy
and Requirements for A Future Global Navigation
SatelliteSystem(GNSS),”Nov.2001.
[5]P. J. G. Teunissen and A.
Kleusberg, GPS for Geodesy.
Springer,1998.
[6]J. B. Rogowski, C. Specht, A. Weintrit, and W.
Leszczynski,“EvaluationofPositioningFunctionalityin
ASG EUPOS for Hydrography and Off‐Shore
Navigation,”TransNavInt.J.Mar.Navig.Saf.SeaTransp.,
vol.9,no.2,pp.221–227,2015.
[7]T. Noack, E.
Engler, A. Klisch, S. Gewies, and D.
Minkwitz, “Integrity concepts for future maritime
Ground Based Augmentation Systems,” in 2nd GNSS
VulnerabilitiesandSolutions2009.
[8]IALA, “Recommendation R‐121: The Performance and
MonitoringOfDGNSSServicesintheFrequency Band
283.5–325KHz,”December2004.
[9]http://www.forschungshafen.de/en/.
[10]S.
Gewies, C. Becker, and T. Noack, “Deterministic
Framework for parallel real‐time Processing in GNSS
Applications,”inNAVITEC2012,2013.
[11]A. Heßelbarth and L. Wanninger, “Performance of
GNSS‐PPP in Postprocessing mode,” presented at the
Hydro,2010.
[12]R. Ziebold, M. Romanovas, andL.Lanca, “Evaluation
ofaLow
CostTacticalGradeMEMSIMUforMaritime
Navigation,”inActivitiesinNavigation,A.Weintrit,Ed.
CRCPress,2015,pp.237–246.
.
