127
1 INTRODUCTION
Position fixing systems are considered as a strategic
key element of the International Maritime
Organizations (IMO) e‐Navigation strategy [1]. The
improvement and the indication of reliability have
beenidentifiedashighlevel user needs withrespect
to electronicposition fixing [2].Analyzing reliability
ofpositionandotherPNTparametersofavessel,not
only the ship‐side components but the whole
int
egrated PNT system needs to be considered [3].
The generic architecture of the maritime integrated
PNT System is shown in Fig. 1. It is the sum of
satellite‐based,ashoreandaboardcomponentsandits
relatedlinks.
Only the int
egrated use of these components
enablestheprovisionofposition,navigationandtime
information taking into account different
requirementsonaccuracyandreliabilitycomingfrom
differentmaritimeapplications.
Figure1.GenericarchitectureoftheIntegratedPNTSystem
ExistingandfutureWorldWideRadioNavigation
Systems (WWRNS) like GPS, GLONASS and
GALILEO are fundamental infrastructures for global
determination of position, navigation and timing
data. Additionally, shore‐side services as part of the
Maritime Service Portfolio (MSP) are used or
considered as candidates to improve the positioning
performance (augmentation services: e.g. IALA
Beacon DGNSS, RTK), to support the ba
ckup
functionality (backup services: e.g. e‐LORAN, R‐
Initial Realization of a Sensor Fusion Based Onboard
Maritime Integrated PNT Unit
R.Ziebold,Z.Dai,L.Lanca,T.Noack&E.Engler
InstituteofCommunicationsandNavigation,GermanAerospaceCentre(DLR),Neustrelitz,Germany
ABSTRACT:Thispa
perintroducesthebasicconceptofthePositionNavigationandTiming(PNT)Moduleas
futurepartofashipsideIntegratedNavigationSystem(INS).CoreofthePNTModuleisasensor fusionbased
processingsystem(PNTUnit).Thepaperwillfocusonimportantaspectsandfirstresultsoftheinit
ialpractical
realization of such a PNT Unit, including a realization of a Consistent Common Reference System (CCRS),
GNSS/IMUtightlycoupledpositioningresultsaswellascontingencyperformanceoftheinertialsensors.
http://www.transnav.eu
the International Journal
on Marine Navigation
and Safety of Sea Transportation
Volume 7
Number 1
March 2013
DOI:10.12716/1001.07.01.17
128
Mode), or to provide PNT relevant Maritime Safety
Information (MSI: e.g. service level, tidal
information).
In our concept, a PNT Module as one part of a
future INS will be responsible for the provision of
PNT information to the shipborne user (see Fig. 1).
Coreof this PNTModule is a PNTUnit, whichuses
theav
ailablenavigationandaugmentationsignalsin
combinationwithadditionaldataofsensorsaboardto
provide accurateand robust PNT information of the
ship[4].
This paper concentrates on this ship‐side part of
the maritime integrated PNT System. Section II
providesanoverviewofthecurrentPNTsystemand
shortly int
roduces our concept of an on‐board
maritime PNT Module. In section III, which is the
mainpartofthe paper, results of theinitialpractical
realizationofaPNTUnitwillpresented.
2 PNTMODULECONCEPT
2.1 Overview
AdetaileddiscussionofourPNTModuleconceptcan
be found in [4], here only a short overview will be
given. Currently, vessels subject to the International
Convention for theSafety ofLife at Sea(SOLAS) [5]
can either use single sensors to provide the PNT
parameters (e.g. position, heading, speed over
ground) individually or use an INS [6]. Fig 2.
represents the single sensor a
pproach. Stand‐alone
equipment provides only sensor‐specific PNT data
e.g. WWRNS sensors for position, velocity and time
data(PVT)andothership‐sidesensorsfornavigation
data(N).Theshipboardprocessinglayerispartofthe
applied sensors and represents the int
ernal used
methodsfortheprovisionofrespectivePNTdata.The
onboard staff has to fuse and asses the navigational
datacomingfromthedifferentsensors.
Figure2.Singlesensorapproach
In the current INS approach, the sensors deliver
their individually determined PNT data to a
shipboardprocessinglayer,whichisillustratedinFig.
3. The INS is performing plausibility checks on the
incoming data and consistency checks on different
sensors utilizing the redundancy within the applied
INS.Integrity,asdefinedinthi
scurrentINSstandard,
isexpected,ifplausibilityandconsistencychecksare
passed[7].
Figure3.ApproachofcurrentINS
Duetothefact,thatnotallpossiblefailuremodes
can be detected, plausibility and consistency tests
alone are insufficient to guarantee the reliability of
INSoutputs(seediscussionin[4]).
Inordertosolvetheseproblemsidentifiedabove,
a PNT data processing unit (short: PNT Unit) is
introduced into the shipboa
rd processing layer of a
futureINS,asillustratedinFig.4.By meansofsensor
fusion techniques, this PNT Unit integrates all
available PVT and N data from onboard sensors in
order to provide optimal PNT output data. In
addition to the current INS approach, the onboard
sensors (here especially the GNSS Receivers) should
also provide their raw data (e.g. ephemerids, code,
doppler and pha
se measurements) to the PNT Unit.
This enables the usage of advanced sensor fusion
techniques and enhanced integrity monitoring
functions in order to improve the resilience of PNT
information. As a new functionality, the PNT Unit
willnotonlyprovideoptimalesti
mationsofthePNT
output data but also integrity information based on
accuracyestimations.
Figure4.PNTUnitapproach
2.2 UnderstandingofIntegrityandIntegrityMonitoring
Integritycanbecategorizedinto“dataintegrity”and
“system integrity”. Data integrity is given, if the
desiredoutputdataisprovidedatthe expected time
inthespecified formats, and meanwhile, thespecific
accuracy requirements are fulfilled. System integrity
is given, if (1) the int
egrity of all output data of a
systemisfulfilledand(2)theoutputdata,additional
statusmessages,andalertmessagesareprovidedina
129
timely,complete,unambiguousandaccuratemanner.
Fromthese definitions it can be seen that the system
integrity can only be given as long as the system
realizes its tasks with the required performance.
According to these definitions, integrity monitoring
needstoincludeerrorestimationsforalloutputdata.
These estimated
errors need to be compared against
given accuracy requirements. Whenever these
requirements are not fulfilled, an alert message
shouldbegeneratedwithinaspecifiedtime.
Integrity monitoring within the PNT Unit can be
carriedoutinthreesequentialsteps,whichhavebeen
discussed in detailin [4]. Thefirst step
is the test of
individual sensors including provided sensor data.
The second step is the compatibility test of similar
datafromdifferentsensors.Thethirdstepisthefault
detection and identification in the integration
algorithm.Examples forspecific realizations of these
stepswillbegiveninthenextsection.
3
INITIALPNTUNITREALIZATION
In orderto demonstrate the functionalities of a PNT
Unit, a prototype of such a PNT Unit is realized
withinourresearchproject.Basisfortherealizationof
thisprototypeisareal‐timecapableC++development
framework, whichhas been developed in ourgroup
and
upgraded especially for the PNT Unit
developments(fordetailssee[8]).Thisframeworkhas
the advantage that identical core algorithms can be
used either in a real time or in a post processing
environment. Furthermore, it enables parallel
processingusingmultiplethreadsofonecomputer.
As a prerequisite for the PNT
Unit prototype
development, a decision about the used sensors had
to be made. The basis for the PNT Unit are the
maritime standard onboard sensors like GNSS
receivers, speed log and gyrocompass. Besides that,
wehavedecidedtouseanInertialMeasurementUnit
(IMU) as an additional sensor. On the
one hand, an
IMUcanbridgeshort‐termoutagesanddisturbations
ofGNSSandenablesthereforetheestablishmentofa
short‐termcontingencyfunctionalityformostofPNT
parameters. On the other hand the diversity of IMU
outputsfurthermore enablesintegritymonitoring for
relevantparameters.AlimitationofIMUinmaritime
navigationliesinaccuracydegradationforlong‐term
operation, so that the integration of IMU with other
navigationsensorsisnecessarytorealizealong‐term
stableoperation.
3.1 Measurementcampaigns
Inorderto collect test datafor the developmentand
validation of the PNT Unit, several measurement
campaigns have
been performed. The examples
shown in this paper will concentrate on a
measurement campaign performed in cooperation
withtheFederalMaritimeandHydrographicAgency
(BSH)onthesurvey andresearchvesselDENEB.The
vessel was equipped with 3 GNSS antennas and
receivers (type: Javad Delta), an IMU (type iMar
IVRUFCAI),
agyrocompass,aDopplerspeedlog,an
electromagnetic speed log and other standard
shipborne sensors. Fig. 5 shows the vessel DENEB,
where the red circles mark the positions of the 3
GNSS antennas and the yellow circle indicates the
position of the IMU installed near the centerline
insidethevessel.
Figure5.VesselDENEBwithsensorlocations
Figure6.TrajectoryofvesselDENEB
Thetrajectoryofthevesselforthetimeslotofone
hour is illustrated in Fig. 6. Leaving the Warnow
River,thevesselperformedananti‐clockwiseturning
maneuverandfinallyshelefttheportandledintothe
Baltic Sea. Based on the master station located near
Rostock port, differential
positioning with carrier
phase measurements have been performed in post‐
processing to obtain the reference trajectory in
centimeteraccuracy.
3.2 ConsistentCommonReferenceSystem(CCRS)
Due to the size of vessels and the distribution of
sensors, the position and velocity information
measuredbydifferentsensorsneedtobeconvertedto
a consistent common reference point (CCRP).
Heading information, as well as the other Euler
angles and their change rates are needed for this
conversion. Therefore, an accurate determination of
ships attitude and their temporal changes are a
prerequisite for PNT parameters determination.
Besides that, the integrity of the other output
parameters
like position and velocity relies also on
the integrity of the attitude information. A detailed
discussion of our PNT Unit based approach of
attitudedeterminationcanbefoundin[10].Hereonly
thebasicideaswillbebrieflyintroduced.
© BSH
130
The maritime standard sensor for heading
determination is the gyrocompass. If it is properly
settled, itprovides long‐term stability. However, the
accuracy depends on the actual ship motion and is
limitedtofewdegrees(see[9][10]).Theusageofa3
antennasGNSS‐Compasswithalargebaselinelengt
h
(aswehaveinstalleditonthe vessel DENEB)yields
an accuracy with a standard deviation:
01.0
forallEulerangles.
100.4
100.6
100.8
angle [°]
heading
100,4 100,8
= 0.01°
0
0.2
0.4
0.6
angle [°]
pitch
0 0.2 0.4
= 0.005°
0.5
0.7
angle [°]
roll
0.4 0.6 0.8
angle [°]
= 0.05°
-0.02 0 0.02
length [m]
=3mm
0:00 0:10 0.20 0:30 0:40 0:50 1:00
-0.1
0
0.1
Local Time
length [m]
base length variation
ant 1-3
ant 1-2
ant 2-3
Figure7. Heading, pitch and roll determination using
GNSS‐compassinquasi‐staticscenarioatport
InFig.7heading,pitchandrollangledetermined
with a GNSS‐compass are shown for quasi‐static
scenario, where the vessel is moored. Additionally,
thechallengesofaGNSS‐compassareshown.
The yellow circle indicates the epochs at which
GNSS‐compassdoesnotprovidereliableresults.The
qualityoftheGNSS‐compasscanbeevaluatedbythe
baseline lengt
h (see lower graph in Fig. 7), which
should stay constant as long as the GNSS carrier
phase measurements are correctly processed.
Unreliable attitude results can be detected by larger
variationsinthebaselinelength.Theseoutliersmight
occur with a failed solution of int
eger ambiguities,
which is the most crucial step within the GNSS‐
compass data processing. In this sense, a GNSS‐
compassoffers high accuracy but limitedavailability
and continuity. In order to overcome these
limitations, a GNSS‐compass should be used in
combination with other sensors, like an IMU. In a
sensor fusion scheme, an IMU can be used for the
detectionofGNSScompassout
liersaswellasforthe
provision of a backup during the times of GNSS
compass outages. Therefore, within our prototype
PNT Unit, an attitude determination based on the
fusion of a GNSS‐compass andan IMU servesas an
accurateandreliableba
sisofaCCRS.
3.3 Integritymonitoringwithcompatibilitytests
As mentioned before, the second step of integrity
monitoring refers to the compatibility test for PNT
dataobtainedfromdifferentsensors.Asanexample,
the compatibility tests for SOG determination are
presentedinthefollowing.
0
1
2
3
4
5
(a) SOG determined by different GNSS antennas
SOG [m/s]
10:15 10:30 10:45 11:00
0
0.2
0.4
(b) SOG difference of antenna 1-2
Local Time
delta SOG [m/s]
antenna 1
antenna 2
antenna 3
sensor raw data
CCRS sensor data
Figure8. (a) SOG determined by the different GNSS
antennas, (b) SOG difference of antenna 2 and antenna1
with/withoutbeingconvertedontoaCCRS
In Fig. 8 (a) SOG data, determined by the three
differentGNSSantennas(seeFig.5)areshown. One
can clearly see systematic differences especially
duringtheturni ng maneuveraround10:30localtime
and at the end of the scenario. As itis illustratedin
Fig.8 (b), these systematicdifferences indeed va
nish
ifthesensorrawdataareconvertedintoaCCRS.For
the SOG compatibility tests within the integrity
monitoring one either accepts larger systematic
differences between distributed sensors or needs to
convert the sensor data into a CCRS before
performingthe compatibility test.The secondoption
has the disadvantage tha
t the integrity tests for one
output parameter (e.g. SOG) depend on the
availabilityandintegrityoftheCCRSitself.
3.4 Integritymonitoringwithinthesensorfusion
ThePNTUnitconcept(seeFig.4)enablestheusageof
sensorrawdatawithinthesensorfusionalgorithm.In
orderto demonstrate theadv
antageofthisapproach
we have implemented a tightly coupled GNSS/IMU
sensor fusion algorithm based on an extended
Kalman Filter. A detailed description of the
implementation can be found in [11]. Here only the
basicideasandresultsarepresented.
In comparison to a loosely coupled GNSS/IMU
Kalman filter, where the position results of a GNSS
receiver is used as an input
, in a tightly coupled
Kalman filter the raw pseudorange measurements
fromeachsatelliteinviewareprocessedinthefilter.
This allows a failure detection of each individual
GNSSobservable.AsanessentialstepoftheKalman
filterroutine,thecalculat
ionoftheinnovationvector
reflects the deviation of the predicted pseudoranges
withrespecttotherealmeasurements.Aslongasthe
dynamic model is working properly, the innovation
vectormainlyreflectsthepotentialfailureshiddenon
each measurement. Based on this fact, the failures
manifest themselves as abnormal jumps in the
innovation sequence ofa specific measurement. This
is the ba
sis for integrity monitoring based on
innovationchecksinaKalmanfilter.
131
-1000 0 1000 2000
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
East [m]
North [m]
GPS only
IMU/GP S
Integration
-10 0 10 20 30 40
600
800
1000
1200
1400
1600
1800
East [m]
North [m]
GPS only
IMU/GP S
Integrati on
Figure9.Tightly‐coupledGPS/IMUintegrationwithandwithoutsatellitefiltering
In the data processing, we use a fixed threshol
d for
innovationmagnitudeofpseudorangesinordertodetect
and remove faulty measurements. The trajectories
processed using single‐point positioning and GPS/IMU
integrationwiththatsatellitefilteraredepictedinFig.9,
where the graph at the left‐hand side shows the whole
tra
jectory within one hour and the graph at the right‐
hand side is a zoom‐in for the first 20 minutes for a
clearerillustration.
10: 00 10:15 10:30 10:45 11:00
0
5
10
15
20
25
30
Local time [hour:minute]
Absolute error [m]
GPS only
GP S/IMU Integ ra ti on
Figure10.Horizontalaccuracyof(1)tightly‐coupledGPS/IMU
integrationwithsatellitefilteringand(2)stand‐aloneGPS
-250 -200 -150 -100 -50 0 50 100
-100
-50
0
50
100
150
200
East [m]
North [m]
GPS only
GPS/IMU (sat. filter)
GPS/IMU (no sat. filter)
Figure11. Tightly‐coupled GPS/IMU integration with and
withoutsatellitefiltering
The GPS/IMU integration gives a much smoother
trajectory comparedto theGPS‐only solutions.In order
to show the improvements in terms of accuracy, the
absolute positioning errors in the horizontal plane are
calculated by comparing with the reference trajectory
andpresentedinFig.10.
Theintegratedsystemshowsstableerrorslimit
edto3
metersformostoftheobservationepochs.Contrarily,the
GPS‐only results show significant outliers. A more
detailedanalysis[11]showsthattheseerrorsarecaused
by signal distortions from low elevation satellites. The
innovationfilterautomaticallydetectsandremovesthese
faultymeasurements.
The conventional GPS/IMU integration works also
without sat
ellite exclusion. The dynamics measured by
IMUcouldsomehowadjustthepositioningerrorscaused
by low measurement quality. In Fig. 11 the
correspondingtrajectoryisalsoillustrated,togetherwith
itscounterpartsusingsingle‐pointpositioningandusing
theintegrationwithsatellitefiltering.
3.5 Contingencyfunctionality
One of the ba
se functionalities of the PNT Unit is the
capabilitytoprovidePNTdataevenincasethatamain
PNT sensor is malfunctioning.For the provision of
positioninformationwecurrentlyhaveonemainsensor:
GNSS receiver 1, with two redundant sensors: GNSS
receiver 2+3 and one contingency sensor: IMU. A
contingencysystem,asdefinedbyIALA[12],allowssafe
completionofa maneuver,butma
ynotbeadequatefor
long‐term use. In order to demonstrate the contingency
functionalityoftheIMUinstrap‐downprocessinginthe
case that all GNSS sensors provide no valid position
information,threeGNSSoutageperiodswithone‐minute
duration are manually set to the GNSS raw
measurements.Beforeandaft
ertheGNSSoutageepochs,
the IMU is tightly integrated with GNSS pseudorange
and Doppler measurements,whereas duringthe GNSS‐
outage epochs, the IMU totally relies on its strap‐down
processing.
132
50 100 150 200 250
0
2
4
6
8
10
Pos Err [m]
Epoch [s]
50 100 150 200 250
0
0.2
0.4
0.6
0.8
1
Vel Err [m/s]
-6 -4 -2 0 2 4 6
0
200
400
600
East [m]
Notrh [m]
GPS/IMU
RTK
IMU only
50 100 150 200 250
0
2
4
6
8
10
Pos Err [m]
Epoch [s]
50 100 150 200 250
0
0.2
0.4
0.6
0.8
1
Vel Err [m/s]
-50 -40 -30 -20 -10 0 10
0
200
400
600
East [m]
Notrh [m]
GPS/IMU
RTK
IMU only
50 100 150 200 250
0
2
4
6
8
10
Pos Err [m]
Epoch [s]
50 100 150 200 250
0
0.2
0.4
0.6
0.8
1
Vel Err [m/s]
120 140 160 180 200 220 240 260 280 300
150
200
250
300
350
East [m]
Notrh [m]
GPS/IMU
RTK
IMU only
Figure12. IMU contingency functionality, left: position error (blue) and velocity error (red), right: trajectories for three
differentGNSSoutagesofoneminute
In Fig. 12, two sub‐figures in the same row
correspondtoaGNSSoutageperiod,wherethesub‐
figure on the left‐hand side shows the horizontal
velocityandpositionerrorsofIMUandthesub‐figure
at the right‐hand side illustrate the trajectory in
horizontal plane.Within one minute, the position
errordrifts toaround10meterandthevelocityerror
to around 0.5 m/s. The concrete performa
nce of the
strap‐down processing depends highly on the
accuracy of initial values of PVT parameters at the
beginningofstrap‐downprocessing.
From Fig. 12 it can also be seen tha
t, once the
GNSS signals become available, the drift effects can
begraduallyadjustedbyGNSS.
In comparison with position parameters, the
velocityerrorisaffectedbyIMUmeasurementerrors
toamuchlargerextent.Thepositiondriftsarecaused
bytheaccumulationofthevelocityerror.Ifwehave
othersensorstoindicat
ethevelocity,thedriftofIMU
velocity error, and moreover the position drifts, will
besignificantlyreduced.Drivenbythis idea,theuse
ofspeedlogcouldserveasaproperbackupsystemto
beintegratedwithanIMU.
Depending on the accuracy requirement in
different operation areas, the IMU can perform the
contingency funct
ionality with different time
durations. For example, in the coastal area, the
position error must be smaller than 10 meter, and
hencetheIMUcontingencefunctionalitycannotwork
for more than one minute. In the deep sea area, the
positionerroristoleratedupto100meters,sotha
tthe
IMU can operate independently for longer time,
however, no more than several minutes, even with
tacticalgradeIMUandgoodinitialization.
3.6 Positionerrorestimation
As discussed in section 2.2 the PNT Unit needs to
provide integrity information to the onboard user,
whichisba
sedonaccuracyestimation.Inafirststep
the GNSS stand‐alone approach based on a classical
Fault Detection Receiver Autonomous Integrity
Monitoring (RAIM) algorithm [13] can be used for
thispurpose.Thistechniquewasfirstlyintroducedin
the aviation sector for using only reliable satellites
during safety‐critical landing a
pproaches. It
determines the integrity of GNSS signals based on
consistency checks among redundant pseudoranges.
The RAIM aims at the determination ofwhether the
system or individual measurements meet the
navigation performance requirements [14]. One
output of RAIM is the Horizontal Protection Level
(HPL). HPL is a statistical bound of the horizontal
position error computed to guarant
ee that the
probabilityofthehorizontalpositionerrorexceeding
that number is smaller than or equal to the target
133
integrityrisk[15].Integrityriskistheprobabilitythat,
at any moment, the position error exceeds a
predefinedmaximum positioningerrorlimitwithout
identification by the integrity monitoring. We
computed the HPL based on the requirements of
integrity risk for future maritime radio navigation
systems as specified by the IMO
[16] for ocean and
coastal navigation mode. The Integrity Risk for
ocean/coastnavigationhereisspecifiedto1e‐5overa
timeperiodof3hours.
Fig.13showsthehorizontalpositioningerror,the
associated error upper bound under 95% confidence
level and the HPL for a time period of
around 40
minutes where the vessel is moored at the port of
Rostock.
The 2D real position accuracy is obtained by
comparingthesingle‐pointpositioningresultswitha
post processed RTK solution. Additionally the
number of visible satellites with the predefined
elevationcut‐offangleof5degreesisshown.
Forthe
calculationofHPLweassume,thatthe pseudorange
error variance for each satellite depends on the
elevationangleinthefollowingform[17]:
2
sin
i
i
i
q
(1)
where
i
is the predefined standard deviation of
the pseudorange error and
i
the elevation angle
ofthe
th
i
satellite.The
i
forallpseudorangeswas
setto3.2meter.
1105 1110 1115 1120 1125 1130
0
5
10
15
20
25
30
35
Minutes of Day (minutes)
HPL [meters]
Horizontal Estimation Error (95%) [meters]
Horizontal Real Error [meters]
Number of Visible Satellites
Figure13. Horizontal protection level, horizontal position
estimation and real error and the number of visible
satellites.
According to Fig. 13, the position 95% error
estimation is around 5 meters which clearly bounds
the real positioning errors for that (short) period of
time.Inordertoreallydrawconclusions,astatistical
analysisofmuchlongertimeperiodsisrequired.The
computedHPLcanbeinterpretedasanoverbound
of
the horizontal error with a probability of 99.99%. It
can also be seen that the HPL will change with the
satelliteconstellation.
This first analysis should be understood as a
starting point forthe discussionwithin the maritime
community how theintegrity should beassessed for
maritimeusers.
4
SUMMARY
Inthispaper,themaritimeintegratedPNTModuleas
the on‐board part of a maritime PNT system is
introduced. The core of the PNT Module is a PNT
data processing Unit enabling the integrated
utilizationofallavailablesensordatatoestablishthe
neededredundancyforimprovedPNTdata
provision
and integrity monitoring. Important aspects and
benefitsofsuchaPNTUnitareshowntogetherwith
examples regarding different integrity monitoring
methods.
TheCCRS,asaprerequisiteforthefusionofdata
from different sensors at different locations, is
introduced.CompatibilitytestsforSOGsensorsshow
the importance of
conversion of sensor data into a
CCRS. A tightly‐coupled GPS/IMU integration
enables the innovation checks for the detection and
removal of outliers in GNSS pseudorange
measurements.The usage of an IMU is not only
beneficial for the integrity monitoring but also
provides contingency functionality.In our case,
where we have
used a tactical grade IMU, the
positionerrorincreasedtoaround10meters and the
velocity error to around 0.5 m/s within one‐minute
GNSS outage. Another important aspect of the PNT
UnitliesintheerrorestimationofPNTparameters.In
a first approach we have adapted a horizontal
position error estimation method for GNSS stand‐
alone positioning from the aviation sector to the
maritimerequirements.Amoredetaileddiscussionof
the required errorestimation, as abasis for integrity
provision for maritime users, will be a subject of
futurework.
ACKNOWLEDGMENTS
TheauthorswouldliketothanktheFederal
Maritime
and Hydrographic Agency of Germany (BSH), in
particular Mr. Tobias Ehlers and the crew of the
DENEBwiththeircaptainMr.Gentesforthesupport
during the measurement activities. The authors are
also grateful for the assistance of their colleagues
PawelBanys,CarstenBecker,AlexanderBorn,Stefan
Gewies, Frank
Heymann, Uwe Netzband, Philipp
Zachhuber.
REFERENCES
[1]IMO, “NAV 54/25 Annex 12 Draft strategy for the
development and implementation of e‐Navigation,”
2008.
134
[2]IMO, “NAV 55/WP.5 Development of an e‐Navigation
StrategyImplementationPlan,”NAV55/WP5,Jul.2009.
[3]GermanyIMO,“NAV58/6/1IMOinputpaper:Proposed
architecture for the provision of resilient PNT data
SubmittedbyGermany.”30‐Mar‐2012.
[4]R.Ziebold,Z.Dai,T.Noack,andE.Engler,“Conceptfor
an onboard integrated PNT Unit,” TransNav, the
International Journal on Marine Navigation and Safety
of Sea Transportation (http://www.transnav.eu), vol. 5,
no.2,pp.149–156,2011.
[5]IMO, “SOLAS CHAPTER V: SAFETY OF
NAVIGATION,”Jul.2002.
[6]IMO,“ResolutionMSC.252(83):AdoptionoftheRevised
Performance Standards for Integrated Navigation
Systems(INS),”
Oktober2007.
[7]IEC, “Maritime navigation and radiocommunication
equipmentandsystems–Integratednavigationsystems,
Operationalandperformancerequirements,methodsof
testingandrequiredtestresults,”2006.
[8]S. Gewies, C. Becker, and T. Noack, “Deterministic
Frame‐workforParallel Real‐Time Processing in GNSS
Applications,”Proceedingsofthe6thESA
Workshopon
Satellite Navigation Technologies. Navitec 2012, 5.‐7.
Dec.2012,Noordwijk,TheNetherlands.,2012.
[9]Z. Dai, R. Ziebold, A. Born, and E. Engler, “Heading‐
determination using the sensor‐fusion based maritime
PNTUnit,”IONPLANS,2012.
[10]IMO,“ResolutionA.424(XI):PerformanceStandardsfor
GyroCompasses,”Nov.1979.
[11]
Z. Dai, R. Ziebold, and E. Engler, “The on‐board
maritimePNTModule–afocusonintegritymonitoring
and preliminary results,” European Navigation
Conference GNSS (ENC GNSS 2011), 28.Nov‐
2.Dez.2011,London,UnitedKindom.
[12]IALA, “Recommendation R‐129: On GNSS
VulnerabilityandMitigationMeasuresEdition2,”2008.
[13]T. Walter
and P. Enge, “Weighted RAIM for Precision
Approach,”1995.
[14]B.Hoffman‐Wellenhof, H.Lichtenegger,andE. Wasle,
GNSS‐Global Navigation System, GPS, GLONASS,
GalileoandMore.2008.
[15]“Integrity,”
http://www.navipedia.net/index.php/Integrity
[16]IMO, “Resolution A.915(22): Revised Maritime Policy
and Requirements for A Future Global Navigation
SatelliteSystem(GNSS),”Nov.2001.
[17]Jean‐Philippe Montillet, Kegen Yu, “On a New
Weighting Matrix to Enhance the Accuracy of Stand‐
AloneGNSSPositioning,” International Symposiumon
CommunicationsandInformationTechnologies(ISCIT),
2012.
