WAASat15ToddWalter,StanfordUniversity

KarlShallberg,ZetaAssociatesIncorporatedEricAltshuler,SequoiaresearchCorporation

WilliamWanner,WilliamJ.HughesFAATechnicalCenterChrisHarris,FAAWAASEngineeringTeamRobertStimmler,RaytheonCompany

AbstractTheWideAreaAugmentationSystem(WAAS)achievedInitialOperatingCapability(IOC)inJuly2003.AtIOC,WAAShad25referencestations intheconterminousUnitedStates,Hawaii,AlaskaandSanJuanPuertoRico, twomasterstations,andfouruplinkstationssupportingtwonarrowbandL1onlyGEOsatellites. Today,WAAShas38referencestations includingfour inCanadaand five inMexico, threemaster stations, and sixuplink stations supporting threewideband L1/L5GEOsatellites.Inadditiontothearchitecturalexpansion,theFAAcontinuedtoevolveandmaintainthealgorithms,hardwareandsoftware. WithWAASnowturningfifteenyearsofage,thispapertakesstockofthesemodificationsanddetailstheimprovedserviceandintegritythathasoccurredsince2003.IntroductionWAASwasthefirstSatelliteBasedAugmentationSystem(SBAS)tobedevelopedwhenitbecameoperationalinJuly2003.ThegoalofanySBASistoprovideprecisenavigationforaircraftdownto200feetabovethegroundoverthewholeofastate’sairspace[1].TheIOCversionofWAASdidnotquitefulfillthatgoal.AlthoughitdidprovidecoverageovermostoftheConterminousUnitedStates(CONUS),itdidnotcoveralloftheairspace,nordiditprovideservicedownto200feet.Inthe ensuing yearsWAAS hasmademany improvements to its service by expanding and upgrading its hardware and bymakingimprovementstoitssoftware[2].ThemodificationstoWAASsinceIOChavetouchedallaspectsofthesystemfromaddingreferencestations,improvingtheionospheric algorithms, refreshing the hardware such as reference receivers and antennas, enhancing communicationsnetworks for increasedcapacityof L5measurements,and integratingnewL1/L5GEOsatellites. Thispaperpresents themajor systemmodificationsover this timeframeanddemonstrates the resulting service improvements. Webelieve thisinformationtobevaluabletootherSatelliteBasedAugmentationSystemproviderslendinginsightintotheevolutionandmaintenanceofasystemascomplexasWAAS.Service improvements are demonstrated using both historical performance and the MatLab Algorithm AvailabilitySimulation Tool (MAAST) to showWAAS availability [3]. MAAST was developed by Stanford and has been proven toaccuratelyreflectavailabilityoftheWAASsystem.ProposedmodificationstoWAASwereevaluatedusingMAASTtoshowtheresultingWAASservice improvementsoverthepastfifteenyearsandwillcontinuetobeusedforupcomingplannedimprovements. WAAS modifications that did not result in performance improvements but were important for safety,continuity,andmaintainabilitywillalsobediscussed. Someofthemorenotablealgorithms implemented inWAASsinceIOCweretheextremestormdetectorandsignalqualitymonitoring.WAASArchitectureOverviewTo provide a frame of reference for the following discussion, Figure 1 provides depiction of the overall architecture ofWAAS.IthasageographicallydiversenetworkofWAASReferenceStations(WRSs)eachcontainingthreeparallelthreadsofequipment.TheseWAASReferenceElements(WREs)eachconsistofaGPSantenna,aGPSreceiver,acesiumclock,anda computer to format the data and send it to theWAAS Master Stations (WMSs). The data is sent along redundantdedicatedcommunicationlinestoensuretimelyarrivalandtominimizetransmissionloss.EachWMShasaCorrectionsandVerification (C&V) subsystem that consists of two parts: a corrections processor (CP) and a safety processor (SP). ThecorrectionsandintegrityboundsestimatedattheWMSsareformattedintomessagesthatareprovidedtoGroundUplink

Stations (GUSs) for transmission to Geostationary Earth Orbit (GEO) satellites. The GEOs downlink this data on the L1frequencywithamodulationsimilartothatusedbyGPS.

Figure1.WAASSystemArchitecture

Figure2.Aschematicofthemajorintegritymonitors.TheCPperformsaninitialscreeningofthedatatoidentifyandremoveoutliers.TheresultingoutputisfedintoafilterthatestimatesthereceiverandsatelliteInter-FrequencyBiases(IFBs),andintofiltersthatestimatetheWREclockoffsets,thesatelliteorbital locations,andthesatelliteclockoffsets. ThesearethenpassedalongtotheSPforevaluation. TheSP isresponsibleforensuringthesafetyoftheWAASoutput.Itwilldecidewhatinformationwillbesenttotheuserandtowhatlevelsuch informationcanbetrusted. TheSPperforms itsown independentdatascreeningonthe inputWREdata. ItsCode Noise andMultiPath (CNMP) monitor [4] performs data screening, carrier smoothing and produces a confidenceboundfortheremaininguncertaintyonthesmoothedpseudorangevalues.Figure2showstheSPmonitoringalgorithmsandinformationflow.TheUserDifferentialRangeError(UDRE)monitortakesinthesmoothediono-freepseudorangesandbounds from the CNMP monitor and uses them to determine a confidence bound on the satellite clock and orbitalcorrectionerrorsfromtheCP.TheCodeCarrierCoherence(CCC)monitorandtheSignalQualityMonitor(SQM)useinputs

from CNMP to determinewhether or not the UDRE bound is sufficiently large to protect against potential code-carrierdivergenceand/orsignaldeformationsrespectively.In parallel, the Grid Ionospheric Vertical Error (GIVE) monitor takes in the smoothed ionospheric delay estimates andboundsfromtheCNMPmonitoraswellastheIFBestimatesfromtheCPtoestimatetheionosphericdelaysandconfidenceboundsforasetofIonosphericGridPoints(IGPs)definedtoexist350kmabovetheWAASservicearea[5].TheuserisabletointerpolatebetweentheseIGPstodetermineanionosphericdelaycorrectionandcorrespondingconfidenceboundforeach of their satellite measurements. The Range Domain Monitor (RDM) then evaluates all of the corrections andconfidencebounds.TheRDMusessmoothedL1measurementsandboundsfromtheCNMPmonitortodeterminewhethercorrections and bounds from the prior monitors combine as expected to bound the fully corrected single frequencymeasurements. Ifthere isaproblem,theRDMmayincreasethecorrespondingUDREandGIVEvaluesor itmayflagthesatellite as unsafe to use. All of this information is then passed to the User PositionMonitor (UPM), which evaluateswhetherallthecorrectedpositionerrorsateachWREareproperlybounded.LiketheRDMittoohastheabilitytoincreasethebroadcastboundsorsetasatelliteasunusable.Finally,thecorrections,UDREsandGIVESarebroadcasttotheuserinasequenceofmessages.The remainder of this paper describes WAAS expansion, significant equipment refresh and maintenance activities,performanceandintegrityenhancements,andanoverviewonaviationadoptionoftheWAASservice.SystemExpansionThe WAAS architectural expansion since IOC focused on improving performance, enhancing robustness, or addressingobsolescence. The addition of both a third C&V and a third GEO increased the robustness of the system by providingadditional redundancy. The addition of Alaska, Canada and Mexico WRSs improved performance by providing bettercoverage inCONUSandAlaskaandalsoexpandingcoveragetoCanadaandMexico. Thereplacementofthe initialGEOswith newer GEOs both improved performance and dealt with obsolescence of aging satellites. The expansion of theterrestrial communications infrastructurewas driven by adding new sites to the system, dealingwith obsolescence andaddingbandwidthinpreparationforfuturedualfrequencycapability.Theparagraphsthatfollowwillprovidemoredetailandrationalforthearchitecturalexpansion.ThirdC&VShortlyafter IOCtheFAAplannedforaseriesofenhancementsthatwouldbe implemented inthesystemoverthenextseveral years. TheFAAhasa stringent requirement that theSignal in Space (SIS)mustbemaintained regardlessof anyupgradeormaintenanceactivitybeingperformedonthesystem.WAASatIOConlyhadtwoC&Vs.Forearlierupgradestothesystem,asoneC&Vwasshutdownfortheupgrade,onlyoneC&Vwasleftrunning.WithonlyoneC&VrunningduringtheseactivitiesWAASwasvulnerabletoalossofSISiftheonlyremainingC&Vfailed.Themitigationforthisriskwastoadda thirdC&V to the system. Witha thirdC&V,asoneof theC&Vs isbeingupgraded thereare still twootherC&VsrunningandprovidingtheneededredundancytomaintaintheSIS.ExpansionofWRSsOne widely recognized enhancement for WAAS has been the addition of reference stations. Although WAAS IOC didprovidecoverageovermostofCONUS,itdidnotcoveralloftheairspace,nordiditprovideservicedownto200feet.TheprimarygoalofaddingWRSstothesystemistomaximizecoverageintheareaandincreaseavailabilityofthoseareasthatdidnot alreadymeet100%availability. The firstpartof thismethodology for choosingnew siteswas to choose criticalpoints. These critical points are the points of the service volumewhere improvements in availability are focused. Thesecondpartofthisanalysisinvolvedrunningcombinationsofsitestoassesstheirbenefitsastheyactuponeachother.Thecombinations of sites are chosen from the list of single sites that perform the best at each critical point. Thesecombinationsarerunwiththefullusergridtomeasureoverallcoverage. A listofpossiblesiteswasproposedbasedonaccessibilityandfeasibilityofalocation.Thesesitesarelimitedtocitieswithproperinfrastructureorplaceswhereproperinfrastructure canbe constructed to support theWRS. Theend resultwas adding thirteen sites. Foursnew siteswereaddedtoAlaska,foursiteswereplacedinCanadaandfivesiteswereplacedinMexico.RefertoFigure3forthelocationofthe original IOCWRSs (red circles), four addedAlaskaWRSs (orange circles) andnine added internationalWRSs (purplediamonds).

Figure3.ThelocationoftheWAASreferencestations

ThirdGEOTwoGEOsaresufficienttoprovidetheneededredundancytomaintainSISfortheWAASsystem.ThereasonforaddingathirdGEOwastomitigatethelengthytimethatwasneededtodeployanotherGEOintheeventthatoneofthetwoWAASGEOsfailed.ThescheduletodeployareplacementGEOcantakeuptothreeyears.Duringthatthree-yearperiod,WAASwouldbeexposedtotheriskofalossofSISiftheremainingGEOfailed,orifthegrounduplinkstationsfailedtoprovideaSIStothatGEO.WAASleasedtheInmarsatAMericanRegion(AMR)satellitefrom2010through2017toprovidethisthirdGEOsignal.GEOReplacementsSince IOC, the InmarsatAtlanticOceanRegion (AOR)andPacificOceanRegion (POR)GEOshavebeen replacedwith theIntelsat Galaxy XV (internally labeled as CRW) and the Telesat Canada Anik F1R (labeled as CRE) satellites. One of thebenefitsoftransitioningtotheCRWandCREGEOswasthattheyprovidedabetterGPSlikeL1rangingsignalthantheIOCGEOs.Withabetterrangingsignal,CRWandCREcanbeusedasarangingsourcetoaugmentthenumberofGPSsatellitesavailable to users. This is especially important tomitigate service gaps causedbyGPSoutages orweakness of theGPSconstellation.AnotherreasonforreplacingtheIOCGEOswastoaddressthefactthattheywereapproachingtheendoftheirexpectedservice life. The FAA typically leases GEO satellites for ten years. Depending of the circumstances, this leasemay beextended,buttowardtheendofthelease,newGEOsatellitesareprocuredtoreplacetheexistingsatellites.CRWandCREreplacedAORandPOR.CRWandCREarenowreachingendoflifeforWAASandtheFAAispursuingthenextsetofGEOsatellites.Eutelsat177WestA(labeledasSM9)willreplaceAMR.TheSES-15satellite(S15)willreplaceCRWandanewseventhGEO (GEO7)will replaceCRE. SM9,S15andGEO7willprovideawidebandwidth rangingcapability. Refer toFigure4forthedifferentGEOcoveragefootprintsoverthehistoryofWAAS.

Figure4.ThedifferentGEOcoveragefootprintsoverthehistoryofWAAS.

TerrestrialCommunicationExpansionTheWAASnetworkcommunicationsinfrastructure,knownastheTerrestrialCommunicationNetwork(TCN),wasdesignedtoredundantlytransportGPSmeasurementandnavigationdatafromeachofthegeographicallydispersedWRSstoeachoftheWMSs,totransporttheto-be-broadcastWAASusermessagesfromtheWMSstotheuplinkstations,andtotransportavarietyofcontrolandstatus informationbetweenallsites.TheTCNusesahubandspokearchitectureconsistingoffourcorecommunicationnodes, twosecondarycommunicationsnodes,anddistributionroutersateachoftheWRSandGUSsites.ThisarchitectureisduplicatedsothatthereareactuallytwofullyredundantnetworksfortransportingWAASdata.The core communications nodeswereoriginally connectedwith 1.544Mbps circuits and the connections from the corenodestoWRSandGUSsiteswerethrough64Kbpscircuits.ThisbandwidthsupportedL1andL2measurementdataonly.InordertosupporttheadditionofL2CandL5signaltypesandpreparefortheeventualtransitionofthesystemintodualfrequencycapability,thenetworkbandwidthwasexpanded.Thisexpansionincludedupgradingtonewrouterhardwareaswell as doubling communications line bandwidths to 128 Kbps for WRS sites and 3.088 Mbps for the core-to-coreconnections.TechRefresh/MaintenanceEveryaspectofWAAShaslargelybeenrefreshedsinceIOC.Thefewexceptionsarerelegatedtosuchitemsasequipmentracksandantennacablesfromtheoriginal IOCsite installations.TherefreshandmaintenanceofWAASdiscussedinthissectionisstructuredbyequipment,operatingsystem/compiler,andmaintenance/siterelocations.Theconcept forWAASwas toutilize to the fullestextentpossibleCommerciallyAvailableOff-The-Shelf (COTS)productswiththebelief thisapproachwouldallow inexpensiveupdatesandgreaterefficiency [6].Overthe lifetimeofWAASthereality has been amixtureof COTS,Non-Development Items (NDI), and customdevelopments specifically for theWAASapplication. Themore significant equipment refresh effortswere the reference receiver, antenna, and processor, along

with the associated Operating System (OS) and compiler. Other equipment refresh efforts more closely followed theCOTS/NDIparadigmandincludedcesiumclocks,communicationrouters,timingreceiversandpowerconditionerstonamejustafew.ReferenceReceiverWAAS is currently using the third generation reference receiver with all three generations being developed andmanufacturedbyNovAtelInc.ThefirstgenerationreferencereceiverwasinternallyfundedanddevelopedatNovAtelandwasessentiallythreeindependentreceiverscontainedwithinthesamechassis.Thekeydifferentiatorsforthisreceiveratits time were L1 C/A multipath mitigation provided by the Multipath-Estimating-Delay-Lock-Loop (MEDLL) and P-Codedelayed correlation technology for semi-codeless L2 tracking. The 2nd and 3rd generation receivers (G-II and G-III) weredirectlyfundedandoverseenbyFAA.TheG-IIreceiverhadthesamecoreL1C/Aandsemi-codelessL2trackingastheG-Ireceiverbutotherkey featureswereaddedsuchasL1C/Acorrelatoroutputs forSignalQualityMonitoring (seeSQM inlatersection), improvedRF interferencerejection,andanexpandablearchitecturesothereceivercouldbeupgradedforGPSL2C/AandL5signalprocessing[7].Ultimately,MEDLLprocessingwasnotusedfromthisreceiverandallWAASGPSL1C/Asignalprocessingsincehasbeenbasedon0.1chipcorrelatormeasurements.TheG-II’s started fielding in2005andwerefullydeployedinWAASwithincorporationofthefinalMexicoandCanadianstationsin2007.TheG-IIIreferencereceiverwasdevelopedtoaddresspendingG-IIEnd-of-Lifeconcernsandincorporateadditionalsignalprocessing needed for dual frequency operations. An important point of emphasiswith this developmentwas ensuringmeasurement performance closely emulated the G-II to minimize algorithm retuning. The receiver development alsorequiredcompliancewithRTCADO-178BLevelDsoftwarecertificationguidelinesandmeetingstabilityrequirements.TheG-IIIhasthechannelcapacitytotrack18GPSsatellitesandassociatedL1C/A,L1C,L2C,L2P(Y)andL5signalsandeightGEOsatellites with L1 C/A and L5 signals. This reference receiver provides improved RF interference mitigation with pulseblanking (G-II had pulse suppression), adaptive carrier tracking, and improved robustness with almanac validation andTiming Receiver Autonomous IntegrityMonitoring (TRAIM) functionality. To address lessons learned fromG-II test andintegration,theFAAaddedatestenvironmentatsixtestsitesatWRSsthatallowedG-IIIdatatobeprovidedinplaceofoperational measurements to a shadow C&V in order to more completely evaluate its performance. The G-III startedfieldinginWAASin2015andcompleteddeploymentin2016.ReferenceAntennaThe reference station antenna used for the first few years ofWAAS servicewas fielded just prior to IOC in 2003. Therelativelylatefinalizationofthisantennawascausedbydiscoveryduringalgorithmvalidationthattheoriginalantennahadunacceptablylargebiaserrorsinpseudorangeasafunctionofazimuthandelevation.TheantennadeployedatIOCwasaNovAtelGPS-600elementintegratedintotheoriginalWAASassembly[8].Whilethisantennaaddressedthecriticalspatialbiasconcerns,oneofitsperformancelimitationswaspoorL2gainwhichresultedinsignalacquisitionathigherelevations(~8 degrees elevation vs. 2 degrees from the original antenna) andultimately degradedperformance since both L1 andL2P(Y)GPSmeasurementsareneededforWAASprocessing.Theoriginalantennavendor,Micropulse,continuedresearchonanantennaelementthatwouldprovideacceptablespatialbiaserrorsforWAASandimproveL2gainperformance.TheywereultimatelysuccessfulandjointlydevelopedthecurrentWAASantennawithFAAwhichincludedmorestringentspatialbiaserrorlimits(<25cmat5degreeselevation)andafilter/lownoiseamplifierwithL5signalreception.Thisantennawasfullydeployedby2007andisstillinservicetoday.ProcessorAtthestartofIOC,WAASusedthreedifferenttypesofprocessorsforitsvarioussubsystems.Thecorrectionsprocessorsandoperations&maintenanceprocessorsattheWMSsiteswereIBMModel397s,theWRSandGUSprocessorswereIBMModel 43ps, and the safety processors at the WMS sites used custom hardware integrating single board computersequippedwithMotorolaMPC7400processors.By2009,allprocessorhardware,exceptforthesafetyprocessors,hadbeenupgradedtoIBMModelp615s.Theseinitialupgradeswererequiredtoaddressperformanceneedsandtoaddressend-of-life concerns. In 2014, another upgradewas performed to replace the p615swith IBMModel 720s. This upgradewasprimarilyforend-of-lifereasons.Sofar,allprocessorsinWAAShavebeenbasedonthePowerPCarchitecture.ThefirstmajorupgradeofthesafetyprocessorssinceIOCisscheduledtooccurin2019.Thiswillupgradewilladdressobsolescenceandwillincreaseperformanceinpreparationfordualfrequencyimprovements.

OS/CompilerRefreshoftheoperatingsystemandcompilerwasasignificantundertakingthattookplacewiththelatestupgradeoftheIBM AIX processors at all sites in the system. The most challenging aspect of the upgrade was outlining an efficientapproach for testing all subsystems that still meets the requirements of DO-178B. The FAA was able to maintain theexisting DO-178B design assurance and verification evidence associated with the Level D software while avoiding theexcessivecostofperforminga100%re-verificationoftheLevelDrequirements.Inordertominimizechangesintroducedduringtheprocessorupgrade,theonlysoftwarechangesallowedaspartoftheprocessorupgradewillbethosechangesnecessary to support theupgradeand FAAdirectedminor changes such as antennapositionupdates. The focusof theanalysis and verificationeffortswason the impactsof the specific changesbeing implemented (processor/OS/Compiler)andtheireffectontheWAASfunctionalbehavior.Thegoaloftheeffortwastoperformsufficientanalysisandverificationto demonstrate that the behavior of the system using the updated system components (i.e., new processor, OS, andcompiler)wasconsistentwiththeexistingapprovedbaseline,theWAASSafetyCaseispreservedandtheresiduallife-cycleriskisreducedtoanacceptablelevel.MaintenanceTheWAASantennapositionupdatesrepresentoneofthemoreimportantperiodicmaintenanceactivities.Fundamentaltothe integrity and accuracy ofWAAS is the surveyed positions of the reference station antennas. At IOC there was noprocess forperiodic resurveyof theantennasoranestablishedcriterion forhow largeantennapositionerrors couldbebeforebecominganintegrityconcern. TheWAASIntegrityPerformancePanel(WIPP)[9]assessedthatpositionerrorforeachWAASreferencestationantennashouldbemaintainedtowithin10cmwhenmeasuredrelativetotheInternationalTerrestrial reference Frame (ITRF) datum for any given epoch. Mexico City was allowed to be within 25 cm given thesignificantsubsidenceatthislocation[10].TheITRFdatumversion(realization)istheoneconsistentwithWGS-84andalsoused forpositionsof theGPSOperationalControlSegmentmonitoringstations.Thebasis for theseerror thresholdswastheyweresmallerthantheantennabiaserror,whichcanbeashighas25cmforlowelevationanglesasmentionedabove.Worst-caseantennabiaserrorsare considered throughout theWAAS integrityanalysis toensure thatall allocations stillbeing satisfied. The reference antenna coordinates are assessed periodically with offline monitoring to determine thetimeframe coordinates should be updated in the operational system to stay within thresholds. Generally, the WAASantennacoordinatesareupdatedonanannualbasis.Othersignificantactivitiesthatcouldbeconsideredundermaintenanceareassociatedwithsiteandequipmentrelocations.OnafewoccasionsbuildingswhereWRSequipmentwaslocatedwasdeemednolongerviableorneededtoberepurposed.Ontheseoccasions,relocationhasrequirednewsiteandmultipathsurveysandinsomeinstancesRFsurveys.Theplanningfor taking one set of equipment down and bringing on new equipment sometimes has resulted inWRSs being out ofoperation for several months until site verification is completed and the new antenna coordinates can be input. TheBillings,MTandColdBay,AKWRSsbothrequiredrelocationinthepast15years.PerformanceImprovementsInadditiontoexpansionoftheWRSnetworkandfieldingofequipmentwithbetterperformance,WAAShasupdatedthealgorithms thatgenerate thecorrectionsandcorrespondingconfidencebounds. Theseupdatesweregenerallymade toimprovetheaccuracyofthecorrections,totightentheconfidencebounds,and/ortoaddressintegrityconcernsthatmaybeidentified.ThissectionwilldescribethealgorithmchangesthatweremadeprimarilytoimproveWAASperformance.AttheendofthesectionwewillshowtheimprovementtothecoverageregionprovidedbyWAAS.Thesubsequentsectionwilladdresschangesthatwereprimarilyaimedatmitigatingintegrityconcerns.CorrectionProcessor(CP)TuningandEnhancementsOneof the firstperformanceupgradesafterWAASwascommissionedaddressedknownanomalies thataffectedsystemperformanceandtheabilitytoanalyzeotheranomaliesorperformanceofthesystem.Incorporationofsoftwarechangesto theC&VCPcapabilitieswere focusedon improvingcontinuity,availability,andaccuracyaswellasaddressingcurrentanomaliesthatwereattributabletothecurrentCPalgorithms.AsummaryoftheCPenhancementsfollows:

Cross-threadcycleslipdetection-Thecross-threadcycleslipalgorithmimprovedtheabilitytodetectcycleslipsthatweremissedbyapolynomialfit.Furthermore,ithelpedpreventfalsecycleslipdetectionsduetophasescintillationbylooseningthecycleslipthreshold inthepolynomial fitwhennocyclesliphasbeendetected. Theneteffectwasasizeablegain insystemavailability.Cross-thread pseudorange edit -This algorithm screens out pseudoranges with excessive multipath by comparing thepseudorangesreceivedbyaWREpairforeachsatellite.Thisscreening,whenperformedbeforeCNMP,allowedbothCNMPandthedataeditortogeneratemoreaccuratephaseoffsetsbyexcludingmultipathmeasurementsfromthephase-offsetestimates.ItalsoallowedWREbiastogenerateamoreaccuratesmoothedpseudorangedifference,whichdirectlyaffectedtheWREbias.WRE Bias - TheWRE bias changes improved the ability to maintain a more consistent bias than the prior algorithms.Furthermore, the mean phase difference calculated in WRE bias allowed for cross-thread cycle slip detection on GEOsatellites.CodeNoiseandMultipath(CNMP)-TheCPCNMPalgorithm,whencombinedwithdataeditorchanges,improvedtheabilitytoscreenoutmeasurementswithexcessmultipathandbettercharacterizethedataweightingusedbytheKalmanfiltersdownstream. When combined, the better screening and more accurate data weighting led to higher overall systemavailability, as well as a noticeable improvement in the convergence of the ionospheric and orbit determination (OD)Kalmanfilters.Orbit Determination Tuning - This consisted of a combination of software enhancements and the tuning of OperationalSystemParameters(OSP)inordertoimproveorbitaccuracy,UDREs,availability,andaddresscurrentsoftwaredeficiencies.ManeuverDetectionTuning-ThischangeincludedmodificationtothemaneuverdetectionsoftwareandOSPsinordertoeliminatefalseGPSsatellitemaneuverdeclarations,improvethecapabilitytodetecttruemaneuversearlier,andeliminateunnecessary processing. The change also protects theOD filter from being degraded due to a failure to recognize themaneuveringsatellites.OldButActiveData(OBAD)Mitigation-ChangesweremadeintheCPtopreventusersfromexperiencingexcessiveerrorwhenusingOBAD,aswellasOBADalarmsinIntegrityDataMonitoring(IDM).ThebasicconceptofalltheOBADavoidancealgorithmswastolimitthechangeincorrectionbythemaximumacceptableerrorforanytimewithinthewindowthatthe“old”correctioncouldpotentiallybeactive.ProcessInputData(PID)ArchitecturalImpact-TheadditionofnewPIDfunctionalityresultedinsomemodificationtothePIDsoftwarearchitectureandsomereorderingoffunctions.Thecross-threadpseudorangeeditwasthefirstnewprocess,followed by cross-thread cycle slip, performpolynomial fit, code noise andmulti-path,WRE bias, and the performdataediting. Also,thereceiverdatamonitoringcapabilitywasreplacedbythecombinationofthenewscreeningcapabilitiesandWREbiasalgorithms.SPCNMPThe SPCNMPmonitor has undergonenumerousmodifications since IOC. Thepurposes of theCNMPalgorithms are toestimateandcorrectforobservedcodenoiseandmultipath,andthentoprovideaconfidenceestimateforresidualerrorinmultipath-corrected L1 and L2pseudorangemeasurements. Toperform this function,CNMPmust check for cycle slips,datagaps,andotheranomaloussignaltrackingconditions.Thealgorithmhascontinuouslybeenpresentedwithchanginginput conditions due to the deployment of different reference station hardware at varied locations with significantlydifferentenvironmentalfactors.Thealgorithmhasundergonemanyimprovementstoimproveitsperformancebothundernominalandchallengingconditions[4].The most important robustness measure added to CNMP was processing all three threads in the SP and then takingadvantage of data editing across threads. Cross thread pseudorange editing in particular allowed mitigation of largemultipatherrorsthatsometimesresultedinCNMPnotpassingitsboundingrequirementatthemostchallengingmultipathsites.Thisaddedlogiccheckedpseudorangebetweenthreadstakingintoaccountgeometrydifferencesbetweenantennas

and satellites as well as average clock differences between threads and compared to a threshold. If the differenceexceededthreshold itwasexcluded fromCNMPprocessing. Thealgorithmalso included isolation logic inanattempt todetermine which thread had the offensive measurement. This logic was also extended to comparing carrier phase tofurther isolate cycle slips. Thiswas especially importantwith northern latitude stations added fromAlaska andCanadawherephasescintillationcanpresentverychallengingtrackingconditionsforL2P(Y)processing[11].Thefullcrossthreadeditlogicwasaddedinthe2007timeframe[4].Inadditiontothisediting,severalspecializeddetectionalgorithmswereadded overtime to assistwith anomalous carrier tracking thatwas found to be commonmode. These conditionsweresourcedtostrongphasescintillationandinoneinstance,specialGPSoperationswithincreasedP-Codesignalpower.The modifications to CNMP mentioned above were intended to improve conditioning of measurements so thatdownstreamintegritymonitorswouldhavemoreaccurateandconsistentrangemeasurementsonwhichtooperate.OneoftherecentmodificationstoCNMPwastoremovetheconservatisminGPSprocessingthatassumesatthestartoftrackworstcasemultipath isalwayspresent. Thisassumption inflatesthe initialerroruncertaintysigmavalue. Generallythisinflation isnotneeded. Thenewlogicutilizes informationfromcrossthreadprocessingtomakeadeterminationonthemultipath environment for each satellite track. The algorithmkeeps trackof instanceswhenboth L1 and L2pass crossthreadpseudorangeedittodetermineifthemultipathhistoryisbenign.Ifthishistoryisbenignandthesatelliteelevationangleisabove30degreesthenCNMPerrorboundisallowedtostartatalowersigmavalue.Thischangehasdemonstratedbenefit forsatelliteanomaliessuchasPRN21where itsuffersperiodic jumps inphasethatcauseCNMPtorestartatallWAASreferencestations[12]orinphasescintillationenvironmentsatstationsonthefringeofWAAScoverage.TheothersignificantperformanceenhancementwithCNMPhasbeenforGEOsatelliteprocessing.ThischangerecognizedthatwiderbandwidthGEOsand0.1chipreceivertrackingresults insmallerworstcasemultipaththanhadbeenassumedwith IOC.The initial unsmoothedCNMPupperbound sigmavalue for theGEOswas reduced from30 to10meters. These tighterconfidence bounds result in the broadcast UDRE for the GEO satellites reaching the floor value of 7.5 meters inapproximately5to6hoursinsteadoftakingmorethanoneday.UDREMonitorTheportionoftheUDREmonitorthatevaluatestheGPSsatellitecorrectionshasnotchangedsignificantlysincetheoriginalIOCbuild.InordertocomputeauseableUDREforasatellite,aminimumofthreeWRSshavetoreturnvalidmeasurementstothatsatellite.IntheIOCbuild,atleastoneobservingWRShadtobewithinCONUSinordertosendaUDREvaluebelow50m. This restriction originally had little impact as 20 of the 25WRSswerewithin CONUS. However, the anticipatedaddition of 13 more WRSs outside of CONUS would cause this rule to severely limit the anticipated performanceimprovement. It was recognized that the restrictionwas not actually necessary in order tomaintain integrity andwasremovedfromtheUDREalgorithmin2005,beforethefieldingof thenewstations. ThresholdsontheallowedvariationarisingfromtroposphericuncertaintywasincreasedduetosomefalsealertingcausedbythenewAlaskanWRSs,andthecalculationoftheintegrityriskwasmadeslightlymoreconservative(occasionallyresultinginlargerrequiredUDREvalues),butotherwisetheGPSsatellitealgorithmsarelargelyunchanged.In contrast, the algorithms that evaluateGEO satellite correctionshaveundergonemore significant changes. Originally,WAASonlysoughttohavetheUDREboundtheGEOrangingerrorwithintheU.S.airspace. However, itwasrecognizedthatusersoutsideofthisairspacedonothaveameanstoidentifywhetherornottheboundappliestothem.AsaresulttheminimumallowablebroadcastUDREvalues,theUDREfloors,wereincreasedfrom7.5mto15mforAOR-Wandfrom15m to50m forPOR in2006. TheMT28parameters [13] for theGEOswereoriginally set so that theprojectederrorboundwasthesameeverywherethroughoutthefootprint. ThesevalueswerechangedtoafixedsetofparametersthatreflectedincreasinguncertaintyforusersfartherawayfromthecenteroftheWRSnetwork. ThesenewfixedGEOMT28parameterswerefieldedin2007alongwithchangestotheUDREcalculationtohaveitoverboundallerrorswithintheGEOfootprint.Finally,thealgorithmthatevaluatesthesafetyofthefixedGEOMT28parametersinreal-timewasimprovedin2008, removing someunnecessary conservatism. Theseupdates, combinedwith the change fromnarrowbandwidth towidebandwidthGEOs,allowedthereductionoftheGEOUDREfloorsbackdownto7.5mTherearemorerecenteffortsunderwaytosignificantlyupdatetheUDREmonitor. Thisnewversion, labeledcovarianceUDRE, envisions amore integrated determination of the UDRE andMT28 parameters [14][15]. The current algorithmsdetermine theUDRE value and theMT28parameters separately. This separation is not optimal as certain conservativemeasuresmayberequiredineachparttoaccountforpossiblebehaviorintheotherpart.ThecovarianceUDREalgorithm

shouldbeabletoprovidetightererrorboundingonboththeGPSandGEOsatellitesandallowWAAStodynamicallyupdatetheGEOMT28parameters.GIVEMonitorIncontrasttotheUDREmonitor,theGIVEmonitorhasundergonesignificantchangesand improvementssince IOC. TheIOCversionusedasimpleplanarmodelforthelocal ionospherearoundeachIGP.Italsocreatedtheconceptofastormdetector todistinguishbetweenquiet ionosphere that couldbewell-modeledanddisturbed ionosphere that required amuchlargererrorbound[5].Whentheionospherewasquiet,theIOCGIVEalgorithmappliedasignificantinflationfactor,calledRirreg,toallowforthepossibilitythattheionospherewasnearlyinthedisturbedstate.ItalsoincludedathreatmodeltoaccountforionosphericanomaliesthatmightnotbeadequatelysampledbythedistributionofWRSmeasurements[16].Thisso-calledundersampledthreatmodelwasempiricallyderivedusingtheworstcollectedionosphericdatapriorto2003.Part of this threat model also described the largest historical ionospheric delay as a function of time of day andgeomagnetic latitude. It exploited the fact that the ionosphere shouldneverhavea largedelay, at certain timesand incertainregions,inordertolowerthecorrespondingGIVEs.ShortlyafterIOCcommissioning,twoofthemostsignificantionosphericdisturbancesevertobesampledwithdenseGPSdataoccurred;oneinOctober2003andtheotherinNovember2003[17].WAASmaintaineditsintegritythroughouttheseevents,butmanyoftheunderlyingmodelsintheGIVEalgorithmsandthreatmodelsrequiredrevisiting.Largeionosphericdelayswereobservedatveryhighgeomagneticlatitude.ItwasdecidedtoremovethealgorithmthatcouldputaceilingontheGIVEbasedontheexpectedmaximumVerticalTotalElectronContent(VTEC).ThismaxVTECalgorithmwasremovedin2006.Theundersampledthreatmodelalsorequiredrevision,however,itwasfoundthatusingtheIOCmethodologywouldlead toa significant reduction inWAAScoverage. Itwas thereforedecided toexpand the ideaof thestormdetector tofurther identify themore significant storms suchas the two thatoccurred in2003. Theseevents,now labeledextremestorms,areveryrarebutwhentheydooccur, isolatedpocketsofdisturbedionospherecanpersist forhoursafterwards.These isolated pockets can temporarily escape sampling by the WRS measurements, leading the GIVE algorithm tomistakenlydeclare the ionosphere tobe ina smoothandquiet state. AnExtremeStormDetector (ESD)wascreated tospecifically react to these rareevents. Anextremestorm isdeclaredonly if very largedisturbancesareobserved foranextendedtime.However,onceanextremestormisdeclared,allGIVEvaluesaresettomaximumuntilatleasteighthoursafternofurtherdisturbancesaredetected. Thus,anESDtripwillpreventWAASfromprovidingpreciseverticalguidanceforatleasthalfaday.ThebenefitisthattheundersampledthreatmodelwhentheESDhasnottrippedcanbedramaticallyreduced. The implementationoftheESDandthe inclusionofthe2003and2004stormdata intothethreatmodeltookplaceinSeptember2007.AlsoincludedinthesameupdatewasareductiontotheRirreginflationterm.Ratherthanassumingthattheionospherewasalwaysinanearstormstate,anupdatedalgorithm,termeddynamicRirreg,utilizestheactualmeasuredlevelofdisturbance[18]. Asthe ionosphere is inaveryquietstatethevastmajorityofthetime,this inflationfactor issignificantlyreduced.Unfortunately,thisreductionexposedaweaknessintheIOCalgorithm.Forsimplicity,Rirregmultipliedboththeionosphericmodelinguncertaintyandthemeasurementnoiseuncertainty.ThissimplificationwasusuallyveryconservativeforstaticRirreg,butwaslesssounderdynamicRirreg.Anewmorecomplicatedanalysiswascreatedthatcorrectlytreatedthesetwoerrorsourcesseparately,creatinganewmeasurementnoiseinflationterm,Rnoise[19].Fortunately,thisnewanalysiswasabletoretainthemajorityoftheGIVEreductionprovidedbydynamicRirregwhilemaintainingtherequiredlevelofintegrity.ThechangetoRirregandinclusionofRnoisemeantthatmanyofthethresholdsandparameters intheGIVEalgorithmwerenottunedtotheiroptimalvalues.AsubsequentroundoftuningidentifiedbetterparametervaluesthatledtoasignificantGIVEreductionwhentheywerefieldedinSeptember2008.Another significant improvement to theGIVEmonitor came from the introduction of kriging [20] [21] [22]. Kriging is atechnique adapted from geospatial statistics tomore accuratelymodel the vertical ionospheric delay at the IGPs. Thekriging GIVE algorithmmodels the ionosphere as having an underlying planar trend with an added stochastic variationsimilartotheIOCalgorithm.Themaindifferenceisthatkrigingassumesthatthestochasticerroriscorrelatedspatially(theIOCalgorithmtreated itas thoughthestochasticcomponentwascompletely independentof location). Krigingaccountscorrelated ionospheric delay variations between eachmeasurement and the IGP as well as from onemeasurement toanother.Themodelofthespatialuncertainty,calledthevariogram,hasanon-zerovalueatzeroseparationdistanceandthen increases as the distance between two points increases. Kriging bettermodels the local variations in the vertical

ionosphere than did the IOC algorithm. This allows the GIVE monitor to function under increasingly disturbed states,creatingmoreaccurateestimateswithouttriggeringanyofthestormdetectors.Thus,minorandmediumdisturbancesintheionospherethatwouldleadtostormdetectortripsandoutagesintheverticalguidanceservice,weremuchlesslikelytotripwithkriging,leadingtosignificantavailabilityandcontinuitygainsforthisservice.MSDWiththemajorityofthethreatsfromtheOctober2003stormremovedbytheadditionoftheESD,asecondextremestormbecamethedriveroftheWAASthreatmodel.OnNovember20thand21st,2003,aplasmajetemittedfromtheequatorialregioncreatinganextremelysharpgradientwhichextendedthroughouttheeasternhalfoftheUS,culminatinginthegreatlakesregion.Unlikeotherextremestorms,thegradientofthisstormwasshowntobesharpenoughthataneventsuchasthis could go undetected when different subsets of the WAAS measurements were excluded. Thus, the threats fromNovember21stwereincludedintheESDthreatmodel.

Figure5.ThedifferentIGPMasksoverthehistoryofWAAS.

AnalysisofaperiodicGIVEspikewasshowntobedrivenby the threatmodeland inparticular, thecriticalpoints in thethreatmodelattachedtothisNovember21st,2003storm.DeeperanalysisofthestormonNovember21st,2003showedthat the maximum chi-squared value of the system, which is the metric that trips the ESD, increased gradually andsignificantly for the November 21st storm. This behavior drove the design for a Moderate Storm Detector, which wasmodeled after the ESD algorithmbutwith a smaller threshold and a short outage time [23]. Aswith the ESD, theMSDallowedthreatstoberemovedatanevenhigherrate,improvingtheperformanceevenmorethanwiththeESD.TheMSDthusdetects lessextremestorms,and in theevent thatno storm ispresent, the threatmodel remainsextremely small.Thus the ionospheric conditionswere split into three states: quiet,moderately disturbed, and extremely disturbed. Byseparatingoutthetrulyquietfromthemoderatelydisturbedanevensmallerthreatmodelcouldbeappliedmostofthe

time.WhentheMSDtrips,buttheESDdoesnot,thepreviousthreatmodelisused,leadingtostillverygoodservice,butwithsomereductionincoverageandwhentheESDtrips,verticalguidanceisnolongerprovided,buthorizontalguidanceisprovidedunderallionosphericconditions.IGPMaskTheIGPmaskforWAAShasalsochangedovertime.Figure5showsthemaskatfourdifferentstagesofoperation.Figure5ashowstheoriginal IOCmaskwith190IGPs,primarilyservingtheCONUSregion. TheIGPsaroundHawaiitransmittheestimatedionosphericdelays,buttheirGIVEsarealwayssettonotmonitored.TheNMsettingisduetohavingonlyoneWRS inHawaiiandnootherreceivermeasurementsoverlapthis region. The lackof redundancymeansthat thesystemcannotfullyruleoutreceiverfaultsfromaffectingthegridestimates. Nevertheless,theestimatesaregenerallyaccurateandmaybeusedfornonsafety-of-lifeapplications.Figure5bshowsthedramaticexpansiontothemaskthatoccurredin2006when131IGPswereaddedtothemask.Thislargernumberincludesdensergridspacingabove55°N,whichallowsbetteraccuracyandlowerGIVEsaboveCanadaandAlaska.In2008someofthewesternmostIGPswereremovedfromthemask,astheywerealmostnevermonitored.Further,thesouthernmost IGPswere also removeddue to equatorial ionospheric behavior observed in that region. Additional IGPswereaddedinthenorthtoprovidebetteravailabilityforAlaskaandCanada.Figure5cshowsthesechangesresultinginamaskcontaining317IGPs.In2010,anadditional12IGPswereremovedfromthesouthagainduetoissueswithmodelingtheequatorialionosphere.ThecurrentmaskwiththesechangesisshowninFigure5d.

Figure6.CoverageareasofLPVatdifferenttimessinceWAAScommissioning

AvailabilityImprovementTo examine the performance improvement provided by these enhancements, we have used our Matlab AlgorithmAvailability Simulation Toolset (MAAST) [3]. This toolset uses the WAAS reference station and satellite locations todeterminewhichmeasurementsshouldbeavailabletoWAASatanygiventimeandthenemulatestheWAASalgorithmstoestimate theexpectedUDREandGIVEvalues. These valuesare thenusedona simulatedgridofusers toestimate theexpectedprotectionlevelvaluesandresultingavailability.IthasbeenfoundthatMAASTisabletoveryaccuratelyestimatetheUDREandGIVEvaluesand isan invaluabletoolto identifythe impactonperformanceduetoproposedalgorithmorreferencestation.Figure6showstheWAASLocalizerPrecisionwithVerticalguidance(LPV)coverageareasoveritshistory.LPVisanapproachproceduresimilartoCategoryIprecisionapproachthatisabletoguideanaircrafttowithin250feetof

the ground [24]. It requires that theVertical Protection Level (VPL) bebelow50mand theHorizontal Protection Level(HPL)bebelow40m.ThedarkpurpleregionsofFigure6showareaswhere100%availabilityofLPVispredictedbyMAAST,whichhasfoundtoalsoaccuratelymatchtheactualperformanceobtainedaftereachWAASupdate.Ascanbeseeninthefigure,thecoverageareahasexpandedsignificantlyovertheyears.InordertodistinguishWAASimprovementsfromchangestotheGPSconstellationstrength,alloftheplotsinFigure6werecalculatedusingthestandard24satelliteGPSconstellationspecifiedinAppendixBof[25].Normally,MAASTisrunusingalmanacscorrespondingtotheGPSconstellationforthedaybeingsimulated. However,whiletheconstellationstrengthhasgenerally improved from2003 to thecurrentday, therecanbeoutagesandothervariations that lead tonoticeablechangesincoverage.Byusingthesamereferenceconstellationforallplots,wecanseparatetheamountofchangeduetoWAASmodificationsfromthoseduetoconstellationchanges.Thisstandard24-satelliteconstellationoftenhassomewhatworseperformancethantheactualconstellationsinceitusuallyhas31operationalsatellites.However,itisnotuncommonfortheretobetimeswherethereferenceconstellationdoesoutperformtheactualone.As identified in previous sections,WAAS has fieldedmany improvements over time and in this section the focus is onavailability after fieldingof themost significant algorithmupdates. WAASdevelopmentwasbroken into several phaseswithPhaseIIcommencingafter IOC.Figure6ashowsthecoveragethatwas initiallyavailableafter IOC. MuchofCONUShadgoodavailability,buttherewerelimitationstowardstheedgesandtherewasverylimitedavailabilityinAlaska.PhaseIIfollowedIOCandoriginallyenvisioned10softwarereleases.Overtimevariousplannedbuildsweresometimesmergedtogetherand/orsplitintomultiplereleasesthatsometimesresultedincreativereleasenames(seetheappendixforafulllistof theseupgradeactivities). Somesoftwarereleasesaddressedprimarilymaintenanceactivitieswith little impactonperformanceandothersincludedmajoralgorithmupdates.ThefirstmajoralgorithmupdatewasincludedinRelease6/7thatwas fieldedon September17, 2007. This release includeddynamicRirreg, ESD, andamuchdenser ionospheric gridnorthof 55°N latitude. In addition to the algorithm changes, this release introduced twoCanadian and threeMexicanWRSs. Prior releaseshadreplacedAORandPORwithCREandCRW,and includedthe fouradditionalAlaskanreferencestations.Figure6bshowstheimprovementinthecoverageareaafterthissetofchanges.Thereissignificantimprovementinthecoverageregion,particularlyontheEastCoast.ThenextmajoralgorithmupgradeoccurredonSeptember22,2008whenRelease8/9.2wasfielded.Thisreleaseincludeda retuningof theGIVEalgorithmparameters toexploit theR6/7algorithmchanges,anda removalof someunusedgridpoints.ThepriorreleaseintroducedthefinaltwoCanadianandtwoMexicanWRSs.TheretuninginparticularresultedinasignificantexpansionofthecoveragetonowprovidehighavailabilitythroughmostofAlaska,CanadaandnorthernMexicoasshowninFigure6c.ThenextphaseofWAASdevelopment,PhaseIII,startedin2009[2].UnderthisWAASFollowOn(WFO)effortthereleasenameswererestartedat1andunderwentsimilarreorganizationandnamechangingovertime.ThenextmajoralgorithmchangewasintroducedonOctober20,2011withWFORelease3A.Thisreleaseincludedkrigingandcorrespondingthreatmodelchanges.AlsotheAMRsatellitewasintroducedshortlybeforethisrelease.Subsequently,Release3A1wasfieldedonJanuary20,2012,toremovesomeofthesouthernmostgridpointsoverMexicoduetopoormodelingoftheequatorialionosphere.ThesechangesareshowninFigure6dandledtobettercoverageinAlaska,butsomewhatworseperformanceon the West Coast and southern Mexico. The kriging however did prevent several ionospheric disturbances over thesubsequent years from tripping the storm detectors. This source of availability improvement is not represented in theMAASTmodeling.AnewphaseofWAASdevelopmentbeganin2014[2]andwithitcameanewnamingconventionforreleasesthatwerenowlinkedtoCalendarYear(CY).ThemostrecentalgorithmchangeintroducedtheMSDwithReleaseCY16.Thischangewas introduced specifically to improve coverage on the West Coast. As can be seen in Figure 6e, the algorithm wassuccessfulinmeetingitsgoalalthoughinpractice,wesometimeslosethiscoverageduetotheactualconstellationstrength(orlackthereof).

IntegrityImprovementsInadditiontotheperformance improvements identified in theprevioussection,sometimesnew integrityconcernswereidentified that requiredalgorithmchanges that led toadegradationof service. Thesechangeswere requireddue toanimproved understanding of the risks. However there was a desire to avoid degrading service. Fortunately, the aboveimprovementswere able to offset and overcome any reduction in coverage thatwould have resulted from fielding thefollowingdesignchangesinisolation.WAASIntegrityResolutionProcess(WIRP)AsWAASwasbeingdeveloped, itbecameclear thataprocess for tracking,assessingandevaluating integrity issueswasneeded.Atthetimeofcommissioning,allknownthreatsweresufficientlymitigated.However,asWAASofflinemonitoringcommenced[10]andasnewandbetterunderstandingoftheintegrityrisksbecameclearer,concernsarosethatneededtobemanagedinsomeway.TheWIRPwasdevelopedtoprovideanefficientmeanstoaddressintegritythreatsagainstthefieldedWAASandaprocessforthemtobeproperlyevaluated.IntegritythreatsareassignedtonodesonthefaulttreeandpreviouslyanintegritythreateithermetitsProbabilityofHazardousMisleadingInformation(PHMI)allocationorfailed.Ifand when an integrity issue surfaces, a problem report is issued. Thematter is deliberated, and if deemed sufficientlyhazardous,aHazardRecordisissuedwithanassociateddeterminationofriskandacceptableexposuretime.Thesevalueswilldeterminethecourseofactionandthetimeframeinwhichtheissueneedstobeaddressed. Higherriskitemswithshortexposurewindowsneedtobeaddressedquickly,potentiallythroughoperatoractionoranexpeditedchangetoanalgorithm.LowerriskitemswithlongerexposurewindowscanbeaddressedbydeeperalgorithmchangesSQMSignal deformations are causedby imperfections in thebroadcast code chips transmittedby the satellites. Rather thanbeing perfectly square waves with upward and downward transitions occurring exactly when desired, the chips havevariationsintheriseandsettimes,finitetransitionslopes,containingovershootandoscillations.Theseimperfectionscanbe further exacerbated by behavior of the receiver front-end filters and tracking loop design [26] [27] [28]. This errorsourcewasfirstnoticedonSVN-19wheretwodifferentreceiversobservedmeterleveldifferencesduetodeformationsinthebroadcastcodechips[29].ThethreatmodelforthistypeofthreatwasverymuchunderinvestigationduringtheinitialdesignandfieldingoftheWAASreferencereceivers.ItwasbelievedthatcomparingmeasurementsfromthetwotrackingloopimplementationsthatfirstobservedtheSV-19threat(i.e.narrowversuswidecorrelatorspacing)wouldbesufficientto mitigate the threat. However, as the threat space became better defined, it became clear that more correlatormeasurementswouldbeneededalongwithamoresophisticateddetectionalgorithm.Unfortunately,itwouldtakeafewyears to design and field this new receiver along with the new Signal QualityMonitoring (SQM) algorithm. Work wasconductedwiththereceivermanufacturerspriorto2003,tobetterunderstandtheirreceiverdesignsandhowtheywouldcompare against our reference receivers. Theunderstanding gained from this effortwas that early user receiversweresufficientlysimilartotemporarilyuseareducedversionofthethreatmodel.ItwasdeterminedthattheremainingriskwasadequatelycoveredbythebroadcastUDREvaluesandthatitwouldbeacceptabletotakenomorethanfiveyearstofieldan algorithm capable of protecting any MOPS compliant receiver design against the full ICAO threat model. The CCCmonitorprovidesprotectionagainstanysuddenchangesinthebroadcastchipshapefromanysatellite.The second generation of receivers provided measurements at nine different correlator spacings. An algorithm wasdesignedthatevaluatedthesymmetryandconsistencyofthechipshapesbroadcastbythedifferentsatellites[26].Initiallythisalgorithmwasused toevaluateperformanceoff-line inorder toensure therewereno latentharmfuldeformations.Overtime,thealgorithmwasevolvedtobecomemoresensitiveandlesslikelytoleadtofalsealerts.ThefirstversionofthisalgorithmwasfieldedinSeptember2008.Itusesfourmetricvaluestoevaluatethedifferencesamongthesatellite.Acommonmodeshapedistortionwould lead to identicalpseudorangeerrorsonall satellites. Suchanerrormodewouldonly affect theuser clockestimateandnot lead to apositionerror. Therefore, themonitor determines and removes acommonmode shapeand themetrics areonlyaffectedbydifferences fromone satellite toanother. Whenanyoneofmetricsexceedsitsthreshold,thesatelliteisflaggedasunusable.Thisflagpersistsfortwelvehoursafterthemetricreturnsbelowthethreshold.ThemagnitudeofthethresholdisafunctionoftheUDREdeterminedbytheUDREmonitor.Thesatellitesignalsarenotall identical. Therearenominaldifferencesthatcreateapersistentlowlevelofdistortiononeachsatellite. Thesenominaldeformations leadtosmallbiasesoneachsatellite. Asignificantefforthasbeenmadetodetermine the upper bounds on the deformation errors as a function of the metric values. An offline analysis has

determinedthemaximumacceptablebiasandthereforetheappropriatethresholdvalueforeachUDRE.SincetheinitialversionofSQM,ithasundergonetuningtoreducetheincidencesoffalsealerts.Noharmfulsignaldeformationfaultshaveoccurred sinceWAAS commissioning, although some smaller changes have been observed [27] [28]. Improved outlierrejectionwasaddedtoreducetheriskofmultipathfromexcessivelyinflatingthemetrics.SomeofthehighersignalpowertestsrunbyGPShascreatedapparentdifferencesandfalsealerts.ImprovedmodelingofthepotentialSQMerrorsallowedthethresholdstobeincreased.IonosphericThreatModelUpdatesThe initial WAAS ionospheric algorithms were outlined in the seminal 2000 paper “Robust Detection of IonosphericIrregularities” [5].Among thesealgorithmswas the ionosphericdelayestimationatan IGP,which computedaweightedaverage of ionospheric Pierce Points (IPPs) in the vicinity of the IGP. The formal error of this fit was used as a majorcomponentoftheGIVE,howeveritwasclearthatthereweresomeerrorslargerthan5.33σwhichwasproblematicinthattheGIVEwasexpectedtoboundto10-7.Assuch,asecondterm,wasaddedtotheGIVEequationwhichensuredthattheGIVEboundedallionosphericirregularities[16].Asthistermwasgeneratedbyremovinglargefractionsofthedatasetandtestingtheionosphericalgorithmsagainstremoveddata,itbecameknownastheundersampledterm.ThistermisasimplelookuptableusingIPPdensity(measuredintermsofthemaximumradiusofthefurthestIPPfromtheIGP)andaRelativeCentroidMetric(RCM).Thethreatmodelisthencomputedbyemployingadatadeprivationschemeinthatsomesubsetofthedataisremovedbeforethealgorithmsareapplied,andallofthedataisthentested.Initially, thedatadeprivationschemeswereanannularscheme,fourhalfplanesandfourthreequadrantschemes.AftertheOctober31standNovember21st2003 storms, thedatadeprivation schemeswereupdated to reflectmore realisticthreats. A single station plus two point removal is used to create large gaps in the IPP distribution to test for localizeddisturbances,anddirectionaldeprivation,whereseveralstationsareremovedsuccessivelyfromone(ofeight)directions.AllIGPsaretestedforthedurationoftheionosphericdisturbance.TheWAAS ionospheric threatmodel isoneof themost significantdriversofperformance, andas such,muchworkandmanyupdateshavebeendone.Theupdatesincludethosethatimproveperformance,suchasIGPmaskupdatesandnewionosphericalgorithms,andthosethatdegradeperformancesuchasaddingnewionosphericstormstotheoverallthreatmodel.OnSeptember17th,2007, thenewstorms from2003wereaddedalongwithother ionospheric algorithms.Thisrepresentsthefirstandonlytimenewstormshavebeenaddedtothethreatmodel,howeveralargeupdateisplannedforCY18whichwillincludeUIVEculling,aschemeforremovingthreatswhichcannotmanifestthemselvesforanyuser.AspartoftheCY18update,thesolarcycle24stormswillbeincluded.RDMTheRDMevaluatestheperformanceofthesatelliteandionosphericcorrectionstogetheroneachWRElineofsight.ItdoesnotusetheinternalIFBandTGDestimatesfromtheCPandthereforeisabletodetecterrorsinthesevaluesthatmaynotbedetectedattheUDREorGIVEmonitor.ItalsodoesnotusetheCPestimatedWREclockbiases.InsteaditdeterminesitsownestimatesoftheWREclockbiasesusingthecorrectedpseudorangeresidualsbasedonthesurveyed locationoftheWREantennas.IntheIOCversionthisclockestimateusedallcorrectedmeasurementsinviewofeachWRE.Further,theRDM used significantly reduced versions of the proposed UDRE and GIVE values for weighting and internal thresholddetermination. When thenewWRSswere fielded inMexico, itwas found that some southward lookingmeasurementsstarted to sample the equatorial anomaly region of the ionosphere. The broadcast GIVE values excluded thesemeasurements, as they did not follow the internalmodels. However, the RDM did use thesemeasurements with lessconservative internal GIVE values. Unfortunately, these measurements occasionally corrupted the RDM WRE clockestimates,whichinturnledtotheflaggingofmultiplesatellitesasunusable.Twoupdateswereputinplacetoaddressthisissue.First,theRDMclockestimateincorporatedmeasurementscreeningtoremoveoutliermeasurements.Second,theinternalRDMGIVEvalueswereincreasedformeasurementsthatwereatriskofpassingthroughtheequatorialanomaly.ThesefixespreventedsubsequentfalsealertsbytheRDM.UPMTheUPMexaminesthecorrectionsandlookstoseeifcorrelatederrorsexisttocreatealargerpositionerrorthanexpected.Like the RDM it uses internal less-conservative versions of theUDRE andGIVE for evaluation. The principle being thatchecking against small thresholdswouldprovideanearly alert compared touserswhoapply themuch largerbroadcast

UDREs and GIVEs. However, before IOC it was discovered that corner cases existed where the UPM did not alwaysguaranteeprotection. Thus, itwas left in as a final check, but itwasnot strictly required tomeet integrity. Instead, aseparateoff-lineanalysiswasperformed todetermine thatadequatemarginexisted in theUDREsandGIVEs, toprotectagainst simultaneous correlated errors. Recently, a new UPM algorithm was developed that does guarantee the userprotectionagainst thecorrelatederror threat[30]. ThisnewalgorithmusesthebroadcastUDREsandGIVEsratherthanhaving to create reduced internal versions. It performs a chi-square checkon the sumof the squareof thenormalizedcorrectedresidualsateachWRE.Amathematicalproofshowsthatuserswillbeprotectedaslongasthischi-squaremetricisbelowaspecifiedresidual.Thisnewchi-squareUPMwasfieldedin2017.Atthemoment,nointegritycreditistakenforthismonitor.However,inthefuture,theUDREand/orGIVEvaluesmaybelowered,toexploittheprotectionnowprovidedbythismonitor.WAASApproachesandEquipageWhentheFAAcommissionedWAASin2003,aparadigmshiftoccurredfortheprecisionapproachservicesofferedbytheFAA.Insteadofhavinglocalsystemsserviceanairportorsinglerunway,therewasnowonesystemthatprovidedservicetotheentireNationalAirspaceSystem(NAS).AviatorscoulduseWAAStoflyverticallyguidedapproachestomanyairportsthefirstdayWAASwasavailable.Onthatfirstdaytherewereover300LNAV/VNAVInstrumentApproachProcedures(IAP)published.Asoflate2017,thenumberofIAPsthatuseWAAShasgrowntoover4000intheNAS.The first LPV IAPs were also published in 2003. Though the number was small (only five), that number has grownsignificantlyoverthe15yearsofWAASoperation. Eachyearsince2003,theFAA’sgoalwastopublishat least400newWAASIAPsperyear,andthisnumberwasalwaysmetandmanytimesexceeded.Figure7showsthegrowthinthenumberof IAPs over the years. At first, IAPs were published primarily to airports and runway ends that have an existing ILS.However,therearenowalmost2,700LPVIAPspublishedtonon-ILSrunwayscomparedtoabout1,100LPVIAPspublishedtorunwayswithanexistingILS.

Figure7.NumberofpublishedLPandLPV/LPV-200procedurespublishedsince2003andtheirlocations

AnotherimprovementtothetypeofIAPspublishedwastheestablishmentoftheso-calledLPV-200IAP[31].Bydefinition,LPV IAPswould not go a decision height any lower than 250 feet. However, after extensive analysis and internationalcoordinationwith ICAO,WAAS (andotherSBAS’s)wereapproved for IAPs thatwent toa200-footdecisionheight. TheanalysisincludedworkbytheFAAthatincludedsimulations,dataanalysisfora3-yearperiod,andotherevaluations.TheresultofthisworkwasthattheFAAbegantopublishLPV-200IAPsinearly2007.Tenyearslater,thereareover1,000LPVIAPspublishedtoadecisionheightoflessthan250feet(i.e.thoseclassifiedasLPV-200IAPs).In 2009, another type of approach procedure was created: the Localizer Precision (LP) approach. This procedure wascreated for runway ends where terrain and/or obstacles do not allow the publication the glideslope portion of theapproach.AnLPprocedureisonewherethepilotmayflyalongaconstantaltituderatherthanataconstantdescentangle.

The tight lateral protection offered by WAAS allows LP IAPs to achieve lower decision altitudes than traditional non-precisionapproachprocedures.Figure7alsoshowsamapoftheairport locationsthathaveLP,LPV,andLPV-200IAPs. Theadditional improvementstoWAASavailabilityandcoveragehaveallowedmoreIAPshavebeenpublished.TheFAAfirstpublishedLPV-200proceduresin CONUS, but there are now LPV-200 procedures in Alaska. LPV approaches are located on theGPS (RNAV) approachcharts.ThereisalineofminimaforLPV.ToknowwhetheranLPVapproachisLPV-200,thedecisionheightmustbebelow250 feet. That is, there isnoLPV-200distinctiononanapproachplate,only theLPV lineofminima is shown. Figure8shows the LPV-200 coverage region over time. As was the case for LPV performance shown in Figure 6, the LPV-200coveragearea increasedsignificantlyover time. Note thatLPV-200approacheswerenotpublishedprior to2007. AsofJanuary4th,2018thereare655LPproceduresat495airportsand3,872LPVproceduresat1,888airports including1029LPV-200approachesat609airports.

Figure8.CoverageareasofLPV-200atdifferenttimessinceWAAScommissioning

Figure8showstheLPV-200coverageregionovertime.AswasthecaseforLPVperformanceshowninFigure6,theLPV-200coverageareaincreasedsignificantlyovertime.ThepreviousWAASMOPS,D0-229D,wasfirstpublishedinDecember2006.ThisversionoftheMOPSwaslaterupdated(toDO-229DChange1)in2013andasubsequentupdate,DO-229Ewaspublishedinlate2016.BecauseasharpincreaseinthenumberofWAASMOPScompliantaviationreceiversweresoldin2007,itcanbeinferredthatmostoftheavionicsinusetodayareDO-229Dcompliant.LargelybasedontheWAASMOPS,theFAApublishedaTechnicalStandardOrder(TSO)thatdescribes the FAA requirements forWAAS avionics. TSO-C145 (“Airborne Navigation Sensors using GPS Augmented byWAAS)andTSO-C146(“AirborneNavigationSensorsusingGPSAugmentedbyWAAS”)arethetwodocumentstheFAAhaspublishedbasedontheWAASMOPS.TheFAAestimatesthatover100,000aircraftareequippedwiththeWAASLPVcapability.TheusageofWAASprovidesalargesafetybenefit–verticalguidanceisavailabletopilotsthatdidn’thavethatcapabilitybefore.InadditiontothelargeusageofWAASbythegeneralandbusinessaviationcommunity,SBASisalsogainingacceptancewithaircarriers.ThefirstscheduledserviceaircarriertoutilizeanLPVapproachoccurredin2009.

TheFAAhasencouragedavionicsmanufacturerstoincludetheWAAScapabilityintheiraircraft.Forexample,theFAAhascontractedwithCMCElectronicstoprovideWAAScapableavionics,withtheLPVfunctionality, intheFAA’sResearchandDevelopmentaircraft.IncludingWAASintheFAAaircraftassistsinsupportofflighttestingotherprograms,suchasADS-B.SummaryandConclusionsWAAShasundergonemanychangesinitsfifteenyearssincecommissioningin2003.Thesechangeswereinresponsetoavariety of needs. The system needed to be maintained so as equipment aged and risked becoming less reliablereplacementswere undertaken. Some changeswere in response to improving the operation ofWAAS. Somewere inresponse to integrity concerns thatwere identifiedafter the initial commissioningofWAAS. Themostobvious changesweretheperformanceimprovementsthatwereputinplacetoexpandtheLPVandLPV-200coverageregions.WAASisneverstatic.Thehardwareandsoftwareareconstantlyevolving.Evenasnewgeostationarysatellitesarebroughtonline,plans for theireventual replacementarealreadybeing formulated. New levelsofservicehavebeencreatedandnewflightproceduresaredevelopedandpublished.Thispaperservesasarecordofthemostsignificantofthesechanges.Afteritscommissioningin2003,WAASperformancewasprimarilylimitedtoCONUS.IntheyearsthatfollowedthatservicewasexpandedtoAlaska,Canada,andNorthernMexico. Theservicealsobecamemorereliableandcapableofprovidingverticalguidancetowithin200feetabovetheground.WAAScontinuestoevolveasnewergenerationsofequipmentarefieldedandnewalgorithmsaredeveloped.Thepurposeoftheseupdatesistocontinuetoprovideverticalandhorizontalguidanceoveragreatercoverageregionandwithincreasingreliability.AcknowledgementsThe WAAS program has benefited from significant contributions across Government, Industry and Academia andacknowledging individualachievementscouldbeapaper in itsownright. Instead,wehavechosentorecognizetheFAAProgramManagersfromIOCtopresentsincetheirleadershipanddevotionwasabsolutelyrequiredforWAAStobefieldedandevolvetowhatitistoday.Tothisend,Mr.LeoEldredgeatIOC,Ms.DeborahLawrencethrough2009,Mr.DeaneBuncethough2016,andcurrently,Mr.DonWilkerson.WewouldalsoliketoacknowledgeMr.TomMcHughwhowastheWAASTechnicalDirectorforalmostthisentireperiod.References[1] Enge, P.,Walter, T., Pullen, S., Kee, C., Chao, Y.C., and Tsai, Y.J., “Wide area augmentation of theGlobal PositioningSystem,”PublishedinProceedingsoftheIEEE,Vol.84,No8,pp.1063-1088,DOI:10.1109/5.533954.[2]Lawrence,D.,Bunce,D.,Mathur,N.G.,andSigler,C.E.,“WideAreaAugmentationSystem(WAAS)-ProgramStatus,”Proceedingsofthe20thInternationalTechnicalMeetingoftheSatelliteDivisionofTheInstituteofNavigation(IONGNSS2007),FortWorth,TX,September2007,pp.892-899.[3]Jan,S.,Chan,W.,andWalter,T.,“MATLABAlgorithmAvailabilitySimulationTool,”PublishedinGPSSolutions,Vol.13.,No.4,September2009.[4]Shallberg,K.andSheng,F.,“WAASMeasurementProcessing;CurrentDesignandPotentialImprovements,”ProceedingsofIEEE/IONPLANS2008,Monterey,CA,May2008,pp.253-262.[5]Walter,T.,etal., “RobustDetectionof Ionospheric Irregularities,”Published inNAVIGATION ,Vol.47,No.2,Summer2001,pp.89-100,DOI10.1002/j.2161-4296.2001.tb00231.x[6] COTS http://www.militaryaerospace.com/articles/print/volume-8/issue-1/news/cots-guides-faa-toward-new-era-of-satellite-aircraft-navigation.html.[7] Auld, J. andManz, A., “TheNext GenerationWAAS Reference Receiver,” Presented January 2005 at the Institute ofNavigation(ION)NationalTechnicalMeeting,SanDiego,California.pp.544-550.

[8]Shallberg,K.andGrabowski,J.,“ConsiderationsforCharacterizingAntennaInducedRangeErrors,”Proceedingsofthe15thInternationalTechnicalMeetingoftheSatelliteDivisionofTheInstituteofNavigation(IONGPS2002),Portland,OR,September2002,pp.809-815.[9] Walter, T., Enge, P., and DeCleene, B., “Integrity Lessons from the WAAS Integrity Performance Panel (WIPP),”Proceedingsof the2003NationalTechnicalMeetingofThe InstituteofNavigation,Anaheim,CA, January2003,pp.183-194.[10]Gordon,S.,Sherrell,C.,andPotter,B.J.,“WAASOfflineMonitoring,”Proceedingsofthe23rdInternationalTechnicalMeetingofTheSatelliteDivisionoftheInstituteofNavigation(IONGNSS2010),Portland,OR,September2010,pp.2021-2030.[11]Altshuler,E.,Shallberg,K.,Potter,B.J.,andWalter,T.,“ScintillationCharacterizationforWAASintheAuroralRegion,”Proceedingsofthe27thInternationalTechnicalMeetingofTheSatelliteDivisionoftheInstituteofNavigation(IONGNSS+2014),Tampa,Florida,September2014,pp.1126-1137.[12]Gordon,S.,Shallberg,K.,Ericson,S.,Grabowski,J.,Morrissey,T.,Lorge,F.,“PRN-21CarrierPhasePerturbationsObservedbyWAAS,”Proceedingsofthe22ndInternationalTechnicalMeetingofTheSatelliteDivisionoftheInstituteofNavigation(IONGNSS2009),Savannah,GA,September2009,pp.1236-1243.[13]Walter,T.,Hansen,A.,andEnge,P.,“MessageType28,”Proceedingsofthe2001NationalTechnicalMeetingofTheInstituteofNavigation,LongBeach,CA,January2001,pp.522-532.[14]Blanch,J.,Walter,T.,Phelts,R.Eric,andEnge,P.,“NearTermImprovementstoWAASAvailability,”Proceedingsofthe2013InternationalTechnicalMeetingofTheInstituteofNavigation,SanDiego,California,January2013,pp.71-77.[15]Blanch,J.,Walter,T.,Enge,P.,Stern,A.,andAltshuler,E.,“EvaluationofaCovariance-basedClockandEphemerisErrorBounding Algorithm for SBAS,” Proceedings of the 27th International TechnicalMeeting of The Satellite Division of theInstituteofNavigation(IONGNSS+2014),Tampa,Florida,September2014,pp.3270-3276.[16]Altshuler,E.S.,etal.,“TheWAASIonosphericSpatialThreatModel,”Proceedingsofthe14thInternationalTechnicalMeeting of the SatelliteDivision of The Institute ofNavigation (IONGPS 2001), Salt Lake City,UT, September 2001, pp.2463-2467.[17]Komjathy,A.,Sparks,L.,Mannucci,A.J.,andCoster,A.,“TheIonosphericImpactoftheOctober2003StormEventonWAAS,”Proceedingsofthe17thInternationalTechnicalMeetingoftheSatelliteDivisionofTheInstituteofNavigation(IONGNSS2004),LongBeach,CA,September2004,pp.1298-1307.

[18]Altshuler,E.,Cormier,D.,andGo,H.,“ImprovementstotheWAAS ionosphericalgorithms,”Proceedingsofthe15thInternational Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 2002), Portland, OR,September2002,pp.2256-2261.[19]Blanch, J.,Walter,T., andEnge,P., “MeasurementNoiseversusProcessNoise in IonosphereEstimation forWAAS,”Proceedingsof the2003NationalTechnicalMeetingofThe InstituteofNavigation,Anaheim,CA, January2003,pp.854-860.[20] Blanch, Juan, “Using Kriging to bound Satellite Ranging Errors due to the Ionosphere,” Ph.D. Dissertation, StanfordUniversity,December2003.

[21]Sparks,L.,J.Blanch,andN.Pandya(2011),“Estimatingionosphericdelayusingkriging:1.Methodology,”RadioSci.,46,RS0D21,doi:10.1029/2011RS004667.[22] Sparks, L., J. Blanch, andN.Pandya (2011), “Estimating ionosphericdelayusingkriging:2. Impacton satellite-basedaugmentationsystemavailability,”RadioSci.,46,RS0D22,doi:10.1029/2011RS004781.[23] Sparks, L. and Altshuler, E., “Improving WAAS Availability Along the Coast of California,” Proceedings of the 27thInternationalTechnicalMeetingofTheSatelliteDivisionofthe InstituteofNavigation(IONGNSS+2014),Tampa,Florida,September2014,pp.3299-3311.[24]Cabler,Hank,DeCleene,Bruce,"LPV:New,ImprovedWAASInstrumentApproach,"Proceedingsofthe15thInternationalTechnicalMeetingoftheSatelliteDivisionofTheInstituteofNavigation(IONGPS2002),Portland,OR,September2002,pp.1013-1021.[25] RTCA, DO-229E, “Minimum Operational Performance Standards for Global Positioning System/Satellite-BasedAugmentationSystemAirborneEquipment(SBASMOPS),”preparedbyRTCASC-159,December,2016.[26]Phelts,R.E.,Walter,T.,Enge,P.,“TowardReal-TimeSQMforWAAS:ImprovedDetectionTechniques,”Proceedingsofthe 16th International Technical Meeting of the Satellite Division of the Institute of Navigation, ION GPS/GNSS-2003,September2003.[27] Phelts, R. Eric, Altshuler, Eric,Walter, Todd, and Enge, Per K “ValidatingNominal Bias Error LimitsUsing 4 years ofWAASSignalQualityMonitoringData,”PresentedApril2015attheInstituteofNavigation(ION)Positioning,NavigationandTimingConference,Honolulu,Hawaii.[28]Shallberg,K.W.,Ericson,S.D.,Phelts,E.,Walter,T.,Kovach,K.,andAltshuler,E.,“CatalogandDescriptionofGPSandWAAS L1 C/A Signal Deformation Events,” Presented January 2017 at the Institute of Navigation (ION) InternationalTechnicalMeeting,Monterey,California.pp.508-520.[29] Edgar, C., Czopek, F., Barker, B., “A Co-operative Anomaly Resolution on PRN-19,” Proceedings of the 12thInternational Technical Meeting of the Satellite Division of The Institute of Navigation (ION GPS 1999), Nashville, TN,September1999,pp.2269-2268.[30]Walter,T.andBlanch,J.,“ImprovedUserPositionMonitorforWAAS,”PublishedinNAVIGATION,Vol.64,No.1,Spring2017,pp.165-175,DOI10.1002/navi.180.[31] BillWanner, BruceDeCleene, David A. Nelthropp, and StephenGordon, “Wide Area Augmentation SystemVerticalAccuracyAssessmentInSupportofLPV200Requirements”,NAVIGATION,JournaloftheInstituteofNavigation,Volume55,Number3,Fall2008,pp.191-203.AppendixWAASReleaseHistory)

Date Release Contents July 10, 2003 04:10 UTC

IOC Initial WAAS Service

March 2004 CR 1 Minor performance improvements and anomaly fixes February 10, 2005 LPV R1 Communication capacity upgrade and minor performance improvements June 23, 2005 LPV R2 Upgrade 15 WRS to GII receivers, two thread pseudorange edit and pseudorange

smoothing, remove 1 in CONUS rule from UDRE, switch from MEDLL to narrow. January 24, 2006 LPV R3 3rd Master Station, expand IGP mask June 23, 2006 LPV R4 Added 4 WRSs in Alaska, slope check in CNMP, dual-frequency cycle slip detection,

increased AOR/POR UDRE Floors, disabled Max VTEC algorithm

August 7, 2006 LPV R4.2 Billings WRS Relocation November 9, 2006

LPV R5.1 Added CRW as a datalink only

July 13, 2007 LPV R5.2 Added CRE with 50 m UDRE floor July 30, 2007 LPV R5.4 Remove AOR/POR September 27, 2007

LPV R6/7 Add 3 Mexico/2 Canada WRS, implemented Dynamic Rirreg and Extreme Storm Detector (ESD). Included 2003 & 2004 storms in undersampled threat model. Expanded IGP mask around Alaska, Mexico and East Coast. Replaced IGPS above 55 degrees north with the much denser Band 9 IGPs (131 IGPs added). CRW UDRE floor set to 50 m

March 6, 2008 LPV R8/9.1 Add 2 Mexico and 2 Canada Reference stations September 22, 2008

LPV R8/9.2 SQM, Kriging coded (but not implemented), GIVE monitor was retuned, Unused IGPs in the pacific and furthest south of Mexico were removed from the IGP mask, new points in the northeast were added, GEO UDRE floor set to 7.5, Relative L1L2 Bias monitor

November 2009 WFO R1 CP measurement processing enhancements, misc. anomalies resolved. Added a SQM Filter Timeout Reset algorithm, change GEO unobserved bias from a static OSP to a value coming from CNMP that reduces after initial warm-up

November 11, 2010

WFO R2A Added AMR as a data only signal

June 16, 2011 WFO R2B Misc. anomalies resolved October 20, 2011 WFO R3A Fielded Kriging, AMR NPA GEO Ranging, SQM minor mod, other anomalies

resolved January 20, 2012 WFO R3A1 Removed southernmost IGPs: removed all nine IGPs at 10 degrees north latitude and

the three western most IGPs at 15 degrees north latitude. July 2012 WFO R3B GUS switch over improvements, TJ comm node, other anomalies resolved July 2013 WFO R4 WAAS build merge/cleanup, other anomalies resolved, changes to RDM: added

median edit on clock estimate and updated phi function, UPM Kfa increased July 2015 R43 Prep for GIII fielding, SQM Smax set to 2 to prepare for mixed GII/GIII state

AMR set as data only signal via operator action (due to chronic performance issues) August 2016 R45 SQM Smax set to 4 for GIII-only state August 24-27, 2016

R46-CY16 Added GPS CNMP fast recovery, GEO CNMP amplitude reduction, and the Moderate Storm Detector (MSD) with Cycle 23 only supertruth V3 threat model. Increased DNU limit from 4 to 8.

October 18 2016 R47-DFO R1

Processors upgrades for WRS, GUS, O&M and Corrections Processors at C&Vs

August 16-22, 2017

R48-CY17 Chi-square UPM, reduce false maneuver detections

November 9, 2017

R48-CY17 Removed AMR

March 2018 R49-DFO R2/3

GEO5 and G-III multicast (L2C and L5 data returned from WRSs)

November 2018 R51-CY18 Minor CNMP Tuning for IIF, Minor SQM Tuning for G-III and “High C/A”. Inclusion of GIVE Floor culling of the undersampled threat model + incorporation of Cycle 24 data using supertruth V5.

February 2019 R52-DFO R4

C&V Safety Processor Upgrade

September 2019 R53-DFO R5

GEO6/GUS Safety Processor Upgrade