Raimondi Final

8/12/2019 Raimondi Final

1/197

1

TTHHSSEEEn vue de l'obtention du

DDOOCCTTOORRAATTDDEELLUUNNIIVVEERRSSIITTDDEETTOOUULLOOUUSSEEDlivr parlInstitut National des Sciences Appliques de Toulouse

Discipline ou spcialit :Traitement du Signal

JURYProf. Dr. Francis Castani - PrsidentProf. Dr. Ren Landry - Rapporteur

Prof. Dr. Emmanuel Duflos - RapporteurDr. Christophe Macabiau - Directeur de thse

Dr. Olivier Julien - ExaminateurIng. Jean-Michel Perre - Examinateur

Ecole doctorale :Mathmatiques Informatique Tlcommunications de ToulouseUnit de recherche :Laboratoire de Traitement du Signal et des Tlcommunications de lENAC

Directeur(s) de Thse :Christophe Macabiau

Prsente et soutenue par Mathieu RAIMONDILe 24 Octobre 2008

Titre : Dveloppement et Caractrisation de techniques de rduction dinterfrencespulses pour rcepteurs GNSS embarqus

(Development and Characterization of Pulsed Interference Mitigation Techniques for on-board GNSS Receivers)


2/197

2


3/197

3

Les organismes de standardisation de laiation ciile (OACI, RTCA, EUROCAE) mnentactellement des tdes sr l'tilisation des sstmes de naigation par satellite fornissantne coertre globale, tels GPS o Galileo, en tant qe moen de naigation embarqniqe. LOACI regrope lensemble de ces sstmes de naigation satellitaires et de lers

sstmes dagmentation sos la dnomination GNSS. Por des raisons de scritidentes, les performances des rcepters GNSS embarqs doient garantir des minimapropres chaqe phase de ol et chaqe procdre dapproche. Ces eigences deperformances sont spcifies dans les spcifications des performances oprationnellesminimales, docments pblis (o en cors de pblication) par les atorits sscites.Le GNSS est en passe dtre amlior par la diffsion de noea signa. Parmi e, lessigna Galilo E5 et GPS L5 deraient permettre lamlioration d serice de naigation parsatellite. Cependant, ces signa seront mis dans ne bande dj tilise par des sstmesradiofrqences. Il est donc primordial de sassrer de la possibilit de la cistence de cessstmes. Pls particlirement, il est ncessaire de sassrer qe les rcepters GNSS

embarqs tilisant les signa sscits respectent les eigences de performance. Lamenace principale a bon fonctionnement des rcepters GNSS tilisant les signa E5/L5 at identifie comme tant les missions plses des sstmes DME, TACAN, JTIDS et MIDS.Sans moen de ltte contre ces interfrences, les performances des rcepters GNSSembarqs peent tre dgrades de manires significaties, empchant les rcepters dese conformer a eigences de scrit sr lensemble d monde, et pls particlirementsr des points chads aant t identifis comme les lie o limpact de cesinterfrences sr lesdits rcepters est la pls importante. De techniqes de rdctiondinterfrences ont t proposes por ltter contre cette menace, le Blanker temporel et leFreqenc Domain Interference Sppressor (FDIS).

Le Blanker temporel est ne techniqe de traitement d signal consistant en n test depissance, relatiement simple mettre en re et dont la capacit de rjection desinterfrences a t dmontre sffisante por assrer les eigences de laiation ciile danstotes les phases de ols sr lensemble d monde por les rcepters GPS et Galilotilisant respectiement les signa L5 et E5, dans [Bastide, 2004]. Totefois, cettetechniqe permet de respecter les eigences aec ne marge faible, dans lesenironnements les pls riches en interfrences, atrement dit les points chads .En reanche, le FDIS est ne techniqe de ltte contre les interfrences plses beacoppls eigeante en termes de ressorces, pisqe base sr lecision des interfrences dansle domaine frqentiel. Cependant, elle permet ne amlioration sensible des performancesd rcepter, et donc ne agmentation des marges isis des eigences fies parlAiation Ciile.


4/197

4

Le FDIS a t propos comme ne alternatie a blanker temporel, mais ses problmesdimplantation et ses performances nont t qe pe tdis. La dissertation a por bt departiciper cette tde de performance afin de alider son intrt. Le plan de la thse est lesiant : tot d'abord, les signa de naigation, Galileo E5a/E5b et GPS L5, les interfrences

plses, ainsi qe ler impact sr les performances des rcepters GNSS sont prsents.Ensite, ne description des techniqes de sppression dinterfrences (blanker temporel,FDIS), lers caractristiqes thoriqes et les dgradations de performances sbies par nrcepter GNSS tilisant ces techniqes en prsence d'interfrences plses sontprsentes. Les conditions dans lesqelles ont t obtenes ces performances, cest dire lechoi des scnarios jos ainsi qe des paramtres obsers, o encore les otils desimlation sont dcrits. La conclsion rsme lanalse des performances, les compare aeigences de lAiation Ciile, et propose des recommandations por la conception dercepters GNSS embarqs.


5/197

5

A

Ciil Aiation standardisation bodies (ICAO, RTCA, EUROCAE) are crrentl inestigating these of the Global Naigation Satellite Sstem (GNSS) as a standalone naigation soltion forciil aircraft. For obios safet reasons, onboard GNSS receiers mst garantee minimmperformance reqirements in gien phases of flights. These reqirements, dependent pon

the sstem and signals sed, are stated in the Minimm Operational PerformanceSpecification (MOPS), pblished (or being pblished) b the corresponding athorities.With that respect, the ftre se of Galileo E5 and GPS L5 bands has raised, among others,interference isses. Indeed, preeistent RF sstems emit in this band, ths interfering iththe E5/L5 signals. The main threat as identified as being DME/TACAN grond beaconsplsed emissions. Withot an mitigation capabilit, these sstems can distrb the properoperation of onboard GNSS receiers, preenting them from compling ith safetreqirements. To Interference Mitigation Techniqes (IMT) hae been proposed to fightthis threat, the Temporal Blanker and the Freqenc Domain Interference Sppressor (FDIS).The Temporal Blanker techniqe offers a fairl simple implementation and as shon to

proide enogh benefits to ensre that the specified reqirements ere met in all phases offlight for a GPS L5 or Galileo E5 receier. Hoeer, it as also demonstrated that thereslting performances ere meeting the reqirements b onl a small margin on the orstDME/TACAN interference enironment that can be fond in Erope and USA, so called theEropean and USA hot spots.In contrast, the FDIS is a more demanding mitigation techniqe against plsed interferencein terms of reqired resorces bt improes the performances of the receier, ths alloinglarger margins ith respect to the ciil aiation reqirements.The core of the std is the analsis of the performances of GNSS receiers sing FDIS asIMT. The dissertation architectre is the folloing: first, the naigation signals, Galileo

E5a/E5b and GPS L5, as long as the interferences that constitte a threat for GNSS naigationand their impact on GNSS receiers operations are presented. Then, a description of thestdied IMTs (Temporal Blanker, FDIS), their theoretical characteristics and the theoreticalderiations of the postcorrelation C/N0 degradation sffered b a receier sing thesetechniqes in presence of plsed interference are depicted.

Afterards, all the reslts obtained concerning the IMTs performance assessments arepresented. Firstl, the Figres Of Merit chosen to anale the performance of bothtechniqes are presented and their choice is motiated. Then, the chosen interference andsignal scenarios, along ith the simlation tools and means are finel detailed. Finall, aconfrontation of Temporal Blanker and FDIS performances is gien sing the preiosldescribed FOMs. The conclsion smmaries the performances analsis, compares them to


6/197

6

ciil aiation performances reqirements, and proposes recommendations for onboardGNSS receiers design.


7/197

7

A

First, I old like to acknoledge the Eropean Commission, ho fnded m PhD thesis,throgh its 6th frameork program and more particlarl the Anastasia project. Moregenerall speaking, I old like to thank the Eropean Commission for fnding ambitiosresearch projects throghot Erope. I also thank some members of the ANASTASIA

consortim, ho I had the chance to meet. It as a pleasre to ork ith KristneBIRKELAND and Reto TRNDLE from DATARESPONS, Lis NUNES from Sksoft, GordanaMIJIC from Ascom, Gioanni GROSSO from Sele commnications, Stephane ROLLET andJeanYes CATROS from Thales Aionics.

En second lie, jaimerais remercier ltat Franais, qi permet chaqe citoen de siredes tdes sprieres, sans distinction de classe sociale ni dorigine ethniqe.

Je tiens assi remercier lEcole Nationale de lAiation Ciile, et pls particlirement ledpartement EL et son Laboratoire d Traitement d signal et des Tlcommnications, a

sein dqel sest drole ma thse. Ces remerciements sadressent assi bien MonsierFarid ZIZI, directer des tdes de lENAC, q Monsier Lcien MAZET, chef ddpartement EL, et q tos les membres d LTST. Je remercie assi Cath MIGOT etColette ROY, por ler aide prciese a secrtariat.

Je remercie assi mon directer de thse Christophe MACABIAU, qi arait p figrer enpremier dans cette liste. Il a s me commniqer sa passion por la recherche, mais assiinstarer ne ambiance de traail dtende et agrable dans son laboratoire, tot enconserant ne trs bonne qalit techniqe ainsi qn grand sens de lorganisation. Il lireste totefois amliorer son toch de balle.

Il mest impossible de ne pas remercier Oliier JULIEN, mon ngre , por aoir sii detrs prs mes traa. Il est assi do dne grande qalit techniqe et il est trs agrablede traailler ses cts, il a donc t trs prcie drant ces 3 ans et je le remercie millefois. Son tocher de balle est bien meiller qe celi de Christophe MACABIAU.

Je remercie assi les doctorants qe je p rencontrer et ctoer a LTST. Tos sont animsdne forte criosit scientifiqe, et il a t agrable et motiant de traailler aec e. Jeremercie ainsi Philippe PAIMBLANC, Emilie REBEYROL, Damien KUBRAK, Hanaa ALBITAR,Adre GIREMUS, Benjamin CHIBOUT, Anas MARTINEAU, Christophe OUZEAU, Na TAO,Marianna SPANGENBERG, Ael GARCIA, Pal THEVENON, Pierre NERI, Adrien CHEN, DamienSERANT, et Gal.


8/197

8

Je remercie assi Antoine BLAIS, enseignant lENAC, por son aide et ses connaissances entraitement nmriqe d signal. Il est, li assi, anim dne grande criosit scientifiqe, ceqi est fort agrable. Je li sohaite bonne chance por sa thse.

Je remercie les professers LANDRY et DUFLOS, mes rapporters de thse, ainsi qe leprofesser Francis CASTANIE, le prsident de mon jr de sotenance, ainsi qe JeanMichelPERRE. Je tiens remercier ce dernier por sa participation en tant qe membre de mon

jr, ainsi qe por sa collaboration tot a long d projet ANASTASIA. Cest n ingnier deqalit, tant a niea techniqe qhmain.

Je remercie assi mes parents, et de manire gnrale ma famille, qi mont permis deporsire mes tdes jsqa bot.

Enfin, je remercie ma femme, Nathalie, por le bonher qe nos aons partag drant cesannes, et qi a contrib me mettre dans de bonnes conditions de traail.


9/197

9

C

EME............................................................................................................................ 3ABAC......................................................................................................................... 5ACKNOLEDGMEN....................................................................................................... 7ABLE OF CONEN......................................................................................................... 9LI OF FIGE.............................................................................................................. 13LI OF ABLE ............................................................................................................... 17LI OF MBOL AND ACONM................................................................................. 19INODCION .............................................................................................................. 22

MOTIVATION ............................................................................................................................22THESIS CONTRIBUTION...............................................................................................................24THESIS ORGANIZATION ...............................................................................................................25

CHAPE I : GALILEO E5A/E5B AND GP L5 IGNAL AND HEI INEFEENCEENIONMEN............................................................................................................... 27

I.1.GPSL5SIGNAL STRUCTURE...................................................................................................27I.1.1. GPS L5 PRN Codes................................................................................................................ 28I.1.2. Naigation Message and Snchronisation seqence .......................................................... 29

I.2.GALILEO E5A/E5B SIGNAL STRUCTURE ....................................................................................30I.2.1. GALILEO E5 Spreading Codes Characteristics ...................................................................... 31I.2.2. Naigation Message and Snchroniation Seqence .......................................................... 32

I.3.INTERFERENCE THREATS IN THE E5BAND ..................................................................................32I.3.1. DME/TACAN Signals............................................................................................................. 33I.3.2. JTIDS/MIDS Signals .............................................................................................................. 36

I.4.GNSSRECEIVERS ARCHITECTURE............................................................................................ 37I.4.1. Antenna and Preamplifier ................................................................................................... 38I.4.2. Donconersion and Filtering............................................................................................ 38I.4.3. Sampling and Qantiation ................................................................................................. 39I.4.4. Signal Processing.................................................................................................................. 41

I.4.4.1. Correlators otpt modelling ................................................................ .................................... 42I.4.4.2. Acqisition Process Principle ................................................................ ...................................... 44I.4.4.3. Tracking Process ......................................................................................................................... 47

I.4.5. Naigation Processing.......................................................................................................... 49I.5.PULSED INTERFERENCE IMPACT ON GNSSRECEIVERS OPERATION..................................................49

I.5.1. Atomatic Gain Control Sensitiit to Interference ............................................................ 49


10/197

10

I.5.2. Correlators sensitiit to DME/TACAN Signals.................................................................... 50I.6.CIVIL AVIATION REQUIREMENTS ............................................................................................. 51

I.6.1. Contet ................................................................................................................................ 51I.6.2. Antenna and RF front end reqirements ............................................................................ 51I.6.3. Acqisition, Tracking and Data demodlation Thresholds.................................................. 54I.6.4. TTFF Reqirements.............................................................................................................. 54I.6.5. Interference Scenarios......................................................................................................... 55

CHAPE II : PLED INEFEENCE MIIGAION ECHNIQE HEOEICAL D . 57II.1.REVIEW OF EXISTING TECHNIQUES......................................................................................... 57II.2.TEMPORAL BLANKER DESCRIPTION......................................................................................... 59

II.2.1. Oerie............................................................................................................................. 59II.2.2. Blanker Dt Ccle.............................................................................................................. 60II.2.3. Temporal Blanker Threshold Determination...................................................................... 61

II.3.FDISDESCRIPTION.............................................................................................................63II.3.1. Oerie............................................................................................................................. 63II.3.2. Windo Sie Effect on FDIS Performance.......................................................................... 64

II.3.2.1. Algorithm Configration ..................................................................... ....................................... 65II.3.2.2. Windoing Effects on FT Estimate ............................................................ ................................ 65II.3.2.3. Simlations ................................................................................................................................ 69II.3.2.4. C/N0 Simlations........................................................................................................................ 71II.3.2.5. Conclsion.................................................................................................................................. 72

II.3.3. Use of eighting indos.................................................................................................. 72II.3.3.1. Weighting indos theor........................................................................................................ 72II.3.3.2. Additional Calclation Indced.................................................................................................. 74

II.3.4. FDIS Threshold Determination ........................................................................................... 75II.3.5. Qantiation impact on FDIS operation............................................................................. 77

II.3.5.1. No interference.......................................................................................................................... 78II.3.5.2. Plsed Interference.................................................................................................................... 78

II.3.6. FDIS Sstem Characteriation............................................................................................. 80II.3.6.1. FDIS phase.................................................................................................................................. 82

II.4.THEORETICAL POSTCORRELATION C/N0DEGRADATION PREDICTIONS.......................................... 82II.4.1. Temporal Blanker postcorrelation C/N0 degradation deriation ...................................... 83II.4.2. FDIS postcorrelation C/N0 degradation deriation............................................................ 84

II.4.2.1. Usefl Signal Poer Degradation .............................................................................................. 85II.4.2.2. Thermal Noise densit degradation........................................................................................... 85II.4.2.3. Interference indced noise densit .......................................................... ................................. 86II.4.2.4. Carrier to noise densit ratio degradation................................................................................. 87II.4.2.5. FDIS Eqialent transfer fnction calclation............................................................................ 88

CHAPE III : IMLAION ENIONMEN.................................................................. 97

III.1.SCENARIO OF INTEREST.......................................................................................................97III.2.STUDIED FIGURES OF MERIT (FOM).....................................................................................97III.3.SIMULATION TOOLS DESCRIPTION......................................................................................... 98

III.3.1. PULSAR............................................................................................................................... 98III.3.1.1. Signal Generation Modle .................................................................. ...................................... 98III.3.1.2. Receier Modle..................................................................................................................... 102

III.3.2. Anastasias Galileo Receier Mockp............................................................................. 111III.3.2.1. Signal Generation Deices ...................................................................................................... 111III.3.2.2. Mockp Receier ................................................................... ................................................ 112

III.3.3. GIRASOLE Mockp receier............................................................................................ 119


11/197

11

III.3.4. Acqisition Time Simlator Description .......................................................................... 119III.3.4.1. Acqisition Time Simlations.................................................................................................. 119III.3.4.2. Assmptions ........................................................................................................................... 120III.3.4.3. IMTs impact on Crosscorrelation peaks ................................................................................ 122

CHAPE I : EL ANALI.................................................................................125IV.1.THEORETICAL DERIVATION TOOL ........................................................................................125

IV.1.1. Temporal Blanker ............................................................................................................ 127IV.1.2. FDIS.................................................................................................................................. 128

IV.2.PULSAR .......................................................................................................................130IV.2.1. C/N0degradation simlation reslts ............................................................................... 130

IV.2.1.1. IMTs Performances in absence of interference...................................................................... 131IV.2.1.2. DME/TACAN Signals Onl, RTCA antenna gain assmptions.................................................. 131IV.2.1.3. DME/TACAN Signals Onl, EUROCAE antenna gain assmptions .......................................... 133IV.2.1.4. Weighting indos .............................................................. .................................................. 134IV.2.1.5. JTIDS/MIDS Signals Onl......................................................................................................... 135IV.2.1.6. DME/TACAN + JTIDS/MIDS Signals......................................................................................... 135

IV.2.2. AGC Gain.......................................................................................................................... 137IV.2.3. Code and Phase tracking accrac .................................................................................. 138

IV.3.ANASTASIAGALILEO MOCKUP RECEIVER .........................................................................139IV.3.1. AGC Performances........................................................................................................... 139IV.3.2. C/N0degradation simlation reslts ............................................................................... 140

IV.4.GIRASOLEMOCKUP RECEIVER ........................................................................................141IV.5.LINK BUDGET..................................................................................................................142IV.6.ACQUISITION TIME SIMULATIONS .......................................................................................143CROSSCORRELATION RESULTS ...................................................................................................146

CONCLION .................................................................................................................147CONCLUSIONS ON FDISPERFORMANCES......................................................................................147RECOMMENDATIONS ON E5/L5RECEIVERS ARCHITECTURE..............................................................150OTHER POSSIBLE INTERFERENCE MITIGATION TECHNIQUES............................................................... 150ORIGINAL CONTRIBUTIONS REVIEW.............................................................................................150FUTURE WORK.......................................................................................................................151

EFEENCE ..................................................................................................................153APPENDI A : AGC LOOP D.....................................................................................158

A.1.LOOP BEHAVIOUR IN STANDARD DEVIATION MODE ................................................................158A.1.1.CLOSED LOOP TRANSFER FUNCTION..................................................................................158A.1.2.TIME CONSTANT DETERMINATION....................................................................................161A.1.3.STABILITY..................................................................................................................... 162A.1.4.MATLABSIMULATION ..................................................................................................163

A.1.1.1. Signal Generation .............................................................................................................. 163A.1.1.2. Epected Simlation Reslts.............................................................................................. 163A.1.1.3. Simlation Reslts ............................................................................................................. 164A.1.1.4. Conclsion ......................................................................................................................... 167

A.2.LOOP BEHAVIOUR IN DISTRIBUTION ESTIMATOR MODE ...........................................................167A.2.1. Closed Loop Transfer Fnction......................................................................................... 168A.2.2. Time Constant .................................................................................................................. 168


12/197

12

A.1.5.STABILITY..................................................................................................................... 169A.1.6.MATLAB SIMULATION RESULTS ........................................................................................170

A.1.1.5. Epected Reslts................................................................................................................ 170A.1.1.6. Simlation Reslts ............................................................................................................. 170

A.3.CONCLUSIONS AND RECOMMENDATIONS ..............................................................................172APPENDI B : EOPEAN HO PO DME/ACAN BEACON CHAACEIIC..............173APPENDI C : M EN FANAI..............................................................................175

C.1.DESCRIPTION DES SIGNAUX .................................................................................................176C.2.DESCRIPTION DES EFFETS DES INTERFRENCES SUR LES RCEPTEURS GNSS ..................................178

C.2.1. Effet des Interfrences sr la Bocle de CAG................................................................... 179C.2.2. Effet des Interfrences Plses sr les Sorties de Corrlaters....................................... 180

C.3.TECHNIQUES DE LUTTE CONTRE LES INTERFRENCES................................................................. 181C.4.DESCRIPTION DES OUTILS DE SIMULATION .............................................................................182

C.4.1. PULSAR ............................................................................................................................. 182

C.4.2. Otil de prdiction de dgradation de C/N0..................................................................... 187C.4.3. Maqette ANASTASIA....................................................................................................... 191

C.5.EXIGENCES AVIATION CIVILE ...............................................................................................193C.6.TEMPS DACQUISITION.......................................................................................................194C.7.CONCLUSION................................................................................................................... 196


13/197

13

L F

FIGURE 1 :NORMALIZED GPSL5PRN1XICODE PSD.................................................................... 29FIGURE 2:NORMALIZED GALILEO E5A PRN1CODE PSD. ................................................................31FIGURE 3:NORMALIZED GALILEO E5B PRN1CODE PSD. ................................................................ 31FIGURE 4:DMESIGNAL PATTERN.................................................................................................34FIGURE 5:NORMALIZED DME/TACANSIGNAL ENERGY SPECTRAL DENSITY. .......................................35FIGURE 6 :EUROPEAN DME/TACANGROUND STATIONS. ...............................................................35FIGURE 7:USDME/TACANGROUND STATIONS............................................................................36FIGURE 8 :NORMALIZED JTIDS/MIDSPULSE. ...............................................................................37FIGURE 9 :GNSSRECEIVER ARCHITECTURE. ..................................................................................37FIGURE 10 :CLOSEUP TO QUANTIZATION:AGCADCBLOCK. ..........................................................39FIGURE 11:EXAMPLE OF UNIFORM CENTRED 8BIT QUANTIZATION LAW............................................... 40FIGURE 12 :SNRDEGRADATION AT CORRELATOR OUTPUT IN PRESENCE OF THERMAL NOISE ONLY.............41FIGURE 13:ARCHITECTURE OF A CORRELATOR. ...............................................................................42FIGURE 14:CROSSCORRELATION FUNCTIONS BETWEEN RECEIVED AND LOCAL E5/L5CODES. ..................43FIGURE 15 :ACQUISITION PROCESS UNCERTAINTY REGION.................................................................45FIGURE 16:TEST CRITERIA PDF. ...................................................................................................47FIGURE 17:GNSSRECEIVERS SIGNAL PROCESSING MODULE SCHEME. ................................................48FIGURE 18:RTCAAND EUROCAEANTENNA GAIN PATTERNS WORST CASE ASSUMPTIONS...................... 52FIGURE 19:E5MAXIMUM OUT OF BAND RFILEVELS. ....................................................................... 53FIGURE 20:E5/L5INTERFERENCE MASKS. .....................................................................................53FIGURE 21:POLYPHASE FILTER IMPLEMENTATION SCHEME. ..............................................................58FIGURE 22:SYNTHESIS BANK IMPLEMENTATION SCHEME. ................................................................ 58FIGURE 23 :PROBABILITY OF EXCEEDING A THRESHOLD..................................................................... 60F

IGURE24:

B

DC AS A FUNCTION OF THE THRESHOLD,ASSUMING A THERMAL NOISE OF DENSITY

200DBW/HZ........................................................................................................................... 62FIGURE 25:C/N0DEGRADATION AS A FUNCTION OF THE THRESHOLD,ASSUMING THERMAL NOISE ONLY.....62FIGURE 26: FDISARCHITECTURE.................................................................................................63FIGURE 27 :PSDESTIMATES OF THERMAL NOISE AND CWUSING THE PERIODOGRAM. ...........................64FIGURE 28:SPECTRAL LEAKAGE ON A CW......................................................................................66FIGURE 29:PSDESTIMATES OF SLICES OF SIGNAL. ...........................................................................67FIGURE 30 :CLOSE UP ON THE SPECTRAL LEAKAGE MECHANISM PULSED INTERFERENCEPSDESTIMATE,USINGOBSERVATION WINDOWS LOCATED AROUND THE CENTRE OF ONE PULSE,MODULATED AT 0HZ. ................ 68FIGURE 31:DIRICHLET KERNEL AMPLITUDE AROUND ZERO.................................................................69FIGURE 32 :FDISEFFICIENCY AS A FUNCTION OF THE NUMBER OF POINTS USED IN FFTCALCULATION,CONSIDERING A PULSE PAIR MODULATED AT 1176MHZ. ..................................................................70


14/197


15/197

15

FIGURE 70 :E5A/L5C/N0DEGRADATION DUE TO DME/TACANBEACONS OVER EUROPE AT FL400,CONSIDERING RTCAANTENNA GAIN ASSUMPTIONS........................................................................126FIGURE 71:E5B C/N0DEGRADATION DUE TO DME/TACANBEACONS OVER EUROPE AT FL400,CONSIDERING RTCAANTENNA GAIN ASSUMPTIONS........................................................................126FIGURE 72:E5A/L5C/N0DEGRADATION DUE TO DME/TACANBEACONS OVER EUROPE AT FL400,CONSIDERING RTCAANTENNA GAIN ASSUMPTIONS AND A BLANKING THRESHOLD OF 117.1DBW. .......127FIGURE 73 :E5B C/N0DEGRADATION DUE TO DME/TACANBEACONS OVER EUROPE AT FL400,CONSIDERING RTCAANTENNA GAIN ASSUMPTIONS AND A BLANKING THRESHOLD OF 120DBW............128FIGURE 74:E5A/L5C/N0DEGRADATION DUE TO DME/TACANBEACONS OVER EUROPE AT FL400,CONSIDERING RTCAANTENNA GAIN ASSUMPTIONS,USING FDISAND A THRESHOLD OF 195DBW/HZ..129FIGURE 75:E5B C/N0DEGRADATION DUE TO DME/TACANBEACONS OVER EUROPE AT FL400,CONSIDERING RTCAANTENNA GAIN ASSUMPTIONS,USING FDISAND A THRESHOLD OF 195DBW/HZ..129FIGURE 76:E5A/L5AGCLOOP GAIN IN VARIOUS INTERFERENCE CONDITIONS. ..................................137FIGURE 77:E5B AGCLOOP GAIN IN VARIOUS INTERFERENCE CONDITIONS......................................... 138FIGURE 78:REQUIRED NUMBER OF CORRELATORS VS.DWELL TIME FOR THE FIRST SATELLITE ACQUISITION.............................................................................................................................................144FIGURE 79:REQUIRED NUMBER OF CORRELATORS VS.DWELL TIME FOR SUBSEQUENT SATELLITE ACQUISITION.............................................................................................................................................145FIGURE 80 :EQUIVALENT LINEAR MODEL SCHEME. ....................................................................... 159FIGURE 81:GALILEO MOCKUP RECEIVER IMPLEMENTED RCFILTER................................................... 159FIGURE 82 :AGCOPEN LOOP TRANSFER FUNCTION BODE DIAGRAM USING VARIANCE........................162FIGURE 83:AGCGAIN EXPECTED BEHAVIOUR USING TEMPORAL BLANKER. ........................................ 164FIGURE 84:AGCGAIN EXPECTED BEHAVIOUR USING FDIS..............................................................164FIGURE 85:AGCGAIN WITH INTERFERENCE FREE SIGNAL AT RECEIVER INPUT USING VARIANCE ESTIMATOR.

............................................................................................................................................165FIGURE 86:AGCGAIN WITH INTERFERENCE AT RECEIVER INPUT USING VARIANCE ESTIMATOR. ............... 165FIGURE 87:AGCGAIN WITH INTERFERENCE AT RECEIVER INPUT USING VARIANCE ESTIMATOR AND TEMPORALBLANKER................................................................................................................................ 166FIGURE 88:AGCGAIN WITH INTERFERENCE AT RECEIVER INPUT USING VARIANCE ESTIMATOR AND TEMPORALBLANKER................................................................................................................................ 166FIGURE 89:AGCGAIN WITH INTERFERENCE AT RECEIVER INPUT USING VARIANCE ESTIMATOR AND FDIS. 167FIGURE 90 :DISTRIBUTION ESTIMATOR VALUE AS A FUNCTION OF THE GAIN. ...................................... 168FIGURE 91 :AGCOPEN LOOP TRANSFER FUNCTION BODE DIAGRAM USING DISTRIBUTION ESTIMATOR...169FIGURE 92 :DISTRIBUTION ESTIMATOR VALUE AS A FUNCTION OF THE GAIN. ...................................... 170FIGURE 93 :AGCGAIN WITH INTERFERENCE FREE SIGNAL AT RECEIVER INPUT USING DISTRIBUTIONESTIMATOR............................................................................................................................. 171FIGURE 94 :AGCGAIN WITH INTERFERENCE AT RECEIVER INPUT USING DISTRIBUTION ESTIMATOR..........171FIGURE 95 :AGCGAIN WITH INTERFERENCE AT RECEIVER INPUT USING DISTRIBUTION ESTIMATOR AND FDIS.............................................................................................................................................172FIGURE 96:OCCUPATION DE LA BANDE E5/L5PAR LES SIGNAUX GNSSET SRNA...............................177 FIGURE 97:REPRESENTATIONS TEMPORELLES ET SPECTRALES DES SIGNAUX DME/TACAN. .................178FIGURE 98:CARACTERISTIQUES TEMPORELLES ET SPECTRALES DES SIGNAUX JTIDS/MIDS. ..................178FIGURE 99:ARCHITECTURE DES RECEPTEURS GNSS.......................................................................179FIGURE 100:BOUCLE DE CAG................................................................................................... 180F

IGURE101:

P

ERTES DEQ

UANTIFICATION EN FONCTION DE LECART TYPE DU SIGNAL ENTRANT

. .............180


16/197


17/197

17

L

TABLE 1:L5SIGNAL CHARACTERISTICS ........................................................................................... 28TABLE 2:1MS XIAND XQCODES PROPERTIES................................................................................29TABLE 3 :E5SIGNAL CHARACTERISTICS.......................................................................................... 30TABLE 4:GALILEO E5PRNCODES CARACTERISTICS.......................................................................... 32TABLE 5:C/N0THRESHOLDS FOR GPSL5AND GALILEO E5A AND E5B SIGNALS....................................54TABLE 6 :IGEBWG1CASE VIIIJTIDS/MIDSSCENARIO .................................................................56TABLE 7 :POSTCORRELATION C/N0DEGRADATION AS A FUNCTION OF THE NUMBER OF SAMPLES USED INFFTCALCULATION. .................................................................................................................... 72TABLE 8:WEIGHTING WINDOWS CHARACTERISTICS. ........................................................................ 73TABLE 9:SNRDEGRADATION DUE TO WEIGHTING WINDOWS.............................................................74TABLE 10:AGCLOOP SETTINGS ................................................................................................117TABLE 11:FRONTEND FILTER CORRELATION LOSSES FOR E5A/L5AND E5B SIGNALS. ...........................130TABLE 12:C/N0MEASUREMENTS IN INTERFERENCE FREE ENVIRONMENT WITH FIXED AGC...................131TABLE 13:C/N0DEGRADATIONS IN DME/TACANEUROPEAN HOT SPOT ENVIRONMENT WITH RTCAANTENNA GAIN ASSUMPTIONS AND OPTIMALLY FIXED AGC..............................................................132TABLE 14 :C/N0MEASUREMENTS IN DME/TACANEUROPEAN HOT SPOT ENVIRONMENT WITH RTCAANTENNA GAIN ASSUMPTIONS AND REGULATED AGC. .................................................................... 133TABLE 15:C/N0MEASUREMENTS IN DME/TACANEUROPEAN HOT SPOT ENVIRONMENT WITH EUROCAEANTENNA GAIN ASSUMPTIONS AND OPTIMALLY FIXED AGC..............................................................134TABLE 16:C/N0MEASUREMENTS IN DME/TACANEUROPEAN HOT SPOT ENVIRONMENT WITH RTCAANTENNA GAIN ASSUMPTIONS AND OPTIMALLY FIXED AGCUSING FDISAND WEIGHTING WINDOWS........134TABLE 17:C/N0MEASUREMENTS IN JTIDS/MIDSIGEBCASE VIIIENVIRONMENT WITH RTCAANTENNAGAIN ASSUMPTIONS AND OPTIMALLY FIXED AGC............................................................................135T

ABLE18:

C/N0

M

EASUREMENTS INDME/TACAN

+

JTIDS/MIDS

ENVIRONMENT WITHRTCA

ANTENNA

GAIN ASSUMPTIONS AND OPTIMALLY FIXED AGC............................................................................136TABLE 19:C/N0MEASUREMENTS IN DME/TACAN+JTIDS/MIDSENVIRONMENT WITH RTCAANTENNAGAIN ASSUMPTIONS AND REGULATED AGC...................................................................................136TABLE 20:CODE PHASE STANDARD DEVIATION (CHIPS) .................................................................. 138TABLE 21:CARRIER PHASE STANDARD DEVIATION (RAD).................................................................139TABLE 22:AGCSENSITIVITY TO PULSED INTERFERENCE.................................................................. 140TABLE 23 :ANASTASIAGALILEO MOCKUP RECEIVER:E5B SIMULATIONS UNDER EUROCAEINTERFERENCECONDITIONS. ..........................................................................................................................141TABLE 24 :GIRASOLEMOCKUP RECEIVER PERFORMANCES OVER EUROPEAN HOT SPOT,WITH EUROCAEAIRCRAFT ANTENNA GAIN ASSUMPTIONS.......................................................................................142TABLE 25:LINK BUDGET FOR SEVERAL IMTS. ...............................................................................143


18/197

18

TABLE 26:L5PRNCODES CROSSCORRELATION SUSCEPTIBILITY TO IMTS......................................... 146TABLE 27:EUROPEAN HOT SPOT DME/TACANBEACONS CHARACTERISTICS......................................173 TABLE 28DEGRADATIONS DE C/N0AVEC LE BLANKER TEMPOREL ET LE FDISAU HOT SPOT POUR LES SIGNAUXE5A/L5OBTENUS AVEC PULSARET LOUTIL DE PREDICTION. ..........................................................189TABLE 29:DEGRADATIONS DE C/N0AVEC LE BLANKER TEMPOREL ET LE FDISAU HOT SPOT POUR LE SIGNALE5B OBTENUS AVEC PULSARET LOUTIL DE PREDICTION.................................................................191TABLE 30:COMPARAISON DES PERFORMANCES AU HOT SPOT POUR LE SIGNAL E5B OBTENUS AVEC LAMAQUETTE ANASTASIA,AVEC SCENARIO DINTERFERENCES EUROCAE...........................................192TABLE 31:PERFORMANCES DE LA BOUCLE DE CAGDE LA MAQUETTE ANASTASIA.............................192 TABLE 32:SEUILS DE C/N0POUR LES SIGNAUX GPSL5ET GALILEO E5A ET E5B. ................................193TABLE 33:BILAN DE LIAISON. ....................................................................................................193


19/197

Introdction

19

L A

Acronms

ADC AnalogtoDigital ConerterAGC Atomatic Gain Control

AltBOC Alternatie Binar Offset CarrierANASTASIA Airborne Ne Adanced Satellite Techniqes and

Technologies in a Sstem Integrated ApproachARNS Aeronatical Radio Naigation SericeBdc Blanker Dt CcleBER Bit Error RateC/N0 Carrier to Noise densit ratioCW Continos WaeDAC Digital to Analog ConerterDFT Discrete Forier Transform

DME Distance Measring EqipmentDPU Digital Processing UnitDSP Digital Signal ProcessorDTI Direction de la Techniqe et de lInnoationE5 E5 freqenc band centred in 1191.18 MHEMLP Earl Mins Late PoerEUROCAE EURopean Organiation for Ciil Aiation

EqipmentsFDAF Freqenc Domain Adaptie FilteringFDIS Freqenc Domain Interference Sppressor

FEC Forard Error CorrectionFFT Fast Forier TransformFIR Finite Implse ResponseFOM Figre Of MeritFPGA FieldProgrammable Gate ArraGIRASOLE Galileo Integrated Receiers for Adanced Safet

Of Life EqipmentGNSS Global Naigation Satellite SstemGPS Global Positioning SstemI&D Integrate and Dmp filterICAO International Ciil Aiation OrganiationICD Interface Control Docment


20/197

Introdction

20

IGEB Interagenc GPS Eectie BoardIIR Infinite Implse ResponseILS Instrment Landing SstemIMT Interference Mitigation Techniqe

ITU International Telecommnication UnionJTIDS Joint Tactical Information Distribtion SstemKcps Kilo chip per secondL5MIDS Mltifnctional Information Distribtion SstemMLS Microae Landing SstemMMI Man Machine InterfaceMOPS Minimm Operational Performance SpecificationOS Open SericePdf Probabilit densit fnction

Pfa Probabilit of false alarmPmd Probabilit of missed detectionP/NRZ/L PolarNonRetrntoZeroLeelPRF Plse Repetition FreqencPRN Psedo Random NoisePSD Poer Spectral DensitPULSAR PULSe Assessment RotinePVT Position, Velocit and TimeQPSK Qadra Phase Shift KeingQZSS QasiZenith Satellite Sstem

RCU Receier Control UnitRF Radio FreqencRTCASNR Signal to Noise RatioSoL Safet Of LifeTACAN TACtical Air Naigation sstemTHAV Thales AionicsTTFF Time To First FiVGA Variable Gain AmplifierVOR Ver high freqenc Omnidirectional Range

Smbols

Eleation angle Thermal noise poer redction at correlator

otpt de to frontend filteringC Total signal transmitted poer

IC Interference coefficient

Earllate spacing

D Naigation message

Phase offset beteen the receied and local


21/197

Introdction

21

carriers

Interference carrier freqenc

FDIS FDIS Eqialent transfer fnction

P

Predetection bandidth

S Sampling freqenc

G Normalied DME/TACAN beacon antenna gain

G Normalied aircraft antenna gain

0H NemanPearson test hpothesis 0

1H NemanPearson test hpothesis 1

BBH Freqenc baseband eqialent filter of thefrontend filter

BBH

L Maimm qantiation leel Noncentral chisqare distribtion noncentralit

parameter Parameter of the Poisson la

L Free space loss

M Nmber of noncoherent integrations

N Nmber of samples sed in FFT calclations

0N Thermal noise densit

IN ,0 Interference eqialent noise densit atcorrelator otpt

Pint Receied interference mean poer after frontend filtering

R Correlation of the local code ith the filteredincoming code

CS Baseband normalied PRN code PSD

IS Baseband normalied interference PSD

Threshold

0 NemanPearson test criterion 0

1 NemanPearson test criterion 1

S Sampling period Weighting indos coefficients


22/197

Introdction

22

INODCION

MOIAION

Todas airliners se nmeros and epensie grond naigation sstems, sch as Ver highfreqenc Omnidirectional Range or Distance Measring Eqipments (VOR/DME) for enrote, or Instrment Landing Sstem (ILS) for landing operations. The International CiilAiation Organiation (ICAO) crrent strateg for introdcing adanced commnication,naigation, and sreillance sstems is to replace the grond naigation sstems b satellitebased ones [Eropean Commnit, 2003]. The Global Naigation Satellite Sstem (GNSS) isdefined b ICAO as A orldide position and time determination sstem that incldes oneor more satellite constellations, aircraft receiers and sstem integrit monitoring,agmented as necessar to spport the reqired naigation performance for the intendedoperation [ICAO]. Using GNSS as a niqe naigation mean, hateer the phase of flight,

old decrease airliners costs bt also simplif the naigation procedres.

The alread eisting Global Positioning sstem (GPS) cannot garantee the safet leelreqired b the ICAO for the most critical phases of flight, sch as landing operations. Theperformances of a GNSS can be obsered along for criteria: accrac, aailabilit,continit, and integrit. As part of the deelopment of satellitebased positioning sstems,(in particlar GPS moderniation or the deelopment of Galileo), a certain nmber of nefeatres ere meant to allo GNSS receiers to match ICAO minimm performancesreqired.

The Galileo E5a/E5b and GPS L5 signals ere mainl designed for ciil aiation prposes.The are located ithin the 960 1215 MH freqenc band, hich as alread allocated toAeronatical Radio Naigation Serices (ARNS). This proides a strong protection againstotofband interferences and ensres that interference beteen sstems present ithinthis band respect the recommendations from the International Telecommnication Union(ITU). Becase other aeronatical sstems (sch as Distance Measring Eqipment (DME),TACtical Air Naigation (TACAN), Joint Tactical Information Distribtion Sstem (JTIDS) andMltifnctional Information Distribtion Sstem (MIDS)) ere alread present in this band, itas necessar to ensre that the proposed GNSS signals cold coeist ith them ithotsignificantl degrade their performance. The particlarit of these interferences is their


23/197

Introdction

23

plsed natre. [Tran, 2001] shoed that the GPS L5 signal cold not fll coeist ith thecrrent sstems as it is, nless the DME/TACAN and JTIDS/MIDS freqencies are reassigned.

In particlar, [Bastide, 2003] and [Tran, 2001] stated that GNSS receiers cold not compl

ith ciil aiation performance reqirements in the orst interference enironments, singE5/L5 signals. It then appeared as a necessit to design antenna or receier basedInterference Mitigation Techniqes (IMTs). Using crrent antennas and forecasts on ftrereceiers architectres, [Bastide, 2004] shoed that GNSS receiers cold meet ciil aiationreqirements sing an IMT called the Temporal Blanker, bt ith er slim margins and nonnegligible constraints on the receier design. This techniqe blanks the parts of the signalthat are aboe a certain threshold de to the presence of a plsed interference.

[Bastide, 2004] is a PhD thesis sponsored b the Direction de la Techniqe et delInnoation (DTI), a French ciil aiation instittion. DTI is deepl inoled in

standardisation actiities so that the interference impact assessment on E5/L5 as a topicalsbject needing a resoltion. Notabl, the DTI is linked member of the EURopeanOrganisation for Ciil Aiation Eqipment Working Grop 62 (EUROCAE WG62), hich is incharge of Galileo standardiation for ciil aiation in Erope. In particlar, the resltsproposed in [Bastide, 2004] largel inspired EUROCAE WG62 orks and ere also sed bthe RTCA SC159, hich is the US eqialent for EUROCAE WG62. The qoted PhD thesisconclded that link bdgets for GPS L5 and Galileo E5a/E5b signals ere positie oer bothErope and the U.S.A. folloing a conseratie orstcase approach and assming the se ofthe temporal blanker. Hoeer, the E5a/L5 margins are slight and the degradation resltssensitie to the blanking threshold. Finall, 1500 to 2500 hardare correlators are reqired

to meet the ciil aiation acqisition time reqirements.

[Monnerat, 2000] also stdied an alternatie IMT: the Freqenc Domain Adaptie Filtering(FDAF). This techniqe detects and remoes the interferences in the freqenc domain,hich is epected to be more efficient than the temporal blanking techniqe, the conterpart being a significantl higher reqired processing poer. This techniqe has also beenstdied in [DiPietro, 1989], here the algorithm is called a Freqenc Domain InterferenceSppressor (FDIS), and not FDAF. Indeed, the name FDAF is sed in [Kinjo, 1997], here theproposed techniqe is completel different from the algorithm proposed in [Monnerat,2000]. In this PhD thesis, the denomination FDIS has been preferred to FDAF. [Monnerat,

2000] also proided FDIS and temporal blanker performance assessments, throghsimlation reslts and theoretical deriations. These reslts shoed that FDIS as a moreefficient IMT than the temporal blanker.

The temporal blanker as thoroghl inestigated b ciil aiation certification bodies(RTCA, EUROCAE, ICAO) as a fairl simple method to mitigate DME/TACAN and JTIDS/MIDSplsed interference hile still alloing ciil aiation operations anhere antime. The FDISas onl briefl inestigated in a second time, mostl de to its significant compleit ithrespect to the temporal blanker. Hoeer, ith the increase in the processing poer, theFDIS is more sedcing becase if applied correctl, it cold lead to more comfortable

margins for ciil aiation sers, ths relaing constraints on the other receier parameters.


24/197

Introdction

24

The Eropean Commission fnded throgh the 6th frameork programme the ANASTASIAproject ([ANASTASIA, 2008]) hich core is to proide onboard Commnication, Naigationand Sreillance (CNS) soltions to cope ith the foreseen dobling of air traffic b 2020.

More particlarl, the objectie of the Naigation Sbproject SP3 Naigation and spacebased technologies, as to define the ne satellitebased technologies able to flfil theciil aiation reqirements in the Naigation domain, and to alidate their performance([ANASTASIA]). One of the otcomes of this project is a Galileo E1/E5 dalfreqencreceier mockp incorporating innoatie signal processing techniqes, mostl to fightmltipath and inband interferences. An important part of the receier as theimplementation of both the temporal blanker and the FDIS techniqes to mitigate plsedinterference. The present PhD thesis as condcted in the frameork of the ANASTASIAproject, hich fnded the PhD.

The objecties of the PhD ere ths: To deepen the FDIS analsis gien in [Monnerat, 2000], to propose a theoretical

deriation of the techniqes epected performances, and to std the benefitsbroght b this techniqe ith respect to the temporal blanker in the contet of ciilaiation considerations,

To take into accont hardare soltions that hae been proposed to implement theFDIS and temporal blanker techniqes, and the effect of implementation constraints,hich hae been neglected in [Bastide, 2004],

To design, based on the preios analsis, the temporal blanker and the FDIS that illbe implemented in the ANASTASIA mockp receier, taking into accont the

receier constraints, To participate to the testing of the ANASTASIA receier and assess the reslting

performance,

To conclde abot the orthiness of the FDIS implementation on airborne GNSSreceiers throgh theoretical deriations, simlations, and tests.

HEI CONIBION

Based on the preios section, the major contribtions of the present Ph.D. thesis are thefolloing:

Consolidation of the reslts on plsed interferences impact on GNSS receiersperformances,

Consolidation of the reslts on GNSS receiers sing the Temporal Blanker as IMT,

Analsis of FDIS algorithm against plsed interference,


25/197

Introdction

25

Theoretical deriation of postcorrelation C/N0 degradation in plsed interferenceenironment sing FDIS as a mitigation techniqe,

Analsis of the impact of plsed interference on Atomatic Gain Control (AGC) deice

accrac,

Implementation and test of a distribtion based AGC shoing good plsedinterference robstness on a Galileo mockp receier,

Implementation and test of Temporal Blanker and FDIS algorithms as IMTs on aGalileo mockp receier,

Analsis of the impact of the FDIS algorithm on PRN codes isolation and trackingloops accrac,

Matching of the mockp receier ith ciil aiation reqirements.

HEI OGANIAION

The core of the std is the analsis of the performances of GNSS receiers sing FDIS asIMT. The dissertation architectre is the folloing.

First, the naigation signals, Galileo E5a/E5b and GPS L5, are presented in Chapter I. Theinterferences that constitte a threat for GNSS naigation are also presented in Chapter I,along ith their impact on GNSS receiers operations. Finall, it presents the performancesreqired b the ciil aiation athorities to certif a GNSS receier as a primar naigationmean, releant in the scope of the std.

Chapter II describes the stdied IMTs (Temporal Blanker, FDIS), their theoreticalcharacteristics and the theoretical deriations of the postcorrelation C/N0 degradationsffered b a receier sing these techniqes in presence of plsed interference. It alsoproposes an analsis of the implementation constraints ssceptible to degrade the

algorithms performances, and std their impact on them.

The analsis approach, the Figres Of Merit chosen to analse the performance of bothtechniqes, the chosen interference and signal scenarios, and the simlation tools(theoretical deriation tool, PULSAR, Galileo receier mockp, GIRASOLE receier mockp,acqisition time simlator) sed to obtain the reslts are finel detailed in Chapter III.

Chapter IV gathers all the reslts obtained concerning the IMTs performance assessments. Aconfrontation of Temporal Blanker and FDIS performances is gien sing the preiosldescribed FOMs and simlation tools. The reslts are also compared ith ciil aiation

reqirements.


26/197

Introdction

26

The conclsion smmaries the performances analsis, and proposes recommendations foronboard GNSS receiers design. Finall, alternatie interference mitigation techniqes areintrodced, and perspecties for ftre ork are proposed.


27/197

Chapter I: Galileo E5a/E5b and GPS L5 signals and their Interference Enironment

27

CHAPE I: GALILEO E5A/E5B AND GP L5 IGNAL

AND HEI INEFEENCE ENIONMEN

This chapter describes the GPS L5 and Galileo E5 signals, along ith the interference thatthreatens their correct se b GNSS sers. Those ne GNSS signals ere chosen to bebroadcast in an Aeronatical Radio Naigation Serices freqenc band (ARNS), as it isprotected b the International Telecommnication Union (ITU). Hoeer, this ARNS bandas alread occpied b radio naigation sstems sch as DME, TACAN, JTIDS, and MIDS,hich impact on GNSS receiers cannot be ignored.The first section of this chapter is dedicated to the description of the GPS L5 signal, and thesecond one to the Galileo E5a/E5b signals. The third section describes the other signalsalread present in the E5 band, hile the forth section briefl introdces tpical GNSSreceiers architectre. Finall, the fifth section describes the impact of these interferences

on GNSS receiers operations.

I.1.GP L5 IGNAL CE

The GPS L5 signal as originall introdced in order to spport the ciil aiation commnit.It brings seeral important featres ith respect to the GPS L1 C/A signal alone: it allosdal freqenc measrements, and ths precise ionospheric dela corrections; it is also aprecise signal de to its ide bandidth (10 times the bandidth of the GPS C/A); finall, italso proides a backp to the L1 signal, improing the serice aailabilit.

The GPS L5 signal can be modelled at the receier antenna port as follos:

( ) ( ) ( ) ( ) ( )

( ) ( ) ( )NHQC

NHIC

L

LL

+

=

520

5105

2sin

2cos I.1

Where: The signal is composed of to components in phase qadratre: the data component

(inphase), and the pilot component (in qadratre),

is the timearing grop dela,

is the GPS signal carrier phase shift,


28/197


28

Cis the total signal transmitted poer,

is the PolarNonRetrntoZeroLeel (P/NRZ/L) materialisation of the naigationmessage. Onl the Inphase component carries this message,

I and Qare respectiel the P/NRZ/L materialisation of the Inphase and

Qadraphase Psedo Random Noise (PRN) codes, 10NH and 20NH are respectiel the P/NRZ/L materialisation of the Inphase and

Qadraphase NemanHoffman codes (also called secondar codes orsnchronisation seqences),

is the time in seconds,

5L =1176.45 MH is the L5 carrier freqenc.

The main characteristics of the GPS L5 signal shon in Table 1 introdce the majorimproements broght in this ne signal (compared to the L1 C/A signal): the increase of the

chip rate, the presence of to different components (Pilot + Data), and the introdction ofsecondar codes and of Forard Error Correction (FEC) codes. The latter ones are meant toredce the Bit Error Rate (BER).

1: L5

ChannelModlation

tpe

ChipRate

[Mcps]

CodePeriod

SmbolRate[sps]

FECperiod[sps]

SecondarCode Rate

SecondarCode Period

L5 data 50 100 1 kH 10 msL5 pilot

QPSK 10.23 1 msN/A N/A 20 ms

A more detailed description of the L5 signal can be fond in the GPS L5 InterfaceSpecification (IS) [ARINC, 2005].

I.1.1.GP L5 PN C

The major interest of the GPS L5 signal is its se of high chipping rate: it indces a iderspectrm compared to L1 C/A hich reslts in a better robstness against narrobandinterference, a better mitigation of longdela mltipath, and loer tracking noise jitter. The

spreading seqence sed b GPS L5 are also significantl longer than the GPS C/A seqences,hich proide a better isolation ith crosscorrelation peaks ([Van Dierendonck, 1999]). 74PRN codes (37 XI sed on the Data channel, and 37 XQ sed on the Pilot one) ere chosenfrom the GPS L5 codes generating process, proided in [ARINC, 2005]. For each satellite, thecode pair (XI and XQ) as chosen to be as orthogonal as possible. Figre 1 shos the poerspectral densit of a tpical GPS L5 PRN code. The main lobe is 20 MH ide, hich ellillstrates the better robstness of sch codes to interference.


29/197


29

30 20 10 0 10 20 30130

120

110

100

90

80

70

60

F (MH)

PD(B/H)

F 1 : N GP L5 PN 1 I PD.

A smmar of the crosscorrelation isolation ith respect to the atocorrelation fnctionmain peak, etracted from [Macabia, 2002], is gien in Table 2, here Doppler freqencoffsets are neglected. GPS L1 C/A PRN codes isolations are also gien, to sho theimproement broght b the ne GPS L5 codes.

2: 1 I Q

XI/XI XQ/XQXI/XQ (same

satellite)L1 C/A

Maimmcrosscorrelation

sidelobe (dB)26.4 26.5 62.1 21.0

I.1.2.N M

The naigation message carries essential information for ser position determination. Aspresented in Table 1, onl the data component of the L5 signal broadcasts the reqired data,the pilot component being meant onl for ranging.As it ill be clarified later, the present inestigation does not reqire the generation of thenaigation message, ths it ill not be described frther herein. Hoeer, the completedescription of the transmitted naigation message can be fond in [ARINC, 2005].Both components (data and pilot) are also encoded respectiel ith a 10bit and a 20bitNemanHoffman code. One code chip of the secondar code has the same dration as thehole primar code period, 1 ms. These codes are sed notabl for bit snchronisation


30/197


30

improement, and also to fight against narroband interference. The are also etensieldescribed in [ARINC, 2005].

I.2.GALILEO E5A/E5B IGNAL CEThe Galileo sstem ill propose nmeros naigation signals broadcasted on 3 bands: E1, E5and E6. The E5 signal, hich is emitted ithin 1164 and 1215 MH is composed of the E5aand E5b signals, modlated coherentl sing an AltBOC(15,10) modlation ([Reberol,2005]). The ALTBOC(15,10) modlation is described in [GSA, 2008]. It can be noted that itsPSD is constitted of to main sidelobes, one centred arond 1176.45 MH representingthe E5a component, and one centred arond 1207.14 MH representing the E5bcomponent. Note that the E5a signal PSD sperimposes ith the GPS L5 signal [Reberol,2007].

The E5a and E5b signals are both composed of a data and a pilot component. The naigationdata carried b the E5a signal corresponds to the Galileo Open Serice (OS), hile the E5bsignal carries the Galileo SafetofLife serice data.The E5 signal can be processed in 2 different as:

As a single ide bandidth signal,

Or considering each of the to signals separatel. In this case, both E5a and E5bsignals can be considered as QPSKmodlated ith a minimal degradation. Hoeer,this configration imposes a high leel of filtering beteen the 2 signals, as it ill beseen later on.

The latter strateg has been chosen b the EURopean Organiation for Ciil Aiation

Eqipment (EUROCAE) in [EUROCAE, 2007], in order to aoid interferences in one band todegrade the Galileo signal in the other band. For eample, interferences receied in eitherband might case the loss of both signals in case the are simltaneosl tracked.Otherise, if the signals are separatel tracked, onl the one concerned b the interferencemight be lost. Moreoer, it is important to keep in mind that the E5b signal is the onl one tocarr the SafetofLife message, and ths the integrit message necessar to ciil aiation.Both signals are modlated b PRN ranging codes, secondar codes. The majorcharacteristics of the E5 signal are smmaried in Table 3. Becase E5b transmits theintegrit message, the data rate is higher than on E5a.

3 : E5

Channel Modlation tpe Chip Rate [Mcps]Smbol Rate

[sps]E5a data 50

E5a pilot N/AE5b data 250

E5b pilot

AltBOC(15,10) 10.23

N/A

The E5a normalied poer spectrm (XQ PRN 1) is represented in Figre 2, hile the E5b

one (XQ PRN 1) is shon in Figre 3.


31/197


31

20 10 0 10 20130

120

110

100

90

80

70

60

F (MH)

P

D(B/H

)

F 2: N G E5 PN 1 PD.

20 10 0 10 20130

120

110

100

90

80

70

60

F (MH)

PD(B/H

)

F 3: N G E5 PN 1 PD.

I.2.1.GALILEO E5 C C

The Galileo E5 primar codes hae the same properties than the L5 ones: chipping rate of10.23 Mcps, period of 1 ms. The major difference beteen E5 and L5 codes lies in the length

of the secondar codes (or snchronisation seqences), as shon in Table 4.


32/197


32

4: G E5 PN

Code length (chips)Channel Code Length (ms)

Primar SecondarE5aI 20 10230 20

E5aQ 100 10230 100

E5bI 4 10230 4E5bQ 100 10230 100

Galileo E5 PRN code isolation properties are epected to be similar to those of GPS L5.

I.2.2.N M

To different tpes of naigation messages are transmitted b the E5 signal: the F/NAVmessage on the E5a data component, and the I/NAV message on the E5b data component.The major difference beteen these to messages is their emission rate: 50 smbols persecond for F/NAV, 250 smbols per second for I/NAV. Moreoer, E5a is part of the Open andthe Commercial Serices (OS and CS), hile E5b is part of the Open, the Commercial and theSafet of Life (SoL) serices ([ESA, 2008]). Indeed, the E5b signal (I/NAV message) broadcastsnaigation and integrit data applicable to E1/E5b dalfreqenc measrements (theintegrit information is not alid for E1/E5a measrements). This is the reason h, from theciil aiation commnit point of ie, at this time the E5b signal is of greater interest than

the E5a one. This sitation ma change as the possibilit of broadcasting integrit data onthe E5a signal is still nder discssion.As for GPS L5, Galileo E5 secondar codes are sed to improe bit snchroniation, solepsedorange measrements ambigities, and also improe the signals robstness tonarroband interference. Their chip rate also eqals 1 kilo chip per second (kcps), bt theirperiod is longer, as shon in Table 4.A fll description of Galileo OS signals can be fond in [GSA, 2008].

I.3.INEFEENCE HEA IN HE E5 BAND

The L5/E5 band occpation has alread been etensiel stdied, in [RTCA, 2004], or [Anon,1997]. DME/TACAN and JTIDS/MIDS signals hae been identified as the onl sstems ecepted nintentional emissions emitting in the band of interest. Otofband andsprios emissions sch as militar radars impact on GNSS receiers hae been assessednegligible for the GPS L5 and Galileo E5a signals [RTCA, 2004]. It is slightl different for theE5b signal, since the militar radar band finishes er close to the E5b pper limit (1217MH). Becase militar radars emit er poerfl signals, the are a potential threat to GNSSreception de to a potential lack of filtering. This problem as thoroghl inestigated bthe EUROCAE ([EUROCAE, 2007]) and the soltion as to specif a sharp E5b filter that has actoff freqenc ell ithin the E5b band, as shon in paragraph I.6.2.


33/197


33

I.3.1.DME/ACAN

DME, and its militar eqialent, TACAN, are to sstems sed b aircrafts to kno theirdistance to a grond station, hich position is knon. The sstems operate as follos

([Borden, 1951]): the aircraft DME eqipment (called interrogator) sends plses to grondstations. Once the interrogation is detected, the station transponder replies to theinterrogator. The distance is then determined b measring the time elapsed beteen eachplse transmitted b the interrogator and the reception of its corresponding repl plsefrom the transponder. This time corresponds to tice the distance beteen the aircraft andthe station, pls fied processing time inside the grond station.According to [RTCA, 2004], onl the signals emitted in the band of interest of the stddistrb GNSS receiers operations. Indeed, the band of interest is either the E5a/L5 one andeqals [1164 MH; 1191 MH], or the E5b one hich eqals [1191 MH; 1215 MH]. Theaircrafts DME interrogators emitting their signals beteen 1025 and 1151 MH, the are

ignored herein. The std focses on DME grond stations, as the emit their signalsbeteen 962 and 1213 MH, hich incldes the aboe defined band of interest. The emittedsignal is composed of a pair of Gassian plses modlated b a cosine, hich can bemodelled as:

( )( ) ( )

( )II

N

P

+

+=

=

2cos1

22

22

I.2

Where:

P is the interference beacon transmitting peak poer (dBW),

I is the carrier freqenc of the DME/TACAN signal (H),

211105.4 = , 12 = is the inter plse time separation,

is the emission time of the kthplse pair and

I is the DME/TACAN signal initial carrier phase shift.

Figre 4 represents a normalied DME/TACAN plse pair, modlated at 14 MH. This carrieras chosen as it is the Intermediate Freqenc (IF) sed in the simlators and the mockpreceier.


34/197


34

0 5 10 15 20 25 30

1

0.5

0

0.5

1

()

A()

F 4: DME .

Theoreticall, the Forier transform (FT) of a DME/TACAN signal as epressed in I.2 has thefolloing epression:

( ) ( )

( )

+

+

+

+

= II

II

AACANDMEF

22

2

22

12

2)/(

22

22

I.3

Moreoer, signals FT are calclated on a bonded spport, hich is materialied b arectanglar indo of dration . Then, the FT of a indoed DME/TACAN signal has thefolloing epression:

[ ]

( ) ( )

( ) ( )

ADMEF I

I

II

sin12

2)( 2

22

22

;0

22

22

+

+

=

+

+

I.4

Figre 5 shos that the energ of a DME/TACAN plse pair is spectrall constrained. Indeed,more than 99.99% of its energ is contained in a bandidth of 1 MH. It can be noticed thatthis band is mch narroer than the stdied GNSS signals ones.


35/197


35

13 13.5 14 14.5 1570

60

50

40

30

20

10

F (MH)

A(B.

2)

F 5: N DME/ACAN E D.

The ansers of each and eer beacon located in the radioelectric range are receied at theairplane leel. The nmber of beacons and their characteristics are therefore a fnction ofthe position (latitde, longitde, altitde) of the aircraft. The DME/TACAN stations located inErope are represented on Figre 6 b red (DME) and ble (TACAN) dots.

10 0 10 20 30 40

35

40

45

50

55

60

65

70

L ()

L()

DME

ACAN

F 6 : E DME/ACAN .

Figre 7 proides the same information in the United States of America. The std focseson these areas as most DME/TACAN beacons are concentrated in Erope and USA.


36/197


36

130 120 110 100 90 80 70 6025

30

35

40

45

50

55

L ()

L()

DME

ACAN

F 7: DME/ACAN .

Each DME/TACAN grond beacon can respectiel emit p to 2700/3600 plse pairs persecond. This nmber depends pon the nmber of aircrafts sending interrogations to thestation, so pon the air traffic.

I.3.2.JID/MID

JTIDS is an L band TDMA netork radio sstem sed b the United States armed forces andtheir allies to spport data commnications needs, principall in the air and missile defencecommnit. It proides highjamresistance, highspeed, crptosecre comptertocompter connectiit in spport of eer tpe of militar platform from Air Force fighters toNa sbmarines. JTIDS is one of the famil of radio eqipment implementing hat is calledLink 16.Mltifnctional Information Distribtion Sstem (MIDS) is the NATO name for thecommnication component of Link16. An older MIDS is the JTIDS.JTIDS/MIDS signals are composed of plses that last 13 s (6.4 s actie and 6.6 s passie).Those plses are modlated b a chip seqence, hich is 32 chips long and each chip lasts200 ns. A tpical JTIDS/MIDS plse is represented in Figre 8.


37/197


37

0 2 4 6 8 10 120

0.2

0.4

0.6

0.8

1

()

A()

F 8 : N JID/MID .

More details abot JTIDS/MIDS signals can be fond in [Nisner, 2003].

I.4.GN ECEIE ACHIECE

In this paragraph, GNSS receiers architectre is briefl described. This description isreqired to introdce ho the interferences described in paragraph I.5 impact thesereceiers. Figre 9 shos a simplified scheme of generic GNSS receiers architectre, hichis detailed in [Van Dierendonck, 1996].

F 9 : GN A.

DOWNCONVERSION (IF)& FILTERING

SAMPLING& QUANTIZATION

SIGNALPROCESSING

PREAMPLIFIER

NAVIGATIONPROCESSING

Antenna


38/197


39/197


39

S is the baseband normalied PRN code PSD,

( )HBB is the freqenc response baseband eqialent filter of the front end filter.

I.4.3. Q

This stage transforms the analog signal into a digital one. In this std, attention ill be paidon the qantiation process hich is highl impacted b interference. A tpical sampling andqantiing architectre is gien in Figre 10.

F 10 : C Q: AGCADC .

Qantiation is achieed b an Analog to Digital Conerter (ADC), hich trns oltages intobit trains. Qantiation las can be centred (0 is a possible otpt ale) or noncentred (0 isnot a possible otpt ale).Figre 11 presents the eample of a niform centred 8bit qantiation la. As it ill be seenlater on, this la is sed in the ANASTASIA mock p receier. The inpt oltages inclded inthe range [L; +L] olts are trned into bit trains hich decimal representation lies in the

range [127; +127]. If the inpt ales do not lie in the [L; +L] range, the ADC is satratedand the otpt ale eqals 127 or +127.

Atomatic Gain

Control AGC

Qantier

ADC

Sampled SignalSignal Processing


40/197


40

F 11: E 8 .

The qantiation process creates a degradation that is sall referred to as qantiationnoise. This noise is tpicall modelled as an additie hite noise. It reslts in a decrease ofthe SignaltoNoise Ratio (SNR), hich ill degrade frther signal processing performance.The SNR degradation de to qantiation is a fnction of the ADC inpt signal and the

conerter characteristics (nmber of bits and qantiation la). In this std, the ADC inptsignal is composed of:

Thermal noise (assmed Gassian and hite), hich poer depends pon thereceier temperatre, the cable losses and amplifications operated beteen theantenna and the ADC,

The GNSS signal of interest. Referring to the link bdget in [Bastide, 2004], hichconsiders the same operational conditions, it can be seen that the leel of this signalis significantl belo the noise floor,

Interference, if receied. As eposed in section I.3, mainl plsed interference(DME/TACAN and JTIDS/MIDS) ill be considered.

In the nominal case, interferences are not considered so that the signal entering the ADC isdominated b thermal noise. The distribtion and poer of the obsered signal aretherefore almost completel set b the thermal noise. The SNR degradation sfferedbecase of qantiation is then a fnction of the nmber of bits sed in the ADC, theqantiation la, the maimm qantiation leel (L), and the noise standard deiation.Figre 12 shos the SNR degradation sffered at the correlator otpt de to qantiation(assmed niform and centred), for different nmber of bits, as a fnction of k, defined asthe ratio beteen the maimm qantiation leel (L) and the inpt signal standarddeiation. Moreoer, the reslts ere obtained assming infinite frontend eqialent

X

XQ

1

2

3

3

2

1

+ 127

+LL

12

28

+

L

12

28

L


41/197


41

bandidth and infinite sampling freqenc [Van Dierendonck, 1996]. Otherise, thedegradation is larger, as shon in [Chang, 1982].

0 2 4 6 8 10

1

2

3

4

5

M M =L/

N(B)

3

4

5

F 12 : N .

According to Figre 12, for a gien ADC implementation (nmber of bits, qantiation la,maimm qantiation leel), the signal standard deiation can be set so as to minimie

qantiation losses.

As mentioned earlier, the thermal noise entering the ADC aries ith time, mainl becaseof temperatre ariations. Therefore, the standard deiation at ADC inpt cannot bepredicted, so that it is not possible to preset amplifiers gains that ill minimie theqantiation loss dring the receier design. The total gain applied to the signal needs to beadapted dring the operation of the receier, reqiring the implementation of a specificamplifier: the AGC. This modle amplifies the signal before the ADC (see Figre 10), andadapts its gain sing a feedback loop. The classical implementation estimates the signalstandard deiation at ADC otpt, calclates the difference beteen the standard deiation

minimiing qantiation loss and the crrent estimate, and sends a command to the AGC.The performances of sch AGCs depend pon the qalit of the standard deiationestimator, the real distribtion of the receied signal, and the design of the loop (filters, timeconstant).More particlarl, the reception of interferences modifies the distribtion and the poer ofthe signal. AGCs sensitiit to interferences is described in paragraph I.5.1.

I.4.4. P

The signal processing stage is composed of to main fnctions: acqisition and tracking. The

acqisition is meant to roghl estimate the signal propagation dela and the Doppler


42/197


42

freqenc. The tracking process consists in a refined estimation of the code and carrierphases. Both processes se correlators, hich are described belo. In addition, othermodles can be added. This is the case in the contet of this thesis, since a specific emphasisis gien to the mitigation of plsed interference. These modles ill be introdced later in

the dissertation.

I.4.4.1.C

The correlators process the signal as follos: The signal is mltiplied b a locall generated carrier,

The obtained signal is mltiplied b a locall generated PRN code, The signal is then integrated oer one or seeral code length.

The correlator scheme, proided in Figre 13, represents the 3 processes described aboe.

F 13: A .

Becase GPS L5, and Galileo E5a and E5b hae a data and a pilot component sing differentspreading seqences, it is necessar to represent separatel the correlator otpt of eachcomponent. The Inphase and Qadraphase correlators otpt can be modelled as follos[Bastide, 2004]:

( ) ( ) ( )

( )( ) ( )( )

( ) ( )

( )

( )( ) ( )( )

( ) ( )

( )( ) ( )( )

( ) ( )

( )( ) ( )( ) QP

P

I

P

P

Q

P

P

I

P

P

R

CQ

R

CI

R

D

C

Q

R

D

CI

,,

,,

,,

,,

cos

sin

4

sin

sin

4

sin

sin

4

cos

sin

4

+=

+=

+=

+=

I.6

Where:

( )D is the naigation data bit sign oer the kthcoherent integration,

DOPDOP , = is the freqenc offset beteen the receied and local carriers,

( )

( ) +

+

P

P

1

2cos 0


43/197


43

is the phase offset beteen the receied and local carriers,

= is the phase offset beteen the receied and local codes,

R , and R , are respectiel the data and pilot crosscorrelations beteen the

filtered receied codes and the nfiltered local ones. A representation of this crosscorrelation fnctions is gien in Figre 14,

I , , Q , , I , , Q , , are respectiel the inphase/qadraphase correlator

otpt noise contribtion of the data and pilot components. These noise samples are

centred and hae a poer of402 P

N = ,

is the predetection bandidth and eqals /1 , here is the coherent

integration time,

( ) ( )

+

= HS BBC2

is the thermal noise poer redction at correlator otpt de

to frontend filtering.

5 0 50.2

0

0.2

0.4

0.6

0.8

1

1.2

P O ()

C

F

N

F

F 14: C E5/L5 .

There are seeral important reslts linked to the epression of the correlator otpt. First, itcan be noticed that the sefl component of the Inphase and Qadraphase correlatorotpt are er similar:

There is a dependence ith the code dela error folloing the correlation fnctionbeteen the incoming and the locall generated spreading code. Note that becaseGalileo E5a and E5b are processed as BPSK(10) signals ith minimal degradation, theassociated correlation fnction can be assmed similar to that of GPS L5.


44/197


44

There is a dependence ith the freqenc error folloing a sinc fnction. The firstero of this sinc fnction appears at /1 hich means that a correlation done oer a

long integration time ill reqire a small freqenc error to fnction properl.

On the other hand, the impact of the phase error is different beteen the I and Qcomponents: the first one has a cosine dependence, hile the second one has a sinedependence.

Finall, note also that a longer coherent integration reslts in a higher SNR. Ths, it isinteresting to se a long integration time if possible. Hoeer, seeral reasons limit thechoice of the integration time, sch as:

The presence of a naigation data bit transition,

The drift of the receier clock,

A strong dnamic stress.

I.4.4.2.A P P

The objecties of the acqisition process are to detect the signal of interest, and ifsccessfl, to roghl assess its characteristics:

The dela beteen the incoming spreading code and the local spreading code, ,

The Doppler freqenc, DOP .

A basic acqisition strateg consists in searching seqentiall all the possible ales of

and DOP ntil the acqisition criterion is maimied and reaches a predefined threshold. In

order to do so, a range of ales and a search step are defined for the to parameters. Thisrange sall depends pon the aailable information regarding the signal of interest(aailabilit of the time, ephemeris, etc). This defines the sie of the todimensional searchspace composed of code/freqenc bins. For each bin, or code/freqenc combination

DOP ,, , the folloing criterion is calclated:

( ) ( ) ( ) ( ) ( )=

+++=M

DOP QIQI1

2222, I.7

Where M is the total nmber of noncoherent integrations.

The aboe criterion as sed b [Bastide, 2004] to take adantage of the pilot/datastrctre of the considered signals, and ths gather the energ of the hole signal. It can benoticed that this acqisition detector remoes the dependence ith the phase error.

The criterion is then compared to a threshold, in order to determine if the searched signal ispresent or not. The test can lead to for different cases:

The signal is present (nderstand the good code/freqenc bin is fond) bt thecriterion does not eceed the threshold. The detection is missed.


45/197


45

The signal is present and the criterion does eceed the threshold. The detectionscceeds.

The signal is not present and the criterion is eceeded. A false alarm occrs.

The signal is not present and the criterion is not eceeded.

The missed detection and false alarm eents are likel to happen becase the testedcriterion is affected b noise, mltipath or interference. The isse raised b these eents isthat the slo don the acqisition process, or, in the orst case, cold lead to erroneosmeasrements.

The considered acqisition process is seqential, hich means that each bin is testedsccessiel. If no detection occrred hen the entire span has been tested, the processstarts oer. Ths, a missed detection reslts in an agmentation of the acqisition timeeqalling at least the time reqired to search the entire search space, hich is represented

in Figre 15.

F 15 : A .

The time reqired to test one bin is the dell time. The time reqired to coer the entiresearch space is therefore:

N

NM

= I.8

Where: M is the dell time,

N is the nmber of bins constitting the ncertaint region,

N is the nmber of correlators allocated to the search.

To hpotheses are sed to assess the acqisition performance:

H0: The signal is not present in the crrentl searched bin. In this case, the detectoreqals 0 ,

Code Phase bins (chips)

Dop

plerBins(H)

Search Di

Raimondi Final

Documents

Transcript of Raimondi Final