On Potential CBOC/TMBOC Common Receiver Architectures

Transcription

1 On Potential CBOC/TMBOC Common Receiver Architecture O. Julien, C. Macabiau Ecole Nationale de l Aviation Civile (ENAC), France J.-A. Avila Rodriguez, S. Wallner, M. Paonni, G. W. Hein Univerity FAF Munich, Germany J.-L. Iler, L. Rie Centre Nationale d Etude Spatiale (CNES), France BIOGRAPHIES Olivier Julien i an aitant profeor at the ignal proceing laboratory of ENAC (Ecole Nationale de l Aviation Civile), Touloue, France. Hi reearch interet are GNSS receiver deign, GNSS multipath and interference mitigation and GNSS interoperability. He received hi B.Eng in in digital communication from ENAC and hi PhD in 5 from the Department of Geomatic Engineering of the Univerity of Calgary, Canada. Joé-Ángel Ávila-Rodríguez i reearch aociate at the Intitute of Geodey and Navigation at the Univerity of the Federal Armed Force Munich. He i reponible for reearch activitie on GNSS ignal, including BOC, BCS, and MBCS modulation. Avila-Rodriguez i one of the CBOC inventor and ha actively participated in the developing innovation of the CBOC multiplexing. He i involved in the GALILEO program, in which he upport the European Space Agency, the European Commiion, and the European GNSS Supervior Authority, through the GALILEO Signal Tak Force. Chritophe Macabiau graduated a an electronic engineer in 99 from the ENAC (Ecole Nationale de l Aviation Civile) in Touloue, France. Since 994, he ha been working on the application of atellite navigation technique to civil aviation. He received hi Ph.D. in 997 and ha been in charge of the ignal proceing lab of the ENAC ince. Stefan Wallner tudied at the Technical Univerity of Munich and graduated in 3 with a Diploma in Techno- Mathematic. He i now reearch aociate at the Intitute of Geodey and Navigation at the Univerity of the Federal Armed Force Germany in Munich. Hi main topic of interet can be denoted a the Spreading Code, the Signal Structure of GALILEO together with Radio Frequency Compatibility of GNSS. Matteo Paonni i reearch aociate at the Intitute of Geodey and Navigation at the Univerity of the Federal Armed Force Munich. He received hi M.S. in Electrical Engineering from the Univerity of Perugia, Italy. Hi main topic of interet are Tracking Algorithm for GNSS Receiver Deign, GNSS Signal Structure and Indoor Poitioning. Guenter W. Hein i Full Profeor and Director of the Intitute of Geodey and Navigation at the Univerity FAF Munich. He i reponible for reearch and teaching in the field of high-preciion GNSS poitioning and navigation, phyical geodey and atellite method. He ha been working in the field of GPS ince 984 and i author of numerou paper on kinematic poitioning and navigation a well a enor integration. In he received the pretigiou Johanne Kepler Award from the US Intitute of Navigation (ION) for utained and ignificant contribution to atellite navigation. Preently he i heavily involved in the Galileo program. Jean-Luc Iler i head of the Tranmiion Technique and ignal proceing department of CNES, whoe main tak are ignal proceing, air interface and equipment in Radionavigation, TT&C, propagation and pectrum urvey. He i involved in the development of everal paceborne receiver in Europe, a well a in tudie on the European RadioNavigation project, like GALILEO and the Peudolite Network. With DRAST, he repreent France in the GALILEO Signal Tak Force of the European Commiion. With Lionel Rie and Laurent Letarquit, he received in 4 the Atronautic Prize of the AAAF ( French aeronautical and pace aociation ) for hi technical work on GALILEO ignal and paceborne GNSS equipment. Lionel Rie i a navigation engineer in the Tranmiion Technique and ignal proceing department, at CNES ince June. He i reponible of reearch activitie on GNSS ignal, including BOC modulation and GPS IIF L5. He i involved in the GALILEO program, in which he upport ESA, EC and GJU, through the GALILEO Signal Tak Force. He graduated in 997 from the Ecole Polytechnique de Bruxelle, at Bruel Free Univerity, Belgium, in 997, and received a M.S. degree from the Ecole Nationale Supérieure de l'aéronautique et de l'epace in Touloue, France, in

2 ABSTRACT Under the 4 Agreement on the Promotion, Proviion, and Ue of Galileo and GPS Satellite-Baed Navigation Sytem and Related Application, the member tate of the European Union and the United State agreed on working together, intenifying thu the cooperation on interoperability and compatibility iue between Galileo and GPS. Among other topic, one important focu wa the E/L frequency band, centred at MHz, where the Galileo E Open Service (OS) ignal and the modernized GPS L civil (LC) ignal are going to be tranmitted along with many other RNSS ignal. Recent effort made by US and European expert identified a common optimized Power Spectral Denity (PSD) frame, known a Multiplexed BOC (MBOC), in which both the Galileo E OS and the GPS LC ignal would fit. Thi normalized MBOC PSD i actually formed by the um of / of the normalized BOC(,) PSD and / of the normalized BOC(6,) PSD. Becaue the MBOC i defined in the frequency domain, the time repreentation cannot be uniquely defined, and at leat two different implementation that would till comply with the MBOC pectrum exit: CBOC and TMBOC. Indeed, the latet development indicate that the main Galileo E OS and GPS LC candidate will exhibit different feature [3],[4]: - The current GPS LC main candidate will have a pure BOC(,) data channel gathering 5% of the total ignal power while the pilot channel will ue a Time- Multiplexed BOC (TMBOC) modulation with 75% of the total civil ignal power.. - The current Galileo E OS main candidate will equally hare it power between it data and pilot channel, with the important difference with repect to TMBOC that in both channel a Compoite BOC (CBOC) modulation with BOC(6,) will be ued. It i well-undertood that the definition of a common PSD for the GPS and Galileo civil ignal on E/L call for an increaed interoperability and compatibility of thee ignal at the uer level. However, to really promote the ue of GPS/Galileo E/L combined receiver, it i of greatet importance to find GPS LC and Galileo E OS tracking architecture that minimize the receiver complexity while maintaining high quality meaurement. Thi i particularly true ince the main candidate for implementation of MBOC for GPS LC and Galileo E OS are already baeline of their repective ytem [3]. The purpoe of thi article i to thoroughly invetigate poible GPS/Galileo receiver architecture that could be adapted to CBOC, TMBOC or both waveform and to ae their performance. The firt tep of the propoed analyi i to ae the interference that both GPS and Galileo ignal will caue on each other. Thi tep i neceary in order to evaluate the degradation that Galileo and GPS ignal will caue on each other, and thu, to ae preciely the quality of the forthcoming tracking loop. In a econd part, everal CBOC/TMBOC tracking architecture meant to minimize the complexity of a combined GPS/Galileo receiver will be preented and their performance in term of reitance to thermal noie and multipath, a well a their complexity, will be compared to optimal architecture dedicated to CBOConly or TMBOC-only receiver. Thi part aim at giving an inight on which key parameter can be modified to find a relevant trade-off between receiver complexity and performance for the general uer. Finally, in the lat ection, different multipath mitigation technique will be teted according to different receiver configuration BOC(,) or MBOC receiver and flexibility in the number of correlator available for each channel pilot or data + pilot tracking. In particular, innovative multipath mitigation technique baed on a multi-correlator receiver are alo invetigated. INTRODUCTION After a long period of dicuion, the Galileo Signal Plan ha finally been frozen with the definitive deciion of both GPS and Galileo to implement the MBOC (Multiplexed BOC) modulation for the Galileo E Open Service (OS) and the GPS L Civil (LC) Signal in E/L ([], [] and [3]). While the baeline of the ret of Galileo ignal in E5 and E6 ha been relatively table along the pat year, the E/L band ha been in continuou evolution. The mot important reaon for thi i the evere degree of congetion that the E/L band preent. MBOC(6,,/) i the reult of the deire to multiplex a wideband ignal the BOC(6,) with a narrow-band ignal the BOC(,) in uch a way that / of the power i allocated in average on the BOC(6,) component [5]. The MBOC normalized Power Spectral Denity (PSD) of the data and pilot component together, pecified without the effect of band-limiting filter and payload imperfection, i given by GMBOC (6,,/) ( f ) = GBOC (,) ( f ) + GBOC (6,) ( f ) () Figure how all the exiting and planned navigation ignal of the four global navigation ytem that are foreeen to play an important role in the future [4]. It can be een that the MBOC ucceed in minimizing it overlap with other navigation ignal in the E/L band. Moreover, it alo fulfill very well the requirement of ma-market uer a it provide a ignal with a narrowband component where mot of the power i allocated. At the ame time, MBOC alo ha a wide band component that i meant to provide future uer with an additional potential to improve performance. A hown in [5], a variety of time waveform can be ued to produce the MBOC(6,,/) PSD. Two fundamentally 53

3 different approache, the Time-Multiplexed BOC (TMBOC) and the Compoite BOC (CBOC) have been elected for the MBOC implementation of GPS and Galileo repectively. Thee are decribed in the next ection It can be noted that both pilot and data component are modulated onto the ame carrier component, with a power plit of 5 percent. The parameter P and Q are choen uch that the power aociated with the BOC(6,) ub-carrier component equal / of the total power of the whole Galileo E OS ignal (data + pilot). Thi yield: P = Q = (3) Figure how a generic view of the Galileo E OS generation cheme. Figure. Spectra of GPS, Galileo, GLONASS and Compa Intended Signal in E/L [4] PRESENTATION OF GALILEO E OS AND GPS LC WAVEFORMS The Galileo E OS Signal The Galileo E Open Service (OS) will ue a CBOC (Compoite Binary Offet Carrier) modulation to implement the MBOC. The CBOC ignal adopted for Galileo i baed on the approach preented in [3], [4], [5] and [6], uing a four-level ub-carrier formed by the weighted um of BOC(,) and BOC(6,) ymbol on both data and pilot. The CBOC modulation i a particular cae of the CBCS multiplexing cheme that wa preented in [6] where the particular BCS equence i in thi cae a BOC(6,). Therefore, all the theory derived in [4] and [6] i alo valid here to decribe the CBOC cae and one only ha to ubtitute in the equation the generic BCS cae by the particular BOC(6,) equence and the power of /. The normalized bae-band Galileo E OS compoite ignal model i given by: E () t = ee e B () t d EB () t [ P x() t + Q y() t ] () t [ P x() t Q y() t ] EC where ee B () t and ee C () t are the data and pilot preading equence repectively, d E B () t i the Galileo E-B navigation meage, and x and y are the ine-boc(,) and ine- BOC(6,) ub-carrier repectively. () Figure. Modulation Scheme of Galileo E Signal If we take a careful look at equation (), it can be recognized that the CBOC modulation ha the data and pilot ub-carrier in anti-phae (with repect to the BOC(6,) component). Thi can be oberved on the equivalent waveform of each channel, hown in Figure 3. A it wa already mentioned in [4], [5], [3], and [6], and a a reult of the light difference in the data and pilot CBOC ub-carrier, light difference are alo oberved in the relative performance of the data and pilot channel, epecially for wide-band receiver. Thee difference favour indeed the performance of the pilot channel where BOC(,) and BOC(6,) are ubtracted. Figure 3. Data (Left) and Pilot (Right) CBOC Sub- Carrier The GPS LC Signal TMBOC i the elected MBOC implementation of GPS [7] and i characterized by uing a binary ub-carrier component that reult from the time-multiplexing of the 53

4 BOC(,) and BOC(6,) ub-carrier according to a determinitic pattern. The GPS LC bae-band ignal can be written a: elc I () t d LC I () t x() t () t = + e () (4) L C LC Q t TMBOC(6,,4 / 33) where x() t if t S TMBOC (6,, p) () t = y() t if t S S i the union of the egment of time when a BOC(,) ub-carrier i ued, while S, the complement of S in the time domain, i the union of the egment of time when a BOC(6,) ub-carrier i ued. Figure 4 how a generic view of a TMBOC modulation time erie. Additionally, all the data channel i excluively modulated by a BOC(,) ub-carrier, while the pilot channel i modulated by a TMBOC ub-carrier, where 9/33 chip of the preading code are modulated by a BOC(,) ub-carrier and the remaining 4/33 chip are modulated by a BOC(6,) ub-carrier. It i important to note that although all the BOC(6,) ub-carrier i on the pilot channel, the average power i hown to fulfil Equation (). Figure 4. Example of TMBOC(6,,4/33) Spreading Time Serie, with all BOC(6,) Spreading Symbol in the 75% Pilot Power Component The exact location of the BOC(6,) preading ymbol obey a rationale a hown in [3] and [8]. In fact, if the BOC(6,) ymbol are properly placed, the preading code auto and cro-correlation can be further improved. A een in equation (4), the elected GPS implementation of TMBOC place 75% of the total power on the pilot channel while the other 5% i reerved for data. CBOC and TMBOC, Same Spectrum, Different Signal During the optimization of the E/L Galileo and GPS ignal, pecial care wa put on deigning a common ignal tructure, from a pectrum point of view, from which both GPS and Galileo could profit the mot. The main idea behind wa that in the future receiver will not only receive GPS or Galileo alone but will try to ue both to exploit the geometry improvement brought by a dual contellation. The firt tep toward that i obviouly that the receiver hould be capable of proceing both Galileo E OS and GPS LC ignal together. A we have een in previou ection, Galileo E OS and GPS LC are very imilar ignal in the ene that they have the ame PSD and it ha been hown that in average they perform the ame. However, the CBOC and TMBOC modulation are quite different and an optimum combined receiver hould account for the difference between both. CBOC i a four-level ignal with data and pilot in antiphae. In addition, the power plit between data and pilot i of 5/5 and both data and pilot have a narrow BOC(,) and a wide-band BOC(6,) component with the ame power. On the contrary, TMBOC i a binary ignal with a data narrow-band channel coniting of only BOC(,) ymbol while the pilot channel contain all the wide-band BOC(6,) component of the ignal. Moreover, 75% of the total power concentrate on the pilot channel while only 5% i allocated to the data channel. Two ignal, one pectrum, two different concept. When conidering future combined GPS/Galileo receiver, the firt tep i to invetigate how both ignal can interfere with each other and if thi can be detrimental for the ignal proceing part. Thi i done in the next ection ASSESSMENT OF CBOC/TMBOC SELF- INTERFERENCE The main objective of thi ection i to how the degradation that a Galileo or GPS receiver i theoretically expected to uffer when other Galileo and GPS ignal interfere with the ueful ignal. In order to meaure thi effect, the equivalent increae of the noie floor due to non-deired ignal will be invetigated at the prompt correlator output. It i well known that the equivalent noie floor at the correlator input when an interferer i preent i the um of: the thermal noie floor, and an additional noie floor that create the ame error variance at the correlator output a the interferer. In the preent cae, the interference can be due to the ame ytem a i the ituation of that interference coming from the deired ignal (intra-ytem interference) or due to a different ytem (inter-ytem interference). Thu: ( N ) = N I eff + INTRA + I INTER (5) The intra- and inter-ytem interference depend on the power of the interfering ignal and a well known figure called Spectral Separation Coefficient (SSC). I INTRA can generally be defined a []: 533

5 I INTRA = t t N G INTRA j= C κ ( f + f ) j j dop df where N INTRA i the number of received ignal belonging to the ame ytem a the deired one, C i the received power of ignal j, j i the receiver front-end-bandwidth, r i the tranmitter front-end-bandwidth, t ( f ) G i the PSD of the deired ignal, (6) f i the Doppler frequency offet of the deired dop ignal, κ j i the SSC between ignal j and the deired ignal, G j i the PSD of the non deired ignal j, and f dop j i the Doppler frequency offet of the non deired ignal j The computation of the equivalent noie power denity I INTER i identical to I INTRA in (6) with the only exception that the ummation in the numerator ha to be done for all ignal that do not belong to the deired ignal ytem. Due to the increaing number of GNSS atellite and thu ignal, the inter- and intra-ytem interference ha to be aeed thoroughly in order to make ure that it doe not harm the acquiition and tracking operation. In particular, it i important to control that certain common aumption ued for the computation of the SSC, uch a the infinite preading code length, are uable in order to minimize the rik of unforeeen interoperability problem. Computation of SSC via imulation A generic model to compute the SSC, repreenting the typical correlation operation, i given in Figure 5. It mut be noted that the input ignal before the correlation i normalized to W of power according to (6). If the only interference i an Additive White Gauian Noie (AWGN), and the front-end filter i aumed rectangular, the PSD of the product between the incoming ignal and the reference ignal i equal to: G r + u ( f ) = N rect G ( u f ) du = N G ( u f ) d r r where N repreent the noie floor. u (7) Thi PSD ha a frequency occupancy ignificantly large compared to the I&D filter bandwidth. Conequently, the PSD of the correlator output can be approximated by: G r ( f ) G () H ID ( f ) = N H ID ( f ) G ( f ) d r where ( f ) H ID i the Integrate and Dump filter. Figure 5. SSC Model to meaure the increae of noie floor due to CDMA noie from other non-deired ignal The correlator output noie variance will then be: r + N σ G ( f ) df = G ( f ) df (9) T r where T i the coherent integration of the interfering ignal and the replica. But a more intereting cae i that of non-white interference. Then, the variance of the correlator output noie i hown to be: r σ = Gk ( u) G ( u f) du H ID ( f) df r () Auming that the multiplier output PSD i ufficiently flat acro the I&D filter, which i true when a very long peudorandom noie code i employed, the aumption of flatne of Gk ( u) G ( u f ) du can be ued a it wa done for the AWGN cae. The SSC i then hown to be approximated by the following expreion: r G r k r r ( f ) G ( f ) ( f ) f (8) df κ () k G df While thi i true for very long code, it ha been hown in the literature ([8], [9], [] and []) that for horter code important difference can be oberved between the ideal flat SSC and that meaured at the receiver. Indeed, it i well known that the PSD of CDMA ignal uing hort code i a line pectrum and thu cannot be conidered a moothed. In thi cae, the SSC can only be expreed a: 534

6 r r ( u) G ( u f ) du H ( f ) Gk ID df κ k () G df r r ( f ) However, thi expreion i very difficult to ae. The mot extreme cae i that of the C/A code where due to it hort code nature and it repetition within a data bit, the SSC can raie by nearly db in the wort cae. It i then neceary to ae the SSC value between GPS and Galileo MBOC ignal through a erie of tet that model exactly the correlator output model hown in Figure 5 uing the true GPS and Galileo preading code. SSC Degradation of CBOC and TMBOC receiver uing imulation It i obviou that the value of the equivalent noie floor will depend upon the power of each interferer. Thu it i of great importance to chooe a cenario repreenting a wort cae with the greatet fidelity. That wa done uing the configuration of Table for the interfering and deired ignal. It repreent a ituation where the interfering ignal (of the ame ytem or from other ytem) are raiing the noie floor to the maximum level with repect to the power of the deired ignal. In addition, it i alo aumed to be repreentative of a ituation where the pectral eparation coefficient will alo be wort cae. However it mut be noted that the increae in noie floor doe not alway correpond to the cae where the SSC adopt it wort cae, but could alo be due to the high interfering power at that pecific location and moment in the time. Table. Parameter of the SSC Simulation Interfering Signal PRN Doppler [Hz] Delay [m] Power [dbw] Deired Signal It mut be noted that the reult would not differ ignificantly if other PRN code were conidered and therefore the reult can be conidered a repreentative of a typical wort cae cenario. It i well known that phae tracking i le robut than code tracking. Conequently, it would be intereting to invetigate the correlator output noie characteritic that would correpond to that ued by the PLL. Two different familie of receiver are conidered herein that correpond to claical receiver configuration to receive an MBOC ignal: TMBOC or CBOC receiver optimized for both ignal waveform, and BOC(,) receiver. In order to implify the number of imulation, only the Galileo E data channel i conidered. Simulation have hown that the inter and intra-ytem interference degradation due to the Galileo E OS pilot channel i comparable to that due to the data channel ince the pectrum i alo very imilar. Thu in the following, CBOC will refer to the CBOC data ignal only. For both receiver configuration, the ame imulated trategy wa employed. It follow a linear approach and the correlator output wa aeed for all poible code phae offet between the interfering normalized ignal and the ideal replica in tep of / chip. That mean that for the cae of a Galileo receiver 49x different code offet flow into the tatitic while for GPS we will have 3x ample. Moreover, for all the imulation a dwell time of m wa ued for the receiver. Thi correpond to the typical PLL integration time. Indeed, due to the receiver clock intability, the correlator that will feed the PLL dicriminator will likely not ue a very long integration. It eem that m, already ued for GPS C/A, i a good duration. Thi mean in other word that auming a deired Galileo ignal, the conidered output i the um of 5 time 4 m coherent integration ince the primary Galileo E OS code duration i 4 m. In the ame manner, ince the length of the GPS LC primary code i of m, the conidered output will conit of two ample reulting from a coherent integration of m. Figure 6 how the output of a CBOC correlator when the interfering ignal are the data channel of Galileo E OS with it correponding code, or the data channel of Galileo E OS pread with random code. A it can be oberved, the Galileo E OS code deliver value lightly higher than thoe of an ideal random code, proving the interet of thi analyi. Table ummarize the reult of all the analyzed cae ( ample were ued for each tet cae). A it can be een, the SSC obtained uing the true Galileo and GPS code are very imilar (within.5 db) to the theoretical and imulated one auming a mooth PSD. An explanation for thi i the fact that the data bit (or the econdary code if the pilot channel i concerned) help in 535

7 making the preading code look longer, and thu more random. It can be concluded from the imulation that the TMBOC and CBOC implementation of MBOC, in pite of being lightly different, will not interfere with each other ignificantly, and more importantly, can be very well approximated by the mooth pectrum approach. Figure 6. m CBOC Correlator Output Amplitude auming an interferer formed by CBOC ignal modulated with random and Galileo E OS code Furthermore, the introduced additional degradation with repect to ideal pectrum SSC i meaured by the following parameter: CDMA σ CDMA Δσ = (3) RND σ RND A it can be een in Table, thi parameter i alway lower than.5 db and indeed fall within the accuracy of the realized imulation. In addition, it i alo expected that more accurate imulation will further reduce the relative increae of the SSC. it i now important to look at the poible architecture that could be ued in a combined GPS/Galileo E/L receiver. POSSIBLE COMMON CBOC/TMBOC TRACKING ARCHITECTURES AND PERFORMANCES A mentioned in the introductory part, the TMBOC and CBOC modulation, although baed on the ame two ubcarrier and giving the ame PSD, have a very different implementation in the time domain. Conequently, achieving a tracking platform that could be adapted to both modulation i a real challenge. One of the idea, invetigated in [3] and [6] for CBOC tracking, conited in the retricted ue of pure ubcarrier in order to limit the receiver complexity. Indeed, by doing o, the receiver doe not require the ue of a multi-bit local replica to track a CBOC ignal. Following that idea, two olution were een a very promiing: the TM6 technique and the dual correlator technique. The TM6 Tracking Technique Thi technique wa deigned to retrict the tracking loop complexity to the lowet poible level, minimizing the number of correlator ued. In that context, TM6 i baed on the ue of Early and Late (E and L) correlation between the incoming CBOC and a pure BOC(6,) local replica and a Prompt (P) correlation between the incoming CBOC and a pure BOC(,). A Dot-Product (DP) dicriminator i then formed uing thee correlator output. The idea behind i to ue the teep BOC(6,) autocorrelation lope to improve the ynchronization (ue of E-L) while uing the power available in the incoming BOC(,) component of the CBOC (ue of P) not to uffer from the high BOC(6,)/CBOC correlation loe. Thi method i well documented in [3]. Since it ha been hown that there wa no major unforeeen threat from the GPS/Galileo elf-interference, Table. Derived Spectral Separation Coefficient for different receiver and interference configuration. BOC refer to BOC(,) and CBOC to the data in-phae component. Interfering Sytem and Signal Deired Sytem and Signal σ CDMA σ RND CDMA Δ σ RND SSC CDMA SSC RND Ideal SSC Galileo CBOC Galileo CBOC GPS BOC Galileo BOC GPS TMBOC Galileo CBOC Galileo CBOC Galileo BOC Galileo CBOC GPS BOC GPS TMBOC Galileo BOC Galileo BOC GPS BOC GPS TMBOC GPS BOC GPS TMBOC GPS TMBOC Galileo CBOC GPS TMBOC Galileo BOC Galileo BOC GPS BOC GPS BOC GPS BOC Galileo CBOC GPS BOC GPS TMBOC

8 It can be hown that under thermal noie TM6 ha a performance in term of equivalent C/N that i.5 db wore than conventional CBOC tracking, but on the other hand.6 db better compared to pure BOC(,) tracking (auming an incoming BOC(,) ignal). Thi i very intereting becaue it mean that even though the tracking technique i very imple, it till outperform BOC(,) tracking, which wa the previou Galileo/GPS E/L baeline modulation for open and civil ignal. It wa alo hown that TM6 wa offering an excellent multipath reitance, even better than the one inherent to the CBOC(6,,/) modulation. The Dual Correlator Technique Thi technique wa deigned in order to realize two parallel correlation: one between the incoming CBOC and a pure local BOC(,) replica and one between the incoming CBOC and a pure local BOC(6,) replica. The two output will be linearly added to form a compoite correlator output uing the linear property of the correlation operation: TI () t ee C () t ( ρboc(,)() t BOC( 6,)() t ) TI ρ T dt = I () t ee C () t BOC(,)() t dt () t ee C () t BOC( 6,)()t t d (4) Thi i done for the three E, L, P correlator. A one can recognize, thi method poe a lightly higher complexity compared to TM6, but till ue binary local replica. Initial work wa conducted in [6] and proved to be extremely promiing. Indeed, by chooing ρ = P and = Q, the dual correlator technique i trictly equivalent to realizing an optimal CBOC correlation. However, by imply changing the repective value of ρ and, which can be done in oftware when forming the dicriminator function, it can be expected that more uitable tracking performance can be achieved according to the uer prioritie. Changing the value of ρ and mean, in a trictly equivalent ene, that the incoming CBOC(6,,/) will be tracked uing a different local ρ + ). Thi mean that different CBOC(6,, ( ) tracking trategie depending on whether we want to pay more or le attention to the BOC(6,) component can be ued. A an example, a hown in Figure 8 (uing the Running Average Multipath Envelope figure of merit), uing a local CBOC(6,,p), with p>/ will help in mitigating multipath better than conventional CBOC(6,,/) tracking becaue the multipath mitigation capacity of the BOC(6,) i higher than that of the BOC(,) modulation. However, if p i choen too high, thi might alo degrade ignificantly the reitance of the code tracking loop to thermal noie (due to the high correlation loe). For CBOC(6,,/, - ) tracking, [6] recommend to ue a ratio ρ between.6 and 3. to have a good compromie between multipath rejection and tracking in white noie (thee two value correpond to a local CBOC(6,,p, - ) with p = 4/ and p = / repectively). The correponding equivalent CBOC local replica are hown in Figure 7. It i alo poible, if a pecific local CBOC waveform i found to fulfill the receiver manufacturer need, to directly generate locally the CBOC waveform of interet. Local Waveform Rho/Beta=3. Rho/Beta= Figure 7. Example of Equivalent CBOC Local Replica to Track an Incoming CBOC(6,,/) Signal Uing the Dual Correlator Technique Running Average Multipath Envelope Error (m) CBOC(6,,/,'-') or Rho/Beta = P/Q Rho/Beta =.6 Rho/Beta = Rho/Beta = Multipath Delay (m) Figure 8. Dual correlator CBOC Multipath mitigation capability for different ρ value In any cae, it i uggeted that, if a DP dicriminator i ued, the prompt correlator trie to gather a much ueful power a poible ince it i reponible for controlling the quaring loe. TMBOC Tracking The correlation degradation, auming perfect ynchronization and an infinite front-end bandwidth, between the TMBOC and a pure BOC(,) or BOC(6,) replica i given by: deg 9 ( ) BOC. 88 (,)/ TMBOC = R (6,,4 / 33) 33 BOC(,) (5) deg 4 ( ) BOC. ( 6,)/ TMBOC = R (6,,4 / 33) 33 BOC( 6,) (6) Comparatively, the ame degradation with an incoming CBOC(6,,/) i: deg ( 95 BOC(,)/ CBOC = R (6,,,' ') BOC (, ) (7) 537

9 deg ( ). 3 BOC( 6,)/ CBOC = R (8) (6,,,' ') BOC ( 6, ) The conequence i that the degradation i ignificantly tronger for the TMBOC modulation. In particular, the lo uffered from the TMBOC/BOC(6,) component i too high to hope uing the TM6 tracking technique. In the ame way, it can be expected that the dual correlator method will be trongly degraded. Auming, for intance, that the BOC(6,)/TMBOC correlation i of interet, a well-known method againt the aforementioned phenomenon i to generate locally a replica compoed of the time-multiplexing of where the BOC(,) ub-carrier i ued in the incoming TMBOC and a BOC(6,) ub-carrier the ret of the time. The ame can be done for the BOC(,)/TMBOC correlation. Thi trongly reduce the correlation loe and put them at the ame level a the CBOC cae. The reulting effect of uing thee time-multiplexed ubcarrier (BOC(,) with and BOC(6,) with ) i that both the TM6 and the dual-correlator method become relevant. In thi cae, the TM6 tracking technique applied to TMBOC lead to performance very imilar to thoe obtained with the CBOC(6,,/, - ) (actually. db better in tracking and approximately the ame multipath mitigation capability). The analyi of the dual correlator method i very intereting. Figure 9 how the code tracking error variance degradation when uing the dual correlator technique receiving a TMBOC(6,,4/33) for different value of ρ and auming an infinite front-end filter. It can be een that optimal tracking i obtained for ρ =, which i normal ince thi correpond to an equivalent local TMBOC(6,,4/33) ub-carrier. It can alo be een that there i a mall window below that optimal ρ value where the tracking degradation i maller than.5 db. Chooing a value in thi intereting window mean a potential improvement of the multipath mitigation capability while not ignificantly reducing the code tracking error due to white noie. To fully analyze the tranpoition of the TM6 and dual correlator tracking technique to TMBOC, it i important to underline certain limitation of that tranpoition: - For the TM6 technique, the ue of a local replica that i a time-multiplexed ignal made of a pure ub-carrier and reult in only a partial correlation ince not all the preading code chip are ued. Thi could degrade the cro-correlation propertie of the preading code. Although it might not be ignificant for the BOC(,) part (it ha to be reminded that in thi cae there are till 9/33 of 3 chip that are ued), it might be detrimental for the BOC(6,) part. Thi i le of a problem for the dual correlator method ince a TMBOC ignal i recompoed after the linear ummation, although becaue of the different recompoition there might be ome loe of cro-correlation propertie. - It might not be intereting, for TMBOC tracking, to replace the ue of a local TMBOC replica by a local replica that i a time-multiplexing of a pure ub-carrier with. Indeed, the gain in receiver complexity in thi cae i not a ignificant a for the CBOC cae. Still, when looking at the dual correlation method, it might be intereting to have the poibility to modify the value of ρ on-the-fly. Code Tracking Noie Variance DEgradation (db) CBOC(6,,/,'-') TMBOC(6,,4/33) Ratio Rho/Beta Figure 9. Code tracking degradation a a function of ρ uing the dual correlator approach HRC ARCHITECTURE AGAINST MULTIPATH The previou ection looked, among other thing, at the impact of multipath on different CBOC/TMBOC tracking technique. When looking at the global approach againt multipath, it i reaonable to invetigate if well-known method that were deigned for the GPS C/A code can be eaily adapted for CBOC and TMBOC tracking. A typical example of uch multipath mitigation technique i the High Reolution Correlator (HRC) preented in [7]. The HRC ue 5 correlator (E, E, P, L, L) located at [-d, - d,, d, d] chip where d i the E-L pacing. The yntheized HRC E and L correlator can then be yntheized a: E HRC ( τ ) = E ( E + P) L HRC ( τ ) = L ( L + P) (9) τ τ = E HRC ( ) L ( ) ( E L) ( E L) HRC Figure how the multipath envelope aociated to the ue of the HRC method with a BOC(,) receiver receiving a BOC(,) ignal, a CBOC(6,,/, - ) 538

10 receiver receiving a CBOC(6,,/, - ) ignal, and a BOC(,) receiver receiving a CBOC(6,,/, - ) ignal. It can be een that the CBOC modulation doe not eem to be very well adapted for the ue of the HRC. Indeed, many bump can be oberved on the multipath error envelope, while thee bump are not preent for the cae of the BOC(,) ignal. We can alo recognize that to mitigate multipath with an incoming CBOC ignal, the ue of a BOC(,) receiver with HRC perform better than a CBOC receiver with HRC. Tracking Error Enveloppe (Chip) BOC(,) BOC(,) HRC - BOC(,) HRC - BOC(,) HRC - CBOC(6,,/,'-')/BOC(,) HRC - CBOC(6,,/,'-')/BOC(,) HRC - CBOC(6,,/,'-') HRC - CBOC(6,,/,'-') Multipath Delay (Chip) Figure Comparion of Multipath Envelope uing HRC Applied to BOC(,), CBOC and TM6 with d=.8 chip and a 4 MHz Front-End Filter Thi implie that a different tracking technique will have to be developed for the CBOC and TMBOC ignal in order to achieve a multipath mitigation equivalent to that of GPS C/A code when the HRC i ued. One way to achieve thi i to ue more than 5 correlator in order to yntheize a dicriminator function that would achieve a deired multipath rejection. Thi i the cope of the following ection. OPTIMUM S-CURVE SHAPING OF THE DIFFERENT MBOC IMPLEMENTATIONS In 5 a new method to derive an optimum dicriminator againt code multipath mitigation wa preented in [4] uing a multi-correlator receiver. At that time, that work wa applied to BPSK() and BOC(,) ignal. In 7, further work ha been realized with the convenient modification to take into account the different MBOC implementation. Detailed information on thi multipath mitigation technique i available in [5]. ~ N D ( Δτ ) = α R i i ( Δτ ) () i= where Δτ i the code tracking error, defined a the difference between the etimated code delay and the true code delay, and α i i the weight of each correlator. Furthermore, R i (Δτ) i defined a: A OFF j ϕat ϕ rec R Δτ = R Δτ + d e () i ( ) ( ) ( ) Thi repreent the band-limited autocorrelation function OFF hifted by d i. Finally, define the dicriminator OFF parameter o a d i and αi to be properly choen for each correlator. The idea behind thi work i firt to define an ideal S- curve. The optimization proce then conit in finding the parameter d i OFF and αi, for a given ignal and a given receiver front-end bandwidth that would reult in an S-curve fitting with the deired one. The typical characteritic of an ideal S-curve are:. A wide linearity range around,. An Unambiguou Tracking Offet (UTO) value that reult in no fale tracking lock point 3. A High-Cut S-Curve (HCS) value The ideal S-Curve ued in the optimization work i depicted in Figure. A hown in figure, the introduction of the UTO value reult in an S-Curve that ha a non-zero value in the outide linear region. Note that thi value, expreed in chip, i very little and ha a different ign on the two ide of the S-Curve. The main reaon to introduce thi offet i that by doing o the code dicriminator will have only one table tracking point in the pull-in region. With repect to the HCS, a evident in figure, it limit the maximum and minimum value of the S-Curve. The price that we have to pay by introducing thee two modification i that we lightly deviate from the ideal, but the advantage in term of performance brought by them reward by far thi deciion. i In [5] two different approache are dicued. We will concentrate on the firt of the two in thi paper. Thi i a conventional tracking loop tructure where multiple correlator are employed. The coherent code phae dicriminator i defined a a linear combination of the correlator output a follow: Figure. Optimum S-Curve 539

11 The fitting proce conit in approaching the ideal S- Curve by a linear combination of hifted replica of the autocorrelation function. It can be initiated after defining the number of correlator and their location d OFF i within a predefined fitting range and with a certain reolution. The weight α i are then calculated, a done in [4], by minimizing the following cot function ~ ~ where D ~ ID [ ( Δτ ) D ( Δτ )] D () ID i the ideal coherent code phae dicriminator. In [5] a realization of the introduced optimum code dicriminator ha been found for the four different MBOC implementation (CBOC Data and Pilot Channel and TMBOC Data and Pilot Channel) at two different value of bandwidth (4 MHz and 4 MHz). For each bandwidth value, a common optimum value of the reolution ha been calculated for all the given ignal and i preented in Table 6. Obviouly a large number of correlator are neceary. However, imilar reult have alo been obtained for all the other ignal of interet not repreented in the previou figure. The reult for relatively hort bandwidth are particularly impreive. In Figure 4 and 5 the obtained multipath envelope are depicted. Figure 4. Multipath Envelope for a bandwidth of 4 MHz Table 3. Common Optimum Code for MBOC Signal: main parameter 4 MHz 4 MHz Reolution [chip].38.8 Extenion of S-Curve [chip].5.5 UTO [chip].5.35 HCS [chip].. Two example of the obtained fitted S-Curve are hown in Figure and Figure 3. Figure 5. Multipath Envelope for a bandwidth of 4 MHz It i important to underline that in both cae the multipath envelope of the four analyzed ignal are almot the ame. Thi eem to be a clear ign that the value of the reolution that have been found for each value of bandwidth eem to be optimal, in the ene that the performance doe not change when the tudied ignal differ. CONCLUSIONS Figure. Example of obtained S-Curve Figure 3. Example of obtained S-Curve It can be een that in both cae, the reulting dicriminator output follow very well the ideal S-curve. The firt paper ha hown that the elf-interference between the future Galileo E OS and GPS LC, given their preading code and ignal tructure, will not bring a ignificant degradation. Moreover, the ue of ideal mooth PSD envelope i a relevant way to etimate their relative SSC. The econd part wa dedicated to the invetigation of two imple tracking technique propoed to track both GPS LC and Galileo E OS pilot channel. The firt method, referred to a TM6 repreent an ultra imple method to reduce the receiver complexity to a minimum while maintaining intereting performance in term of CBOC tracking in white noie and multipath. It ha been hown that thi method only ue pure binary ub-carrier thu avoiding the local generation of a 4-level CBOC replica. 54

12 It ha alo been hown that the adaptation of TM6 for the TMBOC require a more complex local ub-carrier and ome aociated problem uch a the lo of ome cro-correlation propertie that have to be further invetigated. The econd method, the dual correlator tracking technique, ue twice a many correlator a the TM6, realizing once again correlation only between the incoming ignal and pure ub-carrier (BOC(,) and BOC(6,) eparately). It can achieve optimal tracking performance againt both thermal noie and multipath. It can alo be eaily configured in oftware to uit a particular uer requirement (for intance if multipath i the main problem, if low C/N are expected, or if there i an interference located on the BOC(6,) part). Once again, the ue of thi method for CBOC i eaier compared than for TMBOC for which light modification are required in the form of the ue of a time-multiplexed ub-carrier. Still it wa hown to provide an extremely promiing technique. Finally, in the lat part of thi paper, it wa hown that a imple multipath mitigation technique commonly ued in current GPS C/A receiver uch a the HRC technique might not be optimal for CBOC or TMBOC tracking. However, another technique, baed on the optimal ue of multi-correlator output, wa hown to provide excellent multipath mitigation capability for both MBOC implementation, pending a more complex receiver (but that hould not be problematic in the long term when multi-correlator receiver could be the baeline). ACKNOWLEDGEMENTS The author from FAF Univerity, Munich, wih to acknowledge the financial upport for thi work by the German Aeropace Center Deutche Zentrum für Luftund Raumfahrt (DLR). REFERENCES [] 67_Civil_Signal_Accord.ap [] nce=ip/7/8&format=html&aged=&language=en &guilanguage=fr [3] Avila-Rodriguez, J.A., G.W. Hein, S. Wallner, J.- L. Iler, L. Rie, L. Letarquit, A. de Latour, J. Godet, F. Batide, A.R. Pratt and J. Owen (7), The MBOC Modulation: The Final Touch to the Galileo Frequency and Signal Plan, Proceeding of the ION- GNSS 7 (5-8 September, Fort Worth, TX, USA). [4] J.A., Avila-Rodriguez (7), On Optimized Signal Waveform for GNSS, Ph.D. Thei, Univerity FAF Munich, Neubiberg, Germany (awaiting publication) [5] Hein, G.W., J.-A. Avila-Rodriguez, S. Wallner, A.R. Pratt, J.I.R. Owen, J.-L. Iler, J.W. Betz, C.J. Hegarty, Lt L.S. Lenahan, J.J. Ruhanan, A.L. Kraay, and T.A. Stanell (6), MBOC: The New Optimized Spreading Modulation Recommended for GALILEO L OS and GPS LC, Proceeding of IEEE/ION PLANS (4-7 April, San Diego, CA, USA). [6] Hein, G.W., J.-A. Avila-Rodriguez, L. Rie, L. Letarquit, J.-L. Iler, J. Godet, A.R. Pratt (5): A Candidate for the Galileo L OS Optimized Signal, Proceeding of the ION-GNSS (3-6 September, 5, Long Beach, CA, USA). [7] Draft IS-GPS-8 Navtar GPS Space Segment/Uer Segment LC Interface, 9 April 6. [8] Ruhanan, J. J. (7) The Spreading and Overlay Code for the LC Signal, Navigation, the Journal of the ION, Vol. 54, No., pp 43-5 [9] Raghavan, S., K. Tai and L. Cooper (4), GPS CDMA Noie Analyi, AIAA-4-38 nd AIAA International Communication Satellite Sytem Conference and Exhibit 4 (ICSSC), (May 9-, Monterey, CA, USA) [] Kumar, R., S.H. Raghavan, M. Zeitzew, P. Munjal and S. Lazar (999) Analyi of Code Cro Correlation Noie in GPS Receiver Operating in Augmented GPS Sytem, Proceeding of the ION-NTM (Jan. San Diego, CA, USA). [] Raghavan, S., R. Kumar, S. Lazar, M. Zeitzew, R. Wong, J. Michaelon, A. Doran, and M. Bottjer (999), The CDMA Limit of C/A Code in GPS Application - Analyi and Laboratory Tet Reult, Proceeding of the ION-GNSS 999 (4-7 Sept., Nahville, TN, USA), pp [] Wallner, S., G.W. Hein, J.-A. Avila Rodriguez, T. Pany, and A. Pofay (5), Interference Computation Between GPS and Galileo, Proceeding of the ION-GNSS (3-6 Sept., Long Beach, CA, USA). [3] Julien, O., C. Macabiau, J-L. Iler, and L. Rie (6), -bit Proceing of Compoite BOC (CBOC) Signal, t ESA-CNES workhop on GNSS ignal (Touloue, France) 54

13 [4] Pany, T., M. Irigler, and B. Eifeller (5), S- Curve Shaping: A New Method for Optimum Dicriminator Baed Code Multipath Mitigation, Proceeding of the ION-GNSS (3-6 Sept., Long Beach, CA, USA). [5] Paonni, M., J.-A. Avila-Rodriguez, T. Pany, G.W. Hein, and B. Eifeller (7) Looking for an Optimum S- Curve Shaping of the Different MBOC Implementation, Proceeding of the ION-GNSS (5-8 September, 7, Fort Worth, TX, USA). [6] Julien, O., C. Macabiau, J-L. Iler, and L. Rie (7) -Bit Proceing of Compoite BOC (CBOC) Signal and Extenion to Time-Multiplexed BOC (TMBOC) Signal, Proceeding of the ION NTM (San Diego, CA, USA) [7] Mc Graw, G, and M. Braah (999), GNSS Multipath Mitigation Uing Gated and High Reolution Correlator Concept, Proceeding of the US Intitute of Navigation NTM (San Diego, CA, Jan. 5-7), pp