Multipath Mitigation Techniques for CBOC, TMBOC and AltBOC Signals using Advanced Correlators Architectures

Transcription

1 Multipath Mitigation Tehniques for CBOC, TMBOC and AltBOC Signals using Advaned Correlators Arhitetures Aleksandar Jovanovi, Youssef Tawk, Cyril Botteron, Pierre-André Farine Eole Polytehnique Fédérale de Lausanne, Institute of Miroengineering (IMT), Eletronis and Signal Proessing Laboratory, Breguet 2, 2000 Neuhâtel, Switzerland Abstrat Multipath mitigation in urban anyons and indoor environments is an open issue for the reeption of GNSS signals for high preision appliations, as the presene of multipath omponents an lead to signal fading and ranging errors. New families of navigation signals, suh as AltBOC, CBOC and TMBOC bring potential improvements, suh as more signal power, better multipath mitigation apabilities and more robust navigation. Therefore the goal of this paper is to investigate multipath mitigation apabilities of CBOC, TMBOC and AltBOC with different disriminator arhitetures through theoretial analysis and realisti set-up with measurements in order to provide an overview of their performane in different environments. Keywords- Galileo, Multipath, CBOC, TMBOC, AltBOC, Traking I. INTRODUCTION Although GNSS provide aurate information about the position and time on almost every spot on the Earth, there are ommon navigation errors that are unavoidable. One main soure of errors is the effet of multipath oming to the reeiver from the satellites due to signal refletion and sattering. In order to minimize the influene of the indued errors, GPS and Galileo systems reently introdued several modernized modulations speially oneived to provide a better multipath mitigation potential, among them TMBOC (Time Multiplex Binary Offset Carrier), CBOC (Composite Binary Offset Carrier) and AltBOC (Alternate Binary Offset Carrier) [1] [2]. The MBOC modulation was hosen to be implemented on the Galileo E1 and GPS L1C signals. This modulation was speifially designed to improve Galileo and GPS reeivers traking performane at the L1 frequeny while still being bakward ompatible with the previous pure BOC(1,1) baseline. The power density funtion of the MBOC modulation is formed as a sum of the 10/11 normalized BOC(1,1) and of the 1/11 normalized BOC(6,1) signal spetrum. MBOC is defined in the frequeny domain, and two different implementations have been speified for Galileo and GPS, CBOC and TMBOC, respetively [3]. The AltBOC(15,10) modulation is hosen to be implemented on the Galileo E5 band. This modulation allows transmission of four different hannels (E5a-I, E5a-Q, E5b-I and E5b-Q), where the in-phase signals are the data hannels while the quadrature ones are the pilot hannels [1]. A Galileo E5 reeiver will be able to distinguish the four hannels, sine four different quasi-orthogonal Pseudo-Random Noise (PRN) odes are used on eah Galileo satellite. The signal will be transmitted within a 90 MHz bandwidth from the satellites [1], and a minimum of 51 MHz reeiver bandwidth will be needed to aquire the full band. The autoorrelation funtion of the AltBOC, CBOC, and TMBOC signal will be denoted in this paper as R (τ ). For the AltBOC signal it an be approximated as [4]: R( τ ) = β Tri( τ )os( w τ ) (1) where Tri(τ) is the triangular funtion, β the pre-orrelation filter oeffiient, and w s is the angular subarrier frequeny. For CBOC the analytial expression for the autoorrelation funtion an be obtained from [5] : R CBOC = RBOC (1,1) + RBOC (6,1) R (2) BOC (1,1) BOC (6,1) Where R is the ross-orrelation funtion between BOC ( 1,1) BOC(6,1) BOC(1,1) and BOC(6,1), R BOC (1, 1) and R are the BOC( 6,1) autoorrelation funtions for BOC(1,1) and BOC(6,1), respetively. The autoorrelation funtions for the AltBOC, CBOC, and TMBOC are shown in Figure 1 assuming an infinite reeiver bandwidth. For CBOC, the pilot and data hannel autoorrelations are very similar. Therefore, instead of onsidering them separately, we have shown in Figure 1 and used for the simulations the average between the autoorrelation funtions for the CBOC pilot and CBOC data hannels. This orresponds to assuming a ombined pilot and data traking sheme. For TMBOC, as the data hannel has no BOC(6,1) omponent, it is the pilot TMBOC that is shown in s

2 Figure 1 and used for the simulations. It an be seen that the CBOC and TMBOC autoorrelation funtions exhibit narrow main peaks very similar to eah other, and two sideorrelation peaks loated at approximately 0.5 hips around the main peak. For the AltBOC signal, the side peaks are loated around 0.33 hips from the origin. Side peaks an be potential point of false loks, espeially if the multipath signal is present. Figure 1- Autoorrelation funtions of different signals assuming an infinite reeiver bandwidth In fat, with the presene of refleted signals arriving at the reeiver after multiple refletions, diffration, and sattering, the baseband signal, the disriminator output, and the autoorrelation funtion may all be distorted. As the pseudoranges are obtained from these quantities, the distortions may lead to a false or biased lok and onsequently to a range and phase error. While in the reent years several mitigation tehniques have been developed in order to suppress the multipath, some of them showing good multipath mitigation properties, these tehniques may not be appliable to the new BOC modulations. Previously, several papers disussed multipath mitigation tehniques using simplified models and theoretial onsiderations only [6] [7]. The most used and effetive approah, that is used in this work as well, is based on the design of improved reeiver internal orrelations tehniques. More speifially, parameters that haraterize disriminator arhitetures suh as orrelator hip spaing, number of orrelators, pre-orrelation bandwidth, stability and linearity region of the disriminator, are taken into onsideration and their impats on the ode traking auray are presented. The first step of our investigations onsists of analytial derivations and Matlab simulations of the disriminator funtions and multipath mitigation tehniques. These analyses provide insight into the disriminator types that result in minimum ode ranging errors. At a later stage, the investigations are verified pratially using our Spirent GSS8000 Galileo and GPS onstellations simulator [8] onneted to a software reeiver through a triple-band L1, L2 and L5 front-end [9]. With this onfiguration, and the ability of the Spirent to generate realisti senarios, more pratial and relevant results are obtained. Overall, results are ompared by means of multipath error envelopes (MEE). Based on these omparisons, performane of multipath mitigation tehniques is evaluated for eah modulation type. The paper is organized as follows: Setion II provides the general disussion about the ode and multipath, and the way it an be omputed. In setion III and IV narrow and double delta orrelators are desribed with the disriminator urve output analysis. Simulation results for the signals under investigation and different orrelator types are provided in setion V. Setions VI and VII are foused on the desription of the measurement platform and the analysis of the obtained results. Finally, disussion and onlusions are provided in setions VIII and IX. II. MULTIPATH Multipath desribes the effet of satellite signals arriving at the reeiver by more than one path. This means that a diret (line-of-sight) signal overlays with one or more indiret signals (non-line-of-sight). The antenna reeives several replias of the transmitted signal at different moments and with different power amplitudes. Due to the longer propagation path of indiret signals, there is a phase offset (shift in ode and arrier) between multipath signals. The phase offset is a funtion of the geometri onditions but also of time. For automotive appliations in urban environment, the multipath onditions will hange ontinuously, and therefore it is one of the most disturbing problems of GNSS-based navigation systems in urban environments. The pseudo-range may deviate, depending on the signal harateristi, up to 100m or even more, and the position deviation is a funtion of it. Multipath affets ode measurements and phase measurements, but due to the wavelength of the ode observables, the multipath error of ode measurements is higher. The presene of multipath signals an result in ranging errors (ode multipath), arrier phase errors (arrier multipath), Doppler errors, and fading [10]. The following two subsetions provide insights into arrier and ode multipath. A. Carrier Multipath Error In the absene of multipath, the disriminators are able to trak the diret inoming signal phase. In the presene of multipath, the traking loops follow the omposite inoming signal. The traditional traking loops in the reeiver are not able to distinguish diret and omposite signal. They employ the onept of null traking [11] whih results in a non-zero error in the diret arrier phase estimation. Figure 2 Phasor diagram

3 Carrier phase multipath an be studied by onsidering a simple phasor diagram (Figure 2) [11], where the diret signal is assumed to have a zero phase, the phase of the multipath signal with respet to the diret one is θ m, and θ is the phase of the omposite signal. Carrier traking is performed by the PLL whose disriminator reeives as input the orrelation results of the inoming with the loal generated signal. Thus, the magnitudes for the diret and multipath phasor are given by: D = R( τ ) (3) M αr( τ δ ) (4) = where R( τ ) is the autoorrelation funtion for the PRN ode at time lag τ, δ is the delay of the multipath signal, α is the ratio of multipath to diret signal and τ is the ode traking error of the DLL. In order to determine the omposite phaseτ, the multipath is deomposed into its in phase and quadrature omponents: M I = M os( θ m ) (5) M Q = M sin( θ m ) (6) The orrelation values used above are from the prompt orrelator. Having resolved the multipath omponents, it is possible to determine the phase relationship between the omposite and the diret signal, whih is the error due to arrier phase traking. Assuming a perfet lok ontoθ, the arrier phase multipath error is given by: M Q θ = artan (7) D + M I This expression haraterizes the arrier multipath error, and by varying the phase of the multipath relative angle θ m, between 0 and π, a arrier phase multipath error envelope an be found. Carrier multipath error an be haraterized using the equations above and written as [11]: αr( τ δ )sin( θm) θ = artg (8) R( τ ) + αr( τ δ )os( θm) B. Code Multipath Error The multipath environment represents the main fator that affets the pseudoranges through some parameters suh as: number of multipath signals, geometri path delay, phase and relative power. Pseudorange measurement is performed by the ode traking loop that aligns the loally generated PRN ode and the PRN ode of the inoming signal. The lassial approah is to design the disriminator of the DLL using early and late orrelator outputs. However, when multipath is present, the inoming ode phase, orrelation funtion and disriminator funtion are all distorted. Analytially, the line of sight and multipath signals an be treated separately and the total orrelation funtion an be expressed as: R tot ( τ ) = R( τ ) + αr( τ + δ ) (9) where α is the multipath to signal amplitude ratio, and δ the delay of the multipath signal relative to the line of sight signal. The analysis about the multipath error omputation is performed using the general form for the autoorrelation funtion R(τ ) for all investigated signals. The impat of multipath on ode traking auray is often represented as an error envelope representing the maximum error resulting from one single multipath with a ertain phase, delay and amplitude. It is alled multipath error envelope (MEE). It is important to note that, when omputing the multipath-indued ode traking error envelope, it is equivalent to finding the point where the disriminator output rosses the origin, sine this means that it represents the stability point where the loop will lok. This is the approahed we followed in our work. The DLL disriminator funtion obtained by the differene of the early late orrelators with a spaing of one hip was the lassial approah for the ode traking before the digital and software implementations that brought several solutions for the improvement of multipath error, inluding new types of orrelators, suh as narrow and double delta orrelators [10]. The next setion disusses these orrelators, trying to provide an overview of the multipath mitigation strategies, and find the solution for the optimal tehnique for every signal, making the omparison between them and disussing trade-offs. III. NARROW CORRELATOR One of the first approahes to mitigate multipath that was introdued by Novatel in 1992 is alled the narrow orrelator [12]. The onept of narrow orrelator is the redution of the spaing between early (E) and late (L) orrelators. However, band limitation of the front-end filter rounds the autoorrelation peak and therefore the distane between orrelators should not be taken too small. If we assume a arrier phase loked traking, the output of a dot produt (DP) disriminator is equal to: P ( I E I L DP = I ) (10) where the two orrelators are used (early I and late ) with E I L a hip spaing of d hips, and I P is the prompt orrelator. The disriminator outputs for the new signals are shown in Figure 3. It an be observed that the AltBOC, CBOC and TMBOC have almost an idential linear traking region (region in

4 whih the disriminator reats without any bias), whereas BPSK has the widest stability region (region in whih the disriminator reats in the right diretion without losing lok). The stability region of the AltBOC is narrower omparing to CBOC and TMBOC (0.17 omparing to hips in one diretion). traditional orrelator struture. The set of formulas that desribe HRC are provided below [10] [13]: E HRC ( τ ) = 2 E1 ( E2 + P) L HRC ( τ ) = 2 L1 ( L2 + P) (16) τ ) = ( E L ) ( E ) / 2 D HRC ( L2 where E1 and L1 relate to the inner orrelators with a hip spaing d whereas E 2 and L 2 relate to the outer orrelators with a hip spaing of 2d as shown in Figure 4. Te disriminator output for the HRC is shown in Figure 5 for the signals onsidered. Figure 3 Disriminator output for different signals and E-L hip spaing of d=1/12 In the presene of multipath, I and I an be expressed as: E L I E = R ( τ d / 2) + α R( τ d / 2 + δ ) (11) I L = R ( τ + d / 2) + α R( τ + d / 2 + δ ). (12) When no multipath signal is present, and in the ase of ode loking, the disriminator output is equal to zero at zero delay ( τ = 0 ) and hene: Figure 4 Position of the orrelators using HRC ( I I ) = 0 DP = I (13) P E L I E = I L. (14) In the presene of multipath the ode disriminator is equal to zero at non zero delay ( τ 0 ) and equation (14) is still appliable. It an be written as: R ( τ d /2) + αr( τ d /2+ δ) = R( τ + d /2) + αr( τ + d /2+ δ) (15) where τ is the ode ranging error. Equation (15) is solved by iterations and has multiple solutions. The ode ranging error is equal to the minimum absolute value of the multiple solutions obtained. IV. DOUBLE DELTA CORRELATOR Double Delta (ΔΔ) orrelators are a general expression for speial ode disriminators developed in the last deade [13]. The basi idea of this approah is to inrease the number of orrelators in the DLL, using two pairs of early and late orrelators. HRC (High Resolution Correlator) is a speial ase of Double Delta orrelators where the linear ombination of orrelator outputs is ombined to form early, prompt and late orrelators. It uses five orrelators, two more than the Figure 5 HRC Disriminator output for different signals and E-L hip spaing of d=1/12 What an be observed omparing to the narrow orrelator with DP disriminator is that the HRC disriminator behaves better for AltBOC, CBOC and TMBOC in terms of stability traking region, sine there are two more orrelators that help to stay on the main autoorrelation peak. In ontrast, the BPSK stability region is lower, but its linearity region is inreased, and for the other signals the linearity region is similar while the amplitude dynamis of the output is inreased. V. SIMULATION RESULTS The simulations are based on multipath signal simulations aording to the widely used model assuming a diret signal and a single multipath replia with relative amplitude α and delay δ, as given in the following equation:

5 r ( t) = s( t) + α s( t δ ) (17) In our simulations we use a sattered ray with amplitude α=0.5 that is either in-phase or out of phase omparing to the diret ray. The simulation parameters are listed in Table 1. Simulation Assumptions AltBOC CBOC TMBOC Bandwidth 75 MHz 18 MHz 18 MHz Chip spaing (2d) 1/ / / SMR [db] (20 log α) Table 1 Simulation parameters for eah signal The following setions desribe the simulation results for ode and arrier multipath error obtained for eah signal separately as well as their performane omparison. A. AltBOC BPSK(10) The AltBOC signal is haraterized by a very sharp main autoorrelation peak and side-peaks, and it an be traked in two ways: as a pure AltBOC signal, taking advantage of the full bandwidth, or as a BPSK(10) traking either E5a or E5b side-lobe. For the narrow orrelator, Figure 6 shows the multipath error envelopes for both modulations. It an be seen that AltBOC is less sensitive to hip spaing variations than BPSK(10). For a hip spaing of 1/12, BPSK(10) performs better for short multipath delays, and AltBOC has better mitigation for long delays. By inreasing the hip spaing to 0.25, the AltBOC outperforms BPSK(10) for all multipath delays. Figure 7- Carrier Multipath Error for AltBOC for the hip spaing of d=1/12 Figure 8- Carrier Multipath Error for BPSK(10) for the hip spaing of d=1/12 Figure 6 Multipath Error Envelopes for AltBOC and BPSK(10) signal using narrow orrelator The arrier multipath error envelope is plotted in Figure 7 and Figure 8, where it an be seen that the phase error is very small whih is very important for high-preision appliations (e.g. RTK). The range of the error for both ases is the same. Sine the arrier multipath performane depends on the shape of the autoorrelation funtion and not on the orrelator spaing, the arrier multipath errors illustrated in Figure 7 and Figure 8 are valid for different orrelator spaings and will not be signifiantly affeted by the reeiver bandwidth. The MEE s for HRC orrelator are shown in Figure 9. Figure 9 Multipath Error Envelope for AltBOC and BPSK(10) using HRC For the AltBOC signal, the HRC with a hip spaing of 0.25 has a lower performane than the narrow orrelator. However, for a hip spaing of 1/12, the HRC has a better performane than the narrow orrelator. For BPSK(10), the high resolution orrelator is showing a better performane for 1/12 hip spaing. Using 0.25 spaing, for short range multipath delays, the narrow orrelator provides better mitigation and for long delays the HRC is better. In general, in both ases the improvement brought by the HRC is not so great, if we take in onsideration the inreased ode noise traking error indued

6 by using more orrelators. For the arrier multipath error envelope, a similar performane is obtained as when using a narrow orrelator as the PLL depends more on the autoorrelation funtion than on the disriminator type and hip spaing. B. CBOC & TMBOC Sine the CBOC signal is formed as a omposite sum of BOC(1,1) and BOC(6,1) replias, it has a different performane in traking and multipath performane as well, omparing to TMBOC that is a time-domain multiplex of BOC(1,1) and BOC(6,1). The multipath error envelopes for the E-L orrelator with the distane of 1/12 hips and narrow orrelator using DP disriminator are shown in Figure 10. What an be observed is that the CBOC multipath error is dereased omparing to BOC(1,1), whih is less sensitive to multipath delay variations. For a orrelator spaing greater than 0.2 hips, CBOC behaves similarly to BOC(1,1) and the improvement of CBOC is minor. The onlusion is that multipath mitigation is muh better as long as we use a small orrelator spaing, but one should be areful about the tradeoff between the linearity region of the disriminator urve that dereases with smaller E-L spaing and the effet of front-end filtering on the autoorrelation funtion. The multipath error for the TMBOC signal is shown in Figure 10. It is very similar to the CBOC multipath error. of the multipath error and less wavy behavior of the multipath error harateristi for the delays up to 150m. Figure 11 - Carrier Multipath Error for CBOC and the hip spaing of d=1/12 The performane of CBOC and TMBOC using HRC is shown in Figure 12. The onlusion about HRC is that it better suits BOC(1,1) than CBOC and TMBOC for the range of delays between 30m-120m. For the range of delays between 185m- 275m, TMBOC behaves similarly as BOC(1,1) but CBOC shows poorer performane omparing to the two. Multipath mitigation for CBOC and TMBOC is better than for BOC(1,1) for the ranges of delays between m, and a possible mitigation tehnique would be to ombine HRC and NC disriminators depending on the multipath onditions in order to better mitigate multipath. Figure 10 - Multipath Error Envelopes for CBOC and TMBOC signal using narrow orrelator with the hip spaing of d=1/12 The arrier multipath error for the CBOC signal is shown in Figure 11. The arrier multipath error for TMBOC modulation shows similar results as for CBOC and therefore is not shown here. We an observe that the arrier multipath error is not ritial, sine it auses a maximal absolute value of 0.015m. It is interesting to observe that the maximum arrier multipath error for CBOC and TMBOC are smaller than for AltBOC. TMBOC multipath error is more similar to the CBOC(-) one that is used on the pilot hannel. The only differene with respet to ode multipath error that an be observed between the two is that TMBOC behaves slightly better for most of the long multipath delays (>=150m). Therefore for these delays it an better mitigate the multipath. Also it shows less variation Figure 12 - Multipath Error Envelopes for CBOC and TMBOC using HRC with the hip spaing of d=1/12 VI. MEASUREMENT SET-UP The measurement setup onsists of a Spirent GSS8000 simulator [8] for the full onstellation simulations of Galileo E1, E5, and GPS L1C, and a Fraunhofer triple-band front-end [9] to reeive and aquire the signals under investigation on a omputer. The post-proessing is performed using a software reeiver implemented in Matlab and based on modifiations of Kai Borre GPS L1 software reeiver [14] suh that the CBOC, TMBOC and AltBOC an also be reeived and postproessed. The platform set-up is shown on Figure 13.

7 In the following setion, the results using the desribed postproessing arhiteture are provided, analyzed and disussed. Figure 13 Measurement set-up onsisting of front-end, Spirent GSS8000, and the post-proessing software The parameters of the front-end and the software radio used are provided in Table 1. Pre-orrelation parameters BPSK CBOC TMBOC Bandwidth [MHz] IF [MHz] Sampling Frequeny [MHz] Correlator spaing [hips] /12 1/12 Disriminator type DP E-L Power/DP E-L Power/DP Table 2 Parameters of the platform set-up The senario that is investigated in this setup onsists of two satellites signals generated by the Spirent GSS8000 simulator, where the satellites are situated on the same position, and one signal is generated with multipath and the other one without multipath. The differene of the residual ode phase error resulting from the output of the DLL disriminator for the two satellite signals onsidered thus represents the ode multipath error. This is the approah that is used to ompute the error. This way, the performane of the traking loops in the presene of multipath an be investigated and analyzed. The post-proessing arhiteture is shown in Figure 14. Eah blok represents a speifi proessing entity. The data is first aptured by the front-end are formatted in two files: one for aquisition and one for traking. During aquisition, the ode phase of the primary and seondary odes are obtained using a non-oherent aquisition method, as well as the arrier Doppler frequeny. The obtained parameters are then sent to the traking blok that initializes the traking loops and provides residual ode offset, Doppler frequeny offset, I and Q orrelator outputs and ode multipath error. A traditional traking sheme is assumed with two disriminator types: DP (Dot-Produt) and EMLP (Early Minus Late Power). EMLP uses the differene of the squared E and L orrelator outputs in order to form the disriminator funtion (normalization is also used to avoid the amplitude sensitivity). Figure 14 - Post-Proessing Arhiteture VII. MEASUREMENT RESULTS As the front-end bandwidth on the E5 hannel is limited to 18 MHz, the results are only provided for the E5a-Q pilot hannel that an be traked as a pure BPSK(10) signal. Also, the results are only provided for the CBOC pilot signal due to the time onstraint and the omplexity of measurements. As it was shown before, the TMBOC signal shows very similar performane. Finally, the performane of the high resolution orrelator was not onsidered due to the same reasons mentioned above and is left for the future work. A. E5a-Q BPSK(10) The influene of the multipath on the autoorrelation peak during the aquisition proess for one partiular multipath delay is shown on Figure 15. We observe that the autoorrelation funtion is distorted for the satellite having multipath inluded, and the amplitude of the peak is higher sine the multipath is added in a onstrutive way. Figure 15 Autoorrelation funtion with and without multipath for E5a-Q aquisition The residual ode offset error for the ase with and without multipath is shown in Figure 16. The sampling frequeny used results in a four samples per hip for a PRN ode rate of MHz whih leads to the NCO to reset every time the ode is shifted one quarter of a hip. The ode multipath error is

8 omputed using the differene of ode NCO outputs averaged over time (umulative average) as shown in Figure 17. early and late will result in the same error due to the limitation in sampling frequeny, as the minimum possible ode shift is 0.25 (i.e., 4 samples per hip) and any spaing less than that will be onsidered by the software reeiver as Figure 16 Code NCO output for the satellite with and without multipath inluded One the disriminator is stabilized, the ode error is onstant and an be omputed for a speifi multipath delay. This operation is repeated over 15 multipath delays for eah signal and using three different data sets in order to ompute the averaged multipath error envelope as shown in Figure 18. Figure 17 Averaged ode multipath error Figure 19 DLL disriminator output for a 0.25 hips spaing in the ase of E5aQ signal Finally, the ode DLL disriminator output is plotted for the two satellites in Figure 19 where it an be seen that for both ases the disriminator is stabilized slightly around 2s whih means that the multipath does not affet the stability of the disriminator as it will be shown also for CBOC later. In the ase of the AltBOC modulation, similar onlusions ould be stated as in BPSK(10), espeially onsidering the sampling frequeny as both modulation have the same ode rate and to be able to obtain the improvement brought by a narrower hip spaing a higher sampling frequeny is needed. Finally, as the autoorrelation peak is sharper, very narrow hip spaing an results in a longer stabilization time for the disriminator espeially if using low bandwidth, as the effet of filtering will flatness the orrelation peak and hene the early and late orrelators will have almost the same amplitude and onsequently the traking loop may not be able to lok. Therefore a ompromise should be taken between the 3 fators: bandwidth, sampling frequeny and E-L spaing. B. E1 CBOC Galileo CBOC signal has two side-peaks loated approximately 0.5 hips from the main peak and as suh is more suseptible to false loking due to multipath. Bisetion of the 3-D aquisition graph obtained for the multipath delay of 185m is shown in Figure 20. Figure 18 Multipath Error Envelope for BPSK(10) using 0.25 hip spaing and DP disriminator It an be seen that the ode ranging error measurement results are slightly higher than the simulation results. This an be explained sine the front end bandwidth is limited to 18 MHz and the main lobe of the E5aQ is spread over 20 MHz whih leads to a loss of energy and hene an inreased error. Moreover, the effet of noise was not inluded in the simulations and the error between 7 m and 27 m is onstant but for measurement results there was a variation. It should be mentioned that a smaller hip spaing than 0.25 hips between Figure 20 Autoorrelation peak obtained during aquisition with and without multipath of 185m delay

9 For this range of delays as an be observed from Figure 10 the multipath is destrutive and the autoorrelation peak amplitude is deeased. The I and Q outputs from the DLL traking loop for the ase of generated multipath delay of 225 m is shown in Figure 21. The loking proess is stabilized for a short time after the first 200ms. The I outputs are positive and the Q outputs are around zero shortly after the traking proess has started. respetively. Both disriminators provide similar results for the CBOC. Comparing with the urves obtained by simulation it an be observed that the results are similar as the bandwidth that is used (18 MHz) is more than enough for reeiving the full CBOC signal (the minimal bandwidth to reeive CBOC signal is 14 MHz). Figure 21 I and Q outputs from the traking loop for the ase of multipath delay of 225m and CBOC pilot hannel The residual ode offset from the traking loop that has been used to ompute the multipath error is similar as for the BPSK(10) signal on Figure 16 and is not shown here, the only differene is that its maximum range variation is instead of 0.25 in the ase of BPSK. The reason is that in CBOC/TMBOC ase, there are 40 samples per hip, giving a minimal shift of 1/40(0.025) in the ode NCO. The PLL disriminator outputs for the ases with and without multipath using Atan2 disriminator for a multipath delay of 225m are shown in Figure 22. Figure 23 Multipath Error Envelopes obtained by measurements for EMLP disriminator and 1/12 spaing Comparing the two figures, the DP disriminator seems to perform slightly better and the graphs are similar to the ones obtained from theoretial simulations. It has to be mentioned that the urves obtained are fitted using the shape-preserving fitting proess. There were fifteen measurement points inluded on the graphs between 0 and 300m of multipath delay. Disontinuity in the results is attributed to the small number of points for the multipath delay and the low number of data sets to be averaged (only 3 sets of data). More sets of data would be needed for more aurate results. Figure 22 PLL disriminator output using for the ase with and without multipath for Atan2 disriminator It an be observed that both ases lead to similar PLL outputs after the loking stabilization. The only differene is that at the beginning of traking the multipath ase provides the higher amplitude variation until the lok is maintained. The final graphs for the multipath error obtained by averaging the error for eah multipath delay point when EMLP and DP disriminators are used are shown in Figure 23 and Figure 24, Figure 24 - Multipath Error Envelopes obtained by measurements for DP disriminator and 1/12 spaing VIII. DISCUSSION Throughout this work, it was shown that the high resolution orrelator does not bring signifiant improvements omparing to the narrow orrelator for AltBOC and BPSK(10). Moreover, to take advantage of narrowing the hip spaing, a higher sampling rate and a wider bandwidth are needed to maintain the stability and linearity of the traking loops. For CBOC and TMBOC, the onsidered bandwidth is more than

10 enough to proess the full energy of the signals. Also in measurements pilot hannels only were onsidered omparing to the simulations where for CBOC the averaged data and pilot autoorrelation peak was used. Overall, the simulationbased results are relatively well mathed by the ones obtained by measurement, and are similar for both disriminators onsidered, although the DP disriminator shows slightly better results. For HRC, a low multipath error an be ahieved but there are multipath delays for whih it s inreased omparing to BOC(1,1). The performane of the traking loops in the presene of multipath was also investigated and their stability and linearity were taken into onsideration. Table 3 summarizes the performane of eah signal with the speified disriminator and gives an overview about their behaviour. What an be observed from this table is that BPSK with narrow orrelator (E-L=1/12), CBOC/TMBOC with HRC (E-L=1/12), and AltBOC with HRC (E-L=1/4) show good multipath mitigation apabilities and traking stability as well. Finally, eah appliation should hoose a proper reeiver design and traking sheme [15] depending on the trade-off between parameters in Table 3. IX. CONSLUSION In this paper, the traking performane and multipath mitigation ability for the new GNSS signals was presented using different disriminator types. Theoretial and analytial results were derived and validated through the measurement results obtained from the build-in reeiver platform. Work performed in this paper provides additional insights on the robustness of these mitigation tehniques, inferred from the realisti simulation set-up onditions. It also disusses the performane omparison of the CBOC, TMBOC and AltBOC signals in different traking ondition, stressing their advantages and limitations, within eah possible appliation area. ACKNOWLEDGMENT The authors are grateful for the finanial support from the Swiss National Siene Foundation ( who supports this work under grant /1 and the European Community's Seventh Framework Program (FP7/ ) under grant agreement n [228339]. REFERENCES [1] Galileo Joint Undertaking, Galileo Open Servie Signal In Spae InterfaeControl Doument, GAL OS SIS ICD, Draft 1, February 2008 [2] INTERFACE SPECIFICATION, IS-GPS-800, Navstar GPS Spae Segment/User Segment L1C Interfaes, Marh [3] J. Avila-Rodriguez, E. Rebeyrol, L.Ries, CBOC An Implementation of MBOC, CNES workshop, Toulouse, [4] J. M. Sleewaegen, W. De Wilde, and M. Hollreiser, Galileo AltBOC Reeiver, in Proeedings of ION GNSS 2004, Rotterdam, Holland, May 16-19, [5] F. Nunes, F. Sousa, J. Leitao, Code Correlation Referene Waveform Multipath Mitigation in MBOC GNSS Reeiver, ION GNSS, [6] M.Irsigler, B. Eissfeller, Comparison of Multipath Mitigation tehniques with Consideration of Future Signal Strutures, 16th International Tehnial Meeting of the Satellite Division of the Institute of Navigation, ION GPS/GNSS 2003, September 9-12, 2003, Portland, Oregon. [7] M.S.Braash, Performane Comparison of Multipath Mitigating Reeiver Arhitetures, IEEE Aerospae Conferene, vol. 3, pp , Big Sky, Mont, USA, Marh [8] [9] [10] M.Paoni, J-A. Rodriguez, T.Panny, G.Hein, Looking for an optimum S- Curve Shaping of the different MBOC implementations, Navigation journal, Vol. 55, No.4, Winter [11] Braash, M. S. (1996), Multipath Effets, published in Global Positioning System: Theory and Appliations, Volume I, Progress in Astronautis and Aeronautis, Volume 163, Amerian Institute of Aeronautis and Astronautis, In., Washington, 1996, hapter 14, p [12] A.J.Van Dierendonk, P.Fenton and T.Ford, Theory and Performane of Narrow Correlator Spaing in a GPS Reeiver, Navigation 39, No. 3, Fall 1992, [13] Mgraw, G. A., Braash, GNSS Multipath Mitigation Using Gated and High Resolution Correlator Conepts, National Tehnial Meeting of the Satellite Division of the Institute of Navigation, ION - NTM99, January 25-27, 1999, San Diego, [14] K.Borre, D. M. Akos, n. Bertelsen, P. Rinder, S. H. Jensen, A Software-Defined GPS and Galileo Reeiver, Birkhauser Boston, [15] A. Jovanovi, C. Mongredien, C. Botteron, Y. Tawk, G. Rohmer, P-A. Farine, Reqirements and Analysis of the Two-Step Traking Tehnique in the Sope of the GAMMA-A Projet, ION ITM 2010, San Diego CA Narrow Correlator High Resolution Correlator E-L spaing 1/12 1/4 1/12 1/4 Multipath Mitigation Traking Stability Multipath Mitigation Traking Stability Multipath Mitigation Traking Stability Multipath Mitigation Traking Stability AltBOC Very 0.17 Very 0.17 Very good 0.29 Good 0.32 good Good BPSK(10) Very 1 Average 1 Very 0.2 Average 0.5 good Good CBOC Good 0.34 Poor 0.34 Good 0.44 Poor 0.5 TMBOC Good 0.34 Poor 0.34 Good 0.44 Poor 0.5 Table 3 Performane omparison of AltBOC, BPSK(10) and CBOC(TMBOC) for NC and HRC