Demonstration of BOC(15, 2.5) acquisition and tracking with a prototype hardware receiver

Transcription

1 Demonstration of BOC(5, 2.5) acquisition and tracking with a prototype hardware receiver Paul Blunt, Ruediger Weiler, Stephen Hodgart, Surrey Space Centre Martin Unwin Surrey Satellite Technology Limited BIOGRAPHY Paul Blunt is a Ph.D. student at the Surrey Space Centre in the University of Surrey. His current research is GNSS space receiver design with emphasis on acquisition and tracking of BOC modulated signals. He received a MEng in Electrical Engineering from the University of Liverpool in Ruediger Weiler is a Ph.D. student at the Surrey Space Centre in the University of Surrey. The focus of his research is the acquisition and tracking of new GNSS signals. He has a diploma in electrical engineering focused on digital signal processing from the Technical University of Aachen in Stephen Hodgart received a Ph.D. degree from the University of Surrey. He has been with the Surrey Space Centre, since its inception and the launch of their first satellite, UoSAT- in 98. He is currently visiting Reader at the University of Surrey. He was responsible for the first practical low-cost active magnetic and passive gravity gradient control attitude system for a low earth orbiting satellite. He has developed novel error-correcting codes, modulation schemes, and signal processing techniques for application on low earth orbiting satellites. In recent years he has been advising the GNSS group of Surrey Satellite Technology in the understanding and development of receivers for the new BOC modulation. Martin Unwin heads the GNSS team at Surrey Satellite Technology Ltd, responsible for spaceborne GNSS receiver design and operation. He holds a BSc from Lancaster University and a Ph.D. from the University of Surrey. INTRODUCTION Surrey Space Centre (SSC) and Surrey Satellite Technology Ltd (SSTL) have over ten years experience in space-borne GNSS receiver design, applications and research. In collaboration SSTL and SSC have developed a prototype Galileo receiver, which has already demonstrated tracking of the GIOVE-A E-B and E-C signals. In this paper we describe the development and testing of SSTL s prototype Galileo receiver for the E-A signal transmitted by the first Galileo test satellite, GIOVE-A. The reception of the Galileo E-A signal poses a significant challenge to GNSS receiver designers. The E-A signal is transmitted with a BOC(5, 2.5)-cos modulation. This modulation has the highest ratio of subcarrier frequency to code chipping rate of any proposed GPS or Galileo signal. Special techniques are required in order to ensure reliability of the receivers tracking this demanding signal. This paper describes the challenges faced by Galileo E-A receivers and the difficulty of ensuring the reliability of receiver tracking in the presence of noise, multipath and signal distortions. During the course of this development, a novel BOC tracking technique was created called the Double Estimation Technique (DET). SSC has shown the DET to have many advantages over existing BOC tracking techniques, such as the bump-jumping algorithm and multiple gate discriminators. This paper details these advantages which are of particular importance for high ratios of BOC subcarrier frequency to code rate, such as the BOC(5, 2.5) signal specified for Galileo E-A. TRACKING BOC(5, 2.5) SIGNALS The BOC(5, 2.5)-cos signal has a code chipping rate of Mcps and a cosine phased BOC subcarrier modulation of frequency 5.345MHz. Therefore, the squarewave subcarrier has 2 half-cycles or subchips in one code chip. The autocorrelation function of a BOC(5, 2.5)-cos signal is shown in Figure for signal both unfiltered and bandlimited to 40MHz. The high-rate subcarrier creates a large array of correlation peaks either side of the central peak.

2 a) b) Figure BOC(5, 2.5) cosine autocorrelation functions: a) unfiltered b) bandlimited to 40MHz Conventional GNSS receivers use time shifted early and late correlations to form an error signal or discriminator to maintain lock on the incoming signal. While this causes no problem for the single peaked PSK modulation, the multiple peaked correlations synonymous with BOC modulation cause the well known problem of BOC tracking ambiguity. A conventional BOC early-minus-late discriminator (see Figure 2) has several stable code loop states which correspond to the point where the discriminator crosses positively through zero. The early-minus-late discriminator for BOC(5, 2.5) has 23 zero crossings, only one of which delivers the correct correlation between the incoming signal and locally generated replica. It is clearly unacceptable to simply allow the receiver remain at a false-lock point as this can result in a ranging error anywhere from 0 to 00 metres. a) b) Figure 2 BOC(5, 2.5) cosine discriminator curves: a) unfiltered b) bandlimited to 40MHz The BOC(5, 2.5) cosine modulation has the potential to deliver extremely precise timing measurements with very small errors due to multipath interferers. The challenge to receiver designers is to preserve these attributes of the modulation while reliably removing the risk of ambiguous tracking states. BUMP JUMPING FOR BOC(5, 2.5) A number of approaches to solving BOC ambiguity have been suggested by the literature. Some techniques such as single sideband processing proposed in [], offer reliable unambiguous tracking at the considerable cost of much degraded tracking precision and performance in the presence of multipath. One method which has been adopted by the current Galileo BOC(5, 2.5) receivers and preserves the precision of the BOC modulation is the bump-jumping (BJ) algorithm, as described in [2]. This algorithm determines whether or not the correct correlation peak is being tracked by comparing the amplitude of the peak currently being tracked to the amplitude of the adjacent peaks. This is achieved 2

3 through the correlation of two additional time shifted replica codes called very early (VE) and very late (VL). These replicas are separated from the prompt (P) replica by a subchip, ±T S. The algorithm is achieved by using three counters, each associated with the VE, P or VL samples. An illustration of a false-lock condition using the BJ algorithm is shown in Figure 3 for a BOC(2 f C, f C ) signal. Here the receiver has locked one subchip away from the central correct correlation peak and can correct its tracking state by recognising that the magnitude of the VL gate is greater than that of the P gate. V L gate VE gate /2 T S Early gate Late gate Prompt gate Figure 3 BOC(2 f C, f C ) false-lock example with bump jumping gates The BJ algorithm can only correct one peak at a time, shifting the tracking point by a subchip each jump and resetting all counters. The BJ algorithm is effectively blind to the number of subchips required to find the valid timing location and can make only one step at a time in sub-chip steps. The acquisition or slip correction time of the BJ algorithm is not only dependant on the discriminator curve and loop setting time but also depends on the time taken determine a false lock. The receiver s VE or VL counter must pass a predetermined counting threshold in order to determine a false lock state. This threshold must be set sufficiently high so that the level of noise on the VE and VL correlations will not cause a false lock to be declared when in fact the receiver is tracking the correct timing location. Therefore, the BJ threshold must be designed for the most severe noise environment the receiver will operate in, with some margin. To determine a BJ threshold we must know the relative amplitude or comparison amplitude between the main BOC correlation peak and its adjacent peaks. This comparison amplitude decreases with an increasing ratio of sub-carrier frequency to code chipping rate. Hence, the BJ threshold for detecting a false lock condition increases for higher rate BOC signals. A method of choosing a BJ threshold in different noise conditions is given in [2] and in [3] where the effect of correlated noise is also considered. In addition, a number of other factors also influence the choice of the BJ threshold; the effect of bandlimiting the signal, the effect of the true location of false-lock points on the comparison amplitude and the effect predicted multipath inference environment must also be taken into account. In order to develop a reliable BJ receiver we take the approach of determining the worst-case conditions in which the receiver may operate and determining a suitable BJ threshold. Assuming an unfiltered BOC(5, 2.5) signal the correlation peaks of the autocorrelation are separated by a relative amplitude of /2. However, our simulations show the effect of bandlimiting the signal to 40MHz (E transmission bandwidth) rounds off the correlation peaks and reduces the relative amplitude between the peaks by as much as 38%. This has a significant effect on the receivers operating the BJ algorithm. BJ operation also assumes that the false-lock points are at integer sub-chip intervals from the central peak. However this is not strictly true. At false lock locations there is an offset between discriminator s zero crossings and the integer sub-chip, which increases with each integer sub-chip due to the different gradients on either side of each correlation peak. Table shows the reduction of BJ comparison amplitude with false-lock locations. The comparison amplitude is shown to reduce by up to 25% if corrections are required from all false-lock locations. T C 3/4 3

4 Table BOC(5, 2.5) cosine BJ comparison amplitude with false-lock points Early false-lock points Zero crossing location (T S ) P correlation amplitude VE correlation amplitude VL correlation amplitude Comparison amplitude VE P Reduction from nominal value ( 0.083& ) 0% 2.4% 6% 25.6% In order to determine the worse-case effect of multipath on the BJ algorithm we simulate a single multipath interferer at half amplitude relative to the direct signal (coefficient of reflection = 0.5) and in-phase with the incoming carrier. The amplitudes of the P, VE and VL gates are then monitored while tracking the central peak of the BOC(5, 2.5) correlation and varying the relative multipath delay. The comparison amplitude is the difference between either the prompt, P and very-early, VE correlations or the prompt and the very-late, VL correlations. Figure 4 shows how the comparison amplitude varies with the relative delay of the multipath interference. 0.5 Comparison amplitude P k P k TH 0. VE k VL k P - VE P - VL Multipath free value k mincr, k mincr, met 25 Multipath delay (m) Figure 4 Effect of multipath on the BJ gates for a BOC(5, 2.5) signal (in-phase carrier, central peak tracking) The results show that when the relative delay of the multipath interference is at integer sub-chip (9.7 m) intervals the comparison amplitude is halved from the nominal value (no multipath interference present). At short multipath delays however, the comparison amplitude is actually 3/2 times greater than the nominal value, actually improving the performance of the BJ peak detection. In this case the multipath is constructive to the BOC correlation at short delay because the multipath is in-phase with the direct carrier signal. Simulating a multipath which is 80 out of phase with the direct signal results in the opposite effect as shown in Figure 5. 4

5 0.5 Comparison amplitude 0. P k VE k P k VL k TH P - VE P - VL Multipath free value k mincr, k mincr, met 5 Multipath delay (m) Figure 5 Effect of multipath on the BJ gates for a BOC(5, 2.5) signal (80 out of phase carrier, central peak tracking) Table 2 shows the effect of taking into account () the influence of bandlimiting, (2) operation across all false lock locations and (3) half-amplitude multipath on the BJ comparison amplitude. If we design a bump-jump threshold for these considerations the effective comparison amplitude available to perform reliable BJ operation reduces to around a quarter of the unfiltered theoretical value. This reduction has a direct impact on the BJ performance. Simulating for a representative minimum carrier to noise density of 24dB-Hz we find a BJ counting threshold of 5 for the assumption of an unfiltered BOC(5, 2.5) signal (5ms integration period). Taking into account the all reliability considerations shown in Table 2 we find the BJ counting threshold to 62 for a carrier to noise density of 24dB-Hz. To give these numbers some meaning we analysed the average time the receiver would take to make a single subchip correction from a false-lock condition. In the unfiltered case, a subchip correction was achieved on average every 600ms, which may be adequate for some applications. However, for a reliable receiver design we increase the BJ threshold to 62 which in turn increases the average subchip correction time to 4800ms. This is clearly a significant amount of time to maintain a false lock condition and is unacceptable for high integrity applications. Table 2 The effect of worse-case considerations on BJ comparison amplitude Reliability considerations Effective BJ Reduction comparison amplitude Unfiltered Bandlimited to 40 MHz % Bandlimited to 40 MHz % Operation across all false-locks Bandlimited to 40 MHz Operation across all false-locks Robust to multipath % The result of setting the BJ threshold too low can result in the receiver jumping away from the correct timing location to a false-lock point. This phenomenon is illustrated in Figure 6 by introducing interference from a half amplitude multipath signal. Here a receiver has been designed to operate down to an input signal carrier to noise density of 24dB- Hz with a corresponding BJ counting threshold of 5. This threshold level is set too low to account for the strong multipath interference and noise induced increments of the VE counter pass the BJ threshold resulting in a jump to a false-lock condition. Through our analysis and simulation described here we have shown conditions under which the BJ algorithm makes for unreliable tracking of the Galileo BOC(5, 2.5) signal. In appears impractical to raise the BJ threshold sufficiently high to deliver reliable tracking in all conditions. Although the considerations described here may be less dramatic for lower rate BOC signals such as BOC(0, 5) and AltBOC(5, 0) the reliability of the BJ algorithm is questionable. The BJ algorithm has been shown to potentially fail under severe signal conditions even for low rate Galileo BOC(, ) signal [4]. The next section describes an alternative approach to BOC tracking which is believed to deliver a robust solution to the BOC tracking ambiguity even in poor signal conditions. 5

6 Timing error (subchips).5 tt k trc k BJ counter values CNTE k CNTL k VE counter VL counter BJ threshold k Loop iterations k Loop iterations a) b) Figure 6 a) Example simulation showing jumps to invalid tracking states b) BJ counter values during simulation (loop bandwidths B DLL = Hz, C/N 0 = 24 db-hz, half-amplitude multipath) 500 THE DOUBLE ESTIMATION TECHNIQUE FOR BOC(5, 2.5) The double estimation technique (DET) for BOC tracking [3] works on an entirely different principle than the bumpjumping algorithm and does not rely on peak discrimination to determine valid tracking states. In a BOC transmission the sub-carrier is necessarily locked to the code sequence. Indeed it is usually seen as part of the code sequence. The time delay of the received signal must obviously be the same in both the code and the sub-carrier. However, there is nothing to stop us multiplying by code and subcarrier receiver replicas with independent time delay estimates. This is the fundamental concept of a receiver operating the DET, to track separately and independently the subcarrier and code delays using different delay estimates. The DET uses three tracking loops as shown in Figure 7. The innermost delay-locked loop (DLL) tracks the code phase of the received signal. The middle sub-carrier locked loop (SLL) tracks the sub-carrier phase of the received signal. The third loop, either a frequency locked loop (FLL) or a phase locked loop (PLL) tracks the carrier frequency and / or phase of the received signal. Figure 8 shows an example acquisition of all three tracking loops. w III u BOC (t) cos(ω 0 t + φ^ (t)) sin(ω 0 t + φ^ (t)) s(t τ^ *) ~ s (t τ^ *) a(t τ^ ) a ~ (t τ^ ) Carrier sub-carrier Code w IQI w IIQ w QII w QQI S I G N A L P R O C E S S I N G w QIQ e φ FLL / PLL e τ SLL e τ DLL Figure 7 DET tracking loop architecture 6

7 3 3 DLL error (sub-chips) SLL error (sub-chips) PLL error (rads) tt k trc k tt k trs k φ C φr k k 500 Loop iterations Figure 8 Example acquisition of the DET BOC receiver (Loop bandwidths B DLL = B SLL = Hz, C/N 0 = 30 db-hz) Under loop operations the DLL delay estimate provides unambiguous tracking with timing jitter equivalent to that of the underlying PSK modulation. The SLL delay estimate delivers tracking precision equivalent to conventional earlylate processing of the BOC modulation although it is ambiguous, simply locking to the nearest integer subchip value. The ambiguity of the SLL estimate is resolved through the noisier but unambiguous DLL estimate. The existence of two estimates of the received signal delay present in the DET significantly improves the integrity of a BOC receiver. There are no false-lock or invalid tracking states when operating the DET. The SLL settles to integer subchip offsets from the correct, valid delay estimate. The delay estimate derived from the DLL allows correction of the SLL delay estimate wherever the SLL has settled to. Therefore, the corrected delay estimate of the DET, is instantaneously corrected for tracking slips of integer subchips and the integrity of the receiver not compromised. An example acquisition of DET with corrected delay estimate is shown in Figure SLL error (sub-chips) DLL error (sub-chips) Corrected delay estimate tt k trs k tt k trc k.5 tt k τb k k 500 Loop iterations Figure 9 Corrected delay estimate of the DET (loop bandwidths B DLL = B SLL = Hz, C/N 0 = 30 db-hz) In simulation the DET has been shown to maintain reliable and robust acquisition and tracking in high noise conditions (24 db-hz) and with strong multipath interference (half-amplitude), which have been shown to cause the BJ algorithm to fail (see Figure 6). The tracking jitter of the corrected DET delay estimate is equivalent to that achieved by the BJ algorithm (standard early-minus-late discriminators), preserving the precision of the BOC modulated signal. In addition multipath mitigation techniques such as narrow correlators and double-delta correlation can be easily implemented into the DET correlator structure. Protection of this technique has been pursued via the UK patent office. Tracking of the 7

8 BOC(, ) modulated Galileo E-B and E-C signals has been demonstrated using the DET with a prototype Galileo receiver [3]. The next section of this paper describes the further development of the prototype receiver to demonstrate tracking with the Galileo BOC(5, 2.5) signal. GIOVE-A PROTOTYPE RECEIVER DEMONSTRATION GIOVE-A, the first demonstrator satellite of the European Galileo system was designed and built by SSTL. The satellite was successfully launched on the 28 th of December 2005 from the Baikonur Cosmodrome in Kazakhstan, via a Starsem Soyuz-Fregat rocket. GIOVE-A, has been transmitting representative Galileo signals since the 2 th of January 2006, The Galileo E-A signal transmitted from the satellite has the BOC(5, 2.5) -cos modulation. The primary payload of GIOVE-A is a fully space qualified signal generator commissioned by ESA. Driven by one of two rubidium atomic clocks this payload is capable of producing representative Galileo signals in three frequency bands (E, E5, E6) although only two bands may be transmitted at any particular time. The space-qualified units were sourced by ESA and are referred to as customer furnished items (CFI). In order to mitigate risk SSTL also supplied a Galileo signal generator based on commercial of-the-self (COTS) components which was flown on the satellite as a backup to the CFI units. This development consists of the navigational message generation unit (NMGU) and the modulator, frequency generator and up-converter unit (MFUU). Both units are driven by a 0.23MHz reference derived from the onboard atomic clocks. Engineering models of the units are also available for bench testing with receivers. a) b) Figure 0 a) SSTL Galileo signal generator b) BOC(5, 2.5) cosine signal In collaboration between SSTL and SSC a prototype Galileo receiver has been developed to provide a hardware demonstration of BOC tracking techniques such as the DET and bump-jumping algorithm. A block diagram of the receiver is shown in Figure. The Galileo E signal is down-converted to an Intermediate Frequency (IF) of 75MHz with a 35MHz bandwidth. The signal is then digitised by a 4-bit Analogue to Digital Converter (ADC) at a sampling frequency of 00MHz. All the correlation, demodulation and processing functions of the receiver are integrated into a single FPGA. The digital automatic gain control re-quantises the signal to 2-bits which is then fed to the acquisition and tracking channels. The 2-bit signal is also recorded by a data-logger attached to a PC for offline analysis. Figure Block diagram of prototype Galileo receiver The receiver uses the Sub-Carrier Cancellation (SCC) technique [5] during the search process to locate the BOC(5, 2.5) signal. This method combines correlations of the incoming signal with both sine and cosine subcarrier replicas to create an approximation of a single PSK-like peak. Figure 2 shows the resulting SCC correlations for both 8

9 unfiltered and 40MHz bandlimited BOC(5, 2.5) cosine signals. Using this technique removes the risk of missing the signal during acquisition due to the nulls of the BOC correlation. The SCC technique reduces the number of trial code delays required by the search process, which in turn reduces the search time of a serial search process or the complexity of FFT search techniques. a) b) Figure 2 SCC correlations for BOC(5, 2.5) cosine signal a) unfiltered, b) 40MHz bandlimited The SSTL laboratory test setup of the prototype Galileo receiver allows generation of BOC(5, 2.5) signals from two sources, the GIOVE-A CFI payload rack and the SSTL MFUU Galileo signal generator. Live GIOVE-A signals are also available through a prototype triple-frequency GNSS antenna from Roke Manor Research. The current GIOVE-A transmissions show a slight asymmetry has been observed in the BOC(5, 2.5) signal, as shown in [6] and [7]. This is expected to be corrected in the near future by pre-distorting signal transmitted by GIOVE-A. However, due to the large bandwidths occupied by many of the future BOC transmissions, avoiding asymmetry in the receiver front-end filters is particularly demanding. Asymmetry in BOC signals can potentially compromise the integrity of the receiver tracking using the bump-jumping algorithm by reducing the effective peak comparison on which its operation depends. Asymmetry in the received signal also causes a bias in the receiver-estimated pseudorange [8] because the slopes on either side of the central correlation peak are not equal. Figure 3 shows correlation plots from data from the prototype Galileo receiver while processing the BOC(5, 2.5) signal from the GIOVE-A CFI payload rack in the laboratory. This signal is generated from payload units (signal generator, high-power amplifier) that are largely representative of those flown on the GIOVE-A satellite and passed through a non-ideal bandlimited RF front end. The asymmetry in the received correlation can clearly be seen to reduce the difference between the central BOC correlation and its adjacent peaks. In fact the central peak of the BOC correlation is only visible through close inspection. The asymmetry is also visible in the SCC search peak although this has no significant effect on the search performance. It should also be noted that the plots shown in Figure 3 include the effects of the prototype receiver s front-end filter which exacerbates the received asymmetry. Currently the prototype receiver has demonstrated successful search using the SCC technique and coarse tracking of the BOC(5, 2.5) signal using the BJ algorithm. However, tracking of the central BOC(5, 2.5) correlation peak is unstable. The reason for this is the receiver clear when we consider the effect of correlation shown in Figure 3 on the BJ algorithm. The receiver cannot determine the central peak of the correlation and the result is continuous tracking jumps. 9

10 a) b) Figure 3 Correlation plots from the prototype Galileo receiver (signal source - GIOVE-A CFI payload rack) a) BOC(5, 2.5) correlation b) SCC peak for the search process In contrast to the BJ algorithm, asymmetry in the received signal is not believed to pose a severe problem a BOC receiver operating the DET. This is because the DET does not rely on peak discrimination and tracks the sub-carrier of the incoming signal independently of the code. The DET has no false-lock conditions because sub-chip offsets are automatically accounted for in the corrected delay estimate. Figure 4 shows a simulated example acquisition of the DET with a distorted asymmetric signal created by introducing a quarter subchip offset between the subcarrier and the code of the received signal. The received code timing delay is set to /4 of a subchip which corresponds to the correct delay estimation of the received signal. The SLL estimate is biased from the correct delay due to the asymmetry. However, the effect of the subcarrier is removed from the DLL loop which therefore converges to the correct timing location tt k trs k 2 tt k trc k tt k τb k k 500 SLL error (sub-chips) DLL error (sub-chips) Corrected error -/4 sub-chip Figure 4 Example acquisition of the DET with asymmetric correlation, The DET receiver can very accurately measure the subcarrier to code offset, therefore the SLL bias can be calibrated and corrected by using the DLL estimate in order to correct for asymmetry in the received signal. The receiver can then operate reliably even with asymmetric signals. The bias in the BJ algorithm can only be resolved if the receiver has the capability of receiving other signals from the same satellite or through a complex calibration campaign. This combined with the risk to BJ receiver integrity proves the DET to be the preferred choice, particularly for BOC transmissions with high ratios of sub-carrier frequency to code rate, such as the BOC(5, 2.5) signal. Our simulation results show that robust tracking of the BOC(5, 2.5) signal with significant RF front-end asymmetry can be achieved without degrading the timing precision or multipath performance of the receiver provided the receiver 0

11 is operating the DET BOC tracking technique. Work in the near future is focussed on implementing the DET in the prototype receiver to validate our simulations with hardware results. CONCLUSION This paper has described the challenges faced by receiver designers attempting to reliably track BOC signals with a high ratio of subcarrier frequency to code chipping rate with particular emphasis on the high rate BOC(5, 2.5) modulated Galileo E-A signal. This signal has the potential to deliver extremely precise timing measurements with very small errors due to multipath interference. However these benefits will only be available to users provided the receiver tracking techniques can reliably remove the risk of ambiguous BOC tracking states. Two tracking techniques are considered, the bump-jumping (BJ) algorithm and the double estimation technique (DET). The BOC ambiguity resolution provided BJ algorithm is shown to be unreliable for BOC(5, 2.5) tracking in conditions of high noise, strong multipath interference and signal distortion. In contrast, the DET provides a robust solution with no false-lock conditions under the same conditions. For high-integrity applications the reliability of BOC ambiguity techniques should be questioned even for low rate BOC signals. Therefore, we hope that the DET will be the preferred choice in future BOC receiver designs. The development of a prototype hardware receiver is described which will provide a demonstration of BOC(5, 2.5) tracking using the DET. Detailed plots of the BOC(5, 2.5) correlations processed by the receiver are shown and there impact on search and tracking techniques is assessed. The issue of signal distortions is addressed and simulations shown detailing the effect on the receiver. Again the DET proves to be the preferred option as it can be easily calibrated for operation with distorted signals. REFERENCES [] J. W. Betz, The Offset Carrier Modulation for GPS Modernisation, Proceedings of ION 999 National Technical Meeting, San Diego, California, January 999 [2] P. Fine, W. Wilson, Tracking Algorithm for GPS Offset Carrier Signals, Proceedings of ION 999 National Technical Meeting, San Diego, California, January 999 [3] P. Blunt, Advanced GNSS receiver design, PhD thesis, University of Surrey, 2007 [4] A. Simsky, J-M Sleewaegen, M. Hollreiser, M. Crisci, Performance Assessment of Galileo Ranging Signals Transmitted by GSTB-V2 Satellites Proceedings of ION GNSS 2006, Fort Worth, Texas, September 2006 [5] P.W. Ward., A Design Technique to Remove the Correlation Ambiguity in Binary Offset Carrier (BOC) Spread Spectrum Signals, ION 59 th Annual Meeting / CIGTF 22 nd Guidance Test Symposium, Albuquerque, New Mexico, June 2003 [6] S. Graf, C. Günther, Analysis of the GIOVE-A L-signals, Proceedings of ION GNSS 2006, Fort Worth, Texas, September 2006 [7] M. Falcone, M. Lugert, M. Malik, M. Crisci, C. Jackson, E. Rooney, M. Trethewey, GIOVE-A In Orbit Testing Results Proceedings of ION GNSS 2006, Fort Worth, Texas, September 2006 [8] M. Rapisarda, P. Angeletti, E. Casini, A simulation framework for the assessment of navigation payload non-idealities, 2nd Workshop on GNSS Signals & Signal Processing - GNSS SIGNALS 2007, Noordwijk, The Netherlands, April 2007