Unambiguous BOC Acquisition in Galileo Signal

Transcription

1 Unambiguous BO Acquisition in Galileo Signal Wei-Lung Mao, Wei-Yin Zeng, Jyh Sheen, Wei-Ming Wang Department of Electronic Engineering and Graduate of Electro-Optical and Materials Science, National Formosa University, Yunlin ounty 63, Taiwan Abstract Galileo will be Europe s own Global Navigation Satellite System (GNSS), which is aiming to provide highly accurate and guaranteed positioning services. Among several services for separate target groups, Galileo Open Services (OS) are designed for mass-markets, and they will be available worldwide and free of charge for all users. Galileo E1 system has a code period of 4ms which is quadruple that of GPS /A code. In other words, due to the large number of hypotheses in code phase at acquisition stage, a longer searching time or more hardware resource is required and a dual estimate receiver of binary offset carrier (BO) signals for Galileo system applications is described. It is difficult to acquire Galileo signal because of longer code length and the autocorrelation function (AF) of BO modulation. In this paper, we use cyclically shiftand-combine (S) and BPSK-like architectures the impact of unambiguous acquisition algorithms for BO modulation, and that proposed method acquires these satellite signals with hardware complexity reduction. The underlying concept in the method is to modify the code structure and shorten the code period such that the acquisition complexity can be decreased, but the orthogonality of the modified code will not be as good as the original code anymore. In the most cases of this paper, the performance degradation is acceptable when mean acquisition time is used as a performance metric. Simulation results show that our proposed search algorithm can provide better performance with low hardware complexity for acquiring these satellite signals and possible wrong peak selection in the low signal-to-noise ratio (SNR) conditions. Keywords- Acquisition; Ambiguous; BO; BPSK-like; S; Galileo I. INTRODUTION (HEADING 1) Nowadays GNSS (Global Navigation Satellite System) has been widely used in many civil and military applications. In 004, the EU and US make an agreement to establish a common baseline signal BO [4, 8] for the Galileo OS and the modernized civil GPS signal on the L1 frequency. The BO resulting split spectrum signal effectively enables frequency sharing, while providing attributes that include simple implementation, good spectral efficiency, high accuracy, and enhanced multipath resolution. However, BO modulation brings some drawbacks for multiple peaks, which complicates signal acquisition process of miss-detection or wrong peak selection being higher. In order to avoid the ambiguities of the absolute value of AF, the unambiguous BPSK-like algorithm, which can be seen as a superposition of two BPSK modulated signals, is employed to solve this difficulty. The S method is also presented to reduce the search space of code phases and acquisition computation complexities in this paper. There have been few studies on the Galileo satellite system design under distinct acquisition method environ-ments. Several techniques are proposed to solve caused by the BO modulation problems [1, ]. However, the search space is too complicated and time-consuming problem not yet resolved. The only reduce the search complexity to the current references commonly used method is to change the structure of its spreading code, such as the S [3]. Although the orthogonality of the modified code will not be as good as the original code when there is degradation of maximum correlation gain. And modify the spreading code and then through the BO modulation have multiple peaks, which complicate signal acquisition process. With BPSK-like techniques to assist BO for AF signal modulation caused by the ambiguous to change the signal characteristics to become the only single peak. In this use of the Galileo satellite system of two technologies on the issues encountered its processes. Once the acquisition process starts, the Doppler induced frequency offset of the received signal is compensated by a number controlled oscillator (NO) with a preset Doppler bin. After the compensation process for the frequency offset is finished, the original spreading code is cyclically shifted by 1/4 its length for four groups and then four groups cyclically shifted are combined or added to form a S code where the S code is original spreading code by half its length. In this step, the band-pass filtered BO signals with the BPSK-like architecture are correlated with S code stated above. Then, the parallel FFT/IFFT structures are utilized to perform the frequency domain correlation. When the non-coherent integration processing has been accomplished, the maximal peak out of ms search results of code phases is compared to a detection threshold. If the testing statistic has passed the threshold test, the acquisition is finished and can be tried to enter the tracking process. In the simulations, the Galileo signal is sampled at a frequency of 38.19MHz and the signal is down-converted to an IF of 9.548MHz. A SinBO(1,1) signal selected for next generation of GNSS is considered in this work. Three kinds of S length, i.e. one ms, two ms and four ms length of codes are conducted in our experiments to demonstrate the performance of unambiguous acquisition. It is shown that our proposed search algorithm can provide better performance with possible wrong peak selection in the low arrier-to-noise Ratio ISBN:

2 (NR) conditions and have low hardware complexity for acquiring these satellite signals. II. GALILEO SIGNAL MODEL The received Galileo E1B OS signal [7] from a satellite of the Galileo system can be represented as [5, 6] : 0 1 Original code ms 1 ms 1 ms 1 ms ms where is the number of samples in coherent integration; is the coherent integration time; is the sampling period; is the amplitude of the signal; is the navigation data; is the time delay; is the Galileo E1b PRN code; is the sub-carrier of the BO signal; and is the E1b Frequency and the Doppler shift respectively; is the time delay; is the sampling period; is the initial phase of the received signal and is additive white Gaussian noise with zero mean and the variance. The received signal is band-pass filtered, amplified, and down-converted. Due to the down-conversion, the spectrum of the signal is shifted to the intermediate frequency (IF) represented as: A convenient choice is to sample the IF signal with a sampling frequency, where is the front-end bandwidth. In this case, it is easily shown that the noise variance becomes: where is the power spectral density of. III. THE RESEARH SIGNAL AQUISITION METHOD Galileo E1b system has a code period of 4ms which is quadruple that of GPS /A code. In other words, due to the large number of hypotheses in code phase at acquisition stage, a longer searching time or more hardware resource is required. The proposed method acquires these satellite signals with hardware complexity reduction. The underlying concept in the method is to modify the code structure and shorten the code period such that the acquisition complexity can be decreased. However, dividing the spreading code will dilute its orthogonal with respect to the other ones, but we still maintain the orthogonal characteristic as a prerequisite for good. Therefore, we are going to explain and analyze the S in detail. A generalized S code can be constructed as Figure 1. * * 1 * * Sifted-and-ombined ode Assume the received signal Autocorrelation Figure 1. Principle of the cyclically shift-and-combine method The Galileo original spreading code signal (409 chips / 4 ms) were divided into four groups with different code phase of cutting spreading code, and each set of cutting the length of spreading code for the 103 chips / 1 ms. Step one, we will be cutting the spreading code, and order the number is 1,,3,4. The number 1 that the spreading code corresponding to the cutting of the original spreading codes. The number that the spreading code corresponding to the cutting of the original spreading code. The number 3 that the spreading code corresponding to the cutting of the original spreading code. The number 4 that the spreading code corresponding to the cutting of the original spreading code. Step two, we will be cutting marked the spreading code of four groups. And two groups will be divided into new one group. The new groups are made of the number 1,3,34,41. And the each groups composed of four groups corresponding to sub-spreading code. Step three, the four groups of produced by the subspreading code directly for add, have half a head of a new degree of the original spreading code. Here, this new group of spreading codes with code named cyclic shift (yclic Shifted-and-ombined ode), referred to as S code. Using this new group of spreading codes and receive incoming signals for the correlation operation, this method is equivalent to the spreading code for each group of children were associated with the received signal for the operation, after adding; although the basis of this approach will produce two main Peak, but the two peaks will be the impact of the spreading code cutting, while the peak amplitude of the correlation function caused by decay, in addition, the peak and in the resulting phase shift between the actual code, there is a corresponding relationship between the use of this code will be available to assist the judging phase, this relationship is as follows: ISBN:

3 where the ; N = 4; using the above relation, the code phase of the judge, the need to verify the code phase offset four groups (ode Phase Shift), a peak which can be confirmed as the true code phase offset; described as follows: Assuming the original spreading code to, the received signal to, the original spreading code correlation function (Full ode orrelation Function) can be expressed as follows: where the for the modular arithmetic (Module Arithmetic), which is defined as: Positive Integer According to an equation can be found that the correlation function peak code phase shift fall position. S for the proposed method, assuming that the new spreading code (i.e. S code) for the: where the is sub-spreading code, representing the Then the corresponding mathematical correlation function, which can be expressed as follows: the AF of a SinBO(1,1) code has three peaks, not just one, so care must be taken to ensure that the correct one has been found. To obtain a correlation function whose shape is unambiguous, the BPSK-like method uses only one intermediate frequency with a bandwidth including the two principal lobes of the spectrum. The principle of this method is illustrated in Figure. The correlation outputs of the two channels are combined. y IF [n] arrier NO I Q sin cos exp( j f S nts ) exp( jf S nts ) S spreading code R B1 R B R B3 R B4 Figure. Principle of the time domain BPSK-like method A. BPSK-like technology explanation hannel ombining At the output of the correlation, we get the correlation value between the incoming signal and BPSK-like architecture, which can be represented as: where is the S spreading code. The integration and dump function blocks sum up their input data and for I and Q channels, respectively, with each correlation interval equal to (where M is the decimation or down sample factor). Then the outputs of the digital accumulator s become: Among them, the correlation function of the peak falls code phase shift and position, it can be known, S code corresponding to the peak position and the actual code phase shift, there is a relationship in the. Therefore, as long as the use of this relationship to verify four different sets of the corresponding code phase can be asked for the code phase offset. So from the above equation (4) shows, the letter number of possible phase shift position, in the four groups for the possible code phase correlation function, the correlation function from the value of four years. The maximum value out that is the corresponding shift code phase. After calculation, it can be written as: Signal into real and imaginary components: IV. THE BPSK-LIKE METHOD The BO modulations split the signal spectrum into two symmetrical components around the carrier frequency, by multiplying the PRN code with a rectangular sub-carrier. But where is multiplied by architecture; is multiplied by architecture; is the local delay time; is the local Doppler shift; is the local ISBN:

4 phase offset received signal; represents the correlation function between the received BO signal and the local Galileo PRN code modulated by. represents the correlation function between the received BO signal and the local Galileo PRN code modulated by. Noise is divided into real and imaginary components: Finally, divided into real and imaginary parts of the signal that : The final decision variable is obtained as: (4) We use the time domain and frequency domain conversion, it can achieve less computation and simultaneously search multiple code phase process to speed up the acquisition. frequency offset of the received signal is compensated by a number controlled oscillator (NO) with a preset Doppler bin. After the compensation process for the frequency offset is finished, the modified S code is used to replacing the BPSK-like architectures of the local spreading code in the frequency domain processing. When the frequency domain processing has been accomplished, the maximal peak out of ms search results of code phases is compared to a detection threshold. If the testing statistic is less than, the value of the Doppler bin is altered and the correlation/integration process is repeated. Given all Doppler bins are failed in the flow chart, the targeted satellite is considered to be absent, i.e., the acquisition unit is ready for searching next satellite. If the testing statistic has passed the first threshold test, verification by the original spreading code with respect to four code-phase shifts and code phase with maximum peak is chosen. Determine the correlation function value is greater than a threshold that is the corresponding code phase shift and enter the tracking section. Start Band-pass filtering Frequency offset compensation Despreading by the modified S code BPSK-like architectures Frequency domain processing Search the next satellite No All Doppler bins are tested Yes exp( j f nt S ) S S spreading code Fourier Maximal peak > 1 No [n ] y IF Bandpass filter I Fourier Q omplex conjugate Inv Fourier Inv Fourier R B 1 R B Output R B 3 R B 4 Yes Verification by the original spreading code with respect to four code-phase shifts ode phase with maximum peak is chosen hange Doppler bin cos sin omplex conjugate arrier NO Fourier Maximal peak > Yes No exp( jf S nt ) S S spreading code Figure 3. Principle of the frequency domain BPSK-like method B. Acquisition algorithm Search for detection signal and code phase is divided into some judgments of the algorithm with the BPSK-like structure to be integrated. Start acquisition signal, the Doppler induced Tracking start Figure 4. Acquisition flow chart for process ISBN:

5 V. SIMULATION RESULTS In this section, we use MATLAB tool to implement the S method and BPSK-like structure to replace the original E1b Galileo spreading code. The received frequency offset is set to 9.548MHz. Assuming the frequency offset of the received satellite signal is located somewhere between 5 ~+5 khz, and the frequency step during the Doppler search is set to 15 Hz. It is shown in Fig. 6 that the code phase peak is located at the 1554-th chip for different architectures. The AFs of the BPSK-like method seems to be much higher than the performance of the SinBO(1,1) one. The envelope of the AF of SinBO(1,1), where the sidelobes are clearly visible. These sidelobes will cause more challenges to the BO acquisition process, and the BPSK-like method is utilized here to remove the additional peaks to improve the search stage. (Magnitude Highest: ) (a) The AF of BPSK-like architecture Figure 6. omparison of the correlation peak in BPSK-like and SinBO(1,1) architecture (b) The AF at the zero doppler frequency Figure 5. The acquisiiton results of BPSK-like architecture In Figure 5(a), it shows the two-dimensional autocorrelation function from the received Galileo signal and our proposed S method. The integration time is ms and the SNR of all satellites are set as -0 db, which represents an appropriate signal quality. In Figure 5(b), if the integration processing has been accomplished, the maximal peak out of ms search results of code phases is compared to a detection threshold. The Eq. (4) is applied here to determine the true code position between the two peaks in the correlation domain. However, if the maximum of them is larger than the threshold, the code phase and the Doppler frequency are fed into the tracking process, while the acquisition unit restarts to search another satellite signal. (a) NR=5 db-hz (b) NR=30 db-hz ISBN:

6 (c) Probability of detection vs. NR for S method Figure 7. Histogram of 5,000 data samples in BPSK-like method.(a) NR= 5 db-hz; (b) NR= 30 db-hz.(c) Probability of detection vs. NR for S method Figure 7 shows the simulation-based histogram for correct and incorrect bins under different NR environments. A test statistic is calculated in each search window according to the current correlation results. In this simulation, the distribution of the output is based on the chi-square random variable. In Fig. 7(a) (b), the noise only condition is central chi-square distribution and the Galileo signal with noise is non-central chisquare distribution. In Fig. 7(c), it can be observed that S method has a better performance under difference NR. It is shown that the detection probability of S method is good enough for Galileo signal acquisition. Finally, S method will destroy the original spreading code orthogonality, if the SNR less to lead orthogonality even worse, so the peak attenuation effect will result in code phase of the judging error. While there back in the signal code phase search to determine the code phase error is corrected. However, signal attenuation and multi-peak signal to set the threshold mechanism of the phenomenon is more important, so we used BPSK-like structure to lower the threshold value setting difficult. [1] V. Heiries, D. Roviras, L. Ries, V. almettes, Analysis of Non Ambiguous BO Signal Acquisition performance, ION GNSS 17th International Technical Meeting of the Satellite Division, 1-4 Sept. 004 [] Zaixiu Yang, Zhigang Huang, Shengqun Geng, Unambiguous Acquisition Performance Analysis of BO(m,n) Signal /09/ IEEE [3] Pei-Hsueh Lee, Dong-Hong Liu, Wei-Lung Mao, Hen-Wai Tsao, Fan- Ren hang, A Novel Low-omplexity Acquisition Method for Next Generation GNSS Signals [4] Md. Farzan Samad, Effects of MBO Modulation on GNSS Acquisition Stage TAMPERE UNIVERSITY of TEHNOLOGY Department of ommunication Engineering. [5] Kai Borre, Dennis M. Akos, Nicolaj Bertelsen, Peter Rinder, Soren Holdt Jensen, A Software-Defined GPS and Galileo Receiver [6] D. Borio, A statistical theory for GNSS signal acquisition, Ph.D. dissertation, Politecnico di Torino, 008 [7] Galileo Open Service Signal In Space Interface ontrol Document (OS SIS ID Draft 1), 008 European Space Agency / European GNSS Supervisory Authority [8] J. W. Betz. The Offset arrier Modulation for GPS modernization. In Proc. of ION Technical Meeting, June 1999 VI. ONLUSIONS In this paper the performance of unambiguous acquisition algorithms BPSK-like and S for Galileo system signals has been evaluated. The S method can be used to reduce the complexity of search procedures, but also cause attenuation of the orthogonal characteristics of the phase to go through a series of determine. However, through verification of simulation, an appropriate architecture for S method can determine the exact phase of the satellite signal and obtain phase information. In order to avoid the drawbacks of multiple peaks from BO modulation, the BPSK-like search structure is used to yield only single peak. Based on the overall system we proposed, it is aims to accomplish the shortest possible time and do the best quality satellite search, not only for GPS, but covers the EU's Galileo satellite signals. REFERENES ISBN: