兩種捕獲高靈敏度高動態GPS信號的方法

田　嘉，王　偉，史平彥

(空間電子信息技術研究院，西安　710100)

兩種捕獲高靈敏度高動態GPS信號的方法

田嘉，王偉，史平彥

(空間電子信息技術研究院，西安710100)

摘要：針對相較于地面及低軌全球定位系統(global positioning system，GPS)用戶數量眾多，高軌GPS用戶數量極少，以及傳統捕獲方法在高靈敏度及高動態下無法正常工作的問題，在目前國內外主要側重算法的研究，而基于實際信號的具體實現卻很少的現狀基礎上，提出了2種針對高靈敏度高動態信號的捕獲方法：其中一種在僅增加現場可編程門陣列 (field programmable gate array，FPGA)資源的前提下，可以達到對17 dB·Hz及5 Hz/s的GPS L1C/A信號的成功捕獲；另一種則以增加FPGA資源和捕獲時間為代價，達到對15 dB·Hz及5 Hz/s的GPS L1C/A信號的成功捕獲。分別闡述了方法的原理及實現結構，最后使用Spirent 8000生成實際信號，并對捕獲方法進行了驗證，結果表明這2種捕獲方法都可以成功的捕獲信號，解決高靈敏度高動態下GPS L1C/A信號的捕獲問題。

關鍵詞：捕獲；高靈敏度；高動態；空間應用

0Introduction

In the last few years，although GPS is designed for providing PVT (position，velocity and timing，PVT) for land，marine and air users，there is an increasing trend that space applications can also benefit from GPS.In[1]，there is a projection that based on the GPS constellation，over the next twenty years approximately 60 % of space missions will operate in LEO (which ranges from an altitude of 160 km to about 2 000 km)，35 % of space missions will operate at higher altitudes up to geostationary Earth orbit (altitude of approximately 35 786 km)，and the reminding 5% will be a Cislunar / Interplanetary or HEO missions.

The main reason why there are less space applications on HEO orbit is because the signal received on HEO orbit is so different from the signals of current applications.Compared with on-earth indoor application，the dynamic of HEO orbit is much bigger，with respect to LEO application，the sensitivity of HEO orbit is extremely higher，therefore different acquisition methods are required for the HEO applications.

Within this context，this paper aims to develop an acquisition method that can achieve a good performance on HEO orbit.Section 1 and 2 present the signal analysis and system setup，while section 3 reports the theoretical analysis and Matlab simulation.Finally，FPGA implementation analysis is presented in section 4.

1Signal Analysis

The MTO (moon transfer orbit，MTO) is one of traditional orbit on HEO，whose schematic transfer trajectory is depicted in Fig.1.Focusing on this orbit，the exhaustive analysis of received GPS signals is presented in[2].It analyzes the minimum power required to acquire and track at least 1，2，3 or 4 signals simultaneously from GPS by assuming 0 dBi receiver antenna gain.As we all know that，in order to calculate the correct position，the minimum number of visible satellites is 4.

Fig.1　A lunar transfer trajectory from the Earth

According to[2]，it can be seen that the required minimum signal power to detect at least 4 signals does not exceed -169 dBm，if a few peaks are neglected.It is reasonable to assume 10 dBi receiver antenna gain[3-4]，which means a sensitivity of -159 dBm is needed to ensure the simultaneous detection of at least 4 GPS satellites and then a navigation solution.

2System Setup

The GPS signals are generated by a Spirent GSS8000 simulator，which includes facilities to accommodate the special needs of a space-based receiver testing，including[5]：

● Full account for the double atmosphere effect of signals passing through the atmosphere twice for the GPS satellites located on the far side of the Earth；

● Realistic satellite transmit-antenna patterns；

● Spacecraft models and spacecraft motion models；

● Ability to define trajectory data in real time；

The LNA (low noise amplifier，LNA) and front-end used here are ZRL-2400LN[6]and Fraunhofer Triband front-end[7].The full parameters used in our system are presented in Tab.1.

Tab. 1　Front-end worksheet

According to Friis’s formula[8]shown in Eq.(1)，the effective temperature of entire frontend is 301.5 K，which means the noise power density (N0) is.10×log10(k×Teff)≈-174dBm.

Teff=TA+(F1-1)T0+

(1)

where TAis the effective temperature of the antenna which is 130 K，T0is the ambient frontend temperature which is 290 K，Fiis the noise figure of block and Giis the gain of block.

So the minimum carrier-to-noise ratio is 15 dB·Hz in our case.Besides the characteristics mentioned above，the output signal parameters of the front-end are equally important for post-processing which are shown as follows：

● Analog IF (intermediate frequency，IF)：53.78 MHz；

● Digital IF：12.82 MHz；

● Sampling rate：40.96 MHz；

● Resolution：4 bits；

● Output format：real.

Since the sampling frequency is relatively high，a downsampling (down convert，low-pass filter and decimation) is performed before the acquisition process.After the downsampling，the sampling frequency is 4.096 MHz for L1 signal.

3Theoretical Analysis and Matlab Simulation

The full bit method proposed in[9] is chosen as the acquisition method，which means the coherent integration time is 20 ms and the step between 2 Doppler bins is 25 Hz.In order to achieve a good balance between coherent gain and FPGA resource，4 branches implementation is used.

As mentioned in the abstract，2 versions of acquisition method are presented，one relates to 17 dB·Hz and the other corresponds 15 dB·Hz.So the theoretical analysis will focus on these 2 sensitivity values.The evaluation of the total integration time is based on[8]，where it is stated that to make the acquisition successfully，the SNR (signal-to-noise ratio，SNR) after the correlation should be higher than 14 dB.

From Tab.2，it can be seen that the total integration time is 4.98 s when the sensitivity is 17 dB·Hz，which means the maximum Doppler shift during acquisition is almost 25 Hz (5×4.98).In order to accumulate energy in one frequency bin，the version 1 of acquisition method is shown in Fig.2.This structure could guarantee that one branch can accumulate the energy without so much loss which could lead to the successful acquisition.

Tab. 2　Theoretical analysis of L1 acquisition

For the purpose of proving the performance of the presented method in Fig.2，real signal outputted by frontend is saved and sent to software receiver to check whether the correct acquisition can be achieved.The acquisition result is shown in Fig.3，which indicates that the correlation peak of the correct code phase is much bigger than the threshold which is calculated by all correlation values.Therefore，the test result shows that the version 1 of acquisition method works well in this situation.

Fig.2　Version 1 of acquisition method

Fig.3　Acquisition result of method version 1

Similarly，the total integration time is 12.14 s when sensitivity is 15 dB·Hz，which means the maximum Doppler shift during acquisition is 60.7 Hz.So if version 1 of acquisition method is used in this situation，the energy will accumulate in 3 Doppler bins，and which will induce 3.8 dB loss.

Therefore，in order to deal with this problem，version 2 of acquisition method is presented in Fig.4.It can be seen that compared with Fig.2，Doppler rate aiding module is added in Fig.4，the function of this module is generating local carrier and code with certain Doppler rate and ensure all the energy can be accumulated in one frequency bin.In this case，-5 Hz/s，-3 Hz/s，-1 Hz/s，1 Hz/s，3 Hz/s and 5 Hz/s are chosen by Doppler rate aiding module.With the help of this module，the remaining Doppler rate is small enough which could ensure that the energy can be accumulated properly in the whole integration time.

Fig.4　Version 2 of acquisition method

For the purpose of testing this method，real data outputted by frontend is saved and sent to software receiver.The acquisition result is shown in Fig.5，which indicates that the correlation peak at correct code phase surpass the threshold successfully.Therefore，the version 2 of acquisition method can work well in this situation.

4FPGA implementation analysis

The FPGA board used is the Terasic DE3 System[10]，which embeds an Altera Stratix III FPGA (EP3SE260F1152).The resources available in this FPGA are：

● 135 200 ALM (adaptive logic modules，ALM)，equivalent to 254 400 LE (logic elements，LE)；

● 864 blocks of 9216 bits (M9K)；

● 48 blocks of 147 456 bits (M144K = 16 M9K)；

● 768 18-bit multipliers (DSP(digital signal process，DSP)).

The resources needed for version 1 and 2 of acquisition method are almost same，so the analysis will focus on the version 2 of acquisition method.

According to the models proposed in[11]，the required FPGA resource of each module is presented in Tab.3.

Fig.5　Acquisition result of mode version 2

The implementation details of each module in Tab.3 are：

● FFT (IFFT)：4 096 points，18 bits data input precision，18 bits twiddle precision and streaming I/O data flow；

● The carrier and code generator are based on a NCO (numerically controlled oscillator，NCO)；

● The control module will keep each start of coherent integration of the 4 branches delayed by 5 ms；

● The magnitude is computed using the Robertson approximation；

● The coherent and non-coherent accumulator are memory-based；

Tab.3　Resource required for version 2 of acquisition method

● The peak extraction module calculates the peak value，peak index，mean value and standard deviation value；

● The determination module calculates the SNR and selects the biggest one.

Therefore，the total resources needed for version 2 of acquisition method represent 28 % of the ALM available in the FPGA，44 % of the DSP and 42.9 % of the RAM.

Besides resource，acquisition time is also important which can be calculated as

TA=TI+NFBTFBNaid

(2)

Where TIis the time required for saving the data，NFBis the number of frequency bins to be searched，TFBis the time needed to search one frequency bin and Naidis the aiding times from Doppler rate aiding module，for version 1 of acquisition method，Naidequals 1，while for version 2，Naidequals 6.

According to[4]，[12]，the frequency search space can be decreased to at least ±31.55 Hz with the assistance of INS and orbit filter，which means NFBequals 3 here.TFBcan be got as

(3)

Where fs，L1is the sapling rate after down sampling and fFPGAis the processing frequency of the FPGA which is 204.8 MHz in our implementation.

So the total acquisition time for version 1 of acquisition method is 5.27 s (4.98 +3×(4.98×4.096×106/(204.8×106))×1)，and the total acquisition time for version 2 of acquisition method is 16.51 s (12.14+3×(12.14×4.096×106/(204.8×106))×6).

5Conclusion

The reported study has investigated potential possibility of acquiring GPS L1C/A signals on a very high Earth orbit，in particular for a MTO orbit.The obtained results show that although the signal characteristic on HEO is much worse (high sensitivity and high dynamic) compared with the signals from current applications，it still can be acquired correctly.The innovation points of this paper are as follows：1) the research is based on the real signal which is generated by Spirent 8000；2) the presented acquisition methods work well in the real signal situation；3) the hardware resource of these methods are evaluated and reasonable in real case.Finally，in order to reduce the acquisition time，the coupling system will be studied which could reduce the frequency search space (i.e.GPS+INS (inertial navigation system，INS) or GPS + orbit filter).

References

[1]MILLER J J.Enabling a fully interoperable GNSS space service volume[EB/OL].(2012-06-22)[2015-03-20].http：//www.unoosa.org/pdf/icg/2012/wgB2012/1.pdf.

[2]CAPUANO V，BOTTERON C，TIAN J，et al.GNSS to reach the moon[C]//The International Astronautical Federation.Proceedings of the 65th International Astronautical Congress (65thIAC).Toronto，Canada：The International Astronautical Federation，Inc.，2014：340-351.

[3]TIAN J，WANG Y G，WANG W，et，al.Cross-band aided acquisition on HEO orbit[C]//The International Astronautical Federation.Proceedings of the 65th International Astronautical Congress (65thIAC).Toronto，Canada：The International Astronautical Federation，Inc.，2014：1013-1020.

[4]CAPUANO V，BOTTERON C，WANG Y G，et al，GNSS/INS/Star tracker integrated navigation system for Earth-Moon transfer orbit[C]//The Institute of Navigation.Proceedings of the 27nd International Technical Meeting of the Satellite Division of the Institute of Navigation (ION GNSS 2014).Tempa，USA：The Institute of Navigation，Inc.，2014：211-216.

[5]Anonymous.Choosing a GNSS simulator[EB/OL].(2009-08-15)[2014-09-15].http：//www.spirent.cn/ White-Papers/Positioning/eBooks/Choosing_a_GNSS_Simulator_eBook.

[6]Anonymous.ZRL-2400LN 數據表(PDF)-Mini-Citcuits[EB/OL].(2005-10-11)[2014-09-15].http：//www.alldatasheetcn.com/datasheet-pdf/pdf/94680/MINI/ZRL-2400LN.html.

[7]TIAN J，WANG W，WANG D Q，et，al.Whole MTO orbit acquisition method based on FPGA resource restricted platform[C]//The International Astronautical Federation.Proceedings of the 66th International Astronautical Congress (66thIAC).Jerusalem，Israel：The International Astronautical Federation，Inc.，2015：258-264.

[8]DIGGELEN F V.A-GPS：Assisted GPS，GNSS and SBAS[M].Boston/London：Artech House，2009：345-350.

[9]PSIAKI M L，Block acquisition of weak GPS signals in a software receiver[C]//The Institute of Navigation.Proceedings of the 14nd International Technical Meeting of the Satellite Division of the Institute of Navigation (ION GNSS 2001).Salt Lake City，USA：The Institute of Navigation，Inc.，2001：816-820.

[10]Anonymous.ALTERA DE3 development and education board user manual[EB/OL].(2012-06-04)[2014-09-15].http：//download.csdn.net/detail/xieyunlei/4350978.

[11]LECLERE J，BOTTERON C，FARINE P A.Comparison framework of FPGA-based GNSS signals acquisition architectures[J].IEEE Transactions on Aerospace and Electronic Systems.2013，49(3)：1497-1518.

[12]MUSUMECI L，DOVIS F，SILVA P F，et，al.Design of a very high sensitivity acquisition system for a space GNSS receiver[C]//The Institute of Navigation.Proceedings of the 27nd International Technical Meeting of the Satellite Division of the Institute of Navigation (ION GNSS 2014).Tempa，USA：The Institute of Navigation，Inc.，2014：556-568.

Two acquisition methods of high sensitivity and dynamic GPS signal

TIAN Jia，WANG Wei，SHI Pingyan

(CAST-Xi’an Institute of Space Radio Technology，Xi’an 710100，China)

Abstract：Based on the phenomenon that the number of GPS (global positioning system，GPS) HEO (high earth orbit，HEO) users is much smaller than the number of GPS users on earth and in LEO (low earth orbit，LEO) and the problem that the traditional acquisition method cannot work well in high sensitivity and high dynamic situation，the current papers mainly focus on the algorithm research，but the study of real implementations based on the real signal is rarely mentioned.2 acquisition methods which are suitable for high sensitivity and high dynamic signal were presented in this paper.One of them could achieve 17 dB-Hz and 5 Hz/s with only cost of FPFA (field programmable gate array，FPGA) resources；the other could achieve 15 dB-Hz and 5 Hz/s with the cost of FPGA resources and longer acquisition time.This paper elaborated the principle and the implementation structure of the two acquisition methods，and tested them with real signal generated by Spirent 8000.The result showed that both acquisition methods could acquire the signal successfully，for solving the acquisition problem of high sensitivity and high dynamic GPS L1C/A signal.

Keywords：acquisition；high sensitivity；high dynamic；space application

收稿日期：2015-06-15

第一作者簡介：田嘉(1987—)，男，陜西西安人，碩士，講師，研究方向為空間通信與導航。

中圖分類號：V19；P228

文獻標志碼：A

文章編號：2095-4999(2016)02-0036-07

引文格式：田嘉，王偉，史平彥.兩種捕獲高靈敏度高動態GPS信號的方法[J].導航定位學報，2016，4(2)：36-41,46.(TIAN Jia，WANG Wei，SHI Pingyan.Two acquisition methods of high sensitivity and dynamic GPS signal[J].Journal of Navigation and Positioning，2016，4(2)：36-41,46.)DOI：10.16547/j.cnki.10-1096.20160208.