A method for code doppler effect elimination in a signal combining process

Abstract

The application relates to the technical field of signal processing, and discloses a code Doppler effect elimination method in a signal combination process. Before signal combination, a time domain cyclic shift is performed on the autocorrelation function before combination, or a frequency domain rotation factor compensation is performed on the FFT transformation of the autocorrelation function, and then IFFT transformation is performed to the time domain. The application solves the problem that the prior art lacks a technical scheme capable of efficiently and accurately eliminating the code Doppler effect.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of signal processing, and particularly relates to a code Doppler effect elimination method in a signal combination process. BACKGROUND

[0002] Global Navigation Satellite System (GNSS) can provide all-weather high-precision position, velocity and time services for terminal receivers on the earth's surface or near space, which has great significance for national security, daily life and industrial production. GNSS navigation and positioning has been widely used in missile guidance, vehicle navigation, precision agriculture, smart city and other fields.

[0003] The mainstream GNSS receiver system block diagram is shown in Figure 1 , which mainly includes four core modules of all-digital radio frequency module (RF), digital front end (DFE), baseband receiver and system software. Among them, the baseband receiver contains two core units of acquisition and tracking, and the system software is responsible for two main functions of position, velocity, time (PVT) solution and system process control. The RF module performs related processing such as down-conversion, filtering, analog-to-digital conversion (ADC) and the like on the satellite signal to obtain a digital signal into the DFE, and the digital signal is further processed by digital filtering and variable rate processing into the baseband receiver. The acquisition unit realizes the coarse synchronization of the received signal and the local pseudo-random (PN) sequence through two-dimensional search of code phase and frequency, that is, to obtain a rough estimate of the PN code phase and carrier frequency offset. Based on the coarse synchronization result of the acquisition unit, the tracking unit contains multiple tracking channels, which can simultaneously complete the accurate synchronization process of the carrier and PN code phase of multi-mode and multi-frequency point satellite received signals, obtain accurate carrier frequency, phase, code phase and other channel state information, and demodulate the bit of the message. The baseband receiver reports all channel state information and message bits, and the system software can perform real-time PVT solution to provide navigation and positioning and timing services for users.

[0004] In the shielding, multipath interference and other harsh environments, the satellite signal reaching the GNSS receiver is very weak, and even below-145dBm. The high-sensitivity receiving algorithm of weak signal has become the research focus of GNSS receiver. Most satellite signals (first acquisition of the current satellite, unable to drag into the tracking unit) must be coarsely synchronized by the acquisition unit to enter the tracking unit for subsequent fine synchronization process. Therefore, high-sensitivity signal acquisition has become one of the most core technologies of GNSS baseband receiver.

[0005] The high sensitivity signal acquisition strategy mainly contains several large research directions such as improving signal processing gain, anti-interference algorithm, auxiliary acquisition and so on. Based on the periodic characteristics of navigation signal PN sequence, the signal-to-noise ratio can be significantly improved by long-time signal accumulation. The signal accumulation method mainly contains three commonly used algorithms: coherent, non-coherent and differential coherent integration. Among them, the signal-to-noise ratio improvement brought by coherent integration increases linearly with the increase of integration length (reference [odc]: O Driscoll C. Performance analysis of the parallel acquisition of weak GPS signals [D]. Cork: National University of Ireland 2007.), but the coherent integration length is limited by the navigation message bit jump; non-coherent integration (reference [pml]: Psiaki M L. Block Acquisition of Weak GPS Signals in a Software Receiver [C]. ION: GPS salt Lake City Ut Sept, 2001: 2838-2850.) and differential coherent integration (reference [zs]: Zarrabizadeh M H, Sousa E S A. Differentially Coherent PN Code Acquisition Receiver for CDMA Systems [J]. IEEE Transactions on Communications. 1997, 45(11): 1456-1465) are not affected by the navigation message bit jump, and can further increase the signal accumulation length and improve the signal processing gain. Affected by the square loss (reference [jby]: James Bao-Yen T. Fundamentals of Gloabal Positioning System Receivers: A Software Approach [M]. 2ed. John Wiley & Sons. Inc, 2005.), frequency offset and other factors, with the increase of the accumulation length, the non-coherent and differential coherent integration are less and less obvious to improve the signal-to-noise ratio, and the accumulation length cannot be increased indefinitely. In view of the problems existing in the three coherent algorithms, the literature [lxg], [zhl] (Zhang Honglun, Basaxiu, Chenjie, et al. Fine frequency estimation of weak GPS signals based on FFT [J]. Journal of Electronics & Information, 2015, 37(9): 2132-2137.) proposes a series of optimization algorithms such as eliminating message bit flip and frequency compensation to compensate for the performance loss of long-time accumulation, and further improves the signal processing gain.

[0006] The high-speed movement (500 m / s and above) of the receiver, the large frequency deviation (equal to or greater than 10 ppm) of the digital compensation crystal oscillator (DCXO) in the low-cost GNSS positioning chip RF module, and the large frequency deviation of the received satellite signal carrier will cause a large frequency deviation (large frequency deviation), which will also cause a non-negligible PN code phase deviation, which is called "code Doppler effect". The larger the carrier frequency deviation, the more significant the PN code phase deviation over time. When the carrier frequency deviation is large enough (greater than 10 kHz), as the signal accumulation length increases, the signal signal-to-noise ratio does not improve, but deteriorates seriously. Whether it is coherent integration, non-coherent integration, differential coherent integration or other types of signal accumulation, or the combination, optimization and deformation of different signal accumulations such as coherent-non-coherent integration, the large frequency deviation will affect the long-time accumulation of the signal and even cause the signal-to-noise ratio to deteriorate. Therefore, for high-sensitivity signal acquisition, the "code Doppler effect" must be eliminated to obtain the effective processing gain of long-time signal accumulation, achieve signal-to-noise ratio improvement, and improve the probability of weak signal acquisition.

[0007] However, the prior art lacks a technical solution that can efficiently and accurately eliminate the "code Doppler effect". Even in a few related technical solutions, there are problems such as high complexity, high resource occupation, and unclear elimination effect. SUMMARY

[0008] To overcome the shortcomings of the prior art, the present application provides a method for eliminating the code Doppler effect in the signal merging process, which solves the problem of the lack of a technical solution that can efficiently and accurately eliminate the "code Doppler effect" in the prior art.

[0009] The technical solution adopted by the present application to solve the above problems is:

[0010] A method for eliminating the code Doppler effect in the signal merging process, before signal merging, the autocorrelation function before merging is time domain circularly shifted, or the FFT transform of the autocorrelation function is compensated by a frequency domain rotation factor, and then IFFT is transformed to the time domain.

[0011] As a preferred technical solution, the expression of time domain circular shift is: Wherein, R m (n-mΔn) represents the circular shift of the mth correlation sequence by mΔn sampling points.

[0012] As a preferred technical solution, the R m (n) is circularly shifted in the time domain by the integer operation of mΔn.

[0013] As a preferred technical solution, m = 1 to 90 PN cycles.

[0014] As a preferred technical solution, when rounding mΔn, the sampling frequency f is increased. d .

[0015] As a preferred technical solution, when rounding mΔn, the sampling frequency f is selected. s , such that mΔn is an integer.

[0016] As a preferred technical solution, f cd =1.023MHz, f s =2.048MHz, N f =2048.

[0017] As a preferred technical solution, the expression for frequency domain rotation factor compensation is as follows:

[0018]

[0019] express FFT transformation, Y m (k) represents R m FFT transform of (n), N represents the frequency domain rotation factor. f is the number of FFT points, and k is the frequency domain index value.

[0020] As a preferred technical solution, signal combining includes coherent, non-coherent, differential coherent, coherent-non-coherent, and coherent-differential coherent.

[0021] Compared with the prior art, the present invention has the following advantages:

[0022] (1) This invention can effectively improve the probability of weak signal acquisition under large frequency deviation, thereby achieving high-sensitivity GNSS signal acquisition in harsh application scenarios (high-speed movement of terminal receiver, excessive frequency drift of low-end GNSS positioning chip RF module DCXO).

[0023] (2) The present invention can effectively eliminate the impact of Doppler code effect on acquisition success rate, especially for received signals with large carrier frequency offset; for weak signals, only the compensated signal can achieve effective improvement in acquisition success rate through coherent + non-coherent or coherent + differential coherent long-time combining strategies. Attached Figure Description

[0024] Figure 1 Block diagram of a mainstream GNSS receiver system;

[0025] Figure 2 Fig. 1 is a schematic diagram of a PN sequence in a received signal r(n);

[0026] Figure 3 Fig. 2 is a schematic diagram of an autocorrelation R(n) of a received signal r(n) and a local reference PN sequence;

[0027] Figure 4 Fig. 3 is a schematic diagram of a PN sequence autocorrelation R(n) coherent combining process; m

[0028] Figure 5 Fig. 4 is a schematic diagram of a general signal acquisition flowchart for code Doppler effect elimination;

[0029] Figure 6 Fig. 5 is a schematic diagram of a half-bit method coherent + non-coherent frequency domain parallel signal acquisition flowchart for frequency domain rotation factor compensation;

[0030] Figure 7 Fig. 6 is a schematic diagram of a capture success rate P v.s. non-coherent combining number M. DETAILED DESCRIPTION

[0031] The application will be further described below in conjunction with embodiments and drawings, but the embodiments of the application are not limited thereto.

[0032] Embodiment 1

[0033] As shown in the drawings, in order to overcome the "code Doppler effect" caused by the carrier frequency offset of a received GNSS signal and effectively obtain the signal-to-noise ratio gain caused by signal combining, the application provides a code Doppler effect elimination strategy in a signal combining process, which can effectively improve the weak signal capture probability under large frequency offset, thereby realizing high-sensitivity GNSS signal acquisition in a harsh application scenario (terminal receiver high-speed movement, large DCXO frequency drift of a low-end GNSS positioning chip RF module). Figures 1 to 7 1. Code Doppler effect:

[0034] For a GNSS received signal r(n), the carrier frequency is f c , the PN code rate is f cd , the carrier frequency offset caused by receiver movement and DCXO drift is f d , and the PN chip offset rate caused by code Doppler effect is f d,cd . The relationship between f d,cd , f c , f cd , and f d is shown as follows:

[0035]

[0036]

[0037] ​​The PN sequence in the received signal r(n) is shown in Fig. 1(a) as Figure 2 For more intuitive representation of the PN chip offset introduced by code Doppler effect, other related parameters such as navigation message bit modulation, carrier, noise, etc. are omitted. As shown in Fig. 1(b), one PN code period T cd with code length L, each small block represents one chip, and the number represents the chip index i (0~L-1); the received signal sampling rate is f s N is the total number of sampling points in one PN period, and the sampling point index n=0 is the starting point of the PN code. In theory, the total number of sampling points N in one PN period is a fixed value, and the starting point of the PN code in the mth period is mN. Considering the code Doppler effect, N is actually variable.

[0038] As can be seen from formula (1), the greater the carrier frequency offset, the greater the chip offset rate. The autocorrelation function R(n) obtained by correlating the received signal r(n) with the locally generated PN reference sequence is shown in Fig. 2(a). Similarly, in order to intuitively represent the code Doppler effect, only the autocorrelation of the PN sequence after the chip offset in the received signal r(n) and the local reference PN sequence is considered. As can be seen from the definition of the autocorrelation function, the peak value of the autocorrelation function is also shifted by Δn sampling points every PN code period. Figure 3

[0039] 2. Signal accumulation and merging:

[0040] First, three commonly used accumulation and merging methods in GNSS signal acquisition, coherent, non-coherent, and differential coherent merging, are given, and the mathematical representations are shown in formulas (2)-(4) respectively.

[0041]

[0042]

[0043]

[0044] where R m (n) = R(m*N+n) represents the autocorrelation function of the PN sequence in the mth period; Z coh (n), Z n-coh (n), and Z d-coh (n) represent the results of coherent, non-coherent, and differential coherent merging of the PN sequence autocorrelation function R(n) respectively; n=0~N-1, and M is the number of merging.

[0045] Assuming that there is no message bit flip, taking coherent merging as an example, the influence of code Doppler effect on signal merging processing is discussed. Based on formula (2), the coherent merging process of the PN sequence autocorrelation function R(n) is shown in Fig. 3(a). Figure 4 ​The thick dashed curve represents the waveform of the m-th (0 ~ M) autocorrelation function R(n) right-shifted by mN sampling points, i.e., the signal before coherent combination; the thick solid curve represents the M times coherent combination result Z coh (n) (only the trend is shown). After the periodically appearing PN sequence autocorrelation function R(n) is shifted, the peak value is located within 0 ~ MΔn sampling points. Due to the code Doppler effect (Δn≠0), the M PN sequences do not form complete signal energy accumulation after coherent combination, and the actual signal-to-noise ratio cannot be improved by M times.

[0046] The document [wjn] (Wang J, GNSS Acquisition Technology in Weak Signal Environment, Northwest University of Technology, 2019) simply analyzes the code Doppler effect and gives a theoretically feasible solution. The acquisition frequency search range is divided into N r intervals, and N r code rate variable local reference PN sequences are generated according to the intermediate frequency of each interval to compensate for the PN code rate change in the received signal r(n) caused by the Doppler code effect. For actual receiver applications, this method will increase the signal acquisition calculation complexity to N r times the original, especially for large Doppler frequency search range, such complexity increase is unacceptable. In addition, considering the chip precision problem, the generation of variable rate local reference PN sequence requires a high-speed sampling clock, which will additionally increase the hardware and calculation complexity. In summary, the above analysis shows that this solution is only theoretical and cannot be applied to actual GNSS receivers.

[0047] 3. Code Doppler elimination strategy: time domain cyclic shift

[0048] In order to obtain the energy accumulation of M PN sequences as much as possible, the most intuitive method to eliminate the code Doppler effect is to perform time domain cyclic shift on the autocorrelation function R m (n) before combination, which is mathematically expressed as follows:

[0049]

[0050] In most practical applications, whether it is a large drift DCXO or high-speed movement, it can be ensured that f d <<1MHz. For a smaller sampling rate f s , according to formula (1), Δn is a small decimal number, so most of the cyclic shift sampling point number mΔn is also a decimal number. At this time, only the simple rounding operation can be performed on mΔn to obtain u, and the R m (n) time domain u sampling point cyclic shift is realized.

[0051] For smaller Δn, the rounding operation will introduce larger error. Therefore, by rounding operation of mΔn, then time domain cyclic shift is performed on R m (n), and then long time coherent, non-coherent, differential coherent combination, the influence of code Doppler effect on combination gain can be weakened to some extent, but this influence cannot be completely eliminated.

[0052] Based on signal processing common sense, by increasing sampling frequency f d , the time domain resolution of received signal r(n) and R m (n) can be improved. In combination with formula (2), the larger f s , the larger Δn, and the more accurate rounding operation of mΔn. A suitable high sampling frequency f s may be selected to make Δn≈1, that is, mΔn is an integer, to ensure the accuracy of time domain cyclic shift of R m (n), so as to approximately completely eliminate the influence of code Doppler effect on combination gain. However, increasing sampling rate f s will bring a fold increase in signal processing complexity. Whether it is a software or hardware receiver, limited by computing resources, a suitable sampling rate is generally selected under the premise of meeting the sampling theorem, and the sampling rate f s cannot be infinitely increased to meet the accuracy of time domain cyclic shift of R m (n).

[0053] Therefore, for actual receiver applications, the time domain cyclic shift strategy can weaken the influence of Doppler code effect on long time combination gain to some extent, and cannot well eliminate this influence.

[0054] 4. Code Doppler elimination strategy two: frequency domain rotation factor compensation

[0055] Based on the FFT transformation characteristics of digital signal processing: time domain cyclic shift corresponds to frequency domain multiplication of rotation factor. From formula (7), it can be obtained that:

[0056]

[0057] wherein, and Y m (k) represent FFT transformation of r (n) and R m (n) respectively,

[0058] is the frequency domain rotation factor, and N f is the FFT point number.

[0059] Therefore, by performing FFT transformation on R m (n), Y m(k) Perform frequency domain rotation factor compensation, then do IFFT transformation to time domain, and finally perform coherent, non-coherent, differential coherent and other merging operations.

[0060] The frequency domain rotation factor compensation algorithm does not require mΔn to be an integer, has no precision loss, can realize complete elimination of Doppler code effect instead of approximate elimination, and does not need to change the receiver sampling rate f s and any system parameters.

[0061] The frequency domain rotation factor compensation is an operation based on FFT transformation, and is therefore more suitable for a frequency domain parallel GNSS acquisition algorithm.

[0062] The present application takes a GPS L1 C / A signal as an example, f cd = 1.023MHz, the sampling rate f s = 2.048MHz, the FFT point number N f = 2048; the GNSS signal is generated by a GNS8332 multi-constellation navigation signal simulator, signal acquisition is based on a classical half-bit algorithm, combined with coherent and non-coherent merging, to improve signal acquisition sensitivity, and the specific acquisition process is as shown in Figure 6 The preset acquisition threshold Z T needs to balance the false alarm and missed detection probabilities, and can be filtered twice by setting a small threshold and using parallel calculation of tracking channels. When the detection decision quantity is greater than the preset acquisition threshold, the corresponding tracking channel can be fed back, but only the satellite signal that is stably tracked is considered to be successfully acquired.

[0063] Based on the half-bit algorithm, the coherent + non-coherent signal acquisition is simulated to prove that the frequency domain rotation factor compensation algorithm can effectively eliminate the influence of code Doppler effect on long-time signal merging. The following will comprehensively describe the two aspects of different carrier frequency offsets f d and different non-coherent merging numbers M.

[0064] Large carrier drift f d , different non-coherent merging numbers:

[0065] In w times of acquisition, x times of acquisition threshold value and stable tracking decoding, then P = x / w is the effective acquisition success rate. The simulator GNS8332 outputs a signal power of -145dBm, sets a large carrier frequency offset (center frequency point 16KHz), sets the non-coherent merging number M to be 1, 5, 10, 15 in turn, acquires 1000 times for each M value, counts the number of effective acquisition successes, and the acquisition success rate P changes with M, as shown in Figure 7As shown in the figure. For reference comparison, the crosshairs in the figure represent the acquisition success rate without any processing; the squares represent the acquisition success rate after introducing frequency domain rotation factor compensation.

[0066] Depend on Figure 7 It is known that, under the same signal conditions (signal-to-noise ratio), for signal acquisition without any processing, the acquisition success rate P decreases instead of improving when the number of combining operations M increases to a certain extent. However, after introducing frequency domain rotation factor compensation, the acquisition success rate P gradually increases with the increase of the number of combining operations M. Therefore, for weak signal acquisition, the code Doppler effect must be eliminated before long-term combining to effectively improve the signal-to-noise ratio by increasing the number of combining operations M, thereby improving the acquisition success rate, especially under conditions of large carrier frequency offset (above 10kHz).

[0067] in conclusion:

[0068] By introducing a frequency domain rotation factor compensation algorithm, the impact of Doppler code effects on acquisition success rate can be effectively eliminated, especially for received signals with large carrier frequency offsets. For weak signals, only the compensated signal can achieve an effective improvement in acquisition success rate through a coherent + incoherent long-time combining strategy. The above conclusions apply to any form of long-time coherent combining, including but not limited to coherent, incoherent, differential coherent combining, and their various variations and combinations.

[0069] Example 2

[0070] like Figures 1 to 7 As shown, as a further optimization of Embodiment 1, this embodiment also includes the following technical features based on Embodiment 1:

[0071] Introducing a Doppler code effect cancellation strategy, the general GNSS receiver signal acquisition process is as follows: Figure 5 As shown, it includes core modules such as frequency offset compensation, coherent combining, correlation operations, frequency domain (time domain) compensation, and long-term combining. The correlation operations can be implemented using a time domain correlator or a frequency domain cyclic FFT / IFFT. The frequency domain (time domain) compensation module must precede the long-term combining module to implement R... m (n) Frequency domain rotation factor compensation (time domain cyclic shift) operation eliminates the influence of Doppler code effect on long-time combining gain, applicable to all long-time combining strategies including but not limited to coherent, incoherent, and differential signal combining strategies. Based on the long-time combining result Z(n), a capture decision quantity is generated, typically the peak value, peak-to-average power ratio, and their combined deformations; the threshold decision compares the capture decision quantity with the preset capture threshold and determines whether to enter the tracking channel or re-capture.

[0072] Taking GPS L1 C / A signal as an example, based on the classic half-bit algorithm

[48] , the frequency domain cyclic FFT / IFFT parallel acquisition algorithm is adopted, combined with coherent and non-coherent combination, the signal acquisition sensitivity is improved, the system flow is as shown in Figure 6 The specific operation steps are as follows:

[0073] 1. Generate the local reference PN sequence corresponding to the current acquisition satellite pseudo-random sequence number (PRN), and get the frequency domain reference sequence V(k) after FFT transformation;

[0074] 2. Frequency offset compensation is performed on the intermediate frequency signal r(n), that is, sequence point multiplication operation

[0075] 3. The received signal r(n) is grouped in 10ms length, and the grouping is segmented and accumulated according to 1 PN period (1ms), and the combined signal is numbered 0-2M-1 in turn;

[0076] 4. The combined signals numbered 2m and 2m+1 are respectively subjected to FFT transformation to obtain a series of frequency domain sequences X 2m (k) and X 2m+1 (k);

[0077] 5. The frequency domain sequences X 2m (k) and X 2m+1 (k) are respectively subjected to point multiplication operation with the local reference frequency domain sequence V(k) and the frequency domain rotation factor compensation sequence, that is

[0078] 6. The frequency domain sequences after point multiplication are subjected to IFFT transformation to obtain the compensated time domain correlation sequences, that is

[0079] 7. The time domain correlation sequences are respectively subjected to M times of non-coherent combination according to the odd and even numbers, that is

[0080] 8. The MAX module respectively calculates the peak-to-average ratio of the non-coherent combination results of the odd and even sequences out , and selects the larger peak-to-average ratio as the detection decision quantity Z out .

[0081] 9. The detection decision quantity Z out is compared with the preset acquisition threshold Z T , and tracking or re-acquisition decision is made.

[0082] After step 1, two groups (numbered 2m, 2m+1) of adjacent 10ms must have one group without bit flipping, and the 1ms combined signal accumulated in this group for 10ms can obtain 10 times SNR improvement compared with 1ms signal. Then the correlation operation result is compensated by frequency domain rotation factor, and then long time non-coherent combination is performed, regardless of the size of the received signal carrier frequency offset, long time combination can obtain effective long time combination SNR improvement.

[0083] As described above, the application can be better implemented.

[0084] All features disclosed in the above embodiments, or all steps in the disclosed methods or processes, can be combined and / or extended, replaced, or modified, in any manner, except for mutually exclusive features and / or steps.

[0085] The above description is only the preferred embodiment of the application, and does not limit the application in any form. According to the technical essence of the application, any simple modification, equivalent replacement, and improvement of the above embodiments, within the spirit and principles of the application, are still within the protection scope of the technical scheme of the application.

Claims

1. A method for code Doppler effect elimination in a signal combining process, characterized by, Before signal combination, the autocorrelation function before combination is time domain cyclically shifted, or, after frequency domain rotation factor compensation of the FFT transform of the autocorrelation function, IFFT transform is made to the time domain; wherein, the autocorrelation function is the autocorrelation of GNSS receiving signal and local reference PN sequence; signal combination is the combination of the autocorrelation function; The expression of time domain cyclic shift is: wherein, The autocorrelation function after time domain cyclic shift is represented by R m (n-mΔn) represents that the mth correlation sequence is cyclic shifted by mΔn sampling points.

2. The method of claim 1, wherein, By rounding operation of mΔn, R m (n) is circularly shifted in time domain.

3. A method for code Doppler effect elimination in a signal combination process according to claim 2, characterized in that, m = 1 ~ 90 PN periods.

4. A method for code Doppler effect elimination in a signal combination process according to claim 3, characterized in that, mΔn rounded, by increasing the sampling frequency f s to improve the time-domain resolution of the received signal to the local reference PN sequence.

5. A method for code Doppler effect elimination in a signal combination process according to claim 4, characterized in that, mΔn is rounded to an integer, the sampling frequency f is selected s so that mΔn is an integer.

6. A method for code Doppler effect elimination in a signal combination process according to claim 5, characterized in that, f cd = 1.023 MHz, f s = 2.048 MHz, N f = 2048, where f cd is the PN code rate, N f is the number of FFT points.

7. A method for code Doppler effect elimination in a signal combination process according to any one of claims 1 to 6, characterized in that, The expression of the frequency domain rotation factor compensation is: denotes the FFT transform of Y m (k) denotes the R m (n) the FFT transform of denotes the frequency domain twiddle factor, N f is the number of FFT points and k is the frequency domain index value.

Images