Array anti-satellite navigation signal multipath method based on matrix reconstruction algorithm
阅读说明：本技术 基于矩阵重构算法的阵列抗卫星导航信号多径的方法 (Array anti-satellite navigation signal multipath method based on matrix reconstruction algorithm ) 是由 张润萍 任永吉 邱丽原 李蔚 于 2020-05-07 设计创作，主要内容包括：本发明公开了一种基于矩阵重构算法的阵列抗卫星导航信号多径的方法,属于卫星导航定位领域,通过矩阵重构处理算法,使得湮没于噪声下的卫星导航信号被处理后可以直接应用于阵列信号理论,将阵列采集系统接收到的信号,通过矩阵重构、来波方向估计和波束形成加权处理,形成主波束对准卫星直达信号并使零陷位置对准多径信号的天线阵列方向图,达到增强卫星直达信号、抑制多径的目的。本发明不需要惯导辅助确定阵列平台姿态、改进的波束形成准则计算量小,降低了阵列抗多径算法的计算量,并且对码片延迟较小的难以去除的多径具有较好的抑制效果。(The invention discloses an array anti-satellite navigation signal multipath method based on a matrix reconstruction algorithm, which belongs to the field of satellite navigation positioning. The invention does not need inertial navigation to assist in determining the attitude of the array platform, has small calculation amount of the improved beam forming criterion, reduces the calculation amount of the anti-multipath algorithm of the array, and has better inhibiting effect on multipath which has smaller chip delay and is difficult to remove.)
1. A matrix reconstruction algorithm-based method for resisting multipath of satellite navigation signals by an array is characterized by comprising the following steps:
firstly, acquiring and storing satellite navigation signals by adopting a multi-antenna receiver, carrying out coarse acquisition on the satellite navigation signals, and decomposing satellite navigation signals which may be contained in the satellite navigation signals;
step two, sequentially selecting one satellite navigation signal possibly contained in the satellite navigation signals and determining a local pseudo-random code pn of the satellite navigation signals1(t), resampling the stored signal by adopting a preprocessing algorithm, and specifically comprising the following substeps:
(1) respectively carrying out precise capture operation on the stored multi-antenna stored data to obtain the estimated values f of the Doppler frequency and the code phase of each path of antenna signal1,f2,…fN,c1,c2,…cNWherein N is the number of antennas;
(2) clustering the estimated values to obtain the Doppler frequency f of the reference array elementdAnd code phase cdAn estimated value;
(3) reconstructing the pseudo-random code into a new local code pn according to the estimated Doppler and the code phase valuec(t)；
(4) Local code pn to be reconstructedc(t) respectively carrying out correlation operation with signals stored by the multi-antenna receiver to obtain correlated data r1(t),r2(t),…rN(t)；
(5) Re-sampling the correlated data to obtain a self-correlated sampling result r1(T),r2(T),…rN(T) forming a new input signal matrix Y, wherein T is the sampling period of the autocorrelation function;
thirdly, estimating the incoming wave direction of the preprocessed input signal matrix to obtain the azimuth angle and the pitch angle of the satellite navigation signal and the multipath signal;
fourthly, according to the estimation result of the incoming wave direction, beam forming is carried out on the multi-antenna storage signal, and an intermediate frequency signal of the satellite navigation signal after multipath removal is obtained;
and step five, sending the signals after the beam forming into a software receiver for tracking and resolving.
And repeating the second step, the third step, the fourth step and the fifth step until all the satellites obtained by coarse acquisition are tracked in the software receiver, and positioning and resolving according to the tracking results of a plurality of satellites.
2. The method for resisting the multipath of the satellite navigation signal based on the array of the matrix reconstruction algorithm according to claim 1,
the multi-antenna receiver comprises an antenna receiving array, a down-conversion acquisition module, a data acquisition module and a storage module.
3. The method for resisting the multipath of the satellite navigation signal based on the array of the matrix reconstruction algorithm according to claim 1,
and step three, the incoming wave direction estimation adopts a Toeplitz-MUSIC algorithm and is used for analyzing the incoming wave directions of the multipath signals and the satellite navigation signals.
4. The method for resisting the multipath of the satellite navigation signal based on the array of the matrix reconstruction algorithm according to the claim 1, characterized in that:
and step four, the beam forming adopts a linear constraint minimum variance criterion to obtain the weighted weight of the array.
5. The method for resisting the multipath of the satellite navigation signal based on the array of the matrix reconstruction algorithm according to the claim 1, characterized in that:
and the sampling period T in the second step is a pseudo code repetition period of the satellite navigation signal.
6. The method for resisting the multipath of the satellite navigation signal based on the array of the matrix reconstruction algorithm according to the claim 2, characterized in that:
the antenna receiving array is arranged in a way that the distance between each array element is less than one half of the wavelength of the radio frequency signal, and baffles are designed among the array elements.
Technical Field
The invention belongs to the field of satellite navigation positioning, in particular to the field of satellite navigation anti-multipath, and particularly relates to an array anti-multipath algorithm for matrix reconstruction of satellite navigation signals
Background
The application range of the satellite navigation positioning system is wider and wider, the requirement of a user on positioning accuracy is higher and higher, and researches on various error sources of the satellite navigation system and methods and technologies for weakening the influence of the error sources on a receiver are deeper and deeper. At present, the error sources affecting the positioning accuracy mainly include satellite clock error and orbit error, ionosphere delay, troposphere delay, interference, multipath and the like. The errors such as satellite clock error, orbit error, ionosphere delay, troposphere delay and the like have space-time correlation and can be corrected by adopting a differential positioning technology; however, multipath effects do not have space-time correlation, and cannot be corrected by using a difference technique, which is a major bottleneck for improving positioning accuracy and reliability for high-precision users. In order to eliminate the influence of multipath, a method for estimating and suppressing the characteristics of multipath signals becomes a hot point of research in the field of satellite navigation.
The existing multipath suppression technologies are mainly classified into three types, namely a loop suppression algorithm, a data post-processing algorithm and a spatial suppression algorithm, and mainly include a narrow correlation technology, an early-late slope (ELS) algorithm, a Strobe correlator, a Pulse Aperture Correlator (PAC) technology, a delay locked loop (MEDLL) technology and the like. But they all have their own short plates: the loop suppression algorithm has obvious effect under the condition that the time delay difference between a signal and the multipath is long, and the multipath with short time delay (less than 0.01 code sheet) is invalid; a large amount of mathematical operations are needed for a data post-processing algorithm to be used as a support, so that the statistical satellite operation period is long, and the problems of overlarge time consumption and complex calculation in the resolving process exist; the commonly used spatial domain suppression method, single antenna suppression technology, such as antenna choke, High Zenith Antenna (HZA) and the like, has limited suppression on the angle of the received signal of the antenna, and can only partially suppress the multipath signals arriving. Therefore, more and more scholars at home and abroad transfer the research center of gravity to the multipath resisting algorithm of the multi-antenna adaptive array antenna technology.
The anti-multipath technology of the Satellite Navigation signal (GNSS) array antenna designed in China at present can complete the anti-multipath of beam forming by providing prior information such as Satellite direction, platform attitude and the like. The algorithm can sacrifice useful array information, and the iterative processing has large calculation amount and high requirement on the carrier measurement precision. The method for achieving multipath resistance of the GNSS signals without using prior information becomes the problem that multipath resistance of the satellite navigation signal array needs to be solved urgently at present.
Disclosure of Invention
The invention aims to solve the problems, provides a matrix reconstruction GNSS signal array anti-multipath algorithm, and can achieve the aim of array anti-multipath without introducing prior information.
The technical scheme of the invention is as follows:
the method comprises the steps of firstly, acquiring and storing satellite navigation signals by adopting a multi-antenna receiver, wherein the multi-antenna receiver comprises an antenna receiving array, a down-conversion acquisition module, a data acquisition module and a storage module, and coarsely acquiring the satellite navigation signals to decompose satellite navigation signals which may be contained.
The multi-array-element antenna array consists of 16 four-frequency measurement type antennas suitable for navigation signals, the platform can receive frequencies corresponding to GPS/L1 and Beidou B1, B2 and B3, and the antennas support receiving signals with a pitch angle of 0-90 degrees and an azimuth angle of 0-360 degrees.
The down-conversion acquisition module takes a radio frequency chip of a second-generation Beidou RNSS regional signal receiver as a core and can receive signals of frequency points B1, B2, B3, GPS L1, GLONASS L1 and L2 (only one radio frequency channel is arranged in the chip, and one frequency point is selected to work through an SPI interface and signals S0 and S1 selection). The signal enters the chip through the radio frequency input pin, and after being processed by two-stage low noise amplifier (LNA1 and LNA2), down-conversion, filtering and the like, the signal outputs an analog or digital intermediate frequency signal for the baseband chip to use.
The data acquisition module adopts a PCIE X8 XC7K325T development board, the board card is based on an FPGAXC7K325T-2FFG900 chip of Xilinx company, pin _ to _ pin is compatible with the FPGAXC7K410T-2FFG900, PCIeX8 and 64bitDDR3 capacity 2GByte are supported, and the HPC-FMC connector supports various interface inputs.
The storage device adopted by the storage module is a solid state disk 850PRO, 1TB produced by Samsung corporation, the SSD uses a SATA3.0 interface, the theoretical bandwidth of a bus can reach 6Gbps, and the main control chip adopts a Samsung 3-core MEX. The theoretical continuous reading speed can reach 550MB/s, and the continuous writing speed can reach 520 MB/s.
Step two, sequentially selecting one satellite navigation signal possibly contained in the satellite navigation signals and determining a local pseudo-random code pn of the satellite navigation signals1And (t) resampling the storage signal by adopting a preprocessing algorithm.
The preprocessing algorithm is used for performing matrix reconstruction on each path of autocorrelation capture signals received by an array, improving the GNSS signal power to meet the requirement that the applicable signals of the array processing algorithm are generally higher than 0dB, and keeping the superposition information of the phase delay, amplitude attenuation and propagation delay of multipath signals and direct-view signals without loss, and the method specifically comprises the following steps:
(1) respectively carrying out precise capture operation on the stored multi-antenna stored data to obtain the estimated values f of the Doppler frequency and the code phase of each path of antenna signal1,f2,…fN,c1,c2,…cNWherein N is the number of antennas;
(2) clustering the estimated values to obtain the Doppler frequency f of the reference array elementdAnd code phase cdAn estimated value;
(3) reconstructing the pseudo-random code into a new local code pn according to the estimated Doppler and the code phase valuec(t)；
(4) Local code pn to be reconstructedc(t) respectively carrying out correlation operation with signals stored by the multi-antenna receiver to obtain correlated data r1(t),r2(t),…rN(t)；
(5) Re-sampling the correlated data to obtain a self-correlated sampling result r1(T),r2(T),…rN(T) forming a new input signal matrix Y, where T is the sampling period of the autocorrelation function.
And step three, because the vibration directions and the vibration frequencies of the multipath and direct-view signals are the same and the phase difference is kept constant, the multipath and direct-view signals have coherence. When the incoming wave Direction estimation (DOA) is performed on the input multipath signal and the direct view satellite navigation signal, decorrelation is required.
The incoming wave direction estimation algorithm of decoherence uses a Toeplitz-MUSIC algorithm, and comprises the following specific steps:
(1) calculating an autocorrelation matrix of a preprocessed signalWherein m represents the array element number, ym(Ti) Is composed ofThe superscript H of the preprocessed input signal represents conjugate transpose, N represents the number of autocorrelation integration time repeatedly selected by the reconstruction matrix, and TiRepresents the ith time sampled according to the autocorrelation period;
(2) performing decorrelation on the autocorrelation matrix R according to the Toeplitz algorithm to reconstruct the input signal R*＝Yb×Yb H＝IvAHRSATIv+σ2I＝IvRHIvAnd obtaining a covariance matrix ofWherein IvRepresenting an identity matrix;
(3) the rank K (AR) of the covariance matrix is determined according to the MUSIC algorithmSAH) Decomposing eigenvalues of the covariance matrix to obtain K positive real eigenvalues and N-K small eigenvalues, and arranging the K positive real eigenvalues of the covariance matrix as lambda in descending order1≥λ2≥…≥λKCorresponding feature vector is v1,v2,…,vKForming a signal subspace span { v }1,v2,…,vKThe rest M × M-K small characteristic values are related to noise, and each characteristic value corresponds to a characteristic vector vK-1,vK-2,…,vNForming a noise subspace span { v }K+1,vK+2,…,vNAnd constructing a noise matrix U by using the noise characteristic vectors corresponding to the noise characteristic vectorsN＝[vK+1,vK+2,…,vM×M]；
(4) Defining a spatial spectral function from the noise vectorWherein a (theta) is a direction vector, and a sum of theta is changedSo that the spectral function reaches a peak value, the azimuth angle of the corresponding signal at the peak valueThe pitch angle theta is the incoming angle of the signal.
And step four, after the incoming wave directions of the direct-view satellite navigation signals and the multipath signals are estimated, beam forming is carried out by adopting a Linear Constrained Minimum Variance (LCMV) criterion to obtain array weighted weights, and the beams of the array antenna are guided to the direction of the expected signals within a period of time so as to achieve the purposes of enhancing the expected signals and inhibiting the multipath.
The rule is that under the condition that the direction of a direct-view signal and the direction of a multipath signal are known, the output power is minimized under the condition that the power of a satellite direct signal is restrained to be basically unchanged after weighting and 0 is restrained after multipath interference weighting, and the principle can be described as the output power by a mathematical expressionWherein the content of the first and second substances,for the autocorrelation matrix after the coherent elimination, C is (M × M) × K-dimensional constraint matrix, i.e., C ═ v1,v2,…vK]，v1Steering vector, v, for direct-view signals of satellite navigation2,v3…vK-1Is the corresponding direction vector of the multipath signal, K-1 is the number of the multipath signal, f is the (K-1) dimensional constraint vector, that isThe response of the array in the direction of the constrained direct signal is 1, and the response in the multipath direction is zeroed out, i.e. the response value is 0. The constraint ensures that the direct signal is received without distortion and without multipath interference. The mathematical expression of the optimal weight vector under the LCMV criterion is obtained by using the Lagrange multiplier method as
And step five, sending the GNSS signals after beam forming into a software receiver for positioning calculation, multiplying weights obtained by using an LCMV (liquid Crystal display television) rule by input signals correspondingly, summing to obtain intermediate frequency signals of the satellite after multipath removal, and sending the intermediate frequency signals into the software receiver for tracking calculation.
And repeating the second step, the third step, the fourth step and the fifth step until all the satellites obtained by coarse acquisition are tracked in the software receiver, and positioning and resolving according to the tracking results of a plurality of satellites.
Compared with the prior art, the invention has the advantages that:
1. the matrix reconstruction algorithm provided by the invention can enable the GNSS software receiver to realize the synchronization of multipath removal and capturing and positioning of beam forming under the condition of no prior information, thereby simplifying the platform structure and not increasing the calculated amount;
2. the array anti-multipath algorithm provided by the invention can remove multipath signals with long time delay and has good inhibition effect on multipath signals with short time delay (the chip delay is less than 0.01 chip).
Drawings
FIG. 1 is a flowchart of the implementation steps of the method for resisting multipath of satellite navigation signals by an array based on a matrix reconstruction algorithm
Fig. 2 is a flow chart of a GNSS array acquisition system according to the present invention.
Fig. 3 is a flowchart of implementation steps of a matrix reconstruction algorithm for preprocessing a 4-array GNSS signal according to the present invention.
FIG. 4 is a flowchart of the implementation steps of decorrelating the incoming wave direction estimation of 4-array GNSS signals and multi-path signals according to the present invention.
FIG. 5 is a flowchart of the steps for implementing the GPS/L1 matrix reconstruction algorithm of the present invention.
Detailed Description
The satellite navigation system is divided into a United states GPS system, a Russian GLONASS system, a Chinese BDS system and a European Galileo system, and the main difference in the invention is the difference of radio frequency and the difference of pseudo code related duration, and the influence on the algorithm is the difference of array spacing and the difference of integration duration of navigation signal preprocessing. In the implementation, the more common GPS/L1 signal is taken as an example.
In an actual environment, the number of signals generated by multipath of a satellite navigation signal which is really received cannot be determined, and the multipath amplitude and phase are difficult to quantify, so that a semi-physical multipath resistant platform for space transmission and reception is built by adopting a transmitter which simulates the real satellite signal and the multipath signal, and the algorithm processing of the invention is implemented.
The implementation scheme of the GPS/L1 signal with one path of multipath of the invention is as follows:
step one, the multi-array element antenna array adopts a2 x 2 plane array, and the spacing between array elements is designed to be 12.75cm according to the radio frequency of GPS/L1 so as to prevent phase ambiguity. The down-conversion acquisition module selects the radio frequency receiving frequency to be 1575.42Mhz through the SPI interface. The data acquisition module is written into a chip by using an FPGA, the sampling frequency of digital-to-analog conversion is 62.5Mhz, and the intermediate frequency is 46.42Mhz, so that the continuity and the authenticity of signals of each channel are ensured, and the bandwidth of a subsequent transmission link is met. The storage module stores the acquired data of the 8 channels into the PC terminal for subsequent algorithm processing, and adopts an oscilloscope to determine the channels corresponding to 4 array elements in the 8 channels, and respectively stores the data of the 4 channels at the PC terminal.
Step two, reconstructing the satellite navigation signal by using a preprocessing algorithm, determining a trackable satellite number by using fine acquisition, taking the number 1 satellite as an example, and obtaining the Doppler frequency f by the fine acquisition and clusteringdAnd code phase cdAnd is carried into 1023 pseudo-random codes of the GPS/L1 to form a new local code. The input data and the local code are correlated according to 106Hz samples the correlated data, and combines the data re-sampled by 4 channels together to form a new input signal matrix Y.
Thirdly, when estimating the incoming wave direction of the star No. 1, segmenting the matrix output by the preprocessing algorithm, selecting 100 sampling points to form a matrix Y of 4 × 100, inputting the matrix Y into the Toeplitz-MUSIC incoming wave direction estimation algorithm, and obtaining the noise matrix U of the star No. 1 according to the stepsN＝[vK+1,vK+2,…,vM×M]. Will pitch angle theta and azimuth angleCalculating spatial spectrum function P of 0-90 deg. and 0-360 deg. at intervals of 1 deg. respectivelyMUSICThe angle corresponding to the peak value is obtained by comparisonIs the input angle of No. 1 GPS signal, the angle corresponding to the relatively small peak valueIs the input angle of the multipath signal for star No. 1.
Step four, obtaining a covariance matrix according to the preprocessing and the Toeplitz decoherence algorithmAnd the incoming wave angles of the star signal 1 and the multipath signals estimated by the MUSIC algorithmAndthe weight w corresponding to the 4 paths of signals is obtained by being brought into a linear constraint minimum variance criterionLCMVAnd respectively multiplying the weight value by the acquired 4 paths of signals and then summing, wherein the weight value is refreshed once every 100ms according to the preprocessing sampling period.
And step five, removing the multipath satellite signal No. 1, inputting the satellite signal No. 1 into a software receiver, and performing acquisition tracking processing on the satellite navigation signal.
And circulating the steps of two, three, four and five until the satellites which can be acquired by the receiver are all subjected to multipath removal processing and are sent to a software receiver for tracking and acquisition.
It should be noted that this implementation only illustrates the technical solution of the present invention, and for different GNSS signals, the down-conversion frequency and the intermediate frequency data may be changed according to the frequency point thereof, the integration time of the matrix reconstruction algorithm may be changed according to the pseudo random codes of the different GNSS signals, and the number of array elements may be changed according to different implementation environments, which may be increased to 16 array elements. One of ordinary skill in the art may substitute the technical solution of the present invention.