阅读说明：本技术 基于目标空间分集的欠定doa估计方法 (Underdetermined DOA estimation method based on target space diversity ) 是由 吴文 唐辉 缪晨 齐世山 陈春红 汪敏 康炜 杨国 王晶琦 于 2020-05-22 设计创作，主要内容包括：本发明公开了一种基于目标空间分集的欠定DOA估计方法,针对LFMCW体制雷达。方法包括：利用LFMCW体制雷达的特性对每一路阵列天线接收信号进行2维FFT；选择一路信号进行二维恒虚警率CFAR检测以获取目标的距离速度信息；每一路信号根据目标在距离-速度空间的分布划分出目标子空间；最后对每个目标子空间分别进行DOA估计。本发明借助LFMCW雷达的特性,对入射目标信号在距离-速度空间进行分集,能够实现在阵元数固定时估计出更多的目标DOA,提高了阵列的整体性能,并且能够实现目标的距离-速度-角度的联合定位。本发明测角精度高,抗噪声性能好,适用于各种阵列结构,且简单易操作,便于工程实际应用。(The invention discloses an underdetermined DOA estimation method based on target space diversity, which aims at LFMCW system radar. The method comprises the following steps: performing 2-dimensional FFT on each path of array antenna receiving signals by utilizing the characteristics of the LFMCW system radar; selecting one path of signal to carry out two-dimensional constant false alarm rate CFAR detection so as to obtain distance and speed information of a target; dividing each path of signal into a target subspace according to the distribution of the target in a distance-speed space; and finally, performing DOA estimation on each target subspace. According to the invention, diversity is carried out on the incident target signal in the distance-speed space by means of the characteristics of the LFMCW radar, more target DOAs can be estimated when the array element number is fixed, the overall performance of the array is improved, and the distance-speed-angle joint positioning of the target can be realized. The invention has high angle measurement precision and good noise resistance, is suitable for various array structures, is simple and easy to operate and is convenient for practical engineering application.)

1. An underdetermined DOA estimation method based on target space diversity, characterized in that the method comprises the following steps:

step 1, establishing a received data model for each path of signals of an LFMCW radar array;

step 2, performing two-dimensional FFT (fast Fourier transform) on each path of signal;

step 3, selecting two-dimensional FFT data of one path of signal to carry out two-dimensional constant false alarm rate CFAR detection, and obtaining distance and speed information of a target;

step 4, dividing a target subspace by each path of signal according to the distribution of the target in a distance-speed space;

step 5, reconstructing a receiving data model of the LFMCW radar array according to the target subspace;

and 6, performing DOA estimation on each target subspace.

2. The underdetermined DOA estimation method based on target space diversity according to claim 1, wherein the step 1 of establishing a received data model for each path of signals of the LFMCW radar array comprises the following specific processes:

aiming at the uniform circular array with N array elements of incidence of M narrow-band signals, the distance and the speed of the target are r respectively_mAnd v_mM1, 2.. M, the horizontal and pitch angles of the target are expressed asWherein theta is_m∈(0,2π)，The radar array emits a sawtooth LFMCW signal with a carrier frequency f₀The signal wavelength is lambda, the frequency modulation bandwidth is B, the frequency modulation period is T, and the frequency modulation slope is k which is B/T;

taking the circle center of the uniform circular array as a reference point, the intermediate frequency receiving signal x of the nth array element at the time t_n(t) is:

in the formula, for a uniform circular array, the center of the circle is the reference pointThe phase difference frequency of the nth array element relative to the center of a circle when the mth incident signal is received is shown, wherein N is 1,2 …, N; s_m(t) represents the signal of the m-th target at the time t, which is incident to the reference point, namely the circle center; tau is_0m＝2r_mThe inherent time delay of the mth target relative to the reference circle center is represented by c, the light speed is represented by c, and the radius of the uniform circular array is represented by d; f. of_dm＝2v_mf₀C represents the mth target doppler shift; f. of_Rm＝2kr_mThe/c is the frequency difference caused by the mth target distance;represents the frequency modulation period of the time t, whereinRepresents rounding down; n is_n(t) represents the noise of the nth array element at the time t; noise as mean 0, noise power as²White gaussian noise.

3. The underdetermined DOA estimation method based on target space diversity according to claim 1 or 2, wherein the step 2 of performing two-dimensional FFT on each path of signal comprises the following specific processes:

step 2-1, AD sampling is carried out on each path of intermediate frequency receiving signals, the sampling frequency is fs, and the sampling period is Ts-1/fs; sampling Nr repeated frequency modulation periods in a circulating manner once, wherein the number of sampling points in each frequency modulation period is Ns, and making Nr equal to 0, 1. The discrete form of the intermediate frequency received signal at time t of the nth array element is therefore:

in the formula, x_n(nr, ns) represents an ns sampling point of an nr-th repetition period; n is_n(nr, ns) is noise n_n(t) in discrete form; s_m(nr,ns)＝exp[j*(2π(f₀τ_0m+f_dm(nr*T)+(f_Rm+f_dm)ns*Ts)]The discrete form of the signal which represents the m-th target to be incident to the reference point, namely the circle center of the uniform circular array;

step 2-2, performing two-dimensional FFT conversion on the discrete formula of the intermediate frequency receiving signal to obtain the distance and speed information of the target, wherein the two-dimensional FFT expression of the intermediate frequency receiving signal of the nth array element at the time t is as follows:

in the formula, w_n(l, k) is an intermediate frequency received signal x_n(nr, Ns) two-dimensional FFT-transformed result, k ═ 0,1,. Ns-1; 1, Nr-1; here, FFT transformation along the ns axis is called distance dimension transformation, FFT transformation along the nr axis is called velocity dimension transformation, and the obtained data space is called a distance-velocity space of the target, abbreviated as target space; s_m(l,k)、n_n(l, k) each represents s_m(nr, ns) and n_nTwo-dimensional FFT results of (nr, ns); the distance dimension transformation has the spectral line spacing of 1/(Ts Ns), each spectral line corresponds to a distance unit, and the corresponding target distance is r ═ k × cTr/(2B × Ns:); similarly, the line spacing for the velocity dimension transform is 1/(T × Nr), each line corresponds to a doppler cell, and the corresponding target velocity is v ═ l × c/(2f × c)₀*T*Nr)。

4. The underdetermined DOA estimation method based on target space diversity according to claim 1, wherein the step 3 of performing two-dimensional constant false alarm rate CFAR detection on the two-dimensional FFT data of the selected one-path signal to obtain the range and velocity information of the target specifically includes:

step 3-1, CFAR processing is carried out on the two-dimensional FFT data to obtain a constant false alarm detection threshold value;

step 3-2, comparing the threshold value with the two-dimensional FFT data, if the two-dimensional FFT data value is larger than the threshold value, judging that a target exists, recording a spectral line position corresponding to the target position, and solving the target distance and speed; otherwise, the target is judged to be absent.

5. The underdetermined DOA estimation method based on target spatial diversity according to claim 4, wherein the constant false alarm rate CFAR detection specifically employs a cell average constant false alarm detection CA _ CFAR.

6. The underdetermined DOA estimation method based on target space diversity according to claim 3, wherein the step 4 divides each path of signal into target subspaces according to the distribution of the target in the distance-velocity space, and the specific process comprises:

dividing a target space into M target subspaces according to the target number M obtained by constant false alarm detection; for the nth two-dimensional FFT transformed signal, the data matrix w of the mth target subspace_nmExpressed as:

in the formula I_m、k_mRespectively representing a Doppler unit and a speed unit where the mth target is located; nr_m、Ns_mRespectively representing the data length of the mth target subspace in the distance dimension and the speed dimension; the symbol (a: b) represents a continuous value from a to b.

7. The method of claim 3, wherein the step 5 of reconstructing the model of the received data of the LFMCW radar array according to the target subspace comprises:

step 5-1, vectorizing the matrix w_nmThen there is z_nm(h)＝rvec(w_nm) Where h is 1,2, …, Ns_m*Nr_m，rvec (·) denotes a row-wise vectorization operation;

step 5-2, recombining subspace data of each path of signals corresponding to the same target into an array receiving data vector:

wherein the content of the first and second substances,a signal representing a reference point target within the mth target subspace,representing the noise of the nth array element in the mth target subspace, wherein N is 1,2.., N;

the above-mentioned Z is_m(h) The expression is written in matrix form:

Z_m(h)＝AS_m(h)+N(h)

in the formula (I), the compound is shown in the specification,showing the flow pattern matrix of the array,the steering vector is shown to be the same as the array flow pattern matrix, N (h) ═ n_1m(h),n_2m(h),…n_Nm(h)]^TIs the noisy data vector of the array target subspace.

Technical Field

The invention relates to the field of array signal processing, in particular to the field of arrival angle estimation of an LFMCW array radar, and particularly relates to an underdetermined DOA estimation method based on target space diversity.

Background

The array angle of arrival (DOA) has been a research hotspot and difficulty since the last 70 th century, and is widely applied to the field of array signal processing such as radar, sonar, wireless communication, seismic sensing and the like. The Multiple Signal Classification algorithm (MUSIC) proposed by Schmidt greatly promotes the development of subspace DOA estimation algorithm. Although the conventional subspace super-resolution DOA estimation technology can break the Rayleigh limit constraint, most algorithms require that the number of sources is less than the number of antenna array elements, and the number of targets which can be estimated is limited by the array structure. Under the underdetermined condition, namely when the number of the information sources is greater than the number of the array elements, the resolution performance of the algorithm is sharply reduced and even fails. However, in practice, because the number of array elements is small and the number of peripheral targets is large, the underdetermined problem is a common situation when the array system works, especially when an omnidirectional antenna is adopted. The significance of research on underdetermined DOA estimation lies in breaking through the limitation of the physical structure of the array, estimating more target angles by using less array elements and saving resources.

In the existing method for processing the underdetermined DOA estimation problem, the characteristic of an incident signal is mainly utilized, and an array flow pattern matrix is constructed to enlarge an array virtual aperture, so that the underdetermined DOA estimation is realized. The document "DOA Estimation of Quasi-static Signals With Less Sensors Than Sources and Unknown Spatial noise Approach" proposes a method for solving underdetermined DOA Estimation by using the property of Kronecker product, and can realize that the number of targets that can be estimated by the array is increased from N to 2N-1. However, the method requires the incident signal to be a quasi-stationary signal, and is mainly applied to the field of acoustics. The document 'research on practical super-resolution direction finding algorithm based on NC-MUSIC' utilizes the characteristic that an elliptical covariance matrix of a non-circular signal is not zero, provides an NC-MUSIC algorithm, and enables the number of information sources which can be estimated by an array to reach 2N-1 by changing the array flow pattern of a received signal, but the method requires the received signal to be the non-circular signal. Although the methods can solve the underdetermined DOA estimation problem, the number of the information sources which can be estimated is still relatively limited, and especially under the condition of less array element number, the requirement on the characteristics of incident signals is often too high, which is not favorable for popularization and application in practical engineering.

Disclosure of Invention

The invention aims to solve the problems, provides an underdetermined DOA estimation method based on target space diversity, can estimate the number of targets as much as possible by using fewer array elements, improves the overall performance of an array, can realize the distance-speed-angle combined positioning of the targets, and is favorable for accurately positioning the targets.

The technical solution for realizing the purpose of the invention is as follows: an underdetermined DOA estimation method based on target spatial diversity, the method comprising the steps of:

step 1, establishing a received data model for each path of signals of an LFMCW radar array;

step 2, performing two-dimensional FFT (fast Fourier transform) on each path of signal;

step 3, selecting two-dimensional FFT data of one path of signal to carry out two-dimensional constant false alarm rate CFAR detection, and obtaining distance and speed information of a target;

step 4, dividing a target subspace by each path of signal according to the distribution of the target in a distance-speed space;

step 5, reconstructing a receiving data model of the LFMCW radar array according to the target subspace;

and 6, performing DOA estimation on each target subspace.

Further, the step 1 of establishing a received data model for each path of signals of the LFMCW radar array includes:

aiming at the uniform circular array with N array elements of incidence of M narrow-band signals, the distance and the speed of the target are r respectively_mAnd v_mM1, 2.. M, the horizontal and pitch angles of the target are expressed asWherein theta is_m∈(0,2π)，The radar array emits a sawtooth LFMCW signal with a carrier frequency f₀The signal wavelength is lambda, the frequency modulation bandwidth is B, the frequency modulation period is T, and the frequency modulation slope is k which is B/T;

taking the circle center of the uniform circular array as a reference point, the intermediate frequency receiving signal x of the nth array element at the time t_n(t) is:

in the formula, for a uniform circular array, the center of the circle is the reference pointThe phase difference frequency of the nth array element relative to the center of a circle when the mth incident signal is received is shown, wherein N is 1,2, N; s_m(t) represents the signal of the m-th target at the time t, which is incident to the reference point, namely the circle center; tau is_0m＝2r_mThe inherent time delay of the mth target relative to the reference circle center is represented by c, the light speed is represented by c, and the radius of the uniform circular array is represented by d; f. of_dm＝2v_mf₀C represents the mth target doppler shift; f. of_Rm＝2kr_mThe/c is the frequency difference caused by the mth target distance;represents the frequency modulation period of the time t, whereinRepresents rounding down; n is_n(t) represents the noise of the nth array element at the time t; noise as mean 0, noise power as²White gaussian noise.

Further, the step 2 of performing two-dimensional FFT on each path of signal includes:

step 2-1, AD sampling is carried out on each path of intermediate frequency receiving signals, the sampling frequency is fs, and the sampling period is Ts-1/fs; nr repeated frequency modulation periods are sampled in a circulating mode at one time, the number of sampling points in each frequency modulation period is Ns, and Nr is equal to 0,1, … Nr-1, and Ns is equal to 0,1, … Ns-1; the discrete form of the intermediate frequency received signal at time t of the nth array element is therefore:

in the formula, x_n(nr, ns) represents an ns sampling point of an nr-th repetition period; n is_n(nr, ns) is noise n_n(t) in discrete form; s_m(nr,ns)＝exp[j*(2π(f₀τ_0m+f_dm(nr*T)+(f_Rm+f_dm)ns*Ts)]The discrete form of the signal which represents the m-th target to be incident to the reference point, namely the circle center of the uniform circular array;

step 2-2, performing two-dimensional FFT conversion on the discrete formula of the intermediate frequency receiving signal to obtain the distance and speed information of the target, wherein the two-dimensional FFT expression of the intermediate frequency receiving signal of the nth array element at the time t is as follows:

in the formula, w_n(l, k) is an intermediate frequency received signal x_n(nr, Ns) two-dimensional FFT result, k is 0,1, …, Ns-1; 0,1, …, Nr-1; here, FFT transformation along the ns axis is called distance dimension transformation, FFT transformation along the nr axis is called velocity dimension transformation, and the obtained data space is called a distance-velocity space of the target, abbreviated as target space; s_m(l,k)、n_n(l, k) each represents s_m(nr, ns) and n_nTwo-dimensional FFT results of (nr, ns); the distance dimension transformation has the spectral line spacing of 1/(Ts Ns), each spectral line corresponds to a distance unit, and the corresponding target distance is r ═ k × cTr/(2B × Ns:); similarly, the line spacing for the velocity dimension transform is 1/(T × Nr), each line corresponds to a doppler cell, and the corresponding target velocity is v ═ l × c/(2f × c)₀*T*Nr)。

Further, the step 3 of performing two-dimensional constant false alarm rate CFAR detection on the two-dimensional FFT data of the selected one-path signal to obtain the distance and velocity information of the target specifically includes:

step 3-1, CFAR processing is carried out on the two-dimensional FFT data to obtain a constant false alarm detection threshold value;

step 3-2, comparing the threshold value with the two-dimensional FFT data, if the two-dimensional FFT data value is larger than the threshold value, judging that a target exists, recording a spectral line position corresponding to the target position, and solving the target distance and speed; otherwise, the target is judged to be absent.

Further, in step 4, each path of signal divides a target subspace according to the distribution of the target in the distance-velocity space, and the specific process includes:

dividing a target space into M target subspaces according to the target number M obtained by constant false alarm detection; for the nth two-dimensional FFT transformed signal, the data matrix w of the mth target subspace_nmExpressed as:

in the formula I_m、k_mRespectively representing a Doppler unit and a speed unit where the mth target is located; nr_m、Ns_mRespectively representing the data length of the mth target subspace in the distance dimension and the speed dimension; the symbol (a: b) represents a continuous value from a to b.

Further, the step 5 of reconstructing the received data model of the LFMCW radar array according to the target subspace includes the following specific processes:

step 5-1, vectorizing the matrix w_nmThen there is z_nm(h)＝rvec(w_nm) Where h is 1,2, …, Ns_m*Nr_m，rvec (·) denotes a row-wise vectorization operation;

step 5-2, recombining subspace data of each path of signals corresponding to the same target into an array receiving data vector:

wherein the content of the first and second substances,a signal representing a reference point target within the mth target subspace,representing the noise of the nth array element in the mth target subspace, wherein N is 1,2 …, and N;

the above-mentioned Z is_m(h) The expression is written in matrix form:

Z_m(h)＝AS_m(h)+N(h)

in the formula (I), the compound is shown in the specification,showing the flow pattern matrix of the array,the steering vector is shown to be the same as the array flow pattern matrix, N (h) ═ n_1m(h),n_2m(h),…n_Nm(h)]^TIs the noisy data vector of the array target subspace.

Compared with the prior art, the invention has the following remarkable advantages: 1) according to the underdetermined DOA estimation method based on target space diversity, only one target to be estimated is arranged in each target subspace, the limitation of an array structure can be broken through, the estimation of more target DOAs by using fewer array elements is realized, resources are saved, and the full utilization of the resources is realized; although other underdetermined DOA estimation methods can also realize underdetermined DOA estimation, the number of targets which can be estimated is actually limited, and the method can theoretically realize the estimation of infinite target DOAs; 2) compared with other underdetermined DOA estimation methods, the method has the advantages that the precondition is few, the requirement on incident signals is low, only the LFMCW system is adopted by the radar array, the method is simple and ingenious, the DOA estimation method is flexible and selectable, and the actual application in engineering is facilitated; 3) joint estimation among the angle, the speed and the distance of the target can be realized; the target space diversity ensures that only one target exists in each target subspace, reduces the interference of other targets, improves the DOA estimation precision, and simultaneously, because only one target exists in the subspace, once the target DOA is estimated, the speed-distance is matched with the target DOA, so that the positioning and tracking of the target in a three-dimensional space are facilitated on one hand, and on the other hand, the problem that the subspace type algorithm cannot distinguish the targets with different distances, different speeds and the same incoming wave direction is solved; 4) the underdetermined DOA estimation algorithm realizes the periodic accumulation of signals when the two-dimensional FFT is carried out, improves the signal-to-noise ratio, and has stronger robustness to noise.

The present invention is described in further detail below with reference to the attached drawing figures.

Drawings

Fig. 1 is a flow diagram of an underdetermined DOA estimation method based on target spatial diversity in one embodiment.

FIG. 2 is a schematic diagram of an array structure of a uniform circular array in one embodiment.

FIG. 3 is a time-frequency diagram of a sawtooth signal in one embodiment.

FIG. 4 is a schematic diagram of a data acquisition cycle according to an embodiment.

Fig. 5 is a spatial distribution diagram of a target after constant false alarm detection in one embodiment.

FIG. 6 is a diagram illustrating target subspace diversity in one embodiment.

FIG. 7 is a diagram illustrating the estimation result of the underdetermined DOA estimation target spatial spectrum in one embodiment.

FIG. 8 is a graph of the angular accuracy versus the signal-to-noise ratio for different source numbers in one embodiment.

Fig. 9 is a schematic diagram of spatial spectrums of objects with different distances, different speeds and same directions in an embodiment, where (a) is a spatial spectrum of a subspace of an object 1, and (b) is a spatial spectrum of a subspace of an object 2.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

The invention considers the radar of common LFMCW system in engineering application to solve the underdetermined DOA estimation problem. The radar of the LFMCW system is relatively mature at present, the conventional subspace DOA estimation algorithm is also relatively mature, the problem of difficulty is solved by combining two mature simple methods, the method is simple and ingenious, the method is combined with the engineering practice, the engineering realization is facilitated, the joint estimation of the angle, the distance and the speed of a target can be realized, and the accurate positioning of the target is facilitated.

In one embodiment, in conjunction with fig. 1, there is provided a method for underdetermined DOA estimation based on target spatial diversity, the method comprising the steps of:

step S01, establishing a received data model for each path of signal of the LFMCW radar array;

step S02, performing two-dimensional FFT on each path of signal;

step S03, selecting two-dimensional FFT data of one path of signal to carry out two-dimensional constant false alarm rate CFAR detection, and obtaining distance and speed information of a target;

step S04, dividing each path of signal into target subspaces according to the distribution of the target in the distance-speed space;

step S05, reconstructing a receiving data model of the LFMCW radar array according to the target subspace;

step S06, DOA estimation is performed for each target subspace. Here, since the array received data vectors of the subspaces have been reconstructed in step S05, the DOA of the target in each subspace can be estimated thereafter using various common subspace-like DOA estimation algorithms. The present invention thus completes the solution to the problem of underdetermined DOA estimation.

Here, preferably, the DOA estimation algorithm employs the MUSIC algorithm.

Further, in one embodiment, the step S01 of establishing a received data model for each path of signals of the LFMCW radar array includes:

aiming at the uniform circular array with N array elements of incidence of M narrow-band signals, the distance and the speed of the target are r respectively_mAnd v_mM1, 2.. M, the horizontal and pitch angles of the target are expressed asWherein theta is_m∈(0,2π)，The radar array emits a sawtooth LFMCW signal with a carrier frequency f₀The signal wavelength is lambda, the frequency modulation bandwidth is B, the frequency modulation period is T, and the frequency modulation slope is k which is B/T;

taking the circle center of the uniform circular array as a reference point, the intermediate frequency receiving signal x of the nth array element at the time t_n(t) is:

in the formula, for a uniform circular array, the center of the circle is the reference pointThe phase difference frequency of the nth array element relative to the center of a circle when the mth incident signal is received is shown, wherein N is 1,2, N; s_m(t) represents the signal of the m-th target at the time t, which is incident to the reference point, namely the circle center; tau is_0m＝2r_mThe inherent time delay of the mth target relative to the reference circle center is represented by c, the light speed is represented by c, and the radius of the uniform circular array is represented by d; f. of_dm＝2v_mf₀C represents the mth target doppler shift; f. of_Rm＝2kr_mThe/c is the frequency difference caused by the mth target distance;represents the frequency modulation period of the time t, whereinRepresents rounding down; n is_n(t) represents the noise of the nth array element at the time t; noise as mean 0, noise power as²White gaussian noise.

Here, the formula of the intermediate frequency reception signal is common to all LFMCW radar arrays of various array structures, and the different array structures differ only in f_nm，f_nmRepresenting the spatial phase difference of the nth array element relative to the reference array element when the mth incoming signal is received.

Here, in consideration of the fact that the uniform circular array can achieve 360 ° coverage of the azimuth angle and can simultaneously obtain the horizontal angle and the pitch angle of the target, the array structure of the present invention employs the uniform circular array.

Further, in one embodiment, the step S02 of performing two-dimensional FFT on each path of signal includes:

step S02-1, performing AD sampling on each channel of intermediate frequency received signals, where the sampling frequency is fs and the sampling period is Ts ═ 1/fs; sampling Nr repeated frequency modulation periods in a circulating manner once, wherein the number of sampling points in each frequency modulation period is Ns, and making Nr equal to 0, 1. The discrete form of the intermediate frequency received signal at time t of the nth array element is therefore:

in the formula, x_n(nr, ns) represents an ns sampling point of an nr-th repetition period; n is_n(nr, ns) is noise n_n(t) in discrete form; s_m(nr,ns)＝exp[j*(2π(f₀τ_0m+f_dm(nr*T)+(f_Rm+f_dm)ns*Ts)]Discrete form of signal representing m-th target incident to reference point, i.e. centre of uniform circular array(ii) a Here, as can be seen from the above formula, the distance and velocity information of the target is implicit in the frequency and phase terms of exp, so that the distance and velocity information of the target can be derived by performing two-dimensional FFT on the above formula;

step S02-2, performing two-dimensional FFT on the discrete formula of the intermediate frequency received signal to obtain the distance and speed information of the target, where the two-dimensional FFT expression of the intermediate frequency received signal of the nth array element at time t is:

in the formula, w_n(l, k) is an intermediate frequency received signal x_n(nr, Ns) two-dimensional FFT-transformed result, k ═ 0,1,. Ns-1; 1, Nr-1; here, FFT transformation along the ns axis is called distance dimension transformation, FFT transformation along the nr axis is called velocity dimension transformation, and the obtained data space is called a distance-velocity space of the target, abbreviated as target space; s_m(l,k)、n_n(l, k) each represents s_m(nr, ns) and n_nTwo-dimensional FFT results of (nr, ns); the distance dimension transformation has the spectral line spacing of 1/(Ts Ns), each spectral line corresponds to a distance unit, and the corresponding target distance is r ═ k × cTr/(2B × Ns:); similarly, the line spacing for the velocity dimension transform is 1/(T × Nr), each line corresponds to a doppler cell, and the corresponding target velocity is v ═ l × c/(2f × c)₀*T*Nr)。

Further, in one embodiment, the step S03 of performing two-dimensional constant false alarm rate CFAR detection on the two-dimensional FFT data of the selected signal to obtain distance and velocity information of the target specifically includes:

step S03-1, CFAR processing is carried out on the two-dimensional FFT data to obtain a constant false alarm detection threshold value;

step S03-2, comparing the threshold value with the two-dimensional FFT data, if the two-dimensional FFT data value is larger than the threshold value, determining that a target exists, recording a spectral line position corresponding to the target position, and solving the target distance and speed; otherwise, the target is judged to be absent.

Further preferably, in one of the embodiments, the constant false alarm rate CFAR detectionSpecifically, a unit average constant false alarm rate is adopted to detect CA _ CFAR, and for CA _ CFAR, a judgment threshold is only related to the number L of samples near a judgment unit and a scale factor a. Consider the false alarm probability as P_FAThen scale factorThen the expression of the decision threshold T is:

in the formula, w_niAnd representing the data around the nth signal target decision unit for constant false alarm detection.

Further, in one embodiment, in step 4, each of the signals divides the target subspace according to the distribution of the target in the distance-velocity space, and the specific process includes:

in order to reduce the influence of surrounding targets on the current target to be subjected to DOA estimation, only one target to be estimated is required in each divided target subspace, so that the target space is divided into M target subspaces, and only data around the target is taken in each subspace. For the nth two-dimensional FFT transformed signal, the data matrix w of the mth target subspace_nmExpressed as:

in the formula I_m、k_mRespectively representing a Doppler unit and a speed unit where the mth target is located; nr_m、Ns_mRespectively representing the data length of the mth target subspace in the distance dimension and the speed dimension; the symbol (a: b) represents a continuous value from a to b.

Further, in one embodiment, the reconstructing the received data model of the LFMCW radar array according to the subspace in step S05 includes:

step S05-1, vectorizing matrix w_nmThen there is z_nm(h)＝rvec(w_nm) Wherein h is 1,2,…,Ns_m*Nr_m，rvec (·) denotes a row-wise vectorization operation;

step S05-2, the subspace data of each signal corresponding to the same target is recombined into an array received data vector:

wherein the content of the first and second substances,a signal representing a reference point target within the mth target subspace,representing the noise of the nth array element in the mth target subspace, wherein N is 1,2.., N;

the above-mentioned Z is_m(h) The expression is written in matrix form:

Z_m(h)＝AS_m(h)+N(h)

in the formula (I), the compound is shown in the specification,showing an array flow pattern matrix, since there is only one target in the subspace at this time,the steering vector is shown to be the same as the array flow pattern matrix, N (h) ═ n_1m(h),n_2m(h),…n_Nm(h)]^TIs the noisy data vector of the array target subspace.

As a specific example, the present invention is further explained by verification, including the following:

1. establishing a received data model for each path of signal:

the array structure of the uniform circular array is shown in fig. 2, and M narrow-band signals are considered to be incident to the uniform circular array with the array element number of N, and the target is locatedDistance and velocity are r_mAnd v_m(M ═ 1,2,. M), the horizontal and pitch angles of the target are expressed asWherein theta is_m∈(0,2π)，In this example, the target number M is 10, the number of array elements N is 5, the radius of the array elements r is 2 λ, and the signal-to-noise ratio is 15 dB. The specific distance of the target. The speed and angle information is shown in table 1 below.

TABLE 1 object specific range speed and Angle information

The radar transmits LFMCW signals in the form of saw-tooth waves, the time-frequency diagram of which is shown in fig. 3. Wherein the carrier frequency is f₀When the signal wavelength is 9.1GHz, the signal wavelength is 33 mm; the bandwidth of the frequency modulation band is B ═ 100 MHz; the frequency modulation period is T-0.1 ms, and the frequency modulation slope is k-B/T. Taking the circle center of the uniform circular array as a reference point, the intermediate frequency receiving signal x of the nth array element at the time t_n(t) is:

in the formula (I), the compound is shown in the specification,the phase difference frequency of the nth array element relative to the center of a circle when the mth incident signal is received is shown, wherein N is 1,2, N; s_m(t) represents the signal of the m-th target at the time t, which is incident to the reference point, namely the circle center; tau is_0m＝2r_mThe inherent time delay of the mth target relative to the reference circle center is represented by c, the light speed is represented by c, and the radius of the uniform circular array is represented by d; f. of_dm＝2v_mf₀C represents the mth target doppler shift; f. of_Rm＝2kr_mThe/c is the frequency difference caused by the mth target distance;represents the frequency modulation period of the time t, whereinRepresents rounding down; n is_n(t) represents the noise of the nth array element at the time t; noise as mean 0, noise power as²White gaussian noise.

2. And respectively carrying out two-dimensional FFT on each path of signal:

fig. 4 shows a diagram of a data acquisition cycle, and it is considered that AD sampling is performed on each intermediate frequency received signal, where fs is 10MHz, and Ts is 0.1 us. Nr repeated frequency modulation periods are sampled in one cycle, the number of sampling points in each frequency modulation period is Ns, and Nr is made equal to 0_,1,...Nr-1；ns＝0_,Ns-1. The discrete form of the intermediate frequency receiving signal of the nth array element at the time t after AD sampling is as follows:

in the formula, x_n(nr, ns) represents an ns sampling point of an nr-th repetition period; n is_n(nr, ns) is noise n_n(t) in discrete form; s_m(nr,ns)＝exp[j*(2π(f₀τ_0m+f_dm(nr*T)+(f_Rm+f_dm)ns*Ts)]And (3) a discrete form of a signal representing the m-th target incident to the reference point, namely the center of the uniform circular array.

It can be seen from the above formula that the distance and velocity information of the target is implied in the frequency and phase terms of exp, so that the distance and velocity information of the target can be derived by performing two-dimensional FFT on the above formula, and the two-dimensional FFT expression of the intermediate frequency received signal of the nth array element at time t is as follows:

in the formula, w_n(l, k) is an intermediate frequency received signal x_n(nr, ns) two-dimensional FFT transformAs a result, k is 0, 1.., Ns-1; 1, Nr-1; here, FFT transformation along the ns axis is called distance dimension transformation, FFT transformation along the nr axis is called velocity dimension transformation, and the obtained data space is called a distance-velocity space of the target, abbreviated as target space; s_m(l,k)、n_n(l, k) each represents s_m(nr, ns) and n_nTwo-dimensional FFT results of (nr, ns); the distance dimension transformation has the spectral line spacing of 1/(Ts Ns), each spectral line corresponds to a distance unit, and the corresponding target distance is r ═ k × cTr/(2B × Ns:); similarly, the line spacing for the velocity dimension transform is 1/(T × Nr), each line corresponds to a doppler cell, and the corresponding target velocity is v ═ l × c/(2f × c)₀*T*Nr)。

3. Selecting one path of signal to perform two-dimensional CFAR detection, and obtaining distance and speed information of a target:

in order to find the specific position of the target in the distance-velocity space from the data after the two-dimensional FFT, i.e. to find the specific distance and velocity of the target, CFAR detection is required. Firstly, CFAR processing is carried out on input two-dimensional FFT data to obtain a constant false alarm detection threshold, the threshold is compared with the input two-dimensional FFT data, if an input signal is larger than a threshold value, a target is judged to exist, a spectral line position corresponding to the target position is recorded, and the distance and the speed are solved; otherwise, it is determined that no target exists. Here constant false alarm processing employs cell average constant false alarm detection (CA _ CFAR). For CA _ CFAR, the decision threshold is only related to the number of samples L near the decision unit and the scale factor a. Here false alarm probability takes P_FA＝10^-2Then scale factorThen the expression T for the decision threshold is:

in the formula, w_niAnd representing the data around the nth signal target decision unit for constant false alarm detection.

The distance-velocity space profile of a constant false alarm detected target is shown in fig. 5.

4. Dividing a target subspace for each path of signal according to the distribution of the target in a distance-speed space:

through the data processing in the above-described procedure 3, the position information of the targets in the distance-velocity space and the number M of targets can be obtained as 10. Therefore, the target space is divided into M target subspaces, and only the data around the target is taken in each subspace. A schematic diagram of the target spatial diversity is shown in fig. 6. Therefore, the data matrix w of the mth target subspace is the nth two-dimensional FFT transformed signal_nmExpressed as:

in the formula I_m、k_mRespectively representing a Doppler unit and a speed unit where the mth target is located; nr_m、Ns_mRespectively representing the data length of the mth target subspace in the distance dimension and the speed dimension; the symbol (a: b) represents a continuous value from a to b.

5. Reconstructing a received data model of the LFMCW radar array according to the subspace:

first vectorized matrix w_nmThen there is z_nm(h)＝rvec(w_nm) Where h is 1,2, …, Ns_m*Nr_m，rvec (·) denotes a row-wise vectorization operation;

then, the subspace data of each path of signal corresponding to the same target is recombined into an array receiving data vector:

wherein the content of the first and second substances,a signal representing a reference point target within the mth target subspace,representing the noise of the nth array element in the mth target subspace, wherein N is 1,2.., N;

the above-mentioned Z is_m(h) The expression is written in matrix form:

Z_m(h)＝AS_m(h)+N(h)

in the formula (I), the compound is shown in the specification,showing an array flow pattern matrix, since there is only one target in the subspace at this time,the steering vector is shown to be the same as the array flow pattern matrix, N (h) ═ n_1m(h),n_2m(h),…n_Nm(h)]^TIs the noisy data vector of the array target subspace.

6. DOA estimation is performed for each target subspace:

since process 5 above has reconstructed the array received data vectors for the M subspaces, the DOA of the target in each subspace can then be estimated using various common subspace-like DOA estimation algorithms. The DOA estimation in this example uses the MUSIC algorithm:

(1) computing a subspace array received data vector Z_m(h) Covariance matrix of

Wherein, (.)^HDenotes a conjugate transpose, I_NThe subspace array has a finite length of received data, so the data covariance matrix can be obtained by the maximum likelihood estimation, and the expression is:

the subspace data length denoted by Ns × Nr.

(2) To pairAnd decomposing the characteristic value. And according to the change trend of the characteristic values, considering that the large characteristic value corresponds to a target characteristic value and the small characteristic value corresponds to a noise characteristic value. Since the number of targets in the subspace is known to be 1 at this time, the noise subspace matrix U can be accurately obtained_NSubstituting the spectrum estimation function of the MUSIC algorithm with the following functions:

performing spectral peak search on the formula, wherein the point of the spectral peak isThe value is an estimate of the direction of arrival of the signal. The resulting underdetermined DOA spatial spectrum is shown in FIG. 7. In order to make the result display simple, the DOA spatial spectrums of a plurality of target subspaces are superposed on a graph to be displayed, and actually, one DOA spatial spectrum is required to be displayed in each target subspace. The simulation result shows that the underdetermined DOA estimation method based on the target space diversity well solves the underdetermined DOA estimation problem, estimates the correct incoming wave directions of all targets, and realizes the combined positioning of the target distance, the speed and the angle.

Fig. 8 shows the angle measurement accuracy of the present invention at different signal-to-noise ratios, which is used to illustrate that the present invention has better noise immunity. The Monte Carlo experiments are respectively carried out for 50 times when the target numbers are 3, 7 and 10, and the scanning interval of the MUSIC algorithm during the peak searching is 0.5 degrees. Root Mean Square Error (RMSE) is introduced here to measure the accuracy of the angle measurement. The root mean square error is expressed as follows:

wherein G represents the number of Monte Carlo experiments,respectively represents theta at the g-th Monte Carlo experiment_mAndan estimate of (d).

As can be seen from FIG. 8, the method of the present invention has better angle measurement accuracy in the whole simulation region, and especially after the signal-to-noise ratio is greater than 4dB, the error-free estimation of the target angle can be basically realized. Comparing three broken lines of M3, M7 and M10 in the figure, it can be found that the angle measurement precision is reduced when the signal-to-noise ratio is low with the increase of the target number, but the errors are small overall, the precision is high, and the engineering requirements are met. Compared with the DOA estimation result of the MUSIC algorithm under the overdetermined condition, the DOA estimation performance of the method under the condition of low signal-to-noise ratio is obviously improved, and the MUSIC algorithm can hardly correctly estimate the target under the condition of low signal-to-noise ratio.

Fig. 9 considers that two objects having the same azimuth but different range or speed exist in the space, their azimuth angles are (100 °,30 °), and the range and speed information of the objects are (80m, -15m/s), (120m,5m/s), respectively. For the subspace-like algorithm under such a condition, only one of the targets can be estimated, but for the method of the present invention, both targets can be correctly estimated at this time, because the method of the present invention also uses the distance and velocity information of the targets for joint estimation, and the simulation result is shown in fig. 9.

In conclusion, the method can realize DOA estimation of the target under the underdetermined condition. Compared with the existing method, the method has the advantages that diversity is carried out on the incident target signal in the distance-speed space by means of the characteristics of the LFMCW radar, more target DOAs can be estimated when the array element number is fixed, the overall performance of the array is improved, the distance-speed-angle combined positioning of the target can be realized, and the accurate positioning of the target is facilitated. The invention has high angle measurement precision and good noise resistance, is suitable for various array structures, is simple and easy to operate and is convenient for practical engineering application.

The foregoing illustrates and describes the principles, general features, and advantages of the present invention. It will be understood by those skilled in the art that the present invention is not limited by the foregoing examples, which are provided to illustrate the principles of the invention, and that various changes and modifications may be made without departing from the spirit and scope of the invention, which is intended to be protected by the following claims. The scope of the invention is defined by the appended claims and equivalents thereof.