METHOD FOR ESTIMATING THE DIRECTION-OF-ARRIVAL OF A COPRIME ARRAY BASED ON VIRTUAL DOMAIN STATISTICS RECONSTRUCTION OF SINGLE-BIT QUANTIZED SIGNAL

Abstract

The invention discloses a method for estimating the direction-of-arrival of a coprime array based on virtual domain statistics reconstruction of single-bit quantized signal, which mainly solves the problems of difficult realization of software and hardware, limited degree of freedom and the like in the prior art. The realization steps are as follows: arranging a coprime array and a single-bit analog-to-digital converter at a receiving end; calculating equivalent virtual signal corresponding to a single-bit receipt signal of the coprime array; constructing a virtual domain augmented covariance matrix of an initialized single-bit quantized signal; designing, based on statistical correlation analysis between statistics of the single-bit quantized signal and the original unquantized signal, an optimization problem based on virtual domain statistics reconstruction of quantized signal; and performing direction-of-arrival estimation by utilizing the virtual domain augmented covariance matrix corresponding to the optimized single-bit quantized signal.

Claims

1. A method for estimating the direction-of-arrival of a coprime array based on virtual domain statistics reconstruction of single-bit quantized signal, comprising the following steps: (1) arranging a coprime array at a receiving end using M+N−1 antennas, each array element of the coprime array being connected with a single-bit analog-to-digital converter for single-bit quantization of receipt signal, wherein M and N are coprime integers; (2) modeling a single-bit receipt signal of the coprime array: assuming that there are K far-field narrowband incoherent signal sources from directions θ.sub.1, θ.sub.2, . . . , θ.sub.K, receiving an incident signal by adopting the coprime array and the single-bit analog-to-digital converter constructed in step (1), and obtaining the single-bit receipt signal y(l)∈.sup.M+N−1 of the coprime array at the lth time, wherein the modeling is $y\left(l\right)=\mathrm{��}\left(x\left(l\right)\right)=\mathrm{��}\left(\underset{k=1}{\overset{K}{.Math.}}{a}_S\left({\theta }_k\right)s_k\left(l\right)+n_S\left(l\right)\right),$ wherein, (.Math.) is a single-bit quantization operator, x(l) is an unquantized original receipt signal of the coprime array, s.sub.k(l) is a waveform of a k.sup.th signal, (l) is a noise item independent of each signal source, and (θ.sub.k) is a steering vector of the coprime array corresponding to the direction θ.sub.k, expressed as
(θ.sub.k)=┌1,e.sup.−jπu.sup.2.sup.sin(θ.sup.k.sup.), . . . ,e.sup.−jπu.sup.M+N−1.sup.sin(θ.sup.k.sup.)┐.sup.T, wherein, u.sub.1, i=1, 2, . . . , M+N−1 represents an actual position of the i.sup.th physical antenna element in the coprime array, and u.sub.1=0, j=√{square root over (−1)}, [.Math.].sup.T represents a transposition operation, and a sampling covariance matrix of the single-bit receipt signal of the coprime array is obtained by using collected L sampling snapshots, expressed as ${\hat{Q}}_S=\frac{1}{L}\underset{l=1}{\overset{L}{.Math.}}y\left(l\right)y^H\left(l\right),$ wherein (.Math.).sup.H represents conjugate transpose; (3) calculating an equivalent virtual signal corresponding to the single-bit receipt signal of the coprime array: vectorizing the sampling covariance matrix of the single-bit receipt signal of the coprime array, and obtaining the equivalent receipt signal of a virtual array corresponding to the single-bit quantized signal, expressed as ${\hat{q}}_{\mathrm{��}}=\mathrm{vec}\left({\hat{Q}}_{\mathrm{��}}\right)=\frac{2}{\pi }\left(\arcsin \left(\left({.Math.}^{-\frac{1}{2}}.Math.{.Math.}^{-\frac{1}{2}}\right)\mathrm{.Math.}\left(r_{\mathrm{��}}\right)\right)+j.Math.\arcsin \left(\left({.Math.}^{-\frac{1}{2}}.Math.{.Math.}^{-\frac{1}{2}}\right)\mathrm{��}\left(r_{\mathrm{��}}\right)\right)\right),$ corresponding to a non-uniform virtual array,
={v.sub.1−v.sub.D|v.sub.Dv.sub.1∈}, wherein, vec(.Math.) represents a vectorization operation, that is, columns in the matrix are stacked in sequence to form a new vector, Σ=diag(), =E[x(l)x.sup.H(l)] is the covariance matrix of the receipt signal of a unquantized original coprime array, diag(.Math.) represents an operation of taking diagonal elements to form a diagonal matrix, E[.Math.] represents taking expectation operation, .Math. represents Kronecker product, (.Math.) represents taking real part operation, (.Math.) represents taking imaginary part operation, $r_D=\underset{k=1}{\overset{K}{.Math.}}{\sigma }_k^2{a}_D\left({\theta }_k\right)+{\sigma }_n^2\mathrm{vec}\left(I\right),$ wherein, σ.sub.k.sup.2 represents power of the k.sup.th signal source, σ.sub.n.sup.2 represents noise power, I is an identity matrix, (θ.sub.k) is a steering vector of a non-uniform virtual array corresponding to the direction θ.sub.k, and is calculated as (θ.sub.k)=(θ.sub.k).Math.(θ.sub.k), wherein (.Math.)* is a conjugate operation; (4) constructing a virtual domain augmented covariance matrix of an initialized single-bit quantized signal: in order to overcome a signal model mismatch problem caused by the non-uniform virtual array of the coprime array, constructing a virtual domain uniform linear array with a same aperture as a positive half axis of the non-uniform virtual array and a spacing d, wherein the unit spacing d is half of the wavelength of incident narrowband signal, expressed as
={ud|u=0,1,2, . . . ,max()/d}, wherein, max(.Math.) represents an operation of taking the set maximum, correspondingly, the equivalent virtual signal ∈ corresponding to the virtual domain uniform linear array is obtained by the following method: for the equivalent virtual signal corresponding virtual array element position corresponding to , if the virtual array element position is included in the non-uniform virtual array , the equivalent virtual signal at this position is the same as a virtual signal corresponding to the corresponding virtual array element position in ; the equivalent virtual signal corresponding to discontinuous virtual array elements in the remaining non-uniform virtual arrays are set to zero, and then a virtual domain augmented covariance matrix of the initialized single-bit quantized signal can be constructed as
=Toep(), wherein, Toep(.Math.) represents that a vector taken is the first column of a Hermitian Toeplitz matrix; (5) designing, based on statistical correlation analysis between statistics of the single-bit quantized signal and the original unquantized signal, an optimization problem based on virtual domain statistics reconstruction of quantized signal to obtain a single-bit quantized signal covariance matrix corresponding to a virtual uniform array ; according to statistical characteristic analysis, based on the characteristic that the single-bit quantized signal covariance matrix =E[y(l)y.sup.H(l)] is the same as a maximum linear independent set of an original unquantized signal covariance matrix , indicating that a rank of covariance matrix does not change in a single-bit quantization process, and the virtual domain augmented covariance matrix corresponding to the single-bit quantized signal derived from can be regarded as a sampling covariance matrix calculated from the single-bit receipt signal of the virtual uniform array , but some elements are missing; under an ideal condition that all elements are known, the covariance matrix will still retain the matrix rank information related to the incident signal source; based on statistical correlation analysis among the above statistics, the reconstruction problem of augmented covariance matrix can be constrained and optimized by using low rank characteristics of the covariance matrix, and then the following optimization problem for virtual domain statistics reconstruction of single-bit quantized signal is constructed taking as an optimization target: $q_U^{★}=\arg \underset{q_U\in C^{\left[U\right]}}{\min }{.Math.{\mathrm{��}}_{\Omega }\left(\mathrm{Toep}\left(q_U\right)\right)-{\hat{Q}}_U.Math.}_F+\lambda .Math.\mathrm{rank}\left(\mathrm{Toep}\left(q_U\right)\right)$ $\mathrm{subject}\mathrm{to}\mathrm{Toep}\left(q_U\right)\succcurlyeq 0,$ wherein, .sub.Ω(.Math.) represents a projection operation used to select the element in Toep() corresponding to the position of non-zero elements for fitting, λ is a user adjustment parameter, Toep()0 is a positive semi-definite matrix constraint item, ∥.Math.∥.sub.F represents a Frobenius norm, rank(.Math.) represents the rank of the matrix; the single-bit quantized signal covariance matrix Toep() corresponding to the virtual uniform array can be obtained by solving the above optimization problem; and (6) performing direction-of-arrival estimation by utilizing the optimized single-bit quantized signal covariance matrix Toep() corresponding to the virtual uniform array .

2. The method for estimating the direction-of-arrival of a coprime array based on virtual domain statistics reconstruction of single-bit quantized signal according to claim 1, wherein the coprime array in step (1) can be arranged as follows: firstly, a pair of coprime integers M and N are selected, and then a pair of sparse uniform linear subarrays are constructed, wherein the first subarray contains M antenna elements with a spacing of Nd and positions of .sub.1={0, Nd, . . . , (M−1)Nd}, the second subarray contains N antenna elements with a spacing of Md and positions of .sub.2={0, Md, . . . , (N−1)Md}; then, the two sub-arrays are combined according to a way that the first array elements overlap, so as to obtain a non-uniform coprime array =.sub.1+.sub.2 actually containing M+N−1 physical antenna elements.

3. The method for estimating the direction-of-arrival of a coprime array based on virtual domain statistics reconstruction of single-bit quantized signal according to claim 1, wherein an optimal solution of the optimization problem in step (5) is obtained by using convex relaxation technique, and introducing a convex function term trace(Toep()) to replace a penalty term rank(Toep()) in the optimization problem, wherein trace(.Math.) represents a trace of the matrix, and then efficiently solving the optimization problem through various interior point method tools comprising CVX.

4. The method for estimating the direction-of-arrival of a coprime array based on virtual domain statistics reconstruction of single-bit quantized signal according to claim 1, wherein the optimization problem in step (5) can be solved by ADMM, global optimization, approximate approximation or other methods to obtain the covariance matrix Toep() of single-bit quantized signal corresponding to the virtual uniform array .

5. The method for estimating the direction-of-arrival of a coprime array based on virtual domain statistics reconstruction of single-bit quantized signal according to claim 1, wherein the direction-of-arrival estimation in step (6) can be performed by the following method: based on the obtained Toep() single-bit receipt signal corresponding to the virtual uniform array U, the direction-of-arrival estimation can be performed by calculating a following spatial spectrum: $f\left(\theta \right)=\frac{{.Math.a_U\left(\theta \right).Math.}^2}{a_U^H\left(\theta \right)\mathrm{span}\left(\mathrm{Toep}\left(q_U^{★}\right)\right){\mathrm{span}\left(\mathrm{Toep}\left(q_U^{★}\right)\right)}^H{a}_U\left(\theta \right)},\theta \in \left[-90°,90°\right],$ wherein, (θ) is a steering vector of the virtual uniform array U corresponding to the angle μ; span(.Math.) operation is used to collect eigenvectors corresponding to all eigenvalues except the largest K eigenvalues of corresponding matrix, ∥.Math.∥ represents the Euclidean norm; finding all the maximum points in the spatial spectrum f(θ), θ∈[−90°,90°], sorting each maximum point according to the size of a response value f(θ), and taking an angle value θ corresponding to the largest K maximum points of the response value as an result of the direction-of-arrival estimation.

6. The method for estimating the direction-of-arrival of a coprime array based on virtual domain statistics reconstruction of single-bit quantized signal according to claim 1, wherein the direction-of-arrival estimation in step (6) can be processed by a traditional Nyquist methods comprising a subspace method, a sparse method, or an optimization solution method, based on the virtual domain augmented covariance matrix Toep() corresponding to the obtained single-bit quantized signal, to achieve the direction-of-arrival estimation.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0032] FIG. 1 is an overall flow diagram of the present invention.

[0033] FIG. 2 is a schematic structural diagram of a pair of sparse uniform sub-arrays forming a coprime array according to the present invention.

[0034] FIG. 3 is a schematic structural diagram of a coprime array according to the present invention.

[0035] FIG. 4 is a spatial spectrum diagram according to the method provided by the present invention, with K=8.

[0036] FIG. 5 is a spatial spectrum diagram according to the method provided by the present invention, with K=11.

[0037] FIG. 6 is a performance comparison diagram of the root mean square error changing with the signal-to-noise ratio according to the method provided by the present invention.

DESCRIPTION OF THE EMBODIMENTS

[0038] The technical schemes and effects of the present invention will be further described in detail below with reference to the drawings.

[0039] For the application of direction-of-arrival estimation technology in practical systems, especially for the application of the new generation of wireless communication systems characterized by large-scale antenna systems, the existing direction-of-arrival estimation methods face a series of challenges, such as limited degree of freedom, high system deployment cost and operation power consumption, large amount of data and complicated calculation. In order to overcome the above challenges, the present invention provides a method for estimating the direction-of-arrival of a coprime array based on virtual domain statistics reconstruction of single-bit quantized signal by integrating the advantages of coprime array signal processing and single-bit signal processing. Referring to FIG. 1, the implementation steps of the present invention are as follows:

[0040] Step 1: Use M+N−1 physical antenna elements to construct a coprime array at the receiving end, and each receiving end antenna is equipped with a single-bit analog-to-digital converter for signal reception. firstly, a pair of coprime integers M and N are selected, and then a pair of sparse uniform linear subarrays are constructed, wherein the first subarray contains M antenna elements with a spacing of Nd and positions of .sub.1={0, Nd, . . . , (M−1)Nd}, the second subarray contains N antenna elements with a spacing of Md and positions of .sub.2={0, Md, . . . , (N−1)Md}; then, the two sub-arrays are combined according to a way that the first array elements overlap, so as to obtain a non-uniform coprime array =.sub.1+.sub.2 actually containing M+N−1 physical antenna elements. Each receiving antenna is equipped with a single-bit analog-to-digital converter, which is used for binary quantization of the receipt signal.

[0041] Step 2: Modeling an single-bit receipt signal of the coprime array. Assuming that there are K far-field narrowband incoherent signal sources from directions θ.sub.1, θ.sub.2, . . . , θ.sub.K, receiving an incident signal by adopting the coprime array and the single-bit analog-to-digital converter constructed in step (1), and obtaining the single-bit receipt signal y(l)∈.sup.M+N−1 of the coprime array at the lth time, wherein the modeling is

[00007]$y\left(l\right)=\mathrm{��}\left(x\left(l\right)\right)=\mathrm{��}\left(\underset{k=1}{\overset{K}{.Math.}}{a}_S\left({\theta }_k\right)s_k\left(l\right)+n_S\left(l\right)\right),$

[0042] wherein, (.Math.) is a single-bit quantization operator, x(l) is an unquantized original receipt signal of the coprime array, s.sub.k(l) is a waveform of a k.sup.th signal, (l) is a noise item independent of each signal source, and (θ.sub.k) is a steering vector of the coprime array corresponding to the direction, θ.sub.k, expressed as

(θ.sub.k)=┌1,e.sup.−jπu.sup.2 .sup.sin(θ.sup.k.sup.), . . . ,e.sup.−jπu.sup.M+N−1 .sup.sin(θ.sup.k.sup.)┐.sup.T,

[0043] wherein, u.sub.i, i=1, 2, . . . , M+N−1 represents an actual position of the i.sup.th physical antenna element in the coprime array, and u.sub.1=0, j=√{square root over (−1)}, [.Math.].sup.T represents a transposition operation, and a sampling covariance matrix of the single-bit receipt signal of the coprime array is obtained by using collected L sampling snapshots, expressed as

[00008]${\hat{Q}}_S=\frac{1}{L}\underset{l=1}{\overset{L}{.Math.}}y\left(l\right)y^H\left(l\right),$

[0044] wherein (.Math.).sup.H represents conjugate transpose.

[0045] Step 3: Calculating an equivalent virtual signal corresponding to the single-bit receipt signal of the coprime array. Vectorizing the sampling covariance matrix of the single-bit receipt signal of the coprime array, and obtaining the equivalent receipt signal of a virtual array corresponding to the single-bit quantized signal, expressed as

[00009]${\hat{q}}_D=\mathrm{vec}\left({\hat{Q}}_S\right)=\frac{2}{\pi }\left(\arcsin \left(\left({\Sigma }^{-\frac{1}{2}}.Math.{\Sigma }^{-\frac{1}{2}}\right)ℛ\left(r_D\right)\right)+j.Math.\arcsin \left(\left({\Sigma }^{-\frac{1}{2}}.Math.{\Sigma }^{-\frac{1}{2}}\right)\mathrm{��}\left(r_D\right)\right)\right),$

[0046] corresponding to a non-uniform virtual array,

={v.sub.1−v.sub.D|v.sub.D,v.sub.D∈},

[0047] wherein, vec(.Math.) represents a vectorization operation, that is, columns in the matrix are stacked in sequence to form a new vector, Σ=diag(), =E[x(l)x.sup.H(l)] is the covariance matrix of the receipt signal of a unquantized original coprime array, diag(.Math.) represents operation of taking diagonal elements to form a diagonal matrix, E[.Math.] represents taking expectation operation, .Math. represents Kronecker product, (.Math.) represents taking real part operation, (.Math.) represents taking imaginary part operation,

[00010]$r_D=\underset{k=1}{\overset{K}{.Math.}}{\sigma }_k^2{a}_D\left({\theta }_k\right)+{\sigma }_n^2\mathrm{vec}\left(I\right),$

[0048] wherein, σ.sub.k.sup.2 represents power of the k.sup.th signal source, σ.sub.n.sup.2 represents noise power, I is an identity matrix, (θ.sub.k) is a steering vector of a non-uniform virtual array corresponding to the direction θ.sub.k, and is calculated as (θ.sub.k)=(θ.sub.k).Math.(θ.sub.k), wherein (.Math.)* is conjugate operation.

[0049] Step 4: Constructing a virtual domain augmented covariance matrix of an initialized single-bit quantized signal. In order to overcome a signal model mismatch problem caused by the non-uniform virtual array of the coprime array, constructing a virtual domain uniform linear array with a same aperture as a positive half axis of the non-uniform virtual array and a spacing d, wherein the unit spacing d is half of the wavelength of incident narrowband signal, expressed as

={ud|u=0,1,2, . . . ,max()/d},

[0050] wherein, max(.Math.) represents an operation of taking the set maximum, correspondingly, the equivalent virtual signal ∈ corresponding to the virtual domain uniform linear array is obtained by the following method: for the equivalent virtual signal corresponding virtual array element position corresponding to , if the virtual array element position is included in the non-uniform virtual array , the equivalent virtual signal at this position is the same as a virtual signal corresponding to the corresponding virtual array element position in ; the equivalent virtual signal corresponding to discontinuous virtual array elements in the remaining non-uniform virtual arrays are set to zero, and then virtual domain augmented covariance matrix of the initialized single-bit quantized signal can be constructed as

=Toep(),

[0051] wherein, Toep(.Math.) represents that a vector taken is the first column of the Hermitian Toeplitz matrix.

[0052] Step 5: designing, based on statistical correlation analysis between statistics of the single-bit quantized signal and the original unquantized signal, an optimization problem based on virtual domain statistics reconstruction of quantized signal to obtain a single-bit quantized signal covariance matrix corresponding to a virtual uniform array ; according to statistical characteristic analysis, based on the characteristic that the single-bit quantized signal covariance matrix =E[y(l)y.sup.H(l)] is the same as a maximum linear independent set of an original unquantized signal covariance matrix , indicating that a rank of covariance matrix does not change in a single-bit quantization process, and the virtual domain augmented covariance matrix corresponding to the single-bit quantized signal derived from can be regarded as a sampling covariance matrix calculated from the single-bit receipt signal of the virtual uniform array , but some elements are missing; under an ideal condition that all elements are known, the covariance matrix will still retain the matrix rank information related to the incident signal source; based on statistical correlation analysis among the above statistics, the reconstruction problem of augmented covariance matrix can be constrained and optimized by using low rank characteristics of the covariance matrix, and then the following optimization problem for virtual domain statistics reconstruction of single-bit quantized signal is constructed taking as an optimization target:

[00011]$q_U^{\mathrm{.star-solid.}}=\arg \underset{q_U\in C^{\left[U\right]}}{\min }{.Math.{\mathrm{��}}_{\Omega }\left(\mathrm{Toep}\left(q_U\right)\right)-{\hat{Q}}_U.Math.}_F+\lambda .Math.\mathrm{rank}\left(\mathrm{Toep}\left(q_U\right)\right)$$\mathrm{subject}\mathrm{to}\mathrm{Toep}\left(q_U\right)\succcurlyeq 0,$

[0053] wherein, .sub.Ω(.Math.) represents a projection operation used to select the element in Toep(q.sub.U) corresponding to the position of non-zero elements for fitting, λ is a user adjustment parameter, Toep()0 is a positive semi-definite matrix constraint item, ∥.Math.∥.sub.F represents a Frobenius norm, rank(.Math.) represents the rank of the matrix.

[0054] The above optimization problems can be solved by introducing various convex relaxation techniques, for example, by replacing the penalty term rank(Toep()) in the above optimization problems with the convex function term trace(Toep()), where trace(.Math.) represents the trace of matrix, then it can be solved by various interior point methods such as CVX. In addition, the above optimization problems can also be solved by ADMM, global optimization, approximate approximation or other methods to obtain the covariance matrix Toep() of single-bit quantized signals corresponding to the virtual uniform array .

[0055] Step 6: Estimating the direction-of-arrival by using the virtual domain augmented covariance matrix corresponding to the optimized single-bit quantized signal. Based on the obtained Toep() single-bit receipt signal corresponding to the virtual uniform array U, the direction-of-arrival estimation can be performed by calculating a following spatial spectrum:

[00012]$f\left(\theta \right)=\frac{{.Math.a_U\left(\theta \right).Math.}^2}{a_U^H\left(\theta \right)\mathrm{span}\left(\mathrm{Toep}\left(q_U^{\mathrm{.star-solid.}}\right)\right){\mathrm{span}\left(\mathrm{Toep}\left(q_U^{\mathrm{.star-solid.}}\right)\right)}^H{a}_U\left(\theta \right)},\theta \in \left[-90°,90°\right],$

[0056] wherein, (θ) is a steering vector of the virtual uniform array U corresponding to the angle μ; span(.Math.) operation is used to collect eigenvectors corresponding to all eigenvalues except the largest K eigenvalues of corresponding matrix, ∥.Math.∥ represents the Euclidean norm; finding all the maximum points in the spatial spectrum f(θ), δ∈[−90°,90°], sorting each maximum point according to the size of a response value f(θ), and taking an angle value θ corresponding to the largest K maximum points of the response value as an result of the direction-of-arrival estimation.

[0057] In addition, the direction-of-arrival estimation in step (6) can be processed by traditional Nyquist methods, such as subspace method, sparse method, optimization solution method and the like, based on the virtual domain augmented covariance matrix Toep() corresponding to the obtained single-bit quantized signal, to achieve the direction-of-arrival estimation.

[0058] On the one hand, the present invention makes full use of the advantages of coprime array virtual domain signal processing, makes full use of all discontinuous virtual array elements, and at the same time, realizes the improvement of direction-of-arrival estimation degree of freedom from (M+N) to (MN), so that compared with the traditional Nyquist method, the provided method can estimate more incident signal sources under the same number of antennas, reduces the number of radio frequency channels in hardware deployment, and reduces the data scale and computational complexity of receipt signals. On the other hand, the present invention realizes the sparse array virtual domain signal processing based on the single-bit quantized signal by using the single-bit signal processing technology, and performs the direction-of-arrival estimation efficiently by using the covariance rectangle of the single-bit quantized signal corresponding to the augmented virtual array according to the statistical relevance analysis of the quantized signal statistics. At the same time, the method for estimating the direction-of-arrival of single bit in the present invention reduces the average power consumption from several watts for mainstream 12-16-bit analog-to-digital converters in the existing system to several milliwatts, thus greatly reducing the power consumption of the system and avoiding the adverse influence caused by the error between idealized modeling and limited precision quantization of the traditional method.

[0059] The effect of the provided method will be further described with a simulation example.

[0060] Simulation example 1: The parameters of the coprime array are selected as M=3 and N=5, that is, the coprime array of the architecture contains ||=7 antenna elements. It is assumed that the incident direction of the incident narrowband signal is uniformly distributed in [−50°,50°] the signal-to-noise ratio is 0 dB, the sampling snapshot number is L=500, and the user adjustment parameter λ is 0.25. The spatial spectrums of the method for estimating the direction-of-arrival of a coprime array based on virtual domain statistics reconstruction of single-bit quantized signal under the underdetermined conditions K=||+1=8 and K=|| are shown in FIG. 4 and FIG. 5, respectively, where the dashed line represents the real direction-of-arrival of the incident signal. It can be seen from the simulation results that the provided method can effectively distinguish all incident signal sources under the above two conditions, which shows that the degree of freedom is still improved under the background of single-bit quantization signal processing. In addition, the results shown in FIG. 5 further illustrate that the present invention can effectively utilize all discontinuous array elements in the non-uniform array to process Nyquist signals in the virtual domain, thereby achieving the maximum degree of freedom performance.

[0061] Simulation example 2: The parameters of the coprime array are selected as M=3 and N=5, that is, the coprime array of the architecture contains ||=7 antenna elements. It is assumed that the K=10<|| incident directions of narrowband incident signals are uniformly distributed in [−50°,50°] the sampling snapshot number is L=500, and the user adjustment parameter λ is 0.25. A schematic diagram of the performance comparison of the root mean square error versus the signal-to-noise ratio between the method provided by the present invention and the existing sparse signal reconstruction and virtual domain interpolation methods based on non-quantized signals is shown in FIG. 6, in which the pre-defined space grid point spacing for sparse reconstruction and spectral peak search in the contrast method is 0.1, and the root mean square error is obtained by averaging 1000 Monte Carlo tests for each SNR simulation parameter setting. At the same time, the Cramer bound based on virtual domain signal processing is also given to represent the optimal value. It can be seen from the comparison results shown in FIG. 6 that when the signal-to-noise ratio is greater than 5 dB, the method provided by the present invention can still obtain better performance than the non-quantized signal processing method under the condition of single-bit quantization, and keep the same performance trend as the Cramer-Rao bound of virtual domain signal processing.

[0062] In summary, the present invention mainly overcomes the shortcomings of the prior art in terms of degree of freedom and computational complexity, system deployment cost, power consumption, data scale and the like. On the one hand, the virtual domain signal processing characteristics of the coprime array are fully utilized to realize the increase of freedom; on the other hand, based on the design optimization problem of single-bit quantization signal modeling and statistical correlation analysis, the advantages of single-bit signal processing and coprime array signal processing are integrated, which has broad application prospects in practical applications for new generation wireless communication systems, passive positioning, target detection and other applications.