Target parameter super-resolution estimation method based on OCDM ISAC system

Abstract

The invention provides a target parameter super-resolution estimation method based on an OCDM ISAC system. The target parameter super-resolution estimation method mainly solves the problems that in the prior art, distance and speed resolution are limited by time-frequency resources occupied by the system, and sensing precision is poor. According to the scheme adopted by the invention, modulation symbol elimination is carried out on echo signals, after distance and speed coarse estimation is carried out, combined super-resolution estimation of the target distance and speed is realized by adopting a subspace projection method, and the sensing precision of an ISAC system is improved on the premise that the time-frequency overhead of the system is not increased.

Description

Technical Field

[0001] The present invention belongs to the technical field of wireless communication, and mainly relates to a super-resolution estimation method for target distance and speed. Background Art

[0002] Compared with traditional wireless communication and sensing systems, the ISAC (Integrated Sensing and Communication) system can effectively improve the utilization rate of spectrum and hardware devices, and is regarded as one of the key technologies in the 5th Generation Mobile Communication Technology (5G).

[0003] In the ISAC system, the design of the integrated waveform is one of the key issues. The most common one is the integrated waveform based on Orthogonal Frequency Division Multiplexing (OFDM). However, OFDM is more sensitive to Doppler frequency shift. In contrast, Orthogonal Chirp Division Multiplexing (OCDM) has stronger anti-interference ability, higher robustness to Doppler frequency shift, and is more suitable for a large number of high-dynamic scenarios in 5G. Therefore, in terms of communication performance, OCDM is superior to OFDM and is regarded as a more ideal alternative to OFDM. For the sensing function, in the ISAC system based on OCDM, the distance and speed parameters of the target can be obtained by processing the echo signal, so as to achieve the sensing function.

[0004] The literature "Wang Jingqi, Zeng Huan, Tao Zhan, etc. A new radar-communication integrated system based on OCDM [J]. Journal of Microwaves, 2022, 38(06): 14-18." proposed an OCDM radar signal processing method based on 2D Discrete Fourier Transform (DFT), which can effectively remove random communication data and greatly reduce the computational complexity.

[0005] Reference 2 "DE OLIVEIRA L G, ALABD M B, NUSS B, et al. An OCDM radar-communication system[C] / / 2020 14th European Conference on Antennas and Propagation (EuCAP), Copenhagen, Denmark. 2020:1-5." After removing the influence of data symbols, digital pulse compression and target range-velocity estimation are achieved using zero-padding DFT.

[0006] Reference 3 "BHATTACHARJEE S, MISHRA K V, ANNAVAJJALA R, et al. Evaluation of Orthogonal Chirp Division Multiplexing for Automotive Integrated Sensing and Communications[C] / / 2022 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Singapore. 2022:8742-8746" After the received radar signal is processed such as down-conversion, a matched filter is performed to obtain an approximate expression of the radar received signal. Based on this, a likelihood function is constructed, and the parameters that maximize the function are found using a two-dimensional complex periodogram, and then the target range and velocity estimation values are obtained.

[0007] References 1 and 2 are in the self-transmitting and self-receiving mode, and target range and velocity estimation are based on DFT. The operation is simple and easy to implement, but the range and velocity resolution accuracy are limited by the time-frequency resources occupied by the system; Reference 3 considers bistatic ISAC and estimates target parameters based on the matched filter method, but discretizes the range and velocity, with low sensing accuracy and complex calculations. Therefore, to improve the sensing accuracy, the present invention proposes a super-resolution estimation method for target parameters based on an OCDM ISAC system to achieve high-precision target parameter estimation. Summary of the Invention

[0008] To overcome the deficiencies of the prior art, address the limitations in the prior art where range and velocity resolution are restricted by the time-frequency resources occupied by the system, and improve the sensing accuracy, the present invention provides a method for super-resolution estimation of target parameters based on an OCDM ISAC system. After performing modulation symbol cancellation and rough estimation of range and velocity on the echo signal, a subspace projection method is used to achieve joint super-resolution estimation of the target range and velocity, enhancing the sensing accuracy of the ISAC system without increasing the time-frequency overhead of the system.

[0009] A method for super-resolution estimation of target parameters based on an OCDM ISAC system, comprising the following steps:

[0010] Step 1: Use a 2D DFT method to process the sampling results of all symbols in the p-th baseband pulse and the sampling results of the received signal of all symbols in the p-th pulse, obtaining a rough estimation result of the target range and a rough estimation result of the target velocity;

[0011] Step 2: Utilize the rough estimation result of the target range R i and the rough estimation result of the target velocity v i to compensate the sampling result of the received signal of the s-th symbol in the p-th pulse with the known modulation data symbols, obtaining a vector of the s-th symbol;

[0012] Step 3: For the vector of the s-th symbol, use the subspace projection method to perform super-resolution estimation of the target range and velocity, obtaining a super-resolution estimation of the target range and a super-resolution estimation of the velocity;

[0013] Step 4: Use a deblurring method to process the super-resolution estimation of the target range and the super-resolution estimation of the velocity, obtaining a final estimation result of the target range and a final estimation result of the velocity.

[0014] Furthermore, obtaining the rough estimation result of the target range and the rough estimation result of the target velocity The specific process is as follows:

[0015] Step 1-1: To highlight the effects of time delay and Doppler frequency shift, ignore the time delays caused by the pulse period and symbol period, and represent the s-th transmitted symbol x p,s (t) in the p-th pulse as:

[0016]

[0017] where t is the time and j is the imaginary unit; among N p OCDM pulse signals, each pulse contains N s OCDM symbols, and the signal parameters include the number of subcarriers N and the carrier frequency f c, symbol period \(T\) s and pulse period \(T\) p , where \(p = 0, 1, \ldots, N\) p \(-1, s = 0, 1, \ldots, N\) s \(-1\), and \(X(k)\) is the data symbol modulated on the \(k\)th sub - carrier;

[0018] Sampling the \(s\)th transmitted symbol \(x\) p,s (t) in the \(p\)th pulse, then the sampling result \(x\) p,s (n) is expressed as:

[0019]

[0020] where \(n = 0, 1, \ldots, N - 1\);

[0021] Based on Equation (1), the received signal \(y\) p,s,i (t) of the \(i\)th target is:

[0022]

[0023] where \(A\) i is the complex amplitude factor of the attenuation and phase shift that occur during the propagation and scattering of the \(i\)th target; \(\tau\) i is the time delay corresponding to the \(i\)th target, \(f\) d,i is the Doppler frequency shift of the \(i\)th target, \(c\) is the speed of light, \(f\) c is the carrier frequency. Using monostatic ISAC, assuming there are \(N\) t targets, the distance between the \(i\)th target and the transmitter is \(R\) i , and the relative speed is \(v\) i , \(i = 1, 2, \ldots, N\) t ;

[0024] To simplify the derivation, the noise effect is ignored in Equation (3), and sampling the received signal \(y\) p,s,i (t) gives the sampled signal \(y\) p,s,i (n) of the \(i\)th target:

[0025]

[0026] In Equation (4), compared with the phase shift , the phase shift within the symbol period is smaller and is ignored, and Equation (4) is expressed as:

[0027]

[0028] Performing Fourier transform on Equation (2) gives the Fourier transform \(F\) of the sampling result of the \(s\)th symbol in the \(p\)th pulse in the baseband tx,s(k), the Fourier transform of the sampling result of the received signal of the s-th symbol of the p-th pulse is obtained by performing a Fourier transform on Equation (5), denoted as F rx,s (k):

[0029]

[0030] where DFT(·) performs a DFT operation on the content in the parentheses, and k = 0, 1, …, N - 1;

[0031] After performing DFT on the N sampling results of the s-th symbol, the intermediate variable F tx,s and the intermediate variable F rx,s are obtained:

[0032]

[0033] where is an N-dimensional complex space;

[0034] According to Equation (7), the transmit matrix F tx and the receive matrix F rx are expressed as:

[0035]

[0036] where is an N×N s dimensional complex space;

[0037] Step 1-2: Let That is, let the receive matrix be multiplied pointwise by the conjugate matrix of the transmit matrix to remove the influence of communication information on radar detection, where ⊙ is the Hadamard product, and (·) H is the conjugate transpose of the matrix; then F is expressed as:

[0038]

[0039] where F is an intermediate variable, is the delay vector of the i-th target, expressed as:

[0040]

[0041] whose elements are and

[0042] is the Doppler vector of the i-th target, expressed as:

[0043]

[0044] whose elements are and is N s dimensional complex space;

[0045] Step 1-3: From Equation (10), the distance information R of the target i is included in the linear phase shifts of different subcarriers of the same OCDM symbol. By calculating the inverse discrete Fourier transform IDFT, the distance estimate of the reflecting target is obtained; from Equation (11), the Doppler information f of the target d,i is included in the linear phase shifts between the same subcarriers in the frequency domain of adjacent OCDM symbols. By calculating the DFT, the velocity estimate of the reflecting target is obtained; and from Equation (9), the distance influence and Doppler influence of the target are completely orthogonal in the frequency domain; therefore, by performing IDFT and DFT on the columns and rows of matrix F respectively, and then performing a spectral peak search, the distance and velocity estimates of the target are obtained;

[0046] Let the s-th column of the intermediate variable F be f s , where s = 0, 1,..., N s -1. Perform IDFT on each column of F to obtain is the IDFT matrix of F, that is:

[0047]

[0048] where IDFT(·) performs the IDFT operation on the content in the parentheses, and then performs DFT on each row of to obtain is the DFT matrix of; let the n-th row of be where n = 0, 1,..., N - 1, then there is:

[0049]

[0050] where

[0051] Let the element in the l-th row and m-th column of be f l,m , let:

[0052]

[0053] then:

[0054]

[0055] where is rounding down;

[0056]

[0057] Thus, the distance R of the i-th target is obtained i of the rough estimation result and the velocity v of the i-th target i of the rough estimation result

[0058] Furthermore, using the distance R i of the rough estimation result of the target distance the velocity v i of the rough estimation result of the target velocity and the sampling result y of the received signal of the s-th symbol of the p-th pulse with the known modulation data symbol pair p,s to perform compensation, and the vector r corresponding to the s-th symbol is obtained s , and the specific process is as follows:

[0059] Step 2-1: Based on the received signal of the s-th symbol of the p-th pulse and the sampling result of the received signal of the s-th symbol of the p-th pulse, obtain y p,s,i is the i-th component of y p,s , and Taking the i-th target as an example, y p,s,i is expressed as:

[0060]

[0061] wherein, is an intermediate variable, and is an N×N-dimensional complex space;

[0062] is an intermediate variable, and

[0063] D c =diag[X(0),X(1),…,X(N-1)], D c is an intermediate variable, and

[0064] is the distance steering vector, and

[0065] w i =[w i (0),w i (1),…,w i (N-1)] T , w iis a noise vector, and

[0066] Φ is a discrete Fresnel inverse transform matrix, and the element in the n-th row and k-th column of it is diag(·) is an operation for constructing a diagonal matrix;

[0067] Step 2-2: Use the rough estimation result i of the distance R and the rough estimation result i of the velocity v to calculate the rough estimation result of the time delay and the rough estimation result of the Doppler

[0068]

[0069] Obtain the time delay compensation matrix and the Doppler compensation matrix

[0070]

[0071] where Since Φ is a unitary matrix and D c is known, use Φ, D c , the time delay compensation matrix and the Doppler compensation matrix to compensate equation (18), then left multiply equation (18) by where is an intermediate variable, and let:

[0072]

[0073] where D p,s,i is an intermediate variable, define the compensation vector of y p,s,i as r p,s,i , then r p,s,i is expressed as:

[0074]

[0075] and let:

[0076]

[0077] where r p,s is the sum vector of r p,s.i ;

[0078] Step 2-3: In order to jointly estimate the distance and velocity of the target simultaneously, arrange the s-th symbols in all pulses into a column, then the vector r s corresponding to the s-th symbol is obtained:

[0079]

[0080] In the formula, is an intermediate variable, and where is N p N×N p N-dimensional complex space;

[0081] is the range-velocity steering vector, is the Kronecker product of matrices;

[0082] is the velocity steering vector, and is N p dimensional complex space;

[0083] is the range steering vector, and

[0084] Furthermore, for the vector r corresponding to the s-th symbol s , s = 0, 1, …, N s -1, the subspace projection method is used for target range and velocity super-resolution estimation, and the super-resolution estimation of the target range and the super-resolution estimation of the velocity are obtained. The specific process is as follows:

[0085] Step 3-1: Estimate the covariance matrix s from the data of N

[0086]

[0087] Step 3-2: Perform eigenvalue decomposition on , that is, let where Σ is the diagonal matrix composed of the eigenvalues of where the eigenvalues are arranged from large to small, that is, Each column of U is respectively the eigenvector corresponding to the eigenvalue ;

[0088] Step 3-3: Use the existing target source number estimation method in the field of array signal processing to obtain the target number N t , and take the eigenvectors corresponding to the first N t eigenvalues to construct the matrix

[0089] Step 3-4: Discretize R and v at fixed intervals, and calculate the corresponding τ and f d , obtaining the distance-velocity steering vector Construct Search for peaks higher than the threshold γ R,v , and then use the time delay corresponding to the peak and Doppler frequency shift to calculate the super-resolution estimate of the target distance and the super-resolution estimate of the velocity

[0090] where λ is the wavelength, λ = c / f c .

[0091] Furthermore, use the deblurring method to process the super-resolution estimate of the target distance and the super-resolution estimate of the velocity to obtain the final estimated result of the target distance and the final estimated result of the velocity The specific process is as follows:

[0092] Step 4-1: Use the method described in Step 3 to obtain the distance estimate of the target The maximum unambiguous distance of this method is relatively small. To perform distance deblurring, use the pulse compression method to estimate the target distance and combine the two. To reduce the error generated during combination, let:

[0093]

[0094] where α is the distance ambiguity coefficient, is the ambiguous distance estimate, and let q = -1, 0, 1. For different values of q, calculate the absolute value of the difference between and , and take the one with the smallest absolute value of the difference as the final estimated result of the target distance

[0095] Step 4-2: After using the method described in Step 3 to obtain the velocity estimate of the target , the maximum unambiguous velocity is Perform velocity deblurring on , and use the 2D DFT method to obtain the target velocity estimate and combine the two. Let:

[0096]

[0097] where β is the velocity ambiguity coefficient, For fuzzy velocity estimation, for different values of q, calculate respectively and the absolute value of the difference, and take the one with the smallest absolute value of the difference as the final estimated result of the target distance

[0098] The beneficial effects of the present invention are as follows: By compensating the echo signal, the present invention realizes the super-resolution estimation of the target distance and velocity by using the subspace projection method, and combines the pulse compression and 2D DFT methods to eliminate the range and velocity ambiguities. Through simulation verification, the results prove that the method proposed by the present invention can accurately estimate the target parameters and significantly improve the system perception accuracy. Description of the Drawings

[0099] Figure 1 is the OCDM transmitted signal structure diagram;

[0100] Figure 2 is the flowchart of the super-resolution estimation of target parameters;

[0101] Figure 3 is the simulation result of the joint range-velocity estimation, where (a) is the 2D DFT method and (b) is the super-resolution method;

[0102] Figure 4 is the simulation result of range ambiguity resolution, where (a) is before ambiguity resolution and (b) is after ambiguity resolution;

[0103] Figure 5 is the simulation result of velocity ambiguity resolution, where (a) is before ambiguity resolution and (b) is after ambiguity resolution;

[0104] Figure 6 is the relative error of parameter estimation, where (a) is the relative error of range estimation and (b) is the relative error of velocity estimation. Detailed Implementation Manner

[0105] According to the application requirements, determine the signal parameters. The form of the transmitted signal is as Figure 1 shown, and transmit N p OCDM pulse signals, and each pulse contains N s OCDM symbols. The signal parameters include the number of subcarriers N, the carrier frequency f c , the symbol period T s and the pulse period T p ;

[0106] Assume that the continuous time-domain expression of the p-th pulse and the s-th symbol in the baseband is:

[0107] pT p +sT s<t≤pT p +(s + 1)T s

[0108] where p = 0, 1, …, N p -1, s = 0, 1, …, N s -1, and X(k) is the data symbol modulated on the k-th subcarrier;

[0109] Using a monostatic ISAC, assume there are N t targets, the distance between the i-th target and the transmitter is R i , and the relative velocity is v i , i = 1, 2, …, N t , and without considering multipath signals, the received signal of the s-th symbol of the p-th pulse is:

[0110]

[0111] where, A i is the complex amplitude factor of the attenuation and phase shift that occurs during the propagation and scattering of the i-th target, τ i is the time delay corresponding to the i-th target, f d,i is the Doppler frequency shift of the i-th target, c is the speed of light, f c is the carrier frequency, w(t) is Gaussian white noise; sampling x p,s (t) and y p,s (t) with a sampling period of T s / N, the sampling result of the s-th symbol of the p-th pulse in the baseband is

[0112]

[0113] The sampling result of the received signal of the s-th symbol of the p-th pulse is

[0114]

[0115] where

[0116] Figure 2 illustrates the flowchart of the super-resolution estimation of target parameters;

[0117] A method for super-resolution estimation of target parameters based on an OCDM ISAC system, comprising the following steps:

[0118] Step 1: Use the 2D DFT method to process the sampling results of the N s symbols in the p-th pulse in the baseband and the sampling results of the received signals of the N s symbols in the p-th pulse to obtain a rough estimation result of the target distance and the rough estimation result of the target speed The specific process is as follows:

[0119] Step 1-1: To highlight the effects of time delay and Doppler frequency shift, ignore the time delay caused by the pulse period and symbol period, and represent the sth transmitted symbol x p,s (t) in the pth pulse as:

[0120]

[0121] where t is the time and j is the imaginary unit;

[0122] where among N p OCDM pulse signals, each pulse contains N s OCDM symbols, and the signal parameters include the number of subcarriers N, the carrier frequency f c , the symbol period T s and the pulse period T p , where p = 0, 1, …, N p - 1, s = 0, 1, …, N s - 1, and X(k) is the data symbol modulated on the kth subcarrier;

[0123] Sample the sth transmitted symbol x p,s (t) in the pth pulse, then the sampling result x p,s (n) is represented as:

[0124]

[0125] where n = 0, 1, …, N - 1;

[0126] Based on Equation (1), the received signal y p,s,i (t) of the ith target is:

[0127]

[0128] where A i is the complex amplitude factor of the attenuation and phase shift that occurs during the propagation and scattering of the ith target; τ i is the time delay corresponding to the ith target, f d,i is the Doppler frequency shift of the ith target, c is the speed of light, f c is the carrier frequency. Using monostatic ISAC, assume there are N t targets, the distance between the ith target and the transmitter is R i , and the relative speed is v i , i = 1, 2, …, N t ;

[0129] To simplify the derivation, the noise effect is ignored in Equation (3), and the received signal y p,s,i (t) is sampled to obtain the sampled signal y p,s,i (n) of the i-th target:

[0130]

[0131] In Equation (4), compared with the phase shift , the phase shift within the symbol period is relatively small and is ignored, and Equation (4) is expressed as:

[0132]

[0133] The Fourier transform of Equation (2) is performed to obtain the Fourier transform F tx,s (k) of the sampled result of the p-th pulse and the s-th symbol in the baseband. The Fourier transform of Equation (5) is performed to obtain the Fourier transform F rx,s (k) of the sampled result of the received signal of the p-th pulse and the s-th symbol:

[0134]

[0135] where DFT(·) performs the DFT operation on the content in the parentheses, and k = 0, 1,..., N - 1;

[0136] After performing the DFT on the N sampled results of the s-th symbol, the intermediate variable F tx,s and the intermediate variable F rx,s are obtained:

[0137]

[0138] where is the N-dimensional complex space;

[0139] According to Equation (7), the transmit matrix F tx and the receive matrix F rx are expressed as:

[0140]

[0141] where is the N×N s dimensional complex space;

[0142] Step 1-2: Let That is, let the receive matrix be multiplied pointwise by the conjugate matrix of the transmit matrix to eliminate the influence of communication information on radar detection, where ⊙ is the Hadamard product, and (·) H is the conjugate transpose of the matrix; then F is expressed as:

[0143]

[0144] where F is an intermediate variable, is the time delay vector of the i-th target, expressed as:

[0145]

[0146] whose elements are and

[0147] is the Doppler vector of the i-th target, expressed as:

[0148]

[0149] whose elements are and is N s dimensional complex space;

[0150] Step 1-3: As can be seen from Equation (10), the distance information R of the target i is included in the linear phase shifts of different subcarriers of the same OCDM symbol. By calculating the inverse discrete Fourier transform (IDFT) of, the distance estimate of the reflected target is obtained; as can be seen from Equation (11), the Doppler information f of the target d,i is included in the linear phase shifts between the same subcarriers in the frequency domain of adjacent OCDM symbols. By calculating the DFT of, the velocity estimate of the reflected target is obtained; and as can be seen from Equation (9), the distance influence and Doppler influence of the target are completely orthogonal in the frequency domain; therefore, by performing IDFT and DFT on the columns and rows of matrix F respectively, and then performing spectral peak search, the distance and velocity estimates of the target are obtained;

[0151] Let the s-th column of the intermediate variable F be f s , where s = 0, 1,..., N s -1. Perform IDFT on each column of F to obtain is the IDFT matrix of F, that is:

[0152]

[0153] where IDFT(·) performs the IDFT operation on the content in the parentheses, and then performs DFT on each row of to obtain is DFT matrix; Let The n-th row of where n = 0, 1, …, N - 1, then there is:

[0154]

[0155] where

[0156] Let The element in the l-th row and m-th column of l,m is f

[0157]

[0158] Then:

[0159]

[0160] where is floor function, thus obtaining the rough estimation result of the i-th target distance R i and the rough estimation result of the i-th target speed v i i

[0161]

[0162]

[0162] Step 2: To apply the subspace projection method for super-resolution estimation, use the rough estimation result of the target distance R i the rough estimation result of the target speed v i and the known modulation data symbols to compensate the sampling result y p,s of the received signal of the s-th symbol of the p-th pulse, obtaining the vector r s corresponding to the s-th symbol. The specific process is as follows: s

[0163] Step 2-1: Based on the received signal of the s-th symbol of the p-th pulse and the sampling result of the received signal of the s-th symbol of the p-th pulse, obtain y p,s,i is the i-th component of y p,s and Taking the i-th target as an example, y p,s,i is expressed as:

[0164]

[0165] where,

[0166] is an intermediate variable, and is an N×N dimensional complex space;

[0167] is an intermediate variable, and

[0168] D c = diag[X(0), X(1), …, X(N - 1)] is an intermediate variable, and

[0169] is the distance steering vector, and

[0170] w i = [w i (0), w i (1), …, w i (N - 1)] T is the noise vector, and

[0171] Φ is the discrete Fresnel inverse transform matrix, and its (n, k)-th element is diag(·) is the operation of constructing a diagonal matrix;

[0172] Step 2 - 2: Using the rough estimation result i of the distance R and the rough estimation result i of the velocity v calculate the rough estimation result of the time delay and the rough estimation result

[0173]

[0174] Obtain the time delay compensation matrix and the Doppler compensation matrix

[0175]

[0176] where Since Φ is a unitary matrix and D c is known, using Φ, D c , the time delay compensation matrix and the Doppler compensation matrix to compensate Equation (18), then left - multiply Equation (18) by where is an intermediate variable, and let:

[0177]

[0178] where D p,s,i is an intermediate variable, define the compensation vector of y p,s,i as rp,s,i , then r p,s,i is expressed as:

[0179]

[0180] And let:

[0181]

[0182] where r p,s is the sum vector of r p,s.i ;

[0183] Step 2-3: In order to jointly estimate the distance and speed of the target, arrange the s-th symbols in all pulses into a column, then the vector r s corresponding to the s-th symbol is obtained:

[0184]

[0185] In the formula, is an intermediate variable, and

[0186] is the range-velocity steering vector, is the Kronecker product of matrices;

[0187] is the velocity steering vector, and

[0188] is the range steering vector, and

[0189] Step 3: For the vector r s corresponding to the s-th symbol, s = 0, 1,..., N s -1, use the subspace projection method to perform super-resolution estimation of the target distance and speed, and obtain the super-resolution estimation of the target distance and the super-resolution estimation of the speed. The specific process is as follows:

[0190] Step 3-1: Estimate the covariance matrix s from the data of N

[0191]

[0192] Step 3-2: Perform eigenvalue decomposition on , that is, let where Σ is The diagonal matrix formed by the eigenvalues, i.e., where the eigenvalues are arranged from largest to smallest, i.e., Each column of U is respectively the eigenvector corresponding to the eigenvalue;

[0193] Step 3-3: Using the existing target source number estimation method in the field of array signal processing, obtain the target number N t , take the first N t eigenvectors corresponding to the eigenvalues to construct a matrix

[0194] Step 3-4: Discretize R and v at fixed intervals, calculate the corresponding τ and f d , obtain the range-velocity steering vector Construct Search for peaks higher than the threshold γ R,v , and then use the time delay corresponding to the peak and the Doppler frequency shift to calculate the super-resolution estimation of the target range and the super-resolution estimation of the velocity

[0195] where λ is the wavelength, λ = c / f c ;

[0196] Step 4: Use the deblurring method to process the super-resolution estimation of the target range and the super-resolution estimation of the velocity to obtain the final estimation results of the target range and the final estimation results of the velocity The specific process is as follows:

[0197] Step 4-1: Use the method described in Step 3 to obtain the distance estimation of the target The maximum unambiguous distance of this method is relatively small. To perform range deblurring on it, use the pulse compression method to estimate the target range and combine the two. To reduce the error generated during the combination, let:

[0198]

[0199] where α is the range ambiguity coefficient, is the ambiguous distance estimation, and let q = -1, 0, 1. For different values of q, calculate the absolute value of the difference between and As the final estimated result of the target distance

[0200] Step 4-2: After obtaining the speed estimation of the target by using the method described in Step 3 the maximum unambiguous speed is Perform to resolve the speed ambiguity, and use the 2D DFT method to obtain the target speed estimation and combine the two, let

[0201]

[0202] where β is the speed ambiguity coefficient is the ambiguous speed estimation. For different values of q, calculate and the absolute value of the difference, and take the one with the smallest absolute value of the difference as the final estimated result of the target distance

[0203] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0204] The effect of the present invention can be further illustrated by Figures 3 - 6 the simulation results. In the simulation experiment, set N = 256, N s = 32, N p = 8, T s = 5.12us, T p = 4.1ms, f c = 5GHz.

[0205] Figure 3 is the result of the joint distance and speed estimation of two targets with distances and speeds of 93m, 20m / s and 95m, 25m / s respectively. Figure 3 (a) is the result of estimating the target parameters by using the traditional DFT-based method. It can be seen that there is only one peak, and the distance and speed corresponding to this peak are 93m and 21.46m / s respectively, and the accurate target distance and speed estimations cannot be obtained; Figure 3 (b) is the result of estimating the target parameters by using the super-resolution method proposed in this paper. The estimated target distances are 93.1m and 95.2m respectively, and the estimated target speeds are 19.9m / s and 24.8m / s respectively. The errors do not exceed 1m and 1m / s, and compared with the traditional method, the perception accuracy is improved. Similar to the distance estimation error, the magnitude of the error is related to the values of R and v in Embodiment 3. In this simulation, R and v are uniformly valued at intervals of 0.9m and 0.6m / s respectively. If you want to further reduce the estimation error, you can reduce the value interval.

[0206] Figure 4 It is the result of resolving range ambiguity for the estimation result. When the target distance is 850 m, it exceeds the maximum unambiguous distance. Ambiguity as shown in Figure 4 (a) occurs, and two estimation results are obtained, namely 81.9 m and 849.9 m. After processing with the proposed method for resolving range ambiguity, the obtained result is 849.9 m.

[0207] Figure 5 It is the result of resolving velocity ambiguity for the estimation result. When the target distance and velocity are 80 m and 90 m / s respectively, it exceeds the maximum unambiguous velocity. Two groups of estimation results are obtained, namely 79.8 m, 17.1 m / s and 79.8 m, 89.9 m / s. After processing with the proposed method for resolving velocity ambiguity, the obtained result is 79.8 m, 17.1 m / s.

[0208] It can be seen from the simulation results that the proposed super-resolution method can accurately estimate the target parameters without generating ambiguity. Thus, it can be seen that the method proposed by the present invention effectively improves the perception accuracy of the system.

[0209] Figure 6 It is the relative error of parameter estimation.

[0210] Figure 6 It is the relative error of parameter estimation. Figure 6 (a) is the relative error of distance estimation. The target distance is set to be 30 - 90 m, and the parameter estimation methods based on DFT and subspace projection (denoted as sub in the legend) are respectively adopted. It can be seen that except for individual points, the error of the method based on subspace projection is significantly lower than that of the method based on DFT. After calculation, the average relative error of the method based on DFT in the figure is 0.025, and the average relative error of the method based on subspace projection is 0.0038. The fluctuation of the error with the change of the target distance is because the method based on DFT discretizes the distance and velocity into a grid. When the actual target distance is closer to the "grid line", the error is smaller, and vice versa. For the method based on subspace projection, as described in Implementation 3: Step 4, when different Rs are selected, the distance is also discretized, so the error fluctuates with the distance. Figure 6 (b) is the relative error of velocity estimation. The target velocity is set to be 5 - 25 m / s, and the parameter estimation methods based on DFT and subspace projection are respectively adopted. The results show that except for individual points, the error of the method based on subspace projection is significantly lower than that of the method based on DFT. After calculation, the average relative error of the method based on DFT in the figure is 0.12, and the average relative error of the method based on subspace projection is 0.017. The reason for the fluctuation of the error with the change of velocity is similar to the analysis of the relative error of distance estimation.

Claims

1. A target parameter super-resolution estimation method based on OCDM ISAC system, characterized in that: The steps include: Step 1: Use the 2D DFT method to process the sampling results of all symbols in the p-th pulse of the baseband and the sampling results of the received signal of all symbols in the p-th pulse to obtain a rough estimation result of the target distance and a rough estimation result of the target speed; Step 2: Use the target distance R i The rough estimation result of the target speed v i Compensate the sampling result of the received signal of the sth symbol of the pth pulse with the rough estimation result of and the known modulation data symbol to obtain a vector of the sth symbol; Step 3: For the vector of the sth symbol, the subspace projection method is used to perform super-resolution estimation of the target distance and speed to obtain super-resolution estimation of the target distance and super-resolution estimation of the speed; Step 4: Use the defuzzification method to process the super-resolution estimation of the target distance and the super-resolution estimation of the speed to obtain the final estimation result of the target distance and the final estimation result of the speed.

2. The target parameter super-resolution estimation method based on OCDM ISAC system according to claim 1 is characterized in that: Get a rough estimate of the target distance And the rough estimate of the target speed The specific process is as follows: Step 1-1: To highlight the effects of delay and Doppler frequency shift, ignore the delay caused by the pulse period and symbol period, and transform the sth transmitted symbol x in the pth pulse into p,s (t) is expressed as: Where t is the time, j is the imaginary unit; where N p In the OCDM pulse signal, each pulse contains N s OCDM symbols, signal parameters include the number of subcarriers N, carrier frequency f c , symbol period T s and pulse period T p , where p = 0, 1, ..., N p -1,s=0,1,…,N s -1, X(k) is the data symbol modulated on the kth subcarrier; For the sth transmitted symbol x in the pth pulse p,s (t) is sampled, then the sampling result x p,s (n) is expressed as: Where n = 0, 1, ..., N-1; Based on formula (1), the received signal y of the i-th target is p,s,i (t) is: Among them, A i is the complex amplitude factor of the attenuation and phase shift that occurs during the propagation and scattering of the i-th target; τ i is the delay corresponding to the i-th target, f d,i is the Doppler shift of the ith target, c is the speed of light, f c is the carrier frequency, using a single-base ISAC, assuming that there are N t targets, the distance between the i-th target and the transmitter is R i , the relative speed is v i ,i=1,2,…,N t ; To simplify the derivation, the noise effect is ignored in equation (3). p,s,i (t) Sampling is performed to obtain the sampling signal y of the i-th target p,s,i (n): In formula (4), and phase shift Compared to the phase shift within a symbol period is small, so ignore it and express equation (4) as: Perform Fourier transform on equation (2) to obtain the Fourier transform F of the sampling result of the sth symbol of the pth pulse of the baseband tx,s (k), Fourier transform is performed on equation (5) to obtain the Fourier transform F of the sampling result of the received signal of the sth symbol of the pth pulse rx,s (k): Where DFT(·) is the DFT operation performed on the content in brackets, k = 0, 1, ..., N-1; After performing DFT on the N sampling results of the sth symbol, the intermediate variable F is obtained. tx,s and the intermediate variable F rx,s : in is an N-dimensional complex space; According to formula (7), the transmission matrix F tx and the receiving matrix F rx It is expressed as: in is N×N s dimensional complex space; Step 1-2: Order That is, the conjugate matrix of the receiving matrix and the transmitting matrix is ​​multiplied to remove the influence of communication information on radar detection, where ⊙ is the Hadamard product, (·) H is to take the conjugate transpose of the matrix; then F is expressed as: Where F is the intermediate variable, is the delay vector of the i-th target, expressed as: Its elements are and is the Doppler vector of the ith target, It is expressed as: Its elements are and N s dimensional complex space; Step 1-3: According to formula (10), the distance information of the target R i is included in the linear phase shift of different subcarriers of the same OCDM symbol, by calculating The inverse discrete Fourier transform IDFT is used to obtain the distance estimation of the reflecting target; according to formula (11), the Doppler information f of the target d,i It is contained in the linear phase shift between the same subcarriers in the frequency domain of adjacent OCDM symbols, which is calculated by The DFT of the reflected target is used to obtain the velocity estimation of the reflected target. According to formula (9), the distance effect and Doppler effect of the target are completely orthogonal in the frequency domain. Therefore, by performing IDFT and DFT on the columns and rows of the matrix F respectively, and then performing spectrum peak search, the distance and velocity estimation of the target are obtained. Let the sth column of the intermediate variable F be f s , where s = 0, 1, ..., N s -1, perform IDFT on each column of F and get is the IDFT matrix of F, that is: in IDFT(·) is to perform IDFT operation on the contents in brackets, and then Perform DFT on each row of for The DFT matrix of The nth behavior Where n=0,1,…,N-1, then: in set up The element in the lth row and mth column of is f l,m ,make: but: in To round down; So we can get the distance R of the i-th target i The rough estimate result of and the velocity v of the i-th target i The rough estimate result of 3. The target parameter super-resolution estimation method based on the OCDMISAC system according to claim 1 is characterized in that: Using distance R i The rough estimate of the target distance is Speed i The rough estimate of the target speed is The sampling result y of the received signal of the sth symbol of the pth pulse and the known modulated data symbol p,s Compensate and get the vector r corresponding to the sth symbol s The specific process is as follows: Step 2-1: Based on the sampling results of the received signal of the sth symbol of the pth pulse and the received signal of the sth symbol of the pth pulse, obtain y p,s,i for y p,s The i-th component of Taking the i-th target as an example, y p,s,i It is expressed as: in, is an intermediate variable, and is an N×N dimensional complex space; is an intermediate variable, and D c =diag[X(0),X(1),…,X(N-1)], D c is an intermediate variable, and is the distance-oriented vector, and w i =[w i (0),w i (1),…,w i (N-1)] T , w i is the noise vector, and Φ is the discrete Fresnel inverse transform matrix, and its nth row and kth element is diag(·) is the operation of constructing a diagonal matrix; Step 2-2: Using distance R i The rough estimate result of and speed v i The rough estimate result of Calculate the rough delay estimate result And the Doppler rough estimation result The delay compensation matrix is ​​obtained based on the delay rough estimation results and the Doppler rough estimation results. and the Doppler compensation matrix in Since Φ is a unitary matrix and D c , using Φ,D c , the delay compensation matrix and the Doppler compensation matrix compensate for equation (18), then multiply equation (18) by in As intermediate variables, let: Where D p,s,i As the intermediate variable, define y p,s,i The compensation vector is r p,s,i , then r p,s,i It is expressed as: And order: where r p,s For r p,s.i The sum vector of; Step 2-3: In order to jointly estimate the distance and speed of the target at the same time, the sth symbol in all pulses is arranged in a row, and the vector r corresponding to the sth symbol is obtained. s : In the formula, is an intermediate variable, and in N p N×N p N-dimensional complex space; is the distance velocity guidance vector, is the Kronecker product of the matrix; is the velocity vector, and N p dimensional complex space; is the distance-oriented vector, and 4. The target parameter super-resolution estimation method based on the OCDMISAC system according to claim 1 is characterized in that: The vector r corresponding to the sth symbol s ,s=0,1,…,N s -1, using the subspace projection method to perform super-resolution estimation of target distance and speed, and obtain super-resolution estimation of target distance Super-resolution estimation of velocities The specific process is as follows: Step 3-1: From N s The data estimate covariance matrix of the symbols Step 3-2: Perform eigendecomposition, that is, Where Σ is The diagonal matrix of the eigenvalues ​​of The eigenvalues ​​are arranged from large to small, that is, Each column of U The eigenvalues ​​are The corresponding eigenvector; Step 3-3: Use the existing target source quantity estimation method in the field of array signal processing to obtain the target quantity N t , take the first N t The eigenvalues ​​corresponding to the eigenvectors Constructing the Matrix Step 3-4: Discretize R and v at fixed intervals and calculate the corresponding τ and f d , get the distance velocity guidance vector structure Search above threshold γ R,v The peak value, and then use the delay corresponding to the peak value and Doppler shift Calculate the super-resolution estimate of the target distance Super-resolution estimation of velocities in λ is the wavelength, λ=c / f c .

5. The target parameter super-resolution estimation method based on the OCDMISAC system according to claim 1 is characterized in that: Furthermore, the defuzzification method is used to estimate the super-resolution of the target distance. Super-resolution estimation of velocities Processing is performed to obtain the final estimated result of the target distance The final estimate of the speed The specific process is as follows: Step 4-1: Use the method described in step 3 to obtain the distance estimate of the target The maximum unambiguous distance of this method To resolve the distance ambiguity, the pulse compression method is used to estimate the target distance. And combine the two, in order to reduce the error caused by the combination, let: Where α is the distance ambiguity coefficient, is the fuzzy distance estimation, and let q = -1, 0, 1, for different values ​​of q, calculate and The absolute value of the difference, and take the one with the smallest absolute value of the difference As the final estimate of the target distance Step 4-2: Obtain the target speed estimate using the method described in step 3 After that, the maximum unambiguous speed is right To resolve velocity ambiguity, use the 2D DFT method to obtain the target velocity estimate And combine the two, let: Where β is the velocity ambiguity coefficient, To estimate the fuzzy velocity, for different values ​​of q, we calculate and The absolute value of the difference, and take the one with the smallest absolute value of the difference As the final estimate of the target distance 6. A terminal device comprising a processor, a memory and a computer program stored in the memory, characterized in that: When the processor executes the computer program, a target parameter super-resolution estimation method based on an OCDMISAC system according to any one of claims 1 to 5 is implemented.

7. A computer-readable storage medium, wherein a computer program is stored in the medium, characterized in that: When the computer program is executed by a processor, a target parameter super-resolution estimation method based on an OCDM ISAC system according to any one of claims 1 to 5 is implemented.

Images