Near-field channel rapid estimation method in super-large scale RIS assisted multi-user communication system

Abstract

The invention discloses a near-field channel rapid estimation method in a super-large scale intelligent reconfigurable super surface (RIS) assisted multi-user communication system, which comprises the following steps of: S1, establishing a near-field channel model of the super-large scale RIS assisted multi-user communication system under a millimeter wave frequency band; s2, designing a near-field polarization time delay domain sampling matrix for the system; s3, drawing a multi-user pilot frequency receiving signal into a polarization time delay domain pattern by using the near-field polarization time delay domain sampling matrix, and carrying out pilot frequency decontamination processing; s4, designing a sparse recovery algorithm and applying the sparse recovery algorithm to a near-field polarization time delay domain channel estimation algorithm; according to the method, the polarization delay domain characteristics of the near-field channel can be effectively utilized, and the channel state information can be quickly and accurately obtained in the uplink stage.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technologies, and in particular, to a method for rapidly estimating a near-field channel in a multi-user communication system assisted by a very large-scale reconfigurable intelligent surface (RIS). Background Art

[0002] The increasing number of mobile devices has led to a growing demand for large-scale and low-latency communication in the sixth-generation wireless communication network (6G). To improve spectral efficiency, the carrier frequency has been moved to the millimeter-wave (mmWave) and terahertz (THz) frequency bands. Meanwhile, to obtain higher directional beamforming gains, researchers have explored very large-scale array systems, among which very large-scale RIS is one of the potential technologies for future wireless communication. Due to the weak propagation ability of high-frequency signals, when the direct link between communication terminals is blocked, very large-scale RIS ensures smooth links and improves communication quality. However, due to the small wavelength of mmWave / THz and the large size of the very large-scale RIS array, the Rayleigh distance of wireless signals has become significantly larger, which means that each communication device in the cellular network is more likely to be in the near field.

[0003] To ensure the high beamforming gain of a very large-scale RIS-assisted near-field communication system, it is crucial to obtain accurate channel state information. However, due to the spherical wave nature of the near-field channel, the incident and outgoing angles of each antenna are not the same in the near-field region, and the plane wave-related characteristics of the far-field channel are no longer applicable to near-field channel estimation. Many estimation methods that utilize channel sparsity have poor performance or excessive iteration times, thereby introducing significant estimation errors and estimation delays in the channel estimation stage. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for estimating a near-field channel in a multi-user communication system assisted by a very large-scale RIS, which can rapidly and accurately obtain channel state information in the uplink stage and make up for the deficiencies of the existing technology.

[0005] To achieve the above purpose, the technical solution adopted by the present invention is as follows: A method for rapidly estimating a near-field channel in a multi-user communication system assisted by a very large-scale RIS, comprising the following steps:

[0006] S1: Establish a near-field channel model of a multi-user communication system assisted by a very large-scale RIS in the millimeter-wave frequency band;

[0007] S2: For the above system, design the near-field polarization time-delay domain sampling matrix, and plot the multi-user pilot received signals into a polarization time-delay domain pattern;

[0008] S3: Use the multi-user polarization time-delay domain pattern for pilot decontamination processing;

[0009] S4: Design a sparse recovery algorithm and apply it to the near-field polarization time-delay domain channel estimation algorithm.

[0010] Step S1 specifically includes:

[0011] S11: Establish a channel model: Consider a mmWave / THz band time division duplex (TDD) very large-scale RIS-assisted orthogonal frequency division multiplexing (OFDM) communication system. The base station (BS) is equipped with a uniform linear array (ULA) with N RF radio frequency (RF) chains and N BS antennas, where N RF << N BS . At the same time, a RIS of the uniform planar array (UPA) type composed of N = N1N2 phase shift elements is deployed in the cellular network. The array element spacing of both the BS and the RIS is d = λ c / 2 (λ c is the carrier wavelength). The number of single-antenna users is K, and M subcarrier resources are used simultaneously, while the line-of-sight channel between the BS and the users is blocked. The length of the guard interval (such as cyclic prefix (CP)) of the OFDM modulation is M g . The boundary between the near-field regions is generally distinguished by the Rayleigh distance Z = 2D 2 / λ c . Among them, assuming that the BS, RIS, and users are all in the near-field region of each other, the frequency-domain near-field channel between the RIS and the k-th user on the m-th subcarrier

[0012]

[0013] where L is the number of resolvable paths (usually very small), and α l is the complex path gain of the l-th path. In addition, is the frequency of the m-th subcarrier, B and f c correspond to the bandwidth and carrier frequency, is the wavelength at f m , c is the speed of light. The parameter θ l and φ l represent the elevation angle and azimuth angle of the l-th path respectively. r l represents the distance between the scatterer (or user) and the center of the RIS array, where τ l represents the path delay. is the array response vector of the spherical wave channel model in the near-field region, and its (n1,n2)-th element is defined as

[0014]

[0015] represents the distance between the scatterer (or user) and the (n1,n2)-th element of the RIS array, which can be approximated by a second-order Taylor series expansion as

[0016]

[0017] S12: Establish the uplink pilot signal transmission model: In the UL transmission phase, K different users simultaneously send predetermined pilots to the BS on all subcarriers. Assume that the channel coherence time is greater than the duration of the pilot symbol (with length Q) transmission, and multi-user interference is considered. Then, the N RF ×1 spatial frequency domain pilot signal r RF [q,m] received at the q-th pilot interval and the m-th subcarrier is

[0018]

[0019] where m = 1, 2,..., M and q = 1, 2,..., Q. and represent the phase shift of the RIS array and the near-field channel between the BS and the RIS respectively. is a diagonal matrix, where is the phase of the n-th RIS element at the q-th pilot interval, with a precision of 1 bit. is the antenna selection network at the q-th pilot interval, using low-cost 1-bit phase shifters, and its elements are selected from the set {...}. is the Gaussian noise vector at subcarrier m in pilot interval q. Assume that the pilot symbol x[q,m] = 1, then the QN RF ×1 received signal after pilot transmission is expressed as

[0020]

[0021] where

[0022]

[0023] Since RIS is generally deployed in a fixed position in advance to ensure the line-of-sight (LOS) connection between RIS and BS, W[q]W[q], H BR [m] and Φ RIS [q] can be assumed to be known. Finally, since the matrix is full rank, and usually Q≥(N / N RF ), the least square (LS) method can be used to obtain the final observed value of the UL channel after the antenna selection network

[0024]

[0025] Step S2 specifically includes:

[0026] S21: Beam domain sampling design: Similar to the far-field propagation model, the correlation function of the array response vectors corresponding to different angles (θ p , φ p ) and (θ q , φ q ) can be expressed as

[0027]

[0028] where

[0029]

[0030] Therefore, the beam domain sampling can be designed as

[0031]

[0032] to ensure the orthogonality of the array response vectors corresponding to different angles.

[0033] S22: Distance domain sampling design: Use the approximate array transfer response vector for calibration (the error can be ignored)

[0034]

[0035] where

[0036]

[0037] The correlation function of two array response vectors in the same quantization direction but different distances can be expressed as

[0038]

[0039] where

[0040]

[0041] Since the function fluctuates and decreases as β increases, it is easy to find the first pole β * to minimize the relevant function within a reasonable range. Therefore, the distance domain sampling with dictionary size S dis can be designed as

[0042]

[0043] Combining the above beam domain and distance domain sampling methods, a polarization domain transformation matrix W with size N×NS dis can be constructed

[0044]

[0045] S23: Time-delay domain sampling design: Using the common M-dimensional discrete Fourier transform (DFT) matrix F M is an effective method to achieve time-delay domain sampling. The (p,q)-th element of F M is defined as

[0046]

[0047] while denotes the first M M (M g (M g << M) columns of F

[0048] S24: Constructing the sparse pattern in the near-field channel polarization time-delay domain: Let It transfers the multi-user channel observation signal Y in the spatial frequency domain to the polarization time-delay domain. When N = N1N2 and M is large, the near-field polarization time-delay domain channel observation signal of the very large-scale RIS is a sparse matrix with most elements close to zero, and the non-zero elements with each significant modulus correspond to a resolvable user path with a specific elevation angle, azimuth angle, distance, and propagation delay. Therefore, the pilot signal of the target user can be distinguished through the sparse pattern in the near-field channel polarization time-delay domain, that is, pilot decontamination.

[0049] Step S3 specifically includes:

[0050] S31: Transforming the multi-user channel observation signal in the spatial frequency domain to the polarization time-delay domain: Transfer the multi-user channel observation signal Y in the spatial frequency domain to the polarization time-delay domain

[0051]

[0052] wherein is the equivalent noise, which is also independently and identically distributed and follows the CN(0, σ 2 ) distribution. The k-th user is the target user, and the pilot signals of other users are multi-user interference.

[0053] S32: Obtain the polarization delay domain characteristics of the user: In most cases, the positions of the BS and RIS are fixed, and the low mobility of the user ensures that the elevation angle, azimuth angle, distance, and propagation delay of a certain user only experience slight changes over several channel coherence intervals. The only randomly varying parameter is the path gain. Assume that the BS obtains the polarization delay domain characteristics P l and D l of different users during the preamble period, defined as

[0054]

[0055] subject to |P l | = τ P and |D l | = τ D ,

[0056] wherein

[0057]

[0058] and are the polarization delay domain channel and noise terms of the l-th user during the preamble period, which are the channel observation signals. τ P is the size of the polarization domain feature set, and τ D is the size of the delay domain feature set.

[0059] S33: Obtain the channel observation signal of the target user:

[0060]

[0061] Step S4 specifically includes:

[0062] Due to the sparse characteristics of the channel observation signal of the k-th user, channel estimation can be expressed as a sparse noise signal recovery problem. Use the Least Absolute Shrinkage and Selection Operator (LASSO) estimator for channel recovery.

[0063] The specific implementation steps are as follows:

[0064] S41: Input the channel observation signal of the target user Input the noise variance estimate σ2 ; Input polarization domain transformation matrix W; Input time delay domain transformation matrix Input polarization domain feature set P l ; Input time delay domain feature set D l 。

[0065] S42: Initialization p = 1;

[0066]

[0067] S43: Calculate parameters

[0068] S44: If p = 1, then let Otherwise

[0069]

[0070] S45: If Then let λ opt = λ p 。

[0071] S46: Update parameters

[0072] S47: p = p + 1, if p ≤ NS dis M g + 1, then return to S43 for calculation, otherwise go to S48.

[0073] S48: For each element of the matrix Let

[0074]

[0075] S49: Obtain the spatial frequency domain channel estimation result of the final target user

[0076]

[0077] Beneficial effects: The purpose of the present invention is to provide a near - field channel estimation method in a very large - scale RIS - assisted multi - user communication system. This method analyzes and studies the sparse characteristics of the near - field channel polarization time - delay domain pattern, and uses this characteristic for multi - user pilot decontamination and channel estimation, which can quickly and accurately obtain channel state information in the uplink stage, reducing channel estimation error and estimation delay.

[0078] The present invention significantly improves the near - field channel estimation performance in a very large - scale RIS - assisted multi - user communication system through the following innovative points, and the specific beneficial effects are as follows:

[0079] Innovation Point 1: Refined Modeling of Near-Field Spherical Wave Channel Model

[0080] Beneficial Effects:

[0081] Break through the limitations of the traditional far-field plane wave assumption. By introducing a spherical wave model, accurately characterize the near-field channel characteristics of ultra-large-scale RIS in the millimeter-wave / terahertz frequency band (including the coupling effects of distance, angle, and time delay).

[0082] Provide a theoretical basis for subsequent polarization time-delay domain sparsification processing, reducing the channel estimation error by more than 10 dB compared with the traditional LS algorithm (see Appendix Figure 3 ).

[0083] Innovation Point 2: Joint Sampling Matrix Design in Polarization Time-Delay Domain

[0084] Beneficial Effects:

[0085] Beam domain sampling: By orthogonalizing the angle response, solve the azimuth aliasing problem of near-field multi-users, and improve the azimuth resolution accuracy of users by 2 times.

[0086] Distance domain sampling: Utilize the spherical wave distance response characteristics to quantify the exact distance between the user and RIS (error < 0.1 wavelength), avoiding the estimation bias caused by the far-field equidistant assumption.

[0087] Time-delay domain sampling: Combine the DFT matrix to compress the multipath time-delay dimension, reducing the computational complexity from O(NS dis MQL) to O(NS dis M g log2(NS dis M g ).

[0088] Joint effect: The sparsity of the generated polarization time-delay domain pattern reaches more than 90% (see Appendix Figure 1 ), laying a foundation for multi-user interference suppression.

[0089] Innovation Point 3: Multi-User Pilot Decontamination Technology Based on Sparse Pattern

[0090] Beneficial Effects:

[0091] Utilize the prior knowledge of polarization time-delay domain characteristics (such as user historical azimuth / distance templates), and through threshold screening and interference projection, achieve accurate extraction of the pilot signals of target users.

[0092] In a cellular scenario with 4-user dense services, the multi-user interference suppression gain reaches 15 dB.

[0093] Innovation Point 4: Optimization of Low-Complexity LASSO Sparse Recovery Algorithm

[0094] Beneficial Effects:

[0095] Design an adaptive iterative step size and noise variance estimation mechanism for the block sparse characteristics of polarized time-delay domain signals to accelerate the convergence of the algorithm.

[0096] Compared with the traditional OMP algorithm, the calculation time delay is reduced by 60% (see Appendix Figure 4 ), and the estimation accuracy is close to the theoretical lower bound (Genie-aid LS). Brief Description of the Drawings

[0097] Figure 1 Schematic diagram of the sparse characteristics of the near-field channel polarization time-delay domain pattern involved in the present invention;

[0098] Figure 2 Flowchart of the sparse recovery algorithm applied to near-field polarization time-delay domain channel estimation adopted by the present invention;

[0099] Figure 3 Performance comparison of the normalized mean square error (NMSE) between the channel estimation method adopted by the present invention and other channel estimation schemes;

[0100] Figure 4 Comparison of the estimation time delay overhead between the channel estimation method adopted by the present invention and other channel estimation schemes;

[0101] Figure 5 Flowchart of the present invention. Detailed Embodiment

[0102] The present invention will be further described below with reference to the accompanying drawings.

[0103] S1: Establish a near-field channel model for a very large-scale RIS-assisted multi-user communication system in the millimeter wave band;

[0104] S11: Establish a channel model: Consider a mmWave / THz band TDD very large-scale RIS-assisted OFDM communication system. The BS is equipped with a ULA with N RF radio frequency chains and N BS antennas, where N RF << N BS . At the same time, a UPA-type RIS composed of N = N1N2 phase shift elements is deployed in the cellular network. The array element spacing of the BS and the RIS is both d = λ c / 2. The number of single-antenna users is K, and M subcarrier resources are used simultaneously, while the line-of-sight channel between the BS and the users is blocked. The length of the guard interval of OFDM modulation is M gIf the BS, RIS, and users are all in the near-field region of each other, the frequency-domain near-field channel between the RIS and the k-th user on the m-th subcarrier can be expressed as

[0105]

[0106] where L is the number of resolvable paths, and α l is the complex path gain of the l-th path. In addition, is the frequency of the m-th subcarrier, B and f c correspond to the bandwidth and carrier frequency, is the wavelength at f m , c is the speed of light. The parameters θ l and φ l represent the elevation angle and azimuth angle of the l-th path, respectively. r l represents the distance between the scatterer (or user) and the center of the RIS array, where τ l represents the path delay. is the array response vector of the spherical wave channel model in the near-field region, and its (n1n2)-th element is defined as

[0107]

[0108] represents the distance between the scatterer (or user) and the (n1n2)-th element of the RIS array, which can be approximated as

[0109]

[0110] S12: Establish the uplink pilot signal transmission model: In the UL transmission phase, K different users simultaneously send predetermined pilots to the BS on all subcarriers. Assume that the channel coherence time is greater than the duration of the pilot symbol (with length Q) transmission, and multi-user interference is considered. Then, the received N RF ×1 spatial frequency domain pilot signal r RF [q, m] at the q-th pilot interval and the m-th subcarrier is

[0111]

[0112] where m = 1, 2,..., M and q = 1, 2,..., Q. and represent the phase shift of the RIS array and the near-field channel between the BS and the RIS, respectively. is a diagonal matrix, where is the phase of the n-th RIS element at the q-th pilot interval, with a precision of 1 bit. is the antenna selection network at the q-th pilot interval, using a low-cost 1-bit phase shifter, the elements of which are selected from the set . is the Gaussian noise vector at subcarrier m in pilot interval q. Assuming the pilot symbol x[q,m] = 1, the combined QN RF ×1 received signal after pilot transmission is expressed as

[0113]

[0114] where

[0115]

[0116] Since RIS is generally pre-deployed at a fixed position to ensure the LOS connection between RIS and BS, W[q]W[q], H BR [m] and Φ RIS [q] can be assumed to be known. Finally, since the matrix is full-rank and usually Q ≥ (N / N RF ), the LS method can be used to obtain the final observed value of the UL channel after the antenna selection network

[0117]

[0118] S2: Design the near-field polarization time-delay domain sampling matrix, and plot the multi-user pilot received signal into a polarization time-delay domain pattern, as Figure 1 shown;

[0119] S21: Beam domain sampling design: Similar to the far-field propagation model, the correlation function of the array response vectors corresponding to different angles (θ p , φ p ) and (θ q , φ q ) can be expressed as

[0120]

[0121] where

[0122]

[0123] Therefore, the beam domain sampling can be designed as

[0124]

[0125] to ensure the orthogonality of the array response vectors corresponding to different angles.

[0126] S22: Range domain sampling design: Use the approximate array transfer response vector for calibration (the error can be ignored)

[0127]

[0128] Among them

[0129]

[0130] The correlation function of two array response vectors in the same quantization direction but at different distances can be expressed as

[0131]

[0132] Among them

[0133]

[0134] Since the function fluctuates and decreases with the increase of β, it can be easily found the first pole β * , which minimizes the correlation function within a reasonable range. Therefore, the distance domain sampling with a dictionary size of S dis can be designed as

[0135]

[0136] Combining the above beam domain and distance domain sampling methods, a polarization domain transformation matrix W with a size of N×NS dis can be constructed

[0137]

[0138] S23: Time-delay domain sampling design: The commonly used M-dimensional DFT matrix F M is used to implement time-delay domain sampling. The (p,q)-th element of F M is defined as

[0139]

[0140] while represents the first M M (M g (M g << M) columns of F

[0141] S24: Construct the sparse pattern in the near-field channel polarization time-delay domain, as shown in Figure 1 : Let It transfers the multi-user channel observation signal Y in the spatial frequency domain to the polarization time-delay domain. When N = N1N2 and M is large, it can be seen from Figure 1 that the near-field polarization time-delay domain channel observation signal is a sparse matrix with most elements close to zero. Each non-zero element with a significant modulus corresponds to a resolvable user path with a specific elevation angle, azimuth angle, distance, and propagation delay. Therefore, the pilot signals of target users can be distinguished through the sparse pattern in the near-field channel polarization time-delay domain, i.e., pilot decontamination.

[0142] S3: Use the multi-user polarization time-delay domain pattern for pilot decontamination, as Figure 1 shown;

[0143] S31: Transform the multi-user channel observation signal in the spatial frequency domain to the polarization time-delay domain: Transform the multi-user channel observation signal Y in the spatial frequency domain to the polarization time-delay domain

[0144]

[0145] where is the equivalent noise, which is also independently and identically distributed, following the CN(0,σ 2 ) distribution.

[0146] S32: Obtain the polarization time-delay domain characteristics of users: Assume that the BS has obtained the polarization time-delay domain characteristics P l and D l of different users in the preamble period, defined as

[0147]

[0148] subject to |P l | = τ P and |D l | = τ D ,

[0149] where

[0150]

[0151] and are the polarization time-delay domain channel and noise terms of the l-th user in the preamble period, and are the channel observation signals. τ P is the size of the polarization domain feature set, while τ D is the size of the time-delay domain feature set.

[0152] S33: Obtain the channel observation signal of the target user, as Figure 1 shown:

[0153]

[0154] Since the channel observation signal of the k-th user Due to the sparse characteristics, channel estimation can be expressed as a sparse noise signal recovery problem. The LASSO estimator is used for channel recovery.

[0155] S4: Design a sparse recovery algorithm for application to near-field polarization delay-domain channel estimation. The process is as Figure 2 shown;

[0156] S41: Input the channel observation signal of the target user Input the estimated noise variance σ 2 ; Input the polarization-domain transformation matrix W; Input the delay-domain transformation matrix Input the polarization-domain feature set P l ; Input the delay-domain feature set D l .

[0157] S42: Initialize p = 1;

[0158]

[0159] S43: Calculate the parameters

[0160] S44: If p = 1, then let Otherwise

[0161]

[0162] S45: If then let λ opt = λ p .

[0163] S46: Update the parameters

[0164] S47: p = p + 1. If p ≤ NS dis M g + 1, then return to S43 for calculation; otherwise, proceed to S48.

[0165] S48: For each element of the matrix let

[0166]

[0167] S49: Obtain the final spatial frequency-domain channel estimation result of the target user

[0168]

[0169] Figure 3NMSE performance comparison of the channel estimation method (Proposed) adopted in the present invention with other channel estimation schemes (Classical LS, OMP (orthogonal matching pursuit), Simultaneous-OMP, Genie-aid LS). Among them, it is set that the BS is equipped with an N BS = 512-element antenna array, containing N RF = 4 radio frequency chains, serving K = 4 single-antenna users; the RIS is equipped with an array of N = N1N2 = 128×4; the number of system subcarriers is M = 2048; the estimated delay overhead Q = 150. As Figure 3 shown, the estimation accuracy of the adopted channel estimation method is always better than other algorithms at all signal-to-noise ratios and approaches the theoretical benchmark curve at low signal-to-noise ratios.

[0170] Figure 4 Comparison of the estimated delay overhead of the channel estimation method (Proposed) adopted in the present invention with other channel estimation schemes (Classical LS, OMP, Simultaneous-OMP, Genie-aid LS). Among them, it is set that the BS is equipped with an N BS = 512-element antenna array, containing N RF = 4 radio frequency chains, serving K = 4 single-antenna users; the RIS is equipped with an array of N = N1N2 = 128×4; the number of system subcarriers is M = 2048; the signal-to-noise ratio SNR = 0 dB. As Figure 4 shown, when ensuring reliable NMSE performance, the adopted channel estimation algorithm has lower channel estimation delay, which helps to achieve a faster estimation process.

[0171] The present invention provides a near-field channel estimation method in a very large-scale RIS-assisted multi-user communication system. This method analyzes and studies the sparse characteristics of the near-field channel polarization delay domain pattern, and uses this characteristic for multi-user pilot decontamination and channel estimation, and can quickly and accurately obtain channel state information in the uplink stage, reduce the estimation error and estimation delay, and take into account the various requirements of the 6G wireless communication network, and has broad application prospects in the era of 6G wireless large-scale communication.

[0172] The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A fast near-field channel estimation method in a very large-scale RIS-assisted multi-user communication system, characterized in that, It includes the following steps: Step S1: Establish a near-field channel model for a very large-scale RIS-assisted multi-user communication system in the millimeter-wave / terahertz frequency band, where the base station, RIS, and users are all in the near-field region, and the spherical wave model is used to characterize the channel response; Step S2: Based on the near-field channel model established in Step S1, design a near-field polarization delay domain sampling matrix, convert the multi-user pilot received signal from the spatial frequency domain to the polarization delay domain, and generate a polarization delay domain pattern with sparse characteristics; Step S3: Utilize the polarization delay domain pattern generated in Step S2, and through distinguishing the pilot signal distribution characteristics of the target user and interfering users, perform pilot decontamination to eliminate multi-user interference; Step S4: For the polarization delay domain channel observation signal decontaminated in Step S3, design a channel estimation method based on the sparse recovery algorithm to quickly recover the channel state information of the target user.

2. The method according to claim 1, wherein The specific content of Step S1 includes: Construct a near-field channel model for a millimeter-wave / terahertz frequency band time-division multiplexing very large-scale RIS-assisted orthogonal frequency division multiplexing system, where: The BS is equipped with a uniform linear array, the RIS is a uniform planar array, and the array spacing is half a wavelength; The near-field channel model is based on the spherical wave propagation characteristics, and jointly characterizes the multi-user channel response through path gain, elevation angle, azimuth angle, and distance parameters.

3. The method according to claim 1, wherein The specific content of Step S2 includes: Step S21: Design a beam domain sampling matrix, and separate the spatial azimuth characteristics of users by orthogonalizing the array response vectors at different angles; Step S22: Design a distance domain sampling matrix, and utilize the distance-related spherical wave characteristics in the near-field channel to quantify the relative distance information between the users and the RIS; Step S23: Design a delay domain sampling matrix, and extract the multipath delay characteristics by using the discrete Fourier transform; Step S24: Combine the beam domain, distance domain, and delay domain sampling matrices to construct a polarization delay domain transformation matrix, and map the spatial frequency domain pilot signal to a sparse polarization delay domain pattern.

4. The method according to claim 1, wherein The specific content of Step S3 includes: Step S31: Convert the spatial frequency domain multi-user channel observation signal to the polarization delay domain through the polarization delay domain transformation matrix in Step S2; Step S32: Utilize the preset user polarization delay domain feature library to match the signal distribution of the target user; Step S33: Extract the pure channel observation signal of the target user through threshold screening and interference suppression.

5. The method according to claim 1, wherein The specific content of Step S4 includes: Step S41: Use the polarization delay domain signal of the target user output in Step S3 as the input to initialize the parameters of the sparse recovery algorithm; Step S42: Adopt the least absolute shrinkage and selection operator algorithm to iteratively optimize the channel estimation result; Step S43: Through convergence judgment and noise suppression, output the final spatial frequency domain channel estimation value.

6. The method according to claim 3, wherein The construction of the polarization delay domain transformation matrix in Step S24 needs to satisfy: The joint design of the beam domain and distance domain sampling matrices needs to ensure the sparse separability of different user signals in the polarization delay domain; The dimension of the delay domain sampling matrix matches the number of subcarriers of the system to cover the multipath delay range.

7. The method according to claim 4, wherein The establishment of the user polarization delay domain feature library in Step S32 needs to be through: Collect the historical channel data of the user during the leading period; Extract the elevation angle, azimuth angle, distance, and delay features of the user to form a feature template.

8. The method according to any one of claims 1-7, characterized in that The method is applicable to the 6G ultra-large-scale RIS-assisted multi-user communication system.

Images