Channel estimation method of multi-RIS auxiliary OFDM system facing shielding scene
By introducing occlusion environment modeling and group training protocols into a multi-RIS assisted OFDM system, and using tensor decomposition to decouple channel parameters, the complexity and accuracy issues of channel estimation under occlusion conditions are solved, achieving efficient and robust channel estimation.
Patent Information
- Application Number
- CN202511860871.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-11
- Publication Date
- 2026-02-13
AI Technical Summary
In obstructed environments, channel estimation in multi-RIS assisted communication systems faces challenges such as an increase in the number of channel propagation paths, a significant increase in resource consumption, and a decrease in channel estimation performance, making it difficult to achieve high-precision and low-overhead channel estimation.
By introducing occlusion environment modeling, a multi-RIS group training protocol is designed. Effective signal components are extracted by subtracting signals from adjacent time frames, a three-dimensional tensor signal model is constructed, and K-means clustering algorithm and tensor decomposition method are used to decouple channel parameters, thereby achieving efficient occlusion processing and channel estimation.
It significantly reduces signal processing complexity, improves the accuracy and stability of channel estimation, enhances robustness in obstructed environments, and expands the application scope of RIS technology.
Smart Images
Figure CN121530795A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the technical field of wireless communication, in particular to a channel estimation method for a multi-RIS assisted OFDM system in a shielding scenario. BACKGROUND
[0002] With the continuous development of the sixth generation mobile communication (6G) technology, new services, new scenarios and new demands for future society are emerging, and the requirements for data transmission rate, network capacity and coverage of mobile communication systems are significantly improved. The emergence of intelligent metasurface (RIS) technology provides a new solution for building low-power, high-efficiency and controllable wireless propagation environment.
[0003] RIS is a reconfigurable electromagnetic super surface composed of a large number of passive reflecting units. By adjusting the reflection phase shift of each unit, the propagation direction and energy distribution of electromagnetic waves can be reconfigured, and then the signal propagation environment can be adjusted to assist communication. With its advantages of low cost, low energy consumption and flexible deployment, RIS has broad application prospects in future wireless communication systems. For example, it can be deployed in multi-cell wireless networks to reduce inter-cell interference and improve system spectral efficiency and transmission reliability. At the same time, in shielding dense environment and mobile edge computing (MEC) network, RIS can provide additional propagation path and gain to alleviate the impact of shielding and enhance edge coverage capability. In addition, in emerging scenarios such as integrated sensing and communication (ISAC), RIS is also considered as one of the key enabling technologies to improve system performance.
[0004] To fully exploit the potential of RIS assisted communication, accurate channel state information (CSI) is essential. However, the channel estimation of RIS assisted wireless communication system still faces many challenges. On the one hand, the introduction of RIS significantly increases the number of channel propagation paths, expands the scale of the estimated channel, and RIS itself does not have active signal processing capability, which leads to a significant increase in resource consumption in terms of pilot overhead, training duration and feedback information volume. On the other hand, the large-scale RIS array deployed in complex environments is easily affected by shielding and scattering, causing the signal of part of the reflecting units to be severely attenuated or even disabled, resulting in a decline in channel estimation performance. Therefore, how to accurately model the channel characteristics of RIS in a shielding environment and achieve high-precision and low-overhead channel estimation has become a key problem to be solved in current multi-RIS assisted communication systems. SUMMARY
[0005] The purpose of the present application is to provide a channel estimation method for a multi-RIS assisted OFDM system in a shielding scenario. On the basis of traditional channel estimation, the modeling of shielding environment is introduced, and efficient shielding processing and channel estimation are realized with low complexity. The method improves the accuracy of channel estimation, enhances the robustness of shielding environment, and improves the transmission reliability.
[0006] The channel estimation method of the multi-RIS assisted OFDM system for the occlusion scene of the application comprises the following steps: Step S1, system and channel modeling: a multi-RIS assisted OFDM system model for the occlusion scene is established, a RIS occlusion factor is introduced, and a cascaded channel model representing the multi-RIS occlusion state is constructed; Step S2, design of grouping training protocol: a plurality of RISs are divided into several groups, a grouping training protocol is designed according to the OFDM system frame structure, and effective signal components corresponding to each group of RISs are extracted by subtracting the received signals of adjacent time frames; Step S3, tensor construction and occlusion estimation: a three-dimensional tensor signal model is constructed using the extracted grouped signals, and the occlusion state corresponding to each RIS is estimated based on the K-means clustering algorithm; Step S4, channel parameter decoupling and estimation: a occlusion processing scheme is designed, and on this basis, an angle, time delay and fading parameter decoupling and estimation method based on tensor decomposition is constructed to realize complete channel estimation.
[0007] Further, step S1, a multi-RIS assisted OFDM system model for the occlusion scene is established, a RIS occlusion factor is introduced, and a cascaded channel model representing the multi-RIS occlusion state is constructed, which specifically comprises: Consider a multi-RIS assisted downlink OFDM system, in which the base station and the user terminal are respectively equipped with , root antennas, Each block RIS is uniformly deployed between the base station and the user, and each block RIS is equipped with reflective units, , The number of reflective units of the RIS in the vertical and horizontal directions, respectively; the base frequency of the wideband system is , the transmission bandwidth is , which is divided into subcarriers, the beam training protocol uses time frames, and each frame contains slots; the base station-RIS-user cascaded channel is mainly LoS path, and the direct link between the base station and the user is blocked by obstacles; in the time frame, in the slot, the received signal on the subcarrier is: (1); Wherein, , The path loss and time delay parameters of the cascaded channel corresponding to the pth block RIS, is the number of subcarriers; , These are the array response vectors of the base station and the user at the p-th RIS, respectively; , Representing the first Block RIS to the array response vector of the user and the base station; Indicates transpose. The imaginary unit; For transmission pilot, The channel is Gaussian white noise; For the first The first time frame The phase shift matrix of the block RIS. This represents vector matrix operations. For RIS Phase shift coefficient of each unit; Let be the occlusion factor vector of the p-th RIS block, where the corresponding element of the occluded unit is 0 and the element of the unoccluded unit is 1.
[0008] Further, in step S2, the multiple RIS are divided into several groups, and a group training protocol is designed according to the OFDM system frame structure. The effective signal components corresponding to each group of RIS are extracted by subtracting the received signals from adjacent time frames. Specifically, this includes: In adjacent time frames, the same transmission pilot signal is shared. Utilizing vectorization properties, the signal can be... The total received signals from each time slot are combined to form a data vector. : (2); in, For the first Pilot merging matrix for each time frame; For the reconstruction of the first The first time frame; for the reconstruction of the first time frame The time frame, the Pilot matrix for each time slot, As a unit array, Represents the Khatri-Rao product. Represents the Kronecker product. For the first Noise merge vector for each time frame; Will Block RIS is divided into Groups, each group contains The block RIS estimates only one set of channels corresponding to each RIS at a time, and the channels to be estimated are... Phase shift matrix of block RIS As the time frame changes, the phase shifts of other RIS remain constant. The difference between each adjacent time frame signal is calculated, and a difference signal containing only the estimated RIS component is screened out : (3); wherein, is a difference phase shift vector, is the time frame, and the phase shift of the block RIS, is a difference noise vector, 1 , denotes a Hadamard product, , is the time frame, and the received signal and noise on the subcarrier.
[0009] Further, in step S3, a three-dimensional tensor signal model is constructed using the extracted grouped signals, and the occlusion state corresponding to each RIS is estimated based on a K-means clustering algorithm, specifically including: First, since the pilot matrix is known and column full rank, wherein , a least squares estimator is used to eliminate it in the signal to obtain the processed received signal : (4); wherein, is the processed noise, denotes a generalized inverse; The signals of the difference time frames are processed and stacked into a data matrix : (5); wherein, , is a phase shift and noise matrix stacked with the difference time frames; Finally, the signals on the subcarriers are combined to construct a three-dimensional signal tensor , and the original channel estimation problem is converted into a sparse tensor signal recovery problem: (6); wherein, is a combined delay vector of the subcarriers, is a tensor form of noise, Represents the tensor outer product; the sparsity of the tensor is the number of RIS blocks in the grouping. The factor matrix of a tensor is written as: (7); (8); (9); The tensor CP decomposition algorithm is first used to solve for the factor matrix, and then for the tensor. Model 1 unfold Perform SVD decomposition using the factor matrix The van der Mond property, using rotation-invariant techniques to estimate signal parameters, and the ESPRIT algorithm to estimate the factor matrix. and And then according to Reconstructing the factor matrix ; For the reconstructed factor matrix The occlusion state is estimated column-wise; firstly, the known RIS phase shift matrix is eliminated using least squares. Obtain vector : (10); in, Factor matrix The Column; the resulting vector Reconstructed into a two-dimensional matrix: (11); in, For matrix dimension reconstruction function, The elements correspond one-to-one with the UPA array reflection units of the RIS, and will All elements are clustered using K-means based on their magnitude, resulting in two clusters with element coordinate indices set as follows: Based on this, a RIS array occlusion matrix can be constructed. : (12); Ultimately, by The deformation yields the RIS occlusion factor vector. Estimate: (13).
[0010] Further, in step S4, an occlusion handling scheme is designed. Based on this, a method for decoupling and estimating angle, delay, and fading parameters based on tensor decomposition is constructed to achieve complete channel estimation. This specifically includes the following steps: Step 4.1 Occlusion rate statistics and sorting: calculate the occlusion rate of each RIS surface : (14); wherein, denotes the 0 norm; sorting from small to large according to the occlusion rate provides the basis for subsequent screening and grouping; Step 4.2 Severe occlusion RIS screening: set an occlusion threshold When , it is determined that the RIS is a large-area occlusion and does not participate in subsequent estimation; Step 4.3 RIS surface occlusion correction: process each RIS surface one by one, and count the RIS array occlusion matrix The number of non-zero elements in each row and column; in subsequent angle parameter estimation, delete the rows and columns with severe occlusion, i.e. the rows and columns with less than 3 non-zero elements, and only use the signals of the remaining available reflecting elements for parameter estimation, and take the average of the estimation results to improve the estimation stability; Step 4.4 Multi-RIS scheduling strategy: in the positioning and other scenarios that only require part of the RIS channel information, select the optimal RIS group according to the occlusion rate for channel estimation and feedback; in the RIS phase shift optimization and other system design scenarios, in order to ensure the overall performance of the system, on the basis of removing the large-area occlusion RIS, the remaining RIS is grouped and estimated; Subsequently, based on the occlusion processing scheme and the factor matrix obtained by tensor decomposition, the decoupling and estimation of the channel parameters are carried out in turn: according to the estimated value of the factor matrix , the angle parameters at the RIS are searched using the maximum correlation criterion combined with the occlusion processing scheme; according to the estimated value of the factor matrix , the angle parameters at the base station and the user are estimated using the maximum correlation criterion; the delay parameter is solved according to the estimated value of the factor matrix ; finally, the fading parameter is calculated by scaling between the factor matrices, and the complete multi-RIS channel is restored using the estimated channel parameters, achieving robust channel estimation in the occlusion environment.
[0011] Compared with the prior art, the advantages of the present application are: the present application designs a multi-RIS grouping training scheme, uses the difference between adjacent time frames to extract the effective signal components of each group of RIS, significantly reduces the complexity of signal processing and matrix decomposition of the multi-RIS system, and improves the accuracy and stability of channel estimation; the present application designs a multi-RIS occlusion modeling and processing method, which can flexibly adjust the estimation strategy in different occlusion environments, thereby effectively improving the robustness and adaptability of channel estimation in complex propagation environments, and expanding the application range of RIS technology in occlusion environments. BRIEF DESCRIPTION OF DRAWINGS
[0012] Figure 1 Flow chart for channel estimation of the multi-RIS assisted OFDM system for the occlusion scenario of the embodiment of the present application.
[0013] Figure 2 Schematic diagram for multi-RIS occlusion processing of the embodiment of the present application for the occlusion scenario.
[0014] Figure 3 (a) is a diagram of the relationship between the angle parameter at the base station / user and the signal-to-noise ratio.
[0015] Figure 3 (b) is a diagram of the relationship between the angle parameter at the RIS and the signal-to-noise ratio.
[0016] Figure 3 (c) is a diagram of the relationship between the time delay parameter and the signal-to-noise ratio.
[0017] Figure 3 (d) is a diagram of the relationship between the path attenuation parameter and the signal-to-noise ratio.
[0018] Figure 4 Figure for the relationship between the angle parameter estimation accuracy of the multi-RIS and the occlusion rate of the embodiment of the present application for the occlusion scenario. DETAILED DESCRIPTION
[0019] The present application will be described in detail below with reference to the accompanying drawings and further in combination with embodiments, so as to make the purpose, technical scheme and outstanding advantages of the present application clearer, more detailed and more prominent; As Figure 1 shown, the channel estimation method of the multi-RIS assisted OFDM system for the occlusion scenario provided by the embodiment of the present application comprises the following design steps; it should be emphasized that the embodiment specific values do not limit the application scope of the present application on the premise of meeting the operation requirements of the method of the present application.
[0020] S1, system and channel modeling: The embodiment of the present application considers a multi-RIS assisted downlink OFDM system, wherein the base station and the user are both equipped with root antennas, Each block RIS is uniformly deployed between the base station and the user, and each block RIS is equipped with reflector units, respectively, the number of reflector units of the RIS in the vertical and horizontal directions; the base frequency of the wideband system is , the transmission bandwidth is , which is divided into subcarriers, the beam training protocol uses time frames, and each frame contains slots; the base station-RIS-user cascaded channel is mainly in the line-of-sight (LoS) path, and the direct link between the base station and the user is blocked by an obstacle; in the time frame, the slot, and the subcarrier The received signal is: (1); in, , These are the path fading and delay parameters of the cascaded channel corresponding to the p-th RIS block, respectively. The number of subcarriers; , These are the array response vectors of the base station and the user at the p-th RIS, respectively; , Representing the first Block RIS to the array response vector of the user and the base station; Indicates transpose. The imaginary unit; For transmission pilot, The channel is Gaussian white noise; For the first The first time frame The phase shift matrix of the block RIS. This represents vector matrix operations. For RIS Phase shift coefficient of each unit; Let be the occlusion factor vector of the p-th RIS block, where the corresponding element of the occluded unit is 0 and the element of the unoccluded unit is 1.
[0021] S2, Group Training Protocol Design: In adjacent time frames, the same transmission pilot signal is shared. Utilizing vectorization properties, the signal can be... The total received signals from each time slot are combined to form a data vector. : (2); in, For the first Pilot merging matrix for each time frame; For the reconstruction of the first The first time frame; for the reconstruction of the first time frame The time frame, the Pilot matrix for each time slot, As a unit array, Represents the Khatri-Rao product. Represents the Kronecker product. For the first Noise merge vector for each time frame; Will Block RIS is divided into Groups, each group contains The block RIS estimates only one set of channels corresponding to each RIS at a time, and the channels to be estimated are... Phase shift matrix of block RIS Other RIS phase shifts remain unchanged over time frames, and the phase shift of the block RIS is changed as The difference between two adjacent time frame signals is calculated, and the difference signal containing only the estimated RIS component is screened out The difference signal containing only the estimated RIS component is screened out : (3); Wherein, is the difference phase shift vector, is the first time frame, and the phase shift of the block RIS is is the difference noise vector, 1 , represents Hadamard product, , is the received signal and noise on the first time frame, and the first subcarrier.
[0022] S3, tensor construction and occlusion estimation: First, since the pilot matrix is known and the column is full rank, wherein , the least square estimator is used to eliminate it in the signal to obtain the processed received signal : (4); Wherein, is the processed noise, represents the generalized inverse; The signals of the difference time frames are processed and stacked into a data matrix : (5); Wherein, , is the phase shift and noise matrix of stacking difference time frames; Finally, the signals on the subcarriers are combined to construct a three-dimensional signal tensor , and the original channel estimation problem is converted into a sparse tensor signal recovery problem: (6); Wherein, is the combined delay vector of the subcarriers, is the tensor form of noise, denotes tensor outer product; the sparsity of tensor is the number of grouped RIS blocks , the factor matrix of tensor is written as: (7); (8); (9); First, the factor matrix is solved using the tensor CP decomposition algorithm, and the SVD decomposition is performed on the mod 1 expansion of the tensor The Vandermonde property of the factor matrix is used to estimate the signal parameter by the rotation invariant technique, and the ESPRIT algorithm is used to estimate the factor matrix and , and then the factor matrix is reconstructed according to ; The reconstructed factor matrix is estimated according to the column; first, the known RIS phase shift matrix is eliminated by using the least square to obtain the vector : (10); wherein, is the i-th column of the factor matrix ; the obtained vector is reconstructed into a two-dimensional matrix: (11); wherein, is the matrix dimension reconstruction function, the elements of correspond one by one to the UPA array reflecting units of the RIS, and all elements of are K-means clustered according to the amplitude size, divided into two clusters, and the element coordinate index sets are , and accordingly, the RIS array blocking matrix is constructed: (12); Finally, the RIS blocking factor vector is obtained by transforming : (13).
[0023] S4, channel parameter decoupling and estimation: First, as shown in Figure 2 , the multi-RIS is blocked according to the estimated blocking condition, which specifically includes the following steps: Step 4.1 Occlusion Rate Statistics and Sorting: Calculate the occlusion rate for each RIS array. : (14); in, Represents the 0 norm; sorted by occlusion rate from smallest to largest. Figure 2 The RIS3>RIS1>RIS2>RIS4 sequence provides a basis for subsequent screening and grouping. Step 4.2 Severe Occlusion RIS Screening: Set Occlusion Threshold ,when At that time, the RIS was determined to be a large-area occlusion and was not included in subsequent estimations. Figure 2 RIS4); Step 4.3 RIS array occlusion correction: Process the remaining RIS arrays one by one and calculate the RIS array occlusion matrix. The number of non-zero elements in each row and column; in the subsequent angle parameter estimation, rows and columns with severe occlusion, i.e., rows and columns with less than 3 non-zero elements, are deleted, and only the signals of the remaining available reflective units are used for parameter estimation. The estimation results are then averaged to improve the estimation stability. Step 4.4 Multi-RIS Scheduling Strategy: Flexibly select RIS scheduling schemes based on specific application scenarios: In scenarios requiring only partial RIS channel information, the optimal RIS group can be selected for channel estimation and feedback based on the occlusion rate (e.g., in positioning applications requiring at least two RISs, RIS3 and RIS1 are ultimately selected based on the degree of occlusion); in system design scenarios such as RIS phase shift optimization, to ensure overall system performance, the remaining RISs can be estimated in groups after eliminating RISs with large-area occlusion, balancing accuracy and computational overhead. Subsequently, based on the occlusion processing scheme and the factor matrix obtained from tensor decomposition, the channel parameters are decoupled and estimated sequentially: according to the factor matrix... The estimated value is used to search for the angle parameter at RIS using the maximum correlation criterion and combined with the occlusion treatment scheme; based on the factor matrix The estimated values are used to estimate the angle parameters at the base station and the user using the maximum correlation criterion; based on the factor matrix... The estimated values are used to solve for the delay parameters; finally, the fading parameters are calculated by scaling between factor matrices, and the complete multi-RIS channel is restored using all the estimated channel parameters, thus achieving robust channel estimation under obstructed environments.
[0024] Based on the above parameter configuration, MATLAB simulation is run to verify the performance of the proposed channel estimation method of the multi-RIS assisted OFDM system. The relationship between the decoupling accuracy of each channel parameter obtained by channel estimation and the signal-to-noise ratio (SNR) is shown in FIGS. 3(a), 3(b), 3(c) and 3(d), and the relationship between the estimation accuracy of the angle parameter at the RIS and the occlusion rate is shown in FIGS. 4(a), 4(b), 4(c) and 4(d), where the accuracy index is measured by the root mean square error (RMSE); it can be known from the simulation results that the channel estimation method provided in the embodiment of the present application can obtain unique decoupling and accurate recovery of the channel parameters, and greatly improves the robustness of channel estimation in the occlusion environment. Figure 4
[0025] In summary, the channel estimation method of the multi-RIS assisted OFDM system for the occlusion scenario provided in the embodiment of the present application effectively solves the interference problem caused by the occlusion environment to the multi-RIS cascaded channel path parameter estimation by modeling the occlusion state at the RIS and further performing occlusion estimation and processing, and provides strong technical support for multi-RIS scheduling and subsequent system application.
Claims
1. A method for channel estimation of a multi-RIS-assisted OFDM system in an occluded scenario, the method comprising: The method comprises the following steps: Step S1, system and channel modeling: a multi-RIS assisted OFDM system model facing the shielding scene is established, a RIS shielding factor is introduced, and a cascaded channel model representing the shielding state of the multi-RIS is constructed; Step S2, design of grouping training protocol: a plurality of RISs are divided into a plurality of groups, a grouping training protocol is designed according to the OFDM system frame structure, and effective signal components corresponding to each group of RISs are extracted by subtracting adjacent time frame received signals; Step S3, tensor construction and shielding estimation: a three-dimensional tensor signal model is constructed by using the extracted grouping signals, and the shielding state corresponding to each RIS is estimated based on a K-means clustering algorithm; Step S4, channel parameter decoupling and estimation: a shielding processing scheme is designed, and on this basis, an angle, time delay and fading parameter decoupling and estimation method based on tensor decomposition is constructed, to realize complete channel estimation.
2. The method of channel estimation for an occluded environment oriented multi-RIS assisted OFDM system according to claim 1, wherein, The step S1, a multi-RIS assisted OFDM system model facing the shielding scene is established, a RIS shielding factor is introduced, and a cascaded channel model representing the shielding state of the multi-RIS is constructed, specifically comprising: Consider a multi-RIS-aided downlink OFDM system, where the base station and the user terminal are equipped with , , , , , , , , , , , , , (1)； in, , These are the path fading and delay parameters of the cascaded channel corresponding to the p-th RIS block, respectively. The number of subcarriers; , These are the array response vectors of the base station and the user at the p-th RIS, respectively; , Representing the first Block RIS to the array response vector of the user and the base station; Indicates transpose. The imaginary unit; For transmission pilot, The channel is Gaussian white noise; For the first The first time frame The phase shift matrix of the block RIS. This represents vector matrix operations. For RIS Phase shift coefficient of each unit; Let be the occlusion factor vector of the p-th RIS block, where the corresponding element of the occluded unit is 0 and the element of the unoccluded unit is 1.
3. The method of channel estimation for an occluded environment oriented multi-RIS assisted OFDM system according to claim 2, wherein, The step S2, a plurality of RISs are divided into a plurality of groups, a grouping training protocol is designed according to the OFDM system frame structure, and effective signal components corresponding to each group of RISs are extracted by subtracting adjacent time frame received signals, specifically comprising: In adjacent time frames share the same transmission pilot signal, using the vectoring property, the total received signal of one time slot is combined to form a data vector : (2)； wherein is a pilot combining matrix for the th time frame; is a reconstructed th time frame; is a reconstructed th time frame, a pilot matrix for the th time slot, is an identity matrix, denotes a Khatri-Rao product, denotes a Kronecker product, is a noise combining vector for the th time frame; Will Block RIS is divided into Groups, each group contains The block RIS estimates only one set of channels corresponding to each RIS at a time, and the channels to be estimated are... Phase shift matrix of block RIS As the time frame changes, the phase shifts of other RIS remain constant. Calculate the difference between each pair of adjacent time frame signals and filter them out. A differential signal containing only the amount of RIS component to be estimated. : (3)； wherein is a difference phase shift vector, is a phase shift of the th time frame, the th block RIS, is a difference noise vector, 1 , denotes a Hadamard product, , is a received signal and noise on the th time frame, the th subcarrier.
4. The method of channel estimation for an occluded environment oriented multi-RIS assisted OFDM system of claim 3, wherein, The step S3, a three-dimensional tensor signal model is constructed by using the extracted grouping signals, and the shielding state corresponding to each RIS is estimated based on a K-means clustering algorithm, specifically comprising: First, since the pilot matrix is known and column full rank, where , it is removed from the signal using a least squares estimator, resulting in a processed received signal : (4)； wherein is the processed noise, denotes the generalized inverse; The signal processing of the difference time frame is stacked in time dimension to form a data matrix : (5)； wherein , is a stack of phase shifts and noise matrices for the difference time frame Finally, the signals on the subcarriers are combined to construct a three-dimensional signal tensor The original channel estimation problem is converted into a sparse tensor signal recovery problem: (6)； wherein is the combined delay vector of the subcarriers, is a tensor form of the noise, denotes the tensor outer product; the sparsity of the tensor is the number of grouped RIS blocks The factor matrices of the tensor are written as: (7)； (8)； (9)； The tensor CP decomposition algorithm is first used to solve for the factor matrix, and then for the tensor. Model 1 unfold Perform SVD decomposition using the factor matrix The van der Mond property, using rotation-invariant techniques to estimate signal parameters, and the ESPRIT algorithm to estimate the factor matrix. and And then according to Reconstructing the factor matrix ; to the reconstructed factor matrix Estimate occlusion state per column; first use least squares to eliminate known RIS phase shift matrix get vector : (10)； wherein is the factor matrix of the first column; the resulting vector is reshaped into a two-dimensional matrix: (11)； wherein, is a matrix dimension reconstruction function, The elements of are one-to-one corresponding to the reflecting units of the UPA array of the RIS, and All elements are K-means clustered according to the amplitude size, and divided into two clusters, and the element coordinate index sets of the two clusters are and respectively. Accordingly, the RIS array occlusion matrix can be constructed as : (12)； Finally, by deforming the RIS occlusion factor vector to obtain an estimate of (13)。 5. The method of channel estimation for an occluded environment oriented multi-RIS assisted OFDM system of claim 4, wherein, The step S4, a shielding processing scheme is designed, and on this basis, an angle, time delay and fading parameter decoupling and estimation method based on tensor decomposition is constructed, to realize complete channel estimation, specifically comprising the following steps: Step 4.
1. Blockage rate statistics and sorting: Calculate blockage rate for each RIS facet : (14)； wherein, represents 0 norm; sorted from small to large according to the occlusion rate, which provides a basis for subsequent screening and grouping; Step 4.2 Severe occluded RIS screening: set an occlusion threshold When , the RIS is determined to be a large-area occlusion and does not participate in subsequent estimation. Step 4.3 RIS array occlusion correction: process each remaining RIS array one by one, count the RIS array occlusion matrix The number of non-zero elements in each row and column; in subsequent angle parameter estimation, delete the rows and columns with severe occlusion, i.e. the rows and columns with less than 3 non-zero elements, only use the signals of the remaining available reflection units for parameter estimation, and take the average of the estimation results to improve the estimation stability; Step 4.4 multi-RIS scheduling strategy: in the scene of positioning and the like which only needs part of the channel information of the RIS, the optimal RIS group is selected according to the shielding rate for channel estimation and feedback; in the system design scene of RIS phase shift optimization and the like, in order to ensure the overall performance of the system, on the basis of eliminating the RIS with large area shielding, the remaining RISs are grouped and estimated; Subsequently, based on the occlusion processing scheme and the factor matrix obtained by tensor decomposition, the decoupling and estimation of channel parameters are sequentially performed: according to the estimated value of the factor matrix , the angle parameters at the RIS are searched by using the maximum correlation criterion combined with the occlusion processing scheme; according to the estimated value of the factor matrix , the angle parameters at the base station and the user are estimated by using the maximum correlation criterion; the delay parameter is solved according to the estimated value of the factor matrix ; finally, the fading parameter is calculated by scaling between the factor matrices, and the complete multi-RIS channel is restored by using the estimated all channel parameters, so that robust channel estimation in the occlusion environment is realized.