Near-field wideband polarization domain channel estimation method based on multi-candidate bilinear mode detection

Abstract

The near-field broadband polarization domain channel estimation method based on multi-candidate bilinear mode detection disclosed in the present invention belongs to the field of wireless communication technology, and solves the problem of high difficulty and low precision of channel estimation in near-field broadband XL-MIMO system. The existing technology cannot fully handle the complex propagation phenomenon in the polarization domain, resulting in low channel estimation accuracy and difficulty in coping with technical problems of frequency-dependent sparse structures. The near-field broadband polarization domain channel estimation method based on multi-candidate bilinear mode detection provided by the present invention has high precision: multi-candidate mode detection overcomes the limitation that traditional methods can only select a single mode, and significantly improves the accuracy of detection. Strong robustness: adaptive weight adjustment ensures robustness under different channel conditions, especially good performance under low signal-to-noise ratio conditions. Controllable computational complexity: although multi-candidate mode evaluation and weight adjustment are introduced, the overall complexity is still controllable, which is suitable for practical communication applications.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communications, and in particular to a near-field wideband polarization domain channel estimation method based on multi-candidate bilinear pattern detection. Background Art

[0002] With the development of XL-MIMO technology, the capacity and spectrum efficiency of wireless communication systems have been greatly improved, especially in 5G and future 6G networks, where XL-MIMO provides parallel communication services for multiple users through a large number of antenna arrays. However, in near-field scenarios, due to complex propagation phenomena such as spherical wavefront and beam splitting effects, channel modeling and estimation become more complicated, resulting in the inability of traditional channel estimation methods to accurately handle such environments.

[0003] Existing angle-domain channel estimation methods, such as orthogonal matching pursuit (OMP) and sparse Bayesian learning (SBL), are difficult to effectively capture the sparse structure in the polarization domain in near-field scenarios. In particular, in broadband communications, signals of different frequencies propagate in different directions, making the frequency correlation of channel sparsity in the polarization domain more significant. Due to the lack of modeling and processing of this non-uniform sparse structure, the channel estimation performance of traditional methods is poor, especially under low signal-to-noise ratio conditions.

[0004] Therefore, there is an urgent need for a channel estimation method that can effectively capture the complex propagation characteristics in the polarization domain to improve the accuracy and robustness of channel estimation in near-field broadband communications. Summary of the invention

[0005] In view of the shortcomings and deficiencies in the prior art, the present invention provides a method based on multiple candidate bilinear pattern detection (MBPD), which significantly improves the accuracy of channel estimation through multiple candidate pattern evaluation and adaptive weight adjustment strategy.

[0006] To achieve the above objectives, the present invention is implemented by the following technical scheme: The near-field wideband polarization domain channel estimation method based on multi-candidate bilinear mode detection provided by the present invention comprises the following steps:

[0007] S1 builds the near-field channel model of the system;

[0008] S2 multiple candidate bilinear pattern detection MBPD method;

[0009] S3 output results: After multiple rounds of iterations, the MBPD algorithm finally outputs the channel estimation matrix in the polarization domain.

[0010] Preferably, S1 constructs a near-field channel model of the system, and the specific steps are as follows: Assume that the signal transmission process of each user is affected by multiple sparse paths, and the signal strength of each path is given by the complex gain The angle and distance of the path are represented by and It indicates that the channel propagation characteristics conform to the spherical wavefront model; the channel exhibits sparseness in the polarization domain;

[0011] The system uses orthogonal frequency division multiplexing (OFDM) technology. The base station is equipped with a large number of antenna arrays. The signals sent by users propagate in the near-field scenario, and the signals appear as sparse structures in the polarization domain.

[0012] The channel model is modeled by a bilinear mode, and the channel is expressed as:

[0013]

[0014] in, is the number of antenna arrays, is the number of channel paths, Path gain, For distance, is the wavelength, represents the array response vector of the path in the polarization domain.

[0015] Preferably, in the near-field channel model, the array response vector Defined as ,in ,and is scattered from the last hop to the The distance between the base station antennas;

[0016] Based on the geometry of the base station array, for ,and , is the antenna spacing, is the carrier wavelength, is the angle of the path.

[0017] Preferably, step S1 of constructing the system channel model further includes: preprocessing of the received signal,

[0018] First, the signal matrix received by the base station Contains signals from all users and is affected by noise. Pre-whitening is used to reduce the interference of noise on the signal. The signal matrix after processing is: ,in is the inverse matrix of the noise covariance matrix, which is used to pre-whiten the signal.

[0019] Preferably, in step S2, the multi-candidate bilinear pattern detection MBPD method performs bilinear pattern detection on the processed signal matrix, that is, the response value of each pattern is calculated in the angle domain and the distance domain respectively, and the calculation formula is: ,in is the mode response in the polarization domain, which is used to describe the direction and distance of the path;

[0020] In each iteration, select the previous The candidate mode with the largest detection value is used as the basis for updating the support set, and the accumulated detection values ​​are used for subsequent weight adjustment.

[0021] Preferably, the multi-candidate bilinear pattern detection (MBPD) method in step S2 specifically includes two main stages:

[0022] Phase 1: Candidate pattern detection and selection

[0023] (11) Initialization: initial residual matrix Received signal matrix, initial support set is an empty set;

[0024] (12) Mode detection: Detection is performed in the angle domain and distance domain, and the mode response in the polarization domain is calculated: , is the mode response in the polarization domain, which is used to describe the direction and distance of the path;

[0025] (13) Multi-candidate pattern evaluation: Perform cumulative detection on multiple candidate patterns, select the candidate pattern with the largest detection value, and update the support set.

[0026] Preferably, in the MBPD method in step S2,

[0027] Phase 2: Adaptive Weight Adjustment

[0028] (21) Weight initialization: Initialize the weight matrix to the identity matrix Identity matrix;

[0029] (22) Weight update rule: After each iteration, based on the current residual Dynamically adjust the weight matrix; residual Indicates the difference between the current estimation result and the actual received signal, and adjusts the weight to optimize the weighted result of the detection value;

[0030] According to the current detection mode and residual, the weight matrix is ​​dynamically adjusted. The formula for weight adjustment is:

[0031]

[0032] in, is the learning rate, is the loss function; the accuracy of the detection pattern is improved by adjusting the weights;

[0033] (23) Channel estimation support set update: Recalculate the support set based on the updated weight matrix And update the channel estimate.

[0034] Preferably, the channel estimation support set updating method in step (23) is: using the updated support set Recalculate the channel estimation matrix , the residual matrix The update formula is: ,in is the perception matrix contained in the support set.

[0035] Preferably, S3 outputs the result: After multiple rounds of iterations, the MBPD algorithm finally outputs the channel estimation matrix in the polarization domain .

[0036] Preferably, the stopping condition in step S3 is specifically: when the residual matrix When it is less than the set threshold or reaches the preset maximum number of iterations, the iteration process ends and the final channel estimation matrix is ​​output. .

[0037] The present invention provides a near-field broadband polarization domain channel estimation method based on multi-candidate bilinear mode detection. It has the following beneficial effects:

[0038] (1) The near-field wideband polarization domain channel estimation method based on multiple candidate bilinear pattern detection of the present invention improves the accuracy and robustness of channel estimation through a multiple candidate bilinear pattern detection (MBPD) algorithm combined with adaptive weight adjustment, and is particularly suitable for complex near-field wideband communication scenarios; in particular, it relates to a polarization domain channel estimation method suitable for near-field wideband massive multiple-input multiple-output (XL-MIMO) systems.

[0039] (2) Compared with the existing methods, the estimation method of the present invention has the following advantages:

[0040] High precision: Multiple candidate pattern detection overcomes the limitation of traditional methods that can only select a single pattern, significantly improving the accuracy of detection.

[0041] Strong robustness: Adaptive weight adjustment ensures robustness under different channel conditions, especially under low signal-to-noise ratio conditions.

[0042] The computational complexity is controllable: Although multiple candidate mode evaluations and weight adjustments are introduced, the overall complexity is still controllable and suitable for practical communication applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a near-field channel model applicable to any user;

[0044] Figure 2 This is a comparison chart of the normalized mean square error (NMSE) of different channel estimation algorithms under different bandwidth conditions;

[0045] Figure 3 This is a performance comparison chart of signal-to-noise ratio and normalized NMSE of different channel estimation algorithms. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0047] This embodiment demonstrates the application of the method of the present invention in a typical near-field broadband XL-MIMO system, and describes in detail the specific steps and parameter settings of channel estimation. Example 1

[0048] The near-field wideband polarization domain channel estimation method based on multi-candidate bilinear mode detection of the present invention and the polarization domain channel estimation of the near-field wideband XL-MIMO system include the following steps:

[0049] S1. Build the system near-field channel model. The specific steps are as follows:

[0050] S11. System initialization: The system is set to the number of base station antennas , number of users , using OFDM technology, the number of subcarriers is .

[0051] S12. Channel model:

[0052] Assume that the signal transmission process of each user is affected by multiple sparse paths, and the signal strength of each path is given by the complex gain The angle and distance of the path are represented by and It indicates that the channel propagation characteristics conform to the spherical wavefront model; the channel exhibits sparseness in the polarization domain.

[0053] The system uses orthogonal frequency division multiplexing (OFDM) technology. The base station is equipped with a large number of antenna arrays. The signals sent by users propagate in the near-field scenario, and the signals appear as sparse structures in the polarization domain.

[0054] The channel model is modeled by a bilinear mode, and the channel is expressed as:

[0055]

[0056] in, is the number of antenna arrays, is the number of channel paths, Path gain, For distance, is the wavelength, represents the array response vector of the path in the polarization domain.

[0057] like Figure 1 As shown, in the near-field channel model of the present invention, the array response vector Defined as ,in ,and is scattered from the last hop to the Based on the geometric structure of the base station array, It can be approximated as ,and , is the antenna spacing, is the carrier wavelength, is the angle of the path.

[0058] S13. Preprocessing of received signals,

[0059] First, the signal matrix received by the base station Contains signals from all users and is affected by noise. Pre-whitening is used to reduce the interference of noise on the signal. The signal matrix after processing is: ,in is the inverse matrix of the noise covariance matrix, which is used to pre-whiten the signal.

[0060] S14. Bilinear mode detection:

[0061] The processed signal matrix is ​​subjected to bilinear mode detection, that is, the response value of each mode is calculated in the angle domain and the distance domain respectively. The calculation formula is: ,in is the mode response in the polarization domain, which is used to describe the direction and distance of the path;

[0062] In each iteration, select the previous The candidate mode with the largest detection value is used as the basis for updating the support set, and the accumulated detection values ​​are used for subsequent weight adjustment.

[0063] S2. Multiple candidate bilinear pattern detection (MBPD) method, specifically includes two main stages:

[0064] Phase 1: Candidate pattern detection and selection

[0065] (11) Initialization: initial residual matrix (Received signal matrix), initial support set is an empty set;

[0066] (12) Mode detection: Detection is performed in the angle domain and distance domain, and the mode response in the polarization domain is calculated: , is the mode response in the polarization domain, which is used to describe the direction and distance of the path;

[0067] (13) Multi-candidate pattern evaluation: Perform cumulative detection on multiple candidate patterns, select the candidate pattern with the largest detection value, and update the support set.

[0068] Phase 2: Adaptive Weight Adjustment

[0069] (21) Weight initialization: Initialize the weight matrix to the identity matrix (identity matrix);

[0070] (22) Weight update rule: After each iteration, based on the current residual Dynamically adjust the weight matrix; residual Indicates the difference between the current estimation result and the actual received signal, and adjusts the weight to optimize the weighted result of the detection value;

[0071] According to the current detection mode and residual, the weight matrix is ​​dynamically adjusted. The formula for weight adjustment is:

[0072]

[0073] in, is the learning rate, is the loss function; the accuracy of the detection pattern is improved by adjusting the weights;

[0074] (23) Channel estimation support set update: Recalculate the support set based on the updated weight matrix And update the channel estimate.

[0075] Use the updated support set Recalculate the channel estimation matrix , the residual matrix The update formula is: ,in is the perception matrix contained in the support set.

[0076] S3 output results: After multiple rounds of iterations, the MBPD algorithm finally outputs the channel estimation matrix in the polarization domain .

[0077] The specific stopping condition is: when the residual When it is less than the set threshold or reaches the preset maximum number of iterations, the iteration process ends and the final channel estimation matrix is ​​output. . Example 2

[0078] The near-field wideband polarization domain channel estimation method based on multi-candidate bilinear mode detection of the present invention is verified with the performance of other existing estimation methods. The normalized mean square error NMSE of the present invention method and other methods under different bandwidth conditions are compared. The normalized mean square error and signal-to-noise ratio performance of the present invention method and other methods are compared. The specific results and their analysis are as follows:

[0079] Figure 2 A comparison chart of the normalized mean square error (NMSE) between the method of the present invention and other methods under different bandwidth conditions is given. Figure 3 The normalized mean square error and signal-to-noise ratio performance comparison diagram of the method of the present invention and other methods is given. Figure 2 and Figure 3 As shown in the simulation data results, the results show that the method of the present invention exhibits higher channel estimation accuracy than the traditional OMP algorithm in high noise scenarios, especially in near-field broadband systems, and the estimated mean square error NMSE is reduced by about 5dB.

[0080] Figure 2 The performance comparison of various channel estimation algorithms in the bandwidth range of 0 to 10 GHz is shown in Figure 2, and it can be found that the method of the present invention always achieves the lowest NMSE value, starting from below -18 dB and stabilizing at around -17 dB, which is significantly better than other methods. This is because OMP does not fully utilize the multidimensional sparsity of the channel. The method of the present invention maintains excellent performance, with minimal performance degradation as the bandwidth increases, highlighting its robustness and accuracy. In contrast, the angle domain OMP and SOMP methods show higher NMSE values, indicating their limited effectiveness in processing wideband signals. The polarization domain OMP and SOMP methods show moderate performance, but still lag behind the bilinear mode detection method and the method of the present invention. The performance of the beam splitting mode detection method is better than the angle domain and polarization domain methods, but the adaptive weight matrix and multiple candidate evaluations of the method of the present invention ensure the best overall performance.

[0081] Figure 3The normalized mean square error and signal-to-noise ratio performance of the proposed method are compared with other methods. The figure shows that the proposed method consistently achieves the lowest NMSE values ​​at all signal-to-noise ratio levels, demonstrating excellent estimation accuracy and robustness. Specifically, at 0 dB signal-to-noise ratio, the NMSE of the proposed method is about -15 dB, which increases to about -22 dB at 15 dB signal-to-noise ratio. In contrast, the angle domain OMP and SOMP methods show the highest NMSE values, indicating poor performance at all signal-to-noise ratio levels, with NMSE values ​​remaining above -8 dB. The beam splitting mode detection method performs slightly better than the angle domain method, with NMSE improving from -5 dB to about -10 dB. The polarization domain OMP method shows moderate performance, improving from -8 dB to -14 dB, while the polarization domain SOMP method shows a small improvement starting from around -5 dB. The bilinear mode detection method performs better than the angle domain and polarization domain methods, with NMSE improving from -10 dB to -18 dB as the signal-to-noise ratio increases. The method of the present invention improves NMSE by about 5 dB to 10 dB over the BPD algorithm and by about 10 dB to 20 dB over the angle domain and polarization domain methods as the signal-to-noise ratio increases. The superior performance of the method of the present invention is attributed to its adaptive weight matrix and multiple candidate mode evaluations, which enhance robustness and accuracy. The adaptive weight matrix allows the algorithm to dynamically adjust to different channel conditions, while the multiple candidate evaluations help to more efficiently navigate the solution space.

[0082] On the basis of the above embodiments, the present invention continues to describe in detail the technical features involved therein and the functions and roles played by the technical features in the present invention, so as to help technicians in this field fully understand the technical solution of the present invention and reproduce it.

[0083] Finally, although this specification is described according to implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. A near-field wideband polarization domain channel estimation method based on multi-candidate bilinear mode detection, characterized in that: The steps include: S1. Construct the near-field channel model of the system: Assume that the signal transmission process of each user is affected by multiple sparse paths, and the signal strength of each path is determined by the complex gain The angle and distance of the path are represented by and It indicates that the channel propagation characteristics conform to the spherical wavefront model; the channel exhibits sparseness in the polarization domain; The system uses orthogonal frequency division multiplexing (OFDM) technology. The base station is equipped with a large number of antenna arrays. The signals sent by users propagate in the near-field scenario, and the signals appear as sparse structures in the polarization domain. The channel model is modeled by a bilinear mode, and the channel is expressed as: ， in, is the number of antenna arrays, is the number of channel paths, Path gain, For distance, is the wavelength, represents the array response vector of the path in the polarization domain; In the near-field channel model, the array response vector Defined as ,in ,and is scattered from the last hop to the The distance between the base station antennas; Based on the geometry of the base station array, for ,and , is the antenna spacing, is the carrier wavelength, is the angle of the path; S2. Execute the multi-candidate bilinear pattern detection (MBPD) method: It includes two stages: Phase 1: Candidate pattern detection and selection (11) Initialization: initial residual matrix Received signal matrix, initial support set is an empty set; (12) Mode detection: Detection is performed in the angle domain and distance domain, and the mode response in the polarization domain is calculated: , is the mode response in the polarization domain, which is used to describe the direction and distance of the path; (13) Multiple candidate pattern evaluation: cumulatively detect multiple candidate patterns, select the candidate pattern with the largest detection value, and update the support set; Phase 2: Adaptive Weight Adjustment (21) Weight initialization: Initialize the weight matrix to the identity matrix Identity matrix; (22) Weight update rule: After each round of iteration, the weight matrix is ​​dynamically adjusted according to the current detection mode and residual. The formula for weight adjustment is: ， in, is the learning rate, is the loss function; the accuracy of the detection pattern is improved by adjusting the weights; (23) Channel estimation support set update: Use the updated support set Recalculate the channel estimation matrix , the residual matrix The update formula is: ,in is the perception matrix contained in the support set; S3 output results: After multiple rounds of iterations, the MBPD algorithm finally outputs the channel estimation matrix in the polarization domain.

2. The near-field wideband polarization domain channel estimation method based on multi-candidate bilinear mode detection according to claim 1 is characterized in that: Step S1 of constructing the system channel model also includes: preprocessing of the received signal. First, the signal matrix received by the base station Contains signals from all users and is affected by noise. Pre-whitening is used to reduce the interference of noise on the signal. The signal matrix after processing is: ,in is the inverse matrix of the noise covariance matrix, which is used to pre-whiten the signal.

3. The near-field wideband polarization domain channel estimation method based on multi-candidate bilinear mode detection according to claim 1 is characterized in that: S3 output results: After multiple rounds of iterations, the MBPD algorithm finally outputs the channel estimation matrix in the polarization domain .

4. The near-field wideband polarization domain channel estimation method based on multi-candidate bilinear mode detection according to claim 3 is characterized in that: The stopping condition in step S3 is specifically: when the residual matrix When it is less than the set threshold or reaches the preset maximum number of iterations, the iteration process ends and the final channel estimation matrix is ​​output. .

Images