CN-122001719-A - Denoising method and device for channel, electronic equipment and storage medium

CN122001719ACN 122001719 ACN122001719 ACN 122001719ACN-122001719-A

Abstract

The application provides a denoising method, device, electronic equipment and storage medium for a channel, which comprise the steps of obtaining a channel matrix in the channel, constructing an augmented Lagrangian function for removing noise of the channel matrix based on the channel matrix, performing at least one iteration update on preset parameters in the augmented Lagrangian function, and determining a low-rank matrix obtained after the iteration update as a target channel matrix, wherein the target channel matrix is the channel matrix after the noise is removed. By adopting the technical scheme provided by the application, the accuracy and the robustness of channel denoising are improved.

Inventors

WANG CHUNLEI
Xiong Jiu
Meng Fengke
LIU SHEN

Assignees

中国信息安全研究院有限公司

Dates

Publication Date: 20260508
Application Date: 20260309

Claims (10)

1. A method for denoising a channel, the method comprising: acquiring a channel matrix in a channel; constructing an augmented lagrangian function for removing noise of the channel matrix based on the channel matrix; performing at least one iteration update on a preset parameter in the augmented Lagrangian function; and determining the low-rank matrix obtained after iterative updating as a target channel matrix, wherein the target channel matrix is the channel matrix after noise removal.
2. The method for denoising a channel according to claim 1, wherein the iteratively updating the predetermined parameter in the augmented lagrangian function at least once comprises: (A) Determining a first intermediate variable for replacing the current low-rank matrix based on the current low-rank matrix, the sparse matrix, the dual variable and the penalty parameter in the augmented lagrangian function; (B) Determining a new low-rank matrix based on the first intermediate variable, and replacing the current low-rank matrix with the new low-rank matrix; (C) Determining a second intermediate variable for replacing the current low rank matrix based on the new low rank matrix, the current sparse matrix, the current dual variable and the current penalty parameter; (D) Determining a new sparse matrix based on the second intermediate variable, and replacing the current sparse matrix with the new sparse matrix; (E) Determining an iteration residual error based on the channel matrix, the new low-rank matrix and the new sparse matrix; (F) If the quotient of the iteration residual error and the norm of the channel matrix is smaller than a preset tolerance, ending the iteration update; (G) And (C) if the quotient of the norm of the iteration residual error and the norm of the channel matrix is greater than or equal to a preset tolerance, returning to the step (A) to continue to update the preset parameter in an iteration mode.
3. The method of denoising a channel according to claim 2, wherein step (B) comprises: performing singular value decomposition on the first intermediate variable to obtain a singular value vector; performing soft threshold shrinkage processing on the singular value vector to obtain a shrunk singular value vector; and determining a new low-rank matrix according to the singular value vector after the contraction processing, and replacing the current low-rank matrix with the new low-rank matrix.
4. The method of denoising a channel according to claim 2, wherein step (D) comprises: And performing soft threshold shrinkage processing on each element in the second intermediate variable, determining the second intermediate variable after shrinkage processing as a new sparse matrix, and replacing the current sparse matrix with the new sparse matrix.
5. The method of denoising a channel according to claim 2, further comprising: for each iteration update, determining a new dual variable based on the new low rank matrix, the new sparse matrix, the current dual variable and the current penalty parameter, and replacing the current dual variable with the new dual variable.
6. The method of denoising a channel according to claim 2, further comprising: For each iteration update, determining a new penalty parameter based on the current penalty parameter and a preset penalty parameter threshold under the iteration update, and replacing the current penalty parameter with the new penalty parameter.
7. The method of denoising a channel according to claim 1, wherein prior to at least one iterative update of a predetermined parameter in the augmented lagrangian function, the method of denoising further comprises: Initializing the preset parameters according to preset rules.
8. A denoising apparatus of a channel, characterized in that, the denoising device includes: the acquisition module is used for acquiring a channel matrix in the channel; A construction module, configured to construct an augmented lagrangian function for removing noise of the channel matrix based on the channel matrix; the iteration module is used for carrying out at least one iteration update on the preset parameters in the augmented Lagrangian function; And the removing module is used for determining the low-rank matrix obtained after iterative updating as a target channel matrix, wherein the target channel matrix is the channel matrix after noise removal.
9. An electronic device comprising a processor, a memory and a bus, the memory storing machine-readable instructions executable by the processor, the processor and the memory in communication over the bus when the electronic device is in operation, the machine-readable instructions being executable by the processor to perform the steps of the method of denoising a channel as claimed in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium has stored thereon a computer program which, when executed by a processor, performs the steps of the denoising method of a channel as claimed in any one of claims 1 to 7.

Description

Denoising method and device for channel, electronic equipment and storage medium Technical Field The present application relates to the field of channel technologies, and in particular, to a method and apparatus for denoising a channel, an electronic device, and a storage medium. Background In recent years, with the increasing complexity of communication systems, the importance of channel estimation in the communication field is increasing, and the design and implementation of a large number of communication transmission algorithms depend on channels. The channel is usually recovered by a pilot signal, and a channel matrix obtained by pilot recovery is inevitably mixed with a large amount of environmental noise and interference components due to the influence of factors such as complex wireless propagation environment and non-ideal hardware. Therefore, effective denoising of the channel matrix is of great importance. For the problem of channel matrix denoising, various methods have been proposed in the prior art, such as gaussian denoising, denoising based on neural network, generating model and diffusion model. However, the existing method still has the defects in the current channel matrix denoising application that on one hand, part of the denoising results obtained by the existing denoising method under a complex noise environment are low in signal-to-noise ratio and difficult to meet the performance requirement of a high-precision communication system, and on the other hand, the denoising method based on deep learning generally depends on a complex network structure and a large amount of computing resources, so that the computing complexity is high and the processing time is too long. Disclosure of Invention In view of the above, the embodiments of the present application provide a method, an apparatus, an electronic device, and a storage medium for denoising a channel, which improve accuracy and robustness of denoising a channel. The application mainly comprises the following aspects: In a first aspect, an embodiment of the present application provides a denoising method for a channel, where the denoising method includes: acquiring a channel matrix in a channel; constructing an augmented lagrangian function for removing noise of the channel matrix based on the channel matrix; performing at least one iteration update on a preset parameter in the augmented Lagrangian function; and determining the low-rank matrix obtained after iterative updating as a target channel matrix, wherein the target channel matrix is the channel matrix after noise removal. Further, the performing at least one iterative update on the predetermined parameter in the augmented lagrangian function includes: (A) Determining a first intermediate variable for replacing the current low-rank matrix based on the current low-rank matrix, the sparse matrix, the dual variable and the penalty parameter in the augmented lagrangian function; (B) Determining a new low-rank matrix based on the first intermediate variable, and replacing the current low-rank matrix with the new low-rank matrix; (C) Determining a second intermediate variable for replacing the current low rank matrix based on the new low rank matrix, the current sparse matrix, the current dual variable and the current penalty parameter; (D) Determining a new sparse matrix based on the second intermediate variable, and replacing the current sparse matrix with the new sparse matrix; (E) Determining an iteration residual error based on the channel matrix, the new low-rank matrix and the new sparse matrix; (F) If the quotient of the iteration residual error and the norm of the channel matrix is smaller than a preset tolerance, ending the iteration update; (G) And (C) if the quotient of the norm of the iteration residual error and the norm of the channel matrix is greater than or equal to a preset tolerance, returning to the step (A) to continue to update the preset parameter in an iteration mode. Further, step (B) includes: performing singular value decomposition on the first intermediate variable to obtain a singular value vector; performing soft threshold shrinkage processing on the singular value vector to obtain a shrunk singular value vector; and determining a new low-rank matrix according to the singular value vector after the contraction processing, and replacing the current low-rank matrix with the new low-rank matrix. Further, step (D) includes: And performing soft threshold shrinkage processing on each element in the second intermediate variable, determining the second intermediate variable after shrinkage processing as a new sparse matrix, and replacing the current sparse matrix with the new sparse matrix. Further, the denoising method further includes: for each iteration update, determining a new dual variable based on the new low rank matrix, the new sparse matrix, the current dual variable and the current penalty parameter, and replacing the current dual variable with the n