Multi-user communication system channel state information feedback method and device and related equipment
By extracting and fusing the characteristics of channel signals in the angle-delay domain, using the channel state information model for compression and reconstruction, the problem of low accuracy of channel state information feedback in large-scale phased array antennas is solved, and communication performance is improved Signal transmission quality.
Patent Information
- Application Number
- CN202411004735.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-29
- Estimated Expiration
- 2044-07-25
AI Technical Summary
The channel state information feedback method in traditional LEO satellite multi-beam multi-user communication systems is not suitable for large-scale phased array antennas, resulting in low accuracy in channel reconstruction and affecting signal transmission quality.
By extracting the peak characteristics and regional correlation characteristics of the channel signal in the angle-delay domain, fuse it into a sparsity change feature matrix, and compressing and reconstructing it using the channel state information model, reducing the CSI feedback overhead and improving communication performance.
It improves the accuracy of channel state information feedback and system communication performance, reduces the overhead of CSI feedback, and is suitable for large-scale phased array antenna satellite communication.
Smart Images

Figure CN119051701A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of satellite communication technology, and in particular to a method, device and related equipment for feedback of channel state information of a multi-user communication system. Background Art
[0002] Phased array antennas have great flexibility in beam forming, beam reconstruction and beam scanning technologies, and have strong anti-interference capabilities. They have been widely used in low earth orbit (LEO) satellite communication systems. For satellite Internet, in order to give full play to the advantages of phased array antennas, satellites need to accurately obtain channel state information (CSI). The existing idea of obtaining channel state information is that the satellite first sends a pre-agreed pilot to the user terminal (UT), and the user terminal obtains the downlink channel state information through the pilot calculation, and then sends it to the satellite through the feedback link. Channel state information feedback technology includes two categories: explicit feedback and implicit feedback. Implicit feedback is feedback based on the codebook. The feedback principle is that the satellite and the user terminal need to save the codebook in advance before the communication starts. The user receives the pilot sent by the satellite and obtains the downlink channel state information through the channel estimation technology. Then the user calculates the optimal codeword index of the channel state information according to the codeword selection rule and sends it to the satellite through the feedback link. After receiving the feedback codeword index, the satellite searches according to the codebook to obtain the corresponding codeword, that is, the user's feedback information. Although the codebook-based channel state information feedback method has low computational complexity, with the application of large-scale satellite phased array antennas, the number of antennas will increase dramatically, and the codebook size will increase exponentially. At the same time, the computational complexity and complexity of searching for the optimal codeword algorithm will increase dramatically. Therefore, the codebook-based channel state information feedback method is not suitable for large-scale phased array antenna satellite communications. Explicit feedback technology includes channel state information feedback based on compressed sensing. Since the channel characteristics of multi-beam and multi-user satellite phased array antennas are more complex, the sparse matrix commonly used in compressed sensing-based channel state information feedback is difficult to characterize the channel characteristics, resulting in insufficient sparsity of the observed channel state information. Since the accuracy of the reconstructed channel depends on the expression of the sparsity of the channel state information by the observed signal, the insufficient sparsity of the observed channel state information will directly affect the accuracy of the reconstructed channel, thus limiting the application of compressed sensing feedback methods in satellite Internet communications. Summary of the invention
[0003] The present invention provides a multi-user communication system channel state information feedback method, device and related equipment to solve the defects that the traditional LEO satellite multi-beam multi-user communication system channel state information feedback method is not suitable for large-scale phased array antenna satellite communication and the accuracy of reconstructed channel is low, which affects the signal transmission quality.
[0004] The present invention provides a method for feeding back channel state information of a multi-user communication system, comprising: Extracting the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; The peak feature and the regional correlation feature are integrated to obtain a sparse variation feature matrix of the low-orbit satellite channel as the satellite moves in different orbital positions; Inputting the sparsity variation characteristic matrix of the low-orbit satellite channel as the satellite moves in different orbital positions into a channel state information model to obtain a low-dimensional vector; Sending the low-dimensional vector to a satellite, so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; Wherein, the channel state information model is obtained by training based on a high-dimensional channel state information matrix.
[0005] According to the multi-user communication system channel state information feedback method provided by the present invention, the channel state information model includes an encoder; The encoder is arranged at the user end, and is used to compress the input sparse matrix in the angle-delay domain, and output the compressed low-dimensional channel state information vector to feed back to the satellite; During the training process of the channel state information model, the sparse matrix in the input angle-delay domain is passed through two convolutional layers and then through a lightweight self-attention feature extraction module to obtain a high-dimensional feature map, wherein the high-dimensional feature map includes feature information of the peak value and information around the peak; The high-dimensional feature map is passed through two convolutional layers again and then dimensionally transformed to obtain a high-dimensional vector; The high-dimensional vector is reduced in dimension using a fully connected layer to obtain a low-dimensional vector.
[0006] According to the channel state information feedback method for a multi-user communication system provided by the present invention, the channel state information model includes a decoder; The decoder is arranged at the satellite end, and the decoder is used to reconstruct the low-dimensional vector; During the channel state information model training process, the low-dimensional vector is converted into a high-dimensional vector through a fully connected layer; Transforming the dimension of the high-dimensional vector to obtain a high-dimensional feature matrix; Convolving the high-dimensional feature matrix through four convolutional layers respectively, wherein the number of convolution kernels of the four convolutional layers gradually increases, to obtain features of multiple channels; The features of the multiple channels are aggregated and reconstructed into a high-dimensional channel state information matrix.
[0007] According to the multi-user communication system channel state information feedback method provided by the present invention, the extracting peak characteristics of the channel signal in the angle-delay domain includes: Extracting information in the vertical and horizontal directions of the input matrix corresponding to the channel signal through pooling layers with different kernel sizes; The vertical and horizontal information are merged into a feature map containing both angle domain and delay domain information; Extract the peak features of the feature graph that contains both angle domain and delay domain information.
[0008] According to the multi-user communication system channel state information feedback method provided by the present invention, the extraction of regional correlation characteristics of the channel signal in the angle delay domain includes: After the input matrix corresponding to the channel signal is convolved through two convolution layers, the channel signal is processed through two channels, one of which expands the features after the convolution operation to obtain a first matrix, and the rows of the first matrix are summed to obtain a first column vector; Performing dimension transformation through another channel to obtain a second column vector and a third column vector, multiplying the second column vector and the third column vector by the first column vector respectively, and concatenating the obtained results in the third dimension to generate a second matrix; The second matrix is reconstructed into a dimension consistent with the size of the input matrix, and the regional correlation characteristics of the channel signal in the angle delay domain are output.
[0009] According to the multi-user communication system channel state information feedback method provided by the present invention, the feature matrix after the convolution operation is expanded to obtain the first matrix, including: Expanding the periphery of the characteristic matrix by mirroring to add two rows and two columns of elements; A polymerization area of a preset size is set, and it is used to move row by row and column by column on the expanded matrix for raster scanning, and the scanned area is recombined into a row vector with the number of elements being the product of the region height, the region width and the number of channels, and all the scanned deformed row vectors are superimposed to obtain the first matrix.
[0010] According to the multi-user communication system channel state information feedback method provided by the present invention, the fusing of the peak feature and the regional correlation feature comprises: The peak feature and the regional correlation feature are fused using a lightweight feature fusion layer, and the new mapping feature output by the feature fusion layer is for:
[0011] in, is the input matrix corresponding to the channel signal, is the peak feature matrix, is the regional correlation feature matrix, , , are the weight parameters of the input matrix, peak feature matrix and regional correlation feature matrix corresponding to the channel signal, respectively, , , Obtained through model training.
[0012] The present invention also provides a multi-user communication system channel state information feedback device, comprising: An extraction module, used to extract the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; A fusion module, used to fuse the peak feature and the regional correlation feature to obtain a sparse change feature matrix of the low-orbit satellite channel as the satellite moves in different orbital positions; An input module is used to input the sparsity change characteristic matrix of the low-orbit satellite channel as the satellite moves in different orbital positions into a channel state information model to obtain a low-dimensional vector; A reconstruction module, used for sending the low-dimensional vector to a satellite, so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; Wherein, the channel state information model is obtained by training based on a high-dimensional channel state information matrix.
[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method for feedback of channel state information of a multi-user communication system as described in any one of the above items is implemented.
[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements any of the above-mentioned methods for feedback of channel state information of a multi-user communication system.
[0015] The multi-user communication system channel state information feedback method, device and related equipment provided by the present invention extract the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; fuse the peak characteristics and the regional correlation characteristics to obtain the sparse change characteristic matrix of the low-orbit satellite channel with the different orbital positions of the satellite movement; input the sparse change characteristic matrix of the low-orbit satellite channel with the different orbital positions of the satellite movement into the channel state information model to obtain a low-dimensional vector; send the low-dimensional vector to the satellite so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; wherein the channel state information model is trained based on the high-dimensional channel state information matrix. The CSI is compressed and reconstructed through the channel state information model, reducing the burden of CSI display feedback overhead and improving the system communication performance; the channel state information model can extract the changing data characteristics in the channel matrix, improve the accuracy of reconstructed CSI, and further improve the system performance. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0017] Figure 1 This is one of the flow charts of the channel state information feedback method for a multi-user communication system provided by an embodiment of the present invention; Figure 2 is a schematic diagram of the structure of a channel state information model provided by an embodiment of the present invention; Figure 3 is a schematic diagram of a peak extraction process provided by an embodiment of the present invention; Figure 4 is a schematic diagram of a regional correlation extraction process provided by an embodiment of the present invention; Figure 5 is a schematic diagram of the expansion operation flow provided by an embodiment of the present invention; Figure 6 It is a schematic diagram of a lightweight feature fusion process provided by an embodiment of the present invention; Figure 7 This is a second flow chart of a method for feeding back channel state information of a multi-user communication system provided by an embodiment of the present invention; Figure 8 It is a functional structure diagram of a channel state information feedback device for a multi-user communication system provided by an embodiment of the present invention; Fig. 9It is a functional structure diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0018] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0019] Figure 1 A flow chart of a method for feeding back channel state information of a multi-user communication system provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the channel state information feedback method of a multi-user communication system provided by an embodiment of the present invention includes: Step 101, extracting the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; Step 102: The peak feature and the regional correlation feature are merged to obtain a sparsity variation feature matrix of the low-orbit satellite channel as the satellite moves in different orbital positions; Step 103: Input the sparsity variation characteristic matrix of the low-orbit satellite channel as the satellite moves in different orbital positions into a channel state information model to obtain a low-dimensional vector; Step 104: Send the low-dimensional vector to a satellite, so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; The channel state information model is obtained based on high-dimensional channel state information matrix training
[0020] Traditional channel state information feedback methods for multi-user communication systems include implicit feedback and explicit feedback. The overhead of traditional implicit feedback methods is proportional to the number of antennas. As the number of antennas increases, feedback will generate a large amount of overhead, seriously reducing the spectrum efficiency of the communication system. With the application of large-scale satellite-borne phased array antennas, the number of antennas is gradually increasing. Due to the high feedback cost, it is increasingly challenging for satellites to obtain accurate downlink CSI. That is, as the number of antennas increases, downlink CSI feedback occupies a large amount of frequency band resources, and the contradiction between feedback overhead and system performance gradually intensifies. It is crucial to study CSI feedback technology that ensures CSI feedback accuracy and reduces CSI feedback overhead. The massive antenna multi-beam multi-user channel has a sparse feature in the angle-delay domain, which provides support for the compressibility of the channel matrix. Unlike terrestrial communications, the energy peak of the channel of the satellite communication system in the angle-delay domain changes with the movement of the satellite and the change of azimuth, and the channel sparsity characteristics will also change accordingly. The traditional method of extracting features for compression in explicit feedback ignores the changes in the sparse characteristics of the channel in the angle domain and the characteristics that the energy peak changes with the azimuth angle of satellite motion. Therefore, the obtained compressed vector loses the information of the channel in the angle domain, which will lead to a decrease in the accuracy of reconstructing the CSI matrix.
[0021] The channel state information feedback method for a multi-user communication system provided by an embodiment of the present invention extracts the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; the peak characteristics and the regional correlation characteristics are merged to obtain the sparse change characteristic matrix of the low-orbit satellite channel with different orbital positions of the satellite movement; the sparse change characteristic matrix of the low-orbit satellite channel with different orbital positions of the satellite movement is input into the channel state information model to obtain a low-dimensional vector; the low-dimensional vector is sent to the satellite so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; wherein the channel state information model is obtained based on the training of the high-dimensional channel state information matrix. The CSI is compressed and reconstructed through the channel state information model, the burden of CSI display feedback overhead is reduced, and the system communication performance is improved; the channel state information model can extract the changing data characteristics in the channel matrix, improve the accuracy of reconstructed CSI, and further improve the system performance.
[0022] Based on any of the above embodiments, the channel state information model includes an encoder; The encoder is arranged at the user end, and is used to compress the input sparse matrix in the angle-delay domain, and output the compressed low-dimensional channel state information vector to feed back to the satellite; During the training process of the channel state information model, the sparse matrix in the input angle-delay domain is passed through two convolutional layers and then through a lightweight self-attention feature extraction module to obtain a high-dimensional feature map, wherein the high-dimensional feature map includes feature information of the peak value and information around the peak; The high-dimensional feature map is passed through two convolutional layers again and then dimensionally transformed to obtain a high-dimensional vector; The high-dimensional vector is reduced in dimension using a fully connected layer to obtain a low-dimensional vector.
[0023] In the embodiment of the present invention, the low-orbit satellite channel is more complex than the ground channel, and the structural characteristics in the angle-delay domain change with the azimuth of the satellite motion. In order to improve the accuracy of reconstructing the CSI matrix, the sparse characteristics of the channel structure in the angle domain changing with the satellite motion can be extracted and compressed into a low-dimensional vector, which has the characteristic information of the channel in the angle-delay domain.
[0024] Based on any of the above embodiments, the channel state information model includes a decoder; The decoder is arranged at the satellite end, and the decoder is used to reconstruct the low-dimensional vector; During the channel state information model training process, the low-dimensional vector is converted into a high-dimensional vector through a fully connected layer; Transforming the dimension of the high-dimensional vector to obtain a high-dimensional feature matrix; Convolving the high-dimensional feature matrix through four convolutional layers respectively, wherein the number of convolution kernels of the four convolutional layers gradually increases, to obtain features of multiple channels; The features of the multiple channels are aggregated and reconstructed into a high-dimensional channel state information matrix.
[0025] In the embodiment of the present invention, the channel state information model is as follows: Figure 2 As shown in the figure, the encoder compresses the input CSI at the user end and outputs the compressed CSI vector to the satellite through the uplink channel. The decoder at the satellite side inputs the received compressed vector and outputs the recovered reconstructed CSI matrix. The input of the model network is a sparse matrix in the angle-delay domain. During network training, the input matrix First, a convolutional layer with two 3×3 convolution kernels is used, and then a lightweight self-attention feature extraction module is used to obtain a feature map that contains both peak values and information around the peaks. Then, a convolutional layer with two 3×3 convolution kernels is used to increase the nonlinear structure of the encoder, and the output matrix is transformed into a vector form through dimensionality transformation. Finally, a fully connected layer is used to simulate the projection function of the compressed sensing algorithm to reduce the dimensionality of the vector and obtain a low-dimensional vector. , which is the codeword obtained by the compression process.
[0026] Compressed low-dimensional vector The satellite reconstructs the channel matrix through the decoder. In the decoder, the low-dimensional vector It is converted into a high-dimensional vector through a fully connected layer, and then converted into the size of the CSI matrix through a dimensional transformation. The matrix passes through four convolutional layers, each with a convolution kernel size of 3×3 and the number of convolution kernels in each layer gradually increases to 2, 8, 16, and 32. After the last convolution layer with 32 convolution kernels, the number of channels of the matrix is 32. In order to aggregate multiple channels, the last convolution layer is set to have two convolution kernels of size 1×1. In this process, the reconstructed CSI matrix is continuously refined. Since multiple convolutional layers are used, a residual structure is used to avoid the gradient vanishing problem caused by superimposed nonlinear transformations. Finally, the output of the encoder is is the reconstructed CSI matrix obtained.
[0027] In the embodiment of the present invention, reconstructing the CSI matrix through the low-dimensional vector output by the model can improve the accuracy of the CSI feedback matrix.
[0028] Based on any of the foregoing embodiments, the step of extracting a peak feature of a channel signal in an angle-delay domain includes: Step 201: extracting information in the vertical and horizontal directions of the input matrix corresponding to the channel signal through pooling layers with different kernel sizes; Step 202: Fusing the vertical and horizontal information into a feature map containing both angle domain and delay domain information; Step 203: extract the peak features of the feature graph that contains both angle domain and delay domain information.
[0029] As the satellite moves, the azimuth and path delay are constantly changing, so the power peak position in the CSI angle-delay domain is also constantly changing. Accurately identifying and extracting the peak position is of great significance for generating feature maps and subsequent compression processes. In the satellite CSI structure, the vertical and horizontal directions correspond to the delay domain and the angle domain, respectively, and strip pooling kernels can be used to extract peak information separately. Use pooling kernels of sizes 44×1 and 1×26 to extract CSI information in the delay domain and angle domain, respectively, to avoid capturing irrelevant information in the other domain. The extraction process of the peak extraction block is as follows: Figure 3 This process includes: First, two kernels are used to perform pooling operations to extract vertical and horizontal information. The row vector containing the vertical delay domain information The calculation is as follows:
[0030] in is the module input matrix, Represents the input matrix The element in the i-th row and j-th column of . is the output row vector The j-th element of .
[0031] Column vector containing horizontal angle domain information The calculation is as follows:
[0032] in is the output column vector Then the row vector and column vector Expand along columns and rows respectively With matrices of the same size and .
[0033] Then, and Fusion into a new feature map containing both angle domain and delay domain information , calculated as follows:
[0034] Finally, the peak extraction module The output new feature map is calculated as follows:
[0035] in To calculate the Hadamard product. The new feature map output Each position contains The information of the corresponding position in the angle domain and the time delay domain.
[0036] Based on any of the above embodiments, the extracting the regional correlation characteristics of the channel signal in the angle delay domain includes: Step 301: After the input matrix corresponding to the channel signal is convolved through two convolutional layers, the channel signal is processed through two channels, one of which expands the features after the convolution operation to obtain a first matrix, and the rows of the first matrix are summed to obtain a first column vector; Step 302: Perform dimensional transformation through another channel to obtain a second column vector and a third column vector, multiply the second column vector and the third column vector by the first column vector respectively, and connect the obtained results in the third dimension to generate a second matrix; Step 303: reconstruct the second matrix into a dimension consistent with the size of the input matrix, and output the regional correlation characteristics of the channel signal in the angle delay domain.
[0037] In the embodiment of the present invention, considering that the area around the peak has strong regional correlation, extracting regional correlation is also important for extracting channel features. The traditional self-attention model calculates the correlation coefficient of each pixel on the entire feature map, but ignores the positional relationship between different pixels. Considering that the area near the peak of the channel matrix has strong correlation, a new regional self-attention model is adopted, which limits the scope of self-attention to a local range instead of the entire feature map. Since the dimensionality reduction operation is performed in the subsequent network process, the embedding layer of the regional self-attention model is simplified. Simplified self-attention model The calculation is as follows:
[0038] in, It's location The relevant area indicates which elements will be aggregated into new features. Belongs to The position index. Is an aggregate function used to calculate and Region The relevance of the inner elements.
[0039] The generation process of region-related extraction blocks is as follows: Figure 4 As shown. Input matrix After convolution with two convolution layers with 1×1 kernels, two channels are aggregated to obtain tensor1. This process increases the nonlinear characteristics of the model. Then tensor1 is expanded to obtain a two-dimensional matrix tensor2, and the sum of the rows is obtained to obtain the column vector tensor3. At this time, each element of tensor3 contains Each element in All information about the elements in the region. At the same time, tensor1 is reshaped into two column vectors tensor4 and tensor5. Tensor4 and tensor5 are multiplied by tensor3 respectively and concatenated in the third dimension to generate a two-dimensional matrix tensor6. Finally, tensor6 is reshaped into the same Dimensions of the same size, output matrix .
[0040] In the embodiment of the present invention, the step of expanding the features after the convolution operation to obtain a first matrix includes: Expanding the periphery of the characteristic matrix by mirroring to add two rows and two columns of elements; A polymerization area of a preset size is set, and it is used to move row by row and column by column on the expanded matrix for raster scanning, and the scanned area is recombined into a row vector with the number of elements being the product of the region height, the region width and the number of channels, and all the scanned deformed row vectors are superimposed to obtain the first matrix.
[0041] like Figure 5 As shown, the size of each feature matrix is marked. First, the periphery of the feature matrix (26*44*2) is expanded by mirroring to add two rows and two columns of elements. The expanded row above the first row is the last row of the original matrix, and the expanded column to the left of the first column is the last column of the original matrix. The same is true for the last row and the last column. The expanded elements in the corner part are the elements in the original matrix that are symmetrical to its diagonal. Then a 3×3 aggregation area is set, and it is used to move row by row and column by column on the expanded matrix for raster scanning. The purpose is to recombine the scanned area into a row vector with 3×3×2=18 elements, and finally superimpose all the row vectors after scanning deformation into a 1144×18 matrix. At this point, the original matrix of 26×44 is "unfolded" into a first matrix of 1144×18.
[0042] In the embodiment of the present invention, the fusing of the peak feature and the regional correlation feature includes: The peak feature and the regional correlation feature are fused using a lightweight feature fusion layer, and the new mapping feature output by the feature fusion layer is for:
[0043] in, is the input matrix corresponding to the channel signal, is the peak feature matrix, is the regional correlation feature matrix, , , are the weight parameters of the input matrix, peak feature matrix and regional correlation feature matrix corresponding to the channel signal, respectively, , , Obtained through model training.
[0044] In the embodiment of the present invention, after the channel peak position is extracted and the correlation of the area around the peak is extracted, a lightweight feature combination layer is used to fuse these two parts of features together to generate a new mapping feature with both the peak position and the area correlation.
[0045] The structure of the lightweight feature fusion layer is as follows: Figure 6As shown in FIG. 1 , the features of the two feature extraction modules are fused through three bits of trainable parameters. The output new feature map contains both peak information and correlation information of the area around the peak.
[0046] In an embodiment of the present invention, the energy peak and regional correlation information of the channel in the angle-delay domain are first extracted through the attention mechanism, and then the information is fused through a lightweight feature fusion layer to form a feature map with peak information and regional correlation information, and finally a low-dimensional vector is obtained by projection through a fully connected layer to improve the information degree of the low-dimensional vector.
[0047] Based on any of the above embodiments, the embodiment of the present invention includes two stages: initial access and information transmission. Figure 7 As shown in the figure, the dotted line represents the initial access stage, and the solid line represents the information transmission stage. The initial access includes five steps: the user sends the user channel to the satellite, the satellite integrates the user channel into the system channel and sends it to the gateway, the gateway trains the network, the gateway sends the network structure and network parameters to the satellite, and the satellite sends some parameters of the encoder in the network to the user. The information transmission stage includes five steps: user channel processing, the user compresses the CSI into a low-dimensional vector through the encoder, the user feeds back the low-dimensional vector to the satellite, the satellite reconstructs the low-dimensional vector into a high-dimensional CSI matrix through the decoder, and the satellite integrates the reconstructed user matrix into the system channel matrix.
[0048] Initial access phase: Step 1: The user sends the user channel to the satellite, and each UT sends its own space-frequency domain user channel matrix to the satellite .
[0049] Step 2: The satellite integrates the user channels into the system channel and sends it to the gateway. The satellite receives the channel matrix of these users. And combine them into a system channel matrix :
[0050] The system channel matrix Send to the gateway station.
[0051] Step 3: The gateway trains the network and transmits the As the training data of the network, the network is trained. The network consists of two parts: encoder and decoder.
[0052] Step 4: The gateway sends the network structure and network parameters to the satellite. , decoder parameters Sent to satellite.
[0053] Step 5: The satellite sends some parameters of the encoder in the network to the user, and the satellite sends the decoder parameters Sent to the user, the initial access phase ends.
[0054] Information transmission stage: Step 1: User channel processing, converting the user channel matrix from the space-frequency domain to the angle-delay domain through two-dimensional Fourier transform:
[0055] Then the user's value close to zero in the angle-delay domain is truncated.
[0056] Step 2: The user compresses the CSI into a low-dimensional vector through the encoder, inputs the channel matrix after user channel processing into the encoder, and outputs the compressed low-dimensional vector :
[0057] Step 3: The user feeds back the low-dimensional vector to the satellite. The user converts the low-dimensional vector output by the encoder into Feedback to the satellite.
[0058] Step 4: The satellite reconstructs the low-dimensional vector into a high-dimensional CSI matrix through the decoder. The satellite receives the feedback low-dimensional vector Input into the decoder and output the reconstructed user CSI matrix :
[0059] Step 5: The satellite integrates the reconstructed user matrix into the system channel matrix and converts the reconstructed user channel matrix into the space-delay domain through a two-dimensional Fourier transform. :
[0060] Integrate the user channel matrix to form the system channel matrix :
[0061] This is the feedback CSI obtained by the satellite.
[0062] Traditional CSI feedback technology compresses the CSI channel through traditional compression algorithms, but traditional compression algorithms can only capture linear relationships in data. Since there are many nonlinear changes in satellite channels, they cannot be captured by traditional compression algorithms, so the compression rate of traditional compression algorithms is low.
[0063] The multi-user communication system channel state information feedback method provided by the embodiment of the present invention can reduce the CSI feedback burden of the satellite large-scale antenna multi-beam multi-user system and improve the communication system capacity. It also improves the accuracy of the satellite in obtaining reconstructed CSI. The nonlinear data relationship in the CSI data is captured by the machine learning algorithm to achieve more efficient compression, reduce the amount of CSI feedback data, and improve the system communication capacity.
[0064] The channel state information feedback device for a multi-user communication system provided by the present invention is described below. The channel state information feedback device for a multi-user communication system described below and the channel state information feedback method for a multi-user communication system described above can be referred to each other.
[0065] Figure 8 A schematic diagram of a channel state information feedback device for a multi-user communication system provided by an embodiment of the present invention, such as Figure 8 As shown, the channel state information feedback device of a multi-user communication system provided by an embodiment of the present invention includes: An extraction module 801 is used to extract the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; A fusion module 802 is used to fuse the peak feature and the regional correlation feature to obtain a sparse change feature matrix of the low-orbit satellite channel as the satellite moves in different orbital positions; An input module 803 is used to input the sparsity change characteristic matrix of the low-orbit satellite channel as the satellite moves in different orbital positions into a channel state information model to obtain a low-dimensional vector; A reconstruction module 804 is used to send the low-dimensional vector to the satellite, so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; Wherein, the channel state information model is obtained by training based on a high-dimensional channel state information matrix.
[0066] The channel state information feedback device for a multi-user communication system provided by an embodiment of the present invention extracts the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; the peak characteristics and the regional correlation characteristics are merged to obtain the sparse change characteristic matrix of the low-orbit satellite channel with different orbital positions of the satellite movement; the sparse change characteristic matrix of the low-orbit satellite channel with different orbital positions of the satellite movement is input into the channel state information model to obtain a low-dimensional vector; the low-dimensional vector is sent to the satellite so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; wherein the channel state information model is trained based on the high-dimensional channel state information matrix. The CSI is compressed and reconstructed through the channel state information model, the burden of CSI display feedback overhead is reduced, and the system communication performance is improved; the channel state information model can extract the changing data characteristics in the channel matrix, improve the accuracy of reconstructed CSI, and further improve the system performance.
[0067] Fig. 9 An example of a physical structure diagram of an electronic device is shown in FIG. Fig. 9 As shown, the electronic device may include: a processor (processor) 910, a communication interface (Communications Interface) 920, a memory (memory) 930 and a communication bus 940, wherein the processor 910, the communication interface 920, and the memory 930 communicate with each other through the communication bus 940. The memory 930 includes a computer program, an operating system, and acquired data. The processor 910 can call the logic instructions in the memory 930 to execute a channel state information feedback method for a multi-user communication system. The method includes: extracting the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; fusing the peak characteristics and the regional correlation characteristics to obtain a sparse change characteristic matrix of the low-orbit satellite channel as the satellite moves at different orbital positions; inputting the sparse change characteristic matrix of the low-orbit satellite channel as the satellite moves at different orbital positions into a channel state information model to obtain a low-dimensional vector; sending the low-dimensional vector to the satellite so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; wherein the channel state information model is trained based on the high-dimensional channel state information matrix.
[0068] In addition, the logic instructions in the above-mentioned memory 930 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the relevant technology or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.
[0069] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the channel state information feedback method for a multi-user communication system provided by the above-mentioned methods, the method comprising: extracting the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; fusing the peak characteristics and the regional correlation characteristics to obtain a sparse change characteristic matrix of the low-orbit satellite channel that changes with different orbital positions of the satellite movement; inputting the sparse change characteristic matrix of the low-orbit satellite channel that changes with different orbital positions of the satellite movement into a channel state information model to obtain a low-dimensional vector; sending the low-dimensional vector to the satellite so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; wherein the channel state information model is trained based on the high-dimensional channel state information matrix.
[0070] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0071] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiment.
[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for feedback of channel state information in a multi-user communication system, characterized in that: include: Extracting the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; The peak feature and the regional correlation feature are integrated to obtain a sparse variation feature matrix of the low-orbit satellite channel as the satellite moves in different orbital positions; Inputting the sparsity variation characteristic matrix of the low-orbit satellite channel as the satellite moves in different orbital positions into a channel state information model to obtain a low-dimensional vector; Sending the low-dimensional vector to a satellite, so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; Wherein, the channel state information model is obtained by training based on a high-dimensional channel state information matrix.
2. The channel state information feedback method for a multi-user communication system according to claim 1, characterized in that: The channel state information model includes an encoder; The encoder is arranged at the user end, and is used to compress the input sparse matrix in the angle-delay domain, and output the compressed low-dimensional channel state information vector to feed back to the satellite; During the training process of the channel state information model, the sparse matrix in the input angle-delay domain is passed through two convolutional layers and then through a lightweight self-attention feature extraction module to obtain a high-dimensional feature map, wherein the high-dimensional feature map includes feature information of the peak value and information around the peak; The high-dimensional feature map is passed through two convolutional layers again and then dimensionally transformed to obtain a high-dimensional vector; The high-dimensional vector is reduced in dimension using a fully connected layer to obtain a low-dimensional vector.
3. The channel state information feedback method for a multi-user communication system according to claim 1, characterized in that: The channel state information model includes a decoder; The decoder is arranged at the satellite end, and the decoder is used to reconstruct the low-dimensional vector; During the channel state information model training process, the low-dimensional vector is converted into a high-dimensional vector through a fully connected layer; Transforming the dimension of the high-dimensional vector to obtain a high-dimensional feature matrix; Convolving the high-dimensional feature matrix through four convolutional layers respectively, wherein the number of convolution kernels of the four convolutional layers gradually increases, to obtain features of multiple channels; The features of the multiple channels are aggregated and reconstructed into a high-dimensional channel state information matrix.
4. The method for feedback of channel state information of a multi-user communication system according to claim 1, characterized in that: The extracting the peak feature of the channel signal in the angle-delay domain includes: Extracting information in the vertical and horizontal directions of the input matrix corresponding to the channel signal through pooling layers with different kernel sizes; The vertical and horizontal information are merged into a feature map containing both angle domain and delay domain information; Extract the peak features of the feature graph that contains both angle domain and delay domain information.
5. The method for feedback of channel state information of a multi-user communication system according to claim 1, characterized in that: The extracting of regional correlation features of the channel signal in the angle delay domain includes: After the input matrix corresponding to the channel signal is convolved through two convolution layers, the channel signal is processed through two channels, one of which expands the feature matrix after the convolution operation to obtain a first matrix, and the rows of the first matrix are summed to obtain a first column vector; Transforming the dimension of the feature matrix after the convolution operation through another channel to obtain a second column vector and a third column vector, multiplying the second column vector and the third column vector by the first column vector respectively, and concatenating the obtained results in the third dimension to generate a second matrix; The second matrix is reconstructed into a dimension consistent with the size of the input matrix, and the regional correlation characteristics of the channel signal in the angle delay domain are output.
6. The method for feedback of channel state information of a multi-user communication system according to claim 5, characterized in that: The step of expanding the feature matrix after the convolution operation to obtain a first matrix includes: Expanding the periphery of the characteristic matrix by mirroring to add two rows and two columns of elements; A polymerization area of a preset size is set, and it is used to move row by row and column by column on the expanded matrix for raster scanning, and the scanned area is recombined into a row vector with the number of elements being the product of the region height, the region width and the number of channels, and all the scanned deformed row vectors are superimposed to obtain the first matrix.
7. The method for feedback of channel state information of a multi-user communication system according to claim 1, characterized in that: The fusing of the peak feature and the regional correlation feature comprises: The peak feature and the regional correlation feature are fused using a lightweight feature fusion layer, and the new mapping feature output by the feature fusion layer is for: in, is the input matrix corresponding to the channel signal, is the peak feature matrix, is the regional correlation feature matrix, , , are the weight parameters of the input matrix, peak feature matrix and regional correlation feature matrix corresponding to the channel signal, respectively, , , Obtained through model training.
8. A channel state information feedback device for a multi-user communication system, characterized in that: include: An extraction module, used to extract the peak characteristics of the channel signal in the angle-delay domain and the regional correlation characteristics of the channel signal in the angle-delay domain; A fusion module, used to fuse the peak feature and the regional correlation feature to obtain a sparse change feature matrix of the low-orbit satellite channel as the satellite moves in different orbital positions; An input module is used to input the sparsity change characteristic matrix of the low-orbit satellite channel as the satellite moves in different orbital positions into a channel state information model to obtain a low-dimensional vector; A reconstruction module, used for sending the low-dimensional vector to a satellite, so that the satellite reconstructs the low-dimensional vector into a high-dimensional channel state information matrix through the channel state information model; Wherein, the channel state information model is obtained by training based on a high-dimensional channel state information matrix.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the channel state information feedback method for a multi-user communication system according to any one of claims 1 to 7 is implemented.
10. A non-transitory readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the channel state information feedback method for a multi-user communication system as claimed in any one of claims 1 to 7 is implemented.
Citation Information
Patent Citations
Quantization and inverse quantization method in large-scale MIMO channel state information feedback
CN110311718A
Channel state information reconstruction method based on deep learning
CN111464220A
Transform-based large-scale MIMO channel state information feedback method
CN115865145A
Channel state information feedback method suitable for large-scale MIMO system
CN116155333A
Channel state information transmission methods, electronic device, and storage medium
WO2021237423A1
