Channel power spectrum estimation method, device, electronic device and storage medium

Abstract

The channel power spectrum estimation method, device, electronic device and storage medium proposed in the embodiments of the present application include: obtaining the grid position of the grid to be estimated in the target area, and obtaining the cell position and cell antenna parameters of at least one communication cell corresponding to the grid position, wherein the realistic statistical channel model is pre-trained by the reference signal receiving power between multiple grids and the communication cell in the target area; inputting the cell position and cell antenna parameters into the cell feature encoder for feature coding processing to obtain the cell feature code; inputting the grid position and cell position into the grid feature encoder for feature coding processing to obtain the grid feature code; inputting the cell feature code and the grid feature code into the angle power spectrum estimation network for power spectrum estimation to obtain the estimated channel power spectrum between the grid to be estimated and the communication cell, thereby ensuring the accuracy of the channel estimation and improving the channel estimation efficiency in the wireless communication system.

Description

Technical Field

[0001] The present application relates to the field of communication data processing technology, and in particular to a channel power spectrum estimation method, device, electronic device and storage medium. Background Art

[0002] The performance of wireless communication systems is primarily affected by the radio channel. Unlike wired channels, which are typically static and predictable, wireless channels are dynamic and unpredictable, making accurate analysis of wireless communication systems often difficult. Network optimization, for example, can significantly improve user service quality and throughput, but there remains a significant gap between these results and actual application scenarios. A key reason for this is that the complex radio channels in existing networks cannot be accurately characterized and estimated, resulting in ineffective network performance improvements.

[0003] In the related art, channel estimation for complex wireless communication systems is typically performed based on the complex, high-precision environmental information in the wireless communication system. To improve the accuracy of channel estimation, deterministic modeling is then performed to utilize this high-precision environmental information for accurate channel estimation. However, since the environmental information in actual wireless communication systems is often time-varying, the computational complexity of obtaining complex, high-precision environmental information in real time and relying on this real-time, high-precision environmental information for real-time channel estimation is very high, resulting in low channel estimation efficiency in existing channel estimation methods. Summary of the Invention

[0004] The embodiments of the present application provide a channel power spectrum estimation method, apparatus, electronic device, and storage medium, which can improve the channel estimation efficiency in a wireless communication system.

[0005] To achieve the above objectives, a first aspect of an embodiment of the present application proposes a channel power spectrum estimation method, which is performed by a realistic statistical channel model. The realistic statistical channel model includes at least: a cell feature encoder, a grid feature encoder, and an angle power spectrum estimation network. The method includes:

[0006] Obtaining a grid position of a grid to be estimated in a target area, and obtaining a cell position and cell antenna parameters of at least one communication cell corresponding to the grid position, wherein the realistic statistical channel model is pre-trained based on reference signal received powers between a plurality of grids in the target area and the communication cell;

[0007] Inputting the cell location and the cell antenna parameters into the cell feature encoder for feature coding to obtain a cell feature code;

[0008] Inputting the grid position and the cell position into the grid feature encoder for feature coding to obtain a grid feature code;

[0009] The cell feature code and the grid feature code are input into the angle power spectrum estimation network to perform power spectrum estimation, so as to obtain an estimated channel power spectrum between the grid to be estimated and the communication cell.

[0010] In some embodiments, the realistic statistical channel model further includes a converter module, and the step of inputting the cell feature code and the grid feature code into the angle power spectrum estimation network to perform power spectrum estimation to obtain an estimated channel power spectrum between the grid to be estimated and the communication cell includes:

[0011] After performing XOR processing on the cell feature code and the grid feature code, the position embedding of the grid position is superimposed to obtain a feature processing unit;

[0012] Inputting the feature processing unit into the converter module for attention processing to obtain estimated processing data;

[0013] The estimated processed data is input into the angle power spectrum estimation network to perform power spectrum estimation to obtain the estimated channel power spectrum.

[0014] In some embodiments, the converter module includes a multi-head attention module, and inputting the feature processing unit into the converter module for attention processing to obtain estimated processing data includes:

[0015] Acquire other feature processing units between the other grids in the target area and the communication cell, and combine the other feature processing units with the feature processing unit to obtain a feature unit matrix;

[0016] The feature unit matrix is input into the multi-head attention module, multi-head attention processing is performed, and the estimated processing data is obtained.

[0017] In some embodiments, the multi-head attention module includes multiple self-attention processing modules, and inputting the feature unit matrix into the multi-head attention module to perform multi-head attention processing to obtain the estimated processing data includes:

[0018] In the multi-head attention module, a partitioning dimension value of each attention head is obtained based on the ratio of the data dimension of the feature unit matrix to the number of attention heads;

[0019] Generate a projection parameter matrix for each attention head based on the partition dimension value;

[0020] In each of the self-attention processing modules, a matrix query vector, a matrix key vector, and a matrix value vector are generated based on the feature unit matrix, and attention processing is performed on the matrix query vector, the matrix key vector, and the matrix value vector based on the corresponding projection parameter matrix and the linear transformation matrix to obtain sub-estimation processing data;

[0021] Multi-head attention processing is performed on the plurality of sub-estimation processing data to obtain the estimation processing data.

[0022] In some embodiments, inputting the estimated processed data into the angle power spectrum estimation network to perform power spectrum estimation to obtain the estimated channel power spectrum includes:

[0023] In the angle power spectrum estimation network, the estimated processed data is sequentially subjected to linear processing, activation function processing, layer normalization processing, and regularization processing to obtain preliminary processed data;

[0024] Performing residual connection enhancement processing and feature transfer processing on the preliminary processing results in sequence to obtain enhanced processing data;

[0025] Performing non-negative activation processing on the enhanced processed data to obtain activated processed data;

[0026] Estimation processing is performed based on the activation processed data to obtain the estimated channel power spectrum.

[0027] In some embodiments, the step of training the scene-based statistical channel model includes:

[0028] Obtaining a channel impulse response function and a precoding matrix between each antenna in the communication cell and each of the grids;

[0029] Accumulating the products of all the channel impulse response functions and the precoding matrix, squaring the products, and then multiplying the products by the transmit power of the communication cell to obtain a reference signal received power function of each grid;

[0030] Performing time expectation processing on the reference signal received power function to obtain an expected reference signal received power function, wherein the expected reference signal received power function includes a coefficient matrix between the grid and the communication cell;

[0031] Generate a channel power spectrum estimation model based on the coefficient matrix, the reference signal received power function, and the channel power spectrum estimation function between each of the grids and the communication cell;

[0032] The realistic statistical channel model is iteratively updated multiple times based on the channel power spectrum estimation model and the multiple reference signal received powers.

[0033] In some embodiments, the iteratively updating the scene-based statistical channel model multiple times based on the channel power spectrum estimation model includes:

[0034] Obtaining a sparse penalty weight, and obtaining a penalty channel power spectrum function based on a product of the sparse penalty weight and the channel power spectrum estimation function;

[0035] Generating a loss function based on the channel power spectrum estimation model and the accumulated value of the penalty channel power spectrum function;

[0036] The realistic statistical channel model is iteratively updated multiple times based on the loss function and the multiple reference signal received powers.

[0037] To achieve the above objectives, the second aspect of the embodiment of the present application proposes a channel power spectrum estimation device, which is performed by a realistic statistical channel model. The realistic statistical channel model includes at least: a cell feature encoder, a grid feature encoder and an angle power spectrum estimation network. The device includes:

[0038] a data acquisition module, configured to acquire a grid position of a grid to be estimated in a target area, and acquire a cell position and cell antenna parameters of at least one communication cell corresponding to the grid position, wherein the realistic statistical channel model is pre-trained using reference signal received powers between a plurality of grids in the target area and the communication cell;

[0039] a cell feature processing module, configured to input the cell location and the cell antenna parameters into the cell feature encoder for feature coding processing to obtain a cell feature code;

[0040] a grid feature processing module, configured to input the grid position and the cell position into the grid feature encoder for feature coding processing to obtain a grid feature code;

[0041] The channel power spectrum estimation module is used to input the cell feature code and the grid feature code into the angle power spectrum estimation network to perform power spectrum estimation, and obtain the estimated channel power spectrum between the grid to be estimated and the communication cell.

[0042] To achieve the above-mentioned purpose, the third aspect of an embodiment of the present application proposes an electronic device, which includes a memory and a processor, the memory stores a computer program, and the processor implements the channel power spectrum estimation method described in the first aspect when executing the computer program.

[0043] To achieve the above-mentioned purpose, the fourth aspect of an embodiment of the present application proposes a storage medium, which is a computer-readable storage medium and stores a computer program. When the computer program is executed by a processor, it implements the channel power spectrum estimation method described in the first aspect above.

[0044] The channel power spectrum estimation method, device, electronic device and storage medium proposed in the embodiments of the present application are executed by a realistic statistical channel model. The realistic statistical channel model includes at least: a cell feature encoder, a grid feature encoder and an angle power spectrum estimation network. The method includes: first, obtaining the grid position of the grid to be estimated in the target area, and obtaining the cell position and cell antenna parameters of at least one communication cell corresponding to the grid position. The realistic statistical channel model is pre-trained by the reference signal received power between multiple grids and the communication cell in the target area; then, the cell position and cell antenna parameters are input into the cell feature encoder for feature coding processing to obtain the cell feature code; next, the grid position and cell position are input into the grid feature encoder for feature coding processing to obtain the grid feature code; finally, the cell feature code and the grid feature code are input into the angle power spectrum estimation network for power spectrum estimation to obtain the estimated channel power spectrum between the grid to be estimated and the communication cell. The embodiment of the present application adopts a method of pre-utilizing low-cost reference signal received power measurement data between the target area and the communication cell, and uses physical model guidance and deep learning fusion to train to obtain a realistic statistical channel model, thereby avoiding the real-time, complex, high-dimensional environmental information collection and high-computation deterministic modeling required by traditional high-precision methods, thereby reducing the training complexity of the estimation model for channel power spectrum estimation and improving training efficiency. Since the model directly learns from the reference signal received power measurement data that reflects the real environment, it is possible to generate a channel power spectrum estimation result that is closer to reality and more "realistic". In the actual channel power spectrum estimation, the grid position of the estimated grid and the cell position and cell antenna parameters in the communication cell that can be easily obtained in real time are used together with the pre-obtained realistic statistical channel model for efficient feature extraction, thereby avoiding the need for real-time acquisition of complex, time-varying, high-precision environmental information, significantly reducing the computational complexity, and efficiently obtaining the real-time estimated channel power spectrum in the wireless communication system, thereby effectively improving the channel estimation efficiency in the wireless communication system while ensuring the accuracy of the channel estimation.

[0045] Other features and advantages of the present application will be described in the following description, and in part will become apparent from the description, or will be understood by practicing the present application. The purposes and other advantages of the present application can be achieved and obtained through the structures particularly pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flow chart of a channel power spectrum estimation method provided by one embodiment of the present application.

[0047] Figure 2 This is a training flowchart of a scene-based statistical channel model provided by another embodiment of the present application.

[0048] Figure 3 This is a structural diagram of a realistic statistical channel model provided by another embodiment of the present application.

[0049] Figure 4 yes Figure 2 Flowchart of step 205 in FIG.

[0050] Figure 5 This is a structural diagram of a cell feature encoder and a grid feature encoder provided in another embodiment of the present application.

[0051] Figure 6 yes Figure 1 Flowchart of step 104 in FIG.

[0052] Figure 7 yes Figure 6 Flowchart of step 602 in FIG.

[0053] Figure 8 This is a structural diagram of a multi-head attention module provided in another embodiment of the present application.

[0054] Figure 9 yes Figure 7 Flowchart of step 702 in FIG.

[0055] Figure 10 This is a structural diagram of a self-attention processing module provided in another embodiment of the present application.

[0056] Figure 11 yes Figure 6 Flowchart of step 603 in FIG.

[0057] Figure 12 This is a structural diagram of an angle power spectrum estimation network provided in one embodiment of the present application.

[0058] Figure 13 This is a conceptual diagram of a channel power spectrum estimation provided by an embodiment of the present application.

[0059] Figure 14 This is a schematic diagram of a loss reduction curve simulation of a realistic statistical channel model provided by an embodiment of the present application.

[0060] Figure 15 This is a performance simulation diagram of a channel power spectrum estimation method provided in one embodiment of the present application.

[0061] Figure 16 It is a structural diagram of a channel power spectrum estimation device provided in one embodiment of the present application.

[0062] Figure 17 This is a schematic diagram of the hardware structure of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0063] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0064] It should be noted that although the functional modules are divided in the device schematic and the logical order is shown in the flowchart, in some cases, the steps shown or described can be performed in a different order than the module division in the device or the order in the flowchart.

[0065] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application pertains. The terms used herein are for the purpose of describing the embodiments of this application only and are not intended to limit this application.

[0066] The performance of wireless communication systems is primarily affected by the radio channel. Unlike wired channels, which are typically static and predictable, wireless channels are dynamic and unpredictable, making accurate analysis of wireless communication systems often difficult. Network optimization, for example, can significantly improve user service quality and throughput, but there remains a significant gap between these results and actual application scenarios. A key reason for this is that the complex radio channels in existing networks cannot be accurately characterized and estimated, resulting in ineffective network performance improvements.

[0067] In the related art, channel estimation for complex wireless communication systems is typically performed based on the complex, high-precision environmental information in the wireless communication system. To improve the accuracy of channel estimation, deterministic modeling is then performed to utilize this high-precision environmental information for accurate channel estimation. However, since the environmental information in actual wireless communication systems is often time-varying, the computational complexity of obtaining complex, high-precision environmental information in real time and relying on this real-time, high-precision environmental information for real-time channel estimation is very high, resulting in low channel estimation efficiency in existing channel estimation methods.

[0068] In order to improve the channel estimation efficiency in wireless communication systems, the embodiments of the present application adopt a method of pre-using low-cost reference signal received power measurement data between the target area and the communication cell, and train it using a physical model guidance and deep learning fusion method to obtain a realistic statistical channel model, thereby avoiding the real-time, complex, high-dimensional environmental information collection and high-computation deterministic modeling required by traditional high-precision methods, thereby reducing the training complexity of the estimation model for channel power spectrum estimation and improving training efficiency. Because the model directly learns from the reference signal received power measurement data that reflects the real environment, it can generate a channel power spectrum estimation result that is closer to reality and more "realistic". In the actual channel power spectrum estimation, the grid position of the estimated grid, the cell position and cell antenna parameters in the communication cell, which can be easily obtained in real time, and the pre-obtained realistic statistical channel model are used for efficient feature extraction, thereby avoiding the need for real-time acquisition of complex, time-varying, high-precision environmental information, significantly reducing computational complexity, and efficiently obtaining a real-time estimated channel power spectrum in the wireless communication system, thereby effectively improving the channel estimation efficiency in the wireless communication system while ensuring the accuracy of the channel estimation.

[0069] The following describes the channel power spectrum estimation method, apparatus, electronic device, and storage medium provided by the embodiments of the present application. The channel power spectrum estimation method provided in the embodiments of the present application can be applied to any server or computing processor with computing resources.

[0070] The channel power spectrum estimation method in the embodiment of the present application will be described in detail below. Figure 1 , which is an optional flow chart of the channel power spectrum estimation method provided in an embodiment of the present application, Figure 1 The method may include but is not limited to steps 101 to 104. It is also understood that this embodiment is for Figure 1 The order of step 101 to step 104 is not specifically limited, and the order of steps can be adjusted or some steps can be reduced or added according to actual needs.

[0071] Step 101: Obtain a grid position of a grid to be estimated in a target area, and obtain a cell position and cell antenna parameters of at least one communication cell corresponding to the grid position.

[0072] Step 101 is described in detail below.

[0073] In some embodiments, in a wireless communication system including a target area and at least one communication cell, the target area includes multiple grids, and the channel from each communication cell to each grid can be characterized by large-scale fading (path loss, shadow fading) and small-scale fading. The antenna array of each communication cell includes antennas.

[0074] The target area can be a hotspot area of a city or a specific coverage area. Then, within the target area, one or more grids are selected as grids to be estimated (i.e., any one or more grids in the target area, such as The grid position can be expressed in latitude and longitude coordinates, for example (39.9042°N,116.4074°E).

[0075] Next, the wireless communication system determines the communication cells associated with the grid to be estimated, or the communication cells desired by the user. This can be obtained through queries in a geographic location database or network planning information. For each associated communication cell, its location (also expressed in latitude and longitude coordinates) and cell antenna parameters must be obtained. These parameters include antenna gain, antenna angle, and frequency. These parameters are typically stored in the wireless communication system's network management equipment and can be retrieved through interface calls or data exports. This data is easily accessible without the need for complex acquisition equipment or data processing methods.

[0076] After obtaining the grid position of the grid to be estimated, the cell position of the communication cell and the cell antenna parameters, the channel power spectrum is estimated using a realistic statistical channel model pre-trained by the reference signal receiving power between multiple grids in the target area and the communication cell, so as to quickly obtain the estimated channel power spectrum between the grid to be estimated and the communication cell, thereby effectively improving the estimation efficiency of the channel estimation.

[0077] The following first describes how to train and obtain the scene-based statistical channel model.

[0078] Reference Figure 2 The training steps of the realistic statistical channel model include the following steps 201 to 205.

[0079] Step 201: Obtain a channel impulse response function and a precoding matrix between each antenna and each grid in a communication cell.

[0080] Step 202: Accumulate the products of all channel impulse response functions and precoding matrices, square them, and then multiply them by the transmit power of the communication cell to obtain the reference signal received power function of each grid.

[0081] Step 203: Perform time expectation processing on the reference signal received power function to obtain an expected reference signal received power function.

[0082] Step 204: Generate a channel power spectrum estimation model based on the coefficient matrix, the reference signal received power function, and the channel power spectrum estimation function between each grid and the communication cell.

[0083] Step 205: performing multiple iterations to update the scene-based statistical channel model based on the channel power spectrum estimation model and the received powers of multiple reference signals.

[0084] Steps 201 to 205 are described in detail below.

[0085] In some embodiments, with respect to the wireless communication system mentioned above, Each antenna in a communication cell (the antenna position is ) and the first Between the grids The channel impulse response function at time t is shown in the following formula (1).

[0086] (1)

[0087] in, and Corresponding to the total number of divisions of the vertical and horizontal planes in the angle domain, It corresponds to An antenna in a communication cell (the antenna position is ) of the antenna gain, For the The elevation incident angle of each communication cell, For the The azimuth incident angle of each communication cell, and is the spacing between adjacent antennas in the corresponding direction, The phase error between angles is in the interval [ ] obeys uniform distribution, The phase error between antennas has a mean of 0 and a variance of Gaussian distribution, is the wavelength of the electromagnetic wave, For the An antenna in a communication cell (the antenna position is ) and the first Between the grids The channel loss at time t, which obeys the log-normal distribution as shown in the following formula (2).

[0088] (2)

[0089] in, and Respectively An antenna in a communication cell (the antenna position is ) and the first The mean and standard deviation of the log-normal distribution of path loss between grid cells.

[0090] Since the dimension of the channel impulse response is very large in a large-scale antenna, in order to reduce the complexity, the reference signal received power rsrp of multiple beams is considered in this embodiment. The first antenna array in the communication cell The precoding matrix corresponding to the beam is shown in the following formula (3).

[0091] (3)

[0092] in, For the In the communication cell After that, the product of all channel impulse response functions (1) and precoding matrix (3) is accumulated, squared, and then multiplied by the transmission power of the communication cell. , get the reference signal receiving power function of each grid, as shown in the first In the communication cell beam and the The reference signal receiving power function between grids is shown in the following formula (4).

[0093] (4)

[0094] Based on formula (1), the reference signal received power function is further expanded As shown in the following formula (5).

[0095] (5)

[0096] Next, the reference signal received power function (5) is subjected to time expectation processing using the time parameter t to obtain the expected reference signal received power function as shown in the following formula (6).

[0097] (6)

[0098] The expected reference signal received power function includes the coefficient matrix between the grid and the communication cell: , as shown in the following formula (7).

[0099] (7)

[0100] (8)

[0101] Thus, in this embodiment, a statistical relationship between the reference signal received power and the angular power spectrum is established. For a single cell and a single grid, this relationship can be expressed as the following formula (9).

[0102] , (9)

[0103] Among them, the coefficient matrix There are main lobes and side lobes, and the path between the antenna array and the grid can only exist at the angles corresponding to the main lobes and side lobes. The angular power spectrum that needs to be estimated is It is a sparse vector, and the non-zero elements represent the power and corresponding angle of the multipath channel.

[0104] The goal of scene-based statistical channel modeling is to collect the measured data of the received power of the reference signal and calculate the statistical relationship between it and the channel angle power spectrum. To estimate . It characterizes the statistical distribution of channel power in the angular domain, reflects the multi-path topology of the realistic environment, and can be used for performance prediction after network parameter optimization.

[0105] Since the number of communication cells in the actual network optimization corresponding to the wireless communication system is large and the coupling relationship is complex, it is necessary to consider a wide range of channel modeling. Specifically, the goal in the embodiment of the present application is to perform large-scale, realistic statistical channel modeling for multiple communication cells and multiple grids. Therefore, the channel angle power spectrum estimation is represented by Parameterized functions ,but Indicates the The communication cell is in The channel angle power spectrum of the grid; then based on the coefficient matrix , reference signal received power function (5) and channel power spectrum estimation function between each grid and the communication cell , the channel power spectrum estimation model is generated as shown in the following formula (10).

[0106] (10)

[0107] in, is the total number of grids in the target area, P is the maximum number of paths in the multipath channel, and the channel power spectrum is non-negative.

[0108] For the sparse recovery problem of the channel power spectrum estimation model (10), traditional optimization algorithms use greedy or iterative methods to solve it, which cannot effectively utilize environmental information. In this embodiment, with the powerful learning ability of deep neural networks, a transformer-based statistical channel model is designed to learn from the data and establish an implicit mapping from environmental information to the complex nonlinear channel angle power spectrum.

[0109] Reference Figure 3 , is a structural diagram of a realistic statistical channel model provided by an embodiment of the present application. Figure 3 As shown in , the scene-based statistical channel model consists of a cell feature encoder, a grid feature encoder, a transformer, and an angular power spectrum estimation network. The cell feature encoder receives the cell location and cell antenna parameters as input, while the grid feature encoder receives the grid location and cell location as input, and performs feature encoding respectively. The encoded features are added to the geographic location embedding and then input into the transformer module for attention processing. The output of the transformer module is finally input into the angular power spectrum estimation network, which contains structures such as layer normalization, a feedforward network, a multi-head self-attention mechanism, and residual connections, and ultimately outputs the estimated channel power spectrum. Here, "+" represents feature addition, and "C" represents feature concatenation.

[0110] After constructing the realistic statistical channel model, the realistic statistical channel model is iteratively updated multiple times based on the channel power spectrum estimation model (10) and multiple measured reference signal receiving powers between multiple grids and communication cells in the target area, as described below.

[0111] Reference Figure 4 , performing multiple iterative updates on the realistic statistical channel model based on the channel power spectrum estimation model and multiple reference signal received powers, including the following steps 401 to 403.

[0112] Step 401: Obtain a sparse penalty weight, and obtain a penalty channel power spectrum function based on the product of the sparse penalty weight and the channel power spectrum estimation function.

[0113] Step 402: Generate a loss function based on the channel power spectrum estimation model and the accumulated value of the penalty channel power spectrum function.

[0114] Step 403: performing multiple iterations to update the scene-based statistical channel model based on the loss function and the received powers of multiple reference signals.

[0115] Steps 401 to 403 are described in detail below.

[0116] In some embodiments, since the propagation environments of different communication cells are different, when constructing a multi-grid sequence, each grid is paired with the same communication cell. In order to effectively train the entire scene-based statistical channel model, this embodiment designs a model training framework based on self-supervised learning.

[0117] Based on pre-set sparse penalty weights The product of the norm of the channel power spectrum estimation function and the penalty channel power spectrum function is obtained. , then based on the accumulated value of the objective function and the penalty channel power spectrum function in the channel power spectrum estimation model (10), the loss function is generated as shown in the following formula (11).

[0118] (11)

[0119] The loss function includes reconstruction loss and sparse penalty, and the reconstruction target is the maximum value of the multi-beam reference signal received power.

[0120] Afterwards, the realistic statistical channel model is iteratively updated multiple times based on the loss function and the reference signal received powers between multiple grids and communication cells in the target area obtained by actual measurement. Figure 3 In the illustrated realistic statistical channel model, the module parameters updated in each iteration include relevant parameters in the cell feature encoder, grid feature encoder, transformer, and angle power spectrum estimation network.

[0121] Once the training is completed, the model can estimate the channel angle power spectrum of the grid of the specified cell based on the given input, and then combine formula (6) to predict the maximum value of the reference signal received power after the network parameters are adjusted in the wireless communication system, thereby accurately estimating the channel of the wireless communication system.

[0122] Through the above steps 201 to 205, and steps 401 to 403, starting from the basic physical quantities such as channel impulse response, precoding and transmit power, the expected reference signal received power and the underlying channel statistical characteristics (ultimately reflected in the angle power spectrum and the coefficient matrix) are derived. This provides a solid physical foundation and interpretability basis for subsequent model training, avoiding the drawback of the pure black box model that lacks physical meaning. This physical relationship is used to construct the learning target, and the deep learning model is iteratively trained using the actual measured and easily accessible reference signal received power data, so that the model can learn specific "real" The complex propagation characteristics in the "scene" environment overcome the problems of poor generalization ability of traditional statistical models or the dependence of deterministic models on high-precision environmental information and large computational complexity, realizing the "realistic" and efficient model. By introducing a sparse penalty term in the loss function, the prior knowledge that the angular power spectrum of the wireless channel is usually sparse is utilized to guide the model learning to obtain a power spectrum representation that is more in line with physical reality, more concise and effective. This not only helps to improve the generalization ability and estimation accuracy of the model, but also improves the robustness of the model to a certain extent, thus obtaining an efficient, accurate and practical channel power spectrum estimation model that integrates physical principles, measured data and prior knowledge.

[0123] Since in the embodiment of the present application, the data processing flow during the training of the constructed realistic statistical channel model is consistent with the data processing flow for actual channel estimation using the trained realistic statistical channel model, in order to avoid repeated redundant descriptions in this embodiment, the data processing flow for actual channel estimation is described in detail below as an example.

[0124] Step 102: Input the cell location and cell antenna parameters into a cell feature encoder for feature coding processing to obtain a cell feature code.

[0125] Step 103: Input the grid position and the cell position into a grid feature encoder for feature encoding processing to obtain a grid feature code.

[0126] Steps 102 to 103 are described in detail below.

[0127] In such Figure 3 In the realistic statistical channel model shown, in order to extract features of different environmental information, encoders are designed from two dimensions: grid and communication cell.

[0128] The grid feature encoder is used to extract the geographic information of the grid in the propagation environment. Considering that the propagation environment satisfies translation and rotation invariance in geographic space, the relative coordinates of the grid are calculated based on the absolute coordinates of the grid and the communication cell and used as the position feature, as shown in the following formula (12).

[0129] (12)

[0130] in, and For the grid and The absolute coordinates of each communication cell.

[0131] In wireless signal propagation, the effect of propagation distance on signal attenuation is the most direct and obvious. Further, the propagation distance and relative angle between the grid and the cell are calculated and used as physical features to enhance the representation ability of the features. grid and The mechanical horizontal angle, mechanical downtilt angle and frequency between the communication cells are shown in the following formulas (13), (14) and (15) respectively.

[0132] (13)

[0133] (14)

[0134] (15)

[0135] Finally, in the grid feature encoder, the above features are concatenated into the input features of the grid feature encoder, as shown in the following formula (16).

[0136] (16)

[0137] As can be seen from formula (12), For the grid and The x-axis coordinate difference between the communication cells, For the grid and The y-axis coordinate difference between the communication cells, For the grid and The z-axis coordinate difference between the communication cells.

[0138] In addition, the cell feature encoder is used to extract the common parameter information of the cell. From formula (4), we can see that the propagation of the signal is determined by the coefficient matrix and channel angle power spectrum The propagation environment affects the channel angle power spectrum, while the cell antenna parameters affect the coefficient matrix. Therefore, the cell location and antenna parameters are used as input features of the cell feature encoder, as shown in the following formula (17).

[0139] (17)

[0140] In some embodiments, in practical applications, after obtaining the grid position of the grid to be estimated and the cell position and cell antenna parameters of the communication cell, the input features shown in the above formula (16) are respectively used. Input the raster encoder for feature extraction to obtain raster feature encoding , and using the input features shown in formula (17) Input the cell encoder to extract features and obtain the cell feature code .

[0141] Reference Figure 5 , is a schematic diagram of the structure of a cell feature encoder and a grid feature encoder provided in an embodiment of the present application. Figure 5 As shown in the figure, the structure and process of the feature encoder in the realistic statistical channel model are shown. Specifically, it contains two parallel encoding processing paths: the grid feature encoder on the left receives the grid feature input related to a specific communication cell q. It may contain relative coordinates, distance, angle, etc.), and passes through a fully connected layer, a rectified linear unit (ReLU) activation function, a layer normalization layer, a second fully connected layer, and a second layer normalization layer in sequence for deep feature extraction; on the right is the cell feature encoder, which receives the feature input of the communication cell q itself The grid feature encoding and the cell feature encoding are concatenated at the operation point marked "C" to combine the two different but related feature information into a richer feature vector, which serves as input for subsequent processing.

[0142] Step 104: Input the cell feature code and the grid feature code into the angle power spectrum estimation network to perform power spectrum estimation, and obtain the estimated channel power spectrum between the grid to be estimated and the communication cell.

[0143] Step 104 is described in detail below.

[0144] In some embodiments, after obtaining the grid feature code and cell feature coding After that, we will further use raster feature encoding and cell feature coding And the angle power spectrum estimation network performs power spectrum estimation to obtain the real-time estimated channel power spectrum between the grid to be estimated and the communication cell , as described below.

[0145] Reference Figure 6 , inputting the cell feature code and the grid feature code into the angle power spectrum estimation network for power spectrum estimation, and obtaining the estimated channel power spectrum between the grid to be estimated and the communication cell, including the following steps 601 to 603.

[0146] Step 601: After performing XOR processing on the cell feature code and the grid feature code, the position embedding of the grid position is superimposed to obtain a feature processing unit.

[0147] Step 601 is described in detail below.

[0148] In some embodiments, the Transformer uses global information parallel processing to enable the scene-based statistical channel model to efficiently process long input sequences. However, in natural language processing, the position of different tokens in the sequence is crucial. Therefore, positional encoding is often used to identify the absolute or relative position of tokens in the sequence.

[0149] In the channel estimation task corresponding to the embodiments of this application, multi-grid sequence modeling is employed, with each token corresponding to a grid cell pair. To characterize the spatial correlations between different grids, a frequency-based sine-cosine geospatial location embedding method is introduced. This method does not add additional training parameters while enhancing the model's understanding of spatial relationships.

[0150] Therefore, the position of the grid position is embedded As shown in the following formula (18).

[0151] (18)

[0152] in, ,Right now Corresponding to the position embedding of the x-axis, y-axis and z-axis respectively. The dimension of the input token.

[0153] Therefore, the cell characteristics are encoded , raster feature encoding After XOR processing, the grid position is superimposed and embedded , get the feature processing unit As shown in the following formula (19).

[0154] (19)

[0155] Step 602: Input the feature processing unit into the converter module for attention processing to obtain estimated processing data.

[0156] Step 602 is described in detail below.

[0157] Next, in order to enable the model to capture the relationship between each grid in the multi-grid sequence (the consistency of wireless channels in geographic space), whether these relationships are local or global, the self-attention mechanism is used to enable the model to dynamically pay attention to the information of other positions in the sequence when inputting each grid in the multi-grid sequence, thereby effectively combining multiple grids for structured feature extraction. Therefore, the feature processing unit is further Input transformer module to perform attention processing and obtain estimated processing data , as described below.

[0158] Reference Figure 7 , the feature processing unit is input into the converter module for attention processing to obtain estimated processing data, including the following steps 701 to 702.

[0159] Step 701: Acquire other feature processing units between other grids and communication cells in the target area, and combine other feature processing units with feature processing units to obtain a feature unit matrix.

[0160] Step 702: Input the feature unit matrix into the multi-head attention module, perform multi-head attention processing, and obtain estimated processing data.

[0161] Steps 701 to 702 are described in detail below.

[0162] In some embodiments, in the converter module, first, based on other feature processing units between other grids in the target area and the communication cell, and combining other feature processing units and feature processing units, a feature unit matrix input is obtained as a token matrix of a multi-grid sequence, that is, a feature unit matrix ( ), by using the linear transformation matrix ( ), ( )and ( ), and then we get the matrix (Query query), (Key, key) and (Value, value) to facilitate attention processing.

[0163] In addition, multi-head self-attention is an extension of self-attention. By introducing multiple "heads," the model can perform self-attention calculations in parallel in different subspaces while keeping the total number of parameters unchanged, thereby capturing more diverse features and relationships and, in a sense, suppressing overfitting. Therefore, a multi-head attention module is set up in the transformer module, and multiple self-attention processing modules are set up in the multi-head attention module.

[0164] Reference Figure 8 , is a schematic diagram of the structure of a multi-head attention module provided in an embodiment of the present application. Figure 8 As shown in the figure, the structure and process of the multi-head attention module are shown in detail, which is the core component of the Transformer module. The process first performs h independent linear transformations on the input query (Q), key (K), and value (V) vectors (these vectors are usually representations of the previous layer or input embedding) (the "linear transformation" layer at the bottom of the figure has h sets of parallel transformations), generating respective Q, K, and V representations for h different self-attention processing modules; then, within each self-attention processing module, a self-attention calculation (usually scaled dot product attention) is performed using its own Q, K, and V, so that each position can pay attention to the information of all positions in the input sequence and calculate a weighted representation; then, the output vectors obtained by the parallel calculations of the h self-attention processing modules are concatenated (concatenation), combining the different subspace information or dependencies captured by different attention heads; finally, the concatenated vectors are passed through a final linear transformation layer for integration and dimension adjustment to obtain the final output Z of the multi-head self-attention module. This output fuses information from multiple different representation subspaces and can more comprehensively capture the spatial channel feature dependencies between different grid positions in the input sequence.

[0165] Therefore, in the obtained characteristic unit matrix ( ) and then input it into the multi-head attention module for multi-head attention processing to obtain the estimated processing data , as described below.

[0166] Reference Figure 9 , input the feature unit matrix into the multi-head attention module, perform multi-head attention processing, and obtain estimated processing data, including the following steps 901 to 904.

[0167] Step 901: In the multi-head attention module, the partitioning dimension value of each attention head is obtained based on the ratio of the data dimension of the feature unit matrix and the number of attention heads.

[0168] Step 902: Generate a projection parameter matrix for each attention head based on the partition dimension value.

[0169] Step 903: In each self-attention processing module, a matrix query vector, a matrix key vector, and a matrix value vector are generated based on the feature unit matrix, and attention processing is performed on the matrix query vector, the matrix key vector, and the matrix value vector based on the corresponding projection parameter matrix and the linear transformation matrix to obtain sub-estimation processing data.

[0170] Step 904: Perform multi-head attention processing on the multiple sub-estimation processing data to obtain estimated processing data.

[0171] Steps 901 to 904 are described in detail below.

[0172] In some embodiments, Figure 8 In the multi-head attention module shown, firstly based on the data dimension of the feature unit matrix and the number of attention heads The ratio of , to get the division dimension value of each attention head , then, based on the partition dimension value, the projection parameter matrix of each attention head is generated, including , , as well as , is the number of self-attention processing modules in the multi-head attention module. Based on this, the multi-head attention processing flow in the multi-head attention module is shown in the following formula (20).

[0173] (20)

[0174] Next, in the multi-head attention module, the feature unit matrix is processed together based on multiple self-attention processing modules. Perform attention processing. Figure 10 , is a structural diagram of a self-attention processing module provided in an embodiment of the present application. Figure 10 As shown in the figure, the internal structure and data processing flow of the scaled dot-product attention mechanism are illustrated in detail. This is the core computational unit of each "head" in the multi-head attention. The process receives three inputs: the query matrix (Q), the key matrix (K), and the value matrix (V). First, the query matrix Q and the key matrix K are "matrix multiplied" (usually by calculating the dot product of the transpose of Q and K) to obtain the raw attention score, which measures the similarity or association between each query and each key. Then, these raw scores go through a "scaling" step, usually by dividing by the key vector dimension. The square root of is used to prevent gradients from being too small or too large, stabilizing the training process. The scaled attention scores are then fed into a normalization layer, typically by applying a Softmax function to convert them into probability distributions (i.e., attention weights), such that the sum of the weights of all keys corresponding to each query is 1. Finally, these normalized attention weights are matrix-multiplied with the value matrix V to calculate the weighted sum of the values. This final output, Z, is the result of weighted aggregation of value information based on the similarity between the query and the key, representing the context-aware feature representation processed by the attention mechanism.

[0175] like Figure 10 As shown in , in each self-attention processing module, a matrix query vector, a matrix key vector, and a matrix value vector are generated based on the feature unit matrix, and attention processing is performed on the matrix query vector, the matrix key vector, and the matrix value vector based on the corresponding projection parameter matrix and the linear transformation matrix to obtain sub-estimation processing data. That is, the input is the feature unit matrix of the multi-grid sequence ( ), by using the linear transformation matrix ( ), ( )and ( ), and then we get the matrix (Query query), (Key, key) and (Value, value). , , The output value is obtained by the following formula (21), that is, the sub-estimation processing data .

[0176] (twenty one)

[0177] in, is the scaling factor, which acts as a normalization factor. and Determine the similarity The weight distribution method is called dot product attention. Self-attention adaptively calculates weights in parallel and can effectively capture the relationship between any two grids in the sequence.

[0178] Afterwards, based on the multi-head attention processing flow shown in the above formula (20), multi-head attention processing is performed on multiple sub-estimation processing data to obtain estimated processing data.

[0179] Through steps 701 to 702 and steps 901 to 904, the model combines the feature cells of multiple grids into a feature cell matrix and then applies a multi-head self-attention mechanism. This allows the model to focus on and weightedly aggregate information from all other grids (including itself) through the feature representation of each grid (as the query Q). This effectively learns complex spatial patterns, such as how signals propagate in space, the spatial continuity of shadow effects, and the mutual influence of interference between different locations. The multi-head mechanism further allows the model to simultaneously capture spatially correlated features of multiple types or scales from different representation subspaces. This deep modeling of spatial context overcomes the limitations of traditional methods or simple models that may treat each grid as an isolated point or only consider the influence of local neighbors. This enables the final output of estimated processed data to more comprehensively and realistically reflect the spatial heterogeneity and interactions of large-scale wireless environments, significantly improving the accuracy and "realistic" nature of subsequent channel power spectrum estimation, and providing higher-quality input for applications such as network optimization that rely on accurate channel information.

[0180] Step 603: Input the estimated processed data into the angle power spectrum estimation network to perform power spectrum estimation to obtain an estimated channel power spectrum.

[0181] Step 603 is described in detail below.

[0182] In some embodiments, based on Figure 3 In the processing flow of the realistic statistical channel model shown, after obtaining the estimated processing data output by the multi-head attention module, the estimated processing data is further input into the angle power spectrum estimation network for power spectrum estimation to obtain the estimated channel power spectrum, as described below.

[0183] Reference Figure 11 , inputting the estimated processed data into the angle power spectrum estimation network to perform power spectrum estimation to obtain the estimated channel power spectrum, including the following steps 1101 to 1104.

[0184] Step 1101: In the angle power spectrum estimation network, the estimated processed data is sequentially subjected to linear processing, activation function processing, layer normalization processing, and regularization processing to obtain preliminary processed data.

[0185] Step 1102: The preliminary processing results are sequentially subjected to residual connection enhancement processing and feature transfer processing to obtain enhanced processing data.

[0186] Step 1103: Perform non-negative activation processing on the enhanced processed data to obtain activated processed data.

[0187] Step 1104: Perform estimation processing based on the activated processed data to obtain an estimated channel power spectrum.

[0188] Steps 1101 to 1104 are described in detail below.

[0189] In some embodiments, considering the non-negative, sparse and high-dimensional characteristics of the angular power spectrum, an angular power spectrum estimation network is designed. Figure 12 , is a schematic diagram of the structure of an angle power spectrum estimation network provided by an embodiment of the present application. Figure 12 As shown in , progressive dimension expansion is adopted. The basic components of each layer include linear layer, GELU activation function, layer normalization and Dropout. Residual connections are used in the second and third layers to enhance gradient flow and feature transfer. Finally, ReLU activation function is used in the output layer to ensure non-negative output while promoting sparsity.

[0190] Therefore, based on Figure 12 The angle power spectrum estimation network shown first performs linear processing, activation function processing, layer normalization processing and regularization processing on the estimated processed data in sequence to obtain preliminary processed data, so as to introduce nonlinearity while accelerating training and improving the generalization ability of the model, and thereby reducing the dependency between neurons; then, the preliminary processed results are sequentially subjected to residual connection enhancement processing and feature transfer processing to obtain enhanced processed data, so that the model can alleviate the gradient vanishing problem, make it easier for the model to learn the identity mapping, and enable the model to utilize earlier feature information; next, the enhanced processed data is subjected to non-negative activation processing to obtain activation processed data, so that the model can ensure the rationality of the output; finally, estimation processing is performed based on the activation processed data to obtain an estimated channel power spectrum, and the dimension of the output depends on the representation of the channel power spectrum. For example, if the channel power spectrum is represented as a vector, the dimension of the output is the length of the vector.

[0191] Through the above steps 1101 to 1104, the input features are initially transformed and refined through linear transformation, nonlinear activation (such as GELU), layer normalization, and regularization (such as Dropout), while ensuring the stability of training and the generalization ability of the model. The use of residual connections and further feature transfer processing allows the construction of deeper network structures to capture finer feature details, effectively solving the gradient vanishing problem in deep network training, thereby significantly improving the model's ability to learn and infer accurate power spectrum details from complex input features. In addition, the application of non-negative activation functions (such as ReLU) to the results before the final output ensures that the estimated power spectrum values are all non-negative, which directly reflects the constraints of the physical model, making the output results more consistent with physical reality and improving the reliability and credibility of the estimation results. The final estimated channel power spectrum is then directly obtained based on this deeply processed and physically constrained activation data, providing an end-to-end, accurate mapping from high-level features to specific power spectrum distribution, providing high-quality, physically meaningful channel characteristic description for subsequent network optimization applications.

[0192] Reference Figure 13 , is a conceptual diagram of a channel power spectrum estimation provided by an embodiment of the present application. Figure 13 As shown in the figure, by implementing the channel power spectrum estimation method provided by this application, the scene-based statistical channel model receives two types of input information: the first type is parameters, including key information describing the physical scene configuration, such as the geographic coordinates of multiple grids within the target area (multi-grid coordinates), the geographic coordinates of multiple communication cells participating in the communication, and the specific antenna configuration parameters of these cells, such as antenna gain, downtilt angle, azimuth angle (simplified as "and antenna gain angle" in the figure), and operating frequency. The second type is data, specifically the reference signal received power (RSRP) measured in the actual environment, which reflects the impact of the actual wireless propagation environment on the signal. The "+" sign in the figure indicates that these two types of input are combined during model training or operation. This input information is fed into the core "large-scale scene-based statistical channel model" for processing. This model aims to learn the complex mapping relationship between parameter configuration, real RSRP data, and the underlying channel statistical characteristics.

[0193] At the output end, the real-world statistical channel model mainly generates "the statistical characteristics of the channel angle power spectrum of multiple cells and multiple grids." This means that the model can predict the angle power spectrum (APS) of the channel between any specified grid and the relevant cells in the target area, that is, the distribution of signal energy at different arrival angles, which is a key channel statistical characteristic. In addition, the figure also mentions that the output contains information related to the "large-scale environment multipath topology." This means that the real-world statistical channel model can also implicitly or explicitly learn and reflect the characteristics of the multipath propagation structure in a large-scale environment (that is, after obtaining the channel power spectrum, combined with formula (6) to predict the maximum value of the reference signal received power after the network parameters are adjusted in the wireless communication system, thereby accurately estimating the channel of the wireless communication system). In other words, accurate APS output itself contains an understanding of the multipath topology.

[0194] In order to improve the reliability of a channel power spectrum estimation method provided in an embodiment of the present application, this embodiment simulates and verifies the channel power spectrum estimation. The measured data set covers 10,012 10m x 10m grids in 380 communication cells in a certain area, including eight rounds of data. Each round of data switches the codebooks of several cells. We use rounds 1, 3, 5, and 7 as training sets, and rounds 2, 4, 6, and 8 as test sets. The multi-grid sequence length is set to 20, and the number of model parameters is 6,975,736 (6.98M). We measure the model prediction performance with the mean absolute error.

[0195] Reference Figure 14 , is a schematic diagram of a loss reduction curve simulation of a realistic statistical channel model provided in an embodiment of the present application. Figure 14 The figure shows the loss reduction curve after 50 rounds of model training (approximately 5 minutes), where yellow represents the training loss. As can be seen, the training loss of the real-world statistical channel model provided by this application decreases rapidly and smoothly, and the test loss also decreases accordingly, demonstrating the effectiveness of our model design.

[0196] In addition, in this embodiment, the performance of the channel power spectrum estimation method proposed in this application is compared with other baseline algorithms, including the iterative threshold soft threshold algorithm (ISTA), the non-negative orthogonal matching pursuit algorithm (NNOMP) and the weighted non-negative orthogonal matching pursuit algorithm (WNNOMP).

[0197] Reference Figure 15 , is a performance simulation diagram of a channel power spectrum estimation method provided by an embodiment of the present application. Figure 15As shown in , the channel power spectrum estimation method proposed in this application outperforms all baseline algorithms while ensuring training efficiency. It should be noted that the real-scene statistical channel model proposed in this application can predict grids not included in the training set, while other optimization algorithms can only make predictions based on existing grids. The generalization ability of the real-scene statistical channel model proposed in this application is also further improved.

[0198] The channel power spectrum estimation method, device, electronic device and storage medium proposed in the embodiments of the present application are executed by a real-life statistical channel model. The real-life statistical channel model includes at least: a cell feature encoder, a grid feature encoder and an angle power spectrum estimation network. The method includes: first, obtaining the grid position of the grid to be estimated in the target area, and obtaining the cell position and cell antenna parameters of at least one communication cell corresponding to the grid position. The real-life statistical channel model is pre-trained by the reference signal receiving power between multiple grids and the communication cell in the target area. The training step of the real-life statistical channel model includes: obtaining the channel impulse response function and precoding matrix between each antenna and each grid in the communication cell, accumulating the product of all channel impulse response functions and precoding matrices, squaring them, and then multiplying them by the transmission power of the communication cell to obtain the reference signal receiving power function of each grid, performing time expectation processing on the reference signal receiving power function, and obtaining An expected reference signal receiving power function, the expected reference signal receiving power function includes a coefficient matrix between the grid and the communication cell, a channel power spectrum estimation function between each grid and the communication cell is generated based on the coefficient matrix, the reference signal receiving power function and the channel power spectrum estimation function, a channel power spectrum estimation model is generated, a sparse penalty weight is obtained for the realistic statistical channel model based on the channel power spectrum estimation model and multiple reference signal receiving powers, a penalty channel power spectrum function is obtained based on the product of the sparse penalty weight and the channel power spectrum estimation function, a loss function is generated based on the cumulative value of the channel power spectrum estimation model and the penalty channel power spectrum function, and the realistic statistical channel model is iteratively updated multiple times based on the loss function and multiple reference signal receiving powers; then, the cell position and cell antenna parameters are input into the cell feature encoder for feature coding processing to obtain the cell feature code; next, the grid position and cell position are input into the grid feature encoder for feature coding processing to obtain the grid feature code;Finally, the cell feature code and the grid feature code are XORed and then the grid position embedding is superimposed to obtain the feature processing unit. Other feature processing units between other grids in the target area and the communication cell are obtained, and other feature processing units and feature processing units are combined to obtain the feature unit matrix. In the multi-head attention module, the partition dimension value of each attention head is obtained based on the ratio of the data dimension of the feature unit matrix and the number of attention heads. The projection parameter matrix of each attention head is generated based on the partition dimension value. In each self-attention processing module, the matrix query vector, matrix key vector and matrix value vector are generated based on the feature unit matrix, and based on the corresponding The projection parameter matrix and linear transformation matrix perform attention processing on the matrix query vector, the matrix key vector, and the matrix value vector to obtain sub-estimated processed data. Multi-head attention processing is performed on multiple sub-estimated processed data to obtain estimated processed data. In the angle power spectrum estimation network, the estimated processed data is sequentially subjected to linear processing, activation function processing, layer normalization processing, and regularization processing to obtain preliminary processed data. The preliminary processed results are sequentially subjected to residual connection enhancement processing and feature transfer processing to obtain enhanced processed data. The enhanced processed data is subjected to non-negative activation processing to obtain activated processed data. Estimation processing is performed based on the activated processed data to obtain an estimated channel power spectrum.

[0199] The embodiment of the present application adopts a low-cost reference signal receiving power measurement data between the target area and the communication cell, and adopts a physical model guidance and deep learning fusion method to train to obtain a realistic statistical channel model, thereby avoiding the real-time, complex, high-dimensional environmental information collection and high-computation deterministic modeling required by traditional high-precision methods, so as to reduce the training complexity of the estimation model for channel power spectrum estimation and improve the training efficiency. Moreover, since the model directly learns from the reference signal receiving power measurement data that reflects the real environment, it is possible to generate a channel power spectrum estimation result that is closer to reality and more "realistic"; in the actual channel power spectrum estimation, the real-time and easily obtainable grid to be estimated is used. The grid position of the grid and the cell position and cell antenna parameters in the communication cell are efficiently extracted from the pre-obtained realistic statistical channel model, thereby avoiding the need for real-time acquisition of complex, time-varying, high-precision environmental information and significantly reducing the computational complexity, so as to efficiently obtain the real-time estimated channel power spectrum in the wireless communication system, thereby effectively improving the channel estimation efficiency in the wireless communication system while ensuring the accuracy of the channel estimation; in addition, by starting from the basic physical quantities such as channel impulse response, precoding and transmit power, the expected reference signal received power and the underlying channel statistical characteristics (ultimately reflected in the clear mathematical relationship between the angular power spectrum and the coefficient matrix) are derived, which provides a solid physical foundation and guidance for subsequent model training. Based on the interpretability, it avoids the drawback of the pure black box model that lacks physical meaning, and uses this physical relationship to construct the learning goal, and uses the actual measured and easily accessible reference signal receiving power data to iteratively train the deep learning model, so that the model can learn the complex propagation characteristics under a specific "real-life" environment, overcoming the problems of poor generalization ability of traditional statistical models or deterministic models that rely on high-precision environmental information and have a large amount of calculation, and realizes the "real-life" and high efficiency of the model. By introducing a sparse penalty term in the loss function, it utilizes the prior knowledge that the wireless channel angular power spectrum is usually sparse, and guides the model to learn to obtain a power spectrum representation that is more in line with physical reality, more concise and effective, which not only helps to improve the generalization ability of the model, but also It can not only improve the force and estimation accuracy, but also improve the robustness of the model to a certain extent, thereby obtaining an efficient, accurate and practical channel power spectrum estimation model that integrates physical principles, measured data and prior knowledge; and by combining the feature units of multiple grids into a feature unit matrix, and then applying the multi-head self-attention mechanism, the model can make the feature representation of each grid (as the query Q) pay attention to and weightedly aggregate the information from all other grids (including itself), and effectively learn complex spatial patterns such as how the signal propagates in space, the spatial continuity of the shadow effect, and the mutual influence of interference between different positions. The multi-head mechanism further allows the model to simultaneously capture multiple types or scales of spatial correlation features from different representation subspaces.This deep modeling capability of spatial context overcomes the limitations of traditional methods or simple models that may treat each grid as an isolated point or only consider the influence of local neighbors, so that the estimated processing data outputted can more comprehensively and realistically reflect the spatial heterogeneity and interaction of large-scale wireless environments, greatly improving the accuracy and "realistic" degree of subsequent channel power spectrum estimation, and providing higher-quality input for applications such as network optimization that rely on accurate channel information; and, through linear transformation, nonlinear activation (such as GELU), layer normalization and regularization (such as Dropout), the input features are initially converted and refined in depth nonlinearly, while ensuring the stability of training and the generalization ability of the model, and using residual connections and further feature transfer processing, allowing the construction of deeper The hierarchical network structure is used to capture finer feature details, effectively solving the gradient vanishing problem in deep network training, thereby significantly improving the model's ability to learn from complex input features and infer precise power spectrum details, and applying non-negative activation functions (such as ReLU) to process the results before the final output, forcibly ensuring that the estimated power spectrum values are non-negative, which directly reflects the constraints of the physical model, making the output results more consistent with physical reality, and improving the reliability and credibility of the estimation results. The final estimated channel power spectrum is then directly obtained based on this deeply processed and physically constrained activation data, providing an end-to-end, precise mapping from high-level features to specific power spectrum distribution, providing high-quality, physically meaningful channel characteristic descriptions for subsequent network optimization applications.

[0200] The present application also provides a channel power spectrum estimation device, which can implement the above-mentioned channel power spectrum estimation method. Figure 16 , the apparatus 1600 comprises:

[0201] Data acquisition module 1610 is configured to acquire a grid position of a grid to be estimated in a target area, and acquire a cell position and cell antenna parameters of at least one communication cell corresponding to the grid position, wherein the realistic statistical channel model is pre-trained using reference signal received powers between multiple grids and the communication cells in the target area;

[0202] The cell feature processing module 1620 is configured to input the cell location and cell antenna parameters into a cell feature encoder for feature encoding processing to obtain a cell feature code;

[0203] The grid feature processing module 1630 is used to input the grid position and the cell position into the grid feature encoder for feature encoding processing to obtain a grid feature code;

[0204] The channel power spectrum estimation module 1640 is used to input the cell feature code and the grid feature code into the angle power spectrum estimation network to perform power spectrum estimation, and obtain the estimated channel power spectrum between the grid to be estimated and the communication cell.

[0205] In some embodiments, the channel power spectrum estimation module 1640 is further configured to:

[0206] After performing XOR processing on the cell feature code and the grid feature code, the position embedding of the grid position is superimposed to obtain a feature processing unit;

[0207] The feature processing unit is input into the transformer module for attention processing to obtain estimated processing data;

[0208] The estimated processed data is input into the angle power spectrum estimation network for power spectrum estimation to obtain the estimated channel power spectrum.

[0209] In some embodiments, the channel power spectrum estimation module 1640 is further configured to:

[0210] Obtaining other feature processing units between other grids in the target area and the communication cell, and combining the other feature processing units with the feature processing unit to obtain a feature unit matrix;

[0211] The feature unit matrix is input into the multi-head attention module for multi-head attention processing to obtain estimated processed data.

[0212] In some embodiments, the channel power spectrum estimation module 1640 is further configured to:

[0213] In the multi-head attention module, the partitioning dimension value of each attention head is obtained based on the ratio of the data dimension of the feature unit matrix and the number of attention heads;

[0214] Generate the projection parameter matrix of each attention head based on the partition dimension value;

[0215] In each self-attention processing module, a matrix query vector, a matrix key vector, and a matrix value vector are generated based on the feature unit matrix, and attention processing is performed on the matrix query vector, the matrix key vector, and the matrix value vector based on the corresponding projection parameter matrix and the linear transformation matrix to obtain sub-estimation processing data;

[0216] Multi-head attention processing is performed on multiple sub-estimation processing data to obtain estimated processing data.

[0217] In some embodiments, the channel power spectrum estimation module 1640 is further configured to:

[0218] In the angle power spectrum estimation network, the estimated processed data is sequentially subjected to linear processing, activation function processing, layer normalization processing, and regularization processing to obtain preliminary processed data;

[0219] The preliminary processing results are sequentially subjected to residual connection enhancement processing and feature transfer processing to obtain enhanced processing data;

[0220] Perform non-negative activation processing on the enhanced processed data to obtain activated processed data;

[0221] Estimation processing is performed based on the activation processed data to obtain an estimated channel power spectrum.

[0222] In some embodiments, the data acquisition module 1610 is further configured to:

[0223] Obtaining a channel impulse response function and a precoding matrix between each antenna and each grid in a communication cell;

[0224] The product of all channel impulse response functions and precoding matrices is accumulated, squared, and multiplied by the transmit power of the communication cell to obtain the reference signal received power function of each grid;

[0225] Performing time expectation processing on the reference signal received power function to obtain an expected reference signal received power function, wherein the expected reference signal received power function includes a coefficient matrix between the grid and the communication cell;

[0226] Generate a channel power spectrum estimation model based on the coefficient matrix, the reference signal received power function, and the channel power spectrum estimation function between each grid and the communication cell;

[0227] The realistic statistical channel model is iteratively updated multiple times based on the channel power spectrum estimation model and the received powers of multiple reference signals.

[0228] In some embodiments, the data acquisition module 1610 is further configured to:

[0229] Obtaining a sparse penalty weight, and obtaining a penalty channel power spectrum function based on the product of the sparse penalty weight and the channel power spectrum estimation function;

[0230] Generate a loss function based on the channel power spectrum estimation model and the accumulated value of the penalty channel power spectrum function;

[0231] The realistic statistical channel model is updated multiple times iteratively based on the loss function and the received powers of multiple reference signals.

[0232] In the above embodiments, the description of each embodiment has different emphases. For the part not described in detail in a certain embodiment, the specific implementation of the channel power spectrum estimation device is basically the same as the specific implementation of the above channel power spectrum estimation method, and will not be repeated here.

[0233] In the embodiment of the present application, the channel power spectrum estimation device obtains a realistic statistical channel model by using a reference signal receiving power measured data with low acquisition cost between the target area and the communication cell in advance, and adopts the physical model guidance and deep learning fusion method to train it, thereby avoiding the real-time, complex, high-dimensional environmental information acquisition and high-computation deterministic modeling required by traditional high-precision methods, so as to reduce the training complexity of the estimation model for channel power spectrum estimation and improve the training efficiency. Moreover, since the model directly learns from the reference signal receiving power measured data reflecting the real environmental impact, it is possible to generate a channel power spectrum estimation result that is closer to reality and more "realistic"; in the actual channel power spectrum estimation, the real-time and The grid position of the grid to be estimated, the cell position and cell antenna parameters in the communication cell, and the pre-obtained realistic statistical channel model are easily obtained for efficient feature extraction, thereby avoiding the need for real-time acquisition of complex, time-varying, high-precision environmental information, significantly reducing the computational complexity, and efficiently obtaining the real-time estimated channel power spectrum in the wireless communication system, thereby effectively improving the channel estimation efficiency in the wireless communication system while ensuring the accuracy of the channel estimation; in addition, by starting from basic physical quantities such as channel impulse response, precoding, and transmit power, the expected reference signal received power and the underlying channel statistical characteristics (ultimately reflected in the clear mathematical relationship between the angular power spectrum and the coefficient matrix) are derived, which provides a solid foundation for subsequent model training. The physical basis and interpretability of the method avoids the drawback of the pure black box model that lacks physical meaning, and uses this physical relationship to construct the learning goal, and uses the actual measured and easily accessible reference signal receiving power data to iteratively train the deep learning model, so that the model can learn the complex propagation characteristics under a specific "real-life" environment, overcoming the problems of poor generalization ability of traditional statistical models or reliance on high-precision environmental information and large computational complexity of deterministic models, realizing the "real-life" and high efficiency of the model, and by introducing a sparse penalty term in the loss function, utilizing the prior knowledge that the wireless channel angular power spectrum is usually sparse, guiding the model to learn a power spectrum representation that is more in line with physical reality, more concise and effective, which not only helps to improve the model's The generalization ability and estimation accuracy can also improve the robustness of the model to a certain extent, thereby obtaining an efficient, accurate and practical channel power spectrum estimation model that integrates physical principles, measured data and prior knowledge; and by combining the feature units of multiple grids into a feature unit matrix, and then applying the multi-head self-attention mechanism, the model can allow the feature representation of each grid (as the query Q) to pay attention to and weightedly aggregate information from all other grids (including itself), effectively learning complex spatial patterns such as how signals propagate in space, the spatial continuity of shadow effects, and the mutual influence of interference between different positions. The multi-head mechanism further allows the model to simultaneously capture multiple types or scales of spatial correlation features from different representation subspaces.This deep modeling capability of spatial context overcomes the limitations of traditional methods or simple models that may treat each grid as an isolated point or only consider the influence of local neighbors, so that the estimated processing data outputted can more comprehensively and realistically reflect the spatial heterogeneity and interaction of large-scale wireless environments, greatly improving the accuracy and "realistic" degree of subsequent channel power spectrum estimation, and providing higher-quality input for applications such as network optimization that rely on accurate channel information; and, through linear transformation, nonlinear activation (such as GELU), layer normalization and regularization (such as Dropout), the input features are initially converted and refined in depth nonlinearly, while ensuring the stability of training and the generalization ability of the model, and using residual connections and further feature transfer processing, allowing the construction of deeper The hierarchical network structure is used to capture finer feature details, effectively solving the gradient vanishing problem in deep network training, thereby significantly improving the model's ability to learn from complex input features and infer precise power spectrum details, and applying non-negative activation functions (such as ReLU) to process the results before the final output, forcibly ensuring that the estimated power spectrum values are non-negative, which directly reflects the constraints of the physical model, making the output results more consistent with physical reality, and improving the reliability and credibility of the estimation results. The final estimated channel power spectrum is then directly obtained based on this deeply processed and physically constrained activation data, providing an end-to-end, precise mapping from high-level features to specific power spectrum distribution, providing high-quality, physically meaningful channel characteristic descriptions for subsequent network optimization applications.

[0234] An embodiment of the present application further provides an electronic device, including:

[0235] at least one memory;

[0236] at least one processor;

[0237] at least one program;

[0238] The program is stored in the memory, and the processor executes the at least one program to implement the channel power spectrum estimation method described above. The electronic device can be any smart terminal including a mobile phone, a tablet computer, a personal digital assistant (PDA), an in-vehicle computer, etc.

[0239] See also Figure 17 , Figure 17 The hardware structure of an electronic device according to another embodiment is shown. The electronic device includes:

[0240] The processor 1701 can be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0241] The memory 1702 can be implemented in the form of ROM (Read Only Memory), static storage device, dynamic storage device or RAM (Random Access Memory). The memory 1702 can store an operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1702 and is called by the processor 1701 to execute the channel power spectrum estimation method of the embodiment of the present application;

[0242] Input / output interface 1703, used to implement information input and output;

[0243] Communication interface 1704, used to implement communication interaction between this device and other devices, which can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WiFi, Bluetooth, etc.);

[0244] Bus 1705 , which transmits information between various components of the device (e.g., processor 1701 , memory 1702 , input / output interface 1703 , and communication interface 1704 );

[0245] The processor 1701 , the memory 1702 , the input / output interface 1703 and the communication interface 1704 are connected to each other in communication within the device via a bus 1705 .

[0246] An embodiment of the present application further provides a storage medium, which is a computer-readable storage medium and stores a computer program. When the computer program is executed by a processor, the above-mentioned channel power spectrum estimation method is implemented.

[0247] The memory, as a non-transient computer-readable storage medium, can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0248] The embodiments described in the embodiments of this application are intended to more clearly illustrate the technical solutions of the embodiments of this application and do not constitute a limitation on the technical solutions provided by the embodiments of this application. Those skilled in the art will appreciate that with the evolution of technology and the emergence of new application scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0249] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present application, and may include more or fewer steps than shown in the figures, or a combination of certain steps, or different steps.

[0250] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, i.e., they may be located in one place or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of this embodiment.

[0251] Those skilled in the art will appreciate that all or some of the steps in the methods, systems, and functional modules / units in the devices disclosed above may be implemented as software, firmware, hardware, or appropriate combinations thereof.

[0252] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0253] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single items or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0254] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the above-mentioned units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. The mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0255] The units described above as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0256] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0257] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes multiple instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: various media that can store programs, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0258] The preferred embodiments of the present invention are described above with reference to the accompanying drawings, but are not intended to limit the scope of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and essence of the present invention should be within the scope of the present invention.

Claims

1. A channel power spectrum estimation method, characterized in that: The method is performed by a scene-based statistical channel model, wherein the scene-based statistical channel model includes at least: a cell feature encoder, a grid feature encoder, and an angle power spectrum estimation network. The method includes: Obtaining a grid position of a grid to be estimated in a target area, and obtaining a cell position and cell antenna parameters of at least one communication cell corresponding to the grid position, wherein the realistic statistical channel model is pre-trained based on reference signal received powers between a plurality of grids in the target area and the communication cell; Inputting the cell location and the cell antenna parameters into the cell feature encoder for feature coding to obtain a cell feature code; Inputting the grid position and the cell position into the grid feature encoder for feature coding to obtain a grid feature code; The cell feature code and the grid feature code are input into the angle power spectrum estimation network to perform power spectrum estimation, so as to obtain an estimated channel power spectrum between the grid to be estimated and the communication cell.

2. The channel power spectrum estimation method according to claim 1, wherein The scene-based statistical channel model further includes a converter module, which inputs the cell feature code and the grid feature code into the angle power spectrum estimation network to perform power spectrum estimation to obtain an estimated channel power spectrum between the grid to be estimated and the communication cell, including: After performing XOR processing on the cell feature code and the grid feature code, the position embedding of the grid position is superimposed to obtain a feature processing unit; Inputting the feature processing unit into the converter module for attention processing to obtain estimated processing data; The estimated processed data is input into the angle power spectrum estimation network to perform power spectrum estimation to obtain the estimated channel power spectrum.

3. The channel power spectrum estimation method according to claim 2, wherein The converter module includes a multi-head attention module, and the feature processing unit is input into the converter module for attention processing to obtain estimated processing data, including: Acquire other feature processing units between the other grids in the target area and the communication cell, and combine the other feature processing units with the feature processing unit to obtain a feature unit matrix; The feature unit matrix is input into the multi-head attention module, multi-head attention processing is performed, and the estimated processing data is obtained.

4. The channel power spectrum estimation method according to claim 3, wherein The multi-head attention module includes multiple self-attention processing modules, and the feature unit matrix is input into the multi-head attention module to perform multi-head attention processing to obtain the estimated processing data, including: In the multi-head attention module, a partitioning dimension value of each attention head is obtained based on the ratio of the data dimension of the feature unit matrix to the number of attention heads; Generate a projection parameter matrix for each attention head based on the partition dimension value; In each of the self-attention processing modules, a matrix query vector, a matrix key vector, and a matrix value vector are generated based on the feature unit matrix, and attention processing is performed on the matrix query vector, the matrix key vector, and the matrix value vector based on the corresponding projection parameter matrix and the linear transformation matrix to obtain sub-estimation processing data; Multi-head attention processing is performed on the plurality of sub-estimation processing data to obtain the estimation processing data.

5. The channel power spectrum estimation method according to claim 2, wherein The step of inputting the estimated processed data into the angle power spectrum estimation network to perform power spectrum estimation to obtain the estimated channel power spectrum comprises: In the angle power spectrum estimation network, the estimated processed data is sequentially subjected to linear processing, activation function processing, layer normalization processing, and regularization processing to obtain preliminary processed data; Performing residual connection enhancement processing and feature transfer processing on the preliminary processing results in sequence to obtain enhanced processing data; Performing non-negative activation processing on the enhanced processed data to obtain activated processed data; Estimation processing is performed based on the activation processed data to obtain the estimated channel power spectrum.

6. The channel power spectrum estimation method according to claim 1, wherein The training steps of the scene-based statistical channel model include: Obtaining a channel impulse response function and a precoding matrix between each antenna in the communication cell and each of the grids; Accumulating the products of all the channel impulse response functions and the precoding matrix, squaring the products, and then multiplying the products by the transmit power of the communication cell to obtain a reference signal received power function of each grid; Performing time expectation processing on the reference signal received power function to obtain an expected reference signal received power function, wherein the expected reference signal received power function includes a coefficient matrix between the grid and the communication cell; Generate a channel power spectrum estimation model based on the coefficient matrix, the reference signal received power function, and the channel power spectrum estimation function between each of the grids and the communication cell; The realistic statistical channel model is iteratively updated multiple times based on the channel power spectrum estimation model and the multiple reference signal received powers.

7. The channel power spectrum estimation method according to claim 6, characterized in that The iteratively updating the scene-based statistical channel model multiple times based on the channel power spectrum estimation model includes: Obtaining a sparse penalty weight, and obtaining a penalty channel power spectrum function based on a product of the sparse penalty weight and the channel power spectrum estimation function; Generating a loss function based on the channel power spectrum estimation model and the accumulated value of the penalty channel power spectrum function; The realistic statistical channel model is iteratively updated multiple times based on the loss function and the multiple reference signal received powers.

8. A channel power spectrum estimation device, characterized in that: The method is executed by a scene-based statistical channel model, wherein the scene-based statistical channel model includes at least: a cell feature encoder, a grid feature encoder, and an angle power spectrum estimation network. The device includes: a data acquisition module, configured to acquire a grid position of a grid to be estimated in a target area, and acquire a cell position and cell antenna parameters of at least one communication cell corresponding to the grid position, wherein the realistic statistical channel model is pre-trained using reference signal received powers between a plurality of grids in the target area and the communication cell; a cell feature processing module, configured to input the cell location and the cell antenna parameters into the cell feature encoder for feature coding processing to obtain a cell feature code; a grid feature processing module, configured to input the grid position and the cell position into the grid feature encoder for feature coding processing to obtain a grid feature code; The channel power spectrum estimation module is used to input the cell feature code and the grid feature code into the angle power spectrum estimation network to perform power spectrum estimation, and obtain the estimated channel power spectrum between the grid to be estimated and the communication cell.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and is characterized in that when the processor executes the computer program, the channel power spectrum estimation method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the channel power spectrum estimation method according to any one of claims 1 to 7 is implemented.

Images