Channel estimation method and device, electronic equipment, storage medium and program product

Abstract

The invention relates to a channel estimation method and device, electronic equipment, a storage medium and a program product, and relates to the technical field of communication. The channel estimation method comprises the following steps: determining a pilot signal matrix according to the number of transmitting antennas and a pilot length corresponding to a communication signal transmitted by a target channel; constructing a to-be-processed matrix of the target channel according to the pilot signal matrix, the noise power of the communication signal and the number of receiving antennas; determining a time domain weight and a frequency domain weight based on the correlation influence parameters and a weight decision model; generating an iteration initial value according to the time domain weight, a first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, the first time slot and a second reference matrix corresponding to the second subcarrier; performing iterative operation on the to-be-processed matrix according to the iterative initial value and a preset inversion iterative model; and when a preset iteration termination condition is satisfied, obtaining an inverse matrix of the to-be-processed matrix so as to determine a channel matrix of the target channel. The method can improve the communication performance.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to a channel estimation method, device, electronic device, storage medium and program product. Background Art

[0002] Matrix operations are important mathematical tools in the field of communications technology. Matrix inversion is a common operation within matrix operations and a key step in channel estimation scenarios. In channel estimation scenarios, matrix inversion typically uses an approximate iterative inversion method to gradually approximate the inverse result. However, this method requires many iterations and is computationally intensive, resulting in low channel estimation efficiency. Consequently, the channel estimation result may not match the actual channel state, impacting communication performance. Summary of the Invention

[0003] To overcome the problems existing in the related art, the present invention provides a channel estimation method, device, electronic device, storage medium and program product.

[0004] According to a first aspect of an embodiment of the present invention, a channel estimation method is provided, the method comprising: Determining a pilot signal matrix corresponding to the communication signal according to the number of transmitting antennas and a pilot length corresponding to a target channel sending a communication signal, wherein the target channel is a channel corresponding to the first time slot and the first subcarrier; Constructing a to-be-processed matrix of the target channel according to the pilot signal matrix, the noise power of the communication signal, and the number of receiving antennas; Determine the time domain weight and frequency domain weight based on the correlation impact parameters and weight decision model; generating an iteration initial value according to the time domain weight, a first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, and a second reference matrix corresponding to the first time slot and the second subcarrier, wherein the second time slot is used to indicate a time slot previous to the first time slot in continuously distributed time domain resources, and the second subcarrier is used to indicate a subcarrier previous to the first subcarrier in continuously distributed frequency resources; performing an iterative operation on the matrix to be processed according to the iterative initial value and the preset inverse iterative model, determining an iteration step size according to a step size decision model during the iterative operation, and updating the preset inverse iterative model for the next iterative operation according to the iteration step size; When a preset iteration termination condition is met, obtaining the inverse matrix of the matrix to be processed; The channel matrix of the target channel is determined according to the inverse matrix of the matrix to be processed.

[0005] In this embodiment, initial values for iteration are generated based on reference matrices of channels that are correlated with the time-frequency resources of the target channel, namely, a first reference matrix and a second reference matrix. This enables the iteration results to approach the optimal solution more quickly, reduces the number of iterations, and reduces the amount of calculation. In addition, the iteration step size is updated in a timely manner in each iterative operation, which can avoid oscillations while enabling the iteration results to approach the optimal solution more quickly, further reduces the number of iterations, and reduces the amount of calculation, thereby improving the efficiency of channel estimation, avoiding the problem of mismatch between the channel estimation result and the actual channel state, and improving communication performance.

[0006] In an exemplary embodiment, the method further comprises: Obtain a channel matrix of a first channel and a channel matrix of a second channel, where the first channel is a channel corresponding to the second time slot and the first subcarrier, and the second channel is a channel corresponding to the first time slot and the second subcarrier; The channel matrix of the first channel is used as the first reference matrix, and the channel matrix of the second channel is used as the second reference matrix.

[0007] In this embodiment, the channel matrix of the first channel and the channel matrix of the second channel are used to generate an initial value for iteration, which can make the initial value for iteration closer to the channel state of the target channel, thereby reducing the number of iterations and the amount of calculation.

[0008] In an exemplary embodiment, determining the time domain weight and the frequency domain weight based on the correlation impact parameter and the weight decision model includes: Inputting the moving speed of the electronic device, the correlation between the second time slot and the first time slot, and the correlation between the first subcarrier and the second subcarrier into the weight decision model to obtain the time domain weight and the frequency domain weight; The weight decision model is obtained by training based on the reference moving speed of the electronic device, the correlation between reference time domains, the correlation between reference frequency domains, and the reference time domain weight and the reference frequency domain weight.

[0009] In this embodiment, the time domain weights and frequency domain weights are determined by a pre-trained weight decision model, so that the obtained weights are more consistent with the correlation characteristics between time domains and frequency domains, thereby ensuring the reliability of the iterative initial values generated according to the weights.

[0010] In an exemplary embodiment, generating an iteration initial value according to the time domain weight, a first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, a second reference matrix corresponding to the first time slot and the second subcarrier includes: Encoding the first reference matrix, extracting features of the first reference matrix, and then decoding the features of the first reference matrix to obtain a first matrix reconstructed from the first reference matrix; Encoding the second reference matrix, extracting features of the second reference matrix, and then decoding the features of the second reference matrix to obtain a second matrix reconstructed from the second reference matrix; The iteration initial value is generated according to the time domain weight and the first matrix, and the frequency domain weight and the second matrix.

[0011] In this embodiment, by reconstructing the first reference matrix and the second reference matrix, the hidden relationship in the reference matrix can be discovered, and the reconstructed matrix can better reflect the actual channel state, so that the generated iterative initial value is closer to the optimal solution.

[0012] In an exemplary embodiment, the preset inverse iterative model is used to indicate the correlation between the iteration result of the current iteration and the iteration result of the next iteration, and the determining of the iteration step size according to the step size decision model and updating the preset inverse iterative model for the next iteration operation according to the iteration step size include: Inputting the condition number of the matrix to be processed, the iteration residual and the number of current iterations into the step size decision model to obtain the iteration step size; A preset inverse iterative model corresponding to the next iteration is determined according to the iteration step size, the iteration result of the current iteration and the matrix to be processed.

[0013] In this embodiment, the iteration step size is determined based on the iteration residual and the step size decision model, and the iteration step size can be adjusted in time according to the iteration result obtained in each iteration, so that the iteration step size better matches the current iteration situation and effectively avoids oscillation.

[0014] In an exemplary embodiment, the preset iteration termination condition is that the iteration residual is smaller than a first threshold and the convergence probability is greater than a second threshold.

[0015] In this embodiment, the iteration is terminated when the iterative residual is less than the first threshold and the convergence probability is greater than the second threshold, so that the optimal solution can be obtained in time and the inversion efficiency is improved.

[0016] In an exemplary embodiment, determining the channel matrix of the target channel according to the inverse matrix of the matrix to be processed includes: The inverse matrix of the matrix to be processed is input into a preset channel estimation model to obtain the channel matrix of the target channel, wherein the preset channel estimation model is one of a linear minimum mean square error channel estimation model and a least squares channel estimation model.

[0017] In this embodiment, channel estimation is performed using a channel estimation model according to the inverse matrix of the matrix to be processed, so that an accurate channel matrix can be obtained.

[0018] In an exemplary embodiment, the method further comprises: Acquire scenario information, and determine the corresponding preset channel estimation model based on the scenario information.

[0019] In this embodiment, selecting a preset channel estimation model according to an application scenario can make the preset channel estimation model more suitable for the current application scenario, further ensuring the accuracy of the channel matrix.

[0020] According to a second aspect of an embodiment of the present invention, a channel estimation device is provided, the device comprising: A first determining module is configured to determine a pilot signal matrix corresponding to the communication signal according to the number of transmitting antennas and the pilot length corresponding to the communication signal sent by the target channel, wherein the target channel is a channel corresponding to the first time slot and the first subcarrier; A construction module is configured to construct a to-be-processed matrix of the target channel according to the pilot signal matrix, the noise power of the communication signal and the number of receiving antennas; A second determination module is configured to determine the time domain weight and the frequency domain weight based on the correlation influence parameter and the weight decision model; a generating module configured to generate an iteration initial value according to the time domain weight, a first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, and a second reference matrix corresponding to the first time slot and the second subcarrier, wherein the second time slot is used to indicate a time slot previous to the first time slot in continuously distributed time domain resources, and the second subcarrier is used to indicate a subcarrier previous to the first subcarrier in continuously distributed frequency resources; an iterative module configured to perform an iterative operation on the matrix to be processed according to the iteration initial value and a preset inverse iterative model, determine an iteration step size according to a step size decision model during the iterative operation, and update the preset inverse iterative model for the next iterative operation according to the iteration step size; A third determining module is configured to obtain an inverse matrix of the matrix to be processed when a preset iteration termination condition is met; The fourth determining module is configured to determine the channel matrix of the target channel according to the inverse matrix of the matrix to be processed.

[0021] According to a third aspect of an embodiment of the present invention, there is provided an electronic device, including: processor; a memory for storing processor-executable instructions; The processor is configured to execute the method as described in the first aspect of the embodiment of the present invention.

[0022] According to a fourth aspect of an embodiment of the present invention, a non-temporary computer-readable storage medium is provided, which, when instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to execute the method described in the first aspect of the embodiment of the present invention.

[0023] According to a fifth aspect of an embodiment of the present invention, there is provided a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the method as described in the first aspect of the embodiment of the present invention is implemented. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0025] Figure 1 is a flow chart showing a channel estimation method according to an exemplary embodiment; Figure 2 is a flowchart of generating an iteration initial value according to an exemplary embodiment; Figure 3 is a flow chart illustrating a method for determining a preset inversion iterative model according to an exemplary embodiment; Figure 4 is a block diagram of a channel estimation device according to an exemplary embodiment; Figure 5 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0026] Exemplary embodiments are described in detail herein, with examples illustrated in the accompanying drawings. When the following description refers to the drawings, identical numerals in different figures represent identical or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present invention.

[0027] In an exemplary embodiment of the present invention, a channel estimation method is provided. Figure 1 FIG. 1 is a flow chart showing a channel estimation method according to an exemplary embodiment. Figure 1 As shown, the process includes the following steps S101 to S107.

[0028] Step S101, determining a pilot signal matrix corresponding to the communication signal according to the number of transmitting antennas and the pilot length corresponding to the communication signal sent by the target channel, wherein the target channel is the channel corresponding to the first time slot and the first subcarrier.

[0029] Step S102: constructing a matrix to be processed of the target channel according to the pilot signal matrix, the noise power of the communication signal and the number of receiving antennas.

[0030] Step S103 : determining the time domain weight and the frequency domain weight based on the correlation influence parameter and the weight decision model.

[0031] Step S104, generating an initial value for iteration based on the time domain weight, the first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, the second reference matrix corresponding to the first time slot and the second subcarrier, wherein the second time slot is used to indicate the previous time slot of the first time slot in the continuously distributed time domain resources, and the second subcarrier is used to indicate the previous subcarrier of the first subcarrier in the continuously distributed frequency resources.

[0032] Step S105, performing iterative operation on the matrix to be processed according to the initial value of the iteration and the preset inverse iteration model. During the iterative operation, the iteration step size is determined according to the step size decision model, and the preset inverse iteration model for the next iterative operation is updated according to the iteration step size.

[0033] Step S106: When a preset iteration termination condition is met, the inverse matrix of the matrix to be processed is obtained.

[0034] Step S107 : determining the channel matrix of the target channel according to the inverse matrix of the matrix to be processed.

[0035] The method in the embodiments of the present invention is applied to electronic devices, including mobile phones, tablets, smart wearable devices, routers, and other devices with wireless communication capabilities. Optionally, the electronic device includes a symmetric Multiple-Input Multiple-Output (MIMO) system. In a symmetric MIMO system, the transmitter and receiver simultaneously use multiple antennas, and the number of transmitting antennas is the same as the number of receiving antennas.

[0036] In step S101, the target channel is the channel corresponding to the first time slot and the first subcarrier, the first time slot is the time domain resource used by the target channel to send and receive communication signals, and the first subcarrier is the frequency domain resource used by the target channel to send and receive communication signals, wherein the first time slot can be any time slot in the time domain resource, and the first subcarrier can be any subcarrier in the frequency domain resource. The pilot length corresponding to the sent communication signal represents the length of the time domain resource occupied by the sent pilot signal, which can be recorded as , the number of transmitting antennas in an electronic device can be recorded as , according to the pilot length and the number of transmitting antennas, the pilot signal matrix corresponding to the communication signal is constructed, and the dimension of the pilot signal matrix is .

[0037] In step S102, the noise power of the communication signal represents the power of the noise generated when the communication signal is received by the receiving antenna. According to the pilot signal matrix, the noise power of the communication signal and the number of receiving antennas, the matrix to be processed of the target channel is constructed by the following formula:

[0038] in, represents the matrix to be processed of the target channel, Indicates the position of the first time slot in the time domain resource. The first time slot can also be recorded as Time slot, Indicates the number of the first subcarrier in the frequency domain resource. The first subcarrier can also be recorded as subcarrier, Indicates the dimension The pilot signal matrix, Represents the conjugate matrix of the pilot signal matrix, with dimension , represents the number of transmitting antennas, represents the pilot length, represents the noise power of the communication signal, Indicates the dimension The identity matrix, represents the number of receiving antennas, and ,therefore, The dimension is or .

[0039] In step S103, the correlation includes time domain correlation and frequency domain correlation. The time domain correlation represents the correlation between different time slots in the time domain resources, and the frequency domain correlation represents the correlation between different subcarriers in the frequency domain resources. The correlation influencing parameters represent parameters that affect the time domain correlation and / or frequency domain correlation. For example, the parameters that affect the time domain correlation include the moving speed of the electronic device and the communication environment. The faster the moving speed, the more complex the communication environment, and the lower the time domain correlation. The parameters that affect the frequency domain correlation include the subcarrier spacing and the communication environment. The larger the subcarrier spacing, the more complex the communication environment, and the lower the frequency domain correlation. The weight decision model represents a model for determining the time domain weight and the frequency domain weight based on the correlation influencing parameters. The time domain weight represents the weight of the reference matrix corresponding to the channel with different time domain resources from the target channel, and the frequency domain weight represents the weight of the reference matrix corresponding to the channel with different frequency domain resources from the target channel.

[0040] In step S104, the second time slot is used to indicate the previous time slot of the first time slot in the continuously distributed time domain resources. For example, the continuously distributed time domain resources include four time slots numbered t1, t2, t3, and t4, and when the first time slot is t3, the second time slot is t2. The second subcarrier is used to indicate the previous subcarrier of the first subcarrier in the continuously distributed frequency resources. For example, the continuously distributed frequency domain resources include four subcarriers numbered k1, k2, k3, and k4, and when the first subcarrier is k2, the second subcarrier is k1.

[0041] The first reference matrix may be a channel matrix corresponding to the second time slot and the first subcarrier, or may be an inverse matrix of a matrix to be processed for the channel corresponding to the second time slot and the first subcarrier; the second reference matrix may be a channel matrix corresponding to the first time slot and the second subcarrier, or may be an inverse matrix of a matrix to be processed for the channel corresponding to the first time slot and the second subcarrier. The first reference matrix and the second reference matrix are of the same type, both being channel matrices or both being inverse matrices of a matrix to be processed.

[0042] If the first reference matrix is a reference matrix corresponding to a channel with different time domain resources than the target channel, the time domain weights are the weights of the first reference matrix. If the second reference matrix is a reference matrix corresponding to a channel with different frequency domain resources than the target channel, the frequency domain weights are the weights of the second reference matrix. The initial iteration value is generated based on the product of the time domain weights and the first reference matrix, and the product of the frequency domain weights and the second reference matrix. For example, the initial iteration value can be expressed as:

[0043] in, Indicates the target channel ( Time slot and subcarrier) corresponding to the initial value of the iteration, represents the frequency domain weight, Indicates the Time slot and The second reference matrix corresponding to the subcarrier, represents the time domain weight, Indicates the Time slot and A first reference matrix corresponding to the subcarrier.

[0044] In step S105, the preset inverse iterative model is used to perform iterative operations on the matrix to be processed to obtain the inverse matrix of the matrix to be processed. When the iterative operation of the matrix to be processed is performed by the preset inverse iterative model, the initial value of the iteration and the iteration step are input into the preset inverse iterative model, the result of the next iterative operation is output, and the result of the next iterative operation and the iteration step are then input into the preset inverse iterative model, and the result of the next iterative operation is output. Among them, the initial value of the iteration uses the initial value of the iteration calculated in the above steps. The initial value of the iteration can be understood as the initial value of the inverse matrix of the matrix to be processed. The result of each iterative operation can be understood as a value gradually approaching the inverse matrix of the matrix to be processed; the iteration step is determined according to the step decision model, and the iteration step is used to update the preset inverse iterative model for the next iterative operation. The iteration step of each iterative operation is adjusted according to the actual iteration result, which can ensure the adaptability of the iteration step to avoid oscillation.

[0045] In step S106, the preset iteration termination condition is determined based on the iteration residual and the convergence probability. The iteration residual represents the difference between the iteration results obtained from two adjacent iterations. The smaller the iteration residual, the closer the iteration result is to the optimal solution, that is, the closer the iteration result is to the inverse matrix of the matrix to be processed. The convergence probability represents the probability that the iteration result obtained from the current number of iterations will converge to the optimal solution. The greater the convergence probability, the closer the iteration result is to the optimal solution. When the iteration operation meets the preset iteration termination condition, the iteration result obtained at the end of the iteration operation is used as the inverse matrix of the matrix to be processed.

[0046] In step S107 , after the inverse matrix of the matrix to be processed is obtained, channel estimation is performed according to the inverse matrix of the matrix to be processed and a preset channel estimation model to obtain a channel matrix of the target channel.

[0047] In an exemplary embodiment of the present invention, after constructing the matrix to be processed of the target channel, an initial value for iteration is generated based on the time domain weight, the first reference matrix corresponding to the second time slot and the first subcarrier, and the frequency domain weight, the second reference matrix corresponding to the first time slot and the second subcarrier. Compared with the use of fixed initial values for iteration or the use of random numbers as initial values for iteration in related technologies, this method generates initial values for iteration based on the reference matrix of a channel that has correlation with the time-frequency resources of the target channel, which can make the iteration result approach the optimal solution faster, reduce the number of iterations, and reduce the amount of calculation; then, based on the initial value for iteration and the preset inverse iteration model, an iterative operation is performed on the matrix to be processed, and in the process of the iterative operation, the step size decision is made based on the step size decision. The model determines the iteration step size, and updates the preset inverse iteration model of the next iteration operation according to the iteration step size. Compared with the use of a fixed iteration step size or a random number as the iteration step size in the related art, the method updates the iteration step size in time in each iteration operation, which can avoid oscillation while making the iteration result approach the optimal solution faster, reduce the number of iterations, and reduce the amount of calculation; when the preset iteration termination condition is met, the inverse matrix of the matrix to be processed is obtained, and channel estimation is performed according to the inverse matrix of the matrix to be processed to obtain the channel matrix of the target channel, which can improve the channel estimation efficiency and avoid the problem of mismatch between the channel estimation result and the actual channel state, thereby improving the communication performance.

[0048] In some embodiments, the first reference matrix and the second reference matrix are obtained by the following steps: Obtain a channel matrix of a first channel and a channel matrix of a second channel, where the first channel is a channel corresponding to the second time slot and the first subcarrier, and the second channel is a channel corresponding to the first time slot and the second subcarrier; use the channel matrix of the first channel as a first reference matrix, and use the channel matrix of the second channel as a second reference matrix.

[0049] The first time slot is recorded as time slot, the first subcarrier is recorded as subcarrier, the second time slot is time slot, the second subcarrier is recorded as subcarriers, therefore, the channel matrix of the first channel can be recorded as , the channel matrix of the second channel can be recorded as Taking the channel matrix of the first channel as the first reference matrix and the channel matrix of the second channel as the second reference matrix, the initial value of the iteration can be expressed as:

[0050] in, Indicates the target channel ( Time slot and subcarrier) corresponding to the initial value of the iteration, represents the frequency domain weight, Indicates the Time slot and The channel matrix of the second channel corresponding to the subcarrier, represents the time domain weight, Indicates the Time slot and The channel matrix of the first channel corresponding to the subcarrier.

[0051] Since the channel matrix of the first channel can reflect the channel state of the first channel, the channel matrix of the second channel can reflect the channel state of the second channel, and both the first channel and the second channel are correlated with the target channel in terms of time-frequency resources, the channel matrix of the first channel is used as the first reference matrix, and the channel matrix of the second channel is used as the second reference matrix to generate an initial value for iteration. This can make the initial value for iteration closer to the channel state of the target channel, thereby reducing the number of iterations.

[0052] In some embodiments, determining the time domain weight and the frequency domain weight based on the correlation impact parameter and the weight decision model in step S103 includes: The moving speed of the electronic device, the correlation between the second time slot and the first time slot, and the correlation between the first subcarrier and the second subcarrier are input into the weight decision model to obtain the time domain weight and the frequency domain weight; wherein the weight decision model is obtained based on the reference moving speed of the electronic device, the correlation between the reference time domains, the correlation between the reference frequency domains, and the reference time domain weight and the reference frequency domain weight.

[0053] The moving speed of the electronic device is obtained by the speed sensor in the electronic device. The correlation between the second time slot and the first time slot, and the correlation between the first subcarrier and the second subcarrier can be obtained according to any correlation estimation algorithm. For example, the statistical values of the communication signals received in different time slots or different frequency domains, such as the average value, variance, power, etc., are obtained, and the ratio of the difference between the statistical values to the maximum value is used as the correlation. The smaller the difference between the statistical values, the higher the correlation. For example, if the average value of the communication signal received in the first time slot is , the average value of the communication signal received in the second time slot is , then the correlation between the second time slot and the first time slot is The moving speed of the electronic device, the correlation between the second time slot and the first time slot, and the correlation between the first subcarrier and the second subcarrier are input into the weight decision model, and the time domain weight and the frequency domain weight are output.

[0054] Among them, the weight decision model is a model obtained by pre-training based on multiple groups of reference data. The reference data can also be understood as experimental data. Each group of reference data includes the reference moving speed of the electronic device, the correlation between the reference time domains, the correlation between the reference frequency domains, and the reference time domain weight and the reference frequency domain weight. The correlation between the reference time domains represents the correlation between the reference first time domain and the reference second time domain, and the correlation between the reference frequency domains represents the correlation between the reference first subcarrier and the reference second subcarrier. It should be noted that the calculation method of the correlation between the reference time domains is the same as the calculation method of the correlation between the second time slot and the first time slot, and the calculation method of the correlation between the reference frequency domains is the same as the calculation method of the correlation between the first subcarrier and the second subcarrier.

[0055] In this embodiment, the time domain weights and frequency domain weights are determined by a pre-trained weight decision model, so that the obtained weights are more consistent with the correlation characteristics between time domains and frequency domains, thereby ensuring the reliability of the iterative initial values generated according to the weights.

[0056] In some embodiments, in the above step S104, the initial value of the iteration is generated according to the time domain weight, the first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, the second reference matrix corresponding to the first time slot and the second subcarrier, including the following: Figure 2 The following steps S1041 to S1043 are shown.

[0057] Step S1041 : Encode the first reference matrix, extract features of the first reference matrix, and then decode the features of the first reference matrix to obtain a first matrix reconstructed from the first reference matrix.

[0058] Step S1042: encode the second reference matrix, extract features of the second reference matrix, and then decode the features of the second reference matrix to obtain a second matrix reconstructed from the second reference matrix.

[0059] Step S1043 : generating an iteration initial value according to the time domain weight and the first matrix, and the frequency domain weight and the second matrix.

[0060] The first reference matrix is encoded using a first matrix encoder, which is composed of a multi-layer convolutional neural network (CNN), for example, a three-layer CNN. The features of the first reference matrix are then decoded using a first matrix decoder, which is composed of a multi-layer transposed CNN, for example, a three-layer transposed CNN. The first reference matrix is reconstructed using feature mapping, and the reconstructed matrix is used as the first matrix.

[0061] The second reference matrix is encoded using a second matrix encoder to extract features of the second reference matrix. The second matrix encoder is composed of a multi-layer convolutional neural network (CNN), for example, a three-layer CNN. The second matrix encoder and the first matrix encoder may be the same or different. The features of the second reference matrix are then decoded using a second matrix decoder to achieve feature mapping. The second matrix encoder is composed of a multi-layer transposed CNN, for example, a three-layer transposed CNN. The second matrix decoder and the first matrix decoder may be the same or different. The second reference matrix is reconstructed through feature mapping, and the reconstructed matrix of the second reference matrix is used as the second matrix.

[0062] For example, the second matrix encoder is the same as the first matrix encoder, and the second matrix decoder is the same as the first matrix decoder. The matrix encoder is denoted as , the matrix decoder is denoted as , No. Time slot and The first reference matrix corresponding to the subcarrier is recorded as , No. Time slot and The second reference matrix corresponding to the subcarrier is recorded as , then the first matrix Expressed as: , the second matrix Expressed as: .

[0063] After obtaining the first matrix and the second matrix, the initial value of the iteration is generated by the following formula based on the time domain weight and the first matrix, as well as the frequency domain weight and the second matrix:

[0064] in, Indicates the target channel ( Time slot and subcarrier) corresponding to the initial value of the iteration, represents the matrix encoder, represents the matrix decoder, represents the second matrix, represents the first matrix, represents the frequency domain weight, represents the time domain weight.

[0065] Optionally, when the channel matrix of the first channel is used as the first reference matrix and the channel matrix of the second channel is used as the second reference matrix, the above formula for generating the initial value of the iteration is Can be replaced by Time slot and Channel matrix of the second channel corresponding to the subcarrier , Can be replaced by Time slot and Channel matrix of the first channel corresponding to the subcarrier .

[0066] Optionally, when the inverse matrix of the matrix to be processed of the first channel is used as the first reference matrix and the inverse matrix of the matrix to be processed of the second channel is used as the second reference matrix, the above formula for generating the initial value of the iteration is Can be replaced by Time slot and The inverse matrix of the matrix to be processed of the second channel corresponding to the subcarrier , Can be replaced by Time slot and The inverse matrix of the matrix to be processed of the first channel corresponding to the subcarrier .

[0067] In this embodiment, by reconstructing the first reference matrix and the second reference matrix, the hidden relationship in the reference matrix can be discovered, and the reconstructed matrix can better reflect the actual channel state, so that the generated iterative initial value is closer to the optimal solution.

[0068] In some embodiments, the preset inverse iteration model in step S105 is used to indicate the correlation between the iteration result of the current iteration and the iteration result of the next iteration, the iteration step size is determined according to the step size decision model, and the preset inverse iteration model of the next iteration operation is updated according to the iteration step size, including the following: Figure 3 The following steps S1051 to S1052 are shown.

[0069] Step S1051: Input the condition number of the matrix to be processed, the iteration residual and the number of current iterations into the step size decision model to obtain the iteration step size.

[0070] Step S1052: Determine the preset inverse iterative model corresponding to the next iteration according to the iteration step size, the iteration result of the current iteration and the matrix to be processed.

[0071] The condition number of the matrix to be processed is used to characterize the degree of singularity of the matrix. The larger the condition number, the closer the matrix is to singularity, that is, the higher the probability that the matrix is irreversible or the error of the inverse matrix is larger, and the iterative operation is more likely to oscillate. The iteration residual represents the difference between the result of the current iteration and the result of the previous iteration, or the Frobenius norm of the difference, which is used to characterize the amplitude of parameter changes between the two iterative operations. The larger the iteration residual, the more violent the oscillation of the previous iterative operation. The number of current iterations indicates the number of iterations of the current iterative operation. The step size decision model is a pre-trained machine learning model that determines the iteration step size based on the above three parameters. For example, it can be a multi-layer perceptron (MLP) model. The condition number of the matrix to be processed, the iteration residual and the current number of iterations are input into the step size decision model, and the iteration step size is output. The iteration step size can be expressed as:

[0072] in, Indicates the current The iteration step size of the iteration, Indicates the The iterative results of the iterations, Indicates the The iterative results of the iterations, Represents the matrix to be processed of the target channel The condition number of represents the iterative residual, that is and The Frobenius norm of the difference between represents the step size decision model.

[0073] The preset inverse iteration model is used to indicate the correlation between the iteration result of the current iteration and the iteration result of the next iteration. After obtaining the iteration result and iteration step of the current iteration, the preset inverse iteration model corresponding to the next iteration is expressed by the following formula:

[0074] in, Indicates the The iterative results of the iterations, Indicates the The iterative results of the iterations, Indicates the The iteration step size of the iteration, Represents the matrix to be processed of the target channel.

[0075] It should be noted that the first iteration is to iterate the initial value, and the iteration step length of the first iteration is 1, that is, ,in, is the initial value of iteration, , so the iterative step size of the second iteration is , the iterative result of the second iteration is .

[0076] In this embodiment, the iteration step size is determined based on the iteration residual and the step size decision model, and the iteration step size can be adjusted in time according to the iteration result obtained in each iteration, so that the iteration step size better matches the current iteration situation and effectively avoids oscillation.

[0077] In some embodiments, the preset iteration termination condition in the above step S106 is that the iteration residual is less than a first threshold and the convergence probability is greater than a second threshold.

[0078] For the Iterations, the iterative residual is recorded as , the convergence probability is recorded as The first threshold and the second threshold are empirical values. For example, the first threshold is , the second threshold is 0.99, if and , then terminate the iteration and The iterative result of the iteration is used as the inverse matrix of the matrix to be processed.

[0079] In this embodiment, the iteration is terminated when the iterative residual is less than the first threshold and the convergence probability is greater than the second threshold, so that the optimal solution can be obtained in time and the inversion efficiency is improved.

[0080] In some embodiments, determining the channel matrix of the target channel according to the inverse matrix of the matrix to be processed in step S107 includes: The inverse matrix of the matrix to be processed is input into a preset channel estimation model to obtain a channel matrix of the target channel, wherein the preset channel estimation model is one of a linear minimum mean square error channel estimation model and a least squares channel estimation model.

[0081] When the preset channel estimation model is a least squares channel estimation model, the least squares channel estimation model is expressed by the following formula. The inverse matrix of the matrix to be processed is input into the formula, and the channel matrix of the target channel is output:

[0082] in, Indicates the target channel ( Time slot and subcarriers), represents the inverse matrix of the target channel matrix to be processed, Represents the conjugate matrix of the pilot signal matrix when sending the pilot signal, with a dimension of , Represents the pilot signal matrix when receiving the pilot signal, with a dimension of , represents the number of receiving antennas, and ,but The dimension is , refer to the above step S102, ,therefore, The dimension is or .

[0083] When the preset channel estimation model is a linear minimum mean square error channel estimation model, the linear minimum mean square error channel estimation model is expressed by the following formula. The inverse matrix of the matrix to be processed is input into the formula, and the channel matrix of the target channel is output:

[0084] in, Indicates the target channel ( Time slot and subcarriers), Represents the covariance matrix of the target channel, with dimension , Represents the conjugate matrix of the pilot signal matrix when sending the pilot signal, with a dimension of , The inverse matrix of the target channel matrix to be processed is represented by step S102. , Represents the pilot signal matrix when receiving the pilot signal, with a dimension of , represents the number of receiving antennas, and ,therefore, The dimension is .

[0085] In this embodiment, channel estimation is performed using a channel estimation model according to the inverse matrix of the matrix to be processed, so that an accurate channel matrix can be obtained.

[0086] In some embodiments, the step of determining the channel matrix of the target channel further includes: Obtain scene information, and determine a corresponding preset channel estimation model based on the scene information.

[0087] Scenario information is used to characterize the wireless communication scenarios of electronic devices, such as the 5G communication scenarios provided by the 5G communication system, the 6G communication scenarios provided by the 6G communication system, and the WiFi communication scenarios. Different preset channel estimation models are selected in different communication scenarios, and the mapping relationship between the scenario information and the preset channel estimation model is pre-stored. For example, for 5G and 6G communication scenarios, when the electronic device initially accesses the network, the least squares channel estimation model is used. After successfully accessing the network and performing data transmission, the linear minimum mean square error channel estimation model is switched to. For WiFi communication scenarios, if the electronic device is a low-complexity device such as a mobile phone or tablet, the least squares channel estimation model is used. If the electronic device is a high-complexity device such as a high-performance router or wireless network card, the linear minimum mean square error channel estimation model is used. After obtaining the current scenario information, the preset channel estimation model corresponding to the current scenario information is searched from the pre-stored mapping relationship.

[0088] In this embodiment, selecting a preset channel estimation model according to an application scenario can make the preset channel estimation model more suitable for the current application scenario, further ensuring the accuracy of the channel matrix.

[0089] In an exemplary embodiment of the present invention, a channel estimation device is provided. Figure 4 is a block diagram of a channel estimation device according to an exemplary embodiment. Figure 4 As shown, the channel estimation device includes: The first determination module 401 is configured to determine the pilot signal matrix corresponding to the communication signal according to the number of transmitting antennas and the pilot length corresponding to the communication signal sent by the target channel, where the target channel is the channel corresponding to the first time slot and the first subcarrier.

[0090] The construction module 402 is configured to construct a to-be-processed matrix of the target channel according to the pilot signal matrix, the noise power of the communication signal and the number of receiving antennas.

[0091] The second determination module 403 is configured to determine the time domain weight and the frequency domain weight based on the correlation influence parameter and the weight decision model.

[0092] The generation module 404 is configured to generate an initial value for the iteration based on the time domain weight, the first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, the second reference matrix corresponding to the first time slot and the second subcarrier, wherein the second time slot is used to indicate the previous time slot of the first time slot in the continuously distributed time domain resources, and the second subcarrier is used to indicate the previous subcarrier of the first subcarrier in the continuously distributed frequency resources.

[0093] The iteration module 405 is configured to perform iterative operations on the processing matrix according to the initial value of the iteration and the preset inverse iteration model. During the iterative operation, the iteration step size is determined according to the step size decision model, and the preset inverse iteration model of the next iterative operation is updated according to the iteration step size.

[0094] The third determining module 406 is configured to obtain the inverse matrix of the matrix to be processed when a preset iteration termination condition is met.

[0095] The fourth determining module 407 is configured to determine the channel matrix of the target channel according to the inverse matrix of the matrix to be processed.

[0096] In an exemplary embodiment, the generation module 404 is further configured to: Obtain a channel matrix of a first channel and a channel matrix of a second channel, where the first channel is a channel corresponding to the second time slot and the first subcarrier, and the second channel is a channel corresponding to the first time slot and the second subcarrier. Use the channel matrix of the first channel as a first reference matrix, and use the channel matrix of the second channel as a second reference matrix.

[0097] In an exemplary embodiment, the second determining module 403 is further configured to: The moving speed of the electronic device, the correlation between the second time slot and the first time slot, and the correlation between the first subcarrier and the second subcarrier are input into the weight decision model to obtain the time domain weight and the frequency domain weight.

[0098] The weight decision model is obtained by training based on the reference moving speed of the electronic device, the correlation between reference time domains, the correlation between reference frequency domains, and the reference time domain weight and the reference frequency domain weight.

[0099] In an exemplary embodiment, the generation module 404 is further configured to: The first reference matrix is encoded, the features of the first reference matrix are extracted, and the features of the first reference matrix are decoded to obtain a first matrix reconstructed from the first reference matrix; the second reference matrix is encoded, the features of the second reference matrix are extracted, and the features of the second reference matrix are decoded to obtain a second matrix reconstructed from the second reference matrix; and an initial value for iteration is generated based on the time domain weights and the first matrix, as well as the frequency domain weights and the second matrix.

[0100] In an exemplary embodiment, the preset inverse iteration model is used to indicate the correlation relationship between the iteration result of the current iteration and the iteration result of the next iteration, and the iteration module 405 is further configured to: The condition number of the matrix to be processed, the iteration residual and the number of current iterations are input into the step size decision model to obtain the iteration step size; according to the iteration step size, the iteration result of the current iteration and the matrix to be processed, the preset inverse iteration model corresponding to the next iteration is determined.

[0101] In an exemplary embodiment, the preset iteration termination condition is that the iteration residual is smaller than a first threshold and the convergence probability is larger than a second threshold.

[0102] In an exemplary embodiment, the fourth determining module 407 is further configured to: The inverse matrix of the matrix to be processed is input into a preset channel estimation model to obtain a channel matrix of the target channel, wherein the preset channel estimation model is one of a linear minimum mean square error channel estimation model and a least squares channel estimation model.

[0103] In an exemplary embodiment, the fourth determining module 407 is further configured to: Obtain scene information, and determine a corresponding preset channel estimation model based on the scene information.

[0104] Regarding the apparatus in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0105] Figure 5 is a block diagram of an electronic device 500 according to an exemplary embodiment.

[0106] Reference Figure 5 , electronic device 500 may include one or more of the following components: a processing component 502 , a memory 504 , a power component 506 , a multimedia component 508 , an audio component 510 , an input / output (I / O) interface 512 , a sensor component 514 , and a communication component 516 .

[0107] The processing component 502 generally controls the overall operation of the electronic device 500, such as operations associated with display, phone calls, data communications, camera operation, and recording operations. The processing component 502 may include one or more processors 520 to execute instructions to perform all or part of the steps of the above-described method. In addition, the processing component 502 may include one or more modules to facilitate interaction between the processing component 502 and other components. For example, the processing component 502 may include a multimedia module to facilitate interaction between the multimedia component 508 and the processing component 502.

[0108] The memory 504 is configured to store various types of data to support operations on the electronic device 500. Examples of such data include instructions for any application or method operating on the electronic device 500, contact data, phone book data, messages, pictures, videos, etc. The memory 504 can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory, electrically erasable programmable read-only memory, erasable programmable read-only memory, programmable read-only memory, read-only memory, magnetic memory, flash memory, magnetic disk, or optical disk.

[0109] The power supply assembly 506 provides power to the various components of the electronic device 500. The power supply assembly 506 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the electronic device 500.

[0110] The multimedia component 508 includes a screen that provides an output interface between the electronic device 500 and the user. In some embodiments, the screen may include a liquid crystal display and a touch panel. If the screen includes a touch panel, the screen may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, slides, and gestures on the touch panel. The touch sensor can not only sense the boundaries of a touch or slide action, but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 508 includes a front camera and / or a rear camera. When the electronic device 500 is in an operating mode, such as a shooting mode or a video mode, the front camera and / or the rear camera can receive external multimedia data. Each front camera and rear camera can be a fixed optical lens system or have focal length and optical zoom capabilities.

[0111] The audio component 510 is configured to output and / or input audio signals. For example, the audio component 510 includes a microphone, and when the electronic device 500 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode, the microphone is configured to receive external audio signals. The received audio signals can be further stored in the memory 504 or transmitted via the communication component 516. In some embodiments, the audio component 510 also includes a speaker for outputting audio signals.

[0112] I / O interface 512 provides an interface between processing component 502 and peripheral interface modules, such as a keyboard, click wheel, buttons, etc. These buttons may include, but are not limited to, a home button, volume buttons, a start button, and a lock button.

[0113] The sensor assembly 514 includes one or more sensors for providing various status assessments of the electronic device 500. For example, the sensor assembly 514 can detect the open / closed state of the electronic device 500, the relative positioning of components, such as the display and keypad of the electronic device 500. The sensor assembly 514 can also detect changes in the position of the electronic device 500 or a component of the electronic device 500, the presence or absence of user contact with the electronic device 500, the orientation or acceleration / deceleration of the electronic device 500, and changes in the temperature of the electronic device 500. The sensor assembly 514 may include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 514 may also include an optical sensor, such as a CMOS (Complementary Metal Oxide Semiconductor) or CCD (Charge Coupled Device) image sensor, for use in imaging applications. In some embodiments, the sensor assembly 514 may also include an accelerometer, a gyroscope, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0114] The communication component 516 is configured to facilitate wired or wireless communication between the electronic device 500 and other devices. The electronic device 500 can access a wireless network based on a communication standard, such as WiFi, 2G or 3G, or a combination thereof. In an exemplary embodiment, the communication component 516 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 516 also includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on Radio Frequency Identification (RFID) technology, Infrared Data Association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0115] In an exemplary embodiment, the electronic device 500 can be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above method.

[0116] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 504 including instructions. The instructions may be executed by the processor 520 of the electronic device 500 to perform the above method. For example, the non-transitory computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, or the like.

[0117] A non-transitory computer-readable storage medium, when instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to perform a channel estimation method, where the channel estimation method includes any of the above methods.

[0118] A computer program product includes a computer program. When the computer program is executed by a processor, a channel estimation method is implemented. The channel estimation method includes any one of the above methods.

[0119] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein.

Claims

1. A channel estimation method, characterized in that: The method comprises: Determining a pilot signal matrix corresponding to the communication signal according to the number of transmitting antennas and a pilot length corresponding to a target channel sending a communication signal, wherein the target channel is a channel corresponding to the first time slot and the first subcarrier; Constructing a to-be-processed matrix of the target channel according to the pilot signal matrix, the noise power of the communication signal, and the number of receiving antennas; Determine the time domain weight and frequency domain weight based on the correlation impact parameters and weight decision model; generating an iteration initial value according to the time domain weight, a first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, and a second reference matrix corresponding to the first time slot and the second subcarrier, wherein the second time slot is used to indicate a time slot previous to the first time slot in continuously distributed time domain resources, and the second subcarrier is used to indicate a subcarrier previous to the first subcarrier in continuously distributed frequency resources; performing an iterative operation on the matrix to be processed according to the iterative initial value and the preset inverse iterative model, determining an iteration step size according to a step size decision model during the iterative operation, and updating the preset inverse iterative model for the next iterative operation according to the iteration step size; When a preset iteration termination condition is met, obtaining the inverse matrix of the matrix to be processed; The channel matrix of the target channel is determined according to the inverse matrix of the matrix to be processed.

2. The method according to claim 1, characterized in that The method further comprises: Obtain a channel matrix of a first channel and a channel matrix of a second channel, where the first channel is a channel corresponding to the second time slot and the first subcarrier, and the second channel is a channel corresponding to the first time slot and the second subcarrier; The channel matrix of the first channel is used as the first reference matrix, and the channel matrix of the second channel is used as the second reference matrix.

3. The method according to claim 1, characterized in that The determining of the time domain weight and the frequency domain weight based on the correlation influencing parameter and the weight decision model includes: Inputting the moving speed of the electronic device, the correlation between the second time slot and the first time slot, and the correlation between the first subcarrier and the second subcarrier into the weight decision model to obtain the time domain weight and the frequency domain weight; The weight decision model is obtained by training based on the reference moving speed of the electronic device, the correlation between reference time domains, the correlation between reference frequency domains, and the reference time domain weight and the reference frequency domain weight.

4. The method according to claim 1, wherein The generating an iteration initial value according to the time domain weight, the first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, the second reference matrix corresponding to the first time slot and the second subcarrier includes: Encoding the first reference matrix, extracting features of the first reference matrix, and then decoding the features of the first reference matrix to obtain a first matrix reconstructed from the first reference matrix; Encoding the second reference matrix, extracting features of the second reference matrix, and then decoding the features of the second reference matrix to obtain a second matrix reconstructed from the second reference matrix; The iteration initial value is generated according to the time domain weight and the first matrix, and the frequency domain weight and the second matrix.

5. The method according to claim 1, wherein The preset inverse iterative model is used to indicate the correlation between the iterative result of the current iteration and the iterative result of the next iteration, the iterative step size is determined according to the step size decision model, and the preset inverse iterative model for the next iterative operation is updated according to the iterative step size, including: Inputting the condition number of the matrix to be processed, the iteration residual and the number of current iterations into the step size decision model to obtain the iteration step size; A preset inverse iterative model corresponding to the next iteration is determined according to the iteration step size, the iteration result of the current iteration and the matrix to be processed.

6. The method according to claim 1, characterized in that The preset iteration termination condition is that the iteration residual is less than a first threshold and the convergence probability is greater than a second threshold.

7. The method according to claim 1, characterized in that The determining the channel matrix of the target channel according to the inverse matrix of the matrix to be processed includes: The inverse matrix of the matrix to be processed is input into a preset channel estimation model to obtain the channel matrix of the target channel, wherein the preset channel estimation model is one of a linear minimum mean square error channel estimation model and a least squares channel estimation model.

8. The method according to claim 7, characterized in that The method further comprises: Acquire scenario information, and determine the corresponding preset channel estimation model based on the scenario information.

9. A channel estimation device, characterized in that The device comprises: A first determining module is configured to determine a pilot signal matrix corresponding to the communication signal according to the number of transmitting antennas and the pilot length corresponding to the communication signal sent by the target channel, wherein the target channel is a channel corresponding to the first time slot and the first subcarrier; A construction module is configured to construct a to-be-processed matrix of the target channel according to the pilot signal matrix, the noise power of the communication signal and the number of receiving antennas; A second determination module is configured to determine the time domain weight and the frequency domain weight based on the correlation influence parameter and the weight decision model; a generating module configured to generate an iteration initial value according to the time domain weight, a first reference matrix corresponding to the second time slot and the first subcarrier, the frequency domain weight, and a second reference matrix corresponding to the first time slot and the second subcarrier, wherein the second time slot is used to indicate a time slot previous to the first time slot in continuously distributed time domain resources, and the second subcarrier is used to indicate a subcarrier previous to the first subcarrier in continuously distributed frequency resources; an iterative module configured to perform an iterative operation on the matrix to be processed according to the iteration initial value and a preset inverse iterative model, determine an iteration step size according to a step size decision model during the iterative operation, and update the preset inverse iterative model for the next iterative operation according to the iteration step size; A third determining module is configured to obtain an inverse matrix of the matrix to be processed when a preset iteration termination condition is met; The fourth determining module is configured to determine the channel matrix of the target channel according to the inverse matrix of the matrix to be processed.

10. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to execute the method according to any one of claims 1 to 8.

11. A non-transitory computer-readable storage medium, characterized in that When the instructions in the storage medium are executed by a processor of an electronic device, the electronic device is enabled to execute the method according to any one of claims 1 to 8.

12. A computer program product, characterized in that The invention comprises a computer program, which implements the method according to any one of claims 1 to 8 when the computer program is executed by a processor.

Images