A channel estimation method and device for large-scale user asynchronous access

Abstract

The present invention provides a channel estimation method and device for large-scale user asynchronous access, relating to the technical field of wireless communications. The method uses unlicensed random access to satellites to reduce access delay and signaling overhead. Taking into account the actual situation of user asynchronous access, a maximum delay for user access to the satellite is set to expand the pilot sequence, thereby constructing a satellite-to-ground uplink access model that is consistent with the actual access scenario. Furthermore, the target channel estimation model of the present invention is a VAMP-based model-data dual-driven network model. The VAMP-based model-driven network is used to perform channel estimation and perform a first denoising process on the channel estimation results, and the data-driven network is used to perform a second denoising process on the first denoising process results. Therefore, the method can achieve high-precision channel estimation at the satellite receiving end, alleviating the technical problem of low channel estimation accuracy in unlicensed asynchronous access scenarios that exists in existing channel estimation methods.

Description

Technical Field

[0001] The present invention relates to the technical field of wireless communications, and in particular to a channel estimation method and device for asynchronous access of large-scale users. Background Art

[0002] In response to the communication needs of large-scale users for high-reliability and low-latency access to satellites, the research on user access protocols for satellites has far-reaching research significance and high practical value. The traditional authorization-based random access process is complex. When a large number of users access the channel, collisions will occur, resulting in a high probability of access failure, increasing access delay and signaling overhead. In recent years, unauthorized random access technology has become the first choice for large-scale ground users to access satellites because users can directly send data after sending the preamble sequence without waiting for authorization from the base station. This can significantly reduce user access delay and signaling overhead. At the same time, considering that it is difficult for users to perfectly coordinate the transmission time of accessing the satellite, and due to the high-speed movement of the satellite, the arrival time of different users to the satellite is different, and users cannot be completely synchronized. Therefore, the above problems have put forward new requirements for high-precision estimation of channel state information on the satellite side.

[0003] Approximate message passing algorithms (AMPs) are commonly used in existing channel state estimation methods. However, these algorithms typically use independent and identically distributed (IID) sub-Gaussian pilot matrices to achieve accurate channel estimation. However, in practical communication systems, the orthogonality of the pilot matrices is lost due to asynchronous user access to satellites. Therefore, when the pilot matrix actually received by the satellite deviates slightly from the IID sub-Gaussian pilot matrix, the AMPs tend to diverge, significantly degrading channel estimation performance. In summary, existing channel estimation methods suffer from low channel estimation accuracy in unlicensed asynchronous access scenarios. Summary of the Invention

[0004] The object of the present invention is to provide a channel estimation method and apparatus for large-scale user asynchronous access, so as to alleviate the technical problem of low channel estimation accuracy in the existing channel estimation method in the unlicensed asynchronous access scenario.

[0005] In a first aspect, the present invention provides a channel estimation method for large-scale user asynchronous access, comprising: obtaining a received signal at a satellite receiving end, a maximum delay for users to access the satellite, a pilot matrix of all users, and a delay information matrix; wherein the pilot matrix is ​​composed of pilot sequences of all users; the delay information matrix is ​​used to characterize the actual delay of each user accessing the satellite; based on the pilot matrix, the delay information matrix, and the maximum delay, determining the actual delay spread pilot matrix of all users; performing format conversion on the received signal at the satellite receiving end and the actual delay spread pilot matrix to obtain a received signal in real form and an actual delay spread pilot matrix in real form; and utilizing the received signal to obtain a received signal in real form and an actual delay spread pilot matrix in real form. A target channel estimation model is used to process a received signal in real form and an actual delayed spread pilot matrix in real form to obtain a channel estimation result in real form; wherein the target channel estimation model is a model-data dual-driven network model based on the vectorized approximate message passing algorithm VAMP, the model-driven network based on VAMP is used to perform channel estimation based on its input data and perform a first denoising process on the channel estimation result, and the data-driven network is used to perform a second denoising process on the first denoising process result of the model-driven network based on VAMP; the real form of the channel estimation result is format converted to obtain a complex form of the channel estimation result.

[0006] Optionally, based on the pilot matrix, the delay information matrix and the maximum delay, the actual delay extended pilot matrix of all users is determined, including: based on the pilot matrix and the maximum delay, determining the target extended pilot matrix of all users under all optional delays; based on the target extended pilot matrix and the delay information matrix, determining the actual delay extended pilot matrix.

[0007] Optionally, it also includes: constructing a training data set; wherein the training data set includes multiple groups of training data, and each group of training data includes: actual delayed extended pilot matrix samples, effective channel vector samples and received signal samples of the satellite receiving end of all users; converting the format of the training data set to obtain a training data set in real form; using the initial channel estimation model to process the actual delayed extended pilot matrix samples and received signal samples in the target group training data to obtain corresponding real-form channel estimation results; wherein the target group training data represents any group of training data in the real-form training data set; calculating the loss function value based on the effective channel vector samples and the corresponding real-form channel estimation results in the target group training data; iteratively training the initial channel estimation model based on the loss function value until a preset iteration termination condition is reached to obtain a target channel estimation model.

[0008] Optionally, constructing a training data set includes: randomly generating multiple actual delay spread pilot matrix samples; simulating transmission of the multiple actual delay spread pilot matrix samples in satellite multipath fading channels with different signal-to-noise ratios to obtain corresponding received signal samples at multiple satellite receiving ends; wherein the effective channel vector sample of each satellite multipath fading channel is expressed as: ; represents the effective channel vector sample, represents the activity indicator matrix of all users, , Represents a user activity, Represents a user active, Represents a user Inactive, Indicates the number of users, represents the set of complex-valued channel responses of all users, user exist Time, frequency The complex-valued channel response under , represents the Rice factor, hour, Indicates the LoS channel response; hour, represents the NLoS channel response, Indicates the number of propagation paths; , Represents a user In the The complex gain on each path, Represents a user In the The Doppler shift on each path is Represents a user In the The transmission delay on the path, represents the antenna array response, Represents a user In the The azimuth of the arrival angle on the path, Represents a user In the The elevation angle of arrival on each path.

[0009] Optionally, the initial channel estimation model includes: an initial VAMP-based model-driven network and an initial data-driven network; the initial VAMP-based model-driven network includes: a linear estimator, a first decoupler, a nonlinear estimator and a second decoupler; the initial channel estimation model is used to process the actual delayed extended pilot matrix samples and received signal samples in the target group training data, including: using the linear estimator to perform channel estimation processing on the actual delayed extended pilot matrix samples and the received signal samples to obtain a first channel estimation result and a first average variance; the first average variance represents the average of the diagonal values ​​of the covariance of the first channel estimation result; using the first decoupler to process the first channel estimation result and The first mean variance is processed to obtain the mean and mean variance used by the nonlinear estimator to transmit messages; the nonlinear estimator performs channel estimation processing and a first denoising process on the mean and mean variance used to transmit messages to obtain a second channel estimation result and a second mean variance; the second mean variance represents the average of the covariance diagonal values ​​of the second channel estimation result; the data-driven network performs a second denoising process on the second channel estimation result to obtain a real-number channel estimation result; the real-number channel estimation result, the second mean variance, and the mean and mean variance used by the nonlinear estimator to transmit messages are processed using a second decoupler to update the mean and mean variance used by the linear estimator to transmit messages.

[0010] Optionally, the learning parameters of the linear estimator include: environmental noise covariance, a left singular matrix, a singular value matrix, and a right singular matrix obtained by performing singular value decomposition on an actual delay spread pilot matrix in real form.

[0011] Optionally, the initial data-driven network includes: a data preprocessing module, a convolution and pooling module, an upsampling and residual connection module, an SE attention block and a denoising output module connected in sequence; the data preprocessing module is used to adjust the input data of the data-driven network to a specified dimension; the convolution and pooling module includes: a first submodule, a second submodule and a third submodule; the first submodule has the same structure as the second submodule, and the first submodule includes: a convolution layer, an activation layer, a normalization layer and a pooling layer; the third submodule includes: a convolution layer, an activation layer and a normalization layer; the upsampling and residual connection module includes: a fourth submodule and a fifth submodule; the fourth submodule has the same structure as the fifth submodule, and the fourth submodule includes: an upsampling layer, a convolution layer, an activation layer, a normalization layer The residual connection layer in the fourth submodule is used to add the output result of the normalization layer in the fourth submodule and the output result of the second submodule; the residual connection layer in the fifth submodule is used to add the output result of the normalization layer in the fifth submodule and the output result of the data preprocessing module; the SE attention block includes: a global feature compression layer, a first fully connected layer, a second fully connected layer and a feature extraction layer; the feature extraction layer is used to multiply the output result of the second fully connected layer with the output result of the upsampling and residual connection module; the denoising output module includes: a convolution layer, a denoising layer and a soft threshold processing layer; the denoising layer is used to subtract the output result of the data preprocessing module from the output result of the convolution layer in the denoising output module.

[0012] In a second aspect, the present invention provides a channel estimation device for large-scale user asynchronous access, comprising: an acquisition module for acquiring a received signal at a satellite receiving end, a maximum delay for a user to access the satellite, a pilot matrix of all users, and a delay information matrix; wherein the pilot matrix is ​​composed of pilot sequences of all users; the delay information matrix is ​​used to characterize the actual delay of each user accessing the satellite; a determination module for determining an actual delay spread pilot matrix of all users based on the pilot matrix, the delay information matrix, and the maximum delay; and a first format conversion module for performing format conversion on the received signal at the satellite receiving end and the actual delay spread pilot matrix to obtain a received signal in real form and an actual delay spread pilot matrix in real form. ; A channel estimation module is used to use a target channel estimation model to process a received signal in real form and an actual delayed extended pilot matrix in real form to obtain a channel estimation result in real form; wherein the target channel estimation model is a model-data dual-driven network model based on the vectorized approximate message passing algorithm VAMP, the model-driven network based on VAMP is used to perform channel estimation based on its input data and perform a first denoising process on the channel estimation result, and the data-driven network is used to perform a second denoising process on the first denoising process result of the model-driven network based on VAMP; a second format conversion module is used to perform format conversion on the channel estimation result in real form to obtain a channel estimation result in complex form.

[0013] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the computer program, it implements the channel estimation method for large-scale user asynchronous access described in any one of the aforementioned embodiments.

[0014] In a fourth aspect, the present invention provides a computer-readable storage medium storing computer instructions, which, when executed by a processor, implement the channel estimation method for large-scale user asynchronous access described in any one of the aforementioned embodiments.

[0015] In response to the problem of decreased accuracy of traditional channel estimation methods when large-scale users access asynchronously, the present invention provides a channel estimation method for large-scale user asynchronous access. First, it is proposed to adopt unlicensed random access satellite technology to reduce access delay and signaling overhead. Taking into account the actual situation of user asynchronous access, the maximum delay of user access to the satellite is set to expand the pilot sequence, thereby constructing a satellite-to-ground uplink access model that is consistent with the actual access scenario. Furthermore, the target channel estimation model of the present invention is a model-data dual-driven network model based on VAMP. The model-driven network based on VAMP is used to perform channel estimation and perform a first denoising process on the channel estimation result, and the data-driven network is used to perform a second denoising process on the first denoising process result. Therefore, this method can achieve high-precision channel estimation at the satellite receiving end, alleviating the technical problem of low channel estimation accuracy in the existing channel estimation method in the unlicensed asynchronous access scenario. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1 A flow chart of a channel estimation method for large-scale user asynchronous access provided by an embodiment of the present invention;

[0018] Figure 2 A schematic diagram of determining a target extended pilot matrix for all users at all optional delays based on a maximum delay and pilot matrices of all users;

[0019] Figure 3 A schematic structural diagram of a channel estimation model provided by an embodiment of the present invention;

[0020] Figure 4 A network architecture diagram of a data-driven network provided by an embodiment of the present invention;

[0021] Figure 5 A functional module diagram of a channel estimation device for large-scale user asynchronous access provided by an embodiment of the present invention;

[0022] Figure 6 A schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0024] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0025] The following embodiments of the present invention are described in detail with reference to the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments may be combined with each other.

[0026] Example 1

[0027] Figure 1 A flow chart of a channel estimation method for large-scale user asynchronous access provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the method specifically includes the following steps:

[0028] Step S102: Obtain the received signal of the satellite receiving end, the maximum delay of the user accessing the satellite, the pilot matrix and the delay information matrix of all users.

[0029] The pilot matrix is ​​composed of pilot sequences of all users; the delay information matrix is ​​used to characterize the actual delay of each user accessing the satellite.

[0030] Taking into account the actual situation that different users have different arrival times when accessing the satellite and users cannot be completely synchronized, an embodiment of the present invention proposes to construct a satellite-to-ground uplink access model that can match the actual access scenario. Specifically, the maximum delay for users to access the satellite is pre-set. Then, after obtaining the pilot matrix of all users, the pilot matrix of all users is expanded. Then, combined with the delay information matrix of all users, the actual delay expansion pilot matrix of all users is constructed. Finally, combined with the received signal at the satellite receiving end, the process of channel estimation of the satellite multipath fading channel is carried out.

[0031] In the embodiment of the present invention, the satellite-to-ground uplink access model can be expressed as: ; Indicates the received signal at the satellite receiving end, represents the actual delay spread pilot matrix for all users, represents the effective channel vector (i.e., the object to be estimated in the embodiment of the present invention), represents the noise signal, , represents the target extended pilot matrix of all users under all optional delays determined based on the pilot matrix of all users and the maximum delay of users accessing the satellite, represents the pilot length, Indicates the maximum delay for users to access the satellite, Indicates the number of users, represents the delay information matrix of all users, , Represents a user The delayed information matrix, ,set up Represents a user The actual access delay, (i.e., shared optional delay), then when hour, ;when hour, That is to say, Only one element has a value of 1, and the rest of the elements have values ​​of 0, and The element with a value of 1 is used to represent the index of the user's actual access delay, that is, The value of the first element in the value is 1, which means the actual access delay of the user is The first of several optional delays.

[0032] Step S104: determining the actual delay spread pilot matrix of all users based on the pilot matrix, the delay information matrix and the maximum delay.

[0033] Once the maximum delay for a user to access the satellite is known, the pilot matrix for all users at all optional delays can be determined based on this maximum delay and the pilot matrices of all users. This is denoted as the target extended pilot matrix for all users at all optional delays. The target extended pilot matrix is ​​then multiplied by the delay information matrix for all users to determine the actual delay extended pilot matrix for all users.

[0034] Step S106 , performing format conversion on the received signal at the satellite receiving end and the actual delay spread pilot matrix to obtain a received signal in real form and an actual delay spread pilot matrix in real form.

[0035] The received signal at the satellite receiving end and the actual delay spread pilot matrix are both data in complex form. In order to reduce the complexity and computational complexity of the model, the embodiment of the present invention first converts the received signal at the satellite receiving end and the actual delay spread pilot matrix into a format, specifically from a complex form to a real form, before using the target channel estimation model for channel estimation. The format conversion result is then input into the target channel estimation model.

[0036] The above complex to real number format conversion can be expressed as: , ,in, express The real part of express The imaginary part of represents the received signal in real form, express The real part of express The imaginary part of represents the actual delay spread pilot matrix in real number form. In the embodiment of the present invention, , so its matrix dimension is the same as the pilot length , maximum delay and number of users related; , so its dimension is the same as the pilot length and maximum delay related.

[0037] Step S108: Processing the received signal in real form and the actual delay spread pilot matrix in real form by using the target channel estimation model to obtain a channel estimation result in real form.

[0038] In an embodiment of the present invention, an initial channel estimation model is pre-trained and tested using a large amount of sample data to obtain a target channel estimation model capable of outputting highly accurate channel estimation results. The target channel estimation model is a model-data dual-driven network model based on the Vectorized Approximate Message Passing (VAMP) algorithm. The VAMP-based model-driven network is used to perform channel estimation based on its input data and perform a first denoising process on the channel estimation results. The data-driven network is used to perform a second denoising process on the results of the first denoising process of the VAMP-based model-driven network.

[0039] That is to say, the embodiment of the present invention combines the deep learning method with the vectorized approximate message passing algorithm VAMP, expands the VAMP algorithm into a neural network to obtain a VAMP-based model-driven network, and then further combines the data-driven network to achieve multiple denoising after channel estimation, thereby obtaining a high-precision channel estimation result.

[0040] Step S110 , performing format conversion on the channel estimation result in real number form to obtain the channel estimation result in complex number form.

[0041] In the embodiment of the present invention, the input data and output data of the target channel estimation model are both in real number form. Therefore, after obtaining the channel estimation result in real number form, it is necessary to convert it into a complex number form with better readability. That is, the channel estimation result in real number form needs to be converted into a complex number form. Channel estimation results concatenated into complex form , the relationship between the two is .

[0042] To address the problem of decreased accuracy of traditional channel estimation methods when large-scale users access asynchronously, an embodiment of the present invention provides a channel estimation method for large-scale user asynchronous access. First, it is proposed to adopt unlicensed random access satellite technology to reduce access delay and signaling overhead. Taking into account the actual situation of user asynchronous access, the maximum delay for user access to the satellite is set to expand the pilot sequence, thereby constructing a satellite-to-ground uplink access model that is consistent with the actual access scenario. Furthermore, the target channel estimation model used in the embodiment of the present invention is a VAMP-based model-data dual-driven network model. The VAMP-based model-driven network is used to perform channel estimation and perform a first denoising process on the channel estimation result, and the data-driven network is used to perform a second denoising process on the first denoising process result. Therefore, this method can achieve high-precision channel estimation at the satellite receiving end, alleviating the technical problem of low channel estimation accuracy in unlicensed asynchronous access scenarios in existing channel estimation methods.

[0043] In an optional implementation manner, the above step S104, determining the actual delay spread pilot matrix of all users based on the pilot matrix, the delay information matrix and the maximum delay, specifically includes the following steps:

[0044] Step S1041: Determine a target extended pilot matrix for all users at all optional delays based on the pilot matrix and the maximum delay.

[0045] Step S1042: Determine the actual delayed extended pilot matrix based on the target extended pilot matrix and the delay information matrix.

[0046] Figure 2 Schematic diagram of determining the target extended pilot matrix of all users under all optional delays according to the maximum delay and the pilot matrix of all users, as shown in FIG. Figure 2 As shown, after the maximum delay of a user accessing the satellite is known, the extended pilot sequence for each user at each optional delay can be determined based on the maximum delay and the pilot sequence of each user, thereby obtaining the extended pilot matrix of each user at all optional delays. The extended pilot matrices of all users at all optional delays constitute the target extended pilot matrix.

[0047] Specifically, users The pilot sequence is expressed as , the pilot matrix of all users is , then the user In the delay The extended pilot sequence under is expressed as: , express Previously 0, express Afterwards 0, express Based on this, the user The extended pilot matrix at all optional delays is expressed as: , the target extended pilot matrix for all users at all optional delays is expressed as: .

[0048] According to the description above, we can know that when calculating , and obtain the delay information matrix of all users Afterwards, through the formula The actual delay spread pilot matrix of all users can be calculated .

[0049] In an optional implementation manner, the embodiment of the present invention further includes the following steps:

[0050] Step S201: construct a training data set; wherein the training data set includes multiple sets of training data, and each set of training data includes: actual delay spread pilot matrix samples, effective channel vector samples and received signal samples of all users.

[0051] Step S202 : converting the format of the training data set to obtain a training data set in real number form.

[0052] Step S203, using the initial channel estimation model to process the actual delayed spread pilot matrix samples and received signal samples in the target group training data to obtain corresponding real number channel estimation results; wherein the target group training data represents any set of training data in the real number form training data set.

[0053] Step S204 : Calculate the loss function value based on the valid channel vector samples in the target group training data and the corresponding real number channel estimation results.

[0054] Step S205 , iteratively training the initial channel estimation model based on the loss function value until a preset iteration termination condition is reached, thereby obtaining a target channel estimation model.

[0055] Specifically, each parameter in each set of training data should conform to the satellite-to-ground uplink access model constructed in the embodiment of the present invention. To ensure the generalization and universality of the model, after constructing a large number of actual delay spread pilot matrix samples for all users, different actual delay spread pilot matrix samples should be transmitted in satellite multipath fading channels with different signal-to-noise ratios, thereby obtaining a large number of received signal samples at the satellite receiver.

[0056] As described above, the input data for the channel estimation model are real-number samples of the actual delayed spread pilot matrix and received signal samples, and the output data are the corresponding real-number channel estimation results. Therefore, before using the initial channel estimation model for channel estimation, the complex training dataset should be converted to real-number form, and then the initial channel estimation model should be used to perform channel estimation on the sample data. After the model outputs the predicted results (real-number channel estimation results), the loss function value predicted by the model is calculated based on the labels of each set of training data (i.e., valid channel vector samples). The loss function value is then used to iteratively train the initial channel estimation model until the preset iteration termination condition is met.

[0057] The embodiments of the present invention do not impose specific limitations on the functional expression of the loss function; as long as the error between the real-number channel estimation result and the corresponding valid channel vector sample is positively correlated with the loss function value, it is sufficient. In the embodiments of the present invention, the preset iteration termination condition can be when the loss function value converges below a preset threshold or when a specified number of training times is reached. The user can set this condition based on actual needs.

[0058] In an optional implementation, the above step S201, constructing a training data set, specifically includes the following steps:

[0059] Step S2011: randomly generate a plurality of actual delay spread pilot matrix samples.

[0060] Specifically, referring to the method described above for determining the actual delay spread pilot matrix for all users based on the maximum delay of user access to the satellite, the pilot matrix of all users, and the delay information matrix, multiple actual delay spread pilot matrix samples can be generated by customizing the maximum delay, randomly Gaussian generating the pilot matrix of all users, and randomly generating the delay information matrix.

[0061] Step S2012: Simulate the transmission of multiple actual delay spread pilot matrix samples in satellite multipath fading channels with different signal-to-noise ratios to obtain corresponding received signal samples at multiple satellite receiving ends.

[0062] Among them, the effective channel vector sample of each satellite multipath fading channel is expressed as: ; represents the effective channel vector sample, , represents the activity indicator matrix of all users, , Represents a user activity, Represents a user active, Represents a user Inactive, Indicates the number of users, represents the set of complex-valued channel responses for all users.

[0063] In the embodiment of the present invention, it is assumed that most satellite communications are line-of-sight transmissions. is subject to the Rice factor The Rice fading distribution of exist Time, frequency The complex-valued channel response under , represents the Rice factor, hour, Indicates the LoS channel response; hour, represents the NLoS channel response, Indicates the number of propagation paths.

[0064] , Represents a user In the The complex gain on each path, Represents a user In the The Doppler shift on each path is Represents a user In the The transmission delay on the path, represents the antenna array response, Represents a user In the The azimuth of the arrival angle on the path, Represents a user In the The elevation angle of arrival on each path.

[0065] Antenna array response ,in, , represents the response of the antenna decomposition on the x-axis, , represents the response of the antenna decomposition on the y-axis, represents the number of satellite antennas along the x-axis, represents the number of satellite antennas along the y-axis, Represents the total number of antennas in the antenna array. If we only consider the satellite channel model of a single antenna and a single beam, we have , .

[0066] The embodiment of the present invention expands the Vector Approximate Message Passing (VAMP) algorithm into a neural network, uses the VAMP to perform singular value decomposition on a quasi-orthogonal pilot matrix (i.e., an actual delay spread pilot matrix in real form), iteratively learns the measurement noise variance by setting learnable parameters, optimizes the noise adjustment term of the linear estimator to adapt to the noise in the actual observation environment, and iteratively learns the key parameters of the nonlinear shrinkage function to optimize the denoising effect of the nonlinear estimator. Then, a neural network such as a CNN (i.e., a data-driven network) is introduced to further denoise the noisy data, thereby enhancing the adaptability of the channel estimation method to complex noise and ultimately improving the channel estimation accuracy at the satellite receiving end.

[0067] VAMP adapts to channel estimation under a wide range of random matrices by performing singular value decomposition on the known pilot matrix. The specific steps are based on the Bayesian formula. , introduced The variable to be estimated Split into and In one iteration, we first calculate message passing, and in Belief estimation based on the sum-product algorithm ,in, , They represent the mean and average variance used by the linear estimator to transmit messages in the kth iteration, is the noise covariance at the kth iteration. Then the variable node Pass the message to , Pass the message to , then in Estimate in, The prior is a Bernoulli-Gaussian distribution, , Represent the mean and average variance of the nonlinear estimator used to transmit messages in the kth iteration, and finally The message is passed back in reverse and the iteration is continued. Both represent Gaussian distribution, Represents the identity matrix.

[0068] In an optional embodiment, the initial channel estimation model includes: an initial VAMP-based model-driven network and an initial data-driven network; the initial VAMP-based model-driven network includes: a linear estimator, a first decoupler, a nonlinear estimator and a second decoupler.

[0069] Figure 3 A schematic diagram of the structure of a channel estimation model provided by an embodiment of the present invention is shown in FIG. Figure 3 As shown, the model-driven network mainly uses linear estimators and nonlinear estimators , the linear estimator depends on the parameters , the nonlinear estimator depends on the characteristics of the shrinkage function parameters In this embodiment of the present invention, the linear estimation stage Parameterized as ,in, , that is, to After the singular value decomposition process, is a left singular matrix, is the singular value matrix, is a right singular matrix, is the environmental noise covariance. Through multiple rounds of iterative learning and , optimize the noise adjustment term of the linear estimator to adapt to the global noise of the actual observation environment, and at the same time optimize the denoising effect of the nonlinear estimator on local noise.

[0070] Based on the above description, it can be seen that in an embodiment of the present invention, the learning parameters of the linear estimator include: environmental noise covariance, a left singular matrix, a singular value matrix and a right singular matrix obtained by performing singular value decomposition on the actual delayed spread pilot matrix in real form.

[0071] The embodiment of the present invention introduces a data driven network based on the VAMP-based model driven network, which trains its model parameters as the model driven network iteratively , so as to directly denoise the signal to be estimated. After the initial channel estimation model performs the kth round of training, the channel estimation result output by the linear estimator is expressed as: , the channel estimation result output by the nonlinear estimator is expressed as: , the channel estimation result output by the data-driven network is expressed as .

[0072] The output processing flow of the initial channel estimation model is described in detail below. In step S203, the initial channel estimation model is used to process the actual delay spread pilot matrix samples and received signal samples in the target group training data, specifically including the following steps:

[0073] Step S2031: Use a linear estimator to perform channel estimation processing on the actual delay spread pilot matrix samples and the received signal samples to obtain a first channel estimation result and a first average variance; the first average variance represents the average of the diagonal values ​​of the covariance of the first channel estimation result.

[0074] Specifically, when the initial channel estimation model is trained in the kth round, , that is, estimated The essence is Multiplying these two Gaussians gives us an approximation: , so the mean Depend on The mean and Calculated; average variance Depend on The average variance of and Calculated.

[0075] The model parameters of the linear estimator are Therefore, after inputting the actual delay spread pilot matrix samples and the received signal samples, the following calculations are specifically performed: , ;in, represents the first channel estimation result output by the linear estimator during the k-th round of training, represents the first average variance of the linear estimator output during the k-th round of training, represents the average variance of the linear estimator used to pass messages during the k-th round of training, Indicates the mean of the linear estimator used to pass messages during the k-th round of training. During the k-th round of training, , , , that is, express rank, Indicates the kth round of training Positive singular values ​​in .

[0076] Step S2032: Process the first channel estimation result and the first mean variance using the first decoupler to obtain the mean and mean variance used by the nonlinear estimator to transmit the message.

[0077] In the embodiment of the present invention, the first decoupler specifically performs the following calculations: , ;in, represents the mean value of the nonlinear estimator used to pass messages during the k-th round of training, represents the average variance of the message passed by the nonlinear estimator during the kth round of training.

[0078] In step S2033, the nonlinear estimator performs channel estimation processing and first denoising processing on the mean and average variance used to transmit the message to obtain a second channel estimation result and a second average variance; the second average variance represents the average of the covariance diagonal values ​​of the second channel estimation result.

[0079] Specifically, when the initial channel estimation model is trained in the kth round, ,estimate The estimation described above The specific operation performed by the nonlinear estimator depends on the shrinkage function it adopts. Users can choose it according to their actual needs. The shrinkage function of the nonlinear estimator is based on the model parameters of the kth round. Output the second channel estimation result and the second average variance, , represents the second channel estimation result output by the nonlinear estimator during the k-th round of training, It represents the second average variance of the nonlinear estimator output during the k-th round of training.

[0080] Step S2034: The data-driven network performs a second denoising process on the second channel estimation result to obtain a real-number channel estimation result. The data processing flow can be expressed as: , Represents the real-number channel estimation result output by the data-driven network during the k-th round of training.

[0081] Step S2035: Use the second decoupler to process the real channel estimation result, the second mean variance, and the mean and mean variance used by the nonlinear estimator to transmit messages, so as to update the mean and mean variance used by the linear estimator to transmit messages.

[0082] In the embodiment of the present invention, the second decoupler specifically performs the following calculations: , ;in, represents the mean value of the linear estimator used to pass messages during the k+1th round of training, represents the average variance of the linear estimator used to pass messages during the k+1th round of training.

[0083] Figure 4 A network architecture diagram of a data driven network provided by an embodiment of the present invention, a data driven network As the model drives the network iterative training and learning CNN convolution kernel, bias and other parameters, such as Figure 4 As shown in the figure, the initial data driven network includes: sequentially connected data preprocessing module, convolution and pooling module, upsampling and residual connection module, SE attention block and denoising output module. Based on the above network structure, the nonlinear estimator in the model driven network outputs the kth training Perform local denoising and finally output .

[0084] The data preprocessing module is used to adjust the input data of the data-driven network to the specified dimension.

[0085] Optionally, define and initialize the 6-layer convolution weights and biases, add the defined parameters to the parameter list cnn_paras, then define two maximum pooling layers, and then define and initialize the variables lam and tex for controlling the soft threshold shrinkage strength, also add them to the parameter list cnn_paras, and finally input the two-dimensional tensor , which is the signal that needs to be denoised, the input data is dimensionally expanded. Add another dimension after to align the convolution layer input dimension and output a four-dimensional tensor cnn_in.

[0086] The convolution and pooling module includes: a first submodule, a second submodule and a third submodule; the structure of the first submodule is the same as that of the second submodule, the first submodule includes: a convolution layer, an activation layer, a normalization layer and a pooling layer; the third submodule includes: a convolution layer, an activation layer and a normalization layer;

[0087] Optionally, the convolution and pooling module is divided into three submodules (i.e. Figure 4 The three layers marked in the figure are first convolved with cnn_w1 in the convolutional layer to extract local features. The leaky_relu activation function is then used to introduce nonlinearity. The output of the convolutional layer is then batch normalized to accelerate model training, resulting in output data bn1. The pooling layer downsamples data bn1 to reduce the dimensionality of the feature map while retaining important features. The data processing steps for the other two submodules are similar to further extract data features. The output of the convolution and pooling modules is data bn3.

[0088] The upsampling and residual connection module includes: a fourth submodule and a fifth submodule; the fourth submodule has the same structure as the fifth submodule, and the fourth submodule includes: an upsampling layer, a convolution layer, an activation layer, a normalization layer and a residual connection layer.

[0089] The residual connection layer in the fourth submodule is used to add the output result of the normalization layer in the fourth submodule and the output result of the second submodule.

[0090] The residual connection layer in the fifth submodule is used to add the output result of the normalization layer in the fifth submodule and the output result of the data preprocessing module.

[0091] Optionally, the upsampling and residual connection module is divided into two submodules. In the fourth submodule, the data is first upsampled using the nearest neighbor interpolation method to restore the dimension of the feature map. Then, in order to adjust the number of channels, the data is convolved, and the leaky_relu activation function is used to introduce nonlinearity. The output of the convolution layer is then batch normalized to accelerate the training process of the model and output data bn4. Then, in order to alleviate the vanishing gradient and extract data features, the residual connection is used to add the output results of bn4 and the second submodule in the convolution and pooling module to enhance the training effect of the model. In the fifth submodule, the data is also upsampled, convolved, activated and normalized to output data bn5. Then, a residual connection is used to add bn5 and the output result cnn_in of the data preprocessing module to alleviate the vanishing gradient and improve training stability.

[0092] The SE attention block includes: a global feature compression layer, a first fully connected layer, a second fully connected layer, and a feature extraction layer.

[0093] The feature extraction layer is used to multiply the output of the second fully connected layer with the output of the upsampling and residual connection module.

[0094] In this embodiment of the present invention, the SE attention block includes a global feature compression layer, two fully connected layers, and a feature extraction layer to dynamically adjust the channel attention weights of the data and enhance the model's ability to extract features from different channels. First, the global feature compression layer extracts global statistical information for each channel to eliminate interference from the spatial dimension. Then, the two fully connected layers generate channel attention weights. Finally, the feature extraction layer multiplies the original input feature map by the channel weights channel by channel to enhance the features of important channels.

[0095] The denoising output module includes: convolution layer, denoising layer and soft threshold processing layer.

[0096] The denoising layer is used to subtract the output result of the data preprocessing module from the output result of the convolution layer in the denoising output module.

[0097] The data processing flow of the denoising output module is as follows: first, the convolution layer in the denoising output module integrates all feature output data conv6, then the denoising layer subtracts the output result cnn_in of the data preprocessing module from the convolution result conv6 to obtain the denoised feature map, and then reconstructs the feature map into a two-dimensional tensor xhat, and performs soft threshold processing on xhat to calculate the gradient to remove the noise of the input two-dimensional tensor data. Noise, final function output In actual use, the number of network layers and the network parameters of each layer can be modified according to the actual situation to improve the generalization ability and robustness of the network.

[0098] Optionally, the network of the channel estimation model in the embodiment of the present invention is implemented by TensorFlow. In the kth training, the initial learning rate of the channel estimation model is 0.001, and the Adam optimizer is used for training to minimize the normalized mean square error (NMSE). In order to achieve accurate channel estimation, the present invention trains a 0-30dB channel estimation model with a step size of 5dB.

[0099] To verify the performance of the method provided by the embodiment of the present invention, the following experimental case is provided. For a data set generated based on key parameters such as 300 users, a user activity probability of 0.08, a pilot length of 80, a maximum number of delays of 10, and 1 satellite antenna, the model-data dual-drive channel estimation method provided by the embodiment of the present invention improves the NMSE performance by approximately 5.19 dB compared to the single-drive network algorithm at an SNR of 30 dB.

[0100] Comparing the embodiments of the present invention with methods that rely on a single data-driven or model-driven approach for channel estimation, the single-driven approach lacks targeted optimization and is difficult to take into account complex and changeable channel environments and massive amounts of noisy data. The embodiments of the present invention innovate based on the principles of traditional algorithms, breaking through the limitations of the single-driven channel estimation model and proposing a model-data dual-driven channel estimation method based on VAMP. This method utilizes VAMP to perform singular value decomposition on a quasi-orthogonal pilot matrix, and expands VAMP into a neural network to learn user access prior information. A denoising module is then introduced to perform learnable noise reduction on the noisy data, aiming to improve the accuracy of channel estimation at the satellite receiving end. In addition, the present invention is scalable and can be applied to channel estimation in scenarios such as satellite multi-antennas.

[0101] Example 2

[0102] An embodiment of the present invention also provides a channel estimation device for asynchronous access of large-scale users. The device is mainly used to execute the channel estimation method for asynchronous access of large-scale users provided in the above-mentioned embodiment 1. The channel estimation device for asynchronous access of large-scale users provided in an embodiment of the present invention is specifically introduced below.

[0103] Figure 5 A functional module diagram of a channel estimation device for large-scale user asynchronous access provided by an embodiment of the present invention, such as Figure 5 As shown, the device mainly includes: an acquisition module 10, a determination module 20, a first format conversion module 30, a channel estimation module 40, and a second format conversion module 50, wherein:

[0104] The acquisition module 10 is used to obtain the received signal of the satellite receiving end, the maximum delay of the user accessing the satellite, the pilot matrix of all users, and the delay information matrix; wherein the pilot matrix is ​​composed of the pilot sequences of all users; and the delay information matrix is ​​used to represent the actual delay of each user accessing the satellite.

[0105] The determination module 20 is configured to determine the actual delay spread pilot matrix of all users based on the pilot matrix, the delay information matrix and the maximum delay.

[0106] The first format conversion module 30 is used to perform format conversion on the received signal at the satellite receiving end and the actual delay spread pilot matrix to obtain the received signal in real form and the actual delay spread pilot matrix in real form.

[0107] The channel estimation module 40 is used to use the target channel estimation model to process the received signal in real form and the actual delayed spread pilot matrix in real form to obtain a channel estimation result in real form; wherein the target channel estimation model is a model-data dual-driven network model based on the vectorized approximate message passing algorithm VAMP, the model-driven network based on VAMP is used to perform channel estimation based on its input data and perform a first denoising process on the channel estimation result, and the data-driven network is used to perform a second denoising process on the first denoising process result of the model-driven network based on VAMP.

[0108] The second format conversion module 50 is configured to perform format conversion on the channel estimation result in real number form to obtain the channel estimation result in complex number form.

[0109] To address the problem of decreased accuracy of traditional channel estimation methods when large-scale users access asynchronously, an embodiment of the present invention provides a channel estimation device for large-scale user asynchronous access. First, it is proposed to adopt unlicensed random access satellite technology to reduce access delay and signaling overhead. Taking into account the actual situation of user asynchronous access, the maximum delay for user access to the satellite is set to expand the pilot sequence, thereby constructing a satellite-to-ground uplink access model that is consistent with the actual access scenario. Furthermore, the target channel estimation model of the present invention is a model-data dual-driven network model based on VAMP. The model-driven network based on VAMP is used to perform channel estimation and perform a first denoising process on the channel estimation result, and the data-driven network is used to perform a second denoising process on the first denoising process result. Therefore, this method can achieve high-precision channel estimation at the satellite receiving end, alleviating the technical problem of low channel estimation accuracy in unlicensed asynchronous access scenarios in existing channel estimation methods.

[0110] Optionally, the determining module 20 is specifically configured to:

[0111] Based on the pilot matrix and the maximum delay, the target extended pilot matrix for all users at all optional delays is determined.

[0112] An actual delay spread pilot matrix is ​​determined based on the target spread pilot matrix and the delay information matrix.

[0113] Optionally, the device further comprises:

[0114] The construction module is used to construct a training data set; wherein the training data set includes multiple groups of training data, and each group of training data includes: actual delay spread pilot matrix samples of all users, effective channel vector samples and received signal samples of the satellite receiving end.

[0115] The third format conversion module is used to convert the format of the training data set to obtain a training data set in real number form.

[0116] The processing module is used to use the initial channel estimation model to process the actual delayed spread pilot matrix samples and received signal samples in the target group training data to obtain corresponding real number channel estimation results; wherein the target group training data represents any group of training data in the real number form training data set.

[0117] The calculation module is used to calculate the loss function value based on the valid channel vector samples in the target group training data and the corresponding real number channel estimation results.

[0118] The training module is used to iteratively train the initial channel estimation model based on the loss function value until a preset iteration termination condition is reached to obtain a target channel estimation model.

[0119] Optionally, the building block is specifically configured to:

[0120] A plurality of actual delay spread pilot matrix samples are randomly generated.

[0121] A plurality of actual delay spread pilot matrix samples are simulated and transmitted in satellite multipath fading channels with different signal-to-noise ratios to obtain corresponding received signal samples at a plurality of satellite receiving ends.

[0122] Among them, the effective channel vector sample of each satellite multipath fading channel is expressed as: ; represents the effective channel vector sample, represents the activity indicator matrix of all users, , Represents a user activity, Represents a user active, Represents a user Inactive, Indicates the number of users, represents the set of complex-valued channel responses of all users, user exist Time, frequency The complex-valued channel response under , represents the Rice factor, hour, Indicates the LoS channel response; hour, represents the NLoS channel response, Indicates the number of propagation paths.

[0123] , Represents a user In the The complex gain on each path, Represents a user In the The Doppler shift on each path is Represents a user In the The transmission delay on the path, represents the antenna array response, Represents a user In the The azimuth of the arrival angle on the path, Represents a user In the The elevation angle of arrival on each path.

[0124] Optionally, the initial channel estimation model includes: an initial VAMP-based model-driven network and an initial data-driven network; the initial VAMP-based model-driven network includes: a linear estimator, a first decoupler, a nonlinear estimator and a second decoupler.

[0125] The processing module is specifically used to:

[0126] A linear estimator is used to perform channel estimation processing on actual delayed spread pilot matrix samples and received signal samples to obtain a first channel estimation result and a first average variance; the first average variance represents the average of the covariance diagonal values ​​of the first channel estimation result.

[0127] The first channel estimation result and the first mean variance are processed by a first decoupler to obtain a mean value and a mean variance used by a nonlinear estimator to transmit a message.

[0128] The nonlinear estimator performs channel estimation and first denoising on the mean and average variance used to transmit the message to obtain a second channel estimation result and a second average variance; the second average variance represents the average of the covariance diagonal values ​​of the second channel estimation result.

[0129] The data-driven network performs a second denoising process on the second channel estimation result to obtain a real number channel estimation result.

[0130] The real-number channel estimation result, the second mean variance, and the mean and mean variance used by the nonlinear estimator to transmit messages are processed by the second decoupler to update the mean and mean variance used by the linear estimator to transmit messages.

[0131] Optionally, the learning parameters of the linear estimator include: environmental noise covariance, a left singular matrix, a singular value matrix, and a right singular matrix obtained by performing singular value decomposition on an actual delay spread pilot matrix in real form.

[0132] Optionally, the initial data-driven network includes: a data preprocessing module, a convolution and pooling module, an upsampling and residual connection module, a SE attention block and a denoising output module connected in sequence.

[0133] The data preprocessing module is used to adjust the input data of the data-driven network to the specified dimension.

[0134] The convolution and pooling module includes: a first submodule, a second submodule and a third submodule; the structure of the first submodule is the same as that of the second submodule, the first submodule includes: a convolution layer, an activation layer, a normalization layer and a pooling layer; the third submodule includes: a convolution layer, an activation layer and a normalization layer.

[0135] The upsampling and residual connection module includes: a fourth submodule and a fifth submodule; the fourth submodule has the same structure as the fifth submodule, and the fourth submodule includes: an upsampling layer, a convolution layer, an activation layer, a normalization layer and a residual connection layer.

[0136] The residual connection layer in the fourth submodule is used to add the output result of the normalization layer in the fourth submodule and the output result of the second submodule.

[0137] The residual connection layer in the fifth submodule is used to add the output result of the normalization layer in the fifth submodule and the output result of the data preprocessing module.

[0138] The SE attention block includes: a global feature compression layer, a first fully connected layer, a second fully connected layer, and a feature extraction layer.

[0139] The feature extraction layer is used to multiply the output of the second fully connected layer with the output of the upsampling and residual connection module.

[0140] The denoising output module includes: convolution layer, denoising layer and soft threshold processing layer.

[0141] The denoising layer is used to subtract the output result of the data preprocessing module from the output result of the convolution layer in the denoising output module.

[0142] Example 3

[0143] See also Figure 6 An embodiment of the present invention provides an electronic device, which includes: a processor 60, a memory 61, a bus 62 and a communication interface 63, wherein the processor 60, the communication interface 63 and the memory 61 are connected via the bus 62; the processor 60 is used to execute an executable module stored in the memory 61, such as a computer program.

[0144] Memory 61 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk drive. Communication between the system network element and at least one other network element is achieved via at least one communication interface 63 (which may be wired or wireless), and may utilize the Internet, a wide area network, a local area network, a metropolitan area network, or the like.

[0145] The bus 62 may be an ISA bus, a PCI bus, or an EISA bus. The bus may be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, Figure 6 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0146] Among them, the memory 61 is used to store programs, and the processor 60 executes the program after receiving the execution instruction. The method executed by the device defined by the process disclosed in any embodiment of the above-mentioned embodiment of the present invention can be applied to the processor 60 or implemented by the processor 60.

[0147] The processor 60 may be an integrated circuit chip with signal processing capabilities. During implementation, the steps of the above method may be performed by hardware integrated logic circuits or software instructions within the processor 60. The processor 60 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It may implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of the present invention may be directly executed by a hardware decoding processor or by a combination of hardware and software modules within the decoding processor. The software modules may be located in storage media well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or the like. The storage medium is located in the memory 61 , and the processor 60 reads the information in the memory 61 and completes the steps of the above method in combination with its hardware.

[0148] An embodiment of the present invention provides a computer program product of a channel estimation method and apparatus for large-scale user asynchronous access, including a computer-readable storage medium storing a non-volatile program code executable by a processor. The program code includes instructions that can be used to execute the method described in the previous method embodiment. For specific implementation, please refer to the method embodiment and will not be repeated here.

[0149] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0150] If the functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include various media capable of storing program code, such as USB flash drives, mobile hard drives, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical disks.

[0151] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not require further definition or explanation in subsequent drawings.

[0152] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer" and the like indicate positions or locations based on the positions shown in the accompanying drawings, or the positions or locations in which the inventive product is typically placed when in use. These terms are intended solely to facilitate the description of the present invention and to simplify the description, and are not intended to indicate or imply that the devices or components referred to must have a specific orientation, be constructed, or operate in a specific orientation. Therefore, they should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third," etc., are used solely to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0153] Furthermore, terms such as "horizontal," "vertical," and "overhanging" do not necessarily imply that a component must be absolutely horizontal or overhanging, but rather that it can be slightly tilted. For example, "horizontal" simply means that its direction is more horizontal than "vertical," and does not mean that the structure must be completely horizontal, but rather that it can be slightly tilted.

[0154] In the description of the present invention, it should also be noted that, unless otherwise expressly specified or limited, the terms "disposed," "installed," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; they may refer to mechanical connections or electrical connections; they may refer to direct connections or indirect connections through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0155] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A channel estimation method for large-scale user asynchronous access, characterized in that: include: Obtaining a received signal from a satellite receiving end, a maximum delay for a user to access the satellite, a pilot matrix of all users, and a delay information matrix; wherein the pilot matrix is ​​composed of pilot sequences of all users; and the delay information matrix is ​​used to represent the actual delay for each user to access the satellite; determining actual delay spread pilot matrices for all users based on the pilot matrix, the delay information matrix, and the maximum delay; Performing format conversion on the received signal of the satellite receiving end and the actual delay spread pilot matrix to obtain a received signal in real number form and an actual delay spread pilot matrix in real number form; The target channel estimation model is used to process the received signal in real form and the actual delayed spread pilot matrix in real form to obtain a channel estimation result in real form; wherein the target channel estimation model is a model-data dual-driven network model based on the vectorized approximate message passing algorithm VAMP, the model-driven network based on VAMP is used to perform channel estimation based on its input data and perform a first denoising process on the channel estimation result, and the data-driven network is used to perform a second denoising process on the first denoising process result of the model-driven network based on VAMP; the target channel estimation model is obtained by training an initial channel estimation model The model obtained after training, the initial channel estimation model includes: an initial VAMP-based model-driven network and an initial data-driven network; the initial VAMP-based model-driven network includes: a linear estimator, a first decoupler, a nonlinear estimator and a second decoupler; the learning parameters of the linear estimator include: an environmental noise covariance, a left singular matrix, a singular value matrix and a right singular matrix obtained by performing singular value decomposition on an actual delayed spread pilot matrix in real form; the initial data-driven network includes: a data preprocessing module, a convolution and pooling module, an upsampling and residual connection module, an SE attention block and a denoising output module connected in sequence; Performing format conversion on the real number form of the channel estimation result to obtain a complex number form of the channel estimation result.

2. The channel estimation method for large-scale user asynchronous access according to claim 1, characterized in that: Determining actual delay spread pilot matrices for all users based on the pilot matrix, the delay information matrix, and the maximum delay, comprising: Determining, based on the pilot matrix and the maximum delay, a target extended pilot matrix for all users at all optional delays; The actual delay spread pilot matrix is ​​determined based on the target spread pilot matrix and the delay information matrix.

3. The channel estimation method for large-scale user asynchronous access according to claim 1, characterized in that: Also includes: Constructing a training data set; wherein the training data set includes multiple sets of training data, each set of training data includes: actual delay spread pilot matrix samples, effective channel vector samples and received signal samples of the satellite receiving end for all users; Converting the training data set into a real number format. Using the initial channel estimation model, actual delay spread pilot matrix samples and received signal samples in the target group training data are processed to obtain corresponding real number channel estimation results; wherein the target group training data represents any set of training data in the real number training data set; Calculating a loss function value based on valid channel vector samples in the target group training data and corresponding real-number channel estimation results; The initial channel estimation model is iteratively trained based on the loss function value until a preset iteration termination condition is reached to obtain the target channel estimation model.

4. The channel estimation method for large-scale user asynchronous access according to claim 3, characterized in that: Construct a training dataset, including: randomly generating a plurality of actual delay spread pilot matrix samples; Simulating transmission of the plurality of actual delay spread pilot matrix samples in satellite multipath fading channels with different signal-to-noise ratios to obtain corresponding received signal samples of a plurality of satellite receiving ends; Among them, the effective channel vector sample of each satellite multipath fading channel is expressed as: ; represents the effective channel vector sample, represents the activity indicator matrix of all users, , Represents a user activity, Represents a user active, Represents a user Inactive, Indicates the number of users, represents the set of complex-valued channel responses of all users, user exist Time, frequency The complex-valued channel response under , represents the Rice factor, hour, Indicates the LoS channel response; hour, represents the NLoS channel response, Indicates the number of propagation paths; , Represents a user In the The complex gain on each path, Represents a user In the The Doppler shift on each path is Represents a user In the The transmission delay on the path, represents the antenna array response, Represents a user In the The azimuth of the arrival angle on the path, Represents a user In the The elevation angle of arrival on each path.

5. The channel estimation method for large-scale user asynchronous access according to claim 3, characterized in that: The initial channel estimation model is used to process actual delay spread pilot matrix samples and received signal samples in the target group training data, including: Performing channel estimation processing on the actual delayed spread pilot matrix samples and the received signal samples using the linear estimator to obtain a first channel estimation result and a first average variance; the first average variance represents an average of diagonal values ​​of the covariance of the first channel estimation result; Processing the first channel estimation result and the first mean variance using the first decoupler to obtain a mean and a mean variance used by the nonlinear estimator to transmit messages; The nonlinear estimator performs channel estimation processing and a first denoising process on the mean and the average variance used to transmit the message, to obtain a second channel estimation result and a second average variance; the second average variance represents the average of the diagonal values ​​of the covariance of the second channel estimation result; The data-driven network performs a second denoising process on the second channel estimation result to obtain the real number channel estimation result; The real-number channel estimation result, the second mean variance, and the mean and mean variance of the message transmitted by the nonlinear estimator are processed by the second decoupler to update the mean and mean variance of the message transmitted by the linear estimator.

6. The channel estimation method for large-scale user asynchronous access according to claim 5, characterized in that: The data preprocessing module is used to adjust the input data of the data-driven network to a specified dimension; The convolution and pooling module includes: a first submodule, a second submodule and a third submodule; the first submodule has the same structure as the second submodule, the first submodule includes: a convolution layer, an activation layer, a normalization layer and a pooling layer; the third submodule includes: a convolution layer, an activation layer and a normalization layer; The upsampling and residual connection module includes: a fourth submodule and a fifth submodule; the fourth submodule has the same structure as the fifth submodule, and the fourth submodule includes: an upsampling layer, a convolution layer, an activation layer, a normalization layer and a residual connection layer; The residual connection layer in the fourth submodule is used to add the output result of the normalization layer in the fourth submodule and the output result of the second submodule; The residual connection layer in the fifth submodule is used to add the output result of the normalization layer in the fifth submodule and the output result of the data preprocessing module; The SE attention block includes: a global feature compression layer, a first fully connected layer, a second fully connected layer and a feature extraction layer; The feature extraction layer is used to multiply the output result of the second fully connected layer with the output result of the upsampling and residual connection module; The denoising output module includes: a convolution layer, a denoising layer and a soft threshold processing layer; The denoising layer is used to perform subtraction processing on the output result of the data preprocessing module and the output result of the convolution layer in the denoising output module.

7. A channel estimation device for large-scale user asynchronous access, characterized in that: include: An acquisition module is configured to acquire a received signal from a satellite receiving end, a maximum delay for a user to access the satellite, a pilot matrix of all users, and a delay information matrix; wherein the pilot matrix is ​​composed of pilot sequences of all users; and the delay information matrix is ​​used to characterize the actual delay for each user to access the satellite; a determining module, configured to determine an actual delay spread pilot matrix for all users based on the pilot matrix, the delay information matrix, and the maximum delay; a first format conversion module, configured to perform format conversion on the received signal of the satellite receiving end and the actual delay spread pilot matrix to obtain a received signal in real number form and an actual delay spread pilot matrix in real number form; The channel estimation module is used to process the received signal in real form and the actual delayed spread pilot matrix in real form using a target channel estimation model to obtain a channel estimation result in real form; wherein the target channel estimation model is a model-data dual-driven network model based on the vectorized approximate message passing algorithm VAMP, the model-driven network based on VAMP is used to perform channel estimation based on its input data and perform a first denoising process on the channel estimation result, and the data-driven network is used to perform a second denoising process on the first denoising process result of the model-driven network based on VAMP; the target channel estimation model is obtained by performing a first denoising process on the initial channel estimation model. The initial channel estimation model comprises: an initial VAMP-based model-driven network and an initial data-driven network; the initial VAMP-based model-driven network comprises: a linear estimator, a first decoupler, a nonlinear estimator, and a second decoupler; the learning parameters of the linear estimator comprise: an environmental noise covariance, a left singular matrix, a singular value matrix, and a right singular matrix obtained by performing singular value decomposition on an actual delayed spread pilot matrix in real form; the initial data-driven network comprises: a data preprocessing module, a convolution and pooling module, an upsampling and residual connection module, an SE attention block, and a denoising output module connected in sequence; The second format conversion module is configured to perform format conversion on the real number channel estimation result to obtain a complex number channel estimation result.

8. An electronic device comprising a memory and a processor, wherein the memory stores a computer program that can be run on the processor, wherein: When the processor executes the computer program, the channel estimation method for large-scale user asynchronous access according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and when the computer instructions are executed by a processor, the channel estimation method for large-scale user asynchronous access according to any one of claims 1 to 6 is implemented.

Images