Channel estimation method and apparatus, electronic device, and storage medium

Abstract

Embodiments of the present application provide a channel estimation method and device, electronic equipment and storage medium, and relate to the technical field of communication. The method comprises: establishing a current channel with a sending end to be communicated, and determining a target channel estimation model; wherein the current channel is a channel through which the receiving end and the sending end communicate in the current transmission; receiving a pilot signal through the current channel, and predicting a channel matrix corresponding to the current channel based on the received pilot signal through the target channel estimation model; the channel matrix is used to represent the communication state of the current channel. The embodiments of the present application do not limit the dimension of the pilot signal, that is, the low-dimensional pilot signal can be used for effective channel estimation in the dense antenna array scene of the receiving / transmitting end, improve the efficiency of channel estimation, and enable channel estimation within one channel coherence time, thereby improving the accuracy of channel estimation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of communication, in particular, the present application relates to a channel estimation method and device, electronic equipment and storage medium. BACKGROUND

[0002] Channel estimation refers to inferring and predicting the characteristics of a channel transmission path using known information. In the prior art, channel estimation is usually performed based on least squares (LS) and minimum mean squared error (MMSE).

[0003] Existing channel estimation methods require the dimension of the pilot signal to be no less than the number of antennas at the sending end. In a large-scale MIMO (Multiple-Input Multiple-Output) system for 5G and 6G, the number of antennas deployed at the sending end will increase dramatically, which cannot guarantee that the dimension of the pilot signal is no less than the number of antennas at the sending end, so that the existing channel estimation method cannot be applied to channel estimation in a dense antenna array scenario at the receiving / sending end. SUMMARY

[0004] The embodiments of the present application provide a channel estimation method, device, electronic equipment and storage medium, which can solve the problem that the existing channel estimation method cannot be applied to channel estimation in a dense antenna array scenario at the receiving / sending end.

[0005] The technical solution is as follows:

[0006] According to an aspect of the embodiments of the present application, a channel estimation method is provided, applied to a receiving end, the method comprising:

[0007] establishing a current channel with a sending end to be communicated, and determining a target channel estimation model; wherein the current channel is a channel for communication between the receiving end and the sending end in the current transmission;

[0008] receiving a pilot signal through the current channel, and predicting a channel matrix corresponding to the current channel based on the received pilot signal through the target channel estimation model; the channel matrix is used to represent the communication state of the current channel.

[0009] Optionally, the method further comprises:

[0010] receiving a latent variable through the current channel;

[0011] updating the received hidden variable based on the channel matrix through the target channel estimation model, and taking the updated hidden variable as a corresponding hidden variable of next transmission.

[0012] Optionally, if the receiving end is a base station and the sending end is a terminal, the determining of the target channel estimation model comprises:

[0013] obtaining a plurality of candidate channel estimation models;

[0014] receiving the antenna information and the bearer information corresponding to the terminal sent by the terminal; the bearer information is used to indicate a model size supported by the terminal;

[0015] determining a candidate channel estimation model matched with the terminal from the plurality of candidate channel estimation models based on the antenna information and the bearer information, and taking the determined candidate channel estimation model as the target channel estimation model.

[0016] Optionally, if the receiving end is a terminal and the sending end is a base station, the determining of the target channel estimation model comprises:

[0017] obtaining a preset channel estimation model;

[0018] sending the antenna information and the bearer information corresponding to the terminal to the base station, so that the base station determines a target channel estimation model matched with the terminal based on the antenna information and the bearer information; the bearer information is used to indicate a model size supported by the terminal;

[0019] receiving a target parameter of the target channel estimation model sent by the base station;

[0020] updating a preset parameter in the preset channel estimation model to the target parameter, and taking the updated preset channel estimation model as the target channel estimation model.

[0021] Optionally, the candidate channel estimation model is trained by a base station based on the following manner:

[0022] obtaining a training sample set; the training sample set comprises a plurality of sample pilot signals and a plurality of sample channel matrices corresponding to the sample pilot signals;

[0023] performing at least one training operation on an initial channel estimation model based on the training sample set until a preset training end condition is met, and taking the initial channel estimation model meeting the preset training end condition as a trained candidate channel estimation model.

[0024] Optionally, the initial channel estimation model comprises an initial encoder and an initial decoder.

[0025] the training operation comprises:

[0026] inputting the plurality of sample pilot signals and the plurality of sample channel matrices into an initial encoder in the initial channel estimation model to obtain a plurality of sample latent variables corresponding to the plurality of sample channel matrices respectively;

[0027] inputting the plurality of sample latent variables and the plurality of sample pilot signals into an initial decoder in the initial channel estimation model to obtain a plurality of predicted channel matrices corresponding thereto;

[0028] determining a loss function based on each sample channel matrix and each predicted channel matrix;

[0029] adjusting parameters of the initial channel estimation model based on the loss function, and taking the initial channel estimation model after the adjustment of the parameters as an initial channel estimation model corresponding to a next training operation.

[0030] Optionally, the determining of the loss function based on each sample channel matrix and each predicted channel matrix comprises:

[0031] determining a first loss function based on a difference between each sample channel matrix and each predicted channel matrix;

[0032] determining a second loss function based on a difference between each sample latent variable and each preset probability distribution;

[0033] determining the loss function based on the first loss function and the second loss function.

[0034] According to another aspect of the embodiments of the present application, a channel estimation device is provided, which is applied to a receiving end, and comprises:

[0035] a channel establishment module, configured to establish a current channel with a sending end to be communicated with, and determine a target channel estimation model; wherein the current channel is a channel used by the receiving end and the sending end to communicate in a current transmission;

[0036] a channel estimation module, configured to receive a pilot signal through the current channel, and predict a channel matrix corresponding to the current channel based on the received pilot signal through the target channel estimation model; the channel matrix is used to represent a communication state of the current channel.

[0037] According to another aspect of the embodiments of the present application, an electronic device is provided, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of any of the above channel estimation methods when executing the program.

[0038] According to still another aspect of the embodiments of the present application, a computer readable storage medium is provided, and the computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the steps of any of the above channel estimation methods.

[0039] The technical scheme provided by the embodiments of the present application has the beneficial effects that:

[0040] By determining the target channel estimation model, and by the target channel estimation model, based on the pilot information received by the receiving end, predicting the channel matrix of the current channel between the receiving end and the sending end, there is no limitation on the dimension of the pilot signal, i.e. the low-dimensional pilot signal can be used for effective channel estimation in the dense antenna array scene of the receiving / sending end.

[0041] Further, by using the low-dimensional pilot signal for channel estimation, the efficiency of data transmission can be improved, the calculation amount of channel estimation can be reduced, and the efficiency of channel estimation can be improved. The channel will change with external factors, and the channel correlation time refers to the maximum time difference range in which the channel remains constant. By improving the efficiency of channel estimation, channel estimation can be performed within one channel coherence time (i.e. in a state where the channel remains stable), thereby improving the accuracy of channel estimation. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical schemes in the embodiments of the present application, the drawings needed in the description of the embodiments of the present application will be briefly introduced.

[0043] Figure 1 A flowchart of a channel estimation method provided by the embodiments of the present application is shown in FIG. 2;

[0044] Figure 2 A flowchart of another channel estimation method provided by the embodiments of the present application is shown in FIG. 3;

[0045] Figure 3 A flowchart of a target channel estimation model determination method provided by the embodiments of the present application is shown in FIG. 4;

[0046] Figure 4 A structure diagram of an initial channel estimation model provided by the embodiments of the present application is shown in FIG. 5;

[0047] Figure 5 A flowchart of still another channel estimation method provided by the embodiments of the present application is shown in FIG. 6;

[0048] Figure 6 A structure diagram of a channel estimation device provided by the embodiments of the present application is shown in FIG. 7;

[0049] Figure 7A structural schematic diagram of an electronic device is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0050] Embodiments of the present application will be described below in conjunction with the accompanying drawings. It should be understood that the embodiments described below in conjunction with the accompanying drawings are exemplary descriptions of the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions of the embodiments of the present application.

[0051] Those skilled in the art can understand that the singular forms "a", "an" and "the" used herein can include plural forms unless specifically stated otherwise. It should be further understood that the terms "comprise" and "include" used in the embodiments of the present application mean that the corresponding features can be implemented as the presented features, information, data, steps, operations, elements and / or components, but do not exclude other features, information, data, steps, operations, elements, components and / or combinations thereof supported by the present technology. It should be understood that when we say that an element is "connected" or "coupled" to another element, the element can be directly connected or coupled to the other element, or can mean that the element and the other element are connected through an intermediate element. In addition, "connected" or "coupled" used herein can include wireless connection or wireless coupling. The term "and / or" used herein means that at least one of the items defined by the term, for example, "A and / or B" can be implemented as "A", or as "B", or as "A and B".

[0052] In order to make the purpose, technical solutions and advantages of the present application clearer, the embodiments of the present application will be described in further detail below in conjunction with the accompanying drawings.

[0053] With the development of communication technology, massive MIMO (Multiple-Input Multiple-Output) systems have been widely applied. MIMO is a multi-antenna technology that uses multiple transmit antennas and receive antennas to achieve higher data transmission rates and system performance.

[0054] In a MIMO system, due to the presence of multiple antennas, signals can reach the receive antennas through different paths, and will be affected by multipath fading, interference and noise during transmission, resulting in changes in channel transmission characteristics. Therefore, in order to correctly decode the received signal and improve the system performance, it is necessary to accurately understand the state of the channel.

[0055] Channel estimation refers to inferring and predicting the characteristics of the channel transmission path using known information.

[0056] It is assumed that the number of user side (User End, UE) transmitting units is N t, the number of base station side (Bas Station, BS) receiving units is N r , the dimension of the pilot signal is N p . Assuming that only one subcarrier path is analyzed, the user side transmits an initial pilot signal s, which is loaded onto each transmitting antenna array unit N t after the action of the beamforming matrix F, and the base station side receives the pilot signal Y after the channel matrix H. According to the widely used Saleh-Valenzuela channel estimation method, the simplified expression is:

[0057] Y = HFs + N (1)

[0058] From the above formula, it can be seen that the beamforming matrix the base station side shaping matrix and the user side shaping matrix are F UE , F BS , the channel matrix the received pilot signal N is a noise matrix sampled from a white Gaussian distribution The initial pilot signal s is set according to the communication standard, and the beamforming matrix F depends on the inherent structure of the transmitting end and is also a fixed value. Therefore, the core of the entire channel estimation is to solve the channel matrix H.

[0059] In the prior art, the channel matrix is usually solved based on least squares (Least Squares estimation, LS) and minimum mean squared error (Minimum Mean Squared Error, MMSE). The existing channel estimation method requires that the dimension of the pilot signal is not less than the number of antennas of the transmitting end, that is, N p ≥ N t In a large-scale MIMO system facing 5G and 6G, the number of antennas deployed by the transmitting end will increase sharply, and thus it is impossible to guarantee that the dimension of the pilot signal is not less than the number of antennas of the transmitting end.

[0060] In addition, even if the dimension of the pilot signal can be ensured to be not less than the number of antennas of the transmitting end, with the sharp increase in the number of antennas of the transmitting end, the dimension of the pilot signal will inevitably increase sharply, resulting in a significant increase in the data volume of the pilot signal. The increase in the data volume of the pilot signal not only causes the delay of data transmission, but also increases the calculation amount of channel estimation, making it difficult to perform channel estimation within a fixed channel correlation time. Therefore, the existing channel estimation method cannot be applied to channel estimation in a dense antenna array scenario of the receiving / transmitting end.

[0061] The channel estimation method, device, electronic equipment and storage medium provided by the present application aim to solve the above technical problems of the prior art.

[0062] The technical solutions of the embodiments of the present application and the technical effects brought by the technical solutions of the present application are described below by describing several exemplary embodiments. It should be noted that the following embodiments can be mutually referenced, borrowed or combined. The same terms, similar features and similar implementation steps in different embodiments are not described repeatedly.

[0063] Figure 1 A flowchart of a channel estimation method provided by an embodiment of the present application is shown in FIG. 1, which is applied to a receiving end and a sending end, and the method comprises the following steps. Figure 1

[0064] In step S110, a current channel between the receiving end and the sending end is established, and a target channel estimation model is determined.

[0065] Specifically, the channel estimation method provided by the embodiment of the present application is applied to a communication process between two communication devices, wherein one side of the communication device is taken as a receiving end, and the other side of the communication device is taken as a sending end. It should be noted that the receiving end and the sending end are used to distinguish the two sides of the communication device, and do not limit the functions of the communication device, that is, the receiving end can also send signals, and the sending end can also receive signals.

[0066] In the embodiment of the present application, the receiving end and the sending end can include a terminal and a base station, that is, the receiving end can be a base station and the sending end can be a terminal, or the receiving end can be a terminal and the sending end can be a base station.

[0067] The terminal can be a smart phone (such as an Android phone, an iOS phone, etc.), a tablet computer, a notebook computer, a digital broadcast receiver, a MID (Mobile Internet Device), a PDA (Personal Digital Assistant), a desktop computer, a smart home appliance, a vehicle-mounted terminal (such as a vehicle-mounted navigation terminal, a vehicle-mounted computer, etc.), a smart speaker, a smart watch, etc.

[0068] The receiving end can establish a current channel with the sending end, and the current channel can be a channel through which the receiving end and the sending end communicate in the current transmission. The current channel can be understood as a transmission path between the sending end and the receiving end in the current transmission process.

[0069] After establishing the current channel, the receiving end can determine a target channel estimation model, which can be used for channel estimation of the current channel. The target channel estimation model can be constructed based on a deep learning algorithm, for example, the target channel estimation model can be a deep generative model. The determination method of the target channel estimation model will be described in detail below. ​

[0070] In step S120, a pilot signal is received through the current channel, and a channel matrix corresponding to the current channel is predicted based on the received pilot signal through a target channel estimation model. The channel matrix is used to represent the communication state of the current channel.

[0071] Specifically, the pilot refers to a specific known signal used to estimate the channel characteristics in a communication system, which can be generated at the sending end by a modulator.

[0072] The sending end generates an initial pilot signal and sends the initial pilot signal to the receiving end. The initial pilot signal is affected by various factors such as obstacles and weather conditions during transmission in the current channel, so that the signal finally reaching the receiving end can be different from the original signal, i.e., the pilot signal received by the receiving end is different from the initial pilot signal sent by the sending end.

[0073] The receiving end receives the pilot signal through the current channel, wherein the pilot signal can be obtained by updating the initial pilot signal sent by the sending end through the transmission process of the current channel.

[0074] The receiving end can predict the channel matrix corresponding to the current channel based on the mapping relationship between the pilot signal and the channel matrix and the received pilot signal through the determined target channel estimation model. The channel matrix can be used to represent the communication state of the current channel.

[0075] The channel matrix can be a matrix used to describe the signal transmission characteristics between the transmitting antennas and the receiving antennas in a MIMO system. For example, for a MIMO system with n transmitting antennas and m receiving antennas, its channel matrix is a matrix H composed of m rows and n columns of elements, wherein the element h ij may represent the channel coefficient from the i-th transmitting antenna to the j-th receiving antenna.

[0076] The receiving end can obtain the transmission characteristics of the current channel according to the channel matrix corresponding to the current channel, thereby obtaining information such as signal distortion, attenuation, and multipath reflection caused by the transmission process of the current channel. Through the channel matrix, the receiving end can adjust the demodulation strategy and parameter settings according to the obtained channel state information to maximize the reliability of the data and the performance of the system.

[0077] The receiving end and the sending end can perform the above steps to estimate the channel of the current channel in each transmission process, thereby realizing real-time dynamic channel estimation. In addition, the receiving end in the embodiment of the application can include a base station or a terminal, that is, channel estimation can be performed on the base station side and the terminal side, and bidirectional channel estimation can also be realized.

[0078] In the embodiments of the present application, the target channel estimation model is determined, and the channel matrix between the receiving end and the sending end is predicted based on the pilot information received by the receiving end through the target channel estimation model. The channel estimation method provided in the embodiments of the present application does not need to be established on the condition that the dimension of the pilot signal is not less than the number of the antennas of the sending end, and the dimension of the pilot signal is not limited, that is, the low-dimensional pilot signal can be used for effective channel estimation in the dense antenna array scene of the receiving / sending end.

[0079] Further, the low-dimensional pilot signal is used for channel estimation, which can improve the efficiency of data transmission, reduce the calculation amount of channel estimation, and improve the efficiency of channel estimation. The channel changes with external factors, and the channel correlation time refers to the maximum time difference range in which the channel remains constant. By improving the efficiency of channel estimation, channel estimation can be performed within a channel coherence time (that is, in a stable state of the channel), thereby improving the accuracy of channel estimation.

[0080] As an optional embodiment, the method further comprises:

[0081] receiving the hidden variable through the current channel;

[0082] updating the received hidden variable based on the channel matrix through the target channel estimation model, and taking the updated hidden variable as the corresponding hidden variable of the next transmission.

[0083] Specifically, Figure 2 A flowchart of a channel estimation method provided in the embodiments of the present application is shown in FIG. 1. Figure 2 As shown in FIG. 1, in the current transmission process, the sending end can also send the hidden variable to the receiving end, and the receiving end can receive the hidden variable through the current channel. The target channel estimation model can optimize the prediction of the channel matrix based on the hidden variable.

[0084] It should be noted that the hidden variable does not change in the transmission process of the sending end and the receiving end. A specific data format or data identifier can be set for the hidden variable to indicate that the corresponding data of the receiving end does not change in the transmission process.

[0085] After the receiving end obtains the channel matrix based on the received pilot signal through the target channel estimation model, the receiving end can update the received hidden variable based on the channel matrix through the target channel estimation model, obtain the updated hidden variable, and take the updated hidden variable as the corresponding hidden variable of the next transmission. In the next transmission process, the receiving end sends the initial pilot signal corresponding to the next transmission and the updated hidden variable to the sending end.

[0086] Optionally, the target channel estimation model can include an encoder and a decoder. The channel matrix corresponding to the current channel can be obtained based on the received pilot signal and the latent variable through the decoder in the target channel estimation model; the latent variable can be updated based on the channel matrix output by the decoder and the pilot signal through the encoder in the target channel estimation model, to obtain the updated latent variable.

[0087] As an optional embodiment, in the method, if the receiving end is a base station and the sending end is a terminal, the target channel estimation model is determined by:

[0088] A plurality of candidate channel estimation models are obtained.

[0089] The terminal sends antenna information and bearer information corresponding to the terminal to the base station. The bearer information is used to indicate the model size that can be supported by the terminal.

[0090] Based on the antenna information and the bearer information, a candidate channel estimation model matched with the terminal is determined from the plurality of candidate channel estimation models, and the determined candidate channel estimation model is taken as the target channel estimation model.

[0091] Specifically, in the case where the receiving end is a base station and the sending end is a terminal, the method for determining the target channel estimation model includes:

[0092] The base station can obtain a plurality of candidate channel estimation models, wherein the plurality of candidate channel estimation models are pre-trained by the base station, and the training method of the candidate channel estimation model will be described in detail below.

[0093] The terminal sends antenna information and bearer information corresponding to the terminal to the base station. The bearer information is used to indicate the model size that can be supported by the terminal.

[0094] After receiving the antenna information and the bearer information, the base station can select a candidate channel estimation model matched with the terminal from the plurality of candidate channel estimation models according to a preset specification, as the target channel estimation model. The preset specification can include the relationship between the antenna information and the bearer information and the candidate channel estimation model.

[0095] In the embodiment of the present application, the base station receives the antenna information and the bearer information corresponding to the terminal sent by the terminal, and determines the candidate channel estimation model matched with the terminal from the plurality of candidate channel estimation models based on the antenna information and the bearer information, as the target channel estimation model. The target channel estimation model matched with the terminal is determined according to the actual configuration of the terminal, which improves the adaptability between the model and the terminal and is beneficial to improving the accuracy of channel estimation.

[0096] As an optional embodiment, in the method, if the receiving end is a terminal and the sending end is a base station, the determination of the target channel estimation model comprises:

[0097] obtaining a preset channel estimation model;

[0098] sending the antenna information and the bearer information corresponding to the terminal to the base station, so that the base station determines the target channel estimation model matched with the terminal based on the antenna information and the bearer information; the bearer information is used to indicate the model size supported by the terminal;

[0099] receiving the target parameter of the target channel estimation model sent by the base station;

[0100] updating the preset parameter in the preset channel estimation model to the target parameter, and taking the updated preset channel estimation model as the target channel estimation model.

[0101] Specifically, in the case where the receiving end is a terminal and the sending end is a base station, the determination method of the target channel estimation model comprises:

[0102] The terminal can obtain a preset channel estimation model deployed locally in advance, and the preset channel estimation model can be an initialized target channel estimation model.

[0103] The terminal determines the corresponding antenna information and bearer information according to the actual configuration, and sends the corresponding antenna information and bearer information to the base station. The base station can select the target channel estimation model matched with the terminal from the plurality of candidate channel estimation models based on the received antenna information and bearer information.

[0104] After the base station determines the target channel estimation model, the base station can obtain the target parameter of the target channel estimation model, wherein the target parameter can be the model parameter of the target channel estimation model, and send the target parameter to the terminal.

[0105] After the terminal receives the target parameter, the terminal can update the preset parameter in the preset channel estimation model to the target parameter, and then update the preset channel estimation model to the target channel estimation model, so that the base station and the terminal can deploy the same target channel estimation model, for subsequent implementation of bidirectional channel estimation.

[0106] In the embodiment of the present application, the terminal updates the preset parameters in the preset channel estimation model to the target parameters of the target channel estimation model sent by the base station, and takes the updated preset channel estimation model as the target channel estimation model. Through the task of model training performed by the base station, the computing resources and capabilities of the base station are fully utilized, and the terminal only needs to update the parameters in the preset channel estimation model to construct the target channel estimation model, which can not only avoid the problem of poor performance of the trained model due to the limited computing power of the terminal, ensure the accuracy of the target channel estimation model of the terminal, but also save the computing resources of the terminal, avoid the consumption of the terminal resources by model training, and enable the terminal to concentrate limited resources for channel estimation, further improving the accuracy of channel estimation.

[0107] As an optional embodiment, Figure 3 A flowchart of a target channel estimation model determination method provided by the embodiment of the present application is shown in Figure 3 The method comprises the following steps:

[0108] (1) The base station pre-trains a plurality of candidate channel estimation models;

[0109] (2) The terminal pages the base station according to the wireless handover specification standard, and reports the corresponding antenna information and bearer information to the base station;

[0110] (3) The base station determines the target channel estimation model from the plurality of candidate channel estimation models based on the received antenna information and bearer information;

[0111] (4) The base station sends the model information and bearer requirements of the target channel estimation model to the terminal;

[0112] (5) The terminal confirms whether it can bear according to the received model information and bearer requirements, and sends the confirmation information to the base station if it confirms that it can bear;

[0113] (6) After receiving the confirmation information, the base station sends the target parameters of the target channel estimation model to the terminal;

[0114] (7) The terminal updates the preset channel estimation model based on the received target parameters to obtain the target channel estimation model, and returns the update completion information to the base station, so as to complete the deployment of the target channel estimation model on both sides of the base station and the terminal, and complete the preparation work of channel estimation.

[0115] As an optional embodiment, in the method, the candidate channel estimation model is trained by the base station based on the following manner:

[0116] Obtain a training sample set; the training sample set comprises a plurality of sample pilot signals and a plurality of sample channel matrices corresponding thereto;

[0117] performing at least one training operation on the initial channel estimation model based on the training sample set until a preset training end condition is met, and taking the initial channel estimation model meeting the preset training end condition as a trained candidate channel estimation model.

[0118] Specifically, the base station can pre-train a plurality of candidate channel estimation models. In the model training stage, the base station can obtain a training sample set, wherein the training sample set can include a plurality of sample pilot signals and a plurality of sample channel matrices corresponding thereto. The training sample set can be historical measurement channel data information stored by the base station or simulation information.

[0119] The training sample set is input into the initial channel estimation model, and at least one training operation is performed on the initial channel estimation model based on the training sample set until a preset training end condition is met, wherein the training end condition can be convergence of a loss function, for example, the loss function is less than a set value or the loss function calculated for a continuous set number of times is less than a set value; or the training end condition can be that the number of training reaches a preset number.

[0120] The initial channel estimation model meeting the preset training end condition is taken as a candidate channel estimation model. By setting initial channel estimation models of different sizes and training each initial channel estimation model, a plurality of candidate channel estimation models can be obtained.

[0121] Optionally, the initial channel estimation model includes an initial encoder and an initial decoder. The initial channel estimation model can be constructed based on a CVAE (Conditional Variational Auto-Encoding) model.

[0122] The training operation includes:

[0123] The plurality of sample pilot signals and the plurality of sample channel matrices are input into the initial encoder in the initial channel estimation model to obtain a plurality of sample latent variables corresponding to the plurality of sample channel matrices respectively;

[0124] The plurality of sample latent variables and the plurality of sample pilot signals are input into the initial decoder in the initial channel estimation model to obtain a plurality of predicted channel matrices corresponding thereto;

[0125] Based on each sample channel matrix and each predicted channel matrix, a loss function is determined;

[0126] Based on the loss function, the parameters of the initial channel estimation model are adjusted, and the initial channel estimation model after adjusting the parameters is taken as the initial channel estimation model corresponding to the next training operation.

[0127] Specifically, multiple sample pilot signals and multiple sample channel matrices are input into the initial channel estimation model. For each sample pilot signal and its corresponding sample channel matrix, the encoder in the initial channel estimation model encodes the sample channel matrix based on the sample pilot signal to obtain the corresponding sample latent variables. The sample latent variables can be a low-dimensional feature representation of the sample channel matrix.

[0128] Multiple sample pilot signals and multiple sample latent variables are input into the initial channel estimation model. For each sample pilot signal and its corresponding sample latent variable, the decoder in the initial channel estimation model restores the sample channel matrix based on the sample pilot signal and sample latent variable to obtain the predicted sample channel matrix.

[0129] Optionally, after obtaining multiple latent variables, for each latent variable, the latent variable can be concatenated with the corresponding pilot signal, and the concatenated data can be input into the decoder in the initial channel estimation model.

[0130] Based on the sample channel matrix and the predicted channel matrix, the loss function of the initial channel estimation model can be determined.

[0131] The parameters of the initial channel estimation model corresponding to the current training operation can be adjusted based on the loss function, and the adjusted initial channel estimation model can then be used in the next training operation. By continuously executing the above training operation, the model training is constrained by the loss function, making the predicted channel matrix output by the initial channel estimation model increasingly closer to the sample channel matrix, until the preset training termination condition is met.

[0132] Figure 4 This is a schematic diagram of the structure of an initial channel estimation model provided in an embodiment of this application, as shown below. Figure 4 As shown, the initial channel estimation module includes an initial encoder and an initial decoder. The sample channel matrix H used for training can be simulated using the Matlab 5G toolkit. The receiver pilot signal Y can be determined based on specific parameters such as the millimeter-wave frequency band, multipath path, array structure parameters, coherence coefficient, and signal delay. After determining the sample channel matrix H and its corresponding pilot signal Y, the sample channel matrix H and its corresponding pilot signal Y can be normalized and input into the initial encoder. The initial encoder performs convolution and downsampling operations, and outputs a mixture Gaussian probability distribution with a mean vector of μ and a variance vector of σ. The resampling operation samples from a standard normal distribution to form a latent variable z. The distribution of the latent variable z can be expressed by the following formula:

[0133]

[0134] The output latent variable z and pilot signal Y are concatenated and input into the initial decoder. The decoder performs transpose convolution and upsampling operations to multiplex the channel matrix and obtain the predicted sample channel matrix H'.

[0135] As an optional embodiment, in this method, the loss function is determined based on each sample channel matrix and each prediction channel matrix, including:

[0136] The first loss function is determined based on the differences between each sample channel matrix and each predicted channel matrix;

[0137] The second loss function is determined based on the differences between the latent variables of each sample and each preset probability distribution.

[0138] The loss function is determined based on the first loss function and the second loss function.

[0139] Specifically, each sample channel matrix can be used as the true label, and a first loss function is determined based on the difference between each sample channel matrix and its corresponding predicted channel matrix. This first loss function measures the difference between the output generated by the initial channel estimation model and the original input.

[0140] The latent variables of a sample can be understood as a probability distribution. A second loss function can be determined based on the differences between each sample latent variable and its corresponding predefined probability distributions. The second loss function measures the difference between the latent distribution from the input data to the latent space and the predefined prior distribution.

[0141] Based on the first and second loss functions, the loss function for constraining model training is determined. Optionally, the weights corresponding to the first and second loss functions can be obtained, and the two losses can be weighted and summed to obtain the model's loss function.

[0142] Optionally, the specific formula for the loss function is as follows:

[0143]

[0144] In the formula, L represents the loss function, min is the minimization, x represents the true sample channel matrix H, x′ represents the predicted sample channel matrix, y represents the pilot signal Y, and z represents the latent variable.

[0145] The first term D in the loss function KL (q(z|x,y)‖p(z|y)) is the KL divergence measure between the encoded latent variable z and the prior probability distribution p(z|y).

[0146] In the formula, the second term α·E q(z|x,y)(logp(x'|z, y)), which is used to represent that the training target thereof makes the mean square error (MSE) of the sample generated by the decoder and the training sample as small as possible. Wherein, a is a hyperparameter, controls the overall generation ability and diversity, enhances the generalization ability and decoupling ability of the model hidden variable z to different pilot signals and channel matrices, and enables the model to learn the mapping generation relationship between the pilot signal Y and the channel matrix H through training.

[0147] Figure 5 A flowchart of a channel estimation method provided by an embodiment of the present application is shown in FIG. 1. The method includes two stages of channel establishment and normal transmission. In the channel establishment stage, the base station side deploys the trained CVAE model structure parameters to the terminal side UE; in the normal transmission stage, only the low-dimensional pilot signal s and the encoded hidden variable z are transmitted on both sides, so as to establish a duplex communication link within a coherent time. Figure 5

[0148] (1) Channel establishment stage: according to the historical measured channel data information or simulation information stored by the base station side BS, the model updating strategy is started regularly, and various types of CVAE model parameters are trained on the base station side. According to the corresponding specification, the BS side adopts several CVAE models with different model hyperparameter specifications and architecture schemes according to the capability and antenna situation of the terminal UE side.

[0149] The UE side pings to the BS side according to the wireless switching specification standard, and reports the UE local antenna information and model bearing capability specification information to the BS side. The BS side prepares the channel establishment and model transmission work according to the relevant information reported by the UE side, and sends the pre-transmission model information and related bearing requirements to the UE side, and waits for the UE side to confirm.

[0150] After the BS side accepts the confirmation information from the UE side, the corresponding CVAE model is selected, and the corresponding encoder and decoder parameters of the CVAE are transmitted to the UE side. After the UE side accepts the model, the original parameters of the UE side are updated, and the model deployment complete signal is sent to the BS side after the update is completed.

[0151] (2) Normal transmission stage: after the BS side accepts the confirmation information from the UE side, the data uplink and downlink normal transmission is started. Within a coherent communication time: the transmitting end only transmits the low-dimensional pilot signal s and the encoded hidden variable z in the transmission stage, and the receiving end restores the estimated channel matrix H according to the received pilot signal Y and the encoded hidden variable z, generates a highly directional beamforming according to the channel matrix, compensates for the high-frequency attenuation loss and other related functions, so as to establish a duplex communication link, and the receiving end updates and optimizes the encoded hidden variable z' for the initial hidden variable input of the next coherent communication time. decoder encoder

[0152] ​​​In the embodiments of the present application, relying on the conditional variational self-encoding as an efficient data compression representation and generation capability, only low-dimensional pilot signals s and encoded hidden variables z are transmitted on both sides, so that high-precision restoration of the channel matrix can be realized, and the pilot signal transmission load and channel estimation calculation amount can be greatly reduced. The method has the characteristics of high efficiency, light weight and strong applicability. Based on the model training, model deployment, channel establishment, efficient communication standard specification and application process of the variational self-encoding, the process can adapt to the bearing capacity and model requirements of both parties. There is no limitation on the pilot dimension, and the calculation complexity of channel estimation is independent of the pilot sequence dimension, so that the estimation load will not be overloaded due to the increase of the dimension. At the same time, without the assumption of array incoherence, the method can adapt to the channel estimation in the dense scene of receiving / transmitting array, and has generalization for different signal-to-noise ratios.

[0153] Figure 6 A structural schematic diagram of a channel estimation device provided by the embodiments of the present application is shown in FIG. 1. Taking the case that the device is applied to a receiving end, the device includes: Figure 6

[0154] A channel establishment module 210 is configured to establish a current channel with a sending end to be communicated and determine a target channel estimation model. The current channel is a channel on which the receiving end and the sending end communicate in a current transmission.

[0155] A channel estimation module 220 is configured to receive a pilot signal through the current channel, and predict a channel matrix corresponding to the current channel based on the received pilot signal through the target channel estimation model. The channel matrix is used to represent the communication state of the current channel.

[0156] As an optional embodiment, the device further includes a hidden variable optimization module configured to:

[0157] receive a hidden variable through the current channel;

[0158] update the received hidden variable based on the channel matrix through the target channel estimation model, and use the updated hidden variable as a hidden variable corresponding to the next transmission.

[0159] As an optional embodiment, if the receiving end is a base station and the sending end is a terminal, the channel establishment module is specifically configured to:

[0160] obtain a plurality of candidate channel estimation models;

[0161] receive antenna information and bearing information corresponding to the terminal sent by the terminal. The bearing information is used to represent the model size that can be supported by the terminal.

[0162] ​determine a candidate channel estimation model matched with the terminal from the plurality of candidate channel estimation models based on the antenna information and the bearer information, and take the determined candidate channel estimation model as the target channel estimation model.

[0163] As an optional embodiment, if the receiving end is a terminal and the sending end is a base station, the channel establishing module is specifically used for:

[0164] obtaining a preset channel estimation model;

[0165] sending antenna information and bearer information corresponding to the terminal to the base station, so that the base station determines a target channel estimation model matched with the terminal based on the antenna information and the bearer information; the bearer information is used to indicate a model size supported by the terminal;

[0166] receiving a target parameter of the target channel estimation model sent by the base station;

[0167] updating a preset parameter in the preset channel estimation model to the target parameter, and taking the updated preset channel estimation model as the target channel estimation model.

[0168] As an optional embodiment, the device further includes a training module, and the training module includes:

[0169] a sample obtaining submodule, used for obtaining a training sample set; the training sample set includes a plurality of sample pilot signals and a plurality of sample channel matrices corresponding to the sample pilot signals;

[0170] a training submodule, used for performing at least one training operation on an initial channel estimation model based on the training sample set until a preset training end condition is met, and taking the initial channel estimation model meeting the preset training end condition as a trained candidate channel estimation model.

[0171] As an optional embodiment, the initial channel estimation model includes an initial encoder and an initial decoder.

[0172] The training submodule includes:

[0173] an encoding unit, used for inputting the plurality of sample pilot signals and the plurality of sample channel matrices to the initial encoder in the initial channel estimation model, to obtain a plurality of sample latent variables corresponding to the plurality of sample channel matrices respectively;

[0174] a decoding unit, used for inputting the plurality of sample latent variables and the plurality of sample pilot signals to the initial decoder in the initial channel estimation model, to obtain a plurality of predicted channel matrices corresponding thereto;

[0175] a loss function determination unit, configured to determine a loss function based on the sample channel matrices and the predicted channel matrices;

[0176] a parameter optimization unit, configured to adjust parameters of the initial channel estimation model based on the loss function, and take the initial channel estimation model after the parameter adjustment as an initial channel estimation model corresponding to a next training operation.

[0177] As an optional embodiment, the loss function determination unit in the apparatus is specifically configured to:

[0178] determine a first loss function based on differences between the sample channel matrices and the predicted channel matrices;

[0179] determine a second loss function based on differences between the sample latent variables and the preset probability distributions;

[0180] determine the loss function based on the first loss function and the second loss function.

[0181] The apparatus provided in the embodiments of the present application can execute the method provided in the embodiments of the present application, and the implementation principles are similar. The actions performed by each module in the apparatus of the embodiments of the present application are corresponding to the steps in the method of the embodiments of the present application. The detailed function description of each module of the apparatus can be referred to the description in the corresponding method in the foregoing description, and will not be repeated here.

[0182] In the embodiments of the present application, an electronic device is provided, which includes a memory, a processor and a computer program stored in the memory. The processor executes the computer program to implement the steps of the channel estimation method. Compared with the related art, the electronic device can achieve the following effects: by determining a target channel estimation model, and by the target channel estimation model predicting a channel matrix of a current channel between a receiving end and a sending end based on pilot information received by the receiving end, there is no limitation on the dimension of the pilot signal, i.e., the low-dimensional pilot signal can be used for effective channel estimation in a dense antenna array scene of the receiving / sending end. Further, using the low-dimensional pilot signal for channel estimation can improve the efficiency of data transmission, reduce the calculation amount of channel estimation, and improve the efficiency of channel estimation. The channel will change with external factors. The channel correlation time refers to the maximum time difference range in which the channel remains constant. By improving the efficiency of channel estimation, channel estimation can be performed within a channel coherence time (i.e., in a state where the channel remains stable), thereby improving the accuracy of channel estimation.

[0183] In an optional embodiment, an electronic device is provided, as shown in Figure 7 Figure 7 ​The electronic device 4000 shown includes a processor 4001 and a memory 4003. The processor 4001 and the memory 4003 are connected, for example, via a bus 4002. Optionally, the electronic device 4000 can also include a transceiver 4004, which can be used for data interaction between the electronic device and other electronic devices, such as data transmission and / or data reception, etc. It should be noted that the transceiver 4004 is not limited to one in actual application, and the structure of the electronic device 4000 does not constitute a limitation on the embodiments of the present application.

[0184] The processor 4001 can be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array) or other programmable logic device, a transistor logic device, a hardware component or any combination thereof. It can implement or execute various exemplary logical blocks, modules and circuits described in combination with the disclosure. The processor 4001 can also be a combination of computing functions, such as one or more microprocessor combinations, combinations of DSP and microprocessor, etc.

[0185] The bus 4002 can include a path for transmitting information between the above-mentioned components. The bus 4002 can be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The bus 4002 can be divided into an address bus, a data bus, a control bus, etc. For the sake of convenience, Figure 7 Only one thick line is used in the middle, but it does not mean that there is only one bus or one type of bus.

[0186] The memory 4003 can be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, a magnetic disk storage or other magnetic storage devices, or any other medium capable of storing computer instructions and capable of being read by a computer, without limitation.

[0187] The memory 4003 is configured to store a computer program for implementing the embodiments of the present application, and the processor 4001 is configured to control the execution of the computer program stored in the memory 4003. The processor 4001 is configured to execute the computer program stored in the memory 4003 to implement the steps shown in the foregoing method embodiments.

[0188] The embodiments of the present application provide a computer readable storage medium, and the computer readable storage medium stores a computer program. When the computer program is executed by a processor, the steps and corresponding contents of the foregoing method embodiments can be implemented.

[0189] The terms "first", "second", "third", "fourth", "1", "2", and the like (if any) in the specification and claims of the present application and the above-described drawings are used to distinguish similar objects, and do not necessarily have to describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described.

[0190] It should be understood that, although the flowcharts of the embodiments of the present application indicate the respective operation steps by arrows, the implementation order of the steps is not limited to the order indicated by the arrows. Unless otherwise specified herein, in some implementation scenarios of the embodiments of the present application, the implementation steps in each flowchart can be executed in other orders as required. In addition, part or all of the steps in each flowchart can include a plurality of sub-steps or a plurality of stages based on the actual implementation scenario. Part or all of the sub-steps or stages can be executed at the same time, and each of the sub-steps or stages can also be executed at different times. In the scenario where the execution times are different, the execution order of the sub-steps or stages can be flexibly configured as required, and the embodiments of the present application do not limit this.

[0191] The above is only optional implementation of part of the implementation scenarios of the present application. It should be noted that, for those skilled in the art, other similar implementation means based on the technical idea of the present application without departing from the technical concept of the present application also belong to the protection scope of the embodiments of the present application.

Claims

1. A channel estimation method, characterized in that, Applied to the receiving end, including: Establish a current channel with the transmitting end to be communicated, and determine a target channel estimation model; wherein, the current channel is the channel through which the receiving end and the transmitting end communicate in the current transmission; Pilot signals are received through the current channel, and the target channel estimation model predicts the channel matrix corresponding to the current channel based on the received pilot signals; the channel matrix is ​​used to represent the communication state of the current channel. If the receiving end is a base station and the sending end is a terminal, the determination of the target channel estimation model includes: Obtain multiple candidate channel estimation models; The terminal receives antenna information and bearer information corresponding to the terminal; the bearer information is used to indicate the model scale that the terminal can support. Based on the antenna information and the bearer information, a candidate channel estimation model matching the terminal is determined from the plurality of candidate channel estimation models, and the determined candidate channel estimation model is used as the target channel estimation model.

2. The channel estimation method according to claim 1, characterized in that, The method further includes: Hidden variables are received through the current channel; The target channel estimation model updates the received latent variables based on the channel matrix, and uses the updated latent variables as the latent variables for the next transmission.

3. The channel estimation method according to claim 1, characterized in that, If the receiving end is a terminal and the sending end is a base station, the determination of the target channel estimation model includes: Obtain the preset channel estimation model; The antenna information and bearer information corresponding to the terminal are sent to the base station so that the base station can determine the target channel estimation model matching the terminal based on the antenna information and the bearer information; the bearer information is used to represent the model scale that the terminal can support; Receive the target parameters of the target channel estimation model sent by the base station; The preset parameters in the preset channel estimation model are updated to the target parameters, and the updated preset channel estimation model is used as the target channel estimation model.

4. The channel estimation method according to claim 1, characterized in that, The candidate channel estimation model is trained by the base station in the following manner: Obtain a training sample set; the training sample set includes multiple sample pilot signals and their corresponding multiple sample channel matrices; The initial channel estimation model is trained at least once based on the training sample set until a preset training termination condition is met. The initial channel estimation model that meets the preset training termination condition is then used as a trained candidate channel estimation model.

5. The channel estimation method according to claim 4, characterized in that, The initial channel estimation model includes an initial encoder and an initial decoder; The training operations include: Multiple sample pilot signals and multiple sample channel matrices are input into the initial encoder of the initial channel estimation model to obtain multiple sample latent variables corresponding to the multiple sample channel matrices respectively; Multiple sample latent variables and multiple sample pilot signals are input into the initial decoder in the initial channel estimation model to obtain multiple corresponding predicted channel matrices; The loss function is determined based on the sample channel matrix and the predicted channel matrix. The parameters of the initial channel estimation model are adjusted based on the loss function, and the adjusted initial channel estimation model is used as the initial channel estimation model for the next training operation.

6. The channel estimation method according to claim 5, characterized in that, The process of determining the loss function based on each sample channel matrix and each prediction channel matrix includes: The first loss function is determined based on the differences between each sample channel matrix and each predicted channel matrix; The second loss function is determined based on the differences between the latent variables of each sample and each preset probability distribution. The loss function is determined based on the first loss function and the second loss function.

7. A channel estimation device, characterized in that, Applied to the receiving end, including: The channel establishment module is used to establish the current channel with the transmitting end to be communicated and to determine the target channel estimation model; wherein, the current channel is the channel through which the receiving end and the transmitting end communicate in the current transmission; The channel estimation module is used to receive pilot signals through the current channel and predict the channel matrix corresponding to the current channel based on the received pilot signals using the target channel estimation model; the channel matrix is ​​used to represent the communication state of the current channel. If the receiving end is a base station and the sending end is a terminal, the channel establishment module, when determining the target channel estimation model, is used for: Obtain multiple candidate channel estimation models; The terminal receives antenna information and bearer information corresponding to the terminal; the bearer information is used to indicate the model scale that the terminal can support. Based on the antenna information and the bearer information, a candidate channel estimation model matching the terminal is determined from the plurality of candidate channel estimation models, and the determined candidate channel estimation model is used as the target channel estimation model.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images