Communication method and related device

Abstract

A communication method and a related device. In the method, when a network device determines third channel information of a higher dimension by means of a first neural network, input data of the first neural network may comprise first channel information of a lower dimension and Q pieces of second channel information, and the position of a time-domain resource carrying Q second reference signals is before the position of a time-domain resource carrying a first reference signal. Therefore, the first neural network can use a time-domain correlation between the Q pieces of second channel information and the first channel information, and can also use a space-domain correlation and / or a frequency-domain correlation between the first channel information and the third channel information, thereby improving the accuracy of obtaining by the neural network the channel information of the higher dimension on the basis of the channel information of the lower dimension.

Description

A communication method and related equipment

[0001] This application claims priority to the Chinese patent application filed with the State Intellectual Property Office on November 9, 2023, with application number 202311494481.1 and application name “A Communication Method and Related Equipment”, the entire contents of which are incorporated by reference into this application. Technical Field

[0002] The present application relates to the field of wireless technology, and in particular to a communication method and related equipment. Background Art

[0003] Multiple-input, multiple-output (MIMO) technology, a key wireless communication technology, can be used to meet high-speed transmission requirements. Through channel measurement (or channel estimation), network devices use the channel information obtained during the measurement process to calculate precoding information between the network device and the terminal device. This precoding information then enables MIMO communication between the network device and the terminal device.

[0004] Taking the downlink channel measurement process implemented by a network device based on a downlink reference signal as an example, the downlink reference signal sent by the network device may include a channel state information reference signal (CSI-RS). The network device may receive feedback of the CSI-RS, enabling the network device to obtain channel information based on the feedback of the CSI-RS. The CSI-RS is sent through a port of the network device, meaning that the CSI-RS overhead is related to the number of ports of the network device.

[0005] However, in MIMO systems, as frequency bands increase and the demand for high-speed communications increases, the number of ports on network devices may gradually increase, which will increase the overhead of reference signals (such as CSI-RS) used for channel measurement and occupy more transmission resources, thereby increasing the power consumption of terminal devices.

[0006] Summary of the Invention

[0007] The present application provides a communication method and related equipment for obtaining higher-dimensional channel information based on channel information of a lower channel dimension determined by a reference signal through neural network processing, which can reduce the reference signal overhead, thereby improving resource utilization and reducing the power consumption of terminal equipment.

[0008] In a first aspect, the present application provides a communication method, which is executed by a network device, or by a component (e.g., a processor, chip, or chip system) in the network device, or by a logic module or software that implements all or part of the network device's functions. In the first aspect and its possible implementations, the communication method is described as being executed by a network device. In this method, a network device sends a first reference signal; the network device receives first channel information, which is determined based on the first reference signal, and the channel dimension represented by the first channel information is [T1, R1, F1], where T1, R1, and F1 are all positive numbers; the network device processes the first channel information and Q second channel information based on a first neural network to obtain third channel information; wherein the Q second channel information are determined by Q second reference signals, where Q is a positive integer; the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, and the channel dimension represented by the third channel information is [T, R, F], where T, R, and F are all positive numbers; wherein the parameters in the channel dimension represented by the first channel information and the parameters in the channel dimension represented by the third channel information satisfy at least one of the following: T is greater than T1, R is greater than R1, and F is greater than F1.

[0009] Based on the above technical solution, after the network device sends the first reference signal, the network device can receive the first channel information determined based on the first reference signal. Thereafter, the network device can process the first channel information and Q second channel information based on the first neural network to obtain third channel information. Moreover, the channel dimension represented by the first channel information is [T1, R1, F1], and the channel dimension represented by the third channel information is [T, R, F], satisfying at least one of the following: T is greater than the T1, R is greater than R1, and F is greater than F1. In other words, the network device can obtain the third channel information of a higher channel dimension based on the first channel information of a lower channel dimension, and the first channel information of the lower channel dimension is determined by the first reference signal. Thus, the network device can obtain the channel information of a higher dimension through the processing of the neural network based on the channel information of the lower channel dimension determined by the reference signal, which can reduce the reference signal overhead, thereby improving resource utilization and reducing the power consumption of the terminal device.

[0010] In addition, in the process of the above-mentioned network device determining the third channel information of higher dimension through the first neural network, the input data of the first neural network may include the first channel information of lower dimension and Q second channel information, and the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, that is, the Q second channel information is historical channel information of the first channel information. Thus, the first neural network can utilize the time domain correlation between the Q second channel information and the first channel information, and can also utilize the spatial domain correlation and / or frequency domain correlation between the first channel information and the third channel information, thereby improving the accuracy of the neural network in obtaining the higher dimensional channel information based on the channel information of lower channel dimension.

[0011] It should be understood that the channel dimensions represented by the channel information may include multiple dimensions such as dimensions in the spatial domain, dimensions in the frequency domain, and dimensions in the time domain. Generally, the dimensions in the spatial domain may include the dimensions of the antenna port of the signal transmitting end (i.e., the transmitting antenna dimensions) and the dimensions of the antenna port of the signal receiving end (i.e., the receiving antenna dimensions). In the present application, the channel dimensions represented by the channel information include three channel dimensions as an example for explanation. One channel dimension is represented as T or T1 (or other parameters represented by T that may appear later, such as T2 / T3 / T4 / T5, etc.), another channel dimension is represented as R or R1 (or other parameters represented by R that may appear later, such as R2 / R3 / R4 / R5, etc.), and another channel dimension is represented as F or F1 (or other parameters represented by F that may appear later, such as F2 / F3 / F4 / F5, etc.).

[0012] In an implementation example, the spatial transmission process of the reference signal includes the process of the antenna port of the signal transmitting end transmitting the reference signal and the process of the antenna port of the signal receiving end receiving the reference signal. That is, the dimension of the channel information in the spatial domain can correspond to the transceiver port of the reference signal. In other words, the dimension of the reference signal in the spatial domain can be expressed as [T, R] in [T, R, F], where T (or T1, or other parameters represented by T that may appear later, such as T2 / T3 / T4 / T5, etc.) represents the antenna port of the signal transmitting end, and R (or R1, or other parameters represented by R that may appear later, such as R2 / R3 / R4 / R5, etc.) represents the antenna port of the signal receiving end.

[0013] In another implementation example, during the transmission and reception of a reference signal, the dimension of the reference signal in the frequency domain may correspond to the frequency domain resource carrying the reference signal. In other words, the dimension of the reference signal in the spatial domain may be represented as [F] in [T, R, F], where F (or F1, or other parameters denoted by F that may appear later, such as F2 / F3 / F4 / F5, etc.) represents the frequency domain resource carrying the reference signal (e.g., one or more frequency domain units contained in the frequency domain resource).

[0014] It should be noted that the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, which can be understood as at least one of the following: the Q second reference signals are sent by the network device before the network device sends the first reference signal, the Q second reference signals are received by the terminal device before the terminal device receives the first reference signal, and the index number of the time domain resource carrying the Q second reference signals is smaller than the index number of the time domain resource carrying the first reference signal.

[0015] Optionally, the index number of the time domain resource may be any one of the following: an index number of a system frame, an index number of a subframe, an index number of a time slot, an index number of an orthogonal frequency division multiplexing (OFDM) symbol, etc.

[0016] It should be noted that the first neural network is used to process lower-dimensional channel information (such as first channel information) to obtain higher-dimensional channel information (such as third channel information). Accordingly, the first neural network can be understood as a neural network that performs channel prediction.

[0017] In this application, channel prediction can be replaced by other terms, such as channel estimation, channel pre-estimation, channel simulation recovery, channel reconstruction, channel acquisition, channel inference, etc.

[0018] In this application, neural network can be replaced by other terms, such as AI network, machine learning network, AI learning network, AI neural network, etc.

[0019] In a possible implementation of the first aspect, the channel dimension represented by at least one of the Q second channel information is [T2, R2, F2], where T2, R2, and F2 are all positive numbers; the parameters in the channel dimension represented by the second channel information satisfy at least one of the following: T is greater than T2, R is greater than R2, and F is greater than F2.

[0020] Based on the above technical solution, the Q second channel information is determined using Q second reference signals, wherein the channel dimension represented by at least one of the Q second channel information satisfies at least one of the above conditions, i.e., the channel dimension represented by the at least one second channel information is lower than the channel dimension represented by the third channel information. Furthermore, the Q second channel information serves as one of the input data for the first neural network. Thus, this approach can reduce the reference signal overhead during the acquisition of input data for the neural network.

[0021] In a possible implementation of the first aspect, the parameters in the channel dimension represented by the first channel information and the parameters in the channel dimension represented by the second channel information satisfy at least one of the following: the values ​​of T1 and T2 are the same, the values ​​of R1 and R2 are the same, and the values ​​of F1 and F2 are the same.

[0022] Based on the above technical solution, the position of the time domain resource carrying Q second reference signals is located before the position of the time domain resource carrying the first reference signal, that is, the Q second channel information is the historical channel information of the first channel information, and the parameters in the two channel information satisfy at least one of the above items, which enables the network device to obtain the two channel information based on at least one identical dimension, thereby reducing the implementation complexity.

[0023] In a possible implementation of the first aspect, the first neural network includes a first sub-neural network and a second sub-neural network; the network device processes the first channel information and Q second channel information based on the first neural network to obtain the third channel information, including: the network device processes the first channel information based on the first sub-neural network to obtain fourth channel information; wherein the channel dimension represented by the fourth channel information is [T, R, F]; the network device processes the fourth channel information and the Q second channel information based on the second sub-neural network to obtain the third channel information.

[0024] Based on the above technical solution, when the network device obtains the third channel information based on the first neural network, the network device can process the first channel information based on the first sub-neural network to obtain the fourth channel information, and then process the fourth channel information and the Q second channel information based on the second sub-neural network to obtain the third channel information. Thus, the network device can obtain the fourth channel information of higher dimension by processing the first sub-neural network based on the lower dimensional first channel information obtained from the reference signal, and then further process the fourth channel information and the Q second channel information obtained historically based on the second sub-neural network to obtain the third channel information.

[0025] In a possible implementation of the first aspect, after the network device processes the first channel information and Q second channel information based on the first neural network to obtain third channel information, the method further includes: the network device processes the third channel information based on the second neural network to obtain fifth channel information, and the channel dimension represented by the fifth channel information is [T, R, F].

[0026] Based on the above technical solution, after the network device obtains higher-dimensional third channel information based on the first neural network, the network device can also process the third channel information based on the second neural network to obtain fifth channel information of the same channel dimension. Thus, by optimizing the second neural network, the accuracy of the higher-dimensional channel information obtained by the neural network can be further improved.

[0027] It can be understood that the second neural network is used to process the input channel information (e.g., the third channel information) to obtain output channel information (e.g., the fifth channel information). Compared to the input channel information, the output channel information is closer to the channel information of the transmission channel in the actual space. In other words, the second neural network can be a neural network for optimizing the channel information.

[0028] In a possible implementation of the first aspect, the deviation between the fifth channel information and the sixth channel information is less than or equal to the deviation between the third channel information and the sixth channel information, the channel dimension represented by the sixth channel information is [T, R, F], and the sixth channel information is obtained by measuring the reference signal.

[0029] Based on the above technical solution, the sixth channel information is obtained by measuring a reference signal, and the third and fifth channel information are obtained based on a second neural network. Therefore, compared with the third and fifth channel information, the sixth channel information is closer to the channel information of the transmission channel in actual space. In the above technical solution, the second neural network is used to optimize the third channel information to obtain the fifth channel information, so that the deviation between the fifth channel information and the sixth channel information is less than or equal to the deviation between the third and sixth channel information. That is, compared with the third channel information, the fifth channel information is closer to the channel information of the transmission channel in actual space.

[0030] Optionally, the deviation between two channel information (e.g., the fifth channel information and the sixth channel information, the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information described later, the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information, etc.) can be reflected by a mathematical operation result of the two channel information. For example, the deviation can be a normalized mean square error (NMSE), a mean square error (MSE), etc.

[0031] In a possible implementation manner of the first aspect, the method also includes: the network device sends a third reference signal; the network device receives channel information corresponding to the third reference signal, the channel dimension represented by the channel information corresponding to the third reference signal is [T3, R3, F3], and T3, R3, and F3 are all positive numbers; the channel dimension represented by the channel information corresponding to the third reference signal satisfies at least one of the following: T is greater than T3, R is greater than R3, and F is greater than F3; the network device processes the channel information corresponding to the second reference signal, part or all of the Q second channel information, and the third channel information based on the first neural network to obtain seventh channel information; wherein, the channel dimension represented by the seventh channel information is [T, R, F].

[0032] Based on the above technical solution, after the network device obtains the third channel information based on the first neural network, the network device can also send a third reference signal and receive the channel information corresponding to the third reference signal. Subsequently, the network device can obtain the seventh channel information based on the channel information corresponding to the second reference signal. The channel dimension represented by the seventh channel information is [T, R, F], and the channel dimension represented by the channel information corresponding to the third reference signal satisfies at least one of the following: T is greater than T3, R is greater than R3, and F is greater than F3. Thus, the network device can iteratively run the first neural network, based on multiple reference signal transmissions, to obtain higher-dimensional channel information multiple times based on lower-dimensional channel information.

[0033] In a possible implementation of the first aspect, the channel dimension represented by the first channel information and the channel dimension represented by the channel information corresponding to the third reference signal satisfy at least one of the following: the values ​​of T1 and T3 are the same, the values ​​of R1 and R3 are the same, and the values ​​of F1 and F3 are the same.

[0034] Based on the above technical solution, during the process of sending multiple reference signals, the channel dimensions of different channel information obtained based on different reference signals can satisfy at least one of the above items, that is, the different channel information contains at least one identical dimension, which can reduce the implementation complexity.

[0035] In a possible implementation of the first aspect, the method also includes: the network device sends a fourth reference signal; the network device receives first information, the first information being used to indicate a deviation between channel information corresponding to the fourth reference signal and eighth channel information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on a third neural network, the P ninth channel information being determined by P reference signals, the position of the time domain resources carrying the P reference signals being located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

[0036] Based on the above technical solution, the network device may further transmit a fourth reference signal, and the eighth channel information is channel information obtained by processing the channel information corresponding to the fourth reference signal by the receiver (i.e., the terminal device) of the fourth reference signal based on the third neural network. Furthermore, the network device may further receive first information indicating a deviation between the channel information corresponding to the fourth reference signal and the eighth channel information, so that the network device can determine the performance of the channel prediction of the third neural network based on the first information.

[0037] In addition, compared to the method in which the network device receives the channel information corresponding to the fourth reference signal and processes the channel information corresponding to the fourth reference signal based on the first neural network to determine the performance of the channel prediction of the first neural network, in the above technical solution, the network device can determine the performance of the channel prediction of the neural network through the first information. Since the transmission overhead of the first information is less than the transmission overhead of the channel information corresponding to the fourth reference signal, the transmission overhead can be reduced.

[0038] It should be understood that the first information is used to indicate the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information, where the eighth channel information is channel information obtained through processing based on the third neural network. Accordingly, the eighth channel information can be understood as the result of channel prediction performed by the third neural network, and the channel information corresponding to the fourth reference signal is channel information of the transmission channel in actual space. Therefore, the deviation indicated by the first information can be understood as the first information indicating the performance test result of the third neural network performing channel prediction.

[0039] Optionally, the network structure and / or network parameters of the first neural network deployed on the network device and the third neural network deployed on the terminal device are the same.

[0040] In a possible implementation of the first aspect, the channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], where T4, R4, and F4 are all positive numbers; the channel dimension represented by the channel information corresponding to the fourth reference signal and the channel dimension represented by the channel information corresponding to the first reference signal (i.e., the first channel information) satisfy at least one of the following: T is greater than or equal to T4 and T4 is greater than T1, R is greater than or equal to R4 and R4 is greater than R1, and F is greater than or equal to F4 and F4 is greater than F1.

[0041] Based on the above technical solution, the channel dimension represented by the channel information corresponding to the fourth reference signal and the channel dimension represented by the channel information corresponding to the first reference signal (i.e., the first channel information) satisfy at least one of the above items, that is, the channel information corresponding to the fourth reference signal is channel information of higher dimension than the channel information corresponding to the first reference signal (i.e., the fourth reference signal is a reference signal of higher density than the first reference signal), so as to enable the terminal device to detect the channel prediction performance of the third neural network through the reference signal of higher density.

[0042] In addition, the channel information corresponding to the fourth reference signal is channel information of lower dimension than the channel information corresponding to the eighth reference signal, which can reduce the overhead in the detection process of the channel prediction performance of the neural network.

[0043] In a possible implementation manner of the first aspect, the method further includes: the network device sending second information, where the second information is used to indicate whether to adjust the input parameters of the third neural network; wherein the second information is determined based on the first information.

[0044] Based on the above technical solution, after receiving the first information, the network device can also send second information for indicating whether to adjust the input parameters of the third neural network, so that the network device can adjust the input parameters of the third neural network deployed on the terminal device based on the performance detection results of the third neural network.

[0045] In a possible implementation of the first aspect, the method further includes: the network device sends third information, where the third information indicates channel information corresponding to the fourth reference signal processed based on the third neural network; wherein the third information is determined based on the first information.

[0046] Based on the above technical solution, before receiving the first information, the network device may also send third information for instructing the processing of the channel information corresponding to the fourth reference signal based on the third neural network, so that the terminal device can initiate performance detection of the channel information corresponding to the fourth reference signal based on the third information.

[0047] In a possible implementation manner of the first aspect, the method also includes: the network device sends a fourth reference signal; the network device receives channel information corresponding to the fourth reference signal; the channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], and T4, R4, and F4 are all positive numbers; at least one of the following is satisfied: T is greater than or equal to T4 and T4 is greater than T1, R is greater than or equal to R4 and R4 is greater than R1, and F is greater than or equal to F4 and F4 is greater than F1; the network device processes the channel information corresponding to the fourth reference signal and part or all of the Q second channel information based on the first neural network to obtain tenth channel information; wherein the channel dimension represented by the tenth channel information is [T, R, F]; the network device determines the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information.

[0048] Based on the above technical solution, the network device may further transmit a fourth reference signal and receive channel information corresponding to the fourth reference signal. Furthermore, the network device may process the channel information corresponding to the fourth reference signal based on the first neural network to obtain tenth channel information. Furthermore, the network device may further determine a deviation between the channel information corresponding to the fourth reference signal and the tenth channel information, so that the network device can determine the performance of the channel prediction of the first neural network based on the deviation.

[0049] In a possible implementation of the first aspect, the method further includes: the network device sends a fifth reference signal, and the channel dimension represented by the channel information corresponding to the fifth reference signal is determined based on the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information.

[0050] Based on the above technical solution, the network device can also send a fifth reference signal based on the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information, that is, the network device can update or adjust the sending of the reference signal based on the performance of the channel prediction of the first neural network.

[0051] In a possible implementation manner of the first aspect, the channel dimension represented by the channel information corresponding to the fifth reference signal is [T5, R5, F5], where T5, R5, and F5 are all positive numbers; the channel dimension represented by the channel information corresponding to the fifth reference signal is determined based on the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information, including: when the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is good, at least one of the following is satisfied: T5 is less than or equal to T4, R5 is less than or equal to R4, and F5 is less than or equal to F4; when the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is poor, at least one of the following is satisfied: T5 is greater than or equal to T4, R5 is greater than or equal to R4, and F5 is greater than or equal to F4.

[0052] Based on the above technical solution, when the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is better, that is, the network device determines that the channel prediction performance of the first neural network is better. For this reason, after sending the fourth reference signal, the network device can send to a fifth reference signal with the same or smaller pilot density to reduce overhead. Similarly, when the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is worse, that is, the network device determines that the channel prediction performance of the first neural network is worse. For this reason, after sending the fourth reference signal, the network device can send to a fifth reference signal with the same or larger pilot density, so that the network device can obtain channel information with a higher channel dimension and optimize the input parameters of the first neural network based on the channel information with a higher channel dimension to improve the channel prediction performance of the first neural network.

[0053] In a possible implementation manner of the first aspect, the method further includes: the network device adjusting an input parameter of the first neural network based on a deviation between the channel information corresponding to the fourth reference signal and the tenth channel information.

[0054] Based on the above technical solution, since the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information can reflect the performance of the channel prediction of the first neural network, the network device can also adjust the input parameters of the first neural network based on the deviation, and can optimize the input parameters of the first neural network to improve the performance of the channel prediction of the first neural network.

[0055] In a possible implementation manner of the first aspect, the method further includes: the network device sends a sixth reference signal; the network device receives channel information corresponding to the sixth reference signal; the network device processes the channel information corresponding to the sixth reference signal and K eleventh channel information based on the fourth neural network to obtain twelfth channel information; wherein the K eleventh channel information is determined by K reference signals, and K is a positive integer; the time domain resource carrying the Kth reference signal is located before the time domain resource carrying the sixth reference signal; the channel dimension represented by the twelfth channel information is [T, R, F]; the channel dimension represented by the twelfth channel information is [T, R, F]; wherein the network structure of the first neural network is different from the network structure of the fourth neural network; or, the network structure of the first neural network is the same as the network structure of the fourth neural network, and the parameters in the network structure of the first neural network are different from the parameters in the network structure of the fourth neural network; the sixth reference signal satisfies at least one of the following:

[0056] The resources carrying the sixth reference signal are different from the resources carrying the first reference signal;

[0057] The first channel information and the channel information corresponding to the sixth reference signal come from different terminal devices;

[0058] The first channel information and the channel information corresponding to the sixth reference signal come from terminal devices with different mobile information.

[0059] Based on the above technical solution, in addition to transmitting the first reference signal, the network device may also transmit a sixth reference signal and perform channel prediction on the channel information corresponding to the sixth reference signal based on the fourth neural network to obtain twelfth channel information of higher dimension. The transmission of the sixth reference signal satisfies at least one of the above requirements, thereby enhancing the flexibility of the solution implementation.

[0060] The second aspect of the present application provides a communication method, which is executed by a network device, or the method is executed by some components in the network device (such as a processor, a chip or a chip system, etc.), or the method can also be implemented by a logic module or software that can realize all or part of the network device functions. In the second aspect and its possible implementation, the communication method is described as being executed by a network device. In this method, the network device sends a fourth reference signal; the network device receives first information, and the first information is used to indicate the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the eighth channel information is obtained based on the third neural network processing the channel information corresponding to the fourth reference signal and P ninth channel information, and the P ninth channel information is determined by P reference signals, and the position of the time domain resource carrying the P reference signals is located before the position of the time domain resource carrying the fourth reference signal, and P is a positive integer.

[0061] Based on the above technical solution, the network device can transmit a fourth reference signal, and the eighth channel information is channel information obtained by processing the channel information corresponding to the fourth reference signal by the receiver of the fourth reference signal (i.e., the terminal device) based on the third neural network. In addition, the network device can also receive first information indicating a deviation between the channel information corresponding to the fourth reference signal and the eighth channel information, so that the network device can determine the performance of the channel prediction of the third neural network based on the first information.

[0062] In addition, compared to the method in which the network device receives the channel information corresponding to the fourth reference signal and processes the channel information corresponding to the fourth reference signal based on the first neural network to determine the performance of the channel prediction of the first neural network, in the above technical solution, the network device can determine the performance of the channel prediction of the neural network through the first information. Since the transmission overhead of the first information is less than the transmission overhead of the channel information corresponding to the fourth reference signal, the transmission overhead can be reduced.

[0063] In a possible implementation of the second aspect, the channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], where T4, R4, and F4 are all positive numbers; the channel dimension represented by the channel information corresponding to the fourth reference signal and the channel dimension represented by the channel information corresponding to the eighth reference signal satisfy at least one of the following: T is greater than or equal to T4, R is greater than or equal to R4, and F is greater than or equal to F4.

[0064] Based on the above technical solution, the channel information corresponding to the fourth reference signal is channel information of lower dimension than the channel information corresponding to the eighth reference signal, which can reduce the overhead in the detection process of the channel prediction performance of the neural network.

[0065] In a possible implementation manner of the second aspect, the method further includes: the network device sending second information, where the second information is used to indicate whether to adjust the input parameters of the third neural network; wherein the second information is determined based on the first information.

[0066] Based on the above technical solution, after receiving the first information, the network device can also send second information for indicating whether to adjust the input parameters of the third neural network, so that the network device can adjust the input parameters of the third neural network deployed on the terminal device based on the performance detection results of the third neural network.

[0067] In a possible implementation of the second aspect, the method further includes: the network device sends third information, where the third information indicates channel information corresponding to the fourth reference signal processed based on the third neural network; wherein the third information is determined based on the first information.

[0068] Based on the above technical solution, before receiving the first information, the network device may also send third information for instructing the processing of the channel information corresponding to the fourth reference signal based on the third neural network, so that the terminal device can initiate performance detection of the channel information corresponding to the fourth reference signal based on the third information.

[0069] The third aspect of the present application provides a communication method, which is executed by a terminal device, or the method is executed by some components in the terminal device (such as a processor, a chip or a chip system, etc.), or the method can also be implemented by a logic module or software that can realize all or part of the terminal device functions. In the third aspect and its possible implementation, the communication method is described as being executed by a terminal device. In this method, the terminal device receives a fourth reference signal; the terminal device sends first information, and the first information is used to indicate the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on the third neural network, and the P ninth channel information is determined by P reference signals, and the position of the time domain resource carrying the P reference signals is located before the position of the time domain resource carrying the fourth reference signal, and P is a positive integer.

[0070] Based on the above technical solution, the terminal device can receive a fourth reference signal, and the eighth channel information is channel information obtained by processing the channel information corresponding to the fourth reference signal by the receiver of the fourth reference signal (i.e., the terminal device) based on the third neural network. In addition, the network device can also receive first information indicating a deviation between the channel information corresponding to the fourth reference signal and the eighth channel information, so that the network device can determine the performance of the channel prediction of the third neural network based on the first information.

[0071] In addition, compared to the method in which the terminal device sends the channel information corresponding to the fourth reference signal, and enables the network device to process the channel information corresponding to the fourth reference signal based on the first neural network to determine the performance of the channel prediction of the first neural network, in the above technical solution, the network device can determine the performance of the channel prediction of the neural network through the first information. Since the transmission overhead of the first information is less than the transmission overhead of the channel information corresponding to the fourth reference signal, the transmission overhead can be reduced.

[0072] In a possible implementation of the third aspect, the channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], where T4, R4, and F4 are all positive numbers; the channel dimension represented by the channel information corresponding to the fourth reference signal and the channel dimension represented by the channel information corresponding to the eighth reference signal satisfy at least one of the following: T is greater than or equal to T4, R is greater than or equal to R4, and F is greater than or equal to F4.

[0073] Based on the above technical solution, the channel information corresponding to the fourth reference signal is channel information of lower dimension than the channel information corresponding to the eighth reference signal, which can reduce the overhead in the detection process of the channel prediction performance of the neural network.

[0074] In a possible implementation manner of the third aspect, the method further includes: the terminal device receiving second information, the second information being used to indicate whether to adjust the input parameters of the third neural network; wherein the second information is determined based on the first information.

[0075] Based on the above technical solution, after the terminal device sends the first information to the network device, the terminal device can also receive second information for indicating whether to adjust the input parameters of the third neural network, so that the network device can adjust the input parameters of the third neural network deployed on the terminal device based on the performance detection results of the third neural network.

[0076] In a possible implementation of the third aspect, the method further includes: the terminal device receives third information, the third information indicating channel information corresponding to the fourth reference signal processed based on the third neural network; wherein the third information is determined based on the first information.

[0077] Based on the above technical solution, before the terminal device sends the first information to the network device, the terminal device can also receive third information for instructing the processing of the channel information corresponding to the fourth reference signal based on the third neural network, so that the terminal device can start performance detection of the channel information corresponding to the fourth reference signal based on the third information.

[0078] In a fourth aspect, the present application provides a communication device, which is a network device, or a component of a network device (such as a processor, chip, or chip system), or a logic module or software capable of implementing all or part of the network device's functions. In the fourth aspect and its possible implementations, the communication device is described as a network device.

[0079] The device includes a processing unit and a transceiver unit; the transceiver unit is used to send a first reference signal; the transceiver unit is also used to receive first channel information, the first channel information is determined based on the first reference signal, the channel dimension represented by the first channel information is [T1, R1, F1], and T1, R1, and F1 are all positive numbers; the processing unit is used to process the first channel information and Q second channel information based on a first neural network to obtain third channel information; wherein, the Q second channel information are determined by Q second reference signals, and Q is a positive integer; the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, and the channel dimension represented by the third channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the parameters in the channel dimension represented by the first channel information and the parameters in the channel dimension represented by the third channel information satisfy at least one of the following: T is greater than the T1, R is greater than R1, and F is greater than F1.

[0080] In the fourth aspect of the present application, the constituent modules of the communication device can also be used to execute the steps performed in each possible implementation method of the first aspect and achieve corresponding technical effects. For details, please refer to the first aspect and will not be repeated here.

[0081] In a fifth aspect, the present application provides a communication device, which is a network device, or a component of a network device (such as a processor, chip, or chip system), or a logic module or software capable of implementing all or part of the network device functions. In the fifth aspect and its possible implementations, the communication device is described as a network device.

[0082] The device includes a processing unit and a transceiver unit; the transceiver unit is used to send a fourth reference signal; the transceiver unit is also used to receive first information, and the processing unit is used to determine the deviation between channel information corresponding to the fourth reference signal and eighth channel information based on the first information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on a third neural network, the P ninth channel information are determined by P reference signals, the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

[0083] In the fifth aspect of this application, the constituent modules of the communication device can also be used to execute the steps performed in each possible implementation method of the second aspect and achieve corresponding technical effects. For details, please refer to the second aspect and will not be repeated here.

[0084] In a sixth aspect of the present application, a communication device is provided, which is a terminal device, or a component of a terminal device (such as a processor, chip, or chip system), or a logic module or software capable of implementing all or part of the terminal device functions. In the fifth aspect and its possible implementations, the communication device is described as an example of a terminal device.

[0085] The device includes a processing unit and a transceiver unit; the transceiver unit is used to receive a fourth reference signal; the processing unit is used to determine first information; the transceiver unit is also used to send first information, and the first information is used to indicate the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on a third neural network, and the P ninth channel information is determined by P reference signals, and the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

[0086] In the sixth aspect of this application, the constituent modules of the communication device can also be used to execute the steps performed in each possible implementation method of the third aspect and achieve corresponding technical effects. For details, please refer to the third aspect and will not be repeated here.

[0087] In a seventh aspect, the present application provides a communication device, comprising at least one processor, wherein the at least one processor is coupled to a memory; the memory is used to store programs or instructions; wherein the at least one processor is used to execute the program or instructions so that the device implements the method described in any one of the first to third aspects and any possible implementation methods thereof.

[0088] In an eighth aspect, the present application provides a communication device comprising at least one logic circuit and an input / output interface; the logic circuit is used to execute the method described in any one of the first to third aspects and any possible implementation thereof.

[0089] In a ninth aspect, the present application provides a computer-readable storage medium storing instructions. When the instructions are executed by a processor, the processor executes the method described in any one of the first to third aspects above and any possible implementation thereof.

[0090] The tenth aspect of the present application provides a computer program product (or computer program), which includes computer program code. When the computer program code is run on a processor, the processor executes the method described in any one of the first to third aspects above and any possible implementation thereof.

[0091] In the eleventh aspect of the present application, a chip system is provided, which includes at least one processor for supporting a communication device to implement the functions involved in any one of the first to third aspects and any possible implementation methods thereof.

[0092] In one possible design, the chip system may further include a memory for storing program instructions and data necessary for the first communication device. The chip system may be composed of a chip or may include a chip and other discrete components. Optionally, the chip system may further include an interface circuit that provides program instructions and / or data to the at least one processor.

[0093] A twelfth aspect of the present application provides a communication system, which includes the communication device of the third aspect and the communication device of the fourth aspect. Alternatively, the communication system includes the terminal device of any of the above aspects and any implementation thereof, and the network device of any of the above aspects and any implementation thereof.

[0094] It should be understood that the technical effects brought about by any design method in the fourth to twelfth aspects can be referred to the technical effects brought about by the different design methods in the above-mentioned first to third aspects, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0095] FIG1a is a schematic diagram of a communication system provided by the present application;

[0096] FIG1b is another schematic diagram of the communication system provided by the present application;

[0097] FIG1c is another schematic diagram of the communication system provided by this application;

[0098] FIG1d is a schematic diagram of the AI ​​processing process involved in this application;

[0099] FIG1e is another schematic diagram of the AI ​​processing process involved in this application;

[0100] FIG2a is another schematic diagram of the AI ​​processing process involved in this application;

[0101] FIG2 b is another schematic diagram of the AI ​​processing process involved in this application;

[0102] FIG2c is another schematic diagram of the AI ​​processing process involved in this application;

[0103] FIG2 d is another schematic diagram of the AI ​​processing process involved in this application;

[0104] FIG2e is another schematic diagram of the AI ​​processing process involved in this application;

[0105] FIG3 is an interactive schematic diagram of the communication method provided by this application;

[0106] FIG4a is another schematic diagram of the AI ​​processing process provided by this application;

[0107] FIG4 b is another schematic diagram of the AI ​​processing process provided by this application;

[0108] FIG4c is another schematic diagram of the AI ​​processing process provided by this application;

[0109] FIG4 d is another schematic diagram of the AI ​​processing process provided by this application;

[0110] FIG5a is another schematic diagram of the AI ​​processing process provided by this application;

[0111] FIG5 b is another schematic diagram of the AI ​​processing process provided by this application;

[0112] FIG5c is another schematic diagram of the AI ​​processing process provided by this application;

[0113] FIG5 d is another schematic diagram of the AI ​​processing process provided by this application;

[0114] FIG6a is another schematic diagram of the AI ​​processing process provided by this application;

[0115] FIG6 b is another schematic diagram of the AI ​​processing process provided by this application;

[0116] FIG7 is another schematic diagram of the AI ​​processing process provided by this application;

[0117] FIG8 is a schematic diagram of a communication device provided by the present application;

[0118] FIG9 is another schematic diagram of a communication device provided by the present application;

[0119] FIG10 is another schematic diagram of a communication device provided by the present application;

[0120] FIG11 is another schematic diagram of a communication device provided by the present application;

[0121] FIG12 is another schematic diagram of the communication device provided in this application. DETAILED DESCRIPTION

[0122] First, some of the terms used in the embodiments of the present application are explained to facilitate understanding by those skilled in the art.

[0123] (1) Terminal device: It can be a wireless terminal device that can receive network device scheduling and instruction information. The wireless terminal device can be a device that provides voice and / or data connectivity to the user, or a handheld device with wireless connection function, or other processing device connected to a wireless modem.

[0124] Terminal devices can communicate with one or more core networks or the Internet via a radio access network (RAN). Terminal devices can be mobile terminal devices, such as mobile phones (also known as "cellular" phones, mobile phones), computers, and data cards. For example, they can be portable, pocket-sized, handheld, computer-built-in, or vehicle-mounted mobile devices that exchange voice and / or data with the radio access network. Examples include personal communication service (PCS) phones, cordless phones, Session Initiation Protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), tablet computers, and computers with wireless transceiver capabilities. Wireless terminal equipment can also be called system, subscriber unit, subscriber station, mobile station, mobile station (MS), remote station, access point (AP), remote terminal equipment (remote terminal), access terminal equipment (access terminal), user terminal equipment (user terminal), user agent, subscriber station (SS), customer premises equipment (CPE), terminal, user equipment (UE), mobile terminal (MT), etc.

[0125] As an example and not a limitation, in the embodiments of the present application, the terminal device may also be a wearable device. Wearable devices may also be referred to as wearable smart devices or smart wearable devices, etc., which are a general term for wearable devices that are intelligently designed and developed using wearable technology for daily wear, such as glasses, gloves, watches, clothing, and shoes. A wearable device is a portable device that is worn directly on the body or integrated into the user's clothes or accessories. Wearable devices are not only hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are fully functional, large in size, and can achieve complete or partial functions without relying on smartphones, such as smart watches or smart glasses, etc., as well as those that only focus on a certain type of application function and need to be used in conjunction with other devices such as smartphones, such as various smart bracelets, smart helmets, and smart jewelry for vital sign monitoring.

[0126] The terminal may also be a drone, a robot, a terminal in device-to-device (D2D) communication, a terminal in vehicle-to-everything (V2X), a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal in industrial control, a wireless terminal in self-driving, a wireless terminal in remote medical, a wireless terminal in smart grid, a wireless terminal in transportation safety, a wireless terminal in smart city, a wireless terminal in smart home, etc.

[0127] In addition, the terminal device may also be a terminal device in a communication system that has evolved after the fifth generation (5G) communication system (e.g., a sixth generation (6G) communication system) or a terminal device in a future public land mobile network (PLMN). For example, the 6G network can further expand the form and function of 5G communication terminals. 6G terminals include but are not limited to vehicles, cellular network terminals (with integrated satellite terminal functions), drones, and Internet of Things (IoT) devices.

[0128] In an embodiment of the present application, the terminal device may also obtain AI services provided by the network device. Optionally, the terminal device may also have AI processing capabilities.

[0129] (2) Network equipment: It can be a device in a wireless network. For example, the network equipment can be a RAN node (or device) that connects a terminal device to a wireless network, which can also be called a base station. Currently, some examples of RAN equipment include: base station, evolved NodeB (eNodeB), gNB (gNodeB) in a 5G communication system, transmission reception point (TRP), evolved Node B (eNB), radio network controller (RNC), Node B (NB), home base station (e.g., home evolved Node B, or home Node B, HNB), base band unit (BBU), or wireless fidelity (Wi-Fi) access point AP, etc. In addition, in a network structure, the network equipment can include a centralized unit (CU) node, a distributed unit (DU) node, or a RAN device including a CU node and a DU node.

[0130] Alternatively, a RAN node can be a macro base station, micro base station, indoor base station, relay node, donor node, or a wireless controller in a cloud radio access network (CRAN) scenario. A RAN node can also be a server, wearable device, vehicle, or vehicle-mounted device. For example, the access network device in vehicle-to-everything (V2X) technology can be a roadside unit (RSU).

[0131] In another possible scenario, multiple RAN nodes collaborate to assist the terminal in achieving wireless access, and different RAN nodes respectively implement part of the functions of the base station. For example, the RAN node can be a centralized unit (CU), a distributed unit (DU), a CU-control plane (CP), a CU-user plane (UP), or a radio unit (RU). The CU and DU can be set separately, or they can be included in the same network element, such as a baseband unit (BBU). The RU can be included in a radio frequency device or radio frequency unit, such as a remote radio unit (RRU), an active antenna unit (AAU), or a remote radio head (RRH).

[0132] In different systems, CU (or CU-CP and CU-UP), DU or RU may also have different names, but those skilled in the art can understand their meanings. For example, in an open access network (open RAN, O-RAN or ORAN) system, CU may also be called O-CU (open CU), DU may also be called O-DU, CU-CP may also be called O-CU-CP, CU-UP may also be called O-CU-UP, and RU may also be called O-RU. For the convenience of description, this application takes CU, CU-CP, CU-UP, DU and RU as examples for description. Any unit of CU (or CU-CP, CU-UP), DU and RU in this application can be implemented by a software module, a hardware module, or a combination of a software module and a hardware module.

[0133] The communication between the access network device and the terminal device follows a certain protocol layer structure. The protocol layer may include a control plane protocol layer and a user plane protocol layer. The control plane protocol layer may include at least one of the following: a radio resource control (RRC) layer, a packet data convergence protocol (PDCP) layer, a radio link control (RLC) layer, a media access control (MAC) layer, or a physical (PHY) layer. The user plane protocol layer may include at least one of the following: a service data adaptation protocol (SDAP) layer, a PDCP layer, an RLC layer, a MAC layer, or a physical layer.

[0134] For the correspondence between network elements in the ORAN system and their achievable protocol layer functions, please refer to Table 1 below.

[0135] Table 1

[0136] The network device may be any other device that provides wireless communication functionality to the terminal device. The embodiments of this application do not limit the specific technology and device form used by the network device. For ease of description, the embodiments of this application do not limit this.

[0137] The network equipment may also include core network equipment, which may include, for example, a mobility management entity (MME), a home subscriber server (HSS), a serving gateway (S-GW), a policy and charging rules function (PCRF), and a public data network gateway (PDN gateway, P-GW) in a fourth generation (4G) network; and network elements such as an access and mobility management function (AMF), a user plane function (UPF), or a session management function (SMF) in a 5G network. In addition, the core network equipment may also include other core network equipment in a 5G network and a next generation network of a 5G network.

[0138] In an embodiment of the present application, the above-mentioned network device may also have a network node with AI capabilities, which can provide AI services for terminals or other network devices. For example, it can be an AI node on the network side (access network or core network), a computing power node, a RAN node with AI capabilities, a core network element with AI capabilities, etc.

[0139] In the embodiments of the present application, the apparatus for implementing the function of the network device may be the network device, or may be a device capable of supporting the network device in implementing the function, such as a chip system, which may be installed in the network device. In the technical solutions provided in the embodiments of the present application, the technical solutions provided in the embodiments of the present application are described by taking the network device as an example.

[0140] (3) Configuration and pre-configuration: In this application, configuration and pre-configuration are used simultaneously. Configuration refers to the network device / server sending some parameter configuration information or parameter values ​​to the terminal through messages or signaling, so that the terminal can determine the communication parameters or resources during transmission based on these values ​​or information. Pre-configuration is similar to configuration, and can be parameter information or parameter values ​​pre-negotiated between the network device / server and the terminal device, or parameter information or parameter values ​​used by the base station / network device or terminal device as specified in the standard protocol, or parameter information or parameter values ​​pre-stored in the base station / server or terminal device. This application does not limit this.

[0141] Furthermore, these values ​​and parameters can be changed or updated.

[0142] (4) The terms "system" and "network" in the embodiments of the present application can be used interchangeably. "Multiple" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, "at least one of A, B and C" includes A, B, C, AB, AC, BC or ABC. In addition, unless otherwise specified, the ordinal numbers such as "first" and "second" mentioned in the embodiments of the present application are used to distinguish multiple objects, and are not used to limit the order, timing, priority or importance of multiple objects.

[0143] (5) “Sending” and “receiving” in the embodiments of the present application indicate the direction of signal transmission. For example, “sending information to XX” can be understood as the destination of the information being XX, which can include direct sending through the air interface, as well as indirect sending through the air interface by other units or modules. “Receiving information from YY” can be understood as the source of the information being YY, which can include direct receiving from YY through the air interface, as well as indirect receiving from YY through the air interface from other units or modules. “Sending” can also be understood as the “output” of the chip interface, and “receiving” can also be understood as the “input” of the chip interface.

[0144] In other words, sending and receiving can be performed between devices, for example, between a network device and a terminal device, or can be performed within a device, for example, sending or receiving between components, modules, chips, software modules or hardware modules within the device through a bus, wiring or interface.

[0145] It is understandable that information may be processed between the source and destination of information transmission, such as coding, modulation, etc., but the destination can understand the valid information from the source. Similar expressions in this application can be understood similarly and will not be repeated.

[0146] (6) In the embodiments of the present application, "indication" may include direct indication and indirect indication, and may also include explicit indication and implicit indication. The information indicated by a certain information (such as the indication information described below) is called information to be indicated. In the specific implementation process, there are many ways to indicate the information to be indicated, such as but not limited to, directly indicating the information to be indicated, such as the information to be indicated itself or the index of the information to be indicated. The information to be indicated may also be indirectly indicated by indicating other information, wherein the other information is associated with the information to be indicated; or only a part of the information to be indicated may be indicated, while the other part of the information to be indicated is known or agreed in advance. For example, the indication of specific information may be achieved by means of the arrangement order of each information agreed in advance (such as predefined by the protocol), thereby reducing the indication overhead to a certain extent. The present application does not limit the specific method of indication. It is understandable that for the sender of the indication information, the indication information can be used to indicate the information to be indicated, and for the receiver of the indication information, the indication information can be used to determine the information to be indicated.

[0147] In this application, unless otherwise specified, the same or similar parts between the various embodiments can refer to each other. In the various embodiments of this application, and the various methods / designs / implementations in each embodiment, if there is no special explanation and logical conflict, the terms and / or descriptions between different embodiments and the various methods / designs / implementations in each embodiment are consistent and can be referenced to each other. The technical features in different embodiments and the various methods / designs / implementations in each embodiment can be combined to form new embodiments, methods, or implementations according to their inherent logical relationships. The following description of the implementation methods of this application does not constitute a limitation on the scope of protection of this application.

[0148] The present application can be applied to a long term evolution (LTE) system, a new radio (NR) system, or a communication system evolved after 5G (such as 6G, etc.). The communication system includes at least one network device and / or at least one terminal device.

[0149] Please refer to Figure 1a, which is a schematic diagram of a communication system in this application. Figure 1a exemplarily illustrates a network device and six terminal devices, namely terminal device 1, terminal device 2, terminal device 3, terminal device 4, terminal device 5, and terminal device 6. In the example shown in Figure 1a, terminal device 1 is a smart teacup, terminal device 2 is a smart air conditioner, terminal device 3 is a smart gas pump, terminal device 4 is a vehicle, terminal device 5 is a mobile phone, and terminal device 6 is a printer.

[0150] As shown in Figure 1a, the AI ​​configuration information sending entity can be a network device. The AI ​​configuration information receiving entity can be terminal devices 1-6. In this case, the network device and terminal devices 1-6 form a communication system. In this communication system, terminal devices 1-6 can send data to the network device, and the network device needs to receive data sent by terminal devices 1-6. At the same time, the network device can send configuration information to terminal devices 1-6.

[0151] For example, in Figure 1a, terminal devices 4 and 6 can also form a communication system. Terminal device 5 serves as a network device, i.e., the AI ​​configuration information sending entity; terminal devices 4 and 6 serve as terminal devices, i.e., the AI ​​configuration information receiving entities. For example, in a connected vehicle system, terminal device 5 sends AI configuration information to terminal devices 4 and 6, respectively, and receives data from them. Correspondingly, terminal devices 4 and 6 receive AI configuration information from terminal device 5 and send data to terminal device 5.

[0152] Taking the communication system shown in Figure 1a as an example, in addition to executing communication-related services, different devices (including between network devices, between network devices and terminal devices, and / or between terminal devices) may also execute AI-related services.

[0153] As shown in Figure 1b, taking the network device as a base station as an example, the base station can perform communication-related services and AI-related services with one or more terminal devices, and different terminal devices can also perform communication-related services and AI-related services.

[0154] As shown in Figure 1c, taking the terminal devices including a TV and a mobile phone as an example, communication-related services and AI-related services can also be performed between the TV and the mobile phone.

[0155] The technical solution provided in this application can be applied to a wireless communication system (e.g., the system shown in FIG. 1a , FIG. 1b , or FIG. 1c ). For example, an AI network element can be introduced into the communication system provided in this application to implement some or all AI-related operations. The AI ​​network element can also be referred to as an AI node, AI device, AI entity, AI module, AI model, or AI unit, etc. The AI ​​network element can be a network element built into the communication system. For example, the AI ​​network element can be an AI module built into: an access network device, a core network device, a cloud server, or a network management (OAM) to implement AI-related functions. The OAM can be a network management device for a core network device and / or a network management device for an access network device. Alternatively, the AI ​​network element can also be an independently set network element in the communication system. Optionally, the terminal or the chip built into the terminal can also include an AI entity to implement AI-related functions.

[0156] The following is a brief introduction to artificial intelligence (AI) that may be involved in this application.

[0157] Artificial intelligence (AI) can imbue machines with human intelligence. For example, it can enable machines to simulate certain intelligent human behaviors using computer hardware and software. Machine learning methods can be used to achieve AI. In machine learning, a machine uses training data to learn (or train) a model. This model represents the mapping from input to output. The learned model can be used for inference (or prediction), meaning that the model can be used to predict the output corresponding to a given input. This output can also be called an inference result (or prediction result).

[0158] Machine learning can include supervised learning, unsupervised learning, and reinforcement learning. Among them, unsupervised learning can also be called unsupervised learning.

[0159] Supervised learning uses machine learning algorithms to learn the mapping relationship between sample values ​​and sample labels based on collected sample values ​​and sample labels, and then expresses this learned mapping relationship using an AI model. The process of training a machine learning model is the process of learning this mapping relationship. During training, sample values ​​are input into the model to obtain the model's predicted values. The model parameters are optimized by calculating the error between the model's predicted values ​​and the sample labels (ideal values). Once the mapping relationship is learned, the learned mapping can be used to predict new sample labels. The mapping relationship learned by supervised learning can include linear mappings or nonlinear mappings. Based on the type of label, the learning task can be divided into classification tasks and regression tasks.

[0160] Unsupervised learning uses algorithms to discover inherent patterns in collected sample values. One type of unsupervised learning algorithm uses the samples themselves as supervisory signals, meaning the model learns the mapping from one sample to another. This is called self-supervised learning. During training, the model parameters are optimized by calculating the error between the model's predictions and the samples themselves. Self-supervised learning can be used in signal compression and decompression recovery applications. Common algorithms include autoencoders and generative adversarial networks.

[0161] Reinforcement learning, unlike supervised learning, is a type of algorithm that learns problem-solving strategies through interaction with the environment. Unlike supervised and unsupervised learning, reinforcement learning problems lack explicit label data for "correct" actions. Instead, the algorithm must interact with the environment to obtain reward signals from the environment, and then adjust its decision-making actions to maximize the reward signal value. For example, in downlink power control, the reinforcement learning model adjusts the downlink transmit power of each user based on the overall system throughput fed back by the wireless network, hoping to achieve higher system throughput. The goal of reinforcement learning is also to learn the mapping between environmental states and optimal (e.g., optimal) decision-making actions. However, because the labels for "correct actions" cannot be obtained in advance, network optimization cannot be achieved by calculating the error between actions and "correct actions." Reinforcement learning training is achieved through iterative interaction with the environment.

[0162] A neural network (NN) is a specific model in machine learning technology. According to the universal approximation theorem, NNs can theoretically approximate any continuous function, enabling them to learn arbitrary mappings. Traditional communication systems require extensive expert knowledge to design communication modules. However, deep learning communication systems based on neural networks can automatically discover implicit patterns in massive data sets and establish mapping relationships between data, achieving performance superior to traditional modeling methods.

[0163] The idea of ​​a neural network is derived from the neuronal structure of the brain. For example, each neuron performs a weighted sum operation on its input values ​​and outputs the result through an activation function.

[0164] As shown in Figure 1d, it is a schematic diagram of the neuron structure. Assume that the input of the neuron is x=[x0,x1,…,x n ], and the weights corresponding to each input are w=[w,w1,…,w n ], where n is a positive integer, w i and x i It can be a decimal, an integer (such as 0, a positive integer or a negative integer, etc.), or a complex number. i As x i The weight of x i Weighted. The bias of the weighted sum of the input values ​​according to the weight is, for example, b. The activation function can take many forms. Assuming that the activation function of a neuron is: y = f(z) = max(0,z), then the output of the neuron is: For another example, if the activation function of a neuron is: y = f(z) = z, then the output of the neuron is: b can be a decimal, an integer (eg, 0, a positive integer, or a negative integer), or a complex number, etc. The activation functions of different neurons in a neural network can be the same or different.

[0165] Furthermore, neural networks generally include multiple layers, each of which may include one or more neurons. Increasing the depth and / or width of a neural network can improve its expressive power, providing more powerful information extraction and abstract modeling capabilities for complex systems. The depth of a neural network can refer to the number of layers it comprises, and the number of neurons in each layer can be referred to as the width of that layer. In one implementation, a neural network includes an input layer and an output layer. The input layer processes the input information received by the neural network through neurons, passing the processing results to the output layer, which then obtains the output of the neural network. In another implementation, a neural network includes an input layer, a hidden layer, and an output layer. The input layer processes the input information received by the neural network through neurons, passing the processing results to an intermediate hidden layer. The hidden layer performs calculations on the received processing results to obtain a calculation result, which is then passed to the output layer or the next adjacent hidden layer, which ultimately obtains the output of the neural network. A neural network can include one hidden layer or multiple hidden layers connected in sequence, without limitation.

[0166] A neural network is, for example, a deep neural network (DNN). Depending on how the network is constructed, a DNN can include a feedforward neural network (FNN), a convolutional neural network (CNN), and a recurrent neural network (RNN).

[0167] Figure 1e is a schematic diagram of a FNN network. A characteristic of FNN networks is that neurons in adjacent layers are fully connected. This characteristic typically requires a large amount of storage space and results in high computational complexity.

[0168] CNN is a neural network specifically designed to process data with a grid-like structure. For example, time series data (discrete sampling along the time axis) and image data (discrete sampling along two dimensions) can both be considered grid-like data. CNNs do not utilize all input information at once for computation. Instead, they use a fixed-size window to intercept a portion of the information for convolution operations, significantly reducing the computational complexity of model parameters. Furthermore, depending on the type of information intercepted by the window (e.g., people and objects in an image represent different types of information), each window can use a different convolution kernel, enabling CNNs to better extract features from the input data.

[0169] RNNs are a type of DNN that utilizes feedback time series information. Their input consists of a new input value at the current moment and their own output value at the previous moment. RNNs are suitable for capturing temporally correlated sequence features and are particularly well-suited for applications such as speech recognition and channel coding.

[0170] During the machine learning model training process, a loss function can be defined. This function describes the gap or discrepancy between the model's output and the ideal target value. Loss functions can be expressed in various forms, and there are no restrictions on their specific form. The model training process can be viewed as adjusting some or all of the model's parameters to keep the loss function below a threshold or meet the target.

[0171] A model may also be referred to as an AI model, rule, or other name. An AI model can be considered a specific method for implementing an AI function. An AI model represents a mapping relationship or function between the input and output of a model. AI functions may include one or more of the following: data collection, model training (or model learning), model information release, model inference (or model reasoning, inference, or prediction, etc.), model monitoring or model verification, or inference result release, etc. AI functions may also be referred to as AI (related) operations, or AI-related functions.

[0172] The following is an exemplary description of the implementation process of the neural network with reference to the accompanying drawings.

[0173] 1. Fully connected neural network, also known as multilayer perceptron (MLP).

[0174] As shown in Figure 2a, an MLP consists of an input layer (left), an output layer (right), and multiple hidden layers (center). Each layer of the MLP contains several nodes, called neurons. Neurons in adjacent layers are connected to each other.

[0175] Alternatively, considering neurons in two adjacent layers, the output h of the neurons in the next layer is the weighted sum of all neurons x in the previous layer connected to it and passes through the activation function, which can be expressed as:

[0176] h=f(wx+b).

[0177] Among them, w is the weight matrix, b is the bias vector, and f is the activation function.

[0178] Alternatively, the output of the neural network can be recursively expressed as:

[0179] y=f n (w n f n-1 (…)+b n ).

[0180] Where n is the index of the neural network layer, 1<=n<=N, where N is the total number of neural network layers.

[0181] In other words, a neural network can be understood as a mapping from an input data set to an output data set. Neural networks are typically initialized randomly, and the process of obtaining this mapping from random w and b using existing data is called neural network training.

[0182] Optionally, a specific training method is to use a loss function to evaluate the output results of the neural network.

[0183] As shown in Figure 2b, the error can be backpropagated, and the neural network parameters (including w and b) can be iteratively optimized using gradient descent until the loss function reaches a minimum, which is the "better point (e.g., optimal point)" in Figure 2b. It is understood that the neural network parameters corresponding to the "better point (e.g., optimal point)" in Figure 2b can be used as the neural network parameters in the trained AI model information.

[0184] Alternatively, the gradient descent process can be expressed as:

[0185] Among them, θ is the parameter to be optimized (including w and b), L is the loss function, and η is the learning rate, which controls the step size of gradient descent. represents the derivative operation, represents the derivative of θ with respect to L.

[0186] Optionally, the backpropagation process utilizes the chain rule for partial derivatives.

[0187] As shown in Figure 2c, the gradient of the previous layer parameters can be recursively calculated from the gradient of the next layer parameters, which can be expressed as:

[0188] Among them, w ij is the weight of node j connecting to node i, s i is the weighted sum of the inputs to node i.

[0189] 2. Federated Learning (FL)

[0190] The concept of federated learning effectively solves the current difficulties faced by the development of artificial intelligence. On the premise of fully protecting user data privacy and security, it efficiently completes the model learning task by promoting the collaboration between various edge devices and central servers.

[0191] As shown in Figure 2d, the FL architecture is the most widely used training architecture in the current FL field. The FedAvg algorithm is the basic algorithm of FL. Its algorithm flow is roughly as follows:

[0192] (1) The center initializes the model to be trained And broadcast it to all client devices.

[0193] (2) In the round t∈[1,T], client k∈[1,K] based on the local dataset For the received global model Perform E epochs of training to obtain local training results Report it to the central node.

[0194] (3) The central node aggregates and collects the local training results from all (or some) clients. Assume that the client set that uploads the local model in round t is The center will use the number of samples of the corresponding client as the weight to perform weighted averaging to obtain a new global model. The specific update rule is: The center then sends the latest version of the global model Broadcast to all client devices for a new round of training.

[0195] (4) Repeat steps (2) and (3) until the model finally converges or the number of training rounds reaches the upper limit.

[0196] In addition to reporting local models You can also use the local gradient of training After reporting, the central node averages the local gradients and updates the global model according to the direction of the average gradient.

[0197] As you can see, in the FL framework, datasets exist on distributed nodes. Distributed nodes collect local datasets, perform local training, and report the local training results (models or gradients) to the central node. The central node itself does not have a dataset; it is only responsible for fusing the training results of distributed nodes to obtain a global model and send it to the distributed nodes.

[0198] 3. Decentralized learning: Different from federated learning, decentralized learning is another distributed learning architecture.

[0199] As shown in Figure 2e, consider a fully distributed system without a central node. The design goal f(x) of a decentralized learning system is generally the goal f of each node. i The mean of (x), that is Where n is the number of distributed nodes, x is the parameter to be optimized. In machine learning, x is the parameter of the machine learning (such as neural network) model. Each node uses local data and local target f i (x) Calculate local gradient Then it is sent to the neighboring nodes that can be communicated with. After any node receives the gradient information sent by its neighbor, it can update the parameter x of the local model according to the following formula:

[0200] in, represents the parameters of the local model after the k+1th (k is a natural number) update in the i-th node, Represents the parameters of the local model after the kth update in the i-th node (if k is 0, it means is the parameter of the local model of the i-th node that does not participate in the update), α k Represents the tuning coefficient, N i is the set of neighbor nodes of node i, |N i | represents the number of elements in the neighbor node set of node i, that is, the number of neighbor nodes of node i. Through information interaction between nodes, the decentralized learning system will eventually learn a unified model.

[0201] The technical solutions provided in this application can be applied to wireless communication systems (e.g., the systems shown in Figures 1a, 1b, or 1c). In these wireless communication systems, multiple-input, multiple-output (MIMO) technology can be used to meet high-speed transmission requirements. For example, through a channel measurement (or channel estimation) process, a network device can use the channel information obtained in the channel measurement process to calculate precoding information between the network device and a terminal device. Subsequently, MIMO communication can be achieved between the network device and the terminal device using this precoding information.

[0202] Taking the downlink channel measurement process implemented by a network device based on a downlink reference signal as an example, the downlink reference signal sent by the network device may include a channel state information reference signal (CSI-RS). The network device may receive feedback of the CSI-RS, enabling the network device to obtain channel information based on the feedback of the CSI-RS. The CSI-RS is sent through a port of the network device, meaning that the CSI-RS overhead is related to the number of ports of the network device.

[0203] However, in MIMO systems, as frequency bands increase and the demand for high-speed communications increases, the number of ports on network devices may gradually increase, which will increase the overhead of reference signals (such as CSI-RS) used for channel measurement and occupy more transmission resources, thereby increasing the power consumption of terminal devices.

[0204] For example, taking the CSI obtained by CSI-RS as an example, CSI describes some channel characteristics of a wireless link. It can characterize the joint characteristics of path loss, scattering, diffraction, fading, shadowing, etc. in the process of signal propagation from the transmitter to its corresponding receiver. Therefore, obtaining accurate CSI plays a vital role in the communication performance of mobile users in wireless communication systems. More specifically, CSI highly determines the physical layer parameters and schemes deployed for radio communication in wireless communication systems. Obtaining CSI is a crucial issue in wireless communication, which determines whether technologies such as MIMO can release their potential transmission gain. In the new generation of wireless communication systems, with the use of MIMO technology, wider communication frequency bands and support for high mobility, timely and accurate acquisition of CSI faces huge challenges.

[0205] On the one hand, multiple antennas and wider communication bandwidths significantly increase the number of channel dimensions required for acquisition. This necessitates that traditional channel estimation include more pilot signals (also known as "reference signals") in resource blocks, reducing spectral efficiency. On the other hand, high mobility shortens the channel coherence time, causing acquired channels to become invalid in a shorter period of time, necessitating a higher frequency of channel acquisition. These challenges make relying solely on pilot signals to estimate channel state prohibitively expensive, or even unacceptable. Wireless networks urgently need more efficient channel acquisition methods. A key approach to improving channel acquisition efficiency is to fully exploit the spatiotemporal correlation of the channel during the acquisition process.

[0206] In one possible implementation, the spatiotemporal correlation of the channel can be exploited through neural network processing to reduce the reference signal overhead. Some implementation examples are provided below for description.

[0207] In implementation A, the network device can implement channel estimation with channel mapping enhancement through space-frequency correlation based on neural network processing.

[0208] In implementation method A, the channels on some antennas and subcarriers are estimated, and the channels on all antennas and subcarriers are obtained through interpolation and extension, thereby reducing the required pilot overhead. This method essentially utilizes the spatial-frequency domain correlation of the channel.

[0209] For example, for a high-dimensional MIMO-OFDM channel, after sending a reference signal, the network device can receive the measurement results of the reference signal by the terminal device to obtain channel information on some antennas and / or some subcarriers. After that, the network device can obtain channel information on other antennas and / or other subcarriers through neural network processing (such as interpolation processing, extension processing, etc.). Through this technical solution, the channels that need to occupy pilot resources for estimation are reduced from the channels on all antennas and carriers to some antennas and / or subchannels, which can reduce the overhead of the reference signal. Due to the high nonlinearity of space-frequency correlation, the method based on deep learning is significantly ahead in performance.

[0210] Implementation A above initially exploits the channel's space-frequency correlation. However, the channel acquisition results (i.e., the processing results obtained by the neural network) of this solution depend on the quality information of some antennas and / or some subcarriers obtained through reference signal measurement. If the wireless channel path information is too complex or there is a lot of noise interference in the estimation, this solution will face serious performance degradation.

[0211] Implementation method B: Network equipment can achieve channel prediction through time domain correlation based on neural network processing.

[0212] In implementation B, the user channel is calculated based on the current user's location tag by learning the relationship between the user's location coordinates and the channel. Sequence-based channel prediction: Based on the user's past channel states, it is assumed that the user's motion follows a certain motion model to "predict" the user's current channel information.

[0213] For example, for a high-dimensional MIMO-OFDM channel, network equipment transmits reference signals in historical time slots and receives terminal device measurements of these reference signals to obtain channel information for those time slots. Subsequently, the network equipment uses neural network processing to infer the channel state for the current or future time slots using this past time slot channel state information. This technology eliminates the need to expend communication resources to acquire the channel in the current time slot, significantly reducing the cost of channel acquisition. Because the spatiotemporal relationships that need to be modeled are highly complex, channel prediction methods rely heavily on deep learning technology.

[0214] In Implementation B described above, due to factors such as the randomness of user movement and the changing scattering environment, this solution often lacks accuracy in inferring and utilizing the temporal correlation of the channel, resulting in poor channel acquisition quality. Furthermore, as channel acquisition errors accumulate, the quality of channel predictions steadily deteriorates, making it difficult to meet long-term communication requirements. Consequently, practical deployment and application remain difficult.

[0215] As can be seen from the solution description of the above implementation example, the channel information obtained by the above implementation example may still have some problems:

[0216] In implementation method A, by mining and utilizing the space-frequency correlation within MIMO-OFDM, the pilot overhead required for channel estimation in the current time slot can be reduced to a certain extent. However, it ignores the potential reference information in the channel of several past time slots, and its channel acquisition results are completely dependent on the quality of some sub-channels obtained by pilot estimation; therefore, the application of this channel acquisition method is often very limited. In implementation method B,

[0217] Implementation B assumes a deterministic relationship between the current channel and the channels of past time slots. However, due to the inevitable randomness inherent in the communication process within a given time slot, simply modeling this deterministic relationship often fails to reflect reality, leading to severe performance degradation in practical applications. Furthermore, continuously extrapolating current and future channels from the past inevitably introduces randomness and errors, leading to a difficult-to-solve error accumulation problem.

[0218] Furthermore, the method provided by Implementation B undoubtedly has value in mining and utilizing historical channel information. However, pure prediction discards communication information from the current time slot, making the channel results obtained based on Implementation B susceptible to the randomness of communication in that time slot and severely corrupted. In contrast, the method provided by Implementation A uses information obtained during the communication process itself as labels, eliminating the need for additional label information. However, it neglects the utilization of time-domain correlation, resulting in incomplete savings in communication resources and insufficient utilization of channel correlation. Therefore, both Implementation A and Implementation B still have some unresolved issues.

[0219] In order to solve the above problems, the present application provides a communication method and related equipment, which will be described in detail below with reference to the accompanying drawings.

[0220] Please refer to FIG3 , which is a schematic diagram of an implementation of the communication method provided in this application. The method includes the following steps.

[0221] It should be noted that this application uses network devices and terminal devices as examples of the execution subjects of the interactive diagram to illustrate the method provided by this application, but this application does not limit the execution subjects of the interactive diagram. For example, the method executed by the network device can also be executed by a module of the network device (such as a chip, a chip system, or a processor), and can also be implemented by a logical node, a logical module, or software that can implement all or part of the network device. The method executed by the terminal device can also be executed by a module of the terminal device (such as a chip, a chip system, or a processor), and can also be implemented by a logical node, a logical module, or software that can implement all or part of the terminal device functions.

[0222] The method shown in FIG3 includes steps S301 to S303 , and each step will be described below.

[0223] S301. The network device sends a first reference signal, and correspondingly, the terminal device receives the first reference signal.

[0224] S302. The terminal device sends first channel information, and the network device receives the first channel information accordingly. The first channel information is determined based on the first reference signal, and the channel dimension represented by the first channel information is [T1, R1, F1], where T1, R1, and F1 are all positive numbers.

[0225] S303. The network device processes the first channel information and the Q second channel information based on the first neural network to obtain third channel information.

[0226] The Q second channel information are determined by Q second reference signals, where Q is a positive integer; the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, and the channel dimension represented by the third channel information is [T, R, F], where T, R, and F are all positive numbers; the parameters in the channel dimension represented by the first channel information and the parameters in the channel dimension represented by the third channel information satisfy at least one of the following: T is greater than T1, R is greater than R1, and F is greater than F1.

[0227] It should be understood that the channel dimensions represented by the channel information may include multiple dimensions such as dimensions in the spatial domain, dimensions in the frequency domain, and dimensions in the time domain. Generally, the dimensions in the spatial domain may include the dimensions of the antenna port of the signal transmitting end (i.e., the transmitting antenna dimensions) and the dimensions of the antenna port of the signal receiving end (i.e., the receiving antenna dimensions). In the present application, the channel dimensions represented by the channel information include three channel dimensions as an example for explanation. One channel dimension is represented as T or T1 (or other parameters represented by T that may appear later, such as T2 / T3 / T4 / T5, etc.), another channel dimension is represented as R or R1 (or other parameters represented by R that may appear later, such as R2 / R3 / R4 / R5, etc.), and another channel dimension is represented as F or F1 (or other parameters represented by F that may appear later, such as F2 / F3 / F4 / F5, etc.).

[0228] In an implementation example, the spatial transmission process of the reference signal includes the process of the antenna port of the signal transmitting end transmitting the reference signal and the process of the antenna port of the signal receiving end receiving the reference signal. That is, the dimension of the channel information in the spatial domain can correspond to the transceiver port of the reference signal. In other words, the dimension of the reference signal in the spatial domain can be expressed as [T, R] in [T, R, F], where T (or T1, or other parameters represented by T that may appear later, such as T2 / T3 / T4 / T5, etc.) represents the antenna port of the signal transmitting end, and R (or R1, or other parameters represented by R that may appear later, such as R2 / R3 / R4 / R5, etc.) represents the antenna port of the signal receiving end.

[0229] In another implementation example, during the transmission and reception of a reference signal, the dimension of the reference signal in the frequency domain may correspond to the frequency domain resource carrying the reference signal. In other words, the dimension of the reference signal in the spatial domain may be represented as [F] in [T, R, F], where F (or F1, or other parameters denoted by F that may appear later, such as F2 / F3 / F4 / F5, etc.) represents the frequency domain resource carrying the reference signal (e.g., one or more frequency domain units contained in the frequency domain resource).

[0230] It should be noted that, in the above implementation process, the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, which can be understood as at least one of the following: the Q second reference signals are sent by the network device before the network device sends the first reference signal, the Q second reference signals are received by the terminal device before the terminal device receives the first reference signal, and the index number of the time domain resource carrying the Q second reference signals is smaller than the index number of the time domain resource carrying the first reference signal.

[0231] Optionally, the index number of the time domain resource may be any one of the following: an index number of a system frame, an index number of a subframe, an index number of a time slot, an index number of an orthogonal frequency division multiplexing (OFDM) symbol, etc.

[0232] It should be noted that the first neural network is used to process lower-dimensional channel information (such as first channel information) to obtain higher-dimensional channel information (such as third channel information). Accordingly, the first neural network can be understood as a neural network that performs channel prediction.

[0233] In this application, channel prediction can be replaced by other terms, such as channel estimation, channel pre-estimation, channel simulation recovery, channel reconstruction, channel acquisition, channel inference, etc.

[0234] In this application, neural network can be replaced by other terms, such as AI network, machine learning network, AI learning network, AI neural network, etc.

[0235] In one possible implementation, the channel dimension represented by at least one of the Q second channel information is [T2, R2, F2], where T2, R2, and F2 are all positive numbers; the parameters in the channel dimension represented by the second channel information satisfy at least one of the following: T is greater than T2, R is greater than R2, and F is greater than F2. Specifically, the Q second channel information are determined by Q second reference signals, wherein the channel dimension represented by at least one of the Q second channel information satisfies at least one of the above items, that is, the channel dimension represented by the at least one second channel information is lower than the channel dimension represented by the third channel information. Moreover, the Q second channel information is one of the input data of the first neural network. Thus, in this way, the reference signal overhead in the process of acquiring the input data of the neural network can be reduced.

[0236] In one possible implementation, the parameters in the channel dimension represented by the first channel information and the parameters in the channel dimension represented by the second channel information satisfy at least one of the following: the values ​​of T1 and T2 are the same, the values ​​of R1 and R2 are the same, and the values ​​of F1 and F2 are the same. Specifically, the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, that is, the Q second channel information is historical channel information of the first channel information, and the parameters in the two channel information satisfy at least one of the above items, which enables the network device to obtain the two channel information based on at least one identical dimension, thereby reducing the implementation complexity.

[0237] Based on the technical solution shown in Figure 3, after the network device sends the first reference signal in step S301, the network device can receive first channel information determined based on the first reference signal in step S302. Thereafter, in step S303, the network device can process the first channel information and Q second channel information based on the first neural network to obtain third channel information. In addition, the channel dimension represented by the first channel information is [T1, R1, F1], and the channel dimension represented by the third channel information is [T, R, F], satisfying at least one of the following: T is greater than the T1, R is greater than R1, and F is greater than F1. In other words, the network device can obtain third channel information of a higher channel dimension based on the first channel information of a lower channel dimension, and the first channel information of the lower channel dimension is determined by the first reference signal. Therefore, the network device can obtain higher-dimensional channel information through neural network processing based on the channel information of the lower channel dimension determined by the reference signal, which can reduce the reference signal overhead, thereby improving resource utilization and reducing the power consumption of the terminal device.

[0238] In addition, in the process of the above-mentioned network device determining the third channel information of higher dimension through the first neural network, the input data of the first neural network may include the first channel information of lower dimension and Q second channel information, and the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, that is, the Q second channel information is historical channel information of the first channel information. Thus, the first neural network can utilize the time domain correlation between the Q second channel information and the first channel information, and can also utilize the spatial domain correlation and / or frequency domain correlation between the first channel information and the third channel information, thereby improving the accuracy of the neural network in obtaining the higher dimensional channel information based on the channel information of lower channel dimension.

[0239] FIG4 a shows an example implementation of the processing of the first neural network in step S303. In FIG4 a , the input of the first neural network includes first channel information and second channel information, and the output of the first neural network includes third channel information. The first channel information is lower-dimensional channel information, and the third channel information is higher-dimensional channel information.

[0240] Optionally, as can be seen from the above implementation process, in Figure 4a, the second channel information may also be channel information of a lower dimension and the third channel information may be channel information of a higher dimension. In addition, in Figure 4a, at least one channel dimension of the second channel information may be the same as at least one channel dimension of the first channel information to reduce implementation complexity.

[0241] In summary, in the technical solution shown in FIG3, compared with the above-mentioned implementation method A and implementation method B, it is possible to fully utilize the channel information from the past time slot and the partial information of the current time slot, and to achieve accurate and real-time channel information acquisition with low communication resource occupancy. In addition, by introducing the channel information of the past time slot, the correlation of the mobile channel in the time domain is efficiently utilized. At the same time, by introducing partial information of the current time slot, the inference method can well cope with and offset the random changes that may exist in the current time slot channel, and realize a channel acquisition method that deeply mines spatiotemporal correlations. In other words, in the technical solution shown in FIG3, the respective advantages of implementation method A and implementation method B can be combined, and the communication reference signal of the current time slot can be appropriately introduced to resist the randomness of the current time slot, and the information in the channel of the past time slot can be utilized to realize a channel acquisition solution that occupies less communication resources and can resist communication randomness.

[0242] In a possible implementation of step S303 shown in Figure 3, the first neural network used to determine the third channel information includes a first sub-neural network and a second sub-neural network; the network device processes the first channel information and Q second channel information based on the first neural network to obtain the third channel information, including: the network device processes the first channel information based on the first sub-neural network to obtain fourth channel information; wherein the channel dimension represented by the fourth channel information is [T, R, F]; the network device processes the fourth channel information and the Q second channel information based on the second sub-neural network to obtain the third channel information.

[0243] Specifically, in the process of the network device obtaining the third channel information based on the first neural network, the network device may process the first channel information based on the first sub-neural network to obtain the fourth channel information, and then process the fourth channel information and the Q second channel information based on the second sub-neural network to obtain the third channel information. Thus, the network device may obtain the fourth channel information of a higher dimension based on the lower dimensional first channel information obtained by the reference signal through the processing of the first sub-neural network, and then further process the fourth channel information and the Q second channel information obtained historically based on the second sub-neural network to obtain the third channel information.

[0244] FIG4b shows an implementation example of the processing of the first neural network in step S303. Compared to the implementation shown in FIG4a, in FIG4b, the first neural network includes a first sub-neural network and a second sub-neural network, wherein the input of the first sub-neural network includes first channel information, the output of the first sub-neural network includes fourth channel information, and the input of the second sub-neural network includes Q second channel information and the fourth channel information. The first channel information is lower-dimensional channel information and the third channel information is higher-dimensional channel information.

[0245] Optionally, as can be seen from the above implementation process, in FIG4b, the second channel information may also be channel information of a lower dimension and the third channel information may be channel information of a higher dimension. In addition, in FIG4b, at least one channel dimension of the second channel information may be the same as at least one channel dimension of the first channel information to reduce implementation complexity.

[0246] In one possible implementation, in step S303 shown in FIG3 , after the network device processes the first channel information and Q second channel information based on the first neural network to obtain third channel information, the method further includes: the network device processes the third channel information based on the second neural network to obtain fifth channel information, where the channel dimension represented by the fifth channel information is [T, R, F]. Specifically, after the network device obtains the third channel information of higher dimension based on the first neural network, the network device may further process the third channel information based on the second neural network to obtain fifth channel information of the same channel dimension. Thus, by optimizing the second neural network, the accuracy of the higher-dimensional channel information obtained based on the neural network can be further improved.

[0247] It can be understood that the second neural network is used to process the input channel information (e.g., the third channel information) to obtain output channel information (e.g., the fifth channel information). Compared to the input channel information, the output channel information is closer to the channel information of the transmission channel in the actual space. In other words, the second neural network can be a neural network for optimizing the channel information.

[0248] In one possible implementation, the deviation between the fifth channel information and the sixth channel information is less than or equal to the deviation between the third channel information and the sixth channel information, the channel dimension represented by the sixth channel information is [T, R, F], and the sixth channel information is obtained by measuring the reference signal. Specifically, the sixth channel information is obtained by measuring the reference signal, and the third channel information and the fifth channel information are obtained based on the second neural network. Therefore, compared with the third channel information and the fifth channel information, the sixth channel information is closer to the channel information of the transmission channel in the actual space. In the above technical solution, the second neural network is used to optimize the third channel information to obtain the fifth channel information, so that the deviation between the fifth channel information and the sixth channel information is less than or equal to the deviation between the third channel information and the sixth channel information, that is, compared with the third channel information, the fifth channel information is closer to the channel information of the transmission channel in the actual space.

[0249] Figure 4c shows an example implementation of the processing of the first neural network and the second neural network. Compared to the implementation shown in Figure 4a, in Figure 4c, after the first neural network processes to obtain third channel information, this third channel information can also be used as input to the second neural network, and is processed by the second neural network to obtain fifth channel information.

[0250] Figure 4d shows an example implementation of the processing of the first neural network and the second neural network. Compared to the implementation shown in Figure 4b, in Figure 4d, after the first and second sub-neural networks in the first neural network are processed to obtain third channel information, this third channel information can also be used as input to the second neural network, and processed by the second neural network to obtain fifth channel information.

[0251] Optionally, as can be seen from the above implementation process, in FIG. 4 c and FIG. 4 d , the channel dimension of the third channel information and the channel dimension of the fifth channel information may be the same.

[0252] As shown in Figure 5a, which is an application example of Figure 4d, in Figure 5a, the first neural network sub-sub ...

[0253] In addition, in FIG5a, the first channel information of the lower channel dimension may represent the channel information of the t-th time unit (the time unit may be a frame, a sub-array, a time slot, a symbol, etc.) as H′ t , that is, channel information obtained based on a small amount of communication resources (such as fewer transmitting port resources, fewer receiving port resources, fewer frequency domain resources, etc.); Q second channel information can be expressed as channel information on Q time units before the t-th time unit, that is, arrive in, arrive It can be the second channel information on the Q time units. The second channel information can be the channel information obtained based on a small amount of communication resources, or it can be the historical processing result of multiple executions of neural network processing (multiple neural network processing can refer to the implementation process of Figure 6a or Figure 6b later).

[0254] Optionally, the Q time units may be continuous time units or discontinuous time units, which is not limited here. Furthermore, the time intervals between two adjacent time units in the Q time units may be the same or different, which is not limited here.

[0255] The solution shown in Figure 5a can be described as inferring the complete high-dimensional channel information for the current time slot based on the channel information from several past time units and the estimated partial channel information for the current time unit (i.e., the tth time unit), thereby fully utilizing the time-space-frequency correlation. Furthermore, the solution shown in Figure 5a can include the following three steps.

[0256] The first step is to calculate the sparse channel measurement result H′ at the current moment. t After passing through the preliminary mapping network, it is converted into information with complete channel information dimension At this time, spatial-frequency correlation is used.

[0257] The second step is to extract temporal correlation in the past-present information interaction network using an attention-based network (such as a transformer structure) and output information

[0258] Step 3, After the detailed representation network, the final channel information is obtained

[0259] In other words, the functions and input and output implementation examples of the three network modules shown in Figure 5a are as described in Figures 5b to 5c below. For the convenience of subsequent reference, in the following example, the preliminary mapping network is recorded as module 1, the past-present information interaction network is recorded as module 2, and the detailed representation network is recorded as module 3.

[0260] The example shown in Figure 5b is an implementation example of module 1. In Figure 5b, module 1 can expand the sparse channel measurement to the same dimension as the complete channel information through neural network processing. The input of this module 1 is the sparse channel measurement result H′ at the current moment. t , after processing by the preliminary mapping network (in this example, the preliminary mapping network includes a per-dimensional fully-connected subnetwork and a multi-layer perceptron (MLP) subnetwork), the output is a dimensionality with complete channel information Exemplarily, the module 1 may adopt one or more network architectures such as a multi-layer perceptron hybrid (MLPmixer) network structure, an MLP-based network structure, a CNN-based network structure, and a Transformer-based network structure.

[0261] Figure 5c shows an implementation example of module 2. After processing by module 1, the current channel and the past channel have the same dimensions—that is, the dimensions of the complete channel information. These are then input into module 2 as a sequence. As shown in Figure 5c, the past channel is at the beginning of the sequence, and the current channel is at the end.

[0262] It is worth noting that the solution of the present application does not limit the order of the channel information of the past moments (ie, Q second channel information) in the sequence; their placement order can be arbitrary, ie, permutation invariance is maintained, and the number Q of past moments is arbitrary.

[0263] Optionally, in the example shown in Figure 5c, it is not necessary to introduce a time tag. For example, the terminal device moves freely in a small range for 1 hour, and collects channels at 5 moments during this period. At this time, if you want to predict the user's channel at the next moment, because the time span is large and the current position of the terminal device is unknown, even if you know the timestamps of the previous 5 moments, it may not help (or help very little) for this prediction. For this reason, the assumption that time domain correlation will bring benefits is that the closer the historical moment is to the current moment, the greater the correlation with the current moment. In the example shown in Figure 5c, it is possible that the current moment of the terminal device is exactly at the location where the first historical moment was collected, that is, the correlation between the first historical moment and the current moment is the largest, so this example does not need to introduce a time tag.

[0264] In addition, in the example shown in Figure 5c, the first layer of module 2 can be a reshape & fully-connected layer. It first transforms the channel matrix into a vector (assuming that the dimension of the channel information is n (number of antennas) * m (number of carriers), and the dimension of the vector is 2*n*m (real part and imaginary part)). Then, the vector is passed through a fully connected layer to obtain a vector with the embedding size. This allows the sizes of different inputs to be transformed into a unified low dimension, facilitating subsequent network processing.

[0265] Thereafter, in the first addition processing module of the second layer, this vector can be added (Add) to the position embedding vector (for example, the position embedding vector can include the "past / present embedding vector"). The result of the addition (or a backup, copy, etc. of the result) can be used as the input of the first normalization layer (Layer Norm), and after being processed by the multi-head attention layer (Multi-head attention), the processing result can be used as the input of the second addition processing, and the result of the addition (or a backup, copy, etc. of the result) can also be used as the input of the second addition processing. Similarly, the processing result of the second addition processing (or a backup, copy, etc. of the result) can be used as the input of the second normalization layer, and after being processed by the feed forward (Feed Forward), the processing result can be used as the input of the third addition processing; the processing result of the second addition processing (or a backup, copy, etc. of the result) can also be used as the input of the third addition processing. Thereafter, the output of the third addition processing can be used as the input of the third layer network.

[0266] Optionally, the third layer of the network can be represented as (Last output) Fully-connected & Reshape, and the output of the third layer of the network can be

[0267] Optionally, there are one or more networks in the second layer. When there are multiple networks, different networks can be executed in cascade (or series).

[0268] Optionally, the position vectors in the above example include two types, namely the vectors of the past moment and the current moment, that is, the same position embedding vector can be used for multiple past moments. In the processing of the next few layers, the scheme is basically the same as the encoder of the Transformer network. After the multi-head attention layer and the fully connected layer (feed forward), the channel information of the past moment and the current moment undergoes information fusion and feature extraction, and finally outputs a vector. The vector is processed by the fully connected layer and then deformed to obtain a vector with complete channel information dimension.

[0269] The example shown in Figure 5d is an implementation example of module 3. Module 3: In order to further improve the accuracy, the output of module 2 can be After processing by module 3, the final channel information of the same dimension is output Exemplarily, module 3 may adopt one or more network architectures such as an MLPmixer network structure, an MLP-based network structure, a CNN-based network structure, and a Transformer-based network structure.

[0270] It is worth noting that module 3 is optional. Exemplarily, the use of module 3 (i.e., the detail representation network, which can also be the second neural network in Figure 4c / Figure 4d above) is to further improve the accuracy of channel acquisition. Its input and output dimensions are consistent, so only module 1 and module 2 can be retained. When the accuracy of channel acquisition is high, the error is less than a preset threshold, or the communication system has strict requirements on computational complexity (for example, the computing power of IoT devices is weak, and the number of operations is limited when the terminal device is low on power), the method of this embodiment can be used to reduce computational complexity and improve channel acquisition efficiency.

[0271] In one possible implementation, in the method shown in FIG3 , the method may further include: the network device sends a third reference signal; the network device receives channel information corresponding to the third reference signal, the channel dimension represented by the channel information corresponding to the third reference signal is [T3, R3, F3], and T3, R3, and F3 are all positive numbers; the channel dimension represented by the channel information corresponding to the third reference signal satisfies at least one of the following: T is greater than T3, R is greater than R3, and F is greater than F3; the network device processes the channel information corresponding to the second reference signal, part or all of the Q second channel information, and the third channel information based on the first neural network to obtain seventh channel information; wherein the channel dimension represented by the seventh channel information is [T, R, F]. Specifically, after the network device obtains the third channel information based on the first neural network, the network device may further send a third reference signal and receive the channel information corresponding to the third reference signal. Subsequently, the network device may obtain the seventh channel information based on the channel information corresponding to the second reference signal. The channel dimension represented by the seventh channel information is [T, R, F], and the channel dimension represented by the channel information corresponding to the third reference signal satisfies at least one of the following: T is greater than T3, R is greater than R3, and F is greater than F3. Thus, the network device can iteratively run the first neural network and, based on multiple transmissions of the reference signal, obtain higher-dimensional channel information multiple times based on lower-dimensional channel information.

[0272] As shown in Figure 6a, it is an implementation example of the multiple processing process of the first neural network. Compared with the implementation process shown in Figure 4a, in Figure 6a, after the first neural network is processed to obtain the third channel information, the third channel information can also be used as the input of the next first neural network, and the seventh channel information is obtained after the processing of the next first neural network. Among them, the input of the next first neural network also includes the channel information corresponding to the third reference signal, that is, by iteratively running the first neural network, based on multiple reference signals (such as the first reference signal and the third reference signal), multiple acquisitions of higher-dimensional channel information based on lower-dimensional channel information are achieved.

[0273] As shown in Figure 6b, another implementation example of the multiple processing process of the first neural network is shown. Compared with the implementation process shown in Figure 6b, in Figure 6b, after the first sub-neural network and the second sub-neural network in the first neural network are processed to obtain the third channel information, the third channel information can also be used as the input of the next first neural network, and the seventh channel information is obtained after being processed by the first sub-neural network and the second sub-neural network in the next first neural network. Among them, the input of the next first neural network also includes the channel information corresponding to the third reference signal, that is, by iteratively running the first neural network, based on multiple reference signals (such as the first reference signal and the third reference signal), multiple acquisitions of higher-dimensional channel information based on lower-dimensional channel information are achieved.

[0274] It should be understood that the multiple processing processes in the above-mentioned Figures 4c and 4d can also be implemented by the method shown in Figures 6a and 6b. The difference from Figures 6a and 6b is that in the implementation process of Figures 4c and 4d, after each processing of the first neural network, the processing results of the first neural network can be optimized by the second neural network, and the optimized results are used as input for the next iterative processing.

[0275] In one possible implementation, the channel dimension represented by the first channel information and the channel dimension represented by the channel information corresponding to the third reference signal satisfy at least one of the following: T1 and T3 have the same value, R1 and R3 have the same value, and F1 and F3 have the same value. Specifically, during multiple reference signal transmissions, the channel dimensions of different channel information obtained based on different reference signals can satisfy at least one of the above conditions, i.e., the different channel information includes at least one common dimension, thereby reducing implementation complexity.

[0276] In FIG3 and the related implementation process, the neural network used by the network device (such as the first neural network, the second neural network, etc.) can realize the process of determining the channel information of a higher dimension with high accuracy. Ideally, when the accuracy of the prediction performed by the neural network is very high, the channel information determined by the neural network is the same as the channel information of the transmission channel in the actual space. However, in actual applications, there may be more or less deviations between the channel information determined by the neural network and the channel information of the transmission channel in the actual space. To this end, in addition to the implementation process shown in FIG3 above, the performance of the neural network can also be tested and / or optimized. Some implementation methods will be provided below for detailed description.

[0277] Implementation method 1: The network device detects and / or optimizes the performance of the neural network based on the neural network deployed on the terminal device.

[0278] In implementation mode one, the method shown in Figure 3 may further include: the network device sends a fourth reference signal; the network device receives first information, the first information being used to indicate a deviation between the channel information corresponding to the fourth reference signal and the eighth channel information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on the third neural network, the P ninth channel information being determined by P reference signals, and the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

[0279] Specifically, the network device may further transmit a fourth reference signal, and the eighth channel information is channel information obtained by processing the channel information corresponding to the fourth reference signal by a receiver (i.e., a terminal device) of the fourth reference signal based on the third neural network. Furthermore, the network device may further receive first information indicating a deviation between the channel information corresponding to the fourth reference signal and the eighth channel information, so that the network device can determine the performance of the channel prediction of the third neural network based on the first information.

[0280] In addition, compared to the method in which the network device receives the channel information corresponding to the fourth reference signal and processes the channel information corresponding to the fourth reference signal based on the first neural network to determine the performance of the channel prediction of the first neural network, in the above technical solution, the network device can determine the performance of the channel prediction of the neural network through the first information. Since the transmission overhead of the first information is less than the transmission overhead of the channel information corresponding to the fourth reference signal, the transmission overhead can be reduced.

[0281] It should be understood that the first information is used to indicate the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information, where the eighth channel information is channel information obtained through processing based on the third neural network. Accordingly, the eighth channel information can be understood as the result of channel prediction performed by the third neural network, and the channel information corresponding to the fourth reference signal is channel information of the transmission channel in actual space. Therefore, the deviation indicated by the first information can be understood as the first information indicating the performance test result of the third neural network performing channel prediction.

[0282] Optionally, the network structure and / or network parameters of the first neural network deployed on the network device and the third neural network deployed on the terminal device are the same.

[0283] Optionally, the implementation process of the third neural network deployed in the terminal device can refer to the implementation process of the first neural network described above, including but not limited to the examples shown in Figures 4a to 4d, Figures 5a to 5d, and Figures 6a to 6b. For example, historically obtained channel information and lower-dimensional channel information can be used as inputs to the third neural network. After processing by the third neural network, the third neural network can output higher-dimensional channel information.

[0284] In one possible implementation, the channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], where T4, R4, and F4 are all positive numbers; the channel dimension represented by the channel information corresponding to the fourth reference signal and the channel dimension represented by the channel information corresponding to the first reference signal (i.e., the first channel information) satisfy at least one of the following: T is greater than or equal to T4 and T4 is greater than T1, R is greater than or equal to R4 and R4 is greater than R1, and F is greater than or equal to F4 and F4 is greater than F1.

[0285] Specifically, the channel dimension represented by the channel information corresponding to the fourth reference signal and the channel dimension represented by the channel information corresponding to the first reference signal (i.e., the first channel information) satisfy at least one of the above items, that is, the channel information corresponding to the fourth reference signal is channel information of higher dimension than the channel information corresponding to the first reference signal (i.e., the fourth reference signal is a reference signal of higher density than the first reference signal), so as to enable the terminal device to detect the channel prediction performance of the third neural network through the reference signal of higher density.

[0286] In addition, the channel information corresponding to the fourth reference signal is channel information of lower dimension than the channel information corresponding to the eighth reference signal, which can reduce the overhead in the detection process of the channel prediction performance of the neural network.

[0287] In one possible implementation, after the network device receives the first information, the method may further include: the network device sending second information, where the second information is used to indicate whether to adjust the input parameters of the third neural network; wherein the second information is determined based on the first information.

[0288] Specifically, after receiving the first information, the network device may also send second information for indicating whether to adjust the input parameters of the third neural network, so that the network device can adjust the input parameters of the third neural network deployed on the terminal device based on the performance detection results of the third neural network.

[0289] Exemplarily, the input parameters of the third neural network may include the amount of historical channel information input to the third neural network, the channel dimension of the historical channel information input, the channel dimension of the channel information with lower channel dimension, etc.

[0290] In one possible implementation, before the network device receives the first information, the method may further include: the network device sends third information, where the third information indicates channel information corresponding to the fourth reference signal processed based on the third neural network; wherein the third information is determined based on the first information.

[0291] Specifically, before receiving the first information, the network device may also send third information for instructing the processing of the channel information corresponding to the fourth reference signal based on the third neural network, so that the terminal device initiates performance detection of the channel information corresponding to the fourth reference signal based on the third information.

[0292] In a second implementation method, the network device detects and / or optimizes the performance of the neural network based on a locally deployed neural network (eg, the first neural network).

[0293] In implementation mode two, the method shown in Figure 3 may further include: the network device sends a fourth reference signal; the network device receives channel information corresponding to the fourth reference signal; the channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], and T4, R4, and F4 are all positive numbers; at least one of the following items is satisfied: T is greater than or equal to T4 and T4 is greater than T1, R is greater than or equal to R4 and R4 is greater than R1, and F is greater than or equal to F4 and F4 is greater than F1; the network device processes the channel information corresponding to the fourth reference signal and part or all of the Q second channel information based on the first neural network to obtain tenth channel information; wherein the channel dimension represented by the tenth channel information is [T, R, F]; the network device determines the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information.

[0294] Specifically, the network device may further transmit a fourth reference signal and receive channel information corresponding to the fourth reference signal. Furthermore, the network device may process the channel information corresponding to the fourth reference signal based on the first neural network to obtain tenth channel information. Furthermore, the network device may further determine a deviation between the channel information corresponding to the fourth reference signal and the tenth channel information, so that the network device can determine the performance of the channel prediction of the first neural network based on the deviation.

[0295] In one possible implementation, after the network device processes the channel information corresponding to the fourth reference signal and part or all of the Q second channel information based on the first neural network to obtain tenth channel information, the method may further include: the network device sending a fifth reference signal, where the channel dimension represented by the channel information corresponding to the fifth reference signal is determined based on the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information. Specifically, the network device may further send the fifth reference signal based on the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information, that is, the network device may update or adjust the sending of the reference signal based on the performance of the channel prediction of the first neural network.

[0296] In one possible implementation, when the network device sends a fifth reference signal, the channel dimension represented by the channel information corresponding to the fifth reference signal is [T5, R5, F5], where T5, R5, and F5 are all positive numbers; the channel dimension represented by the channel information corresponding to the fifth reference signal is determined based on the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information, including: when the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is good, at least one of the following is satisfied: T5 is less than or equal to T4, R5 is less than or equal to R4, and F5 is less than or equal to F4; when the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is poor, at least one of the following is satisfied: T5 is greater than or equal to T4, R5 is greater than or equal to R4, and F5 is greater than or equal to F4.

[0297] Specifically, when the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is better, that is, the network device determines that the channel prediction performance of the first neural network is better. For this reason, after sending the fourth reference signal, the network device can send a fifth reference signal with the same or smaller pilot (in this application, "pilot" can be replaced by "reference signal") density to reduce overhead. Similarly, when the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is worse, that is, the network device determines that the channel prediction performance of the first neural network is worse. For this reason, after sending the fourth reference signal, the network device can send a fifth reference signal with the same or larger pilot density to facilitate the network device to obtain channel information with a higher channel dimension, and optimize the input parameters of the first neural network based on the channel information with a higher channel dimension to improve the channel prediction performance of the first neural network.

[0298] In one possible implementation, after the network device processes the channel information corresponding to the fourth reference signal and part or all of the Q second channel information based on the first neural network to obtain tenth channel information, the method may further include: the network device adjusts the input parameters of the first neural network based on the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information.

[0299] Specifically, since the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information can reflect the channel prediction performance of the first neural network, the network device can also adjust the input parameters of the first neural network based on the deviation, and can optimize the input parameters of the first neural network to improve the channel prediction performance of the first neural network.

[0300] Exemplarily, the input parameters of the first neural network may include the amount of historical channel information input to the first neural network (i.e., the value of Q), the channel dimension of the historical channel information input (i.e., the channel dimension of the second channel information), the channel dimension of the channel information with lower channel dimension (i.e., the channel dimension of the first channel information), etc.

[0301] Optionally, the deviation between two channel information (e.g., the deviation between the fifth channel information and the sixth channel information, the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information, the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information, etc.) can be reflected by a mathematical operation result of the two channel information. For example, the deviation can be a normalized mean square error (NMSE), a mean square error (MSE), etc.

[0302] For example, let's take the deviation as NMSE. The current complete channel information estimated by the neural network will be input into the neural network as the channel information at the past moment, that is, the output of the current neural network will be used as the input of the subsequent neural network. Therefore, there is an error accumulation problem. If the error NMSE of the current channel (denoted as NMSE_A) is smaller than the error NMSE of the channel at the past moment (denoted as NMSE_B), the channel acquisition can continue to run, and the estimation performance will be improved (the estimation error will be reduced). If the error NMSE_A of the current channel is larger than the error NMSE_B of the channel at the past moment, due to error accumulation, the performance of the channel acquisition will become worse and worse until it fails to work. If the error NMSE_A of the current channel is equal to the error NMSE_B of the channel at the past moment, then the channel acquisition will maintain stable performance and can continue to run.

[0303] Furthermore, if the error increases and channel acquisition performance becomes unstable, the density of pilot signals placed on the air interface resources or the number of input samples n in the past can be increased to make the current measured channel more accurate, thereby reducing NMSE_A and ensuring stable channel acquisition. Therefore, the communication system needs to monitor the error and performance of the current channel acquisition.

[0304] Exemplarily, the network device can place a higher density of pilots on the air interface resources at regular intervals (for example, the pilot density of the fourth reference signal is higher than the pilot density of the first reference signal), and the network device notifies the terminal device to start performance monitoring of channel acquisition (for example, the notification is implemented through the third information in the first implementation method above). Among them, the extreme scenario of high-density pilots is to fill the air interface resources allocated to the terminal device with pilots. At this time, a channel matrix with complete channel information dimensions can be directly obtained by channel measurement. After that, the terminal device performs channel measurement of high-density pilots, and then uses traditional channel estimation methods (such as least squares (LS) estimation, Wiener filtering, etc.) to obtain a complete channel matrix, and regards this channel as the true channel value. The low-density part of the high-density pilot is still used for channel measurement, and the measurement results are input into the channel acquisition module to obtain a predicted channel matrix. The channel true value is compared with the predicted channel matrix to obtain the error NMSE_A at this time, and NMSE_A is compared with NMSE_B at the past moment to determine the performance trend of the current channel acquisition. If NMSE_A is greater than NMSE_B by more than a preset threshold, the terminal device believes that the performance of the current channel acquisition will be unsustainable and needs to find a way to reduce the error. At this time, the terminal device notifies the network device of the performance monitoring results, and the network device changes the pilot pattern to increase the pilot placement density, or increases the number of input samples n in the past, thereby improving the accuracy of channel acquisition. Next, the terminal device continues to monitor the channel error and performance at regular intervals. When NMSE_A is less than NMSE_B by more than a preset threshold, the terminal device can notify the network device to reduce the pilot density, or reduce the number of input samples n in the past, to avoid wasting air interface resources, because it only needs to ensure that NMSE_A is close to NMSE_B to ensure the continuous operation of the neural network channel acquisition.

[0305] Optionally, at the beginning of the channel acquisition task, since there may not be enough samples from the past between the network device and the terminal device, performance monitoring of channel acquisition can also be enabled. The main purpose at this time is to obtain an accurate channel matrix with a sufficient number of samples from the past for subsequent channel acquisition.

[0306] As can be seen from the description of the above implementation methods 1 and 2, the first information (or the parameters involved in the process of the network device determining the deviation) based on the feedback of the performance monitoring results in implementation method 1 can include the channel acquisition error value at the current moment, the error change of the channel acquisition, and the error threshold of the channel acquisition. After the network device obtains this information, it can make decisions based on this information, make changes to the subsequent communication configuration for channel acquisition (such as pilot density, the number of samples Q of historical channel information), and notify the terminal device. For example, the network device records the channel acquisition error value during each performance monitoring. When the network device receives a new channel acquisition error value, it can compare it with the channel acquisition error value at the past moment. If the change in the channel acquisition error value exceeds the error threshold of the channel acquisition (the error threshold can include an upper limit and a lower limit), the network device changes the communication configuration for channel acquisition accordingly. In addition, the error change of the channel acquisition can be calculated by the network device or sent to the network device after the terminal device calculates it. The calculation of the error change of channel acquisition needs to consider the channel acquisition error value at the current moment and the channel acquisition error value at at least one past moment. When using channel acquisition error values ​​at multiple past moments, the error change can be calculated by fitting and extrapolation methods.

[0307] Therefore, through the feedback of performance monitoring, channel acquisition performance feedback and configuration instructions are sent in the air interface signaling interaction, allowing network devices to flexibly control communication configuration, reduce resources and computing overhead when channel acquisition performance is good, and make timely adjustments when channel acquisition performance is poor to ensure the sustainability of channel acquisition.

[0308] In one possible implementation, the method also includes: the network device sends a sixth reference signal; the network device receives channel information corresponding to the sixth reference signal; the network device processes the channel information corresponding to the sixth reference signal and K eleventh channel information based on the fourth neural network to obtain twelfth channel information; wherein the K eleventh channel information is determined by K reference signals, and K is a positive integer; the time domain resource carrying the Kth reference signal is located before the time domain resource carrying the sixth reference signal; the channel dimension represented by the twelfth channel information is [T, R, F]; wherein the network structure of the first neural network is different from the network structure of the fourth neural network; or, the network structure of the first neural network is the same as the network structure of the fourth neural network, and the parameters in the network structure of the first neural network are different from the parameters in the network structure of the fourth neural network; the sixth reference signal satisfies at least one of the following conditions A to C.

[0309] Condition A: The resources carrying the sixth reference signal are different from the resources carrying the first reference signal.

[0310] Condition B: The first channel information and the channel information corresponding to the sixth reference signal come from different terminal devices.

[0311] Condition C: The first channel information and the channel information corresponding to the sixth reference signal come from terminal devices with different mobile information.

[0312] Specifically, in addition to transmitting the first reference signal, the network device may also transmit a sixth reference signal and perform channel prediction on the channel information corresponding to the sixth reference signal based on the fourth neural network to obtain twelfth channel information of a higher dimension. The transmission of the sixth reference signal satisfies at least one of the above conditions, thereby enhancing the flexibility of the solution implementation.

[0313] It should be noted that the fourth neural network may be the same as the first neural network used in step S303, i.e., the network structure of the first neural network is different from the network structure of the fourth neural network. Alternatively, the fourth neural network may be different from the first neural network used in step S303, i.e., the network structure of the first neural network is the same as the network structure of the fourth neural network, and the parameters in the network structure of the first neural network are different from the parameters in the network structure of the fourth neural network.

[0314] In addition, the sixth reference signal is different from the first reference signal sent in step S301. For ease of understanding, the above conditions A to C are described below as examples. For example, on the air interface time-frequency resource map, the network device can use different neural networks for channel information at different time-frequency resource locations and group them. The grouping method can be based on at least one of conditions A to C.

[0315] As an implementation example of condition A, resources of different reference signals may be grouped.

[0316] For example, different reference signal resources may include reference signal transmit port resources and / or reference signal receive port resources. To this end, different reference signal resources may correspond to different pilot densities. In other words, since different pilot densities correspond to different input dimensions of the neural network, channel acquisition requirements with the same pilot density can be grouped together. For example, the following Table 2 can be divided into different groups based on different pilot densities. Accordingly, in the above technical solution, the pilot density of the first reference signal and the pilot density of the sixth reference signal may be any two different reference signal pilot densities in Table 2.

[0317] Table 2

[0318] For another example, the resources of different reference signals may include the number of frequency domain resources. Take the number of frequency domain resources as the number of resource blocks (RBs) that carry reference signals as an example. The network device may allocate 10 RBs to some terminal devices, and 20 RBs to other terminal devices. The difference in the number of resources will cause the number of pilots and the dimension of the complete channel information to change, that is, the input and output dimensions of the neural network change. Therefore, the resources can be grouped according to the number of resources allocated to the user. For example, the following Table 3 can be divided into different groups based on different numbers of resources. Accordingly, in the above technical solution, the number of resources of the first reference signal and the number of resources of the sixth reference signal can be any two different resource numbers in Table 3.

[0319] Table 3

[0320] As an implementation example of condition B, different terminal devices may be grouped.

[0321] For example, the number of signal transmission paths (i.e., multipath) between different terminal devices and network devices is likely to be different. For this reason, the channel dimensions of the channel information corresponding to the reference signals fed back by different terminal devices input into the neural network may be different. Therefore, different terminal devices can be divided into different groups. For example, the following Table 4 can be divided into different groups based on different terminal devices. Accordingly, in the above technical solution, the terminal device that feeds back the channel information of the first reference signal (i.e., the first channel information) and the terminal device that feeds back the channel information of the sixth reference signal can be any two different terminal devices in Table 4.

[0322] Table 4

[0323] As an implementation example of condition C, terminal devices with different mobility information may be grouped.

[0324] For example, the speed and range of movement of the terminal device affect the accuracy of the neural network output results. Generally, this solution works better when the range is small and the speed is low. Therefore, users can be grouped according to their mobility.

[0325] An example is shown in Table 5, which can be divided into different groups based on different mobile speeds. Accordingly, in the above technical solution, the mobile speed of the terminal device that feeds back the channel information of the first reference signal (i.e., the first channel information) and the mobile speed of the terminal device that feeds back the channel information of the sixth reference signal can be any two different mobile speeds in Table 5.

[0326] Table 5

[0327] Another example is shown in Table 6, which can be divided into different groups based on different movement ranges. Accordingly, in the above technical solution, the movement range of the terminal device that feeds back the channel information of the first reference signal (i.e., the first channel information) and the movement range of the terminal device that feeds back the channel information of the sixth reference signal can be any two different movement speeds in Table 6.

[0328] Table 6

[0329] It is worth noting that grouping can also be performed based on a combination of the above conditions. For example, by using different neural networks for different scenarios, such as different pilot density, different number of resources, different user mobility information, etc., the flexibility and scalability of the solution can be increased.

[0330] Please refer to FIG7 , which is a schematic diagram of an implementation of the communication method provided in this application. The method includes the following steps.

[0331] The method shown in FIG7 includes steps S701 to S702 , and each step will be described below.

[0332] S701. The network device sends a fourth reference signal, and correspondingly, the terminal device receives the fourth reference signal.

[0333] S702. The terminal device sends first information, and the network device receives the first information accordingly. The first information is used to indicate a deviation between channel information corresponding to a fourth reference signal and eighth channel information; the channel dimension represented by the eighth channel information is [T, R, F], where T, R, and F are all positive numbers; and the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on a third neural network, the P ninth channel information being determined by P reference signals, the positions of time domain resources carrying the P reference signals being located before the positions of time domain resources carrying the fourth reference signal, and P being a positive integer.

[0334] Based on the technical solution shown in FIG7 , the network device may transmit a fourth reference signal in step S701, and the eighth channel information is channel information obtained by processing the channel information corresponding to the fourth reference signal by the receiver (i.e., the terminal device) of the fourth reference signal based on the third neural network. Furthermore, the network device may also receive first information indicating a deviation between the channel information corresponding to the fourth reference signal and the eighth channel information in step S702, so that the network device can determine the performance of the channel prediction of the third neural network based on the first information.

[0335] In addition, compared to the method in which the network device receives the channel information corresponding to the fourth reference signal and processes the channel information corresponding to the fourth reference signal based on the first neural network to determine the performance of the channel prediction of the first neural network, in the above technical solution, the network device can determine the performance of the channel prediction of the neural network through the first information. Since the transmission overhead of the first information is less than the transmission overhead of the channel information corresponding to the fourth reference signal, the transmission overhead can be reduced.

[0336] In one possible implementation, the channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], where T4, R4, and F4 are all positive numbers. The channel dimension represented by the channel information corresponding to the fourth reference signal and the channel dimension represented by the channel information corresponding to the eighth reference signal satisfy at least one of the following: T is greater than or equal to T4, R is greater than or equal to R4, and F is greater than or equal to F4. Specifically, the channel information corresponding to the fourth reference signal is of lower dimensionality than the channel information corresponding to the eighth reference signal, thereby reducing overhead in detecting the channel prediction performance of a neural network.

[0337] In one possible implementation, the method further includes: the network device sending second information indicating whether to adjust input parameters of the third neural network; wherein the second information is determined based on the first information. Specifically, after receiving the first information, the network device may further send second information indicating whether to adjust the input parameters of the third neural network, so that the network device adjusts the input parameters of the third neural network deployed on the terminal device based on the performance test results of the third neural network.

[0338] In one possible implementation, the method further includes: the network device transmitting third information indicating processing of channel information corresponding to the fourth reference signal based on the third neural network; wherein the third information is determined based on the first information. Specifically, before receiving the first information, the network device may further transmit third information indicating processing of the channel information corresponding to the fourth reference signal based on the third neural network, so that the terminal device can initiate performance testing of the channel information corresponding to the fourth reference signal based on the third information.

[0339] It should be noted that the implementation shown in FIG7 can also refer to the description of FIG3 and related implementations (such as the implementation process of the above implementation method 1).

[0340] Referring to Figure 8 , an embodiment of the present application provides a communication device 800. This communication device 800 can implement the functions of the communication device (the communication device being a terminal device or a network device) in the above-described method embodiment, thereby also achieving the beneficial effects of the above-described method embodiment. In this embodiment of the present application, the communication device 800 can be a communication device, or it can be an integrated circuit or component within the communication device, such as a chip.

[0341] It should be noted that the transceiver unit 802 may include a sending unit and a receiving unit, which are respectively used to perform sending and receiving.

[0342] In one possible implementation, when the apparatus 800 is used to execute the method executed by the network device in the aforementioned FIG. 3 and related embodiments, the apparatus 800 includes a processing unit 801 and a transceiver unit 802; the transceiver unit 802 is used to send a first reference signal; the transceiver unit 802 is further used to receive first channel information, the first channel information is determined based on the first reference signal, the channel dimension represented by the first channel information is [T1, R1, F1], T1, R1, F1 are all positive numbers; the processing unit 801 is used to perform the first channel information and Q based on the first neural network. The Q second channel information are processed to obtain third channel information; wherein, the Q second channel information are determined by Q second reference signals, and Q is a positive integer; the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, and the channel dimension represented by the third channel information is [T, R, F], and T, R, and F are all positive numbers; wherein, the parameters in the channel dimension represented by the first channel information and the parameters in the channel dimension represented by the third channel information satisfy at least one of the following: T is greater than T1, R is greater than R1, and F is greater than F1.

[0343] In one possible implementation, when the device 800 is used to execute the method executed by the network device in Figure 7 and related embodiments, the device 800 includes a processing unit 801 and a transceiver unit 802; the transceiver unit 802 is used to send a fourth reference signal; the transceiver unit 802 is also used to receive first information, and the processing unit 801 is used to determine the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information based on the first information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on the third neural network, and the P ninth channel information is determined by P reference signals, and the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

[0344] In one possible implementation, when the device 800 is used to execute the method executed by the terminal device in the aforementioned Figure 7 and related embodiments, the device 800 includes a processing unit 801 and a transceiver unit 802; the transceiver unit 802 is used to receive a fourth reference signal; the processing unit 801 is used to determine the first information; the transceiver unit 802 is also used to send the first information, which is used to indicate the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on the third neural network, and the P ninth channel information is determined by P reference signals, and the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

[0345] It should be noted that, for details on the information execution process of the units of the above-mentioned communication device 800, please refer to the description in the method embodiment shown above in this application, and no further details will be given here.

[0346] Please refer to Figure 9, which is another schematic structural diagram of a communication device 900 provided in this application. The communication device 900 includes a logic circuit 901 and an input / output interface 902. The communication device 900 may be a chip or an integrated circuit.

[0347] The transceiver unit 802 shown in FIG8 may be a communication interface, which may be the input / output interface 902 in FIG9 , which may include an input interface and an output interface. Alternatively, the communication interface may be a transceiver circuit, which may include an input interface circuit and an output interface circuit.

[0348] Optionally, the input-output interface 902 is used to send a first reference signal; the input-output interface 902 is also used to receive first channel information, the first channel information is determined based on the first reference signal, the channel dimension represented by the first channel information is [T1, R1, F1], and T1, R1, and F1 are all positive numbers; the logic circuit 901 is used to process the first channel information and Q second channel information based on the first neural network to obtain third channel information; wherein, the Q second channel information are determined by Q second reference signals, and Q is a positive integer; the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, and the channel dimension represented by the third channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the parameters in the channel dimension represented by the first channel information and the parameters in the channel dimension represented by the third channel information satisfy at least one of the following: T is greater than the T1, R is greater than R1, and F is greater than F1.

[0349] Optionally, the input-output interface 902 is used to send a fourth reference signal; the input-output interface 902 is also used to receive first information, and the logic circuit 901 is used to determine the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information based on the first information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on the third neural network, and the P ninth channel information is determined by P reference signals, and the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

[0350] Optionally, the input-output interface 902 is used to receive a fourth reference signal; the logic circuit 901 is used to determine the first information; the input-output interface 902 is also used to send the first information, which is used to indicate the deviation between the channel information corresponding to the fourth reference signal and the eighth channel information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; wherein the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on the third neural network, and the P ninth channel information is determined by P reference signals, and the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

[0351] The logic circuit 901 and the input / output interface 902 may also execute other steps executed by the first communication device or the second communication device in any embodiment and achieve corresponding beneficial effects, which will not be described in detail here.

[0352] In a possible implementation, the processing unit 801 shown in FIG. 8 may be the logic circuit 901 in FIG. 9 .

[0353] Optionally, the logic circuit 901 may be a processing device, and the functions of the processing device may be partially or entirely implemented by software.

[0354] Optionally, the processing device may include a memory and a processor, wherein the memory is used to store a computer program, and the processor reads and executes the computer program stored in the memory to perform corresponding processing and / or steps in any one of the method embodiments.

[0355] Alternatively, the processing device may include only a processor. A memory for storing the computer program is located outside the processing device, and the processor is connected to the memory via circuits / wires to read and execute the computer program stored in the memory. The memory and processor may be integrated or physically separate.

[0356] Optionally, the processing device may be one or more chips, or one or more integrated circuits. For example, the processing device may be one or more field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), system-on-chips (SoCs), central processor units (CPUs), network processors (NPs), digital signal processors (DSPs), microcontroller units (MCUs), programmable logic devices (PLDs), or other integrated chips, or any combination of the above chips or processors.

[0357] Please refer to Figure 10, which shows the communication device 1000 involved in the above-mentioned embodiments provided in an embodiment of the present application. The communication device 1000 can specifically be a communication device serving as a terminal device in the above-mentioned embodiments. The example shown in Figure 10 is that the terminal device is implemented through the terminal device (or a component in the terminal device).

[0358] Herein, a possible logical structure diagram of the communication device 1000 is shown. The communication device 1000 may include but is not limited to at least one processor 1001 and a communication port 1002 .

[0359] The transceiver unit 802 shown in FIG8 may be a communication interface, which may be the communication port 1002 in FIG10 , which may include an input interface and an output interface. Alternatively, the communication port 1002 may be a transceiver circuit, which may include an input interface circuit and an output interface circuit.

[0360] Further optionally, the device may also include at least one of a memory 1003 and a bus 1004. In an embodiment of the present application, the at least one processor 1001 is used to control and process the actions of the communication device 1000.

[0361] In addition, the processor 1001 can be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a 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 the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a digital signal processor and a microprocessor, and so on. Those skilled in the art will clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0362] It should be noted that the communication device 1000 shown in Figure 10 can be specifically used to implement the steps implemented by the terminal device in the aforementioned method embodiment and achieve the corresponding technical effects of the terminal device. The specific implementation methods of the communication device shown in Figure 10 can refer to the description in the aforementioned method embodiment and will not be repeated here.

[0363] Please refer to Figure 11, which is a structural diagram of the communication device 1100 involved in the above-mentioned embodiments provided in an embodiment of the present application. The communication device 1100 can specifically be a communication device as a network device in the above-mentioned embodiments. The example shown in Figure 11 is that the network device is implemented through the network device (or a component in the network device), wherein the structure of the communication device can refer to the structure shown in Figure 11.

[0364] The communication device 1100 includes at least one processor 1111 and at least one network interface 1114. Further optionally, the communication device also includes at least one memory 1112, at least one transceiver 1113 and one or more antennas 1115. The processor 1111, the memory 1112, the transceiver 1113 and the network interface 1114 are connected, for example, via a bus. In an embodiment of the present application, the connection may include various interfaces, transmission lines or buses, etc., which are not limited in this embodiment. The antenna 1115 is connected to the transceiver 1113. The network interface 1114 is used to enable the communication device to communicate with other communication devices through a communication link. For example, the network interface 1114 may include a network interface between the communication device and the core network device, such as an S1 interface, and the network interface may include a network interface between the communication device and other communication devices (such as other network devices or core network devices), such as an X2 or Xn interface.

[0365] The transceiver unit 802 shown in FIG8 may be a communication interface, which may be the network interface 1114 in FIG11 , which may include an input interface and an output interface. Alternatively, the network interface 1114 may be a transceiver circuit, which may include an input interface circuit and an output interface circuit.

[0366] Processor 1111 is primarily used to process communication protocols and communication data, control the entire communication device, execute software programs, and process software program data, for example, to support the communication device in performing the actions described in the embodiments. The communication device may include a baseband processor and a central processing unit. The baseband processor is primarily used to process communication protocols and communication data, while the central processing unit is primarily used to control the entire terminal device, execute software programs, and process software program data. Processor 1111 in Figure 11 may integrate the functions of both a baseband processor and a central processing unit. Those skilled in the art will appreciate that the baseband processor and the central processing unit may also be independent processors interconnected via a bus or other technology. Those skilled in the art will appreciate that a terminal device may include multiple baseband processors to accommodate different network standards, multiple central processing units to enhance its processing capabilities, and various components of the terminal device may be connected via various buses. The baseband processor may also be referred to as a baseband processing circuit or a baseband processing chip. The central processing unit may also be referred to as a central processing circuit or a central processing chip. The functionality for processing communication protocols and communication data may be built into the processor or stored in memory as a software program, which is executed by the processor to implement the baseband processing functionality.

[0367] The memory is primarily used to store software programs and data. Memory 1112 can exist independently and be connected to processor 1111. Alternatively, memory 1112 can be integrated with processor 1111, for example, within a single chip. Memory 1112 can store program code for executing the technical solutions of the embodiments of the present application, and execution is controlled by processor 1111. The various computer program codes executed can also be considered drivers for processor 1111.

[0368] Figure 11 shows only one memory and one processor. In an actual terminal device, there may be multiple processors and multiple memories. The memory may also be referred to as a storage medium or a storage device. The memory may be a storage element on the same chip as the processor, i.e., an on-chip storage element, or an independent storage element, which is not limited in the present embodiment.

[0369] The transceiver 1113 can be used to support the reception or transmission of radio frequency signals between the communication device and the terminal. The transceiver 1113 can be connected to the antenna 1115. The transceiver 1113 includes a transmitter Tx and a receiver Rx. Specifically, one or more antennas 1115 can receive radio frequency signals. The receiver Rx of the transceiver 1113 is used to receive the radio frequency signal from the antenna, convert the radio frequency signal into a digital baseband signal or a digital intermediate frequency signal, and provide the digital baseband signal or digital intermediate frequency signal to the processor 1111 so that the processor 1111 can further process the digital baseband signal or digital intermediate frequency signal, such as demodulation and decoding. In addition, the transmitter Tx in the transceiver 1113 is also used to receive a modulated digital baseband signal or digital intermediate frequency signal from the processor 1111, convert the modulated digital baseband signal or digital intermediate frequency signal into a radio frequency signal, and transmit the radio frequency signal through one or more antennas 1115. Specifically, the receiver Rx can selectively perform one or more stages of down-mixing and analog-to-digital conversion on the RF signal to obtain a digital baseband signal or a digital intermediate frequency signal. The order of the down-mixing and analog-to-digital conversion processes is adjustable. The transmitter Tx can selectively perform one or more stages of up-mixing and digital-to-analog conversion on the modulated digital baseband signal or digital intermediate frequency signal to obtain a RF signal. The order of the up-mixing and digital-to-analog conversion processes is adjustable. The digital baseband signal and the digital intermediate frequency signal may be collectively referred to as digital signals.

[0370] The transceiver 1113 may also be referred to as a transceiver unit, a transceiver, a transceiver device, etc. Optionally, a device in the transceiver unit that implements a receiving function may be referred to as a receiving unit, and a device in the transceiver unit that implements a transmitting function may be referred to as a transmitting unit. That is, the transceiver unit includes a receiving unit and a transmitting unit. The receiving unit may also be referred to as a receiver, an input port, a receiving circuit, etc., and the transmitting unit may be referred to as a transmitter, a transmitter, or a transmitting circuit, etc.

[0371] It should be noted that the communication device 1100 shown in Figure 11 can be specifically used to implement the steps implemented by the network device in the aforementioned method embodiment, and to achieve the corresponding technical effects of the network device. The specific implementation methods of the communication device 1100 shown in Figure 11 can refer to the description in the aforementioned method embodiment, and will not be repeated here one by one.

[0372] Please refer to FIG12 , which is a schematic structural diagram of the communication device involved in the above-mentioned embodiment provided in an embodiment of the present application.

[0373] It can be understood that the communication device 120 includes, for example, modules, units, elements, circuits, or interfaces, which are appropriately configured together to implement the technical solutions provided in this application. The communication device 120 can be the terminal device or network device described above, or a component (such as a chip) in these devices, used to implement the method described in the following method embodiment. The communication device 120 includes one or more processors 121. The processor 121 can be a general-purpose processor or a dedicated processor. For example, it can be a baseband processor or a central processing unit. The baseband processor can be used to process communication protocols and communication data, and the central processing unit can be used to control the communication device (such as a RAN node, terminal, or chip, etc.), execute software programs, and process data of software programs.

[0374] Optionally, in one design, the processor 121 may include a program 123 (sometimes also referred to as code or instructions), which may be executed on the processor 121 to cause the communication device 120 to perform the methods described in the following embodiments. In yet another possible design, the communication device 120 includes circuitry (not shown in FIG12 ).

[0375] Optionally, the communication device 120 may include one or more memories 122 on which a program 124 (sometimes also referred to as code or instructions) is stored. The program 124 can be run on the processor 121, so that the communication device 120 executes the method described in the above method embodiment.

[0376] Optionally, the processor 121 and / or the memory 122 may include an AI module 127, 128, which is used to implement AI-related functions. The AI ​​module may be implemented through software, hardware, or a combination of software and hardware. For example, the AI ​​module may include a wireless intelligent control (RIC) module. For example, the AI ​​module may be a near real-time RIC or a non-real-time RIC.

[0377] Optionally, data may be stored in the processor 121 and / or the memory 122. The processor and the memory may be provided separately or integrated together.

[0378] Optionally, the communication device 120 may further include a transceiver 125 and / or an antenna 126. The processor 121 may also be sometimes referred to as a processing unit, and controls the communication device (e.g., a RAN node or terminal). The transceiver 125 may also be sometimes referred to as a transceiver unit, a transceiver, a transceiver circuit, or a transceiver, and is configured to implement the transceiver functions of the communication device through the antenna 126.

[0379] The transceiver unit 802 shown in FIG8 may be a communication interface, which may be the transceiver 125 in FIG12 . The transceiver 125 may include an input interface and an output interface. Alternatively, the transceiver 125 may be a transceiver circuit, which may include an input interface circuit and an output interface circuit.

[0380] An embodiment of the present application further provides a computer-readable storage medium, which is used to store one or more computer-executable instructions. When the computer-executable instructions are executed by a processor, the processor executes the method described in the possible implementation methods of the first communication device or the second communication device in the aforementioned embodiment.

[0381] An embodiment of the present application also provides a computer program product (or computer program). When the computer program product is executed by the processor, the processor executes the method that may be implemented by the above-mentioned first communication device or second communication device.

[0382] An embodiment of the present application also provides a chip system, which includes at least one processor for supporting a communication device to implement the functions involved in the possible implementation methods of the above-mentioned communication device. Optionally, the chip system also includes an interface circuit, which provides program instructions and / or data to the at least one processor. In one possible design, the chip system may also include a memory, which is used to store the necessary program instructions and data for the communication device. The chip system can be composed of chips, or it can include chips and other discrete devices, wherein the communication device can specifically be the first communication device or the second communication device in the aforementioned method embodiment.

[0383] An embodiment of the present application further provides a communication system, wherein the network system architecture includes the first communication device and the second communication device in any of the above embodiments.

[0384] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0385] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0386] In addition, the functional units in the various embodiments of the present application can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the contributing part or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

Claims

1. A communication method, characterized in that: The method comprises: sending a first reference signal; receiving first channel information, where the first channel information is determined based on the first reference signal, where a channel dimension represented by the first channel information is [T1, R1, F1], and where T1, R1, and F1 are all positive numbers; The first channel information and Q second channel information are processed based on a first neural network to obtain third channel information; wherein the Q second channel information are determined by Q second reference signals, where Q is a positive integer; the position of the time domain resource carrying the Q second reference signals is located before the position of the time domain resource carrying the first reference signal, and the channel dimension represented by the third channel information is [T, R, F], where T, R, and F are all positive numbers; At least one of the following is satisfied: the T is greater than the T1, the R is greater than the R1, and the F is greater than the F1.

2. The method according to claim 1, characterized in that The channel dimension represented by at least one of the Q second channel information is [T2, R2, F2], where T2, R2, and F2 are all positive numbers; At least one of the following is satisfied: the T is greater than the T2, the R is greater than the R2, and the F is greater than the F2.

3. The method according to claim 2, characterized in that At least one of the following is met: The values ​​of T1 and T2 are the same, the values ​​of R1 and R2 are the same, and the values ​​of F1 and F2 are the same.

4. The method according to any one of claims 1 to 3, characterized in that: The first neural network includes a first sub-neural network and a second sub-neural network; The processing of the first channel information and Q second channel information based on the first neural network to obtain third channel information includes: Processing the first channel information based on the first sub-neural network to obtain fourth channel information; wherein the channel dimension represented by the fourth channel information is the [T, R, F]; The fourth channel information and the Q second channel information are processed based on the second sub-neural network to obtain the third channel information.

5. The method according to any one of claims 1 to 4, characterized in that: After the first channel information and Q second channel information are processed based on the first neural network to obtain third channel information, the method further includes: The third channel information is processed based on the second neural network to obtain fifth channel information, and the channel dimension represented by the fifth channel information is the [T, R, F].

6. The method according to claim 5, characterized in that The deviation between the fifth channel information and the sixth channel information is less than or equal to the deviation between the third channel information and the sixth channel information, the channel dimension represented by the sixth channel information is the [T, R, F], and the sixth channel information is obtained by measuring the reference signal.

7. The method according to any one of claims 1 to 6, characterized in that: The method further comprises: sending a third reference signal; receiving channel information corresponding to the third reference signal, where the channel dimension represented by the channel information corresponding to the third reference signal is [T3, R3, F3], where T3, R3, and F3 are all positive numbers; and at least one of the following is satisfied: T is greater than T3, R is greater than R3, and F is greater than F3; Based on the first neural network, the channel information corresponding to the second reference signal, part or all of the Q second channel information, and the third channel information are processed to obtain seventh channel information; wherein the channel dimension represented by the seventh channel information is the [T, R, F].

8. The method according to claim 7, characterized in that At least one of the following is met: The values ​​of T1 and T3 are the same, the values ​​of R1 and R3 are the same, and the values ​​of F1 and F3 are the same.

9. The method according to any one of claims 1 to 8, characterized in that: The method further comprises: sending a fourth reference signal; Receive first information, where the first information is used to indicate a deviation between channel information corresponding to a fourth reference signal and eighth channel information; wherein the channel dimension represented by the eighth channel information is the [T, R, F], and the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on a third neural network, the P ninth channel information are determined by P reference signals, and the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

10. The method according to claim 9, characterized in that The channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], where T4, R4, and F4 are all positive numbers; and at least one of the following is satisfied: The T is greater than or equal to the T4 and the T4 is greater than the T1, the R is greater than or equal to the R4 and the R4 is greater than the R1, and the F is greater than or equal to the F4 and the F4 is greater than the F1.

11. The method according to claim 9 or 10, characterized in that: The method further comprises: Sending second information, where the second information is used to indicate whether to adjust an input parameter of the third neural network; wherein the second information is determined based on the first information.

12. The method according to any one of claims 9 to 11, characterized in that: The method further comprises: Send third information, wherein the third information indicates channel information corresponding to the fourth reference signal processed based on the third neural network; wherein the third information is determined based on the first information.

13. The method according to any one of claims 1 to 8, characterized in that: The method further comprises: sending a fourth reference signal; Receive channel information corresponding to the fourth reference signal; the channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], and the T4, the R4, and the F4 are all positive numbers; at least one of the following is satisfied: the T is greater than or equal to the T4 and the T4 is greater than the T1, the R is greater than or equal to the R4 and the R4 is greater than the R1, and the F is greater than or equal to the F4 and the F4 is greater than the F1; Processing the channel information corresponding to the fourth reference signal and part or all of the Q second channel information based on the first neural network to obtain tenth channel information; wherein the channel dimension represented by the tenth channel information is the [T, R, F]; Determine a deviation between the channel information corresponding to the fourth reference signal and the tenth channel information.

14. The method according to claim 13, characterized in that The method further comprises: A fifth reference signal is sent, wherein the channel dimension represented by the channel information corresponding to the fifth reference signal is determined based on a deviation between the channel information corresponding to the fourth reference signal and the tenth channel information.

15. The method according to claim 14, characterized in that The channel dimension represented by the channel information corresponding to the fifth reference signal is [T5, R5, F5], where T5, R5, and F5 are all positive numbers; the channel dimension represented by the channel information corresponding to the fifth reference signal is determined based on the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information, including: When the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is better, at least one of the following is satisfied: the T5 is less than or equal to the T4, the R5 is less than or equal to the R4, and the F5 is less than or equal to the F4; When the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information indicates that the channel quality corresponding to the tenth channel information is poor, at least one of the following is satisfied: the T5 is greater than or equal to the T4, the R5 is greater than or equal to the R4, and the F5 is greater than or equal to the F4.

16. The method according to any one of claims 13 to 15, characterized in that The method further comprises: The input parameters of the first neural network are adjusted based on the deviation between the channel information corresponding to the fourth reference signal and the tenth channel information.

17. The method according to any one of claims 1 to 16, characterized in that: The method further comprises: sending a sixth reference signal; receiving channel information corresponding to the sixth reference signal; Based on the fourth neural network, the channel information corresponding to the sixth reference signal and K eleventh channel information are processed to obtain twelfth channel information; wherein the K eleventh channel information are determined by K reference signals, and K is a positive integer; the time domain resource carrying the Kth reference signal is located before the time domain resource carrying the sixth reference signal; the channel dimension represented by the twelfth channel information is the [T, R, F]; wherein the network structure of the first neural network is different from the network structure of the fourth neural network; or, the network structure of the first neural network is the same as the network structure of the fourth neural network, and the parameters in the network structure of the first neural network are different from the parameters in the network structure of the fourth neural network; The sixth reference signal satisfies at least one of the following: The resources carrying the sixth reference signal are different from the resources carrying the first reference signal; The first channel information and the channel information corresponding to the sixth reference signal come from different terminal devices; The first channel information and the channel information corresponding to the sixth reference signal come from terminal devices with different mobile information.

18. A communication method, characterized in that: include: sending a fourth reference signal; Receive first information, where the first information is used to indicate a deviation between channel information corresponding to a fourth reference signal and eighth channel information; wherein a channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; Among them, the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on the third neural network, the P ninth channel information is determined by P reference signals, the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

19. The method according to claim 18, characterized in that The channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], where T4, R4, and F4 are all positive numbers; and at least one of the following is satisfied: The T is greater than or equal to the T4, the R is greater than or equal to the R4, and the F is greater than or equal to the F4.

20. The method according to claim 18 or 19, characterized in that The method further comprises: Sending second information, where the second information is used to indicate whether to adjust an input parameter of the third neural network; wherein the second information is determined based on the first information.

21. The method according to any one of claims 18 to 20, characterized in that The method further comprises: Send third information, wherein the third information indicates channel information corresponding to the fourth reference signal processed based on the third neural network; wherein the third information is determined based on the first information.

22. A communication method, characterized in that: include: receiving a fourth reference signal; Sending first information, where the first information is used to indicate a deviation between channel information corresponding to the fourth reference signal and eighth channel information; wherein the channel dimension represented by the eighth channel information is [T, R, F], and T, R, and F are all positive numbers; Among them, the eighth channel information is obtained by processing the channel information corresponding to the fourth reference signal and P ninth channel information based on the third neural network, the P ninth channel information is determined by P reference signals, the position of the time domain resources carrying the P reference signals is located before the position of the time domain resources carrying the fourth reference signal, and P is a positive integer.

23. The method according to claim 22, characterized in that The channel dimension represented by the channel information corresponding to the fourth reference signal is [T4, R4, F4], where T4, R4, and F4 are all positive numbers; and at least one of the following is satisfied: The T is greater than or equal to the T4, the R is greater than or equal to the R4, and the F is greater than or equal to the F4.

24. The method according to claim 22 or 23, characterized in that The method further comprises: Receive second information, where the second information is used to indicate whether to adjust an input parameter of the third neural network; wherein the second information is determined based on the first information.

25. The method according to any one of claims 22 to 24, characterized in that The method further comprises: Receive third information, wherein the third information indicates channel information corresponding to the fourth reference signal processed based on the third neural network; wherein the third information is determined based on the first information.

26. A communication device, characterized in that: Comprising means for performing the method as claimed in any one of claims 1 to 25.

27. A communication device, characterized in that: The method comprises at least one processor coupled to a memory; the at least one processor is configured to execute the method according to any one of claims 1 to 25.

28. The communication device according to claim 27, characterized in that The communication device is a chip or a chip system.

29. A readable storage medium, characterized in that: The storage medium stores a computer program or an instruction, and when the computer program or the instruction is executed by the communication device, the method according to any one of claims 1 to 25 is implemented.

30. A computer program product, characterized in that The method comprises instructions which, when executed on a computer, cause the computer to perform the method according to any one of claims 1 to 25.

Images