A method for channel state information feedback

Abstract

An embodiment of the present application provides a method for channel state information feedback. The method includes: measuring N groups of frequency-domain channel state information reference signals sent by a base station to obtain channel matrices of N frequency-domain channels; performing singular value decomposition on each channel matrix to obtain a left singular matrix, a diagonal matrix containing singular values, and a right singular matrix; for each diagonal matrix, determining whether it is a characteristic diagonal matrix according to the ratio of the largest singular value in the diagonal matrix to the sum of all singular values; counting the number of characteristic diagonal matrices; if the counted number is greater than or equal to a first set value, generating first channel state information to feedback to the base station; if the counted number is less than a second set value, generating second channel state information to feedback to the base station; if the counted number is between the first and second set values, generating third channel state information to feedback to the base station. The embodiment of the present application can solve the problem of poor flexibility in channel state information feedback and improve the service quality of the network.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of computer technologies, and in particular, to a method for feedback of channel state information, a computer device, and a computer-readable storage medium. Background Art

[0002] Future mobile communication systems will meet the diverse service requirements of people in various areas such as residence, work, leisure, and transportation. Even in scenarios with ultra-high traffic density, ultra-high connection density, and ultra-high mobility characteristics, such as dense residential areas, offices, stadiums, open-air gatherings, subways, expressways, high-speed rails, and wide-area coverage, they can also provide users with extreme service experiences such as ultra-high-definition videos, virtual reality, augmented reality, cloud desktops, and online games. At the same time, future mobile communication systems will also penetrate into the Internet of Things and various industrial fields, deeply integrate with industrial facilities, medical instruments, transportation vehicles, etc., and effectively meet the diverse service requirements of vertical industries such as industry, medical care, and transportation, realizing true "Internet of Everything".

[0003] The application scenarios of future mobile communication systems can be divided into two categories, namely mobile broadband (MBB) and the Internet of Things (IoT). Among them, the main technical requirement for mobile broadband access is high capacity to provide high data rates to meet the growing demand for data services. The Internet of Things is mainly driven by the needs of machine type communication (MTC), and can be further divided into two types, including low-rate massive machine communication (MMC) and low-latency and high-reliability machine communication. Among them, for low-rate massive machine communication, a large number of nodes access at low rates, the transmitted data packets are usually small, and the interval time is relatively long. The cost and power consumption of such nodes are usually very low; for low-latency and high-reliability machine communication, it is mainly for machine communication with relatively high requirements for real-time performance and reliability, such as real-time alarms and real-time monitoring.

[0004] In a mobile communication system, feedback information related to a wireless channel is the basis for effective data transmission. How to obtain high-quality channel state information is an urgent problem to be solved at present. Summary of the Invention

[0005] The purpose of the embodiments of the present application is to provide a method for feedback of channel state information, a computer device, and a computer-readable storage medium, which can solve the problem of poor flexibility in feedback of channel state information and improve the service quality of the network.

[0006] One aspect of the embodiments of the present application provides a method for feedback of channel state information, including:

[0007] Measure N groups of frequency-domain channel state information reference signals sent by the base station to obtain channel matrices of N frequency-domain channels;

[0008] Perform singular value decomposition on the channel matrix of each frequency-domain channel to obtain the left singular matrix, diagonal matrix containing singular values, and right singular matrix corresponding to each frequency-domain channel;

[0009] For each diagonal matrix, calculate the ratio of the largest singular value in the diagonal matrix to the sum of all singular values. If the calculated ratio is greater than or equal to the set threshold, determine the diagonal matrix as the characteristic diagonal matrix;

[0010] Count the number of characteristic diagonal matrices; if the counted number is greater than or equal to the first set value, generate first channel state information based on the first column of each right singular matrix and feedback it to the base station; if the counted number is less than the second set value, generate second channel state information based on the time-domain channel vector obtained by Fourier transform of the channel matrices of N frequency-domain channels and feedback it to the base station; if the counted number is between the first and second set values, generate third channel state information based on the first two columns of each right singular matrix and feedback it to the base station.

[0011] Preferably, the generating first channel state information based on the first column of each right singular matrix specifically includes:

[0012] Use a first encoding neural network based on an autoencoder structure to process the first column of each right singular matrix to generate first channel state information.

[0013] Furthermore, the method further includes:

[0014] After receiving the first channel state information, the base station uses a first decoding neural network based on an autoencoder structure to decode the first channel state information to obtain the first column of each right singular matrix.

[0015] Preferably, the generating second channel state information based on the time-domain channel vector obtained by Fourier transform of the channel matrices of N frequency-domain channels specifically includes:

[0016] Use a second encoding neural network based on an autoencoder structure to process the time-domain channel vector to generate second channel state information.

[0017] Furthermore, the method further includes:

[0018] After receiving the second channel state information, the base station uses a second decoding neural network based on an autoencoder structure to decode the second channel state information to obtain the time-domain channel vector.

[0019] Preferably, the generating third channel state information based on the first two columns of each right singular matrix specifically includes:

[0020] Obtain the first two columns of each right singular matrix, project each obtained column vector onto L mutually orthogonal basis vectors to obtain 2N×L projection values, and perform quantization processing on the 2N×L projection values to obtain the third channel state information;

[0021] Wherein, the value of L is an integer greater than or equal to T / 4, and T is the number of transmit antennas of the base station.

[0022] Furthermore, the method further includes:

[0023] After receiving the third channel state information, the base station uses L mutually orthogonal basis vectors to obtain the first two columns of N right singular matrices according to the 2N×L projection values.

[0024] An aspect of an embodiment of the present application further provides a computer device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above channel state information feedback method are implemented.

[0025] An aspect of an embodiment of the present application further provides a computer-readable storage medium, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the steps of the above channel state information feedback method are implemented.

[0026] The channel state information feedback method, device, and computer-readable storage medium provided by the embodiments of the present application measure N groups of frequency-domain channel state information reference signals sent by the base station to obtain the channel matrices of N frequency-domain channels; perform singular value decomposition on the channel matrix of each frequency-domain channel to obtain the left singular matrix, diagonal matrix containing singular values, and right singular matrix corresponding to each frequency-domain channel; for each diagonal matrix, calculate the ratio of the largest singular value in the diagonal matrix to the sum of all singular values. If the calculated ratio is greater than or equal to a set threshold, then determine the diagonal matrix as a characteristic diagonal matrix; count the number of characteristic diagonal matrices; if the counted number is greater than or equal to a first set value, then generate first channel state information according to the first column of each right singular matrix and feedback it to the base station; if the counted number is less than a second set value, then generate second channel state information according to the time-domain channel vectors obtained by Fourier transform of the channel matrices of N frequency-domain channels and feedback it to the base station; if the counted number is between the first and second set values, then generate third channel state information according to the first two columns of each right singular matrix and feedback it to the base station.

[0027] Since the result obtained from the singular value decomposition of the channel matrix, that is, the number of statistical eigen - diagonal matrices, can be used to analyze the spatial characteristics of the channel, corresponding channel processing methods are adopted to generate channel state information based on different spatial characteristics of the channel; that is to say, the terminal can feedback channel state information matching the spatial characteristics of the channel to the base station, providing a flexible way of feedback of channel state information, thereby solving the problem of poor flexibility in feedback of channel state information and improving the service quality of the network. Brief Description of the Drawings

[0028] Figure 1 Schematically shows a flowchart of a method for feedback of channel state information according to an embodiment of the present application;

[0029] Figure 2 Schematically shows a hardware architecture diagram of a computer device suitable for implementing the method for feedback of channel state information according to an embodiment of the present application. Detailed Description of the Embodiment

[0030] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present application without creative efforts shall fall within the scope of protection of the present application.

[0031] It should be noted that the descriptions involving "first", "second", etc. in the embodiments of the present application are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between various embodiments can be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement them. When the combination of technical solutions appears to be contradictory or unable to be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present application.

[0032] In the description of the present application, it should be understood that the numerical labels before the steps do not identify the order of execution of the steps, but are only used to facilitate the description of the present application and to distinguish each step, so they cannot be understood as a limitation to the present application.

[0033] The technical solutions of the embodiments of the present invention will be described in detail below with reference to the drawings.

[0034] Figure 1 Schematically shows a specific flowchart of a method for feedback of channel state information according to Embodiment 1 of the present application, including the following steps:

[0035] Step S101: The terminal measures N groups of frequency-domain channel state information reference signals sent by the base station to obtain channel matrices of N frequency-domain channels.

[0036] Specifically, a terminal with R receive antennas receives N groups of frequency-domain channel state information reference signals sent by a base station with T transmit antennas. The terminal measures the frequency-domain channel state information reference signals to obtain channel matrices of N frequency-domain channels from the base station to the terminal, denoted as FH-i, where the value of i ranges from 1 to N. FH-i is the channel matrix of the i-th frequency-domain channel, and its dimension is R×T.

[0037] Step S102: The terminal performs singular value decomposition on the channel matrix of each frequency-domain channel to obtain the left singular matrix, the diagonal matrix containing singular values, and the right singular matrix corresponding to each frequency-domain channel.

[0038] Specifically, the terminal performs singular value decomposition on the channel matrix of each frequency-domain channel to obtain the left singular matrix, the diagonal matrix containing singular values, and the right singular matrix corresponding to each frequency-domain channel. Among them, the dimension of the left singular matrix is R×R, the dimension of the diagonal matrix is R×T, and the dimension of the right singular matrix is T×T.

[0039] That is to say, by performing singular value decomposition on the channel matrix of each frequency-domain channel, the terminal can obtain N left singular matrices, N diagonal matrices containing singular values, and N right singular matrices.

[0040] Step S103: For each diagonal matrix, calculate the ratio of the largest singular value in the diagonal matrix to the sum of all singular values. If the calculated ratio is greater than or equal to the set threshold, then determine the diagonal matrix as the characteristic diagonal matrix.

[0041] In an exemplary embodiment, the above set threshold can be set to 0.8.

[0042] Step S104: Count the number of characteristic diagonal matrices. If the counted number is greater than or equal to the first set value, then execute the following step S105; if the counted number is less than the second set value, then execute the following step S107; if the counted number is between the first and second set values, then execute the following step S109.

[0043] Specifically, the result obtained by performing singular value decomposition on the channel matrix through the above steps, that is, the counted number of characteristic diagonal matrices, can be used to analyze the spatial characteristics of the channel. Based on different spatial characteristics of the channel, corresponding channel processing methods are adopted to generate channel state information. That is to say, the terminal can feedback channel state information matching the spatial characteristics of the channel to the base station, providing a flexible way of feedback for channel state information, thereby solving the problem of poor flexibility in feedback of channel state information and improving the service quality of the network.

[0044] According to simulation and actual measurement analysis, when the number of statistical characteristic diagonal matrices is greater than or equal to the first set value, the base station to terminal channel energy is mainly concentrated in specific few directions. Therefore, perform the following step S105, adopt the first channel state processing method to generate the first channel state information; in a preferred embodiment, the first set value can be N / 2;

[0045] If the number of statistical characteristic diagonal matrices is less than the second set value, it indicates that the base station to terminal channel energy is dispersed in relatively many directions. Therefore, perform the following step S107, adopt the second channel state processing method to generate the second channel state information; wherein, the first set value is greater than the second set value. In a preferred embodiment, the second set value can be N / 4;

[0046] If the number of statistical characteristic diagonal matrices is between the first and second set values, it indicates that the base station to terminal channel energy is mainly concentrated in specific several directions. Therefore, perform the following step S109, adopt the third channel state processing method to generate the third channel state information.

[0047] Step S105: The terminal generates the first channel state information based on the first column of each right singular matrix and feeds it back to the base station;

[0048] In this step, the terminal uses the first channel state processing method: the terminal uses the first coding neural network based on the autoencoder structure to process the first column of each right singular matrix to generate the first channel state information;

[0049] In an exemplary embodiment, the first coding neural network includes at least 1 convolutional layer, 1 pooling layer, and 1 fully connected layer. The input of the first coding neural network is the first column of N right singular matrices. The terminal performs quantization processing on the output of the fully connected layer to obtain the first channel state information and feeds the first channel state information back to the base station.

[0050] Since the base station to terminal channel energy is mainly concentrated in specific few directions at this time, according to simulation and actual measurement analysis, the neural network using the autoencoder can well describe this channel.

[0051] More preferably, when the channel signal-to-interference-plus-noise ratio from the base station to the terminal is greater than or equal to 10 dB, the terminal transmits the first channel state information using at least half of the full power. In fact, when the channel signal-to-interference-plus-noise ratio from the base station to the terminal is greater than or equal to 10 dB, it indicates that the downlink channel energy is relatively concentrated and the channel changes relatively slowly, so less power can be used to transmit relevant information.

[0052] Step S106: The base station decodes the first channel state information.

[0053] In this step, after receiving the first channel state information, the base station decodes the first channel state information using a first decoding neural network based on an autoencoder structure to obtain the first column of each right singular matrix.

[0054] According to the simulation and hardware test results, the results of optimization considering channel recovery accuracy, terminal power consumption, and hardware implementation complexity can be the following exemplary embodiments: The first decoding neural network includes at least 3 convolutional layers, 2 pooling layers, and one fully connected layer; the parameters in the first encoding neural network and the first decoding neural network are jointly trained.

[0055] The computational complexity of the first decoding neural network is 3 times that of the first encoding neural network, so that the base station can perform as many complex operations as possible to reduce the burden on the terminal.

[0056] Step S107: The terminal generates second channel state information based on the time-domain channel vector obtained by performing Fourier transform on the channel matrix of N frequency-domain channels, and feeds back the second channel state information to the base station.

[0057] In this step, the terminal uses a second channel state processing method: uses a second encoding neural network based on an autoencoder structure to process the time-domain channel vector to generate second channel state information.

[0058] In an exemplary embodiment, the second channel state processing method is a second encoding neural network based on an autoencoder structure. The second encoding neural network includes at least 2 convolutional layers, 2 pooling layers, and 1 fully connected layer; the input of the second encoding neural network is R×T time-domain channel vectors. The terminal performs quantization processing on the output of the fully connected layer to obtain the second channel state information, and feeds back the second channel state information to the base station.

[0059] Since the channel energy from the base station to the terminal is dispersed in more directions at this time, according to simulation and actual measurement analysis, a neural network using an autoencoder can well describe this channel.

[0060] More preferably, when the signal-to-interference-plus-noise ratio of the channel from the base station to the terminal is greater than or equal to 10 dB, the terminal transmits the second channel state information at full power. The reason is that the downlink channel energy is relatively dispersed at this time, the channel changes are relatively complex, and the probability of this user performing multi-user transmission is relatively high. Therefore, it is necessary to improve the probability of successful transmission of this channel state information as much as possible.

[0061] Step S108: The base station decodes the second channel state information.

[0062] In this step, after receiving the second channel state information, the base station decodes the second channel state information using a second decoding neural network based on an autoencoder structure to obtain the time-domain channel vector.

[0063] According to the simulation and hardware test results, the optimization results considering the channel recovery accuracy, terminal power consumption, and hardware implementation complexity can be the following exemplary embodiments: The second decoding neural network includes at least 6 convolutional layers, 5 pooling layers, and one fully connected layer; the parameters in the second encoding neural network and the second decoding neural network are jointly trained.

[0064] The computational complexity of the second decoding neural network is 6 times that of the second encoding neural network, so that the base station can perform more complex operations as much as possible to reduce the burden on the terminal.

[0065] Step S109: The terminal generates third channel state information based on the first two columns of each right singular matrix and feeds it back to the base station;

[0066] In this step, the terminal processes the first two columns of N right singular matrices using the third channel state processing method: obtain the first two columns of each right singular matrix, project each obtained column vector onto L mutually orthogonal basis vectors to obtain 2N×L projection values, perform quantization processing on the 2N×L projection values to obtain the third channel state information, and feed back the third channel state information to the base station.

[0067] The dimension of the time-domain channel vector is X*1. Since the dimension X of the time-domain channel vector describes the sparsity of the channel, and this channel sparsity is related to the characteristics of the downlink channel, through simulation and actual measurement analysis, and comprehensively considering the complexity of hardware implementation, in an exemplary embodiment, the value of X is an integer greater than or equal to N / 5 and an integer multiple of 2.

[0068] The third channel state information includes the description information of L mutually orthogonal basis vectors, where the value of L is an integer greater than or equal to T / 4. This can, on the basis of controlling the feedback overhead, use as many orthogonal basis vectors as possible to describe the corresponding right singular vector.

[0069] Since the channel energy from the base station to the terminal mainly concentrates in specific directions at this time, according to simulation and actual measurement analysis, using the method of spatial orthogonal projection on the right singular vector can well describe this channel.

[0070] More preferably, when the signal-to-interference-plus-noise ratio of the channel from the base station to the terminal is greater than or equal to 10 dB, the terminal transmits the third channel state information using at least 0.75 times the full power. The reason is that the downlink channel energy is relatively dispersed at this time, there is a certain correlation between time-domain channels, and the possibility of this user performing multi-user transmission is relatively large, so it is necessary to improve the probability of successful transmission of this channel state information as much as possible.

[0071] Step S110: The base station decodes the third channel state information.

[0072] In this step, after receiving the third channel state information, the base station uses L mutually orthogonal basis vectors to obtain the first two columns of the right singular matrix of N according to 2N×L projection values.

[0073] In the technical solution of the present invention, N frequency-domain channel state information reference signals sent by the base station are measured to obtain the channel matrices of N frequency-domain channels; the channel matrices of each frequency-domain channel are subjected to singular value decomposition to obtain the left singular matrix, the diagonal matrix containing singular values, and the right singular matrix corresponding to each frequency-domain channel; for each diagonal matrix, calculate the ratio of the largest singular value in the diagonal matrix to the sum of all singular values. If the calculated ratio is greater than or equal to the set threshold, then determine the diagonal matrix as the characteristic diagonal matrix; count the number of characteristic diagonal matrices; if the counted number is greater than or equal to the first set value, generate the first channel state information based on the first column of each right singular matrix and feedback it to the base station; if the counted number is less than the second set value, generate the second channel state information based on the time-domain channel vector obtained by Fourier transform of the channel matrices of N frequency-domain channels and feedback it to the base station; if the counted number is between the first and second set values, generate the third channel state information based on the first two columns of each right singular matrix and feedback it to the base station.

[0074] Since the result obtained by the singular value decomposition of the channel matrix, that is, the counted number of characteristic diagonal matrices, can be used to analyze the spatial characteristics of the channel, and based on different spatial characteristics of the channel, corresponding channel processing methods are adopted to generate channel state information; that is to say, the terminal can feedback channel state information matching the spatial characteristics of the channel to the base station, providing a flexible way of feedback of channel state information, thereby solving the problem of poor flexibility in feedback of channel state information and improving the service quality of the network.

[0075] Figure 2 FIG. schematically shows a hardware architecture diagram of a computer device 1300 suitable for implementing the channel state information feedback method according to an embodiment of the present application. In this embodiment, the computer device 1300 is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions. For example, it can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a rack server, a blade server, a tower server or a cabinet server (including an independent server, or a server cluster composed of multiple servers), etc. As Figure 2 shown, the computer device 1300 at least includes, but is not limited to: a memory 1310, a processor 1320, and a network interface 1330 that can be communicatively linked to each other through a system bus. Among them:

[0076] The memory 1310 includes at least one type of computer-readable storage medium. The readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 1310 may be an internal storage module of the computer device 1300, such as the hard disk or memory of the computer device 1300. In other embodiments, the memory 1310 may also be an external storage device of the computer device 1300, such as a plug-in hard disk equipped on the computer device 1300, a Smart Media Card (abbreviated as SMC), a Secure Digital (abbreviated as SD) card, a Flash Card, etc. Of course, the memory 1310 may also include both the internal storage module and the external storage device of the computer device 1300. In this embodiment, the memory 1310 is generally used to store the operating system and various application software installed on the computer device 1300, such as the program code of the channel state information feedback method. In addition, the memory 1310 can also be used to temporarily store various data that have been output or will be output.

[0077] In some embodiments, the processor 1320 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips. The processor 1320 is generally used to control the overall operation of the computer device 1300, such as performing control and processing related to data interaction or communication with the computer device 1300. In this embodiment, the processor 1320 is used to run the program code stored in the memory 1310 or process data.

[0078] The network interface 1330 may include a wireless network interface or a wired network interface, which is generally used to establish a communication link between the computer device 1300 and other computer devices. For example, the network interface 1330 is used to connect the computer device 1300 to an external terminal through a network, and establish a data transmission channel and a communication link between the computer device 1300 and the external terminal. The network may be a wireless or wired network such as an enterprise intranet (Intranet), the Internet, the Global System of Mobile communication (GSM for short), Wideband Code Division Multiple Access (WCDMA for short), 4G network, 5G network, Bluetooth, Wi-Fi, etc.

[0079] It should be noted that Figure 2 Only the computer device with components 1310 - 1330 is shown, but it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively.

[0080] In this embodiment, the channel state information feedback method stored in the memory 1310 can also be divided into one or more program modules and executed by one or more processors (processor 1320 in this embodiment) to complete the embodiments of the present application.

[0081] The embodiments of the present application also provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the channel state information feedback method in the embodiments are implemented.

[0082] In this embodiment, the computer-readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (such as SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the computer-readable storage medium may be an internal storage unit of a computer device, such as the hard disk or memory of the computer device. In other embodiments, the computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk equipped on the computer device, a Smart Media Card (SMC for short), a Secure Digital (SD for short) card, a Flash Card, etc. Of course, the computer-readable storage medium may also include both the internal storage unit and the external storage device of the computer device. In this embodiment, the computer-readable storage medium is generally used to store the operating system and various application software installed on the computer device, such as the program code of the channel state information feedback method in the embodiment. In addition, the computer-readable storage medium can also be used to temporarily store various data that have been output or will be output.

[0083] Obviously, those skilled in the art should understand that the above-mentioned modules or steps of the embodiments of the present application can be implemented by a general-purpose computing device. They can be concentrated on a single computing device or distributed on a network composed of multiple computing devices. Optionally, they can be implemented by program codes executable by the computing device, so that they can be stored in a storage device and executed by the computing device. And in some cases, the steps shown or described can be executed in a different order from here, or they can be separately made into individual integrated circuit modules, or multiple modules or steps among them can be made into a single integrated circuit module to be implemented. In this way, the embodiments of the present application are not limited to any specific combination of hardware and software.

[0084] The above are only the preferred embodiments of the present application, and do not limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made by using the content of the specification and drawings of the present application, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present application.

Claims

1. A method for channel state information feedback, characterized in that Including: Measuring N groups of frequency-domain channel state information reference signals sent by a base station to obtain channel matrices of N frequency-domain channels; Performing singular value decomposition on the channel matrix of each frequency-domain channel to obtain a left singular matrix, a diagonal matrix containing singular values, and a right singular matrix corresponding to each frequency-domain channel; For each diagonal matrix, calculating the ratio of the largest singular value in the diagonal matrix to the sum of all singular values. If the calculated ratio is greater than or equal to a set threshold, determining the diagonal matrix as a characteristic diagonal matrix; Counting the number of characteristic diagonal matrices; if the counted number is greater than or equal to a first set value, generating first channel state information based on the first column of each right singular matrix and feeding it back to the base station; if the counted number is less than a second set value, generating second channel state information based on the time-domain channel vector obtained by performing Fourier transform on the channel matrices of N frequency-domain channels and feeding it back to the base station; if the counted number is between the first and second set values, generating third channel state information based on the first two columns of each right singular matrix and feeding it back to the base station.

2. The method according to claim 1, characterized in that, The generating of the first channel state information based on the first column of each right singular matrix specifically includes: Processing the first column of each right singular matrix using a first encoding neural network based on an autoencoder structure to generate the first channel state information.

3. The method according to claim 2, wherein Also including: After receiving the first channel state information, the base station decodes the first channel state information using a first decoding neural network based on an autoencoder structure to obtain the first column of each right singular matrix.

4. The method according to claim 1, characterized in that The generating of the second channel state information based on the time-domain channel vector obtained by performing Fourier transform on the channel matrices of N frequency-domain channels specifically includes: Processing the time-domain channel vector using a second encoding neural network based on an autoencoder structure to generate the second channel state information.

5. The method according to claim 4, characterized in that Also including: After receiving the second channel state information, the base station decodes the second channel state information using a second decoding neural network based on an autoencoder structure to obtain the time-domain channel vector.

6. The method according to claim 1, characterized in that, The generating of the third channel state information based on the first two columns of each right singular matrix specifically includes: Obtaining the first two columns of each right singular matrix, projecting each obtained column vector onto L mutually orthogonal basis vectors to obtain 2N×L projection values, and performing quantization processing on the 2N×L projection values to obtain the third channel state information; wherein, the value of L is an integer greater than or equal to T / 4, and T is the number of transmitting antennas of the base station.

7. The method according to claim 6, wherein Also including: After receiving the third channel state information, the base station uses L mutually orthogonal basis vectors to obtain the first two columns of N right singular matrices based on the 2N×L projection values.

8. The method according to any one of claims 1-7, characterized in that The set threshold is specifically 0.8; and the first set value is specifically N / 2, and the second set value is specifically N / 4.

9. A computer device, the computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it is used to implement the steps of the channel state information feedback method described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and the computer program can be executed by at least one processor, so that the at least one processor executes the steps of the channel state information feedback method described in any one of claims 1 to 8.

Images