Communication methods and communication devices

Abstract

A communication method and apparatus are disclosed. The method includes: after a first communication device acquires a first downlink feature vector matrix, it determines a codeword corresponding to the first downlink feature vector matrix based on an encoder and a processing function, and sends the codeword to a second communication device. The processing function includes at least one of a first function and a second function. The first function adjusts the feature vectors of the matrix input to the first function, and the second function adjusts the data format of the matrix input to the second function. In determining the codeword to be fed back, compression processing is not only performed based on the encoder, but preprocessing can also be performed based on the processing function. The first function can optimize the feature vectors of the downlink feature vector matrix to improve the correlation between uplink and downlink and reduce feedback overhead. The second function can adjust the data format of the downlink feature vector matrix to generalize the application scenarios of the encoder. This can improve the performance of CSI feedback schemes based on deep learning.

Description

Technical Field

[0001] This application relates to the field of wireless communication technology, and more specifically, to a communication method and a communication device. Background Technology

[0002] Multiple-input multiple-output (MIMO) technology, as a hallmark of current wireless communication, has significantly improved the spectrum and energy efficiency of systems. The success of massive MIMO in downlink operation hinges on the accurate estimation of the base station's channel state information (CSI) to achieve efficient transmission precoding.

[0003] For frequency division duplex (FDD) massive MIMO systems, the use of large antenna arrays leads to a significant increase in the amount of feedback data. Deep learning (DL) based CSI feedback methods have shown great promise in FDD massive MIMO systems, improving the accuracy and time efficiency of channel state information recovery.

[0004] However, current CSI feedback methods based on dataset-adapted deep learning models exhibit poor performance. For example, deep learning models trained for specific RF environments lose effectiveness when applied to different environments due to model mismatch. Furthermore, while using a shared encoder with multiple task-specific decoders may reduce storage and update costs, it risks reducing CSI recovery accuracy because it fails to capture environment-specific features. Therefore, improving the performance of deep learning-based CSI feedback methods is a pressing issue. Summary of the Invention

[0005] This application provides a communication method to improve the performance of a DL-based CSI feedback scheme.

[0006] Firstly, a communication method is provided. This method can be executed by a first communication device. Unless otherwise specified, the "first communication device" in this application can refer to the first communication device itself (e.g., a terminal device), or a component of the first communication device (e.g., a processor, chip, or chip system, such as the circuit or chip responsible for communication functions in the terminal device (e.g., a modem chip, also known as a baseband chip, or a system-on-chip (SoC) chip containing a modem core, or a system-in-package (SIP) chip), or it can be a logic module or software that can implement all or part of the first communication device. For ease of description, the following description uses the execution of the first communication device as an example.

[0007] The communication method includes: acquiring a first downlink feature vector matrix, the first downlink feature vector matrix being used to indicate downlink channel state information; determining a codeword corresponding to the first downlink feature vector matrix based on an encoder and a processing function; and transmitting the codeword to a second communication device. The processing function includes at least one of a first function and a second function, the first function being used to adjust the feature vector of the matrix input to the first function, and the second function being used to adjust the data format of the matrix input to the second function.

[0008] Based on the above technical solution, after the first communication device obtains the first downlink feature vector matrix, it can process the first downlink feature vector matrix based on the encoder and the first function and / or the second function to obtain the codewords fed back to the second communication device. The encoder can compress the data to be fed back, reducing feedback overhead.

[0009] Furthermore, in this technical solution, during the processing of the first downlink feature vector matrix by the first communication device, a first function and / or a second function can be utilized. The first function can adjust the phase of the eigenvectors of the matrix input to it, thereby optimizing the first downlink feature vector matrix based on the first function. This makes the eigenvectors of the first downlink feature vector matrix as close as possible to the reference eigenvectors, improving the correlation between uplink and downlink and further reducing feedback overhead. Additionally, the second function can adjust the data format of the matrix input to it, thus generalizing the application scenarios of the encoder.

[0010] Therefore, if the first function and / or the second function are used to process the first downlink feature vector matrix during the process of determining the codeword to be fed back, the performance of the DL-based CSI feedback scheme can be improved.

[0011] In conjunction with the first aspect, in certain implementations of the first aspect, when the processing function includes both the first function and the second function, determining the codeword corresponding to the first downlink feature vector matrix based on the encoder and the processing function includes: obtaining a second downlink feature vector matrix based on the first function and the first downlink feature vector matrix, wherein the second downlink feature vector matrix is ​​related to the downlink channel matrix and the downlink channel matrix is ​​related to the uplink channel matrix; obtaining a third downlink feature vector matrix based on the second downlink feature vector matrix, the downlink reference feature vector matrix, and the second function, wherein the data format of the third downlink feature vector matrix is ​​the same as the data format of the downlink reference feature vector matrix; and determining the codeword based on the third downlink feature vector matrix and the encoder.

[0012] Based on the above technical solution, after the first communication device obtains the first downlink feature vector matrix, it can process the first downlink feature vector matrix based on the encoder, as well as the first and second functions, to obtain the codewords fed back to the second communication device. Specifically, before compression processing by the encoder, the first downlink feature vector matrix to be fed back is optimized using the first function to obtain the second downlink feature vector matrix. Then, the second downlink feature vector matrix is ​​adjusted in data format using the second function to obtain the third downlink feature vector matrix input to the encoder. It should be understood that the feature vector optimization and data format adjustment of the first downlink feature vector matrix are preprocessing steps before encoder compression processing, thereby reducing the complexity of the encoder model through efficient preprocessing.

[0013] In conjunction with the first aspect, in certain implementations of the first aspect, the encoder, the first function, the second function, the first downlink feature vector matrix, and the codeword satisfy the following relationship:

[0014] W BCE，DL =f BCE (W DL )

[0015] W IFA，DL b DL =f IFA (W BCE，DL W Ben，DL )

[0016] c = f en (W IFA，DL ;Φ)

[0017] Among them, f BCE (·) represents the first function, W DL W represents the first downlink eigenvector matrix. BCE，DL Let f represent the second downlink eigenvector matrix.IFA (·) represents the second function, W IFA，DL W represents the third downlink eigenvector matrix. Ben，DL Let b represent the downlink reference eigenvector matrix. DL This represents first control information, which is used by the second communication device to reconstruct the second downlink feature vector matrix, wherein f en (·;Φ) represents the encoder, Φ represents the network parameters of the encoder, and c represents the codeword.

[0018] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: sending the first control information to the second communication device.

[0019] Based on the above technical solution, the first communication device can send the first control information for reconstructing the second downlink feature vector matrix to the second communication device, so that the second communication device can accurately complete the reconstruction of the second downlink feature vector matrix.

[0020] In conjunction with the first aspect, in some implementations of the first aspect, obtaining the second downlink feature vector matrix based on the first function and the first downlink feature vector matrix includes: aligning the feature vector corresponding to the s-th sub-band in the first downlink feature vector matrix with the reference vector of the downlink channel matrix corresponding to the s-th sub-band based on the first function, to obtain the feature vector corresponding to the s-th sub-band in the second downlink feature vector matrix. The N assigned to the communication system to which the first communication device belongs... s N corresponding to each sub-band s The normalized eigenvectors are concatenated to obtain the second downlink eigenvector matrix, where N s s is a positive integer, and the value of s is less than or equal to N. s Positive integers.

[0021] Based on the above technical solution, in the process of enhancing the correlation between uplink and downlink channels by optimizing the eigenvectors of the first downlink eigenvector matrix in the feature space using the first function, the reference vector of the downlink channel matrix is ​​taken into account. That is, the inherent strong correlation between the uplink and downlink channels is utilized to enhance the bidirectional correlation of the eigenvector matrix. Since they experience the same propagation environment, the uplink and downlink channel matrices exhibit strong correlation before eigenvector decomposition. Therefore, more closely aligning the downlink eigenvector with the downlink channel vector in its subband can improve its correlation with the uplink eigenvector. Thus, the bidirectional correlation of the first downlink eigenvector matrix can be enhanced using the downlink channel matrix measured by the first communication device without additional transmission costs.

[0022] In conjunction with the first aspect, in some implementations of the first aspect, the eigenvector corresponding to the s-th sub-band in the second downlink eigenvector matrix is ​​represented by at least one of the following:

[0023] or, or,

[0024] in, Let ||·|| represent the eigenvector corresponding to the s-th sub-band in the second downlink eigenvector matrix, ||·|| represent the l2 norm, and ε s，DL Represents the eigenvalue λ s，DL The corresponding feature space, h s，DL w represents the reference vector of the downlink channel matrix corresponding to the s-th subband. s，DL This represents the eigenvector corresponding to the s-th sub-band in the first downlink eigenvector matrix. Used to h s，DL Projected onto feature space ε s，DL , The columns constitute ε s，DL Orthogonal basis.

[0025] In conjunction with the first aspect, in some implementations of the first aspect, obtaining the third downlink eigenvector matrix based on the second downlink eigenvector matrix and the second function includes: converting the second downlink eigenvector matrix into a downlink sparse eigenvector matrix using a two-dimensional discrete Fourier transform; obtaining a first row cyclic shift step size and a first column cyclic shift step size based on the downlink sparse eigenvector matrix and the downlink reference eigenvector matrix; and obtaining the third downlink eigenvector matrix based on the downlink sparse eigenvector matrix, the first row cyclic shift step size, the first column cyclic shift step size, and the second function.

[0026] Based on the above technical solution, in the process of obtaining the third downlink feature vector matrix by adjusting the data format of the second downlink feature vector matrix input to the second function, the input format of the encoder is standardized based on the predefined benchmark matrix to ensure a consistent input data distribution, which significantly improves the generalization performance of a single encoder model.

[0027] In conjunction with the first aspect, in some implementations of the first aspect, the second downlink eigenvector matrix and the downlink sparse eigenvector matrix satisfy the following formula:

[0028]

[0029] Among them, W Spar，DL W represents the downlink sparse eigenvector matrix. BCE，DLLet F represent the second downlink eigenvector matrix. d and F a The Discrete Fourier Transform (DFT) matrix is... This represents the two-dimensional discrete Fourier transform.

[0030] The downlink sparse eigenvector matrix, the first row cyclic shift step size, the first column cyclic shift step size, the second function, and the third downlink eigenvector matrix satisfy the following formula:

[0031]

[0032] Among them, W IFA，DL Let f represent the third downlink eigenvector matrix. IFA (·) represents the second function. This indicates the cyclic shift step size of the first row. This indicates the cyclic shift step size of the first column.

[0033] In conjunction with the first aspect, in some implementations of the first aspect, determining the codeword based on the third downlink feature vector matrix and the encoder includes: compressing the third downlink feature vector matrix based on the encoder to obtain the codeword.

[0034] Secondly, a communication method is provided. This method can be executed by a second communication device. Unless otherwise specified, the "second communication device" in this application can refer to the second communication device itself (e.g., a network device), or a component of the second communication device (e.g., a processor, chip, or chip system, such as a circuit or chip in a network device responsible for communication functions (e.g., a modem chip, also known as a baseband chip, or a SoC chip or SIP chip containing a modem core), or it can be a logic module or software that can implement all or part of the functions of the second communication device. For ease of description, the following description uses the execution by a second communication device as an example.

[0035] The communication method includes: receiving a codeword from a first communication device, the codeword being a codeword corresponding to a first downlink feature vector matrix indicating downlink channel state information; determining a second uplink feature vector matrix based on a first uplink feature vector matrix and a processing function, the first uplink feature vector matrix being used to indicate uplink channel state information; and obtaining a reconstructed third downlink feature vector matrix based on a decoder, the codeword, and the second uplink feature vector matrix, wherein the processing function includes at least one of a first function and a second function, the first function being used to adjust the feature vectors of the matrix input to the first function, the second function being used to adjust the data format of the matrix input to the second function, and the second downlink feature vector matrix being determined by the first function and the first downlink feature vector matrix.

[0036] In conjunction with the second aspect, in some implementations of the second aspect, when the processing function includes the first function and the second function, determining the second uplink feature vector matrix based on the first uplink feature vector matrix and the processing function includes: determining a third uplink feature vector matrix based on the first uplink feature vector matrix and the first function, wherein the third uplink feature vector matrix is ​​related to the uplink channel matrix and the uplink channel matrix is ​​related to the downlink channel matrix; and determining the second uplink feature vector matrix based on the third uplink feature vector matrix, the uplink reference feature vector matrix, and the second function, wherein the data format of the second uplink feature vector matrix is ​​the same as the data format of the uplink reference feature vector matrix.

[0037] In conjunction with the second aspect, in some implementations of the second aspect, the first uplink eigenvector matrix, the first function, the second function, the third uplink eigenvector matrix, and the second uplink eigenvector matrix satisfy the following relationship:

[0038] W BCE，UL =f BCE (W UL )

[0039] W IFA，UL b UL =f IFA (W BCE，UL W Ben，UL )

[0040] Among them, f BCE (·) represents the first function, W UL W represents the first uplink eigenvector matrix. BCE，UL f represents the third upward eigenvector matrix. IFA (·) represents the second function, the W Ben，UL Let W represent the reference matrix. Ben，ULLet b represent the uplink reference eigenvector matrix. UL This indicates the second control information, the W IFA，UL This represents the second upward eigenvector matrix.

[0041] In conjunction with the second aspect, in some implementations of the second aspect, determining the third uplink feature vector matrix based on the first uplink feature vector matrix and the first function includes: aligning the feature vector corresponding to the s-th sub-band in the first uplink feature vector matrix with the reference vector of the uplink channel matrix corresponding to the s-th sub-band based on the first function, to obtain the feature vector corresponding to the s-th sub-band in the third uplink feature vector matrix; and allocating N to the communication system to which the second communication device belongs. s N corresponding to each sub-band s The normalized eigenvectors are concatenated to obtain the third upward eigenvector matrix, where N s s is a positive integer, and the value of s is less than or equal to N. s Positive integers.

[0042] In conjunction with the second aspect, in some implementations of the second aspect, the eigenvector corresponding to the s-th sub-band in the third uplink eigenvector matrix is ​​represented by at least one of the following:

[0043] or, or,

[0044] in, This represents the eigenvector corresponding to the s-th sub-band in the third upward eigenvector matrix, ||·|| represents the l2 norm, and ε s，UL Represents the eigenvalue λ s，UL The corresponding feature space, h s，UL w represents the reference vector of the uplink channel matrix corresponding to the s-th sub-band. s，UL This represents the eigenvector corresponding to the s-th sub-band in the first uplink eigenvector matrix. Used to h s，UL Projected onto feature space ε s，UL , The columns constitute ε s，UL Orthogonal basis.

[0045] In conjunction with the second aspect, in some implementations of the second aspect, determining the second upward feature vector matrix based on the third upward feature vector matrix, the upward reference feature vector matrix, and the second function includes: converting the third upward feature vector matrix into an upward sparse feature vector matrix through a two-dimensional discrete Fourier transform; obtaining the second row cyclic shift step size and the second column cyclic shift step size based on the upward sparse feature vector matrix and the upward reference feature vector matrix; and obtaining the second upward feature vector matrix based on the upward reference feature vector matrix, the second row cyclic shift step size, the second column cyclic shift step size, and the second function.

[0046] In conjunction with the second aspect, in some implementations of the second aspect, the third uplink eigenvector matrix and the uplink sparse eigenvector matrix satisfy the following formula:

[0047]

[0048] Among them, W Spar，UL W represents the uplink sparse eigenvector matrix. BCE，UL F represents the third upward eigenvector matrix. d and F a The Discrete Fourier Transform (DFT) matrix is... This represents the two-dimensional discrete Fourier transform.

[0049] The up-row sparse eigenvector matrix, the second row cyclic shift step size, the second column cyclic shift step size, the second function, and the second up-row eigenvector matrix satisfy the following formula:

[0050]

[0051] Among them, W IFA，UL Let f represent the second upward eigenvector matrix. IFA (·) represents the second function. This indicates the cyclic shift step size for the second row. This indicates the cyclic shift step size of the second column.

[0052] In conjunction with the second aspect, in some implementations of the second aspect, the decoder, the codeword, the second uplink feature vector matrix, and the reconstructed third downlink feature vector matrix satisfy the following relationship:

[0053]

[0054] Among them, f de (·;Ψ) represents the decoder, Ψ represents the network parameters of the decoder, |W IFA，UL| represents the modulus of the second uplink feature vector matrix, and c represents the codeword. This represents the third row eigenvector matrix after reconstruction.

[0055] In conjunction with the second aspect, in some implementations of the second aspect, the method further includes: obtaining a reconstructed second downlink feature vector matrix based on the inverse function of the second function, the reconstructed third downlink feature vector matrix, and the second control information, wherein the third downlink feature vector matrix is ​​determined by the second downlink feature vector matrix and the second function.

[0056] In conjunction with the second aspect, in some implementations of the second aspect, the inverse function of the second function, the reconstructed third downlink eigenvector matrix, the second control information, and the reconstructed second downlink eigenvector matrix satisfy the following relationship:

[0057]

[0058] in, Denotes the inverse function of the second function, where b UL This indicates the second control information, the This represents the third downlink eigenvector matrix after reconstruction. This represents the second downlink feature vector matrix after reconstruction.

[0059] Regarding the beneficial effects not described in detail in the second aspect, please refer to the relevant description in the first aspect, which will not be repeated here.

[0060] Thirdly, a communication device is provided, which may be a first communication device, or a device or module for performing the functions of the first communication device.

[0061] One possible implementation is that the communication device may include modules or units corresponding to the methods / operations / steps / actions described in the first aspect, which may be hardware circuits, software, or a combination of hardware circuits and software.

[0062] In one design, the device may include a processing module and a communication module. The communication module is used to perform the sending and receiving actions performed by the first communication device in the method described in the first aspect above, while the processing module is used to perform processing-related actions performed by the first communication device in the method described in the first aspect above.

[0063] In one design, the device can be a terminal device, or a device, module, circuit, or chip configured in the terminal device, or a device that can be used in conjunction with the terminal device.

[0064] Fourthly, a communication device is provided, which may be a second communication device, or a device or module for performing the functions of a second communication device.

[0065] One possible implementation is that the communication device may include modules or units corresponding to the methods / operations / steps / actions described in any of the second aspects, wherein the modules or units may be hardware circuits, software, or a combination of hardware circuits and software.

[0066] In one design, the device may include a processing module and a communication module. The communication module is used to perform the sending and receiving actions performed by the second communication device in the method described in the second aspect above, while the processing module is used to perform processing-related actions performed by the second communication device in the method described in the second aspect above.

[0067] In one design, the device can be a network device, or a device, module, circuit, or chip configured in the network device, or a device that can be used in conjunction with the network device, such as an intelligent network element with a deployed radio intelligent controller (RIC).

[0068] Fifthly, a communication apparatus is provided, comprising: at least one processor for executing a computer program or instructions to perform the methods described in the first aspect and any possible implementations of the first and second aspects. Optionally, the apparatus further comprises a memory for storing the computer program or instructions. Optionally, the apparatus further comprises a communication interface through which the processor reads the computer program or instructions.

[0069] In one implementation, the device is a communication device (such as a terminal device or a network device).

[0070] In another implementation, the device is a chip, chip system, or circuit for communication equipment (such as terminal equipment or network equipment).

[0071] Sixthly, a processor is provided for performing the methods provided in the first and second aspects described above.

[0072] Unless otherwise specified, or if it does not contradict its actual function or internal logic in the relevant description, the transmission and acquisition / reception operations involved in the processor can be understood as processor output and reception, input and other operations, or as transmission and reception operations performed by radio frequency circuits and antennas. This application does not limit them in this regard.

[0073] Optionally, the device further includes: a memory for storing a program; correspondingly, at least one processor for executing the computer program or instructions in the memory.

[0074] Optionally, the device also includes a communication interface. The communication interface is coupled to the processor and can be used to input information to the processor or output information from the processor.

[0075] A seventh aspect provides a computer-readable storage medium storing program code for execution by a device, the program code including methods for performing any possible implementation of the first and second aspects described above.

[0076] Eighthly, a computer program product containing instructions is provided, which, when run on a computer, causes the computer to perform the method in any possible implementation of the first and second aspects described above.

[0077] Ninth aspect, a chip is provided, the chip including a processing circuit and a communication interface, the processing circuit reading instructions from a memory through the communication interface and executing the method provided by any of the implementations of the first and second aspects above.

[0078] Optionally, the processing circuit is one or more processors, or all or part of the control or processing circuitry included in one or more processors.

[0079] Optionally, as one implementation, the chip also includes a memory storing computer programs or instructions, and a processor for executing the computer programs or instructions in the memory. When the computer programs or instructions are executed, the processor is used to perform the methods provided by any of the implementations of the first and second aspects described above.

[0080] In a tenth aspect, a communication system is provided, including a first communication device and a second communication device. The second communication device is used to implement the method provided in any possible implementation of the second aspect; the first communication device is used to implement the method provided in any possible implementation of the first aspect. Attached Figure Description

[0081] Figure 1 This is an architecture diagram of a communication system applicable to embodiments of this application.

[0082] Figure 2 This is a schematic diagram of an Open Radio Access Network (ORAN) system architecture.

[0083] Figure 3 This is a schematic diagram of a neural network.

[0084] Figure 4 This is a schematic diagram of AI-CSI feedback.

[0085] Figure 5 This is a schematic flowchart of a communication method provided in this application.

[0086] Figure 6 This is a schematic diagram of a CSI feedback method provided in this application.

[0087] Figure 7 This is a schematic diagram of the CSI feedback performance provided in this application.

[0088] Figure 8 This is a schematic diagram of another CSI feedback performance provided in this application.

[0089] Figure 9 This is a schematic diagram of an encoder and decoder provided in this application.

[0090] Figure 10 This is a schematic diagram of another type of CSI feedback provided in this application.

[0091] Figure 11 This is a schematic block diagram of a communication device provided in an embodiment of this application.

[0092] Figure 12 This is a schematic diagram of another communication device provided in an embodiment of this application.

[0093] Figure 13 This is a schematic diagram of a chip system provided in an embodiment of this application. Detailed Implementation

[0094] To facilitate understanding of the embodiments of this application, the following points will be explained first.

[0095] First, in this application, "for indicating" can include both direct and indirect indication. When describing an indication message as indicating A, it can include whether the indication message directly indicates A or indirectly indicates A, but does not necessarily mean that the indication message carries A.

[0096] The information indicated by the instruction is called the information to be instructed. In the specific implementation process, there are many ways to indicate the information to be instructed, such as, but not limited to, directly indicating the information to be instructed, such as the information to be instructed itself or its index. It can also be indirectly indicated by indicating other information, where there is a relationship between the other information and the information to be instructed. It can also indicate only a part of the information to be indicated, while the other parts are known or pre-agreed upon. For example, the instruction of specific information can be achieved by using a pre-agreed (e.g., protocol-defined) arrangement of various pieces of information, thereby reducing instruction overhead to some extent. At the same time, common parts of various pieces of information can be identified and indicated uniformly to reduce the instruction overhead caused by individually indicating the same information.

[0097] Second, in this application, "at least one" refers to one or more, and "more than one" refers to two or more (including two). Furthermore, in the embodiments of this application, "first," "second," and various numerical designations (e.g., "#1," "#2," etc.) are merely for descriptive convenience and are not intended to limit the scope of the embodiments of this application. The sequence numbers of the processes below do not imply an order of execution; the execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application. It should be understood that the objects described in this way can be interchanged where appropriate to describe solutions other than those in the embodiments of this application. Moreover, in the embodiments of this application, terms such as "S510" are merely identifiers for descriptive convenience and do not limit the order of execution steps.

[0098] Third, in the embodiments of this application, the words "exemplary" or "for example" are used to indicate that they are examples, illustrations, or descriptions. Any embodiment or design that is described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design options. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0099] Fourth, the term "storage" in the embodiments of this application can refer to storage in one or more memories. These memories can be separate installations or integrated into an encoder, decoder, processor, or communication device. Alternatively, some memories can be separately installed, while others can be integrated into the decoder, processor, or communication device. The type of memory can be any form of storage medium, and this application does not limit this.

[0100] Fifth, in the implementation of this application, "protocol" may refer to standard protocols in the field of communications, such as the NR protocol and related protocols applied in future communication systems, and this application does not limit it.

[0101] Sixth, in the embodiments of this application, the terms "of", "corresponding (relevant)", "corresponding", and "associate" can sometimes be used interchangeably. It should be noted that when their differences are not emphasized, their intended meanings are consistent.

[0102] Seventh, in the embodiments of this application, "under the circumstances", "when", and "if" can sometimes be used interchangeably. It should be noted that when the distinction is not emphasized, their intended meanings are consistent.

[0103] Eighth, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0104] Ninth, the terms "message", "information", or "information element (IE)" can be used interchangeably in this article. There are no restrictions on the names of messages, information, or frames, as long as they can achieve the corresponding functions.

[0105] Tenth, in this application, "send" and "receive" indicate the direction of signal transmission. For example, "send information to XX" can be understood as the destination of the information being XX, and "send information" can include direct transmission or indirect transmission through other units or modules. "Receive information from YY" can be understood as the source of the information being YY, and "receive information" can include direct reception from YY or indirect reception from YY through other units or modules. Besides air interface transmission or reception signals implemented at the system level, such as network devices or terminal devices, "send" can also be understood as the "output" of a chip interface, and "receive" can also be understood as the "input" of a chip interface. For example, a modem or system-on-a-chip (SoC) chip or system-in-package (SIP) chip transmits or receives signals. "Send" or "receive" can also be performed through device components, for example, by using buses, traces, or interfaces to transmit or receive signals through several parts, modules, or chips of a device.

[0106] The technical solutions in this application will now be described with reference to the accompanying drawings.

[0107] The embodiments of this application can be applied to various communication systems, including but not limited to: 5th generation (5G) systems, LTE systems, long term evolution-advanced (LTE-A) systems, LTE frequency division duplex (FDD) systems, LTE time division duplex (TDD) systems, etc. They can also be applied to future communication systems, such as 6th generation mobile communication systems. Furthermore, they can be applied to device-to-device (D2D) communication, vehicle-to-everything (V2X) communication, machine-to-machine (M2M) communication, machine-type communication (MTC), Internet of Things (IoT) communication systems, narrowband Internet of Things (NB-IoT) systems, or other communication systems. Furthermore, it can be extended to similar wireless communication systems, such as Wireless-Fidelity (WiFi), Worldwide Interoperability for Microwave Access (WIMAX), and communication systems related to the 3rd Generation Partnership Project (3GPP), without limitation.

[0108] The communication system applicable to embodiments of this application may include one or more data transmitters and one or more data receivers. Optionally, one of the transmitters and receivers may be a terminal device and the other a network device.

[0109] Figure 1 This is an architecture diagram of a communication system applicable to embodiments of this application. (See diagram below.) Figure 1 As shown, the embodiments of this application can be applied to both uplink and downlink transmissions. Figure 1 This example uses only uplink or downlink transmission between one network device and two terminal devices (e.g., terminal device 1 and terminal device 2). In uplink transmission, the data sender is the terminal device, and the data receiver is the network device; in downlink transmission, the sender is the network device, and the receiver is the terminal device.

[0110] For ease of understanding, the following describes the equipment (or network elements, nodes, etc.) that may be involved in this application.

[0111] Terminal equipment: can be called user equipment (UE), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication equipment, user agent, or user device.

[0112] Terminal devices can be devices that provide voice / data connectivity to users, such as handheld devices with wireless connectivity, in-vehicle devices, etc. Currently, examples of terminals include: mobile phones, tablets, laptops, PDAs, mobile internet devices (MIDs), wearable devices, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, wireless terminals in self-driving, wireless terminals in remote medical surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, cellular phones, cordless phones, session initiation protocol phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), handheld devices with wireless communication capabilities, computing devices or other processing devices connected to wireless modems, in-vehicle devices, wearable devices, terminal devices in 5G networks, or terminal devices in future evolved public land mobile networks (PLMNs), etc., and the embodiments of this application are not limited to these.

[0113] By way of example and not limitation, in this embodiment, the terminal device can also be a wearable device. Wearable devices, also known as wearable smart devices, are a general term for devices that utilize wearable technology to intelligently design and develop everyday wearables, such as glasses, gloves, watches, clothing, and shoes. Wearable devices are portable devices that are worn directly on the body or integrated into the user's clothing or accessories. Wearable devices are not merely hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are feature-rich, large in size, and can achieve complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses, as well as those that focus on a specific type of application function and require the use of other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0114] Furthermore, in this embodiment, the terminal device can also be a terminal device in an IoT system. IoT is an important component of the future development of information technology. Its main technical feature is to connect objects to the network through communication technology, thereby realizing an intelligent network of human-machine interconnection and object-to-object interconnection.

[0115] Network equipment, also known as access network equipment, provides network access functionality for terminal devices and can use transmission tunnels of different quality depending on the user's level and service requirements. Access networks can employ different access technologies. Currently, there are two types of wireless access technologies: 3GPP (3rd Generation Partnership Project) access technologies (such as those used in 3G, 4G, or 5G systems) and non-3GPP access technologies. 3GPP access technologies refer to those that conform to 3GPP standards and specifications; for example, access network equipment in 5G systems is called a next-generation node base station (gNB). Non-3GPP access technologies refer to those that do not conform to 3GPP standards and specifications; for example, air interface technologies represented by access points (APs) in Wireless Fidelity (WiFi).

[0116] An access network that uses wireless communication technology to implement access network functions can be called a radio access network (RAN). The RAN manages radio resources, provides access services to terminal devices, and forwards control signals and user data between the terminal and the core network. The RAN can also be an open RAN (O-RAN).

[0117] RAN nodes, also known as radio access network devices, RAN entities, or access nodes, are used to help terminals access communication systems wirelessly. In one application scenario, an RAN node can be a base station, an evolved NodeB (eNodeB), a transmission reception point (TRP), a next-generation NodeB (gNB) in a 5G mobile communication system, a next-generation base station in a future mobile communication system, a base station in a future mobile communication system, or an access node in a WiFi system. RAN nodes can be macro base stations, micro base stations, indoor stations, relay nodes, or donor nodes.

[0118] In another application scenario, multiple RAN nodes can collaborate to help terminals achieve wireless access, with different RAN nodes implementing different functions of the base station. For example, a RAN node can be a central unit (CU), a distributed unit (DU), or a radio unit (RU). The CU performs the functions of the base station's radio resource control (RRC) protocol and packet data convergence protocol (PDCP), and can also perform the functions of the service data adaptation protocol (SDAP). The DU performs the functions of the base station's radio link control (RLC) layer and medium access control (MAC) layer, and can also perform some or all of the physical layer (PHY) functions. For specific descriptions of these protocol layers, refer to the relevant 3GPP technical specifications. The RU can be used to implement radio frequency signal transmission and reception. The CU and DU can be two independent RAN nodes, or they can be integrated into the same RAN node, such as within a baseband unit (BBU). RUs can be included in radio frequency equipment, such as remote radio units (RRUs) or active antenna units (AAUs). CUs can be further divided into two types of RAN nodes: CU-control plane and CU-user plane.

[0119] In different systems, RAN nodes can have different names. For example, in an Open RAN (O-RAN) system, a CU can also be called an Open CU (O-CU), a DU can also be called an Open DU (O-DU), and a RU can be called an Open RU (O-RU).

[0120] Figure 2 An exemplary schematic diagram of an ORAN system architecture provided in an embodiment of this application is shown. The ORAN system in this embodiment may include... Figure 2 Other components besides those shown. For example... Figure 2 As shown, access network devices can communicate with the core network (CN) via a backhaul link and with terminals via an air interface. For example, a BBU in an access network device communicates with the core network via a backhaul link, and an RU in the access network device communicates with at least one terminal via an air interface. A BBU communicates with at least one RU via a fronthaul link; the BBU and RU may or may not be co-located. A BBU includes at least one CU and at least one DU, which can communicate via at least one midhaul link.

[0121] In this application, the RAN node can be implemented through software modules, hardware modules, or a combination of software and hardware modules. For example, the RAN node can be a server loaded with the corresponding software module. The embodiments of this application do not limit the specific technology or device form used in the RAN node.

[0122] For example, the aforementioned terminal equipment and / or access network equipment includes one or more functional modules for signal processing. Taking physical layer functions as an example, the terminal equipment and / or access network equipment includes one or more of the following functions: coding, rate matching, scrambling, modulation, layer mapping, precoding, resource element (RE) mapping, digital beamforming (BF), inverse fast Fourier transformation (IFFT) / adding a cyclic prefix (CP), decoding, rate matching dematching, descrambling, demodulation, inverse discrete Fourier transformation (IDFT), channel equalization (or channel estimation), RE demapping, digital BF, fast Fourier transform (FFT) / CP removal, digital-to-analog (DA) conversion, analog BF, analog-to-digital (AD) conversion, or analog BF.

[0123] It should be understood that the access network can provide services to the cell. Terminal devices can communicate with the cell through the transmission resources (e.g., frequency domain resources, or spectrum resources) allocated by the access network devices.

[0124] The communication systems and service scenarios described in the embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided in the embodiments of this application. As those skilled in the art will know, with the evolution of network architecture and the emergence of new service scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0125] To facilitate understanding of the embodiments of this application, some basic concepts involved in this application will be briefly explained.

[0126] 1. Artificial Intelligence (AI): This refers to enabling machines to learn, accumulate experience, and solve problems that humans can solve through experience, such as natural language understanding, image recognition, and chess. AI can be understood as the intelligence exhibited by machines created by humans. Generally, AI refers to the technology of using computer programs to represent human intelligence. The goals of AI include understanding intelligence by constructing computer programs that demonstrate symbolic reasoning or reasoning.

[0127] 2. Machine Learning (ML): This is an implementation method of artificial intelligence. Machine learning is a method that endows machines with the ability to perform functions that cannot be done directly by programming. In practical terms, machine learning is a method of training a model using data and then using the model to make predictions. There are many methods of machine learning, such as neural networks (NN), decision trees, and support vector machines. Machine learning theory mainly involves designing and analyzing algorithms that enable computers to learn automatically. Machine learning algorithms are a class of algorithms that automatically analyze data to obtain patterns and use these patterns to predict unknown data.

[0128] 3. Neural Networks: Neural networks are a specific manifestation of machine learning methods. A neural network is a mathematical model that mimics the behavioral characteristics of animal neural networks to process information. For example... Figure 3 As shown, a neural network can be composed of three types of computational layers: input layer, hidden layer, and output layer. Each layer has one or more logical decision units, called neurons. Common neural network structures include feedforward neural networks (FNN), convolutional neural networks (CNN), and recurrent neural networks (RNN), all of which are based on neurons. Each neuron performs a weighted summation operation on its input values ​​and outputs the result through a nonlinear function. The weights of the neuron's weighted summation operation and the nonlinear function are called the parameters of the neural network. The connections between neurons in the neural network are called the structure of the neural network, and the parameters of all neurons constitute the parameters of the neural network.

[0129] 4. Deep neural network: A neural network with multiple hidden layers.

[0130] 5. Deep learning (DL): Machine learning that utilizes deep neural networks.

[0131] 6. AI Model: An AI model is an algorithm or computer program that enables AI functionality. It represents the mapping relationship between the model's input and output; in other words, it's a function model that maps a certain dimension of input to a certain dimension of output. The parameters of this function model can be obtained through machine learning training. For example, f(x) = ax 2+b is a quadratic function model, which can be viewed as an AI model. a and b are the parameters of this AI model, and a and b can be obtained through machine learning training. For example, the AI ​​model mentioned in the following embodiments of this application is not limited to neural networks, linear regression models, decision tree models, support vector machines (SVM), Bayesian networks, Q-learning models, or other machine learning (ML) models.

[0132] The implementation of an AI model can be a hardware circuit, software, or a combination of both; there are no restrictions. Non-restrictive examples of software include: program code, program, subroutine, instruction, instruction set, code, code segment, software module, application program, or software application, etc.

[0133] 7. Deep Learning (DL) Model: A learning model that represents data through multiple layers of non-linear processing units, possessing the ability to automatically extract features from raw data. Deep learning models have demonstrated powerful performance in many fields, especially in computer vision, natural language processing, and tabular data analysis.

[0134] 8. Reference signal: also known as pilot signal. The reference signal involved in this application includes, but is not limited to, the following reference signals:

[0135] Demodulation reference signals (DMRS), channel state information-reference signals (CSI-RS), tracking reference signals (TRS), sounding reference signals (SRS), phase tracking reference signals (PT-RS), positioning reference signals (PRS), and sensing reference signals (SeRS), etc.

[0136] The reference signal in this application may also be a reference signal other than those listed above, which will not be listed here.

[0137] 9. CSI-RS Measurement: In LTE and NR communication systems, the base station needs to obtain downlink CSI to determine the resources, modulation and coding scheme (MCS), precoding, and other configurations of the downlink data channel for scheduling the UE.

[0138] In a time division duplex (TDD) system, due to the reciprocity of uplink and downlink channels, the base station can obtain the uplink CSI by measuring the uplink reference signal, and then infer a more accurate downlink CSI, for example, using the uplink CSI as the downlink CSI.

[0139] In FDD systems, uplink and downlink reciprocity cannot be guaranteed. Downlink CSI is obtained by the UE by measuring downlink reference signals, such as downlink channel state information reference signal (CSI-RS) or synchronization signal / physical broadcast channel block (SSB), and then feeding back the CSI report to the base station so that it can obtain the downlink CSI.

[0140] For FDD massive MIMO systems, the application of large antenna arrays leads to a significant increase in the amount of feedback data. The DL-based CSI feedback method has shown great promise in FDD massive MIMO systems. The DL-based CSI feedback method can improve the accuracy and time efficiency of channel state information recovery.

[0141] For example, if a massive MIMO system base station is configured with N Tx One antenna, user equipment configured with N Rx There are N antennas, and the system is allocated N sub-bands. s Each subband contains N gran Subcarriers. Downlink channel Represented as:

[0142]

[0143] Wherein, in equation (1-1) This represents the downlink channel of the s-th subband, where the subband index s takes values ​​of 1, 2, ..., N. s .

[0144] Assuming ideal channel estimation for user equipment, the normalized eigenvector corresponding to the largest eigenvalue of the s-th subband... As a precoding vector, ||w s，DL || 2=1. For example, the precoding vector of the s-th subband can be obtained through eigenvector decomposition, as shown in equation (1-2):

[0145]

[0146] In equation (1-2) The largest eigenvalue λ s，DL Characterize the precoding power gain of the MIMO system.

[0147] It should be noted that, in order to achieve effective downlink precoding, the user equipment needs to report N to the base station. s Each sub-band corresponds to N s For each feature vector, UE and BS can use various neural network architectures to implement the feature vector matrix. Compression and reconstruction.

[0148] For example, the squared generalized cosine similarity (SGCS) criterion Used to measure the channel state information feedback and reconstruction accuracy of the s-th sub-band:

[0149]

[0150] In equation (1-3) Let represent the reconstructed feature vector of the s-th subband.

[0151] Furthermore, N s The CSI feedback and reconstruction performance of each sub-band can be measured by the average SGCS:

[0152]

[0153] In equation (1-4) A value close to 1 indicates high accuracy in CSI feedback and reconstruction.

[0154] For example, the optimization objective of CSI feedback can be defined as:

[0155]

[0156] In equation (1-5) This represents a set of CSI feedback schemes. For example, the set of CSI feedback schemes includes codebook-based feedback and deep learning-based autoencoder feedback methods.

[0157] 10. AI-based Channel State Information (CSI) Feedback (AI-CSI Feedback): With the development of AI research, the application of neural network models in communication systems is constantly expanding. For example, neural network models can be used for compressed feedback of channel information. The UE obtains downlink channel information based on a reference signal, and then uses this downlink channel information as input to the UE-side neural network to obtain the compressed amount of channel information. The UE feeds back the channel information to the base station, which can then input the feedback amount into the neural network to recover the downlink channel information. By leveraging the nonlinear feature extraction capability of neural networks, the accuracy of channel measurement can be improved. Therefore, it is worth considering using neural networks to solve the channel measurement problem under large arrays and large bandwidths.

[0158] For example, an auto-encoder (AE) model consists of two sub-models: an encoder and a decoder. AE can generally refer to a network structure composed of these two sub-models. AE models can also be called bilateral models, two-end models, or collaborative models. The encoder and decoder of an AE are usually trained together and can be used in a matched manner. Figure 4 As shown, AI-CSI feedback can be implemented based on the AI ​​model of the AE (Autonomous Equipment). For example, the UE side compresses and quantizes the CSI using an encoder, while the base station recovers the CSI using a decoder. For the base station, the input to the AI ​​model is the CSI fed back by the UE, and the output is the recovered CSI. During the training of the AI ​​model on the base station side, the CSI measured on the UE side can serve as the ground truth label for the recovered CSI.

[0159] For example, suppose f en (·;Φ) represents Figure 4 The encoder shown in the figure has Φ representing the encoder's network parameters; f de (·;ψ) Figure 4 The decoder shown is an example where ψ represents the network parameters of the decoder. A CSI feedback scheme based on feature vectors can then be expressed as:

[0160] c = f en (W DL ;Φ) (1-6)

[0161]

[0162] In equation (1-6), W DL Let f represent the eigenvector matrix, and c represent the vector matrix transmitted via f. en (·;Φ) is the compressed codeword; in addition, in equation (1-7) It is via f de (·;ψ) is the reconstructed eigenvector matrix, |WDL | represents the modulus of the eigenvector matrix.

[0163] The above text combined Figure 1 This paper briefly introduces the application scenarios of the communication method provided in the embodiments of this application, and describes the basic concepts that may be involved in the embodiments of this application. Among these basic concepts, AI-CSI feedback is introduced. One possible AI-CSI feedback scheme is deep learning-based CSI feedback, which can improve the efficiency of CSI feedback. The deep learning-based CSI feedback scheme relies on a deep learning model, and the performance of different deep learning models is related to the dataset corresponding to the model. For example, the advantages and disadvantages of models corresponding to different datasets are shown in Table 1 below.

[0164] Table 1

[0165]

[0166]

[0167] As can be seen from the above, CSI feedback based on dataset adaptation models has the following problems:

[0168] 1) Performance degradation under different environments. For example, a deep learning model trained for a specific radio frequency environment loses its effectiveness when applied to different environments due to model mismatch.

[0169] 2) Limitations of training in mixed environments. For example, training on datasets from mixed environments can help with generalization, but it still leads to performance degradation compared to environment-specific models, and faces practical deployment challenges on energy-constrained devices due to the complexity and size of the models.

[0170] 3) High memory and bandwidth requirements. For example, multi-encoder to multi-decoder strategies require a large amount of memory to store a large number of deep learning networks, or a large amount of bandwidth to download new encoder parameters.

[0171] 4) The trade-off between model accuracy and model size / complexity is complex. For example, using a shared encoder with multiple task-specific decoders may reduce storage and update costs, but it risks reducing CSI recovery accuracy because it cannot capture environment-specific features.

[0172] 5) Additional control information feedback payloads are required. For example, input format normalization improves the consistency of the input data distribution for a single deep learning model, but additional control bits must be transmitted to restore the original format.

[0173] To address the problems of the aforementioned CSI feedback schemes, this application provides a communication method to improve the performance of DL-based CSI feedback schemes.

[0174] It should be understood that the communication method provided in the embodiments of this application can be applied to multi-antenna communication systems, for example, Figure 1 The communication system shown is illustrated above. The application scenarios described in the embodiments of this application are merely examples and do not constitute any limitation on the scope of protection of this application.

[0175] It should also be understood that the embodiments shown below do not particularly limit the specific structure of the execution subject of the method provided in the embodiments of this application, as long as it is possible to communicate according to the method provided in the embodiments of this application by running a program that records the code of the method provided in the embodiments of this application. For example, the execution subject of the method provided in the embodiments of this application can be a device, or a functional module in the device that can call and execute a program.

[0176] Figure 5 This is a schematic flowchart illustrating a communication method provided in this application. It includes the following steps:

[0177] S510, the first communication device acquires the first downlink feature vector matrix.

[0178] Specifically, the first downlink eigenvector matrix is ​​used to indicate the state information of the downlink channel. For example, the first communication device measures the downlink reference signal to obtain the state information of the downlink channel, and determines the first downlink eigenvector matrix that can characterize the state information of the downlink channel based on the state information of the downlink channel.

[0179] Optionally, the first communication device may be a terminal device.

[0180] As an example and not a limitation, the first communication device may obtain the first downlink feature vector matrix by performing the following steps:

[0181] Step 1.1: The first communication device measures the downlink reference signal to obtain the downlink CSI, such as the UE measuring the downlink CSI-RS or SSB, etc. This downlink CSI is used to indicate the state of the downlink channel, for example, the downlink channel... Represented as:

[0182]

[0183] In equation (2-1) This represents the downlink channel of the s-th subband, where the subband index s takes values ​​of 1, 2, ..., N. s .

[0184] Step 1.2: Determine the normalized eigenvector corresponding to the largest eigenvalue of the s-th sub-band.

[0185] Step 1.3: Determine the first downlink eigenvector matrix as follows

[0186] It should be understood that the process by which the first communication device obtains the first downlink feature vector matrix described above is merely an example and does not constitute any limitation on the scope of protection of this application. The first communication device may also obtain the first downlink feature vector matrix in other ways. For example, the first communication device may determine the first downlink feature vector matrix based on historical communication data; or, for example, the first communication device may receive the first downlink feature vector matrix from another device. For instance, the first communication device may provide the information required to determine the first downlink feature vector matrix to a management device, which then determines the first downlink feature vector matrix and provides it to the first communication device.

[0187] Furthermore, after the first communication device obtains the first downlink feature vector matrix, it can compress and feed back the first downlink feature vector matrix to reduce feedback overhead.

[0188] Furthermore, in the process of compressing and feeding back the first downlink eigenvector matrix, the first communication device in this application, in order to improve the compression feedback performance, in addition to compressing the first downlink eigenvector matrix based on the encoder, can also adjust the first downlink eigenvector matrix based on the first function to optimize the eigenvectors of the first downlink eigenvector matrix, enhance the correlation between uplink and downlink channels, and thus reduce the redundancy of ultra-low rate feedback; and / or,

[0189] The data format of the first downlink feature vector matrix can also be adjusted based on the second function in order to maintain a consistent data distribution at the encoder and decoder ends without incurring additional transmission overhead.

[0190] therefore, Figure 5 The method flow shown also includes:

[0191] S520, the first communication device determines the codeword corresponding to the first downlink feature vector matrix based on the encoder and processing function.

[0192] Specifically, the processing function includes at least one of a first function and a second function, wherein the first function is used to adjust the eigenvector of the matrix input to the first function, and the second function is used to adjust the data format of the matrix input to the second function.

[0193] As an example and not a limitation, the first communication device may determine the codeword corresponding to the first downlink feature vector matrix based on the encoder and processing function in the following possible ways:

[0194] Method 1.1: The processing function includes a first function, and the first communication device determines the codeword corresponding to the first downlink feature vector matrix based on the encoder and the first function.

[0195] Method 1.2: The processing function includes a second function, and the first communication device determines the codeword corresponding to the first downlink feature vector matrix based on the encoder and the second function.

[0196] Method 1.3: The processing function includes a first function and a second function. The first communication device determines the codeword corresponding to the first downlink feature vector matrix based on the encoder, the first function, and the second function.

[0197] The following section uses the method described in Method 1.3 above, where the first communication device determines the codeword corresponding to the first downlink feature vector matrix based on the encoder, the first function, and the second function, as an example to illustrate how the first communication device in this application determines the codeword corresponding to the first downlink feature vector matrix. The methods for determining the codeword corresponding to the first downlink feature vector matrix shown in Methods 1.1 and 1.2 above can refer to the description of determining the codeword corresponding to the first downlink feature vector matrix in Method 1.3, with the difference being:

[0198] In the case shown in method 1.1, it is not necessary to adjust the data format of the downlink eigenvector matrix based on the second function; or,

[0199] In the case shown in Method 1.2, it is not necessary to optimize the eigenvectors of the downlink eigenvector matrix based on the first function.

[0200] For example, the first communication device determines the codeword corresponding to the first downlink feature vector matrix based on the encoder, the first function, and the second function, including the following steps:

[0201] Step 2.1: Obtain the second downlink eigenvector matrix based on the first function and the first downlink eigenvector matrix. The second downlink eigenvector matrix is ​​related to the downlink channel matrix, and the downlink channel matrix is ​​related to the uplink channel matrix. Alternatively, the second downlink eigenvector matrix is ​​determined based on the first downlink eigenvector matrix, and both the first and second downlink eigenvector matrices satisfy the first function. Or, the first downlink eigenvector matrix serves as the input to the first function, and the second downlink eigenvector matrix serves as the output of the first function.

[0202] The correlation between the second downlink eigenvector matrix and the downlink channel matrix can be understood as the minimization of the Euclidean distance between them. Furthermore, since uplink and downlink transmissions experience the same propagation environment, the uplink and downlink channel matrices exhibit a strong correlation before eigenvector decomposition.

[0203] Step 2.2: Obtain the third downlink eigenvector matrix based on the second downlink eigenvector matrix, the downlink reference eigenvector matrix, and the second function. The data format of the third downlink eigenvector matrix is ​​the same as that of the downlink reference eigenvector matrix. Alternatively, the third downlink eigenvector matrix is ​​determined based on the second downlink eigenvector matrix, and both the second and third downlink eigenvector matrices satisfy the second function. Or, the second downlink eigenvector matrix serves as the input to the first function, and the third downlink eigenvector matrix serves as the output of the second function, etc.

[0204] The downlink reference eigenvector matrix can be a downlink matrix. For example, it can be a downlink channel matrix including line-of-sight (LOS) paths; or it can be a downlink channel matrix including the strongest path in non-line-of-sight (NLOS) scenarios, and so on. Without considering the cost of generating the downlink reference eigenvector matrix, it can also be constructed using other more complex methods. For example, the downlink matrix can be the downlink channel matrix of all paths in the NLOS scenario. These will not be illustrated further here.

[0205] Step 2.3: Determine the codeword corresponding to the first downlink feature vector matrix based on the third downlink feature vector matrix and the encoder.

[0206] As an example and not a limitation, steps 2.1 to 2.3 above can be simply described as: determining the codeword based on the encoder and the third downlink feature vector matrix, the third downlink feature vector matrix being determined based on the first downlink feature vector matrix, the first function, and the second function.

[0207] Optionally, the encoder, the first function, the second function, the first downlink feature vector matrix, and the codeword satisfy the following relationship:

[0208] W BCE，DL =f BCE (W DL (2-2)

[0209] W IFA，DL b DL =f IFA (W BCE，DL W Ben，DL (2-3)

[0210] c = f en (W IFA，DL ;Φ) (2-4)

[0211] In the above equation (2-2), f BCE(·) represents the first function, W DL W represents the first downlink eigenvector matrix. BCE，DL Let f represent the second downlink eigenvector matrix. In equation (2-3) above, f IFA (·) represents the second function, W IFA，DL W represents the third downlink eigenvector matrix. Ben，DL Let b represent the downlink reference eigenvector matrix. DL This represents the first control information, which is used by the second communication device to reconstruct the second downlink feature vector matrix. In equation (2-4) above, f... en (·;φ) represents the encoder, φ represents the network parameters of the encoder, and c represents the codeword.

[0212] It should be understood that the first control information is used by the second communication device to recover the second downlink feature vector matrix. Therefore, the first communication device can send the first control information to the second communication device. For example, the first communication device can send message #1 to the second communication device, which includes the first control information. This message #1 can be a CSI measurement report. The first control information can be information such as downlink channel delay or amplitude.

[0213] Furthermore, after the first communication device determines the codeword corresponding to the first downlink feature vector matrix, it can send the codeword to the second communication device, which then reconstructs the relevant channel information based on the codeword. Figure 5 The method flow shown also includes:

[0214] S530, the first communication device sends a codeword to the second communication device, and correspondingly, the second communication device receives the codeword from the first communication device.

[0215] Specifically, the first communication device sends the codeword to the second communication device via an air interface message. It should be understood that this application does not limit the specific process of the first communication device sending the codeword to the second communication device; reference can be made to the description of how a terminal device sends codewords to a network device in current related solutions, which will not be repeated here.

[0216] Furthermore, after receiving the codeword, the second communication device can reconstruct the downlink feature vector matrix based on the codeword and its self-determined second uplink feature vector matrix. During the process of determining the second uplink feature vector matrix, the second communication device can adjust the phase and / or data format of the first uplink feature vector matrix based on a processing function. Figure 5 The method flow shown also includes:

[0217] S540, the second communication device determines the second uplink feature vector matrix based on the first uplink feature vector matrix and the processing function.

[0218] Specifically, the processing function includes at least one of a first function and a second function, wherein the first function is used to adjust the eigenvector of the matrix input to the first function, and the second function is used to adjust the data format of the matrix input to the second function.

[0219] The first uplink eigenvector matrix is ​​used to indicate the state information of the uplink channel.

[0220] Optionally, referring to the description of the first communication device obtaining the first downlink feature vector matrix in step S510 above, the second communication device obtaining the first uplink feature vector matrix in this application may be: the second communication device measures the uplink reference signal to obtain the uplink channel state information, and determines the first uplink feature vector matrix that can characterize the uplink channel state information based on the uplink channel state information.

[0221] As an example and not a limitation, the second communication device can obtain the first uplink feature vector matrix by performing the following steps:

[0222] Step 3.1: The second communication device measures the uplink reference signal to obtain the uplink CSI, such as the BS measuring the SRS or other uplink reference signals to obtain the uplink CSI. This uplink CSI is used to indicate the state of the uplink channel, for example, the uplink channel... Represented as:

[0223]

[0224] Among them, in the above formula (2-5) This represents the downlink channel of the s-th subband, where the subband index s takes values ​​of 1, 2, ..., N. s .

[0225] Step 3.2: Determine the normalized eigenvector corresponding to the largest eigenvalue of the s-th sub-band.

[0226] Step 3.3: Determine the first upward eigenvector matrix as follows

[0227] It should be understood that the process by which the second communication device obtains the first uplink feature vector matrix described above is merely an example and does not constitute any limitation on the scope of protection of this application. The second communication device may also obtain the first uplink feature vector matrix in other ways. For example, the second communication device may determine the first uplink feature vector matrix based on historical communication data; or, for example, the second communication device may receive the first uplink feature vector matrix from another device. For instance, the second communication device may provide the information required to determine the first uplink feature vector matrix to the management device, which then determines the first uplink feature vector matrix and provides it to the second communication device.

[0228] As an example and not a limitation, the second communication device may determine the second uplink feature vector matrix based on the first uplink feature vector matrix and the processing function in the following possible ways:

[0229] Method 2.1: The processing function includes a first function, and the second communication device determines the second uplink feature vector matrix based on the first uplink feature vector matrix and the first function.

[0230] Method 2.2: The processing function includes a second function, and the second communication device determines the second uplink feature vector matrix based on the first uplink feature vector matrix and the second function.

[0231] Method 2.3: The processing function includes a first function and a second function. The second communication device determines the second uplink feature vector matrix based on the first uplink feature vector matrix, the first function, and the second function.

[0232] The following section uses the second communication device shown in method 2.3 above, which determines the second uplink feature vector matrix based on the first uplink feature vector matrix, the first function, and the second function, as an example to illustrate how the second communication device in this application determines the second uplink feature vector matrix. The methods for determining the second uplink feature vector matrix shown in methods 2.1 and 2.2 above can refer to the description of determining the second uplink feature vector matrix in method 2.3, with the difference being:

[0233] In the case shown in method 2.1, it is not necessary to adjust the data format of the first uplink eigenvector matrix based on the second function; or,

[0234] In the case shown in Method 2.2, it is not necessary to optimize the eigenvectors of the first upward eigenvector matrix based on the first function.

[0235] For example, the second communication device determines the second uplink feature vector matrix based on the first uplink feature vector matrix and the processing function, including the following steps:

[0236] Step 4.1: Determine the third uplink eigenvector matrix based on the first uplink eigenvector matrix and the first function. The third uplink eigenvector matrix is ​​related to the uplink channel matrix, and the uplink channel matrix is ​​related to the downlink channel matrix. Alternatively, the third uplink eigenvector matrix is ​​determined based on the first uplink eigenvector matrix, and both the first and third uplink eigenvector matrices satisfy the first function. Or, the first uplink eigenvector matrix serves as the input to the first function, and the third uplink eigenvector matrix serves as the output of the first function, etc.

[0237] The correlation between the third uplink eigenvector matrix and the uplink channel matrix can be understood as the minimization of the Euclidean distance between them. Furthermore, since uplink and downlink transmissions experience the same propagation environment, the uplink and downlink channel matrices exhibit a strong correlation before eigenvector decomposition.

[0238] Step 4.2: Determine the second upward eigenvector matrix based on the third upward eigenvector matrix, the upward reference eigenvector matrix, and the second function. The data format of the second upward eigenvector matrix is ​​the same as that of the upward reference eigenvector matrix. Alternatively, the second upward eigenvector matrix is ​​determined based on the third upward eigenvector matrix, and both the third and second upward eigenvector matrices satisfy the second function. Or, the third upward eigenvector matrix serves as the input to the first function, and the second upward eigenvector matrix serves as the output of the second function.

[0239] The uplink reference feature vector matrix can be an uplink matrix. For example, it can be an uplink channel matrix that includes line-of-sight (LOS) paths; or it can be an uplink channel matrix that includes the strongest path in non-line-of-sight (NLOS) scenarios, and so on. Without considering the cost of generating the uplink reference feature vector matrix, it can also be constructed using other more complex methods. For example, the uplink matrix can be the uplink channel matrix of all paths in the NLOS scenario. These will not be illustrated further here.

[0240] Optionally, the first uplink eigenvector matrix, the first function, the second function, the third uplink eigenvector matrix, and the second uplink eigenvector matrix satisfy the following relationship:

[0241] W BCE，UL =f BCE (W UL (2-6)

[0242] W IFA，UL b UL =fIFA (W BCE，UL W Ben，UL (2-7)

[0243] In the above equation (2-6), f BCE (·) represents the first function, W UL Let f represent the first upward eigenvector matrix, and let f represent the third upward eigenvector matrix. In equation (2-7) above, f... IFA (·) denotes the second function, W Ben，UL Let b represent the uplink reference eigenvector matrix. UL Indicates the second control information, W IFA，UL This represents the second upward eigenvector matrix.

[0244] As one possible implementation, this second control information is related to b shown in equation (2-3) above. DL The first control information represented is the same, namely b DL and b UL This indicates the same control information. For example, the second control information could also be information such as downlink channel delay or amplitude.

[0245] In this implementation, the second control information can be the first control information that the first communication device sends to the second communication device through additional transmission overhead.

[0246] As another possible implementation, this second control information is related to b shown in equation (2-3) above. DL The first control information they represent is different. For example, the second control information may be information such as the uplink channel delay or amplitude.

[0247] In this implementation, the correlation between the uplink and downlink channels can be utilized to replace the first control information with the second control information. The first communication device does not need to send the first control information to the second communication device through additional transmission overhead, which can reduce signaling overhead.

[0248] Furthermore, after the second communication device receives the codeword from the first communication device and determines the aforementioned second uplink feature vector matrix, it can reconstruct the downlink feature vector matrix based on the decoder, the received codeword, and the second uplink feature vector matrix. Figure 5 The method flow shown may also include:

[0249] S550, the second communication device obtains the reconstructed third downlink feature vector matrix based on the decoder, codeword, and second uplink feature vector matrix.

[0250] For example, the decoder, the codeword, the second uplink feature vector matrix, and the reconstructed third downlink feature vector matrix satisfy the following relationship:

[0251]

[0252] In the above equation (2-8), f de (·;Ψ) represents the decoder, Ψ represents the network parameters of the decoder, |W IFA，UL | represents the modulus of the second uplink eigenvector matrix, and c represents the codeword. This represents the third row eigenvector matrix after reconstruction.

[0253] Optionally, the data format of the reconstructed third downlink eigenvector matrix can be restored based on the second function to obtain the reconstructed second downlink eigenvector matrix. Figure 5 The method flow shown may also include:

[0254] S560, the second communication device obtains the reconstructed second downlink feature vector matrix based on the inverse function of the second function, the reconstructed third downlink feature vector matrix, and the second control information.

[0255] For example, the inverse function of the second function, the reconstructed third downlink eigenvector matrix, the second control information, and the reconstructed second downlink eigenvector matrix satisfy the following relationship:

[0256]

[0257] Among them, in the above formula (2-9) This represents the inverse function of the second function, used for data format restoration. The b... UL This indicates the second control information, the This represents the third downlink eigenvector matrix after reconstruction. This represents the second downlink feature vector matrix after reconstruction.

[0258] It should be noted that the downlink eigenvector matrix of the second communication device reconfiguration and recovery in this application can be... Instead of W DL The reason is and W DL The corresponding eigenvalues ​​are the same.

[0259] Figure 5 In the communication method shown, after the first communication device obtains the first downlink feature vector matrix, it can process the first downlink feature vector matrix based on the encoder and a first function and / or a second function to obtain the codewords fed back to the second communication device. The encoder can compress the data to be fed back, reducing feedback overhead.

[0260] Furthermore, in this technical solution, during the processing of the first downlink feature vector matrix by the first communication device, a first function and / or a second function can be utilized. The first function can adjust the phase of the eigenvectors of the matrix input to it, thereby optimizing the first downlink feature vector matrix based on the first function. This makes the eigenvectors of the first downlink feature vector matrix as close as possible to the reference eigenvectors, improving the correlation between uplink and downlink and further reducing feedback overhead. Additionally, the second function can adjust the data format of the matrix input to it, thus generalizing the application scenarios of the encoder.

[0261] Therefore, if the first function and / or the second function are used to process the first downlink feature vector matrix during the process of determining the codeword to be fed back, the performance of the DL-based CSI feedback scheme can be improved.

[0262] Additionally, it should be noted that the aforementioned processing of the feature vector matrix based on the first function and / or the second function is performed before input to the encoder or decoder, which is a preprocessing step. Efficient preprocessing can significantly reduce model complexity.

[0263] To facilitate understanding, the following will be combined with... Figure 6 This application provides a detailed description of the details. Figure 5 The relevant steps performed by the first and second communication devices in the communication method shown.

[0264] For example, Figure 6 This illustrates CSI feedback based on eigenvectors in a large-scale MIMO communication system. From... Figure 6 As can be seen, the deep learning-based CSI feedback process is divided into three main stages: preprocessing, neural network processing, and postprocessing.

[0265] The UE preprocessing stage includes:

[0266] Bidirectional correlation enhancement: eigenvector matrix W DL After optimization, it has stronger correlation and unique feature vectors, thus yielding W. BCE，DL .

[0267] Input format alignment: Align the enhancement matrix W BCE，DL With reference format W Ben，DL Alignment, generating the final aligned feature vector matrix W IFA，DL .

[0268] The BS preprocessing stage includes:

[0269] Bidirectional correlation enhancement: eigenvector matrix W UL After optimization, it exhibits stronger relevance and uniqueness, thus yielding W. BCE，UL .

[0270] Input format alignment: Align the enhancement matrix W BCE，UL With reference format W Ben，UL Alignment, generating the final aligned feature vector matrix W IFA，UL and control information b UL .

[0271] The UE neural network stage includes:

[0272] Encoder network: Normalized matrix W IFA，DL It is compressed into codeword c by the encoder network.

[0273] The BS neural network stage includes:

[0274] Decoder network: The decoder network then reconstructs the matrix from the codeword c and the uplink magnitude matrix. To restore the aligned downlink eigenvector matrix.

[0275] The BS post-processing stage includes:

[0276] Original format recovery: based on uplink control information b UL The downlink matrix will be restored. Converted to the original format of the eigenvector matrix

[0277] from Figure 6 As can be seen from the above, the communication method provided in this application is a DL-based CSI feedback scheme that utilizes uplink auxiliary neural networks and input format alignment methods to reduce redundancy in compressed codewords and enhance the generalization performance of CSI feedback in large-scale MIMO systems. This DL-based CSI feedback scheme goes through preprocessing stages (including input format alignment and correlation enhancement in the first and second communication devices), neural network processing (with encoder and decoder networks), and post-processing (restoring the original matrix format). Exemplarily, the DL-based CSI feedback scheme in this application incorporates a bidirectional correlation enhancement method in the preprocessing stage to overcome the challenges of correlation degradation and model mismatch caused by non-unique feature vectors. By implementing this bidirectional correlation enhancement step, the initial feature vector matrices in the downlink and uplink are optimized to a unique and more correlated format before format alignment. Therefore, this enhancement method produces better and more stable performance in dynamic wireless communication environments.

[0278] In addition, for ease of understanding, combined with Figure 7 and Figure 8 illustrate Figure 6 The CSI feedback scheme shown performs better than other CSI feedback schemes.

[0279] like Figure 7As shown, for the same CSI recovery performance, the performance of the communication method in this application is mainly reflected in its low feedback bit overhead. Figure 7 As shown, under the same CSI recovery performance (e.g., SGCS equals 0.85), the CSI feedback scheme provided in this application has less than 10 feedback bits, the CSI feedback scheme based on UniversalNet has nearly 20 feedback bits, and the uplink-assisted CSI feedback method (Ubi-ImCsiNet) has nearly 30 feedback bits, etc. In other words, the communication method provided in this application, when applied to CSI feedback, utilizes enhanced bidirectional correlation and can reduce the feedback payload overhead by about 70% compared to UniversalNet, and by about 80% compared to the uplink-assisted CSI feedback method Ubi-ImCsiNet.

[0280] like Figure 8 As shown, different CSI feedback schemes exhibit varying CSI recovery performance when the feedback bits are 24 bits. The performance of the communication method in this application is primarily reflected in its superior CSI recovery performance. Figure 8 As shown, in different environments, for a feedback bit length of 24 bits, the CSI feedback scheme provided in this application has the highest SGCS value, indicating that the CSI recovery performance is the best.

[0281] As one possible implementation, the first communication device in this application needs to use the first function during the process of determining the codeword, and the second communication device also needs to use the first function during the process of reconstructing the downlink feature vector matrix. For ease of understanding, the determination and use of the first function in this application will be described in detail below with reference to Scheme 1.

[0282] Option 1: Determining and using the first function.

[0283] In the aforementioned CSI feedback scheme based on eigenvectors, for a given channel matrix of the s-th sub-band... Assuming that H s normalized eigenvector w s Corresponding to the eigenvalue λ s There are multiple normalized feature vectors. satisfy This non-uniqueness reduces the correlation between uplink and downlink feature vectors, limiting the efficiency of CSI compression.

[0284] This application designs a first function (for ease of description, the first function will be denoted as f below) to achieve this. BCE (·)) so that the upward eigenvector matrix W BCE，UL =f BCE (WUL The upward eigenvectors and the downward eigenvector matrix W) BCE，DL =f BCE (W DL The downlink feature vector is unique in the feature space, which can maximize the correlation between uplink and downlink feature vectors and improve the compression and recovery efficiency of CSI.

[0285] It should be noted that the function f is designed to meet the above requirements. BCE The key challenge is the lack of CSI in the opposite direction without increasing transmission overhead; that is, the first communication device lacks an uplink matrix for measuring and enhancing correlation, and the second communication device lacks a downlink matrix for measuring and enhancing correlation. To address this issue, this application proposes to leverage the inherent strong correlation between the uplink and downlink channels to enhance the bidirectional correlation of the eigenvector matrix.

[0286] Since uplink and downlink experience the same propagation environment, the uplink channel matrix... and downlink channel matrix Before eigenvector decomposition, they exhibit strong correlation. Therefore, more closely aligning the downlink eigenvector with the downlink channel vector in its subband can improve its correlation with the uplink eigenvector.

[0287] Therefore, this application uses the first communication device to measure Enhance W DL The second communication device can also perform a similar operation to correlate with the uplink feature vector matrix, for example, using the data measured by the first communication device. Enhance W UL Correlation with the downlink feature vector matrix.

[0288] For example, for a first communication device, obtaining the second downlink feature vector matrix based on the first function and the first downlink feature vector matrix includes the following steps:

[0289] Step 5.1: Based on the first function, align the eigenvector corresponding to the s-th sub-band in the first downlink eigenvector matrix with the reference vector of the downlink channel matrix corresponding to the s-th sub-band to obtain the eigenvector corresponding to the s-th sub-band in the second downlink eigenvector matrix. Alternatively, determine the eigenvector corresponding to the s-th sub-band in the second downlink eigenvector matrix based on the eigenvector corresponding to the s-th sub-band in the first downlink eigenvector matrix, where the eigenvector corresponding to the s-th sub-band in the first downlink eigenvector matrix and the eigenvector corresponding to the s-th sub-band in the second downlink eigenvector matrix satisfy the first function. Or, the eigenvector corresponding to the s-th sub-band in the first downlink eigenvector matrix is ​​used as the input to the first function, and the eigenvector corresponding to the s-th sub-band in the second downlink eigenvector matrix is ​​used as the output of the first function.

[0290] Step 5.2: Assign N to the communication system to which the first communication device belongs. s N corresponding to each sub-band s The normalized eigenvectors are concatenated to obtain the second downlink eigenvector matrix. Alternatively, based on N... s The eigenvectors corresponding to each sub-band are used to determine the second downlink eigenvector matrix. Where N... s s is a positive integer, and the value of s is less than or equal to N. s Positive integers.

[0291] For example, optimizing the eigenvector matrix The goal is to approximate the reference vectors derived from the uplink and downlink channel matrices as closely as possible while preserving their respective feature spaces. Specifically, the bidirectional correlation enhancement problem can be represented on the first communication device side as:

[0292]

[0293] In the above formula (3-1) Let ||·|| represent the eigenvector corresponding to the s-th sub-band in the second downlink eigenvector matrix, ||·|| represent the l2 norm, and ε s，DL Represents the eigenvalue λ s，DL The corresponding feature space, h s，DL w represents the reference vector of the downlink channel matrix corresponding to the s-th subband. s，DL This represents the feature vector corresponding to the s-th sub-band in the first downlink feature vector matrix.

[0294] To illustrate how to obtain the downlink eigenvector matrix with enhanced correlation, since The dimension is greater than choose This represents the channel between the first receiving antenna and all transmitting antennas in the s-th subband. The channel vector from any other receiving antenna can also be represented as h. s，DL.

[0295] Because h s，DL Projected onto feature space ε s，DL Approximate to vector h s，DL The adjusted feature vector can be represented as:

[0296]

[0297] Among them, in the above formula (3-2) This indicates a projection operation. Used to h s，DL Projected onto feature space ε s，DL .

[0298] Optionally, when the eigenvalue λ s，DL When the algebraic multiplicity k > 1, the above equation (3-2) can be expanded to:

[0299]

[0300] Among them, the matrix in equation (3-3) above The columns constitute ε s，DL An orthogonal basis. The algebraic multiplicity of an eigenvalue is the number of times the eigenvalue appears in the characteristic polynomial of the matrix.

[0301] Through normalization This ensures the unity norm. (This applies to all N...) s Normalized eigenvectors on each subband The optimized downlink eigenvector matrix W is obtained by cascading. BCE，DL Similarly, the related enhanced uplink eigenvector matrix W can be obtained. BCE，UL .

[0302] For example, for a second communication device, determining a third uplink feature vector matrix based on the first uplink feature vector matrix and the first function includes the following steps:

[0303] Step 6.1: Based on the first function, align the eigenvector corresponding to the s-th sub-band in the first uplink eigenvector matrix with the reference vector of the uplink channel matrix corresponding to the s-th sub-band to obtain the eigenvector corresponding to the s-th sub-band in the third uplink eigenvector matrix. Alternatively, determine the eigenvector corresponding to the s-th sub-band in the third uplink eigenvector matrix based on the eigenvector corresponding to the s-th sub-band in the first uplink eigenvector matrix, where the eigenvector corresponding to the s-th sub-band in the first uplink eigenvector matrix and the eigenvector corresponding to the s-th sub-band in the third uplink eigenvector matrix satisfy the first function. Or, the eigenvector corresponding to the s-th sub-band in the first uplink eigenvector matrix is ​​used as the input of the first function, and the eigenvector corresponding to the s-th sub-band in the third uplink eigenvector matrix is ​​used as the output of the first function.

[0304] Step 6.2: The N assigned to the communication system to which the second communication device belongs s N corresponding to each sub-band s The normalized eigenvectors are concatenated to obtain the third upward eigenvector matrix. Alternatively, based on N... s The eigenvectors corresponding to each sub-band are used to determine the third upward eigenvector matrix. Where N... s s is a positive integer, and the value of s is less than or equal to N. s Positive integers.

[0305] For example, optimizing the eigenvector matrix The goal is to approximate the reference vectors derived from the uplink and downlink channel matrices as closely as possible while preserving their respective feature spaces. Specifically, the bidirectional correlation enhancement problem can be represented on the first communication device side as:

[0306]

[0307] Among them, in the above formula (3-4) This represents the eigenvector corresponding to the s-th sub-band in the third upward eigenvector matrix, ||·|| represents the l2 norm, and ε s，UL Represents the eigenvalue λ s，UL The corresponding feature space, h s，UL w represents the reference vector of the uplink channel matrix corresponding to the s-th sub-band. s，UL This represents the eigenvector corresponding to the s-th sub-band in the first uplink eigenvector matrix.

[0308] To illustrate how to obtain the uplink eigenvector matrix with enhanced correlation, since The dimension is greater than choose This represents the channel between the first transmit antenna and all receive antennas in the s-th subband. The channel vector from any other transmit antenna can also be used as h. s，UL.

[0309] Because h s，UL Projected onto feature space ε s，UL Approximate to vector h s，UL The adjusted feature vector can be represented as:

[0310]

[0311] Among them, in the above formula (3-5) This indicates a projection operation. Used to h s，UL Projected onto feature space ε s，UL .

[0312] Optionally, when the eigenvalue λ s，UL When the algebraic multiplicity k > 1, the above equation (3-5) can be expanded to:

[0313]

[0314] Among them, the matrix in equation (3-6) above The columns constitute ε s，UL An orthogonal basis. The algebraic multiplicity of an eigenvalue is the number of times the eigenvalue appears in the characteristic polynomial of the matrix.

[0315] Through normalization This ensures the unity norm. (This applies to all N...) s Normalized eigenvectors on each subband By cascading, the optimized uplink eigenvector matrix W is obtained. BCE，UL Similarly, the related enhanced uplink eigenvector matrix W can be obtained. BCE，UL .

[0316] As one possible implementation, the first communication device in this application needs to use the second function during the process of determining the codeword, and the second communication device also needs to use the second function during the process of reconstructing the downlink feature vector matrix. For ease of understanding, the determination and use of the second function in this application will be described in detail below with reference to Scheme 2.

[0317] Option 2: Determining and using the second function.

[0318] Due to input distribution transformations and model mismatches, deep learning-based CSI feedback can lead to performance degradation in unpredictable environments. To address this issue, this application proposes an input format alignment function that uses a pair of predefined benchmarks to ensure consistency and compatibility with the pre-trained CSI feedback model.

[0319] By integrating bidirectional correlation into the original format recovery, the method proposed in this application can recover the original format before alignment on the second communication device side without additional transmission overhead.

[0320] For example, for the first communication device, obtaining the third downlink feature vector matrix based on the second downlink feature vector matrix and the second function includes:

[0321] Step 7.1: Convert the second downlink eigenvector matrix into a downlink sparse eigenvector matrix using a two-dimensional discrete Fourier transform. The downlink sparse eigenvector matrix can be understood as a finite number of non-zero downlink channel coefficients existing in the time delay angle domain, including the amplitude and phase of the downlink channel and the corresponding time delay angle coordinates. Alternatively, the downlink sparse eigenvector matrix is ​​determined based on the second downlink eigenvector matrix, obtained by matrix transformation of the second downlink eigenvector matrix, such as by using a two-dimensional discrete Fourier transform.

[0322] Additionally, it should be noted that the method of determining the downlink sparse eigenvector matrix based on the transformation of the second downlink eigenvector matrix in the above-mentioned application, which is a two-dimensional discrete Fourier transform, is only an example and does not constitute any limitation on the scope of protection of this application. Other methods can also be used to determine the downlink sparse eigenvector matrix based on the second downlink eigenvector matrix, such as fast Fourier transform, etc., which will not be illustrated here.

[0323] Step 7.2: Based on the downlink sparse eigenvector matrix and the downlink reference eigenvector matrix, obtain the first row cyclic shift step size and the first column cyclic shift step size. Step 7.3: Based on the downlink sparse eigenvector matrix, the first row cyclic shift step size, the first column cyclic shift step size, and the second function, obtain the third downlink eigenvector matrix.

[0324] For example, the second downlink eigenvector matrix is ​​first transformed into a downlink sparse eigenvector matrix using a two-dimensional discrete Fourier transform (2D-DFT), thereby achieving input format alignment. For instance, the eigenvector matrix W... BCE，DL Convert to W Spar，DL :

[0325]

[0326] Among them, the above formula (4-1)W Spar，DL W represents the downlink sparse eigenvector matrix. BCE，UL F represents the third upward eigenvector matrix. d and F a The Discrete Fourier Transform (DFT) matrix is... This represents the two-dimensional discrete Fourier transform.

[0327] Secondly, a coupling reference matrix is ​​constructed and deployed on the UE and BS for alignment. For example, a pair of downlink and uplink matrices is selected as the reference matrix, which corresponds to the physical channel with only line-of-sight (LoS) path.

[0328] In order to W Spar，DL Align with reference W Ben，DL You can use f IFA (·), for example, using the function f cs (·, m, n) implements cyclic shift, where m and n represent the shift step size of the row and column, respectively.

[0329] The shifted eigenvector matrix W cs，DL =f cs (W Spar，DL The elements in (m, n) are represented as:

[0330]

[0331] Determine the optimal shift step size W SparDL With W BenDL Alignment.

[0332] To facilitate understanding, the solution process will be explained below. and The process:

[0333] r DL [i] is defined as W Spar，DL The sum of the magnitudes of the row vectors Similarly, r Ben，DL [i] is defined as the sum of the magnitudes of the base row vectors.

[0334] Determine the circular shift Maximize r DL and r Ben，DL Intercorrelation in Indicates r DL Circular shift m steps. Optimal shift. It can be represented as:

[0335]

[0336] r DL [j] is defined as W Spar，DL The sum of the magnitudes of the row vectors Similarly, r Ben，DL [j] is defined as the sum of the magnitudes of the baseline row vectors.

[0337] Determine the circular shift Maximize r DL and r Ben，DL Intercorrelation in Indicates r DL Circular shift n steps. Optimal shift. It can be represented as:

[0338]

[0339] The aligned downlink eigenvector matrix is ​​represented as follows:

[0340]

[0341] Among them, control information b DL Include

[0342] For example, for the second communication device, determining the second uplink feature vector matrix based on the third uplink feature vector matrix, the uplink reference feature vector matrix, and the second function includes:

[0343] Step 8.1: Convert the third uplink eigenvector matrix into an uplink sparse eigenvector matrix using a two-dimensional discrete Fourier transform. The uplink sparse eigenvector matrix can be understood as a finite number of non-zero uplink channel coefficients existing in the time delay angle domain, including the amplitude and phase of the uplink channel and the corresponding time delay angle coordinates. Alternatively, the uplink sparse eigenvector matrix is ​​determined based on the third uplink eigenvector matrix, obtained by matrix transformation of the third uplink eigenvector matrix, such as by performing a two-dimensional discrete Fourier transform on the third uplink eigenvector matrix.

[0344] Additionally, it should be noted that the method of determining the upward sparse eigenvector matrix based on the transformation of the third upward eigenvector matrix in the above-mentioned application, which is a two-dimensional discrete Fourier transform, is only an example and does not constitute any limitation on the scope of protection of this application. Other methods can also be used to determine the upward sparse eigenvector matrix based on the third upward eigenvector matrix, such as inverse discrete Fourier transform, fast Fourier transform, etc., which will not be illustrated here.

[0345] Step 8.2: Based on the uplink sparse feature vector matrix and the uplink reference feature vector matrix, obtain the second row cyclic shift step size and the second column cyclic shift step size. Step 8.3: Based on the uplink reference feature vector matrix, the second row cyclic shift step size, the second column cyclic shift step size, and the second function, obtain the second uplink feature vector matrix.

[0346] For example, the third uplink eigenvector matrix is ​​first transformed into an uplink sparse eigenvector matrix using a two-dimensional discrete Fourier transform (2D-DFT), thereby achieving input format alignment. For instance, the eigenvector matrix W... BCE，UL Convert to W Spar，UL :

[0347]

[0348] In equation (4-6) above, W Spar，UL W represents the uplink sparse eigenvector matrix. BCE，UL F represents the third upward eigenvector matrix. d and F a The Discrete Fourier Transform (DFT) matrix is... This represents the two-dimensional discrete Fourier transform.

[0349] Secondly, a coupling reference matrix is ​​constructed and deployed on the UE and BS for alignment. For example, a pair of downlink and uplink matrices is selected as the reference matrix, which corresponds to the physical channel with only line-of-sight (LoS) path.

[0350] In order to W Spar，UL Align with reference W Ben，UL You can use f IFA (·), for example, using the function f cs (·, m, n) implements cyclic shift, where m and n represent the shift step size of the row and column, respectively.

[0351] The shifted eigenvector matrix W cs，UL =f cs (W Spar，UL The elements in (m, n) are represented as:

[0352]

[0353] Determine the optimal shift step size W Spar，UL With W Ben，UL Alignment.

[0354] To facilitate understanding, the solution process will be explained below. and The process:

[0355] r UL [i] is defined as W SparUL The sum of the magnitudes of the row vectors Similarly, r Ben，UL [i] is defined as the sum of the magnitudes of the base row vectors.

[0356] Determine the circular shift Maximize r UL and r Ben，UL Intercorrelation in Indicates r UL Circular shift m steps. Optimal shift. It can be represented as:

[0357]

[0358] r UL [j] is defined as W Spar，UL The sum of the magnitudes of the row vectors Similarly, r Ben，UL [j] is defined as the sum of the magnitudes of the baseline row vectors.

[0359] Determine the circular shift Maximize r UL and r Ben，UL Intercorrelation in Indicates r UL Circular shift n steps. Optimal shift. It can be represented as:

[0360]

[0361] The aligned downlink eigenvector matrix is ​​represented as follows:

[0362]

[0363] Among them, control information b UL Include

[0364] As another possible implementation, the first communication device in this application needs to use an encoder to determine the codeword, and the second communication device needs to use a decoder to reconstruct the downlink feature vector matrix. For ease of understanding, the encoder and decoder in this application will be described in detail below with reference to Scheme 3.

[0365] Option 3: Encoder and decoder configuration.

[0366] In this application, the first communication device needs to rely on an encoder to determine the codeword, and the second communication device needs to rely on a decoder to reconstruct the downlink feature vector matrix.

[0367] like Figure 9As shown, the encoder on the terminal device side can separate the complex feature vector matrix into real and imaginary parts, and then cascade them along sub-bands to form a feature map. Two transformation-based single-transformer network encoder layers are used for feature extraction, and a fully connected layer with M units is used for dimensionality reduction. Subsequently, a quantizer module is used to quantize the dimensionality-compressed floating-point vector.

[0368] The decoder (or decoder) on the network device side reconstructs the downlink feature vector matrix using compressed codewords and the uplink amplitude matrix. A fully connected layer restores the codewords to their original length. A conjugate layer merges the reconstructed downlink features with the uplink amplitude for decoding. To improve reconstruction accuracy by leveraging uplink-downlink correlation, five residual blocks are used, each containing two 3×3 convolutional layers with 32 and 2 channels respectively. A 3×3 convolutional layer reduces the feature map from 3 to 2, followed by a scaling layer that concatenates the feature maps. Two TransformerSingle TransNet-based decoder layers are used for fine-grained recovery. A reshaping layer separates the real and imaginary parts of the recovered downlink feature vector matrix. Unlike networks that use uplink amplitude to optimize the CSI feedback network output, which relies solely on compressed codewords for recovery (potentially sacrificing performance if the initial recovery is inaccurate), this application embeds the dimension-recovered uplink amplitude into the network, fully utilizing bidirectional channel correlation to effectively compensate for lost downlink amplitude information and achieve ultra-low-rate feedback.

[0369] It should be understood that Figure 9 The encoder and decoder structures shown are merely examples and do not constitute any limitation on the scope of protection of this application. The encoder used by the first communication device in determining the codeword in this application may have other possible forms, which will not be illustrated here. Similarly, the decoder used by the second communication device in reconstructing the downlink feature vector matrix may also have other possible forms, which will not be illustrated here.

[0370] Furthermore, the above communication method can be adaptively adjusted for use in multi-user Massive MIMO scenarios. For example, the CSI matrix H can be used instead of the eigenvector-based matrix W, and the first function for enhancing correlation can be replaced with a third function for sparse transformation. The specific process can be found above. Figure 5 The communication method shown will not be described in detail here.

[0371] For ease of understanding, combined with Figure 10 This section provides a brief overview of the CSI feedback process in a multi-user Massive MIMO scenario.

[0372] For example, Figure 10 CSI feedback in a multi-user Massive MIMO communication system is illustrated. From Figure 10 As can be seen, the deep learning-based CSI feedback process is divided into three main stages: preprocessing, neural network processing, and postprocessing.

[0373] The UE preprocessing stage includes:

[0374] Sparse transformation: CSI matrix H DL After sparse transformation, the CSI matrix H after sparse transformation is obtained. ST，DL .

[0375] Input format alignment: Align the CSI matrix H after sparse transformation ST，DL With reference format H Ben，DL Alignment, generating the final aligned eigenvector matrix H IFA，DL .

[0376] The BS preprocessing stage includes:

[0377] Sparse transformation: CSI matrix H UL After sparse transformation, the CSI matrix H after sparse transformation is obtained. ST，UL .

[0378] Input format alignment: Align the CSI matrix H after sparse transformation ST，UL With reference format H Ben，UL Alignment, generating the final aligned eigenvector matrix H IFA，UL and control information b UL .

[0379] The UE neural network stage includes:

[0380] Encoder Network: Normalization Matrix H IFA，DL It is compressed into codeword c by the encoder network.

[0381] The BS neural network stage includes:

[0382] Decoder network: The decoder network then reconstructs the matrix from the codeword c and the uplink magnitude matrix. To restore the aligned downlink eigenvector matrix.

[0383] The BS post-processing stage includes:

[0384] Original format recovery: based on uplink control information b UL The downlink matrix will be restored. Converted to the original format of the eigenvector matrix

[0385] It should be understood that the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0386] It should also be understood that, in the various embodiments of this application, unless otherwise specified or logically conflicting, the terminology and / or descriptions between different embodiments are consistent and can be referenced mutually. Technical features in different embodiments can be combined to form new embodiments based on their inherent logical relationships. For example, the above... Figure 5 The illustrated embodiments can be combined to achieve downlink positioning and uplink positioning.

[0387] It should also be understood that in some of the above embodiments, the examples are mainly based on devices in existing network architectures (such as the first communication device, the second communication device, etc.). It should be understood that the specific form of the device is not limited in the embodiments of this application. For example, any device that can achieve the same function in the future is applicable to the embodiments of this application.

[0388] It is understood that in the above-described method embodiments, the methods and operations implemented by the device (such as the first communication device, the second communication device, etc.) can also be implemented by the device's components (such as chips or circuits).

[0389] The above, combined with Figure 5 The communication method provided in the embodiments of this application is described in detail. The above-described communication method is mainly introduced from the perspective of the interaction between, for example, a first communication device and a second communication device. It is understood that, in order to achieve the above functions, the first communication device and the second communication device include hardware structures and / or software modules corresponding to the execution of each function.

[0390] Those skilled in the art will recognize that, based on the units and algorithm steps described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is implemented in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0391] The following combination Figures 11 to 13 The communication device provided in this application is described in detail. It should be understood that the description of the device embodiments corresponds to the description of the method embodiments. Therefore, for details not described in detail, please refer to the method embodiments above; for brevity, some details are omitted.

[0392] This application embodiment can divide the transmitting or receiving device into functional modules according to the above method examples. For example, each function can be divided into its own functional modules, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods. The following description uses the division of functional modules according to each function as an example.

[0393] Figure 11 This is a schematic block diagram of a communication device 10 provided in an embodiment of this application. The device 10 includes a transceiver module 11 and a processing module 12. The transceiver module 11 can implement corresponding communication functions, and the processing module 12 is used for data processing. In other words, the transceiver module 11 is used to perform operations related to receiving and sending, while the processing module 12 is used to perform other operations besides receiving and sending. The transceiver module 11 can also be referred to as a communication interface or a communication unit.

[0394] In one possible implementation, the device 10 may further include a storage module 13, which can be used to store instructions and / or data. The processing module 12 can read the instructions and / or data in the storage module to enable the device to perform the actions of the device in the aforementioned method embodiments.

[0395] In one design, the device 10 may correspond to the first communication device in the above method embodiments, or to a component of the first communication device (such as a chip).

[0396] The device 10 can implement the steps or processes corresponding to those performed by the first communication device in the above method embodiments. The transceiver module 11 can be used to perform the transceiver-related operations of the first communication device in the above method embodiments, and the processing module 12 can be used to perform the processing-related operations of the first communication device in the above method embodiments.

[0397] In one possible implementation, processing module 12 is used to acquire a first downlink feature vector matrix, which indicates the state information of the downlink channel. Processing module 12 is also used to determine the codeword corresponding to the first downlink feature vector matrix based on the encoder and processing function. Transceiver module 11 is used to send the codeword to the second communication device. The description of the processing function can be found in the above method embodiments and will not be repeated here.

[0398] When the device 10 is used to perform Figure 5When the method is in use, the transceiver module 11 can be used to execute the steps of sending and receiving information in the method, such as step S530; the processing module 12 can be used to execute the processing steps in the method, such as steps S510 and S520.

[0399] It should be understood that the specific process of each unit performing the above-mentioned corresponding steps has been described in detail in the above method embodiments, and will not be repeated here for the sake of brevity.

[0400] In another design, the device 10 may correspond to the second communication device in the above method embodiment, or to a component of the second communication device (such as a chip).

[0401] The device 10 can implement the steps or processes corresponding to those performed by the second communication device in the above method embodiments. The transceiver module 11 can be used to perform transceiver-related operations of the second communication device in the above method embodiments, and the processing module 12 can be used to perform processing-related operations of the second communication device in the above method embodiments.

[0402] In one possible implementation, transceiver module 11 is configured to receive codewords from a first communication device, the codewords being codewords corresponding to a first downlink feature vector matrix indicating downlink channel state information. Processing module 12 is configured to determine a second uplink feature vector matrix based on the first uplink feature vector matrix and a processing function, the first uplink feature vector matrix being used to indicate uplink channel state information. Processing module 12 is further configured to obtain a reconstructed third downlink feature vector matrix based on the decoder, the codewords, and the second uplink feature vector matrix, wherein the processing function includes at least one of a first function and a second function, the first function being used to adjust the feature vectors of the matrix input to the first function, the second function being used to adjust the data format of the matrix input to the second function, and the second downlink feature vector matrix being determined by the first function and the first downlink feature vector matrix.

[0403] When the device 10 is used to perform Figure 5 When the method is in use, the transceiver module 11 can be used to execute the steps of sending and receiving information in the method, such as step S530; the processing module 12 can be used to execute the processing steps in the method, such as steps S540, S550 and S560.

[0404] It should be understood that the specific process of each unit performing the above-mentioned corresponding steps has been described in detail in the above method embodiments, and will not be repeated here for the sake of brevity.

[0405] It should also be understood that the device 10 here is embodied in the form of a functional module. The term "module" here can refer to an application-specific integrated circuit (ASIC), electronic circuitry, a processor (e.g., a shared processor, a proprietary processor, or a group processor, etc.) and memory for executing one or more software or firmware programs, integrated logic circuitry, and / or other suitable components supporting the described functions. In an alternative example, those skilled in the art will understand that device 10 may specifically be a first communication device in the above embodiments, used to execute the various processes and / or steps corresponding to the first communication device in the above method embodiments; or, device 10 may specifically be a second communication device in the above embodiments, used to execute the various processes and / or steps corresponding to the second communication device in the above method embodiments. To avoid repetition, further details are omitted here.

[0406] The apparatus 10 of each of the above-described schemes has the function of implementing the corresponding steps performed by the devices (such as the first communication device, the second communication device, etc.) in the above-described methods. This function can be implemented by hardware or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above-described functions; for example, the transceiver module can be replaced by a transceiver (for example, the transmitting unit in the transceiver module can be replaced by a transmitter, and the receiving unit in the transceiver module can be replaced by a receiver), and other units, such as processing modules, can be replaced by processors, which respectively execute the transceiver operations and related processing operations in each method embodiment.

[0407] In addition, the transceiver module 11 can also be a transceiver circuit (for example, it may include a receiving circuit and a transmitting circuit), and the processing module can be a processing circuit.

[0408] Figure 12 This is a schematic diagram of another communication device 20 provided in an embodiment of this application. The device 20 includes a processor 21, which is used to execute computer programs or instructions stored in a memory 22, or to read data / signaling stored in the memory 22, to perform the methods in the above-described method embodiments. In one possible implementation, the processor 21 may be one or more.

[0409] One possible implementation is, such as Figure 12 As shown, the device 20 also includes a memory 22 for storing computer programs or instructions and / or data. The memory 22 may be integrated with the processor 21 or it may be disposed separately. In one possible implementation, there may be one or more memories 22.

[0410] One possible implementation is, such as Figure 12As shown, the device 20 also includes a transceiver 23 for receiving and / or transmitting signals. For example, the processor 21 controls the transceiver 23 to receive and / or transmit signals.

[0411] As one approach, the device 20 is used to implement the operations performed by the first communication device and the second communication device in the various method embodiments described above.

[0412] It should be understood that the processor mentioned in the embodiments of this application can be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0413] It should also be understood that the memory mentioned in the embodiments of this application can be volatile memory and / or non-volatile memory. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM). For example, RAM can be used as an external cache. By way of example and not limitation, RAM includes the following forms: static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM).

[0414] It should be noted that when the processor is a general-purpose processor, DSP, ASIC, FPGA, or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component, the memory (storage module) can be integrated into the processor.

[0415] It should also be noted that the memory described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0416] Figure 13 This is a schematic diagram of a chip system 30 provided in an embodiment of this application. The chip system 30 (or processing system) includes logic circuitry 31 and an input / output interface 32.

[0417] The logic circuit 31 can be a processing circuit in the chip system 30. The logic circuit 31 can be coupled to a memory unit, calling instructions from the memory unit, enabling the chip system 30 to implement the methods and functions of the embodiments of this application. The input / output interface 32 can be an input / output circuit in the chip system 30, outputting processed information from the chip system 30, or inputting data or signaling information to be processed into the chip system 30 for processing.

[0418] As one option, the chip system 30 is used to implement the operations performed by the first communication device or the second communication device in the various method embodiments described above.

[0419] For example, logic circuit 31 is used to implement processing-related operations performed by the first communication device or the second communication device in the above method embodiments; input / output interface 32 is used to implement sending and / or receiving-related operations performed by the first communication device or the second communication device in the above method embodiments.

[0420] This application also provides a computer-readable storage medium storing computer instructions for implementing the methods executed by the first communication device or the second communication device in the above-described method embodiments.

[0421] For example, when the computer program is executed by a computer, it enables the computer to implement the methods executed by the first communication device or the second communication device in the various embodiments of the above methods.

[0422] This application also provides a computer program product comprising instructions which, when executed by a computer, implement the methods performed by the first communication device or the second communication device in the above-described method embodiments.

[0423] This application also provides a communication system, including the aforementioned first communication device and second communication device.

[0424] The explanations and beneficial effects of the relevant contents in any of the devices provided above can be found in the corresponding method embodiments provided above, and will not be repeated here.

[0425] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0426] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0427] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0428] The units described as separate components may or may not be physically separate. 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 the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0429] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0430] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to related technologies, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0431] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A communication method, characterized in that, Applied to a first communication device, the method includes: Obtain the first downlink feature vector matrix, which is used to indicate the state information of the downlink channel; The codewords corresponding to the first downlink feature vector matrix are determined based on the encoder and processing function; Send the codeword to the second communication device. The processing function includes at least one of a first function and a second function, wherein the first function is used to adjust the eigenvector of the matrix input to the first function, and the second function is used to adjust the data format of the matrix input to the second function.

2. The method according to claim 1, characterized in that, When the processing function includes the first function and the second function, determining the codeword corresponding to the first downlink feature vector matrix based on the encoder and the processing function includes: A second downlink eigenvector matrix is ​​obtained based on the first function and the first downlink eigenvector matrix. The second downlink eigenvector matrix is ​​related to the downlink channel matrix, and the downlink channel matrix is ​​related to the uplink channel matrix. A third downlink feature vector matrix is ​​obtained based on the second downlink feature vector matrix, the downlink reference feature vector matrix, and the second function. The data format of the third downlink feature vector matrix is ​​the same as that of the downlink reference feature vector matrix. The codeword is determined based on the third downlink feature vector matrix and the encoder.

3. The method according to claim 2, characterized in that, The encoder, the first function, the second function, the first downlink feature vector matrix, and the codeword satisfy the following relationship: W BCE,DL =f BCE (W DL ) W IFA,DL ,b DL =f IFA (W BCE,DL ,W Ben,DL ) c=f en (W IFA,DL ;Φ) Among them, f BCE (·) represents the first function, W DL W represents the first downlink eigenvector matrix. BCE,DL Let f represent the second downlink eigenvector matrix. IFA (·) represents the second function, W IFA,DL W represents the third downlink eigenvector matrix. Ben,DL Let b represent the downlink reference eigenvector matrix. DL This represents first control information, which is used by the second communication device to reconstruct the second downlink feature vector matrix, wherein f en (·;φ) represents the encoder, φ represents the network parameters of the encoder, and c represents the codeword.

4. The method according to claim 3, characterized in that, The method further includes: The first control information is sent to the second communication device.

5. The method according to any one of claims 2 to 4, characterized in that, The step of obtaining the second downlink feature vector matrix based on the first function and the first downlink feature vector matrix includes: Based on the first function, the feature vector corresponding to the s-th sub-band in the first downlink feature vector matrix is ​​aligned with the reference vector of the downlink channel matrix corresponding to the s-th sub-band, and the feature vector corresponding to the s-th sub-band in the second downlink feature vector matrix is ​​obtained; The communication system to which the first communication device belongs is assigned N s N corresponding to each sub-band s The normalized eigenvectors are concatenated to obtain the second downlink eigenvector matrix. Where, N s s is a positive integer, and the value of s is less than or equal to N. s Positive integers.

6. The method according to claim 5, characterized in that, The eigenvector corresponding to the s-th sub-band in the second downlink eigenvector matrix is ​​represented by at least one of the following: or, or, in, Let ε represent the eigenvector corresponding to the s-th sub-band in the second downlink eigenvector matrix, ‖·‖ represent the l2 norm, and ε s,DL Represents the eigenvalue λ s,DL The corresponding feature space, h s,DL w represents the reference vector of the downlink channel matrix corresponding to the s-th subband. s,DL This represents the eigenvector corresponding to the s-th sub-band in the first downlink eigenvector matrix. Used to h s,DL Projected onto feature space ε s,DL , The columns constitute ε s,DL The orthogonal basis.

7. The method according to any one of claims 2 to 6, characterized in that, The process of obtaining the third downlink feature vector matrix based on the second downlink feature vector matrix and the second function includes: The second downlink eigenvector matrix is ​​transformed into a downlink sparse eigenvector matrix using a two-dimensional discrete Fourier transform. Based on the downlink sparse feature vector matrix and the downlink reference feature vector matrix, the first row cyclic shift step size and the first column cyclic shift step size are obtained; The third downlink feature vector matrix is ​​obtained based on the downlink sparse feature vector matrix, the first row cyclic shift step size, the first column cyclic shift step size, and the second function.

8. The method according to claim 7, characterized in that, The second downlink eigenvector matrix and the downlink sparse eigenvector matrix satisfy the following formula: Among them, W Spar,DL W represents the downlink sparse eigenvector matrix. BCE,DL Let F represent the second downlink eigenvector matrix. d and F a The Discrete Fourier Transform (DFT) matrix is... Represents a two-dimensional discrete Fourier transform; The downlink sparse eigenvector matrix, the first row cyclic shift step size, the first column cyclic shift step size, the second function, and the third downlink eigenvector matrix satisfy the following formula: Among them, W IFA,DL Let f represent the third downlink eigenvector matrix. IFA (·) represents the second function. This indicates the cyclic shift step size of the first row. This indicates the cyclic shift step size of the first column.

9. The method according to any one of claims 2 to 8, characterized in that, Determining the codeword based on the third downlink feature vector matrix and the encoder includes: The codeword is obtained by compressing the third downlink feature vector matrix based on the encoder.

10. A communication method, characterized in that, Applied to a second communication device, the method includes: Receive a codeword from a first communication device, wherein the codeword is the codeword corresponding to a first downlink feature vector matrix indicating downlink channel state information; The second uplink feature vector matrix is ​​determined based on the first uplink feature vector matrix and the processing function. The first uplink feature vector matrix is ​​used to indicate the state information of the uplink channel. The reconstructed third downlink feature vector matrix is ​​obtained based on the decoder, the codeword, and the second uplink feature vector matrix. The processing function includes at least one of a first function and a second function. The first function is used to adjust the eigenvector of the matrix input to the first function, and the second function is used to adjust the data format of the matrix input to the second function. The second downlink eigenvector matrix is ​​determined by the first function and the first downlink eigenvector matrix.

11. The method according to claim 10, characterized in that, When the processing function includes the first function and the second function, determining the second uplink feature vector matrix based on the first uplink feature vector matrix and the processing function includes: A third uplink feature vector matrix is ​​determined based on the first uplink feature vector matrix and the first function. The third uplink feature vector matrix is ​​related to the uplink channel matrix, and the uplink channel matrix is ​​related to the downlink channel matrix. The second uplink feature vector matrix is ​​determined based on the third uplink feature vector matrix, the uplink reference feature vector matrix, and the second function. The data format of the second uplink feature vector matrix is ​​the same as that of the uplink reference feature vector matrix.

12. The method according to claim 11, characterized in that, The first uplink eigenvector matrix, the first function, the second function, the third uplink eigenvector matrix, and the second uplink eigenvector matrix satisfy the following relationship: W BCE,UL =f BCE (W UL ) W IFA,UL ,b UL =f IFA (W BCE,UL ,W Ben,UL ) Among them, f BCE (·) represents the first function, W UL W represents the first uplink eigenvector matrix. BCE,UL f represents the third upward eigenvector matrix. IFA (·) represents the second function, the W Ben,UL Let W represent the reference matrix. Ben,UL Let b represent the uplink reference eigenvector matrix. UL This indicates the second control information, the W IFA,UL This represents the second upward eigenvector matrix.

13. The method according to claim 11 or 12, characterized in that, The step of determining the third uplink feature vector matrix based on the first uplink feature vector matrix and the first function includes: Based on the first function, align the feature vector corresponding to the s-th sub-band in the first uplink feature vector matrix with the reference vector of the uplink channel matrix corresponding to the s-th sub-band to obtain the feature vector corresponding to the s-th sub-band in the third uplink feature vector matrix; The communication system to which the second communication device belongs is assigned N s N corresponding to each sub-band s The normalized eigenvectors are concatenated to obtain the third upward eigenvector matrix. Where, N s s is a positive integer, and the value of s is less than or equal to N. s Positive integers.

14. The method according to claim 13, characterized in that, The eigenvector corresponding to the s-th sub-band in the third uplink eigenvector matrix is ​​represented by at least one of the following: or, or, in, This represents the eigenvector corresponding to the s-th sub-band in the third upward eigenvector matrix, where ||·|| represents the l2 norm, and ε s,UL Represents the eigenvalue λ s,UL The corresponding feature space, h s,UL w represents the reference vector of the uplink channel matrix corresponding to the s-th sub-band. s,UL This represents the eigenvector corresponding to the s-th sub-band in the first uplink eigenvector matrix. Used to h s,UL Projected onto feature space ε s,UL , The columns constitute ε s,UL The orthogonal basis.

15. The method according to any one of claims 11 to 14, characterized in that, The step of determining the second uplink feature vector matrix based on the third uplink feature vector matrix, the uplink reference feature vector matrix, and the second function includes: The third upward eigenvector matrix is ​​transformed into an upward sparse eigenvector matrix using a two-dimensional discrete Fourier transform. Based on the uplink sparse feature vector matrix and the uplink reference feature vector matrix, the second row cyclic shift step size and the second column cyclic shift step size are obtained. The second upward feature vector matrix is ​​obtained based on the upward reference feature vector matrix, the second row cyclic shift step size, the second column cyclic shift step size, and the second function.

16. The method according to claim 15, characterized in that, The third upward eigenvector matrix and the upward sparse eigenvector matrix satisfy the following formula: Among them, W Spar,UL W represents the uplink sparse eigenvector matrix. BCE,UL F represents the third upward eigenvector matrix. d and F a The Discrete Fourier Transform (DFT) matrix is... Represents a two-dimensional discrete Fourier transform; The up-row sparse eigenvector matrix, the second row cyclic shift step size, the second column cyclic shift step size, the second function, and the second up-row eigenvector matrix satisfy the following formula: Among them, W IFA,UL Let f represent the second upward eigenvector matrix. IFA (·) represents the second function. This indicates the cyclic shift step size for the second row. This indicates the cyclic shift step size of the second column.

17. The method according to any one of claims 10 to 16, characterized in that, The decoder, the codeword, the second uplink feature vector matrix, and the reconstructed third downlink feature vector matrix satisfy the following relationship: Among them, f de (·;Ψ) represents the decoder, Ψ represents the network parameters of the decoder, |W IFA,UL | represents the modulus of the second uplink feature vector matrix, and c represents the codeword. This represents the third row eigenvector matrix after reconstruction.

18. The method according to any one of claims 10 to 17, characterized in that, The method further includes: The reconstructed second downlink feature vector matrix is ​​obtained based on the inverse function of the second function, the reconstructed third downlink feature vector matrix, and the second control information. The third downlink feature vector matrix is ​​determined by the second downlink feature vector matrix and the second function.

19. The method according to claim 18, characterized in that, The inverse function of the second function, the reconstructed third downlink eigenvector matrix, the second control information, and the reconstructed second downlink eigenvector matrix satisfy the following relationship: in, Denotes the inverse function of the second function, where b UL This indicates the second control information, the This represents the third downlink eigenvector matrix after reconstruction. This represents the second downlink feature vector matrix after reconstruction.

20. A communication device, characterized in that, The communication device includes a processor and a memory coupled together. The memory is used to store a computer program. When the processor runs the computer program, the communication device performs the method as described in any one of claims 1-9; or, the communication device performs the method as described in any one of claims 10-19.

21. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed on a communication device, cause the communication device to perform the method as described in any one of claims 1-19.

22. A computer program product, characterized in that, The computer instructions, when executed on the communication device, cause the communication device to perform the method as described in any one of claims 1-19.

23. A chip, characterized in that, The chip includes a processor and a communication interface. The processor reads and executes instructions through the communication interface. When the chip is installed in a communication device, the communication device performs the method as described in any one of claims 1-16.

24. A communication system, characterized in that, The communication system includes a terminal device and a network device, wherein the terminal device is used to perform the method as described in any one of claims 1-9, and the network device is used to perform the method as described in any one of claims 10-19.

Images