Method for feeding back channel state information and communication apparatus

Abstract

A method for feeding back channel state information and a communication apparatus, which can update a contrastive learning network of a decoding network model on the basis of a high-compression-rate codeword and uplink channel information, and estimate downlink channel information of a current scenario on the basis of a low-compression-rate codeword, the uplink channel information and the decoding network model, thus reducing the overhead for feeding back CSI. The method comprises: after receiving a first codeword and a first uplink measurement pilot frequency, updating a contrastive learning network on the basis of the first codeword and first uplink channel information corresponding to the first uplink measurement pilot frequency; after receiving a second codeword and a second uplink measurement pilot frequency, inputting the second codeword and second uplink channel information corresponding to the second uplink measurement pilot frequency into the contrastive learning network; determining CSI common information by means of the contrastive learning network; adjusting a channel reconstruction network on the basis of the CSI common information and the second codeword; and, on the basis of the CSI common information and the parameter-adjusted channel reconstruction network, determining target downlink channel information.

Description

A method and communication device for feeding back channel state information Technical Field

[0001] The embodiments of the present application relate to the field of wireless communications, and in particular to a method and a communication device for feeding back channel state information. Background Art

[0002] As the number of antennas increases, the dimension of channel state information (CSI) becomes larger and larger, so feeding back CSI requires more overhead.

[0003] Currently, there is a method for feeding back CSI that is roughly as follows: after receiving the downlink measurement pilot sent by the base station, the terminal determines the downlink channel vector based on the downlink measurement pilot. The downlink channel vector is encoded into a codeword using a coding network model at a certain compression rate. The codeword is the compressed data of the downlink channel vector, and the compression rate is the ratio of the codeword length to the downlink channel vector length. The codeword is sent to the base station via a feedback link. The base station decodes the codeword based on the decoding network model. The decoding result is the estimated downlink channel vector.

[0004] The above method reduces CSI distortion by feeding back high-compression codewords, but the high-compression codewords incur large resource overhead.

[0005] Summary of the Invention

[0006] The present application provides a method for CSI feedback, which can determine a comparative learning network based on high-compression codewords and uplink channel information, and then obtain target downlink channel information based on low-compression codewords, uplink channel information, and the comparative learning network, thereby reducing the overhead of CSI feedback. The present application also provides a communication device capable of implementing the CSI feedback method.

[0007] In a first aspect, a method for feeding back CSI is provided, comprising: after sending a first downlink measurement pilot to a terminal, the receiving terminal determines first uplink channel information based on a first codeword sent by the first downlink measurement pilot and a first uplink measurement pilot sent by the terminal, and then updates a parameter value of a comparative learning network based on the first codeword and the first uplink channel information; then sending a second downlink measurement pilot to the terminal; after the receiving terminal determines second uplink channel information based on a second codeword sent by the second downlink measurement pilot and a second uplink measurement pilot sent by the terminal, inputs the second codeword and the second uplink channel information into the updated comparative learning network, outputs CSI common information through the comparative learning network, inputs the CSI common information and the second codeword into a super network, outputs a network parameter adjustment amount through the super network, adjusts the parameter value of a channel reconstruction network based on the network parameter adjustment amount, inputs the CSI common information into the adjusted channel reconstruction network, and outputs target downlink channel information through the adjusted channel reconstruction network. The compression rate of the second codeword is lower than the compression rate of the first codeword, and the super network includes but is not limited to a multilayer perceptron. The CSI public information includes one or more of angle, delay, Doppler component, and power.

[0008] Among them, the first downlink measurement pilot and the first uplink measurement pilot are measurement pilots for the first time period, and the second downlink measurement pilot and the second uplink measurement pilot are measurement pilots for the second time period. The method of the present application can determine the comparative learning network for the first time period based on the high compression rate codeword (i.e., the first codeword) and the first uplink channel information, and then obtain the target downlink channel information for the second time period based on the low compression rate codeword (i.e., the second codeword), the second uplink channel information and the comparative learning network, thereby reducing the overhead of CSI feedback in the second time period. Moreover, the super network can determine the network parameter adjustment amount of the current scenario based on the CSI public information and the low compression rate codeword, and can adjust the channel reconstruction network in the decoding network model based on the network parameter adjustment amount of the current scenario, so that the decoding network model can reconstruct the downlink channel information of different scenarios.

[0009] In one possible implementation, the first uplink channel information, the second uplink channel information, the first downlink channel information, the second downlink channel information, and the target downlink channel information are represented by channel vectors. Optionally, the channel vectors are obtained by performing singular value decomposition on a channel matrix. Alternatively, the channel vectors are obtained by transforming the channel matrix.

[0010] In another possible implementation manner, the first uplink channel information, the second uplink channel information, the first downlink channel information, the second downlink channel information, and the target downlink channel information are represented by a channel matrix.

[0011] In another possible implementation, a first downlink measurement pilot is sent to the terminal in a first measurement pilot period, and a second downlink measurement pilot is sent to the terminal in a second measurement pilot period. When the second measurement pilot period is longer than the first measurement pilot period, the duration of feeding back the first codeword is shorter than the duration of feeding back the second codeword, and the second codeword is a low-compression codeword. Therefore, feeding back the low-compression codeword for a long period of time can significantly reduce the overhead of feeding back CSI.

[0012] In another possible implementation, the first downlink measurement pilot and the second downlink measurement pilot meet at least one of the following conditions: the measurement pilot bandwidth of the first downlink measurement pilot is greater than the measurement pilot bandwidth of the second downlink measurement pilot, the measurement pilot density of the first downlink measurement pilot is greater than the measurement pilot density of the second downlink measurement pilot, and the number of measurement pilot ports of the first downlink measurement pilot is greater than the number of measurement pilot ports of the second downlink measurement pilot. In this way, a large number of first downlink measurement pilots can be sent, thereby improving the accuracy of the terminal's estimation of the downlink channel state.

[0013] In another possible implementation, the CSI feedback method of the present application further includes: sending a first compression ratio identifier to the terminal; and sending a second compression ratio identifier to the terminal. The first compression ratio identifier corresponds to the compression ratio of the first codeword, and the second compression ratio identifier corresponds to the compression ratio of the second codeword, thereby explicitly instructing the terminal to send high-compression ratio codewords and low-compression ratio codewords.

[0014] In another possible implementation, the contrastive learning network includes a first feature extraction network, a second feature extraction network, and a common information network. The first feature extraction network is used to extract codeword features, and the second feature extraction network is used to extract uplink channel features. The first and second feature extraction networks are identical. The common information network is used to determine CSI common information based on the codeword features and uplink channel features. The common information network is a deep neural network, a convolutional neural network, or a recurrent neural network.

[0015] In another possible implementation, the network parameter adjustment amount includes at least one of network structure data, neuron weights, bias, and activation function. The network structure data includes one or more of network layer identifiers, the number of neurons in the network, and neuron identifiers in the network.

[0016] In another possible implementation, the method for feeding back CSI in the present application further includes: precoding data according to target downlink channel information; sending the precoded data to the terminal, and then measuring a communication quality indicator based on the precoded data, and when the communication quality indicator is greater than a preset threshold, triggering the step of sending a first downlink measurement pilot to the terminal. The target downlink channel information can represent the downlink CSI of the current scenario, and data can be transmitted based on the downlink CSI. The communication quality of the data can be measured by a communication quality indicator, and the communication quality indicator includes a signal-to-noise ratio and / or a block error rate. When the communication quality is poor, CSI can be fed back again to obtain new downlink channel information.

[0017] A second aspect provides a method for feeding back CSI, the method comprising: receiving a first downlink measurement pilot sent by an access network device, determining first downlink channel information based on the first downlink measurement pilot, inputting the first downlink channel information into a first compression network model, and outputting a first codeword through the first compression network model; sending the first codeword to the access network device, and sending a first uplink measurement pilot to the access network device; receiving a second downlink measurement pilot sent by the access network device, determining second downlink channel information based on the second downlink measurement pilot, inputting the second downlink channel information into a second compression network model, outputting a second codeword through the second compression network model, sending the second codeword to the access network device, and sending a second uplink measurement pilot to the access network device. The first codeword is a codeword for a first time period, and the second codeword is a codeword for a second time period. The compression ratio of the second codeword is lower than the compression ratio of the first codeword, indicating that a high-compression-ratio codeword and an uplink measurement pilot are sent in the first time period, and a low-compression-ratio codeword and an uplink measurement pilot are sent in the second time period. The access network device can determine target downlink channel information based on the low-compression-ratio codeword and the uplink measurement pilot, thereby reducing the overhead of feeding back CSI in the second time period.

[0018] In one possible implementation, a first downlink measurement pilot signal sent by an access network device in a first measurement pilot period is received, and a second downlink measurement pilot signal sent by the access network device in a second measurement pilot period is received. Because the second measurement pilot period is longer than the first measurement pilot period, the terminal feeds back a low-compression codeword in the second measurement pilot period, significantly reducing feedback overhead.

[0019] In another possible implementation, after receiving a first compression rate identifier from an access network device, the first downlink channel information is input into a first compression network model based on the first compression rate identifier. After receiving a second compression rate identifier from the access network device, the second downlink channel information is input into a second compression network model based on the second compression rate identifier. In this way, codewords with different compression rates can be fed back based on the compression rate identifier.

[0020] The third aspect provides a method for feeding back channel state information (CSI), which includes: after sending a first downlink measurement pilot to a terminal, receiving a first codeword sent by the terminal according to the first downlink measurement pilot; after receiving a first uplink measurement pilot sent by the terminal, determining first uplink channel information according to the first uplink measurement pilot; updating a parameter value of a comparative learning network according to the first codeword and the first uplink channel information; after sending a second downlink measurement pilot to the terminal, receiving a second codeword sent by the terminal according to the second downlink measurement pilot; after receiving a second uplink measurement pilot sent by the terminal, determining second uplink channel information according to the second uplink measurement pilot; inputting the second codeword and the second uplink channel information into a comparative learning network, and outputting CSI public information through the comparative learning network; sending the CSI public information to the terminal; receiving a network parameter adjustment amount sent by the terminal; adjusting a parameter value of a channel reconstruction network according to the network parameter adjustment amount; inputting the CSI public information into a channel reconstruction network after parameter adjustment, and then outputting target downlink channel information through the channel reconstruction network after parameter adjustment. The compression rate corresponding to the second codeword is less than the compression rate corresponding to the first codeword.

[0021] When a super network is deployed on a terminal, it can determine the network parameter adjustment for the current scenario, reducing the computational overhead of the base station. After the terminal sends this adjustment to the access network device, the access network device adjusts the channel reconstruction network in the decoding network model based on the adjustment, enabling the decoding network model to reconstruct downlink channel information for different scenarios.

[0022] In a possible implementation of the third aspect, a first downlink measurement pilot is sent to the terminal in a first measurement pilot period, and a second downlink measurement pilot is sent to the terminal in a second measurement pilot period, where the second measurement pilot period is longer than the first measurement pilot period. Because the second measurement pilot period is longer than the first measurement pilot period, the terminal feeds back a low-compression-rate codeword in the second measurement pilot period, significantly reducing feedback overhead.

[0023] In another possible implementation of the third aspect, the first downlink measurement pilot and the second downlink measurement pilot meet at least one of the following conditions: a measurement pilot bandwidth of the first downlink measurement pilot is greater than a measurement pilot bandwidth of the second downlink measurement pilot, a measurement pilot density of the first downlink measurement pilot is greater than a measurement pilot density of the second downlink measurement pilot, and the number of measurement pilot ports of the first downlink measurement pilot is greater than the number of measurement pilot ports of the second downlink measurement pilot. In this way, a large number of first downlink measurement pilots can be sent, thereby improving the accuracy of downlink channel state estimation by the terminal.

[0024] In another possible implementation of the third aspect, the method for feeding back CSI in the present application further includes: after sending the first downlink measurement pilot to the terminal, sending a first compression ratio identifier to the terminal; and after sending the second downlink measurement pilot to the terminal, sending a second compression ratio identifier to the terminal. The first compression ratio identifier corresponds to the compression ratio of the first codeword; and the second compression ratio identifier corresponds to the compression ratio of the second codeword. This can explicitly instruct the terminal to feed back codewords with different levels of compression ratios.

[0025] In another possible implementation of the third aspect, the contrastive learning network includes a first feature extraction network, a second feature extraction network, and a common information network. The first feature extraction network is used to extract codeword features, and the second feature extraction network is used to extract uplink channel features. The first and second feature extraction networks are identical. The common information network is used to determine CSI common information based on the codeword features and uplink channel features.

[0026] In another possible implementation of the third aspect, the network parameter adjustment amount includes at least one of network structure data, neuron weights, bias and activation function.

[0027] In another possible implementation of the third aspect, the method for feeding back CSI of the present application further includes: precoding data according to target downlink channel information; sending the precoded data to a terminal; and measuring a communication quality indicator based on the data, and when the communication quality indicator is greater than a preset threshold, triggering the step of sending a first downlink measurement pilot to the terminal. The communication quality indicator includes a signal-to-noise ratio and / or a block error rate.

[0028] In another possible implementation manner of the third aspect, the method for feeding back CSI of the present application further includes: sending a network parameter value of the super network to the terminal.

[0029] The fourth aspect provides a method for feeding back channel state information (CSI), which comprises receiving a first downlink measurement pilot sent by an access network device; determining first downlink channel information based on the first downlink measurement pilot; inputting the first downlink channel information into a first compression network model, and outputting a first codeword through the first compression network model; sending the first codeword to the access network device; sending a first uplink measurement pilot to the access network device; receiving a second downlink measurement pilot sent by the access network device; determining second downlink channel information based on the second downlink measurement pilot; inputting the second downlink channel information into a second compression network model, and outputting a second codeword through the second compression network model, wherein the compression rate of the second codeword is less than that of the first codeword; sending the second codeword to the access network device; sending the second uplink measurement pilot to the access network device; receiving CSI public information sent by the access network device, wherein the CSI public information is determined by the access network device based on the second codeword and the second uplink measurement pilot; inputting the CSI public information into a super network, and outputting a network parameter adjustment amount through the super network; and sending the network parameter adjustment amount to the access network device.

[0030] In one possible implementation, a first downlink measurement pilot signal sent by an access network device in a first measurement pilot period is received; and a second downlink measurement pilot signal sent by the access network device in a second measurement pilot period is received. The second measurement pilot period is longer than the first measurement pilot period. The second codeword sent by the terminal in the second measurement pilot period is a low-compression codeword, thereby reducing the overhead of CSI feedback.

[0031] In another possible implementation, a first compression rate identifier is received from an access network device, and the first downlink channel information is input into a first compression network model based on the first compression rate identifier. A second compression rate identifier is received from the access network device, and the second downlink channel information is input into a second compression network model based on the second compression rate identifier. In this way, codewords with different levels of compression rates can be fed back based on the compression rate table identifier.

[0032] In another possible implementation, before receiving the first downlink measurement pilot signal sent by the access network device, a network parameter value of a super network sent by the access network device is received; and the super network is configured according to the network parameter value of the super network. The network parameter value of the super network includes a super parameter of the super network.

[0033] The fifth aspect provides a method for feeding back CSI, which includes: after sending a downlink measurement pilot to the terminal, the receiving terminal determines the first uplink channel information based on the codeword sent by the downlink measurement pilot and the first uplink measurement pilot sent by the receiving terminal; after inputting the codeword and the first uplink channel information into the comparative learning network, the CSI common information is determined through the comparative learning network, after receiving the second uplink measurement pilot sent by the terminal, the second uplink channel information is determined based on the second uplink measurement pilot, the second uplink channel information is input into the super network, and the network parameter adjustment amount is output through the super network; the parameter value of the channel reconstruction network is adjusted according to the network parameter adjustment amount; the CSI common information is input into the channel reconstruction network after parameter adjustment, and the target downlink channel information is output through the channel reconstruction network after parameter adjustment.

[0034] According to this implementation, the CSI common information for the first period can be determined based on the high-compression codeword and uplink channel information. Then, the downlink channel information for the second period can be determined based on the uplink channel information and the CSI common information from the first period, thereby reducing the overhead of CSI feedback in the second period. Furthermore, the super network can determine the network parameter adjustment amount for the current scenario based on the uplink channel information. Based on the network parameter adjustment amount for the current scenario, the channel reconstruction network in the decoding network model can be adjusted, enabling the decoding network model to reconstruct downlink channel information for different scenarios.

[0035] In one possible implementation, a terminal receives a first uplink measurement pilot signal sent in a first measurement pilot period, and receives a second uplink measurement pilot signal sent in a second measurement pilot period. Because the first measurement pilot period is shorter than the second measurement pilot period, no codeword is fed back in the second measurement pilot period, thereby reducing the overhead of CSI feedback.

[0036] In another possible implementation, data is precoded based on target downlink channel information; the precoded data is sent to the terminal, and a communication quality indicator is obtained based on the data. When the communication quality indicator is greater than a preset threshold, indicating poor communication quality, a downlink measurement pilot is sent to the terminal to update the target downlink channel information. The communication quality indicator includes a signal-to-noise ratio and / or a block error rate.

[0037] A sixth aspect provides a method for feeding back CSI, comprising: receiving a downlink measurement pilot signal from an access network device, determining downlink channel information based on the downlink measurement pilot signal; inputting the downlink channel information into a compression network model, and outputting a codeword from the compression network model; sending the codeword to the access network device; and after sending a first uplink measurement pilot signal to the access network device, sending a second uplink measurement pilot signal to the access network device. In this manner, the terminal feeds back a high-compression codeword signal during a first time period and does not feed back a codeword signal during a second time period, thereby reducing the overhead of feeding back CSI.

[0038] In a possible implementation, the first uplink measurement pilot is sent to the access network device in a first measurement pilot period, and the first uplink measurement pilot is sent to the access network device in a second measurement pilot period. The first measurement pilot period is shorter than the second measurement pilot period.

[0039] The seventh aspect provides a communication device, comprising a module for executing the method as described in the first aspect or any one of the implementations of the first aspect, or comprising a module for executing the method as described in the third aspect or any one of the implementations of the third aspect, or comprising a module for executing the method as described in the fifth aspect or any one of the implementations of the fifth aspect.

[0040] The eighth aspect provides a communication device, comprising a module for executing the method as described in the second aspect or any one of the implementations of the second aspect, or comprising a module for executing the method as described in the fourth aspect or any one of the implementations of the fourth aspect, or comprising a module for executing the method as described in the sixth aspect or any one of the implementations of the sixth aspect.

[0041] The ninth aspect provides a communication device, comprising a processor and an interface circuit, wherein the interface circuit is used to receive signals from other communication devices and transmit them to the processor or send signals from the processor to other communication devices, and the processor is used to implement the method as described in the first aspect or any one of the implementations of the first aspect through a logic circuit or executing code instructions, or to implement the method as described in the third aspect or any one of the implementations of the third aspect, or to implement the method as described in the fifth aspect or any one of the implementations of the fifth aspect.

[0042] The tenth aspect provides a communication device, comprising a processor and an interface circuit, wherein the interface circuit is used to receive signals from other communication devices and transmit them to the processor or send signals from the processor to other communication devices, and the processor is used to implement the method as described in the second aspect or any one of the implementations of the second aspect through a logic circuit or executing code instructions, or to implement the method as described in the fourth aspect or any one of the implementations of the fourth aspect, or to implement the method as described in the sixth aspect or any one of the implementations of the sixth aspect.

[0043] The eleventh aspect provides a computer-readable storage medium, which stores a computer program or instruction. When the computer program or instruction is executed by a communication device, it implements the method described in the first aspect, any one of the implementations of the first aspect, the third aspect, or any one of the implementations of the third aspect, or is used to implement the method described in the fifth aspect or any one of the implementations of the fifth aspect.

[0044] The twelfth aspect provides a computer-readable storage medium, which stores a computer program or instruction. When the computer program or instruction is executed by a communication device, it implements the method described in the second aspect, any one of the implementations of the second aspect, the fourth aspect, or any one of the implementations of the fourth aspect, or is used to implement the method described in the sixth aspect or any one of the implementations of the sixth aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] FIG1 is a schematic diagram of the architecture of a mobile communication system used in an embodiment of the present application;

[0046] FIG2 is a flow chart of a limited CSI feedback method according to an embodiment of the present application;

[0047] FIG3 is another flowchart of the limited CSI feedback method according to an embodiment of the present application;

[0048] FIG4A is a schematic diagram of a method for feeding back CSI in an embodiment of the present application;

[0049] FIG4B is another schematic diagram of a method for feeding back CSI in an embodiment of the present application;

[0050] FIG5A is a schematic diagram of a decoding network model processing data in an embodiment of the present application;

[0051] FIG5B is another schematic diagram of a decoding network model processing data in an embodiment of the present application;

[0052] FIG6 is a schematic diagram of a Markov random field block in an embodiment of the present application;

[0053] FIG7 is a schematic diagram of the codeword bit number and the achievable rate ratio in an embodiment of the present application;

[0054] FIG8 is a flowchart of a zero-feedback CSI method according to an embodiment of the present application;

[0055] FIG9A is a schematic diagram of a zero-feedback CSI method according to an embodiment of the present application;

[0056] FIG9B is another schematic diagram of the zero-feedback CSI method according to an embodiment of the present application;

[0057] FIG10A is another schematic diagram of a decoding network model processing data in an embodiment of the present application;

[0058] FIG10B is another schematic diagram of a decoding network model processing data in an embodiment of the present application;

[0059] FIG11 is a schematic diagram of uplink and downlink frequency domain spacing and achievable rate ratios in an embodiment of the present application;

[0060] FIG12 is a flowchart of a method for feeding back CSI when a super network is deployed in a terminal according to an embodiment of the present application;

[0061] FIG13 is another flowchart of a method for feeding back CSI when a super network is deployed in a terminal according to an embodiment of the present application;

[0062] FIG14 is a structural diagram of a communication device according to an embodiment of the present application;

[0063] FIG15 is another structural diagram of the communication device in an embodiment of the present application. DETAILED DESCRIPTION

[0064] Figure 1 is a schematic diagram of the architecture of a communication system 1000 used in an embodiment of the present application. As shown in Figure 1 , the communication system includes a radio access network 100 and a core network 200. Optionally, the communication system 1000 may also include the Internet 300. The radio access network 100 may include at least one radio access network device (such as 110a and 110b in Figure 1 ) and at least one terminal (such as 120a-120j in Figure 1 ). The terminal is wirelessly connected to the radio access network device, and the radio access network device is wirelessly or wiredly connected to the core network. The core network device and the radio access network device may be independent, distinct physical devices, or the core network device's functions and the radio access network device's logical functions may be integrated into the same physical device, or a single physical device may integrate some of the core network device's functions and some of the radio access network device's functions. Terminals and radio access network devices may be interconnected via wired or wireless connections. Figure 1 is merely a schematic diagram. The communication system may also include other network devices, such as wireless relay devices and wireless backhaul devices, which are not shown in Figure 1 .

[0065] Radio access network equipment is the access device that terminals use to wirelessly access a communication system. Radio access network equipment can be a base station, an evolved NodeB (eNodeB), a transmission reception point (TRP), a next-generation NodeB (gNB) in fifth-generation (5G) mobile communication systems, a next-generation base station in sixth-generation (6G) mobile communication systems, a base station in future mobile communication systems, or an access node in a WiFi system. It can also be a module or unit that performs some of the functions of a base station, such as a centralized unit (CU) or a distributed unit (DU). The CU here completes the functions of the radio resource control protocol and the packet data convergence protocol (PDCP) of the base station, and can also complete the function of the service data adaptation protocol (SDAP); the DU completes the functions of the radio link control layer and the medium access control (MAC) layer of the base station, and can also complete the functions of part of the physical layer or all of the physical layer. For the specific description of each of the above protocol layers, please refer to the relevant technical specifications of the 3rd Generation Partnership Project (3GPP). The wireless access network device can be a macro base station (such as 110a in Figure 1), a micro base station or an indoor station (such as 110b in Figure 1), or a relay node or a donor node, etc. The embodiments of the present application do not limit the specific technology and specific device form adopted by the wireless access network device. For the convenience of description, the following description takes the base station as an example of the wireless access network device.

[0066] A terminal is a device with wireless transceiver capabilities that can send signals to or receive signals from a base station. A terminal may also be referred to as a terminal device, user equipment (UE), mobile station, or mobile terminal. Terminals can be widely used in various scenarios, such as device-to-device (D2D), vehicle-to-everything (V2X) communication, machine-type communication (MTC), the Internet of Things (IoT), virtual reality, augmented reality, industrial control, autonomous driving, telemedicine, smart grids, smart furniture, smart offices, smart wearables, smart transportation, and smart cities. A terminal can be a mobile phone, tablet computer, computer with wireless transceiver capabilities, wearable device, vehicle, aircraft, ship, robot, robotic arm, smart home appliance, or the like. The embodiments of this application do not limit the specific technology or device form factor used by the terminal.

[0067] Base stations and terminals can be fixed or mobile. They can be deployed on land, indoors or outdoors, handheld or vehicle-mounted; on water; or on aircraft, balloons, and satellites. The embodiments of this application do not limit the application scenarios of base stations and terminals.

[0068] The roles of base stations and terminals can be relative. For example, the helicopter or drone 120i in Figure 1 can be configured as a mobile base station. To terminals 120j accessing the wireless access network 100 via 120i, terminal 120i is a base station. However, to base station 110a, 120i is a terminal, meaning that communication between 110a and 120i occurs via a wireless air interface protocol. Of course, communication between 110a and 120i can also occur via a base station-to-base station interface protocol. In this case, 120i is also a base station relative to 110a. Therefore, base stations and terminals can be collectively referred to as communication devices. 110a and 110b in Figure 1 can be referred to as communication devices with base station functionality, while 120a-120j in Figure 1 can be referred to as communication devices with terminal functionality.

[0069] Communication between base stations and terminals, between base stations, and between terminals can be carried out through authorized spectrum, unauthorized spectrum, or both; communication can be carried out through spectrum below 6 gigahertz (GHz), spectrum above 6 GHz, or spectrum below 6 GHz and spectrum above 6 GHz. The embodiments of the present application do not limit the spectrum resources used for wireless communication.

[0070] In the embodiments of the present application, the functions of the base station may also be performed by a module (such as a chip) in the base station, or by a control subsystem that includes the base station functions. The control subsystem that includes the base station functions here may be a control center in the above-mentioned application scenarios such as smart grid, industrial control, smart transportation, and smart city. The functions of the terminal may also be performed by a module (such as a chip or modem) in the terminal, or by a device that includes the terminal functions.

[0071] In this application, a base station sends downlink signals or downlink information to a terminal, and the downlink information is carried on a downlink channel; the terminal sends uplink signals or uplink information to the base station, and the uplink information is carried on an uplink channel. In order to communicate with the base station, the terminal needs to establish a wireless connection with the cell controlled by the base station. The cell with which the terminal has established a wireless connection is called the serving cell of the terminal. When the terminal communicates with the serving cell, it will also be subject to interference from signals in neighboring cells.

[0072] The following is an introduction to the network model involved in this application:

[0073] A neural network is a mathematical or computational model that mimics the structure and function of biological neural networks (the central nervous system of animals, particularly the brain). Neural networks perform computations by connecting a large number of neurons. A neural network can include multiple layers with different functions, each with parameters and computational rules. Different layers within a neural network have different names depending on the computational formula or function. For example, a layer that performs convolution is called a convolutional layer, and convolutional layers are often used to extract features from input signals (such as images). A neural network can also be composed of multiple sub-neural networks. Different neural network structures can be applied to different scenarios (such as classification and recognition) or provide different results when used in the same scenario. Differences in neural network structure can include one or more of the following: the number of layers within the neural network, the order of the layers, and the weights, parameters, or computational formulas within each layer. A variety of highly accurate neural networks exist in the industry for applications such as recognition and classification. Some neural networks can be trained with specific datasets and then used independently to complete a task or combined with other neural networks (or other functional modules) to complete a task.

[0074] Deep learning models are machine learning models with complex neural network structures. Depending on whether the training model relies on labels corresponding to the training data, deep learning models can be divided into supervised and unsupervised learning models, which will not be discussed here. Classic deep learning models include convolutional neural networks (CNNs), recurrent neural networks (RNNs), and recursive neural networks (RNNs). In RNNs, the output of a neuron can directly affect itself at the next moment. That is, the input of a neuron in layer i at time m includes not only the output of the neuron in layer (i-1) at that moment, but also its own output at time (m-1). In CNNs, not all neurons in the upper and lower layers are directly connected; instead, they are connected through a "convolution kernel" as an intermediary. In this network, the information to be processed retains its original positional relationship after the convolution operation.

[0075] Contrastive learning networks are either self-supervised or unsupervised networks. They can encode similar data in a similar way and make the encoding results of different data as different as possible.

[0076] Multilayer perceptron (MLP) is a forward-structured artificial neural network that includes an input layer, an output layer, and multiple hidden layers. The multiple hidden layers are fully connected.

[0077] Base stations can be deployed in cities, towns, deserts, oceans, grasslands, and mountainous areas. They can be deployed indoors or outdoors. In each scenario, the base station uses a decoding network model to decode the codeword into a target downlink channel matrix. The model parameters of the decoding network model are adjusted based on the loss between the target downlink channel matrix and the original downlink channel matrix. The decoding network model is then updated based on the model parameters. The base station iterates these steps until the loss between the reconstructed downlink channel matrix and the original downlink channel matrix is ​​less than a preset loss value.

[0078] When a terminal switches from one scenario to another, the channel matrix between the base station and the terminal also changes. Because the downlink channel matrices vary significantly across different scenarios, if the compression network model or decoding network model doesn't fit the current scenario, the downlink channel information generated by the decoding network model also won't fit the current scenario. This can lead to a high block error rate in the transmitted data, necessitating retraining of the decoding network model.

[0079] To address the high overhead of training a decoding network model using high-compression codewords, this application can, after training the decoding network model using high-compression codewords, update the decoding network model using low-compression codewords or an uplink channel matrix, thereby reducing the overhead of CSI feedback. In this application, the method in which the terminal feeds back low-compression codewords and the access network device updates the decoding network model based on the low-compression codewords is called a limited-feedback CSI method. The method in which the terminal does not feed back compression codewords and the access network device updates the decoding network model based on the uplink channel matrix is ​​called a zero-feedback CSI method.

[0080] In the limited feedback CSI method, the terminal will feedback high-compression codewords and low-compression codewords, which are generated by different compression network models. The trigger condition for feedback codewords can be a preset time period or a compression rate indicator sent by the access network equipment.

[0081] The following implementation subject takes a terminal and an access network device as an example to introduce a method for feeding back CSI without a compression rate identifier. Referring to FIG2 , an embodiment of the method for feeding back CSI with limited information in this application includes:

[0082] Step 201: The access network device sends a first downlink measurement pilot to the terminal.

[0083] Step 202: The terminal determines first downlink channel information according to the first downlink measurement pilot.

[0084] The first downlink channel information may be a downlink channel matrix for the first time period or a downlink channel vector for the first time period. The second downlink channel information may be a downlink channel matrix for the second time period or a downlink channel vector for the second time period. In the present application, a downlink channel vector may be a vector obtained by sequentially arranging the rows of a downlink channel matrix, a vector obtained by sequentially arranging the columns of a downlink channel matrix, or a vector obtained by performing singular value decomposition of the downlink channel matrix. It should be understood that the method for obtaining the downlink channel vector is not limited to the above examples.

[0085] Step 203: The terminal inputs the first downlink channel information into the first compression network model, and outputs the first codeword through the first compression network model.

[0086] Step 204: The terminal sends the first codeword to the access network device.

[0087] Step 205: The terminal sends the first uplink measurement pilot to the access network device.

[0088] Step 206: The access network device determines first uplink channel information according to the first uplink measurement pilot.

[0089] In the present application, the first uplink channel information may be the uplink channel matrix of the first time period or the uplink channel vector of the first time period. The second uplink channel information may be the uplink channel matrix of the second time period or the uplink channel vector of the second time period. The uplink channel vector may be a vector obtained by sequentially arranging the rows of the uplink channel matrix, a vector obtained by sequentially arranging the columns of the uplink channel matrix, or a vector obtained by performing singular value decomposition of the uplink channel matrix. It should be understood that the method for obtaining the uplink channel vector is not limited to the above examples.

[0090] Step 207: The access network device updates and compares the first codeword and the first uplink channel information to learn the parameter values ​​of the network.

[0091] The access network equipment is deployed with a pre-trained decoding network model, which includes a contrastive learning network, a channel reconstruction network, and a super network. Optionally, the contrastive learning network includes a first feature extraction network, a second feature extraction network, and a public information network. The first feature extraction network is used to extract codeword features, the second feature extraction network is used to extract uplink channel features, and the public information network is used to determine CSI public information based on the codeword features and uplink channel features. Optionally, both the first feature extraction network and the second feature extraction network are neural networks.

[0092] The public information network can be a deep neural network, a recurrent neural network, or a convolutional neural network, or the public information network is a single network layer. When the public information network is a convolutional neural network, the public information network includes a convolution layer and an activation layer, or the public information network includes a convolution layer, a pooling layer, and an activation layer. When the public information network is a single network layer, the network layer is used to implement the function.

[0093] Step 208: The access network device sends a second downlink measurement pilot to the terminal.

[0094] Step 209: The terminal determines the second downlink channel information according to the second downlink measurement pilot.

[0095] Step 210: The terminal inputs the second downlink channel information into the second compression network model, and outputs a second codeword through the second compression network model.

[0096] The condition for triggering the execution of step 203 or step 210 may be a preset period. For example, steps 201 to 203 are executed in the first period, and steps 208 to 210 are executed in the second period. The length of the period can be set according to actual conditions and is not limited in this application.

[0097] The compression rate of the second codeword is less than the compression rate of the first codeword. For example, the compression rate of the first codeword is 1 / 4 or 1 / 8, and the compression rate of the second codeword is 1 / 16, 1 / 32, or 1 / 64. It should be understood that the compression rate of the first codeword and the compression rate of the second codeword can be set according to actual conditions and are not limited in this application. The first compression network model is used to compress the downlink channel information into a high-compression codeword, and the second compression network model is used to compress the downlink channel information into a low-compression codeword.

[0098] Step 211: The access network device receives the second codeword sent by the terminal.

[0099] Step 212: The terminal sends the second uplink measurement pilot to the access network device.

[0100] Step 213: The access network device determines second uplink channel information according to the second uplink measurement pilot.

[0101] Step 214: The access network device inputs the second uplink channel information into the comparative learning network, which then outputs CSI common information. CSI common information includes identical and similar portions of the uplink and downlink channel information. When the difference between the uplink and downlink channel information is less than a threshold, they are determined to be CSI common information. When the difference between the uplink and downlink channel information is greater than or equal to a threshold, they are determined not to be CSI common information. Specifically, CSI common information includes one or more of angle, delay, Doppler component, and power.

[0102] Step 215: The access network device inputs the CSI public information and the second codeword into the super network, and outputs the network parameter adjustment value through the super network. The super network may include but is not limited to a multilayer perceptron (MLP).

[0103] Step 216: The access network device adjusts the parameter value of the channel reconstruction network according to the network parameter adjustment amount.

[0104] The network parameter adjustment amount includes at least one of network structure data, neuron weights, bias and activation function.

[0105] The network structure data includes one or more of a network layer identifier, a number of neurons, and a neuron identifier. When the network parameter adjustment includes the network structure data, the parameter values ​​of one or more network layers in the channel reconstruction network may be adjusted. Optionally, the parameter value of the last network layer in the channel reconstruction network may be adjusted.

[0106] The present application can adjust the neuron weights, biases, and activation functions of the network layers in the channel reconstruction network based on the network parameter adjustment amount. Specific operations may include, but are not limited to, multiplication, convolution, substitution, and exclusive-or. For example, the activation function of the network layer in the channel reconstruction network can be replaced with the activation function of the network parameter adjustment amount. Alternatively, the activation function of the network parameter adjustment amount can be multiplied with the activation function of the network layer in the channel reconstruction network. Other operations are similar and will not be described in detail here.

[0107] Step 217: The access network device inputs the CSI public information into the adjusted channel reconstruction network, and outputs the target downlink channel information through the adjusted channel reconstruction network.

[0108] In the present application, the first downlink measurement pilot and the first uplink measurement pilot belong to the measurement pilot of the first time period, and the second downlink measurement pilot and the second uplink measurement pilot belong to the measurement pilot of the second time period. The downlink measurement pilot can be, but is not limited to, a channel state information-reference signal (CSI-RS). The uplink measurement pilot can be, but is not limited to, a channel sounding reference signal (SRS). The duration of the first time period is less than the duration of the second time period, so that a long, low-compression codeword can be used to obtain the downlink channel CSI.

[0109] It should be noted that the terminal may receive the first downlink measurement pilot and send the first uplink measurement pilot in the first measurement pilot period, and receive the second downlink measurement pilot and send the second uplink measurement pilot in the second measurement pilot period. The first measurement pilot period is shorter than the second measurement pilot period, so the terminal may send a low-compression codeword in the second measurement pilot period, thereby significantly reducing feedback overhead.

[0110] The first downlink measurement pilot and the second downlink measurement pilot satisfy at least one of the following conditions: the measurement pilot bandwidth of the first downlink measurement pilot is greater than the measurement pilot bandwidth of the second downlink measurement pilot, the measurement pilot density of the first downlink measurement pilot is greater than the measurement pilot density of the second downlink measurement pilot, and the number of measurement pilot ports of the first downlink measurement pilot is greater than the number of measurement pilot ports of the second downlink measurement pilot. In this way, a large number of first downlink measurement pilots can be sent in the first measurement pilot period, so that the terminal can improve the accuracy of estimating downlink state information based on the large number of first downlink measurement pilots.

[0111] In this embodiment, the access network device can determine a comparative learning network for the first period based on high-compression codewords and uplink channel information, adjust a channel reconstruction network based on low-compression codewords and uplink channel information, and use this channel reconstruction network to output target downlink channel information, which can be considered downlink channel CSI. Feedback of low-compression codewords in the second period enables estimation of downlink channel CSI, thereby reducing CSI feedback overhead.

[0112] Secondly, the super network can determine the network parameter adjustment amount for the current scenario based on the CSI public information and low-compression codewords. Based on the network parameter adjustment amount for the current scenario, the channel reconstruction network in the decoding network model can be adjusted, so that the decoding network model can accurately estimate the downlink channel CSI in different scenarios.

[0113] In the present application, the first compression network and the second compression network of the terminal belong to the encoding network, and the first compression network and the decoding network model can constitute an end-to-end network. The loss function of the end-to-end network includes end-to-end loss and contrast loss. The sending end and the receiving end exchange training information to complete the training of the end-to-end network. For example, after the receiving end obtains the downlink channel matrix, it sends the downlink channel matrix to the sending end, and the downlink channel matrix is ​​used as a training label. The downlink channel matrix estimated by the sending end is used as a training sample. The training sample can be offline data or online data, which is not limited in this application. The training sample can also be a downlink channel vector.

[0114] End-to-end loss includes bit error indicator loss and contrast loss. The bit error indicator can be, but is not limited to, the achievable rate. Contrast loss refers to the loss between the output data of the downlink channel codeword after the contrastive learning network and the output data of the uplink channel information after the contrastive learning network.

[0115] In an optional embodiment, the loss function of the end-to-end network is:

[0116] In this loss function, is the achievable rate loss, is the contrast loss.

[0117] N b is the total number of training batches, is the total number of subcarriers, O c is the CSI common information part corresponding to the uplink channel information, O down is the CSI public information part corresponding to the codeword, is the downlink channel vector of the i-th subcarrier estimated by the terminal based on the l-th batch of downlink measurement pilots, It is the downlink channel vector of the i-th subcarrier in the l-th batch of data output by the decoding network model, where l is the training batch index and i is the subcarrier index.

[0118] In an optional embodiment, after step 207, the method for feeding back CSI in the present application further includes: the access network device outputs the CSI public information of the first time period through the comparative learning network, inputs the CSI public information of the first time period into the channel reconstruction network, and outputs the downlink channel information of the first time period through the channel reconstruction network.

[0119] In this embodiment, the access network device can reconstruct the downlink channel CSI of the first time period based on the first codeword and the first uplink channel information, precode the data based on the downlink channel information of the first time period, send the precoded data to the terminal, measure the communication quality index based on the above data, and when the communication quality index is greater than the preset threshold, re-execute steps 201 to 217.

[0120] In another optional embodiment, after step 217, the method for feeding back CSI in the present application further includes: precoding data according to target downlink channel information; sending the precoded data to the terminal; measuring a communication quality indicator based on the data, and when the communication quality indicator is greater than a preset threshold, re-executing steps 201 to 217. When the communication quality indicator is less than or equal to the preset threshold, it indicates that the communication quality of the current downlink channel can guarantee data transmission, and data transmission can continue based on the current downlink channel information. When the communication quality indicator is greater than the preset threshold, it indicates that the communication quality is poor, that is, the estimated downlink channel information does not meet the current scenario, and the target downlink channel information needs to be re-estimated.

[0121] The communication quality of the data can be measured based on a communication quality indicator, which includes a signal-to-noise ratio and / or a block error rate (BER). The signal-to-noise ratio may be, but is not limited to, a signal-to-interference-plus-noise ratio (SINR). It should be understood that the preset threshold corresponding to the signal-to-noise ratio and the preset threshold corresponding to the block error rate may be set to different values, without specific limitation. The communication quality indicators of the present application are not limited to the above examples.

[0122] The following implementation subject takes a terminal and an access network device as an example to introduce a method for feeding back CSI including a compression rate identifier. Referring to FIG3 , another embodiment of the method for limited CSI feedback in this application includes:

[0123] Step 301: The access network device sends a first downlink measurement pilot and a first compression rate identifier to a terminal respectively.

[0124] Optionally, the first downlink measurement pilot and the first compression ratio identifier are sent on different time domain resources, and the second downlink measurement pilot and the second compression ratio identifier are sent on different time domain resources.

[0125] Step 302: The terminal determines first downlink channel information according to the first downlink measurement pilot.

[0126] Step 303: The terminal inputs the first downlink channel information into the first compression network model according to the first compression rate identifier, and outputs the first codeword through the first compression network model.

[0127] Step 304: The terminal sends the first codeword to the access network device.

[0128] Step 305: The terminal sends the first uplink measurement pilot to the access network device.

[0129] Step 306: The access network device determines first uplink channel information according to the first uplink measurement pilot.

[0130] Step 307: The access network device compares the first codeword and the first uplink channel information to learn network parameter values.

[0131] Step 308: The access network device sends a second downlink measurement pilot and a second compression rate identifier to the terminal.

[0132] Step 309: The terminal determines the second downlink channel information according to the second downlink measurement pilot.

[0133] Step 310: The terminal inputs the second downlink channel information into the second compression network model according to the second compression rate identifier, and outputs a second codeword through the second compression network model.

[0134] Step 311: The access network device receives the second codeword sent by the terminal.

[0135] Step 312: The terminal sends the second uplink measurement pilot to the access network device.

[0136] Step 313: The access network device determines second uplink channel information according to the second uplink measurement pilot.

[0137] Step 314: The access network device inputs the second uplink channel information into the comparison learning network, and learns the network CSI public information through comparison.

[0138] Step 315: The access network device inputs the CSI common information and the second codeword into the super network, and outputs the network parameter adjustment value through the super network.

[0139] Step 316: The access network device adjusts the parameter value of the channel reconstruction network according to the network parameter adjustment amount.

[0140] Step 317: The access network device inputs the CSI public information into the adjusted channel reconstruction network, and outputs the target downlink channel information through the adjusted channel reconstruction network.

[0141] In this application, compression ratios include but are not limited to 1 / 4, 1 / 8, 1 / 32, and 1 / 64. The corresponding relationship between the compression ratio and the compression level can be set according to actual conditions and is not limited in this application.

[0142] The terminal may receive a first downlink measurement pilot and send a first uplink measurement pilot in a first measurement pilot period, and receive a second downlink measurement pilot and send a second uplink measurement pilot in a second measurement pilot period. Since the terminal sends a low-compression rate codeword in the second measurement pilot period, and the second measurement pilot period is greater than the first measurement pilot period, it can send a high-compression rate codeword for a short period of time and a low-compression rate codeword for a long period of time, thereby reducing feedback CSI overhead. In an optional embodiment, the second measurement pilot period may be less than a preset duration, and the first measurement pilot period may be greater than or equal to the preset duration. The second measurement pilot period, the first measurement pilot period, and the preset duration may be set according to actual conditions and are not limited in this application.

[0143] This embodiment has all the technical effects of the embodiment shown in Figure 2, and this embodiment can explicitly instruct the terminal to feedback codewords of different levels of compression rates. The optional embodiments of the embodiment shown in Figure 3 can refer to the optional embodiments of the embodiment shown in Figure 2, and will not be repeated here. It should be noted that the present application can also set any compression rate as the default compression rate. If no compression rate identifier is received, the codeword of the default compression rate is fed back. If a compression rate identifier is received, the codeword corresponding to the compression rate identifier is fed back.

[0144] The following describes the limited CSI feedback method in this application with reference to Figures 4A and 4B. Referring to Figure 4A , the access network device includes a decoding network model, which includes a contrastive learning network, a channel reconstruction network, and a super network (not shown in Figure 4A ). The contrastive learning network includes a first feature extraction network, a second feature public information network, and a public information network.

[0145] In the first time period, after the signal processing unit of the terminal obtains the first downlink measurement pilot, it determines the first downlink channel vector based on the first downlink measurement pilot, and then uses the first compression network model to compress the first downlink channel vector into a first codeword, and sends the first codeword to the first feature extraction network. The codeword features are output through the first feature extraction network. After the signal processing unit of the access network device receives the first uplink measurement pilot sent by the terminal, it determines the first uplink channel vector based on the first uplink measurement pilot, inputs the first uplink channel vector into the second feature extraction network, outputs the uplink channel features through the second feature extraction network, and after inputting the first uplink channel vector into the public information network, the CSI public information of the first time period is output through the public information network based on the codeword features and the uplink channel features. After inputting the first CSI public information into the channel reconstruction network, the downlink channel vector of the first time period is estimated through the channel reconstruction network.

[0146] Referring to Figure 4B , during the second time period, after the terminal's signal processing unit obtains a second downlink measurement pilot, it determines a second downlink channel vector based on the second downlink measurement pilot. It then compresses the second downlink channel vector into a second codeword using a second compression network model. The second codeword is sent to the first feature extraction network, and the first feature extraction network outputs codeword features. After receiving the second uplink measurement pilot sent by the terminal, the access network device's signal processing unit determines a second uplink channel vector based on the second uplink measurement pilot. The second uplink channel vector is input into the second feature extraction network, and the second feature extraction network outputs second uplink channel features. The public information network outputs CSI public information for the second time period based on the codeword features and uplink channel features. The CSI public information and codeword for the second time period are input into the super network, and the super network outputs a network parameter adjustment value. The channel reconstruction network is adjusted based on the network parameter adjustment value, and the channel reconstruction network estimates the downlink channel vector for the second time period based on the CSI public information for the second time period.

[0147] The data flow of Figure 4A is described below in conjunction with the network of Figure 5A. Referring to Figure 5A, in one example, the first feature extraction network includes four network units connected in sequence. The first network unit includes a convolutional layer, a batch normalization (BN) layer, and an activation layer. The convolution kernel of the convolutional layer can be, but is not limited to, a 3*3 kernel, and the activation layer function can be, but is not limited to, a LeakyReLU activation function. The second and third network units are both Markov random field blocks (MRFBs), and the fourth network unit is identical to the first network unit. Referring to Figure 6, in one example, the Markov random field block includes a first network subunit, a second network subunit, a third network subunit, a connection layer, a fourth network subunit, and a fifth network subunit. The first, second, and third network subunits are respectively connected to the input of the connection layer, and the output of the connection layer is connected to the fourth network subunit. The fourth network subunit is then connected to the fifth network subunit.

[0148] In one example, the first, second, and third network subunits all include a convolutional layer, a batch normalization layer, and an activation layer. The convolution kernel of the convolutional layer of the first network subunit is 3*3, the convolution kernel of the convolutional layer of the second network subunit is 7*7, and the convolution kernel of the convolutional layer of the third network subunit is 9*9. The fourth network subunit includes a convolutional layer and a batch normalization layer. The convolution kernel of the convolutional layer of the fourth network subunit is 1*1. The fifth network subunit includes an addition layer and an activation layer. The activation function of the activation layer in the first, second, third, and fifth network subunits can be, but is not limited to, a LeakyReLU activation function.

[0149] The channel reconstruction network includes a first mapping subnet and a second mapping subnet, wherein the first mapping subnet includes a reshape layer, a dense layer, and an activation layer, wherein the activation layer may be, but is not limited to, a ReLU activation function. The second mapping subnet includes a dense layer and a reshape layer.

[0150] After the first codeword is processed by the first network unit, the second network unit, the third network unit and the fourth network unit in sequence, the codeword characteristics are output through the fourth network unit. After the first uplink channel vector is processed by the first network unit, the second network unit, the third network unit and the fourth network unit, the uplink channel characteristics are output through the fourth network unit. The public information network outputs the CSI public information of the first time period based on the codeword characteristics and the uplink channel characteristics. After the CSI public information of the first time period is processed by the first mapping subnet and the second mapping subnet, the downlink channel vector of the first time period can be output through the second mapping subnet.

[0151] The data flow of Figure 4B is described below in conjunction with the network of Figure 5B. Referring to Figure 5B, in one example, a super network includes a super network subnet and a dense layer, and the super network subnet includes a reshaping layer, a dense layer, and an activation layer. The second codeword passes through the first network unit, the second network unit, the third network unit, and the fourth network unit of the first feature extraction subnet, and the fourth network unit of the first feature extraction subnet outputs the codeword feature. The second uplink channel vector passes through the first network unit, the second network unit, the third network unit, and the fourth network unit of the second feature extraction subnet, and the fourth network unit of the second feature extraction subnet outputs the uplink channel feature. The public information network outputs the CSI public information of the second time period based on the codeword feature and the uplink channel feature. The CSI public information of the second time period and the second codeword are processed by the super network subnet and the dense layer to obtain a network parameter adjustment amount. After adjusting the parameter values ​​of the channel reconstruction network according to the network parameter adjustment amount, the adjusted channel reconstruction network outputs the downlink channel vector of the second time period based on the CSI public information of the second time period.

[0152] It should be noted that the network units or subnetworks included in the contrastive learning network, channel reconstruction network, and super network are not limited to the above examples, nor are the network layers included in the network units or subnetworks. The parameter values ​​(such as size) of the network layers are also not limited to the above examples.

[0153] The results of limited feedback CSI are introduced below. Referring to FIG7 , in one example, the base station antenna parameters and carrier parameters are shown in Table 1, and the codeword bit number and achievable rate are shown in FIG7 .

[0154] Table 1

[0155] The training samples and test samples are extracted in the sparse domain after performing a two-dimensional discrete Fourier transform on the downlink channel matrix according to the direction vector of each carrier.

[0156] As can be seen from FIG7 , in this application, when the number of codeword bits is 120, the achievable rate ratio is 97%. When the number of codeword bits is 160, the achievable rate ratio is 97.5%. When the number of codeword bits is 200, the achievable rate ratio is 98%. When the number of codeword bits is 240, the achievable rate ratio is 98.5%.

[0157] In TransNet, when the codeword has 120 bits, the achievable rate ratio is 80%. When the codeword has 160 bits, the achievable rate ratio is 81%. When the codeword has 200 to 240 bits, the achievable rate ratio is 82%. It can be seen that, given the same feedback overhead, the CSI feedback method of this application can significantly increase the achievable rate and improve signal decoding performance.

[0158] The following describes the method for zero CSI feedback in this application. Referring to FIG8 , an embodiment of the method for zero CSI feedback in this application includes:

[0159] Step 801: The access network device sends a downlink measurement pilot to the terminal.

[0160] Step 802: The terminal determines downlink channel information according to the downlink measurement pilot.

[0161] Step 803: The terminal inputs the downlink channel information into a compression network model, and outputs a codeword through the compression network model. The compression network model is used to generate a high-compression-rate codeword from the downlink channel information.

[0162] Step 804: The terminal sends the codeword to the access network device.

[0163] Step 805: The terminal sends the first uplink measurement pilot to the access network device.

[0164] Step 806: The access network device determines first uplink channel information according to the first uplink measurement pilot.

[0165] Step 807: The access network device inputs the codeword and the first uplink channel information into a comparative learning network, and determines the CSI public information through the comparative learning network. For details about the comparative learning network and the CSI public information, please refer to the corresponding description in the embodiment shown in FIG2 .

[0166] Step 808: The terminal sends the second uplink measurement pilot to the access network device.

[0167] Step 809: The access network device determines second uplink channel information according to the second uplink measurement pilot.

[0168] Step 810: The access network device inputs the second uplink channel information into the super network, and outputs the network parameter adjustment value through the super network.

[0169] Step 811: The access network device adjusts the parameter value of the channel reconstruction network according to the network parameter adjustment amount.

[0170] Step 812: The access network device inputs the CSI public information into the channel reconstruction network after parameter adjustment, and outputs the target downlink channel information through the channel reconstruction network after parameter adjustment.

[0171] The terminal may receive a downlink measurement pilot and send a first uplink measurement pilot in a first measurement pilot period, and send a second uplink measurement pilot in a second measurement pilot period. Since the terminal does not send codewords in the second measurement pilot period, and the second measurement pilot period is longer than the first measurement pilot period, it can send high-compression codewords in a short period of time and not send codewords for a long period of time, thereby reducing feedback CSI overhead. In an optional embodiment, the second measurement pilot period may be less than a preset duration, and the first measurement pilot period may be greater than or equal to the preset duration. The second measurement pilot period, the first measurement pilot period, and the preset duration may be set according to actual circumstances and are not limited in this application.

[0172] In this embodiment, the access network device determines the CSI common information for the first period based on the high-compression codeword and uplink channel information, adjusts the channel reconstruction network based on the uplink channel vector, and uses this channel reconstruction network to output target downlink channel information, which can be considered downlink channel CSI. In the second period, downlink channel CSI can be estimated without feeding back codewords, thereby reducing the overhead of CSI feedback.

[0173] Secondly, the super network can determine the network parameter adjustment amount of the current scenario based on the uplink channel information, and can adjust the channel reconstruction network in the decoding network model based on the network parameter adjustment amount of the current scenario, so that the decoding network model can reconstruct the downlink channel information of different scenarios.

[0174] The following describes the zero feedback CSI method in the present application in conjunction with Figures 9A and 9B. Referring to Figure 9A, in the first time period, after the signal processing unit of the terminal obtains the downlink measurement pilot, it determines the downlink channel vector based on the downlink measurement pilot, then compresses the downlink channel vector into a codeword using a compression network model, and sends the codeword to the feature extraction network of the access network device. After receiving the first uplink measurement pilot sent by the terminal, the signal processing unit of the access network device determines the first uplink channel vector based on the first uplink measurement pilot, inputs the codeword into the first feature extraction network, and inputs the first uplink channel vector into the second feature extraction network. The first feature extraction network outputs the codeword features and the uplink channel features output by the second feature extraction network. The public information network determines and saves the CSI public information for the first time period based on the codeword features and the uplink channel features.

[0175] Referring to Figure 9B, in the second time period, the signal processing unit of the access network device receives the second uplink measurement pilot sent by the terminal, determines a second uplink channel vector based on the second uplink measurement pilot, inputs the second uplink channel vector into the super network, outputs a network parameter adjustment amount through the super network, adjusts the channel reconstruction network based on the network parameter adjustment amount, and outputs a downlink channel vector for the second time period based on the CSI public information of the first time period.

[0176] The first feature extraction network, the second feature extraction network, and the channel reconstruction network of FIG10A can be referred to FIG5A , and the first feature extraction network, the second feature extraction network, the channel reconstruction network, and the super network of FIG10B can be referred to FIG5B .

[0177] The following introduces the data passing through the neural network model in Figure 9A in conjunction with the neural network model of Figure 10A. Referring to Figure 10A, in one example, the codeword is processed by the first network unit, the second network unit, the third network unit and the fourth network unit of the first feature extraction network respectively, and the codeword feature is output through the fourth network unit. The first uplink channel vector is processed by the first network unit, the second network unit, the third network unit and the fourth network unit of the second feature extraction network, and the uplink channel feature is output through the fourth network unit. The public information network outputs the CSI public information of the first time period based on the codeword feature and the uplink channel feature.

[0178] The data processed by the neural network model in Figure 9B will be described below in conjunction with the neural network model in Figure 10B. Referring to Figure 10B , in one example, the second uplink channel vector is processed by the super subnet and dense layer to obtain a network parameter adjustment. After adjusting the channel reconstruction network based on the network parameter adjustment, the adjusted channel reconstruction network outputs a downlink channel vector for the second time period based on the CSI public information from the first time period.

[0179] In this application, the uplink and downlink frequency domain interval is the interval between the uplink frequency band of FDD and the downlink frequency band of FDD. The results of zero feedback CSI are introduced below. Refer to Figure 11. In an example, when the uplink and downlink frequency domain interval is 120M, the reachable rate ratio of this application is 98%, the reachable rate ratio of baseline-Z1 is 90%, and the reachable rate ratio of Up2Down is 38%. When the uplink and downlink frequency domain interval is 360M, the reachable rate ratio of this application is 98%, the reachable rate ratio of baseline-Z1 is 89%, and the reachable rate ratio of Up2Down is 29%. When the uplink and downlink frequency domain interval is 720M, the reachable rate ratio of this application is 98%, the reachable rate ratio of baseline-Z1 is 89%, and the reachable rate ratio of Up2Down is 25%. It can be seen from this that when the uplink and downlink frequency domain interval is the same, the reachable rate ratio of this application is higher than the other two comparison schemes, which improves the signal decoding performance.

[0180] In the above embodiments, the super network is deployed on the access network device. The following describes a method for feeding back CSI that does not include a compression rate indicator when the super network is deployed on the terminal side. Referring to Figure 12, another embodiment of the method for limited CSI feedback in this application includes:

[0181] Step 1201: The access network device sends a first downlink measurement pilot to the terminal.

[0182] Step 1202: The terminal determines first downlink channel information according to the first downlink measurement pilot.

[0183] Step 1203: The terminal inputs the first downlink channel information into the first compression network model, and outputs the first codeword through the first compression network model.

[0184] Step 1204: The terminal sends the first codeword to the access network device.

[0185] Step 1205: The terminal sends the first uplink measurement pilot to the access network device.

[0186] Step 1206: The access network device determines first uplink channel information according to the first uplink measurement pilot.

[0187] Step 1207: The access network device updates the parameter values ​​of the comparative learning network according to the first codeword and the first uplink channel information.

[0188] Step 1208: The access network device sends a second downlink measurement pilot to the terminal.

[0189] Step 1209: The terminal determines second downlink channel information according to the second downlink measurement pilot.

[0190] Step 1210: The terminal inputs the second downlink channel information into the second compression network model, and outputs a second codeword through the second compression network model.

[0191] The condition for triggering step 1203 or step 1210 may be a preset period. For example, steps 1201 to 1203 are performed during the first measurement pilot period, and steps 1208 to 1210 are performed during the second measurement pilot period. The compression ratio of the second codeword is lower than the compression ratio of the first codeword. The first compression network model is configured to output codewords with a high compression ratio, and the second compression network model is configured to output codewords with a low compression ratio.

[0192] Step 1211: The access network device receives the second codeword sent by the terminal.

[0193] Step 1212: The terminal sends the second uplink measurement pilot to the access network device.

[0194] Step 1213: The access network device determines second uplink channel information according to the second uplink measurement pilot.

[0195] Step 1214: The access network device inputs the second codeword and the second uplink channel information into the comparative learning network, and outputs the CSI public information through the comparative learning network.

[0196] Step 1215: The access network device sends the CSI public information to the terminal.

[0197] Step 1216: The terminal inputs the CSI common information and the second codeword into the super network, and outputs the network parameter adjustment value through the super network.

[0198] It should be noted that before step 1208, the access network device may send network parameter values ​​of the super network to the terminal, and the terminal may configure or update the super network based on the network parameter values ​​of the super network. The network parameters of the super network include one or more of network structure data, neuron weights, bias, and activation function.

[0199] Step 1217: The terminal sends the network parameter adjustment value to the access network device.

[0200] Step 1218: The access network device adjusts the parameter value of the channel reconstruction network according to the network parameter adjustment amount.

[0201] Step 1219: The access network device inputs the CSI public information into the channel reconstruction network after parameter adjustment, and outputs the target downlink channel information through the channel reconstruction network after parameter adjustment.

[0202] In this embodiment, when the super network is deployed at the terminal, the terminal can feed back low-compression codewords and network parameter adjustment amounts. The access network device can estimate the downlink channel CSI of the second time period based on the low-compression codewords and network parameter adjustment amounts, thereby reducing the overhead of feeding back CSI.

[0203] Secondly, the super network can extract the network parameter adjustment amount of the current scenario based on the CSI public information and low-compression codewords, and adjust the channel reconstruction network in the decoding network model based on the network parameter adjustment amount of the current scenario, so that the decoding network model can reconstruct the downlink channel information of different scenarios.

[0204] Again, deploying the super network at the terminal can reduce the computing overhead of the base station.

[0205] The following describes a method for feeding back CSI including a compression rate indicator in a terminal in conjunction with deployment of a super network. Referring to FIG13 , another embodiment of the method for limited CSI feedback in this application includes:

[0206] Step 1301: The access network device sends a first downlink measurement pilot and a first compression rate identifier to a terminal respectively.

[0207] Step 1302: The terminal determines first downlink channel information according to the first downlink measurement pilot.

[0208] Step 1303: The terminal inputs the first downlink channel information into the first compression network model according to the first compression rate identifier, and outputs the first codeword through the first compression network model.

[0209] Step 1304: The terminal sends the first codeword to the access network device.

[0210] Step 1305: The terminal sends the first uplink measurement pilot to the access network device.

[0211] Step 1306: The access network device determines first uplink channel information according to the first uplink measurement pilot.

[0212] Step 1307: The access network device updates the parameter values ​​of the comparative learning network according to the first codeword and the first uplink channel information.

[0213] Step 1308: The access network device sends a second downlink measurement pilot and a second compression rate identifier to the terminal.

[0214] Step 1309: The terminal determines the second downlink channel information according to the second downlink measurement pilot.

[0215] Step 1310: The terminal inputs the second downlink channel information into the second compression network model according to the second compression rate identifier, and outputs a second codeword through the second compression network model.

[0216] Step 1311: The access network device receives the second codeword sent by the terminal.

[0217] Step 1312: The terminal sends the second uplink measurement pilot to the access network device.

[0218] Step 1313: The access network device determines second uplink channel information according to the second uplink measurement pilot.

[0219] Step 1314: The access network device inputs the second codeword and the second uplink channel information into a comparative learning network, and outputs CSI public information through the comparative learning network.

[0220] Step 1315: The access network device sends the CSI public information to the terminal.

[0221] Step 1316: The terminal inputs the CSI common information and the second codeword into the super network, and outputs the network parameter adjustment value through the super network.

[0222] Step 1317: The terminal sends the network parameter adjustment value to the access network device.

[0223] Step 1318: The access network device adjusts the parameter value of the channel reconstruction network according to the network parameter adjustment amount.

[0224] Step 1319: The access network device inputs the CSI public information into the channel reconstruction network after parameter adjustment, and outputs the target downlink channel information through the channel reconstruction network after parameter adjustment.

[0225] In this embodiment, when the super network is deployed at the terminal, the terminal can feed back low-compression codewords and network parameter adjustment amounts. The access network device can estimate the downlink channel CSI of the second time period based on the low-compression codewords and network parameter adjustment amounts, thereby reducing the overhead of feeding back CSI.

[0226] Secondly, the super network can extract the network parameter adjustment amount of the current scenario based on the CSI public information and low-compression codewords, and adjust the channel reconstruction network in the decoding network model based on the network parameter adjustment amount of the current scenario, so that the decoding network model can reconstruct the downlink channel information of different scenarios.

[0227] Again, deploying the super network at the terminal can reduce the computing overhead of the base station.

[0228] In addition, this embodiment provides a method for explicitly instructing a terminal to feed back codewords with different levels of compression rates.

[0229] It is understood that, in order to implement the functions described in the above embodiments, the access network equipment and terminals include hardware structures and / or software modules corresponding to the respective functions. Those skilled in the art will readily appreciate that, in conjunction with the various exemplary units and method steps described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is implemented in hardware or in a hardware-driven manner by computer software depends on the specific application scenario and design constraints of the technical solution.

[0230] Figures 14 and 15 are schematic diagrams of the structures of possible communication devices provided in embodiments of the present application. These communication devices can be used to implement the functions of the terminal or access network device in the above-mentioned method embodiments, thereby also achieving the beneficial effects of the above-mentioned method embodiments. In the embodiments of the present application, the communication device can be one of the terminals 120a-120j shown in Figure 1, or it can be the base station 110a or 110b shown in Figure 1, or it can be a module (such as a chip) applied to the terminal or base station.

[0231] As shown in Figure 14, the communication device 1400 includes a processing unit 1410 and a transceiver unit 1420. The communication device 1400 is used to implement the functions of the terminal or base station in the method embodiments shown in Figures 2, 3, 8, 12 or 13 above.

[0232] When the communication device 1400 is used to implement the function of the terminal in the method embodiment shown in Figure 2, the processing unit 1410 is used to execute steps 202, 203, 209 and 210, and the transceiver unit 1420 is used to execute steps 201, 204, 205, 208, 211 and 212.

[0233] When the communication device 1400 is used to implement the functions of the access network device in the method embodiment shown in FIG2 , the transceiver unit 1420 is used to execute steps 201, 204, 205, 208, 211, and 212, and the processing unit 1410 is used to execute steps 206, 207, and 213 to 217. For a more detailed description of the processing unit 1410 and the transceiver unit 1420, please refer to the relevant description of the method embodiment shown in FIG2 .

[0234] When the communication device 1400 is used to implement the function of the terminal in the method embodiment shown in Figure 3, the processing unit 1410 is used to execute steps 302, 303, 309 and 310, and the transceiver unit 1420 is used to execute steps 301, 304, 305, 308, 311 and 312.

[0235] When the communication device 1400 is used to implement the functions of the access network device in the method embodiment shown in FIG3 , the transceiver unit 1420 is used to execute steps 301, 309, 312, and 319, and the processing unit 1410 is used to execute steps 306, 307, 308, and 313 to 317. For a more detailed description of the processing unit 1410 and the transceiver unit 1420, please refer to the relevant description of the method embodiment shown in FIG3 .

[0236] When the communication device 1400 is used to implement the terminal function in the method embodiment shown in Figure 8, the processing unit 1410 is used to execute steps 802 and 803, and the transceiver unit 1420 is used to execute steps 801, 804, 805 and 808.

[0237] When the communication device 1400 is used to implement the functions of the access network device in the method embodiment shown in FIG8 , the transceiver unit 1420 is used to execute steps 801, 804, 805, and 808, and the processing unit 1410 is used to execute steps 806, 807, and 809 to 812. For a more detailed description of the processing unit 1410 and the transceiver unit 1420, please refer to the relevant description of the method embodiment shown in FIG8 .

[0238] When the communication device 1400 is used to implement the function of the terminal in the method embodiment shown in Figure 12, the processing unit 1410 is used to execute steps 1202, 1203, 1209, 1210 and 1216, and the transceiver unit 1420 is used to execute steps 1201, 1204, 1205, 1208, 1211, 1212, 1215 and 1217.

[0239] When the communication device 1400 is used to implement the functions of the access network device in the method embodiment shown in FIG12 , the transceiver unit 1420 is used to execute steps 1201, 1204, 1205, 1208, 1211, 1212, 1215, and 1217, and the processing unit 1410 is used to execute steps 1206, 1207, 1213, 1214, 1218, and 1219. For a more detailed description of the processing unit 1410 and the transceiver unit 1420, please refer to the relevant description of the method embodiment shown in FIG12 .

[0240] When the communication device 1400 is used to implement the function of the terminal in the method embodiment shown in Figure 13, the processing unit 1410 is used to execute steps 1302, 1303, 1309, 1310 and 1316, and the transceiver unit 1420 is used to execute steps 1301, 1304, 1305, 1308, 1311, 1312, 1315 and 1317.

[0241] When the communication device 1400 is used to implement the functions of the access network device in the method embodiment shown in FIG13 , the transceiver unit 1420 is used to execute steps 1301, 1304, 1305, 1308, 1311, 1312, 1315, and 1317, and the processing unit 1410 is used to execute steps 1306, 1307, 1313, 1314, 1318, and 1319. For a more detailed description of the processing unit 1410 and the transceiver unit 1420, please refer to the relevant description of the method embodiment shown in FIG13 .

[0242] The present application provides a wireless communication system, comprising a plurality of communication devices 1400, one or more of which are configured to implement terminal functions, and one or more other communication devices are configured to implement access network device functions. The communication devices configured to implement terminal functions are deployed with a compression network model, while the communication devices configured to implement access network device functions are deployed with a decoding network model. The compression network model and the decoding network model can form an end-to-end network.

[0243] As shown in Figure 15, communication device 1500 includes a processor 1510 and an interface circuit 1520. Processor 1510 and interface circuit 1520 are coupled to each other. It is understood that interface circuit 1520 can be a transceiver or an input / output interface. Optionally, communication device 1500 may also include a memory 1530 for storing instructions executed by processor 1510, input data required by processor 1510 to execute instructions, or data generated by processor 1510 after executing instructions.

[0244] When the communication device 1500 is used to implement the method shown in Figure 2, Figure 3, Figure 8, Figure 12 or Figure 13, the processor 1510 is used to implement the functions of the above-mentioned processing unit 1410, and the interface circuit 1520 is used to implement the functions of the above-mentioned transceiver unit 1420.

[0245] When the communication device is a chip used in a terminal, the terminal chip implements the functions of the terminal in the above method embodiments. The terminal chip receives information from other modules in the terminal (such as a radio frequency module or antenna), and the information is sent by the base station to the terminal; or the terminal chip sends information to other modules in the terminal (such as a radio frequency module or antenna), and the information is sent by the terminal to the base station.

[0246] When the above-mentioned communication device is a module applied to a base station, the base station module implements the functions of the access network device in the above-mentioned method embodiment. The base station module receives information from other modules in the base station (such as a radio frequency module or an antenna), and the information is sent by the terminal to the base station; or the base station module sends information to other modules in the base station (such as a radio frequency module or an antenna), and the information is sent by the base station to the terminal. The base station module here can be the baseband chip of the base station, or it can be a DU or other module. The DU here can be a DU under the open radio access network (O-RAN) architecture.

[0247] It is understood that the processor in the embodiments of the present application may be a central processing unit (CPU), or may be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. The general-purpose processor may be a microprocessor or any conventional processor.

[0248] The method steps in the embodiments of the present application can be implemented in hardware or in software instructions that can be executed by a processor. The software instructions can be composed of corresponding software modules, and the software modules can be stored in random access memory, flash memory, read-only memory, programmable read-only memory, erasable programmable read-only memory, electrically erasable programmable read-only memory, registers, hard disk, mobile hard disk, CD-ROM or any other form of storage medium well known in the art. An exemplary storage medium is coupled to the processor so that the processor can read information from the storage medium and write information to the storage medium. The storage medium can also be an integral part of the processor. The processor and storage medium can be located in an ASIC. In addition, the ASIC can be located in a base station or a terminal. The processor and storage medium can also exist in a base station or a terminal as discrete components.

[0249] In the above embodiments, all or part of the embodiments may be implemented using software, hardware, firmware, or any combination thereof. When implemented using software, all or part of the embodiments may be implemented in the form of a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are performed in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, a network device, a user device, or other programmable device. The computer program or instructions may be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program or instructions may be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; an optical medium, such as a digital video disk; or a semiconductor medium, such as a solid-state drive. The computer-readable storage medium may be a volatile or nonvolatile storage medium, or may include both volatile and nonvolatile types of storage media.

[0250] In the various embodiments of the present application, unless otherwise specified or there is a logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced by each other. The technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.

[0251] In this application, "at least one" means one or more, and "more" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone, where A and B can be singular or plural. In the text description of this application, the character " / " generally indicates that the previous and next associated objects are in an "or" relationship; in the formula of this application, the character " / " indicates that the previous and next associated objects are in a "division" relationship. "Including at least one of A, B and C" can mean: including A; including B; including C; including A and B; including A and C; including B and C; including A, B and C.

[0252] It is understood that the various numbers used in the embodiments of this application are merely for ease of description and are not intended to limit the scope of the embodiments of this application. The order of the sequence numbers of the above-mentioned processes does not necessarily imply a specific order of execution; the order of execution of the processes should be determined by their functions and inherent logic.

Claims

1. A method for feeding back channel state information CSI, characterized in that: include: Sending a first downlink measurement pilot to the terminal; receiving a first codeword sent by the terminal according to the first downlink measurement pilot; Receiving a first uplink measurement pilot sent by the terminal; Determine first uplink channel information according to the first uplink measurement pilot; updating a parameter value of a comparative learning network according to the first codeword and the first uplink channel information; Sending a second downlink measurement pilot to the terminal; receiving a second codeword sent by the terminal according to the second downlink measurement pilot, where a compression rate corresponding to the second codeword is less than a compression rate corresponding to the first codeword; Receiving a second uplink measurement pilot sent by the terminal; determining second uplink channel information according to the second uplink measurement pilot; Inputting the second codeword and the second uplink channel information into the contrastive learning network, and outputting CSI public information through the contrastive learning network; Inputting the CSI public information and the second codeword into a super network, and outputting a network parameter adjustment amount through the super network; Adjusting a parameter value of the channel reconstruction network according to the network parameter adjustment amount; The CSI public information is input into the channel reconstruction network after parameter adjustment, and the target downlink channel information is output through the channel reconstruction network after parameter adjustment.

2. The method according to claim 1, characterized in that The sending the first downlink measurement pilot to the terminal includes: sending the first downlink measurement pilot to the terminal in a first measurement pilot period; The sending the second downlink measurement pilot to the terminal includes: sending the second downlink measurement pilot to the terminal in a second measurement pilot period, where the second measurement pilot period is greater than the first measurement pilot period.

3. The method according to claim 1, characterized in that The first downlink measurement pilot and the second downlink measurement pilot satisfy at least one of the following conditions: a measurement pilot bandwidth of the first downlink measurement pilot is greater than a measurement pilot bandwidth of the second downlink measurement pilot, a measurement pilot density of the first downlink measurement pilot is greater than a measurement pilot density of the second downlink measurement pilot, and a number of measurement pilot ports of the first downlink measurement pilot is greater than a number of measurement pilot ports of the second downlink measurement pilot.

4. The method according to claim 1, characterized in that The method further comprises: After sending the first downlink measurement pilot to the terminal, sending a first compression rate identifier to the terminal, where the first compression rate identifier corresponds to the compression rate of the first codeword; After sending the second downlink measurement pilot to the terminal: sending a second compression rate identifier to the terminal, where the second compression rate identifier corresponds to the compression rate of the second codeword.

5. The method according to claim 1, characterized in that The contrastive learning network includes a first feature extraction network, a second feature extraction network and a public information network, the first feature extraction network is used to extract codeword features, the second feature extraction network is used to extract uplink channel features, the public information network is used to determine CSI public information based on codeword features and uplink channel features, and the first feature extraction network and the second feature extraction network are the same.

6. The method according to claim 1, characterized in that The network parameter adjustment amount includes at least one of network structure data, neuron weights, bias and activation function.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: precoding the data according to target downlink channel information; Sending precoded data to the terminal; Measuring a communication quality indicator according to the data, the communication quality indicator comprising at least one of a signal-to-noise ratio and a block error rate; When the communication quality indicator is greater than a preset threshold, the step of sending a first downlink measurement pilot to the terminal is triggered.

8. A method for feeding back channel state information CSI, characterized in that: include: Receiving a first downlink measurement pilot sent by an access network device; determining first downlink channel information according to the first downlink measurement pilot; Inputting the first downlink channel information into a first compression network model, and outputting a first codeword through the first compression network model; Sending the first codeword to the access network device; Sending a first uplink measurement pilot to the access network device; Receiving a second downlink measurement pilot sent by the access network device; determining second downlink channel information according to the second downlink measurement pilot; Inputting the second downlink channel information into a second compression network model, and outputting a second codeword through the second compression network model, wherein a compression rate of the second codeword is less than a compression rate of the first codeword; Sending the second codeword to the access network device; Sending a second uplink measurement pilot to the access network device.

9. The method according to claim 8, characterized in that The receiving a first downlink measurement pilot sent by the access network device comprises: receiving a first downlink measurement pilot sent by the access network device in a first measurement pilot period; The receiving a second downlink measurement pilot sent by the access network device includes: receiving a second downlink measurement pilot sent by the access network device in a second measurement pilot period, where the second measurement pilot period is greater than the first measurement pilot period.

10. The method according to claim 8 or 9, characterized in that: The method further includes: receiving a first compression rate identifier sent by the access network device; The inputting the first downlink channel information into the first compression network model comprises: inputting the first downlink channel information into the first compression network model according to the first compression rate identifier; The method further includes: receiving a second compression rate identifier sent by the access network device; The inputting the second downlink channel information into the second compression network model includes: inputting the second downlink channel information into the second compression network model according to the second compression rate identifier.

11. A method for feeding back channel state information CSI, characterized in that: include: Sending a first downlink measurement pilot to the terminal; receiving a first codeword sent by the terminal according to the first downlink measurement pilot; Receiving a first uplink measurement pilot sent by the terminal; Determine first uplink channel information according to the first uplink measurement pilot; updating a parameter value of a comparative learning network according to the first codeword and the first uplink channel information; Sending a second downlink measurement pilot to the terminal; receiving a second codeword sent by the terminal according to the second downlink measurement pilot, where a compression rate corresponding to the second codeword is less than a compression rate corresponding to the first codeword; Receiving a second uplink measurement pilot sent by the terminal; determining second uplink channel information according to the second uplink measurement pilot; Inputting the second codeword and the second uplink channel information into the contrastive learning network, and outputting CSI public information through the contrastive learning network; Send CSI public information to the terminal; A network parameter adjustment amount sent by a receiving terminal; Adjusting a parameter value of the channel reconstruction network according to the network parameter adjustment amount; The CSI public information is input into the channel reconstruction network after parameter adjustment, and the target downlink channel information is output through the channel reconstruction network after parameter adjustment.

12. The method according to claim 11, characterized in that The sending the first downlink measurement pilot to the terminal includes: sending the first downlink measurement pilot to the terminal in a first measurement pilot period; The sending the second downlink measurement pilot to the terminal includes: sending the second downlink measurement pilot to the terminal in a second measurement pilot period; The second measurement pilot period is greater than the first measurement pilot period.

13. The method according to claim 11, characterized in that The method further comprises: After sending the first downlink measurement pilot to the terminal, sending a first compression rate identifier to the terminal, the first compression rate identifier corresponding to a compression rate of the first codeword; After sending the second downlink measurement pilot to the terminal: sending a second compression rate identifier to the terminal, where the second compression rate identifier corresponds to the compression rate of the second codeword.

14. The method according to any one of claims 11 to 13, characterized in that The method further comprises: precoding the data according to target downlink channel information; Sending precoded data to the terminal; Measuring a communication quality indicator according to the data, the communication quality indicator comprising at least one of a signal-to-noise ratio and a block error rate; When the communication quality indicator is greater than a preset threshold, the step of sending a first downlink measurement pilot to the terminal is triggered.

15. A method for feeding back channel state information CSI, characterized in that: include: Receiving a first downlink measurement pilot sent by an access network device; determining first downlink channel information according to the first downlink measurement pilot; Inputting the first downlink channel information into a first compression network model, and outputting a first codeword through the first compression network model; Sending the first codeword to the access network device; Sending a first uplink measurement pilot to the access network device; Receiving a second downlink measurement pilot sent by the access network device; determining second downlink channel information according to the second downlink measurement pilot; Inputting the second downlink channel information into a second compression network model, and outputting a second codeword through the second compression network model, wherein a compression rate of the second codeword is less than a compression rate of the first codeword; Sending the second codeword to the access network device; Sending a second uplink measurement pilot to the access network device; receiving CSI common information sent by the access network device, where the CSI common information is determined by the access network device according to the second codeword and the second uplink measurement pilot; Inputting the CSI public information into a super network, and outputting a network parameter adjustment amount through the super network; The network parameter adjustment amount is sent to the access network device.

16. The method according to claim 15, characterized in that The receiving a first downlink measurement pilot sent by the access network device comprises: receiving a first downlink measurement pilot sent by the access network device in a first measurement pilot period; The receiving a second downlink measurement pilot sent by the access network device comprises: receiving a second downlink measurement pilot sent by the access network device in a second measurement pilot period; The second measurement pilot period is greater than the first measurement pilot period.

17. The method according to claim 15, characterized in that The method further includes: receiving a first compression rate identifier sent by the access network device; The inputting the first downlink channel information into the first compression network model comprises: inputting the first downlink channel information into the first compression network model according to the first compression rate identifier; The method further includes: receiving a second compression rate identifier sent by the access network device; The inputting the second downlink channel information into the second compression network model includes: inputting the second downlink channel information into the second compression network model according to the second compression rate identifier.

18. The method according to any one of claims 15 to 17, characterized in that Before the receiving the first downlink measurement pilot sent by the access network device, the method further includes: Receiving a network parameter value of a super network sent by the access network device; The super network is configured according to the network parameter values ​​of the super network.

19. A method for feeding back channel state information CSI, characterized in that: include: Sending a downlink measurement pilot to the terminal; receiving a codeword sent by the terminal; Receiving a first uplink measurement pilot sent by the terminal; Determine first uplink channel information according to the first uplink measurement pilot; Inputting the codeword and the first uplink channel information into a contrastive learning network, and determining CSI common information through the contrastive learning network; Receiving a second uplink measurement pilot sent by the terminal; determining second uplink channel information according to the second uplink measurement pilot; inputting the second uplink channel information into a super network, and outputting a network parameter adjustment amount through the super network; Adjusting a parameter value of a channel reconstruction network according to the network parameter adjustment amount, wherein the channel reconstruction network is respectively connected to the contrastive learning network and the super network; The CSI public information is input into the channel reconstruction network after parameter adjustment, and the target downlink channel information is output through the channel reconstruction network after parameter adjustment.

20. The method according to claim 19, characterized in that The receiving the first uplink measurement pilot sent by the terminal includes: receiving the first uplink measurement pilot sent by the terminal in a first measurement pilot period; The receiving the second uplink measurement pilot sent by the terminal includes: receiving a second uplink measurement pilot sent by the terminal in a second measurement pilot period; The first measurement pilot period is smaller than the second measurement pilot period.

21. The method according to claim 19 or 20, characterized in that The method further comprises: precoding the data according to the target downlink channel information; Sending precoded data to the terminal; Acquire a communication quality indicator according to the data, the communication quality indicator comprising at least one of a signal-to-noise ratio and a block error rate; When the communication quality indicator is greater than a preset threshold, the step of sending a downlink measurement pilot to the terminal is triggered.

22. A method for feeding back channel state information CSI, characterized in that: include: Receiving a downlink measurement pilot sent by an access network device; Determine downlink channel information according to the downlink measurement pilot; Inputting the downlink channel information into a compression network model, and outputting a codeword through the compression network model; Sending the codeword to the access network device; Sending a first uplink measurement pilot to the access network device; Sending a second uplink measurement pilot to the access network device.

23. The method according to claim 22, characterized in that The sending the first uplink measurement pilot to the access network device comprises: sending the first uplink measurement pilot to the access network device in a first measurement pilot period; The sending the second uplink measurement pilot to the access network device comprises: sending the second uplink measurement pilot to the access network device in a second measurement pilot period; The first measurement pilot period is smaller than the second measurement pilot period.

24. A communication device comprising means for executing the method of any one of claims 1 to 7, 11 to 14, 19 to 21.

25. A communication device comprising means for executing the method of any one of claims 8 to 10, 15 to 18, 22 to 23.

26. A communication device, characterized in that: It includes a processor and an interface circuit, wherein the interface circuit is used to receive signals from other communication devices and transmit them to the processor or send signals from the processor to other communication devices, and the processor is used to implement the method as described in any one of claims 1 to 7, 11 to 14, and 19 to 21 through a logic circuit or executing code instructions.

27. A communication device, characterized in that: It includes a processor and an interface circuit, wherein the interface circuit is used to receive signals from other communication devices and transmit them to the processor or send signals from the processor to other communication devices, and the processor is used to implement the method as described in any one of claims 8 to 10, 15 to 18, and 22 to 23 through a logic circuit or executing code instructions.

28. A computer-readable storage medium, characterized in that: The storage medium stores a computer program or an instruction. When the computer program or the instruction is executed by the communication device, the method as claimed in any one of claims 1 to 7, 11 to 14, and 19 to 21 is implemented.

29. A computer-readable storage medium, characterized in that: The storage medium stores a computer program or instruction. When the computer program or instruction is executed by the communication device, the method as claimed in any one of claims 8 to 10, 15 to 18, and 22 to 23 is implemented.

Images