CN117751559A

CN117751559A - CSI feedback method, device, equipment and storage medium

Info

Publication number: CN117751559A
Application number: CN202180101282.2A
Authority: CN
Inventors: 刘文东; 田文强; 肖寒
Original assignee: Guangdong Oppo Mobile Telecommunications Corp Ltd
Current assignee: Guangdong Oppo Mobile Telecommunications Corp Ltd
Priority date: 2021-11-02
Filing date: 2021-11-02
Publication date: 2024-03-22
Also published as: WO2023077274A1

Abstract

A CSI feedback method, a device, equipment and a storage medium relate to the field of mobile communication. The method comprises the following steps: encoding the CSI by using an encoder (12) to obtain CSI feedback information; the encoder (12) is trained from a real training set and a first supplemental training set, the first supplemental training set being generated by a generator in an countermeasure generation network, the countermeasure generation network being trained based on the real training set; the CSI feedback information is sent to the access network device (20).

Description

CSI feedback method, device, equipment and storage medium

Technical Field

The present invention relates to the field of mobile communications, and in particular, to a method, an apparatus, a device, and a storage medium for feeding back channel state information (Channel State Information, CSI).

Background

In the current New Radio (NR) system, for the CSI feedback scheme, a terminal generally adopts codebook-based eigenvector feedback, so that a base station obtains CSI of a downlink channel. Specifically, the base station sends a downlink CSI reference signal (CSI Reference Signals, CSI-RS) to the user, the terminal estimates CSI of the downlink channel by using the CSI-RS, and performs eigenvalue decomposition on the estimated downlink channel to obtain an eigenvector corresponding to the downlink channel. Further, NR provides two codebook designs of Type 1 and Type 2, where Type 1 codebook is used for CSI feedback with conventional accuracy and transmission of Single-User MIMO (SU-MIMO) and Multi-User MIMO (MU-MIMO), and Type 2 codebook is used to improve transmission performance of MU-MIMO.

Disclosure of Invention

The embodiment of the application provides a CSI feedback method, a device, equipment and a storage medium, and provides a CSI feedback scheme based on a countermeasure generation network. The technical scheme is as follows.

According to an aspect of the present application, there is provided a CSI feedback method applied to a terminal, the method including:

encoding the CSI by using an encoder to obtain CSI feedback information; the encoder is trained by a real training set and a first supplemental training set, the first supplemental training set is generated by a generator in an countermeasure generation network, and the countermeasure generation network is trained based on the real training set;

and sending the CSI feedback information to access network equipment.

According to an aspect of the present application, there is provided a CSI feedback method applied to an access network device, where the method includes:

receiving CSI feedback information sent by a terminal, wherein the CSI feedback information is obtained by encoding the CSI by the terminal through an encoder;

decoding the CSI feedback information by using a decoder to obtain the CSI measured by the terminal; the encoder and the decoder are trained from a real training set and a first supplemental training set, the first supplemental training set being generated by a generator in an countermeasure generation network, the countermeasure generation network being trained based on the real training set;

And sending the CSI feedback information to access network equipment.

According to an aspect of the present application, there is provided a CSI feedback apparatus, the apparatus comprising:

the coding module is used for coding the CSI by using an encoder to obtain CSI feedback information; the encoder is trained by a real training set and a first supplemental training set, the first supplemental training set is generated by a generator in an countermeasure generation network, and the countermeasure generation network is trained based on the real training set;

and the sending module is used for sending the CSI feedback information to the access network equipment.

the receiving module is used for receiving the CSI feedback information sent by the terminal, wherein the CSI feedback information is obtained by encoding the CSI by the terminal through an encoder;

the decoding module is used for decoding the CSI feedback information by using a decoder to obtain the CSI measured by the terminal; the encoder and the decoder are trained from a real training set and a first supplemental training set, the first supplemental training set being generated by a generator in an countermeasure generation network, the countermeasure generation network being trained based on the real training set.

According to one aspect of the present application, there is provided a terminal comprising: a processor; a transceiver coupled to the processor; a memory for storing executable instructions of the processor; wherein the processor is configured to load and execute the executable instructions to implement the CSI feedback method as described in the above aspect.

According to one aspect of the present application, there is provided a network device comprising: a processor; a transceiver coupled to the processor; a memory for storing executable instructions of the processor; wherein the processor is configured to load and execute the executable instructions to implement the CSI feedback method as described in the above aspect.

According to an aspect of the present application, there is provided a computer readable storage medium having stored therein executable instructions that are loaded and executed by the processor to implement the CSI feedback method as described in the above aspect.

According to an aspect of the present application, there is provided a computer program product or computer program comprising computer instructions stored in a computer readable storage medium, the computer instructions being read from the computer readable storage medium by a processor of a computer device, the computer instructions being executed by the processor to cause the computer device to perform the CSI feedback method of the above aspect.

According to an aspect of the present application, there is provided a chip including a programmable logic circuit or a program, the chip being configured to implement the CSI feedback method as described in the above aspect.

The technical scheme provided by the embodiment of the application at least comprises the following beneficial effects:

under the condition that training samples in a real training set are fewer, a complementary training set is generated by utilizing an countermeasure generation network, so that an encoder and a decoder with excellent performance are obtained through training, the encoder and the decoder are used for finishing the feedback of the CSI, the feedback accuracy of the CSI between a terminal and network equipment can be improved, and less feedback data volume is used for representing more complete and detailed channel information.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

Fig. 1 is a block diagram of a CSI feedback method provided in an exemplary embodiment of the present application;

Fig. 2 is a flowchart of a CSI feedback method provided in an exemplary embodiment of the present application;

FIG. 3 is a flow chart of an countermeasure generation network provided by an exemplary embodiment of the present application;

FIG. 4 is a training schematic of an countermeasure generation network provided in an exemplary embodiment of the present application;

FIG. 5 is a flow chart of a method of training an encoder and decoder provided in one exemplary embodiment of the present application;

FIG. 6 is a training schematic of an encoder and decoder provided in an exemplary embodiment of the present application;

fig. 7 is a flowchart of a CSI feedback method provided in an exemplary embodiment of the present application;

FIG. 8 is a flow chart of a model update method provided by an exemplary embodiment of the present application;

FIG. 9 is a flowchart of a model update method provided by an exemplary embodiment of the present application;

FIG. 10 is a flow chart of a model update method provided by an exemplary embodiment of the present application;

FIG. 11 is a flowchart of a model update method provided by an exemplary embodiment of the present application;

FIG. 12 is a flowchart of a model update method provided by an exemplary embodiment of the present application;

fig. 13 is a block diagram of a CSI feedback device provided in an exemplary embodiment of the present application;

Fig. 14 is a block diagram of a CSI feedback device provided in an exemplary embodiment of the present application;

fig. 15 is a schematic structural diagram of a communication device according to an exemplary embodiment of the present application.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the present application more apparent, the embodiments of the present application will be described in further detail below with reference to the accompanying drawings.

Fig. 1 shows a schematic diagram of a mobile communication system according to an embodiment of the present application. The mobile communication system may include: a terminal 10 and an access network device 20.

The number of terminals 10 is typically plural and one or more terminals 10 may be distributed within the cell managed by each access network device 20. The terminal 10 may include various handheld, in-vehicle, wearable, computing, or other processing devices connected to a wireless modem, as well as various forms of User Equipment (UE), mobile Station (MS), and the like. For convenience of description, in the embodiment of the present application, the above-mentioned devices are collectively referred to as a terminal.

The access network device 20 is a means deployed in the access network for providing mobile communication functionality for the terminal 10. The access network device 20 may include various forms of macro base stations, micro base stations, relay stations, access points, location management function entities (Location Management Function, LMF), and so forth. The names of access network device-capable devices may vary in systems employing different radio access technologies, for example in 5G NR systems, called gndeb or gNB. As communication technology evolves, the name "access network device" may change. For convenience of description, in the embodiments of the present application, the above-mentioned devices for providing mobile communication functions for the terminal 10 are collectively referred to as an access network device. A connection may be established between the access network device 20 and the terminal 10 over an air interface so that communication, including interaction of signaling and data, may take place over the connection. The number of access network devices 20 may be plural, and communication between two adjacent access network devices 20 may be performed by wired or wireless means. The terminal 10 may switch between different access network devices 20, i.e. establish a connection with different access network devices 20.

The "5G NR system" in the embodiments of the present disclosure may also be referred to as a 5G system or an NR system, but a person skilled in the art may understand the meaning thereof. The technical scheme described in the embodiment of the disclosure can be applied to a 5G NR system and also can be applied to a subsequent evolution system of the 5G NR system.

In the embodiment of the present application, the encoder 12 is provided in the terminal 10, and the decoder 22 is provided in the access network device 20. Access network device 120 transmits CSI-RS to terminal 10 on a downlink channel. The terminal 10 obtains CSI of the downlink channel by measurement based on the CSI-RS. The terminal 10 encodes the CSI through the encoder 12 to obtain CSI feedback information. The terminal 10 reports the CSI feedback information to the access network device 20. The access network device 20 decodes the CSI of the terminal 10 by the decoder 22.

Since the encoder 12 and decoder 22 are artificial intelligence based models, they need to be pre-trained using real training samples. However, as there are fewer training samples in the real training set, the present application also proposes a sample supplementing scheme based on the countermeasure generation network (Generative Adversarial Networks, GAN) that can supplement a sufficient number of training samples, and the similarity between the supplemented training samples and the real training samples is extremely high.

Fig. 2 shows a flowchart of a CSI feedback method provided in an exemplary embodiment of the present application. This embodiment is exemplified by the application of the method to the terminal 10 and the network device 20 shown in fig. 1. The method comprises the following steps:

step 202: the terminal encodes the CSI by using an encoder to obtain CSI feedback information;

the encoder is an AI coding model for encoding CSI into CSI feedback information. The access network equipment sends the CSI-RS to the terminal in a downlink channel, wherein the CSI is obtained after the terminal measures the CSI-RS.

The CSI feedback information is a feedback bit sequence or a feedback codebook obtained after the encoder encodes CSI. And the terminal encodes or compresses the CSI by using an encoder to obtain the CSI feedback information. That is, the AI coding model has a nonlinear fitting capability with which to compress feedback the CSI. The encoder is also called a channel encoder.

Illustratively, the CSI feedback information is at least one of a feedback codebook, a eigenvector, a matrix, and a bit sequence.

Step 204: the terminal sends CSI feedback information to the access network equipment;

and the terminal sends the CSI feedback information to the access network equipment through an uplink feedback channel. The uplink feedback channel may be a physical uplink control channel (Physical Uplink Control Channel, PUCCH) and the uplink feedback channel may also be a physical uplink shared channel (Physical Uplink Shared Channel, PUSCH).

Step 206: the access network equipment receives the CSI feedback information sent by the terminal, wherein the CSI feedback information is obtained by encoding the CSI by the terminal through an encoder;

and the access network equipment receives the CSI feedback information sent by the terminal through the uplink feedback channel.

Step 208: and the access network equipment decodes the CSI feedback information by using the decoder to obtain the CSI measured by the terminal.

The encoder and the decoder are trained by a real training set and a first supplementary training set, wherein the first supplementary training set is generated by a generator in an countermeasure generation network, and the countermeasure generation network is trained based on the real training set. The decoder is also called a channel decoder. The encoder and decoder may be collectively referred to as a CSI self-encoder.

And the access network equipment decodes or reconstructs the CSI feedback information by using the decoder to obtain the CSI of the downlink channel measured by the terminal.

Illustratively, the neural network structure of the generator and the arbiter may employ at least one of a deep neural network (Deep Neural Networks, DNN), a convolutional neural network (Convolutional Neural Network, CNN), a Long short-term memory (LSTM), a gated loop unit (Gate Recurrent Unit, GRU), a loop neural network (Recurrent Neural Network, RNN), and any other possible neural network architecture, and the specific network architecture of the generator and the arbiter is not limited in this embodiment.

In summary, in the method provided in this embodiment, under the condition that training samples in the real training set are fewer, the complementary training set is generated by using the countermeasure generation network, so that the encoder and the decoder with excellent performance are obtained through training, and the encoder and the decoder are used to complete the feedback of CSI, so that the accuracy of CSI feedback between the terminal and the network device can be improved, and less feedback data volume is used to represent more complete and detailed channel information.

It should be noted that, step 202 and step 204 in the embodiment of fig. 2 may be implemented separately as a CSI feedback method on the terminal side, and step 206 and step 208 in the embodiment of fig. 2 may be implemented separately as a CSI feedback method on the access network device side. Likewise, the steps performed by the terminal may be implemented solely as a corresponding method on the terminal side in other embodiments, and the steps performed by the access network device may be implemented solely as a corresponding method on the access network device side in other embodiments.

Training process against the generation network:

FIG. 3 illustrates a flowchart of a training method for an countermeasure generation network provided in an exemplary embodiment of the present application. The method may be performed by an access network device or a terminal or other device, the method comprising:

The countermeasure generation network includes: a generator neural network (Generator Neural Network) and a arbiter neural network (Discriminator Neural Network). The countermeasure network is also called a generation type countermeasure network. The generator neural network is called a generator for short, and the arbiter neural network is called a arbiter for short.

GAN is inspired by zero and gaming in the game theory, and the generated problem is regarded as the countermeasure and gaming of two networks, namely a discriminator and a generator: the generator produces synthetic data from a given noise (typically referred to as a uniform or normal distribution) and the arbiter resolves the output of the generator with the real data. The former attempts to produce more realistic data and, correspondingly, the latter attempts to more perfectly distinguish the realistic data from the generated data. Thus, two networks have advanced in the countermeasure, and after the advancement, the data obtained by the generation type network is more and more perfect and approaches the real data, so that the data which is wanted can be generated.

Illustratively, the neural network structure of the generator and the arbiter may employ at least one of DNN, CNN, LSTM, GRU, RNN and any other possible neural network architecture, and the specific network architecture of the generator and the arbiter is not limited in this embodiment.

Step 302: inputting training samples in a real training set into a discriminator to obtain a first discrimination result;

the training samples in the real training set include at least one of:

·CSI

CSI and CSI feedback information occurring in pairs (e.g., feedback codebook obtained by high-precision quantization).

The form of the training samples is not limited in this application, and the present embodiment is exemplified by taking the training samples in the real training set as CSI.

And inputting training samples in the real training set into a discriminator to obtain a first discrimination result. Illustratively, the first discrimination result may be 0 or 1,0 representing false and 1 representing true. Alternatively, the first discrimination result may be a probability value in the form of a percentage, which is used to represent the probability that the discrimination result is true, for example, 80% represents the probability that the discrimination result is true is 80%, and if the probability exceeds the threshold value by 50%, the discrimination result is considered to be true, and if the probability is less than 50%, the probability is considered to be false.

In this embodiment, the discriminator may also be referred to as a channel discriminator.

Step 304: inputting the noise signal into a generator to obtain a supplementary training sample;

the noise signal may be random noise, such as a noise signal conforming to a gaussian distribution, or a noise signal conforming to a uniform distribution, a noise signal conforming to a bernoulli two-dimensional distribution, or a noise signal conforming to another distribution. The noise signal may also be a noise signal containing known information.

The noise signal is input to a generator which generates a supplemental training sample based on the noise signal, the supplemental training sample being aimed to be as similar or identical as possible to the real training sample in the real training set. The generator may also be referred to as a channel generator.

Step 306: inputting the supplementary training sample into a discriminator to obtain a second discrimination result;

and inputting the supplementary training sample into a discriminator to obtain a second discrimination result.

It should be noted that, the steps 302 and the steps 304-306 may be performed alternately or simultaneously, and the sequence of performing the steps 302 and the steps 304-306 is not limited in this application. For example, step 302 may be performed before steps 304-306 are performed; steps 304-306 may also be performed before step 302. For another example, steps 302 and 304-306 are alternately performed based on different training samples; alternatively, step 302 is performed once and steps 304-306 are performed multiple times; still alternatively, steps 304-306 are performed once and step 302 is performed multiple times.

Referring to fig. 4 in combination, taking an access network device as an example of a training device, the access network device may collect CSI of a plurality of terminals as a real training set 31, for example, CSI of a plurality of terminals in the same geographic range as the real training set 31. On the one hand, the real training samples 32 in the real training set 31 are input to the discriminator D, and the discriminator D outputs the corresponding discrimination result 35, i.e. the first discrimination result. On the other hand, the random noise 33 is input to the generator G, which outputs a dummy (complementary) training sample 34, and the dummy training sample 34 is input to the discriminator D, which outputs a corresponding discrimination result 35, i.e., a second discrimination result. Then, the generator and the arbiter are trained based on the loss functions of the first and second discrimination results.

Step 308: based on the first discrimination result and the second discrimination result, the generator and the discriminator are trained.

The loss function of the generator is set by a target that the first discrimination result and the second discrimination result are both true, and the loss function of the discriminator is set by a target that the first discrimination result is true and the second discrimination result is false.

Illustratively, in the training stage of the discriminator, the neural model parameters of the fixed generator are unchanged, and the discriminator is trained based on the loss function of the discriminator; in the training stage of the generator, the neural model parameters of the fixed discriminant are unchanged, and the generator is trained based on the loss function of the generator. And alternately executing the two training processes until the training ending condition is met.

Illustratively, the training end conditions include: the number of training times reaches a number of times threshold, or the loss function converges.

The training method of the countermeasure generation network is not limited in this embodiment, and may be implemented in a modified form such as wasperstein GAN (WGAN-GP) with gradient penalty.

In summary, the method provided in this embodiment can train to obtain the countermeasure generation network based on the real training samples in the real training set. The countermeasure generation network is capable of generating as many false training samples as possible as true training samples, and in the case where there are limited true training samples in the true training set, the countermeasure generation network is capable of generating enough false training samples as supplemental training samples.

Training process of encoder and decoder:

fig. 5 shows a flowchart of a training method of an encoder and a decoder according to an exemplary embodiment of the present application. The method may be performed by an access network device or a terminal or other device, the method comprising:

step 402: generating a first supplemental training set with a generator in the countermeasure generation network;

let M be the number of training samples needed to adequately support the training encoder and decoder, and M be the number of real training samples in the real training set (or raw data set). If the real training set has acquisition limit, time limit or cost limit, and the size M of the real training set is far smaller than the size M of the required training set, namely M < M, the real training set is directly used for training the encoder, and a coding model with better performance cannot be obtained.

After the challenge-generating network is trained, the generator has better sample generating capability. A first supplemental training set can be generated based on a generator in the countermeasure generation network. The first supplemental training set is not less than (M-M) in size.

Step 404: mixing the real training set and the first supplementary training set to obtain a combined training set;

And mixing the real training sample in the real training set and the supplementary training sample in the first supplementary training set to obtain a combined training set. The size of the joint training set is equal to or greater than M.

That is, the joint training set is of a size sufficient to support the number of training samples required by the encoder and decoder.

Step 406: and training the encoder and/or the decoder by adopting the joint training set to obtain the trained encoder and/or decoder.

And training the encoder and/or the decoder by adopting the joint training set to obtain the trained encoder and/or decoder.

In one example, each training sample in the joint training set is CSI, and the encoder and decoder are trained in an end-to-end training manner. In another example, each training sample in the joint training set includes a set of CSI and CSI feedback codebooks, based on which the encoder may be trained separately, the decoder may be trained separately, and the encoder and decoder may be trained end-to-end.

The present embodiment is not limited to the manner in which the encoder and/or decoder is trained. Referring schematically to fig. 6, the training device mixes the real training set 41 and the supplemental training set 42 into a joint training set 43. The joint training set 43 includes a plurality of training samples, such as one CSI for each training sample, and the encoder and decoder are trained using the CSI in the joint training set 43. I.e., the training samples are input to the encoder, which trains the training samples to obtain CSI feedback information 44. The decoder decodes the CSI feedback information 44 and outputs a recovered channel 45. The error between the recovered channel 45 and the training samples is calculated based on the loss function 46 and the encoder and decoder are end-to-end trained using an error back propagation method.

In summary, in the method provided in this embodiment, the first complementary training set is generated by the countermeasure generation network, and then the combined training set obtained by mixing the real training set and the first complementary training set is used, so that the encoder and the decoder with better performance can be obtained by using sufficient training samples, and the compression efficiency and the feedback accuracy in CSI feedback can be improved.

CSI feedback process for model training device based on access network device:

fig. 7 shows a flowchart of a CSI feedback method provided in an exemplary embodiment of the present application. The method may be performed by an access network device or terminal, the method comprising:

step 502: the access network equipment trains to obtain an countermeasure generation network based on the real training set;

the training process of the countermeasure generation network may refer to the training process shown in the embodiment shown in fig. 3 and will not be described again.

Step 504: the access network equipment trains to obtain an encoder and a decoder based on the joint training set;

the training process of the encoder and the decoder may refer to the training process shown in the embodiment shown in fig. 5 and will not be described again.

Step 506: the access network equipment transmits an encoder to the terminal;

the access network equipment transmits the encoder to the terminal through the downlink signaling, or transmits the model parameters of the encoder, and the terminal constructs the encoder according to the model parameters of the encoder.

Illustratively, the downlink signaling includes at least one of downlink control information (Downlink Controllnformation, DCI), radio resource control (Radio Resource Control, RRC), medium access control (Medium Access Control Control Element, MAC CE). The downlink signaling may also be dedicated signaling and channel resources dedicated to model delivery, which is not limited by the embodiments of the present application.

Illustratively, the model parameters of the encoder and/or decoder include: at least one of a neural network type, a number of layers of the neural network, a type of a neural network layer, a type of a neuron in the neural network, a number of neurons in the neural network, a matrix weight of a neuron in the neural network.

Step 508: the terminal encodes the CSI by using an encoder to obtain CSI feedback information;

step 510: the terminal sends CSI feedback information to the access network equipment;

and the terminal sends the CSI feedback information to the access network equipment by using the uplink feedback channel.

Step 512: the access network equipment receives the CSI feedback information sent by the terminal, wherein the CSI feedback information is obtained by encoding the CSI by the terminal through an encoder;

step 514: and the access network equipment decodes the CSI feedback information by using the decoder to obtain the CSI measured by the terminal.

In other embodiments, the training device may also be implemented by a terminal. The terminal trains to obtain an countermeasure generation network based on the real training set; the terminal trains to obtain an encoder and a decoder based on the joint training set; the terminal reports the decoder or the model parameters of the decoder to the access network device. And the terminal encodes the CSI by using an encoder to obtain the CSI feedback information. And the terminal sends the CSI feedback information to the access network equipment. The access network equipment receives the CSI feedback information sent by the terminal, wherein the CSI feedback information is obtained by encoding the CSI by the terminal through an encoder. And the access network equipment decodes the CSI feedback information by using the decoder to obtain the CSI measured by the terminal.

In other embodiments, the countermeasure network may also be trained by the first device, and the countermeasure network or model parameters of the countermeasure network may be sent to the second device. The second device generates a supplemental training set based on the challenge-generating network, and trains the resulting encoder and decoder based on the joint training set. The first device is an access network device, and the second device is a terminal; alternatively, the first device is a terminal and the second device is an access network device.

In summary, the method provided in this embodiment can quickly train to obtain the encoder and decoder with excellent performance by using the strong computing capability of the access network device. And then the access network equipment transmits an encoder to the terminal to complete the CSI feedback process based on the AI, and the compression efficiency and the feedback precision during the CSI feedback are improved.

Update training process of encoder and/or decoder based on periodic feedback:

limited by the generalization of the encoder and the complex variability of the channel environment. When the channel environment changes and the encoder does not adapt to the channel environment, the model parameters need to be updated when the performance is reduced. However, since the encoder of the CSI self-encoder is disposed on the UE side and the decoder is disposed on the network side, when the model is not adapted, the network side cannot acquire a large amount of changed high-quality valid channel data sets through CSI feedback. And continuously adopting a high-precision codebook to perform compression feedback on a changed channel, and introducing higher feedback overhead. Therefore, the embodiment adopts a mode of antagonizing the generation network, reduces the feedback density of the high-precision codebook, generates the second supplementary data set based on a small amount of channel feedback, and completes the model update of the encoder.

Fig. 8 shows a flowchart of a model updating method according to an exemplary embodiment of the present application. The method may be performed by an access network device and a terminal, the method comprising:

step 602: the terminal periodically sends a first CSI feedback codebook based on codebook quantification to access network equipment;

The first CSI feedback codebook is a CSI feedback codebook obtained based on a codebook quantization mode, and is not a CSI feedback codebook obtained based on an AI model or encoder compression. Illustratively, the first CSI feedback codebook is a high quality CSI feedback codebook, and can effectively express a changed channel environment.

Since the first CSI feedback codebook may also be plural, the first CSI feedback codebook may also be referred to as a second original training set, a second real training set, an updated training set, or the like.

Illustratively, the first CSI feedback codebook includes CSI feedback information of a plurality of sampling time points, such as a plurality of CSI feedback information within a time window W, or a plurality of CSI feedback information sampled according to a specified period within the time window W.

Illustratively, the first CSI feedback codebook includes CSI feedback information on a plurality of bandwidths, subbands, or frequency points, such as a plurality of CSI feedback information within frequency window B, or a plurality of CSI feedback information on a plurality of bandwidths and/or subbands of frequency window B.

Illustratively, the feedback period of the first CSI feedback codebook may be determined by the configuration parameter T. The configuration parameter T of the first CSI feedback codebook may also carry other configuration information needed in the codebook feedback process.

In one example, prior to step 602, the access network device sends a first reporting configuration to the terminal, such as by downlink signaling. The terminal receives a first reporting configuration sent by the access network device, wherein the first reporting configuration is used for indicating a first reporting parameter of a first CSI feedback codebook. Illustratively, the first reporting parameter includes at least one of: a time window W; a frequency window B; the configuration parameter T of the first CSI feedback codebook.

In another example, before step 602, the terminal determines a first reporting configuration by itself, where the first reporting configuration is used to indicate a first reporting parameter of the first CSI feedback codebook. And the terminal sends the first reporting configuration to the access network equipment. For example, the terminal sends the first reporting configuration to the access network device through uplink signaling.

Step 604: the access network equipment receives a first CSI feedback codebook sent by a terminal;

step 606: the access network equipment builds a second supplementary training set based on the first CSI feedback codebook through the countermeasure generation network;

the access network equipment builds a second supplementary training set based on the first CSI feedback codebook through the countermeasure generation network; or, the access network equipment determines the CSI corresponding to the first CSI feedback codebook, and constructs a second supplementary training set based on the CSI corresponding to the first CSI feedback codebook through the countermeasure generation network; or the access network equipment determines the CSI corresponding to the first CSI feedback codebook, and constructs a second supplementary training set based on the first CSI feedback codebook and the CSI corresponding to the first CSI feedback codebook through the countermeasure generation network.

The manner of training the countermeasure generation network based on the first CSI feedback codebook is similar to the embodiment shown in fig. 3, and will not be described in detail.

In one example, the countermeasure network is the countermeasure network shown in the embodiment of fig. 3.

In one embodiment, the countermeasure network is a further trained countermeasure network based on the first CSI feedback codebook, the countermeasure network being different from the countermeasure network shown in the embodiment shown in fig. 3.

Step 608: the access network equipment updates and trains at least one of the encoder and the decoder through the first CSI feedback codebook and the second complementary training set;

illustratively, the access network device performs joint update training on the encoder and the decoder through a first joint update training set formed by the first CSI feedback codebook and the second supplemental training set. The updated encoder can be well adapted to the changed channel information.

The access network equipment can only update and train the encoder; or, only updating and training the decoder; alternatively, the encoder and decoder are updated and trained.

Step 610: the access network device issues the updated encoder to the terminal.

In the case of updating the encoder, the access network device issues the updated encoder to the terminal, and the whole process may be referred to as shown in fig. 9. The quantization accuracy of the high-accuracy codebook in step 602 may be further enhanced based on the Type2 codebook, so as to improve the recovery accuracy of CSI and ensure the accuracy of the first CSI feedback codebook; meanwhile, feedback related configuration of the first CSI feedback codebook: the period parameter T, the time window W, the frequency window B, etc. may be configured by the network side, and the UE is notified through DCI. The process of generating the second supplemental data set based on the first CSI feedback codebook in step 606 is the same as the method of producing the first supplemental data set in the embodiment of fig. 3. However, because the model is updated, the second supplemental data set is typically generated with a smaller sample size than the required sample size M of the data set of embodiment 1 shown in fig. 3, so as to ensure that model convergence against the generation network and rapid updating of the CSI self-encoder model can be accomplished quickly.

In another embodiment, the device for updating the encoder and/or decoder may also be a terminal. The terminal periodically measures the first CSI to form an updated training set; the terminal builds a second supplementary training set based on the updated training set through the countermeasure generation network; and the terminal updates and trains at least one of the encoder and the decoder through the first CSI feedback codebook and the second supplementary training set. And the terminal reports the updated decoder to the access network equipment. The specific process is similar to the embodiment shown in fig. 8, and will not be described again.

In summary, in the method provided in this embodiment, by configuring a high-precision codebook with a lower density for feeding back an update data set and matching with an antagonism generation network, the on-line update of the CSI self-encoder model can be implemented, and the accuracy of CSI feedback and recovery is ensured by using the method based on periodic update in the on-line deployment process of the CSI self-encoder.

Update training process of encoder and/or decoder based on aperiodic feedback:

fig. 10 shows a flowchart of a model updating method according to an exemplary embodiment of the present application. The method may be performed by an access network device and a terminal, the method comprising:

step 702: the terminal sends a second CSI feedback codebook based on codebook quantification to the access network equipment under the condition that the triggering condition is met;

the second CSI feedback codebook is a CSI feedback codebook obtained based on a codebook quantization mode, and is not a CSI feedback codebook obtained based on an AI model or encoder compression. Illustratively, the second CSI feedback codebook is a high quality CSI feedback codebook, and can effectively express a changed channel environment.

Since the second CSI feedback codebook may also be plural, the second CSI feedback codebook may also be referred to as a second original training set, a second real training set, an updated training set, or the like.

The triggering conditions include: the change in channel state or channel information or channel parameters is greater than a preset threshold. The channel state or channel information or channel parameters include: reference signal received power (Reference Signal Received Power, RSRP), reference signal received quality (Reference Signal Received Quality, RSRQ), received signal strength indicator (Reference Signal Strength Indicator, RSSI), CSI.

The preset threshold may be predefined by the communication protocol; alternatively, the preset threshold may be preconfigured; alternatively, the preset threshold may be configured by the access network device towards the terminal.

Illustratively, the second CSI feedback codebook includes CSI feedback information of a plurality of sampling time points, such as a plurality of CSI feedback information within the time window W, or a plurality of CSI feedback information sampled at a specified period within the time window W.

Illustratively, the second CSI feedback codebook includes CSI feedback information on a plurality of bandwidths, subbands, or frequency points, such as a plurality of CSI feedback information within frequency window B, or a plurality of CSI feedback information on a plurality of bandwidths and/or subbands of frequency window B.

Illustratively, the feedback period of the second CSI feedback codebook may be determined by the configuration parameter T. The triggering condition of the second CSI feedback codebook may also carry other configuration information needed in the codebook feedback process.

In one example, prior to step 702, the access network device sends a second reporting configuration to the terminal, such as by downlink signaling. The terminal receives a second reporting configuration sent by the access network device, wherein the second reporting configuration is used for indicating a second reporting parameter of a second CSI feedback codebook. Illustratively, the second reporting parameter includes at least one of: a time window W; a frequency window B; the trigger condition of the second CSI feedback codebook is as shown in fig. 11.

In another example, before step 702, the terminal determines a second reporting configuration by itself, where the second reporting configuration is used to indicate a second reporting parameter of a second CSI feedback codebook. And the terminal sends the second reporting configuration to the access network equipment. For example, the terminal sends the second reporting configuration to the access network device through uplink signaling, as shown in fig. 12.

Step 704: the access network equipment receives a second CSI feedback codebook sent by the terminal;

step 706: the access network equipment builds a third supplementary training set based on the second CSI feedback codebook through the countermeasure generation network;

the access network equipment builds a third supplementary training set based on the second CSI feedback codebook through the countermeasure generation network; or, the access network equipment determines the CSI corresponding to the second CSI feedback codebook, and constructs a third supplementary training set based on the CSI corresponding to the second CSI feedback codebook through the countermeasure generation network; or the access network equipment determines the CSI corresponding to the second CSI feedback codebook, and constructs a third supplementary training set based on the second CSI feedback codebook and the CSI corresponding to the second CSI feedback codebook through the countermeasure generation network.

The manner of training the countermeasure generation network based on the second CSI feedback codebook is similar to the embodiment shown in fig. 3, and will not be described in detail.

In one embodiment, the countermeasure network is a countermeasure network based on the second CSI feedback codebook additional training, the countermeasure network being different from the countermeasure network shown in the embodiment shown in fig. 3.

Step 708: the access network equipment performs update training on at least one of the encoder and the decoder through the second CSI feedback codebook and the third supplemental training set;

illustratively, the access network device performs joint update training on the encoder and the decoder through a second joint update training set formed by the second CSI feedback codebook and the third supplemental training set. The updated encoder can be well adapted to the changed channel information.

Step 710: the access network device issues the updated encoder to the terminal.

In case of updating the encoder, the access network device issues the updated encoder to the terminal. The quantization accuracy of the high-accuracy codebook in step 702 may be further enhanced based on the Type2 codebook, so as to improve the recovery accuracy of CSI and ensure the accuracy of the second CSI feedback codebook; meanwhile, feedback related configuration of the second CSI feedback codebook: the period parameter T, the time window W, the frequency window B, etc. may be configured by the network side, and the UE is notified through DCI. The process of generating the second supplemental data set based on the second CSI feedback codebook in step 706 is the same as the method of producing the first supplemental data set in the embodiment of fig. 3. However, because the model is updated, the second supplemental data set is typically generated with a smaller sample size than the required sample size M of the data set of embodiment 1 shown in fig. 3, so as to ensure that model convergence against the generation network and rapid updating of the CSI self-encoder model can be accomplished quickly.

In another embodiment, the device for updating the encoder and/or decoder may also be a terminal. The terminal periodically measures the second CSI to form an updated training set; the terminal builds a third supplementary training set based on the updated training set through the countermeasure generation network; and the terminal updates and trains at least one of the encoder and the decoder through the second CSI feedback codebook and the third supplementary training set. And the terminal reports the updated decoder to the access network equipment. The specific process is similar to the embodiment shown in fig. 8, and will not be described again.

In summary, in the method provided in the present embodiment, during the online deployment process of the CSI self-encoder, the CSI self-encoder model is updated online based on the trigger condition, so that compared with the previous embodiment, the communication data volume during communication between the terminal and the server can be reduced.

Fig. 13 is a block diagram of a CSI feedback device according to an exemplary embodiment of the present application, which may be implemented as a terminal or a functional module within a terminal, where the device includes:

an encoding module 1320, configured to encode the CSI using an encoder to obtain CSI feedback information; the encoder is trained by a real training set and a first supplemental training set, the first supplemental training set is generated by a generator in an countermeasure generation network, and the countermeasure generation network is trained based on the real training set;

a sending module 1340, configured to send the CSI feedback information to an access network device.

In an alternative embodiment, the encoder is trained in the following manner:

generating the first supplemental training set with a generator in the countermeasure generation network;

mixing the real training set and the first supplementary training set to obtain a combined training set;

And training the encoder by adopting the combined training set to obtain the trained encoder.

In an alternative embodiment, the countermeasure generation network includes a generator and a arbiter, the countermeasure generation network being trained based on:

inputting training samples in the real training set into the discriminator to obtain a first discrimination result;

inputting a noise signal into the generator to obtain a supplementary training sample; inputting the supplementary training sample into the discriminator to obtain a second discrimination result;

training to obtain the generator and the discriminator based on the first discrimination result and the second discrimination result;

In an alternative embodiment, the apparatus further comprises:

a receiving module 1360, configured to receive the encoder issued by the access network device, where the encoder is trained by the access network device.

In an optional embodiment, the sending module 1340 is further configured to periodically send a first CSI feedback codebook based on codebook quantization to the access network device; the receiving module 1360 is configured to receive an updated encoder issued by the access network device, where the updated encoder is obtained after the access network device performs update training on the encoder based on the first CSI feedback codebook and a second supplemental training set, and the second supplemental training set is constructed by the countermeasure generation network based on the first CSI feedback codebook.

In an optional embodiment, the receiving module 1360 is further configured to receive a first reporting configuration sent by the access network device, where the first reporting configuration is used to indicate a first reporting parameter of the first CSI feedback codebook;

or alternatively, the first and second heat exchangers may be,

the sending module 1340 is further configured to send a first reporting configuration to the access network device, where the first reporting configuration is used to indicate a first reporting parameter of the first CSI feedback codebook.

In an alternative embodiment, the first reporting parameter includes at least one of: a time window W; a frequency window B; and the configuration parameter T of the first CSI feedback codebook.

In an optional embodiment, the sending module 1340 is further configured to send, to the access network device, a second CSI feedback codebook based on codebook quantization if a trigger condition is met; the receiving module 1360 is further configured to receive an updated encoder issued by the access network device, where the updated encoder is obtained after the access network device performs update training on the encoder based on the second CSI feedback codebook and a third supplemental training set, and the third supplemental training set is constructed by the countermeasure generation network based on the second CSI feedback codebook.

In an alternative embodiment, the triggering condition includes:

the variation value of the channel parameter is larger than a preset threshold value;

wherein the channel state or channel information or channel parameters include: RSRP, RSRQ, RSSI, CSI.

In an alternative embodiment, the apparatus further comprises:

the receiving module 1360 is further configured to receive a second reporting configuration sent by the access network device, where the second reporting configuration is used to indicate a second reporting parameter of the second CSI feedback codebook; or, the sending module 1340 is further configured to send a second reporting configuration to the access network device, where the second reporting configuration is used to indicate a second reporting parameter of the second CSI feedback codebook.

In an alternative embodiment, the second reporting parameter includes at least one of: a time window W; a frequency window B; and triggering conditions of the second CSI feedback codebook.

Fig. 14 is a block diagram of a CSI feedback apparatus according to an exemplary embodiment of the present application, which may be implemented as an access network device or as a functional module within an access network device, where the apparatus includes:

a receiving module 1420, configured to receive CSI feedback information sent by a terminal, where the CSI feedback information is obtained by encoding CSI by the terminal through an encoder;

a decoding module 1440, configured to decode the CSI feedback information by using a decoder to obtain CSI measured by the terminal; the encoder and the decoder are trained from a real training set and a first supplemental training set, the first supplemental training set being generated by a generator in an countermeasure generation network, the countermeasure generation network being trained based on the real training set.

In an alternative embodiment, the encoder and decoder are trained as follows:

and training the encoder and the decoder by adopting the joint training set to obtain the trained encoder and decoder.

In an alternative embodiment, a transmitting module 1460 is configured to issue the encoder to the terminal.

In an alternative embodiment, the receiving module 1420 is configured to periodically receive a first CSI feedback codebook based on codebook quantization sent by a terminal;

a training module 1480 for a second supplemental training set constructed based on the first CSI feedback codebook through the countermeasure generation network; and updating and training at least one of the encoder and the decoder through the first CSI feedback codebook and the second supplemental training set.

In an optional embodiment, the receiving module 1420 is configured to receive a first reporting configuration sent by the terminal, where the first reporting configuration is configured to indicate a first reporting parameter of the first CSI feedback codebook; or, a sending module 1460 is configured to send a first reporting configuration to the terminal, where the first reporting configuration is used to indicate a first reporting parameter of the first CSI feedback codebook.

In an alternative embodiment, the first reporting parameter includes at least one of:

a time window W;

a frequency window B;

and the configuration parameter T of the first CSI feedback codebook.

In an alternative embodiment, the receiving module 1420 is configured to receive a second CSI feedback codebook based on codebook quantization sent by the terminal if a trigger condition is met;

In an alternative embodiment, the sending module 1460 is configured to send the updated encoder to the terminal in a case of performing update training on the encoder.

In an alternative embodiment, the triggering condition includes: the variation value of the channel parameter is larger than a preset threshold value. The channel state or channel information or channel parameters include: RSRP, RSRQ, RSSI, CSI.

In an optional embodiment, the receiving module 1420 is configured to receive a second reporting configuration sent by the terminal, where the second reporting configuration is configured to indicate a second reporting parameter of the second CSI feedback codebook; or, a sending module 1460 is configured to send a second reporting configuration to the terminal, where the second reporting configuration is used to indicate a second reporting parameter of the second CSI feedback codebook.

Fig. 15 shows a schematic structural diagram of a communication device (terminal or access network device) according to an exemplary embodiment of the present application, where the communication device includes: a processor 101, a receiver 102, a transmitter 103, a memory 104, and a bus 105.

The processor 101 includes one or more processing cores, and the processor 101 executes various functional applications and information processing by running software programs and modules.

The receiver 102 and the transmitter 103 may be implemented as one communication component, which may be a communication chip.

The memory 104 is connected to the processor 101 via a bus 105.

The memory 104 may be used to store at least one instruction that the processor 101 is configured to execute to implement the various steps of the method embodiments described above.

Further, the memory 104 may be implemented by any type of volatile or nonvolatile storage device or combination thereof, including but not limited to: magnetic or optical disks, electrically erasable programmable Read-Only Memory (EEPROM), erasable programmable Read-Only Memory (Erasable Programmable Read Only Memory, EPROM), static random access Memory (Static Random Access Memory, SRAM), read-Only Memory (ROM), magnetic Memory, flash Memory, programmable Read-Only Memory (Programmable Read-Only Memory, PROM).

In an exemplary embodiment, there is also provided a computer readable storage medium having stored therein at least one instruction, at least one program, a code set, or a set of instructions, which are loaded and executed by the processor to implement the CSI feedback method performed by the first terminal or the second terminal or the network device provided in the above respective method embodiments.

In an exemplary embodiment, a computer program product or a computer program is also provided, the computer program product or computer program comprising computer instructions stored in a computer readable storage medium, the computer instructions being read from the computer readable storage medium by a processor of a communication device, the computer instructions being executed by the processor, causing the communication device to perform the CSI feedback method performed by the first terminal or the second terminal or the network device as described in the above aspects.

The foregoing description of the preferred embodiments is merely exemplary in nature and is in no way intended to limit the invention, since it is intended that all modifications, equivalents, improvements, etc. that fall within the spirit and scope of the invention.

Claims

A channel state information CSI feedback method, applied in a terminal, the method comprising:

coding the CSI by using an encoder to obtain CSI feedback information; the encoder is trained by a real training set and a first supplemental training set, the first supplemental training set is generated by a generator in an countermeasure generation network, and the countermeasure generation network is trained based on the real training set;

and sending the CSI feedback information to access network equipment.
The method of claim 1, wherein the encoder is trained by:

generating the first supplemental training set with a generator in the countermeasure generation network;

mixing the real training set and the first supplementary training set to obtain a combined training set;

and training the encoder by adopting the combined training set to obtain the trained encoder.
The method of claim 1, wherein the countermeasure generation network includes a generator and a arbiter, the countermeasure generation network being trained based on:

inputting training samples in the real training set into the discriminator to obtain a first discrimination result;

Inputting a noise signal into the generator to obtain a supplementary training sample; inputting the supplementary training sample into the discriminator to obtain a second discrimination result;

training to obtain the generator and the discriminator based on the first discrimination result and the second discrimination result;

the loss function of the generator is set by a target that the first discrimination result and the second discrimination result are both true, and the loss function of the discriminator is set by a target that the first discrimination result is true and the second discrimination result is false.
A method according to any one of claims 1 to 3, wherein the method further comprises:

and receiving the coder issued by the access network equipment, wherein the coder is trained by the access network equipment.
A method according to any one of claims 1 to 3, wherein the method further comprises:

periodically sending a first CSI feedback codebook based on codebook quantization to the access network equipment;

and receiving an updated encoder issued by the access network equipment, wherein the updated encoder is obtained after the access network equipment performs updating training on the encoder based on the first CSI feedback codebook and a second supplementary training set, and the second supplementary training set is constructed by the countermeasure generation network based on the first CSI feedback codebook.
The method of claim 5, wherein the method further comprises:

receiving a first reporting configuration sent by the access network equipment, wherein the first reporting configuration is used for indicating a first reporting parameter of the first CSI feedback codebook;

or alternatively, the first and second heat exchangers may be,

and sending a first reporting configuration to the access network equipment, wherein the first reporting configuration is used for indicating a first reporting parameter of the first CSI feedback codebook.
The method of claim 6, wherein the first reporting parameter comprises at least one of:

a time window W;

a frequency window B;

and the configuration parameter T of the first CSI feedback codebook.
A method according to any one of claims 1 to 3, wherein the method further comprises:

transmitting a second CSI feedback codebook based on codebook quantization to the access network equipment under the condition that a trigger condition is met;

and receiving an updated encoder issued by the access network equipment, wherein the updated encoder is obtained after the access network equipment performs updating training on the encoder based on the second CSI feedback codebook and a third supplementary training set, and the third supplementary training set is constructed by the countermeasure generation network based on the second CSI feedback codebook.
The method of claim 8, wherein the trigger condition comprises:

the variation value of the channel parameter is larger than a preset threshold value.
The method of claim 8, wherein the method further comprises:

receiving a second reporting configuration sent by the access network device, wherein the second reporting configuration is used for indicating a second reporting parameter of the second CSI feedback codebook;

or alternatively, the first and second heat exchangers may be,

and sending a second reporting configuration to the access network equipment, wherein the second reporting configuration is used for indicating a second reporting parameter of the second CSI feedback codebook.
The method of claim 10, wherein the second reporting parameter comprises at least one of:

a time window W;

a frequency window B;

and the configuration parameter T of the second CSI feedback codebook.
A channel state information CSI feedback method, applied to an access network device, the method comprising:

receiving CSI feedback information sent by a terminal, wherein the CSI feedback information is obtained by encoding the CSI by the terminal through an encoder;

decoding the CSI feedback information by using a decoder to obtain the CSI measured by the terminal; the encoder and the decoder are trained from a real training set and a first supplemental training set, the first supplemental training set being generated by a generator in an countermeasure generation network, the countermeasure generation network being trained based on the real training set.
The method of claim 12, wherein the encoder and the decoder are trained by:

generating the first supplemental training set with a generator in the countermeasure generation network;

mixing the real training set and the first supplementary training set to obtain a combined training set;

and training the encoder and the decoder by adopting the joint training set to obtain the trained encoder and decoder.
The method of claim 12, wherein the countermeasure generation network includes a generator and a arbiter, the countermeasure generation network being trained based on:

inputting training samples in the real training set into the discriminator to obtain a first discrimination result;

inputting a noise signal into the generator to obtain a supplementary training sample; inputting the supplementary training sample into the discriminator to obtain a second discrimination result;

training to obtain the generator and the discriminator based on the first discrimination result and the second discrimination result;

the loss function of the generator is set by a target that the first discrimination result and the second discrimination result are both true, and the loss function of the discriminator is set by a target that the first discrimination result is true and the second discrimination result is false.
The method according to any one of claims 12 to 14, further comprising:

and issuing the encoder to the terminal.
The method according to any one of claims 12 to 14, further comprising:

periodically receiving a first CSI feedback codebook based on codebook quantization sent by a terminal;

a second supplemental training set constructed based on the first CSI feedback codebook through the countermeasure generation network;

and updating and training at least one of the encoder and the decoder through the first CSI feedback codebook and the second supplemental training set.
The method of claim 16, wherein the method further comprises:

receiving a first reporting configuration sent by the terminal, wherein the first reporting configuration is used for indicating a first reporting parameter of the first CSI feedback codebook;

or alternatively, the first and second heat exchangers may be,

and sending a first reporting configuration to the terminal, wherein the first reporting configuration is used for indicating a first reporting parameter of the first CSI feedback codebook.
The method of claim 17, wherein the first reporting parameter comprises at least one of:

a time window W;

a frequency window B;

And the configuration parameter T of the first CSI feedback codebook.
The method according to any one of claims 12 to 14, further comprising:

receiving a second CSI feedback codebook based on codebook quantization, which is sent by the terminal under the condition that the triggering condition is met;

a second supplemental training set constructed based on the first CSI feedback codebook through the countermeasure generation network;

and updating and training at least one of the encoder and the decoder through the first CSI feedback codebook and the second supplemental training set.
The method according to claim 16 or 19, characterized in that the method further comprises:

and under the condition of updating and training the encoder, issuing the updated encoder to the terminal.
The method of claim 19, wherein the trigger condition comprises:

the variation value of the channel parameter is larger than a preset threshold value.
The method of claim 19, wherein the method further comprises:

receiving a second reporting configuration sent by the terminal, wherein the second reporting configuration is used for indicating a second reporting parameter of the second CSI feedback codebook;

or alternatively, the first and second heat exchangers may be,

and sending a second reporting configuration to the terminal, wherein the second reporting configuration is used for indicating a second reporting parameter of the second CSI feedback codebook.
The method of claim 22, wherein the second reporting parameter comprises at least one of:

a time window W;

a frequency window B;

and triggering conditions of the second CSI feedback codebook.
A channel state information, CSI, feedback apparatus, the apparatus comprising:

the coding module is used for coding the CSI by using an encoder to obtain CSI feedback information; the encoder is trained by a real training set and a first supplemental training set, the first supplemental training set is generated by a generator in an countermeasure generation network, and the countermeasure generation network is trained based on the real training set;

and the sending module is used for sending the CSI feedback information to the access network equipment.
The apparatus of claim 24, wherein the encoder is trained by:

generating the first supplemental training set with a generator in the countermeasure generation network;

mixing the real training set and the first supplementary training set to obtain a combined training set;

and training the encoder by adopting the combined training set to obtain the trained encoder.
The apparatus of claim 24, wherein the countermeasure generation network includes a generator and a arbiter, the countermeasure generation network being trained based on:

inputting training samples in the real training set into the discriminator to obtain a first discrimination result;

inputting a noise signal into the generator to obtain a supplementary training sample; inputting the supplementary training sample into the discriminator to obtain a second discrimination result;

training to obtain the generator and the discriminator based on the first discrimination result and the second discrimination result;

the loss function of the generator is set by a target that the first discrimination result and the second discrimination result are both true, and the loss function of the discriminator is set by a target that the first discrimination result is true and the second discrimination result is false.
The apparatus according to any one of claims 24 to 26, further comprising:

and the receiving module is used for receiving the encoder issued by the access network equipment, and the encoder is trained by the access network equipment.
The apparatus according to any one of claims 24 to 26, further comprising:

The sending module is further configured to periodically send a first CSI feedback codebook based on codebook quantization to the access network device;

the receiving module is configured to receive an updated encoder issued by the access network device, where the updated encoder is obtained after the access network device performs update training on the encoder based on the first CSI feedback codebook and a second supplemental training set, and the second supplemental training set is constructed by the countermeasure generation network based on the first CSI feedback codebook.
The apparatus of claim 28, wherein the apparatus further comprises:

the receiving module is further configured to receive a first reporting configuration sent by the access network device, where the first reporting configuration is used to indicate a first reporting parameter of the first CSI feedback codebook;

or alternatively, the first and second heat exchangers may be,

the sending module is further configured to send a first reporting configuration to the access network device, where the first reporting configuration is used to indicate a first reporting parameter of the first CSI feedback codebook.
The apparatus of claim 29, wherein the first reporting parameter comprises at least one of:

a time window W;

a frequency window B;

And the configuration parameter T of the first CSI feedback codebook.
The apparatus according to any one of claims 24 to 27, further comprising:

the sending module is further configured to send a second CSI feedback codebook based on codebook quantization to the access network device when the triggering condition is met;

the receiving module is further configured to receive an updated encoder issued by the access network device, where the updated encoder is obtained after the access network device performs update training on the encoder based on the second CSI feedback codebook and a third supplemental training set, and the third supplemental training set is constructed by the countermeasure generation network based on the second CSI feedback codebook.
The apparatus of claim 31, wherein the trigger condition comprises:

the variation value of the channel parameter is larger than a preset threshold value.
The apparatus of claim 31, wherein the apparatus further comprises:

the receiving module is further configured to receive a second reporting configuration sent by the access network device, where the second reporting configuration is used to indicate a second reporting parameter of the second CSI feedback codebook;

or alternatively, the first and second heat exchangers may be,

the sending module is further configured to send a second reporting configuration to the access network device, where the second reporting configuration is used to indicate a second reporting parameter of the second CSI feedback codebook.
The apparatus of claim 33, wherein the second reporting parameter comprises at least one of:

a time window W;

a frequency window B;

and triggering conditions of the second CSI feedback codebook.
A channel state information, CSI, feedback apparatus, the apparatus comprising:

the receiving module is used for receiving the CSI feedback information sent by the terminal, wherein the CSI feedback information is obtained by the terminal through encoding the CSI by an encoder;

the decoding module is used for decoding the CSI feedback information by using a decoder to obtain the CSI measured by the terminal; the encoder and the decoder are trained from a real training set and a first supplemental training set, the first supplemental training set being generated by a generator in an countermeasure generation network, the countermeasure generation network being trained based on the real training set.
The apparatus of claim 35, wherein the encoder and the decoder are trained by:

generating the first supplemental training set with a generator in the countermeasure generation network;

mixing the real training set and the first supplementary training set to obtain a combined training set;

And training the encoder and the decoder by adopting the joint training set to obtain the trained encoder and decoder.
The apparatus of claim 35, wherein the countermeasure generation network includes a generator and a arbiter, the countermeasure generation network being trained based on:

inputting training samples in the real training set into the discriminator to obtain a first discrimination result;

inputting a noise signal into the generator to obtain a supplementary training sample; inputting the supplementary training sample into the discriminator to obtain a second discrimination result;

training to obtain the generator and the discriminator based on the first discrimination result and the second discrimination result;

the loss function of the generator is set by a target that the first discrimination result and the second discrimination result are both true, and the loss function of the discriminator is set by a target that the first discrimination result is true and the second discrimination result is false.
The apparatus according to any one of claims 35 to 37, further comprising:

and the sending module is used for sending the encoder to the terminal.
The apparatus according to any one of claims 35 to 37, further comprising:

the receiving module is used for periodically receiving a first CSI feedback codebook based on codebook quantification sent by the terminal;

a training module for constructing a second supplemental training set based on the first CSI feedback codebook through the countermeasure generation network; and updating and training at least one of the encoder and the decoder through the first CSI feedback codebook and the second complementary training set.
The apparatus of claim 39, wherein the apparatus further comprises:

the receiving module is configured to receive a first reporting configuration sent by the terminal, where the first reporting configuration is used to indicate a first reporting parameter of the first CSI feedback codebook;

or alternatively, the first and second heat exchangers may be,

and the sending module is used for sending a first reporting configuration to the terminal, wherein the first reporting configuration is used for indicating a first reporting parameter of the first CSI feedback codebook.
The apparatus of claim 40, wherein the first reporting parameter comprises at least one of:

a time window W;

a frequency window B;

and the configuration parameter T of the first CSI feedback codebook.
The apparatus according to any one of claims 35 to 37, further comprising:

the receiving module is used for receiving a second CSI feedback codebook based on codebook quantification, which is sent by the terminal under the condition that the triggering condition is met;

a training module for constructing a second supplemental training set based on the first CSI feedback codebook through the countermeasure generation network; and updating and training at least one of the encoder and the decoder through the first CSI feedback codebook and the second supplemental training set.
The apparatus according to claim 39 or 42, further comprising:

and the sending module is used for sending the updated encoder to the terminal under the condition of updating and training the encoder.
The apparatus of claim 42, wherein the trigger condition comprises:

the variation value of the channel parameter is larger than a preset threshold value.
The apparatus of claim 42, further comprising:

the receiving module is configured to receive a second reporting configuration sent by the terminal, where the second reporting configuration is used to indicate a second reporting parameter of the second CSI feedback codebook;

Or alternatively, the first and second heat exchangers may be,

and the sending module is used for sending a second reporting configuration to the terminal, wherein the second reporting configuration is used for indicating a second reporting parameter of the second CSI feedback codebook.
The apparatus of claim 45, wherein the second reporting parameter comprises at least one of:

a time window W;

a frequency window B;

and triggering conditions of the second CSI feedback codebook.
A terminal, the terminal comprising:

a processor;

a transceiver coupled to the processor;

a memory for storing executable instructions of the processor;

wherein the processor is configured to load and execute the executable instructions to implement the CSI feedback method of any of claims 1 to 11.
A network device, the network device comprising:

a processor;

a transceiver coupled to the processor;

a memory for storing executable instructions of the processor;

wherein the processor is configured to load and execute the executable instructions to implement the CSI feedback method of any of claims 12 to 23.
A computer readable storage medium having stored therein executable instructions that are loaded and executed by the processor to implement the CSI feedback method of any of claims 1 to 23.
A computer program product having stored therein executable instructions that are loaded and executed by the processor to implement the CSI feedback method of any of claims 1 to 23.
A chip for implementing the CSI feedback method of any of claims 1 to 23.