CN117353849A

CN117353849A - Channel feedback model determining method, terminal and network side equipment

Info

Publication number: CN117353849A
Application number: CN202210748510.1A
Authority: CN
Inventors: 李刚; 韩双锋; 温子睿; 石子烨
Original assignee: China Mobile Communications Group Co Ltd; China Mobile Communications Ltd Research Institute
Current assignee: China Mobile Communications Group Co Ltd; China Mobile Communications Ltd Research Institute
Priority date: 2022-06-28
Filing date: 2022-06-28
Publication date: 2024-01-05
Also published as: WO2024002003A1

Abstract

The invention provides a channel feedback model determining method, a terminal and network side equipment, and relates to the technical field of communication, wherein the method comprises the following steps: receiving a channel state information reference signal sent by network side equipment, and carrying out channel measurement on the channel state information reference signal to obtain channel information; model training is carried out on the target channel feedback model based on the channel information, or the target channel feedback model is selected based on the channel information; and sending the model information of the decoding model in the target channel feedback model to the network side equipment. The embodiment of the invention can reduce the cost of the network side.

Description

Channel feedback model determining method, terminal and network side equipment

Technical Field

The present invention relates to the field of communications, and in particular, to a method, a terminal, and a network side device for determining a channel feedback model.

Background

Currently, machine learning-based channel state information compression feedback is receiving a great deal of attention. For machine learning based channel state information compression feedback, the channel feedback model is generally divided into two parts, namely an encoder (encoder) and a decoder (decoder). In actual deployment, an encoder needs to be deployed at a terminal, a decoder needs to be deployed at a network side, the terminal uses the encoder to compress the estimated channel state information (Channel State Information, CSI) into a string of bit streams, the bit streams are sent to the network side through an uplink feedback channel, and the network side finally recovers the original CSI based on the bit streams. At present, the training of the channel feedback model is deployed on the network side, and the network side needs to acquire a large amount of channel data to train the channel feedback model due to the need of adapting to different terminals, so that the cost of the network side is high.

Disclosure of Invention

The embodiment of the invention provides a channel feedback model determining method, a terminal and network side equipment, which are used for solving the problem that the network side needs to acquire a large amount of channel data to train the channel feedback model, so that the cost of the network side is high.

In order to solve the technical problems, the invention is realized as follows:

in a first aspect, an embodiment of the present invention provides a method for determining a channel feedback model, which is applied to a terminal, where the method includes:

receiving a channel state information reference signal sent by network side equipment, and carrying out channel measurement on the channel state information reference signal to obtain channel information;

model training is carried out on the target channel feedback model based on the channel information, or the target channel feedback model is selected based on the channel information;

and sending the model information of the decoding model in the target channel feedback model to the network side equipment.

Optionally, the model information includes at least one of:

model identification of the decoding model;

all weight coefficients of the decoding model, or part of weight coefficients of the decoding model.

Optionally, the training the target channel feedback model based on the channel information includes:

Training the weight coefficient of a target layer neural network in the decoding model based on the channel information in the process of training the target channel feedback model, and keeping the weight coefficient of a coding model in the target channel feedback model and the weight coefficient of other layer neural networks except the target layer neural network in the decoding model unchanged;

the target layer neural network is at least one layer of neural network of the decoding model.

Optionally, the method further comprises:

transmitting the channel information to the network side equipment, and receiving training configuration for model training, which is transmitted by the network side equipment based on the channel information;

the model training of the target channel feedback model based on the channel information comprises the following steps:

and performing model training on the target channel feedback model based on the training configuration.

Optionally, before the model training of the target channel feedback model based on the channel information, the method further includes:

and selecting a target channel feedback model based on the channel information.

Optionally, the selecting a target channel feedback model based on the channel information includes:

Determining target parameters based on the channel information, wherein the target parameters comprise static characteristic parameters and/or dynamic environment parameters;

and selecting a target channel feedback model from at least two channel feedback models based on the target parameters.

In a second aspect, an embodiment of the present invention provides a method for determining a channel feedback model, which is applied to a network side device, where the method includes:

transmitting a channel state information reference signal to a terminal so that the terminal performs channel measurement on the channel state information reference signal to acquire channel information;

and receiving model information of a decoding model in a target channel feedback model sent by the terminal, wherein the target channel feedback model is obtained by the terminal through model training based on the channel information, or the target channel feedback model is selected by the terminal based on the channel information.

Optionally, the model information includes at least one of:

model identification of the decoding model;

Optionally, before receiving the model information of the decoding model in the target channel feedback model sent by the terminal, the method further includes:

Receiving the channel information sent by the terminal, and sending training configuration for model training to the terminal based on the channel information;

the training configuration is used for carrying out model training on the target channel feedback model.

Optionally, the target channel feedback model is obtained by training a basic model by using training samples for the network side equipment;

before the transmitting training configuration for model training to the terminal based on the channel information, the method further includes:

acquiring a first channel characteristic based on the channel information;

and comparing the first channel characteristics with second channel characteristics, and determining training configuration for model training based on the comparison result, wherein the second channel characteristics are average channel characteristics determined based on the training samples.

Optionally, the first channel characteristic includes at least one of:

angle domain information;

time delay domain information;

doppler domain information.

In a third aspect, an embodiment of the present invention provides a terminal, including:

the receiving module is used for receiving the channel state information reference signal sent by the network side equipment, and carrying out channel measurement on the channel state information reference signal to obtain channel information;

The processing module is used for carrying out model training on the target channel feedback model based on the channel information or selecting the target channel feedback model based on the channel information;

and the first sending module is used for sending the model information of the decoding model in the target channel feedback model to the network side equipment.

Optionally, the model information includes at least one of:

model identification of the decoding model;

Optionally, the processing module is specifically configured to:

Optionally, the apparatus further comprises:

the second sending module is used for sending the channel information to the network side equipment and receiving training configuration for model training sent by the network side equipment based on the channel information;

The processing module is specifically configured to:

Optionally, the apparatus further comprises:

and the selection module is used for selecting a target channel feedback model based on the channel information.

Optionally, the processing module is specifically configured to:

In a fourth aspect, an embodiment of the present invention provides a network side device, where the network side device includes:

the first sending module is used for sending a channel state information reference signal to the terminal so that the terminal can perform channel measurement on the channel state information reference signal to obtain channel information;

the receiving module is used for receiving the model information of the decoding model in the target channel feedback model sent by the terminal, wherein the target channel feedback model is obtained by performing model training on the basis of the channel information for the terminal, or the target channel feedback model is selected on the basis of the channel information for the terminal.

Optionally, the model information includes at least one of:

model identification of the decoding model;

Optionally, the apparatus further comprises:

the second sending module is used for receiving the channel information sent by the terminal and sending training configuration for model training to the terminal based on the channel information;

the apparatus further comprises:

the acquisition module is used for acquiring a first channel characteristic based on the channel information;

and the determining module is used for comparing the first channel characteristics with second channel characteristics, determining training configuration for model training based on the comparison result, wherein the second channel characteristics are average channel characteristics determined based on the training samples.

Optionally, the first channel characteristic includes at least one of:

angle domain information;

time delay domain information;

doppler domain information.

In a fifth aspect, embodiments of the present invention provide a terminal, comprising a transceiver and a processor,

the transceiver is used for receiving a channel state information reference signal sent by the network side equipment, and performing channel measurement on the channel state information reference signal to obtain channel information;

the processor is used for carrying out model training on the target channel feedback model based on the channel information or selecting the target channel feedback model based on the channel information;

the transceiver is further configured to send model information of a decoding model in the target channel feedback model to the network side device.

Optionally, the model information includes at least one of:

model identification of the decoding model;

Optionally, the processor is configured to:

Optionally, the transceiver is further configured to: transmitting the channel information to the network side equipment, and receiving training configuration for model training, which is transmitted by the network side equipment based on the channel information;

the processor is further configured to:

Optionally, the processor is further configured to: and selecting a target channel feedback model based on the channel information.

Optionally, the processor is further configured to:

In a sixth aspect, an embodiment of the present invention provides a network-side device, including a transceiver and a processor,

the transceiver is used for sending a channel state information reference signal to the terminal so that the terminal can perform channel measurement on the channel state information reference signal to obtain channel information;

the transceiver is further configured to receive model information of a decoding model in a target channel feedback model sent by the terminal, where the target channel feedback model is obtained by performing model training on the basis of the channel information for the terminal, or the target channel feedback model is selected on the basis of the channel information for the terminal.

Optionally, the model information includes at least one of:

model identification of the decoding model;

Optionally, the transceiver is further configured to:

the processor is configured to:

acquiring a first channel characteristic based on the channel information;

Optionally, the first channel characteristic includes at least one of:

angle domain information;

time delay domain information;

doppler domain information.

In a seventh aspect, an embodiment of the present invention provides a terminal, including: a processor, a memory, and a program stored on the memory and executable on the processor, which when executed by the processor, implements the steps of the channel feedback model determination method described in the first aspect.

In an eighth aspect, an embodiment of the present invention provides a network side device, including: a processor, a memory, and a program stored on the memory and executable on the processor, which when executed by the processor, implements the steps of the channel feedback model determination method of the second aspect described above.

In a ninth aspect, an embodiment of the present invention provides a computer readable storage medium, where a computer program is stored, where the computer program is executed by a processor to implement the steps of the channel feedback model determining method described in the first aspect; or the computer program when executed by a processor implements the steps of the channel feedback model determination method of the second aspect described above.

In the embodiment of the invention, a terminal receives a channel state information reference signal sent by network side equipment, and channel measurement is carried out on the channel state information reference signal to obtain channel information; model training is carried out on the target channel feedback model based on the channel information, or the target channel feedback model is selected based on the channel information; and sending the model information of the decoding model in the target channel feedback model to the network side equipment. Therefore, the terminal performs model training on the target channel feedback model or selects the target channel feedback model, so that the cost of a network side can be reduced.

Drawings

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

Fig. 1 is a flowchart of a method for determining a channel feedback model according to an embodiment of the present invention;

fig. 2 is a schematic structural diagram of a channel feedback model according to an embodiment of the present invention;

FIG. 3 is a second flowchart of a method for determining a channel feedback model according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of a model information structure according to an embodiment of the present invention;

FIG. 5 is a schematic diagram of a model training result provided by an embodiment of the present invention;

FIG. 6 is a schematic diagram of a classification of a channel feedback model according to an embodiment of the present invention;

FIG. 7a is a schematic diagram of a channel feedback model according to an embodiment of the present invention;

FIG. 7b is a schematic diagram of a channel feedback model according to an embodiment of the present invention;

FIG. 7c is a third schematic diagram of a channel feedback model according to an embodiment of the present invention;

FIG. 7d is a schematic diagram of a channel feedback model according to an embodiment of the present invention;

FIG. 7e is a schematic diagram of a channel feedback model according to an embodiment of the present invention;

FIG. 8 is a third flowchart of a method for determining a channel feedback model according to an embodiment of the present invention;

FIG. 9 is a flowchart of a method for determining a channel feedback model according to an embodiment of the present invention;

FIG. 10 is a flowchart of a method for determining a channel feedback model according to an embodiment of the present invention;

fig. 11 is a schematic structural diagram of a terminal according to an embodiment of the present invention;

fig. 12 is a schematic structural diagram of a network side device according to an embodiment of the present invention;

FIG. 13 is a second schematic diagram of a terminal according to an embodiment of the present invention;

fig. 14 is a second schematic structural diagram of a network side device according to an embodiment of the present invention.

Detailed Description

The following description of the embodiments of the present invention will be made clearly and fully with reference to the accompanying drawings, in which it is evident that the embodiments described are some, but not all embodiments of the invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

In the embodiment of the invention, a channel feedback model determining method, a terminal and network side equipment are provided, so that the problem that the network side needs to acquire a large amount of channel data to train the channel feedback model, and the cost of the network side is high is solved.

Referring to fig. 1, fig. 1 is a flowchart of a channel feedback model determining method provided by an embodiment of the present invention, for a terminal, as shown in fig. 1, where the method includes the following steps:

step 101, receiving a channel state information reference signal sent by a network side device, and performing channel measurement on the channel state information reference signal to obtain channel information.

The channel state information reference signal (Channel state information Reference Signal, CSI-RS) is sent by the network side equipment and used for channel measurement by the terminal to acquire channel information. The channel information may be channel state information (Channel State Information, CSI).

Step 102, training a target channel feedback model based on the channel information, or selecting a target channel feedback model based on the channel information.

The target channel feedback model may include, among other things, an encoding model and a decoding model.

In the process of model training the target channel feedback model, as shown in fig. 2, a part of layers of the decoding model may be frozen, only the rest of the layers outside the frozen part of layers may be trained, the frozen layers may be referred to as frozen layers, and the trained layers may be referred to as fine-tuning layers. In one embodiment, in the process of model training of a target channel feedback model, training the weight coefficient of a target layer neural network in the decoding model based on the channel information, and keeping the weight coefficient of a coding model in the target channel feedback model and the weight coefficient of other layer neural networks except the target layer neural network in the decoding model unchanged, wherein the target layer neural network is at least one layer neural network of the decoding model.

The target channel feedback model may be model trained in two ways:

(1) The training configuration of the online training is determined by the terminal. The terminal can determine the optimal training configuration and the model layer number and weight coefficient which need to be fed back to the network side equipment. The terminal may train starting from the last layer of the decoding model until the performance is not significantly improved. For example, the terminal may train the penultimate layer according to priority 1, train the penultimate layer and the penultimate layer according to priority 2, train the penultimate layer, the penultimate layer and the penultimate layer according to priority 3, …, train the penultimate layer, …, the penultimate layer, and the mth layer according to priority m, and stop training until the training accuracy does not increase, where priority 1 is the highest priority.

(2) A training configuration for online training is determined by the network side device. The network side equipment can configure the training configuration with better terminal configuration according to the historical big data analysis result. The model structure can be divided into a frozen layer and a fine tuning layer, for the frozen layer, the terminal does not update the weight coefficient when the model training is carried out counter propagation, and for the fine tuning layer, the terminal updates the weight coefficient through gradient descent when the model training is carried out counter propagation. The terminal only needs to feed back the weight coefficient of the fine adjustment layer, so that training expenditure and feedback expenditure can be reduced. In one embodiment, the training configuration is as follows: fine tuning layer: 1,2 … m; and (3) a frozen layer: m … n; training the number of iterations.

And step 103, sending the model information of the decoding model in the target channel feedback model to the network side equipment.

The terminal can send the model information of the decoding model in the target channel feedback model after model training to the network side equipment.

In one embodiment, as shown in fig. 3, after the terminal performs online training on the target channel feedback model by using the channel information acquired by the terminal, after uplink transmission resources are acquired through an uplink service resource transmission request, model information of a decoding model in the target channel feedback model may be transmitted in a service data adaptation protocol (Service Data Adaptation Protocol, SDAP) data packet, where, as shown in fig. 4, the model information may include a model ID; model layer number information (e.g., L1 to Ln) that needs to be updated; the model weight coefficients need to be updated. Wherein, the uplink service resource transmission request may include: the terminal sends a scheduling request (Scheduling Request, SR) signaling to a radio access network (Radio Access Network, RAN) through a Physical Uplink Control Channel (PUCCH), the RAN sends an Uplink (UL) grant signaling to the terminal through a physical downlink Control channel (Physical downlink Control channel, PDCCH), the terminal sends a buffer status report (Buffer Status Report, BSR) signaling to the RAN through a medium access Control (Medium Access Control, MAC) Control unit (Control Element, CE), and the RAN sends the UL grant signaling to the terminal through downlink Control information (Downlink Control Information, DCI) or PDCCH.

It should be noted that, the network side device (e.g., the base station) may periodically send CSI-RS; after receiving the CSI-RS, the terminal carries out channel estimation, uses a coding model in a target channel feedback model to carry out compression quantization, and then sends compressed and quantized channel information to network side equipment; and the network side equipment receives the compressed and quantized channel information, decompresses the channel information by adopting the updated decoding model, and acquires complete downlink channel estimation information.

It should be noted that, the coding model and the decoding model can be trained in an end-to-end manner in model training, and are also matched for use in deployment; significant degradation of channel information recovery accuracy can result if a non-matching coding model and decoding model are used. In the related art, the model training is performed by the network side equipment, and the network side equipment is required to collect a large amount of channel data due to the fact that a large amount of calculation power is required for the model training, so that the cost is high. And, because the channel environment where the terminal is located is changeable, the fixed channel feedback model is difficult to obtain better performance in all wireless scenes, namely the generalization problem exists. In the embodiment of the invention, the terminal performs model training on the target channel feedback model or selects the target channel feedback model based on the channel information, so that the problem of reduced recovery precision caused by generalization of the coding model and the decoding model in different environments can be solved.

Optionally, the model information includes at least one of:

model identification of the decoding model;

In this embodiment, the terminal sends the model identifier of the decoding model to the network side device, and/or all weight coefficients of the decoding model or part of weight coefficients of the decoding model, so that the network side device can update the decoding model based on the model identifier and/or the weight coefficients sent by the terminal.

For example, the target channel feedback model may be obtained by training the base model with training samples for the network side device. As shown in fig. 2, the coding model in the base model is composed of four fully connected layers, and the decoding model in the base model is also composed of four fully connected layers. The initial weight coefficient of the basic model can be obtained by training the network side equipment through the collected big data, and the terminal can select three channel data sets with different sizes to retrain the basic model. For example, as shown in FIG. 5, the base model may be retrained by a large dataset, a medium dataset, and a small dataset. And, the base model may be retrained by freezing different layers of the base model. As can be seen from fig. 5, there is a better training strategy for retraining at the terminal side, where the Normalized Mean Square Error (NMSE) of the last layer is higher, but the NMSE drops rapidly after training the last two layers, and the marginal effect is reduced afterwards, and the last two layers are the choices of better balance efficiency and performance. The performance of the decoding model may be optimized by optimizing the weight coefficients of the partial layers of the decoding model. For the terminal, only the layer weight coefficient of the decoding model of the feedback part can greatly reduce the overhead of air interface transmission.

In this embodiment, in the process of model training on the target channel feedback model, the weight coefficient of the target layer neural network in the decoding model is trained based on the channel information, and the weight coefficient of the coding model in the target channel feedback model and the weight coefficients of the other layer neural networks in the decoding model except the target layer neural network are kept unchanged, so that only the weight coefficient of a part of layers is trained during model training, and only the weight coefficient of a part of layers of the decoding model is transferred to the network side device, and overhead of air interface transmission can be reduced.

Optionally, the method further comprises:

In an implementation manner, a terminal may send an uplink sounding reference signal (Sounding Reference Signal, SRS) to a network side device, and receive a training configuration for model training sent by the network side device, where the training configuration may be determined based on channel characteristics, and the channel characteristics may be determined based on channel information sent by the terminal and the SRS.

It should be appreciated that in a wireless communication system, multipath, fading, time-varying, frequency-selective characteristics of the wireless channel can have a significant impact on the performance of the information transmission. The wireless channel characteristics comprise parameters such as the number, angle and multipath time delay of uplink and downlink multipaths, and the like, for a time division multiplexing (Time Division Duplex, TDD) system, the wireless channel has good interoperability, a terminal can send an uplink sounding reference signal SRS for acquiring channel information to network side equipment, the network side equipment acquires an uplink channel matrix H through measuring the SRS, and utilizes uplink and downlink reciprocity to obtain a downlink channel matrix H ', and calculates a precoding matrix and a beam forming vector for downlink data transmission through the downlink channel matrix H'. For a frequency division multiplexing (Frequency Division Duplex, FDD) system, uplink and downlink are deployed on different frequencies, and complete channel reciprocity does not exist, and angles and time delays of various paths on a wireless channel are mainly determined by factors such as a spatial angle relation of a receiving and transmitting end, a position and a material of a reflector in a wireless environment, and the like, and the frequencies of wireless signals do not have strong relation. The frequency of the wireless channel can greatly influence the components such as path loss, penetration loss, polarization leakage factors and the like of the wireless channel, so that the amplitude variation and the phase variation experienced by each path can be greatly influenced. Therefore, the frequency domain base vectors and the space domain base vectors of the uplink and downlink channels in the FDD system have stronger correlation, and the weighting coefficients represent amplitude changes and phase changes experienced by each path in the wireless channel, and are independent, namely, have partial reciprocity.

In one embodiment, the network side device may obtain a part of the channel characteristics of the downlink channel according to the part of the reciprocity of the channel, and may also obtain the channel characteristics according to the channel information fed back by the terminal.

In this embodiment, the terminal sends the channel information to the network side device, and receives training configuration for model training sent by the network side device based on the channel information; and the terminal carries out model training on the target channel feedback model based on the training configuration, so that the training configuration can be determined through the network side equipment.

and selecting a target channel feedback model based on the channel information.

Wherein the static characteristic parameters may include, but are not limited to: network configuration parameters such as macro/micro station, network side multi-antenna port number, etc.; terminal capability parameters such as terminal multi-antenna port number, terminal AI reasoning capability, etc. The static characteristic parameters generally belong to information to be reported when the terminal is initially accessed into the network, and basically remain unchanged in the occurrence process of communication service. The dynamic environment parameters can correspond to various dynamic scenes, the dynamic environment parameters can comprise channel environment parameters, channel quality parameters and the like, the channel environment parameters can be channel environment parameters under line-of-sight or non-line-of-sight transmission, and the channel quality parameters can comprise signal-to-noise ratio ranges and the like.

In one embodiment, the network side device may form different training sets based on the target parameters in the stage of training the basic model, train a plurality of models based on the different training sets, and divide the plurality of models into a plurality of categories according to different applicable scenes.

In one embodiment, the terminal may select the target channel feedback model from multiple models, as shown in fig. 6, where the multiple models may be classified according to static feature parameters to obtain multiple semi-static classes, and then each sub-class determined by the static feature parameters may be further classified according to dynamic environment parameters, and each sub-class may be divided into multiple dynamic sub-class models.

It should be noted that there may be a coding model or a decoding model sharing among multiple models within each subclass. The coding model may be referred to as an encoder and the decoding model may be referred to as a decoder. As shown in fig. 7a to 7e, 5 modes of coding model-decoding model are enumerated, one decoding model may be shared by a plurality of coding models, or one coding model may be shared by a plurality of decoding models. For the 5 modes listed, there may be two categories, acde/b, or abde/c, depending on whether there are multiple alternative coding models or decoding models. Each pair of coding model-decoding model corresponds to a group of dynamic categories, and when in actual use, a specific group of coding model-decoding model can be selected to compress and recover the CSI.

In one embodiment, the target channel feedback model is selected from at least two channel feedback models based on the target parameters, which may be that, for each pair of coding model-decoding models, the applicable static characteristic parameters and/or dynamic environment parameters are marked, and the corresponding coding model-decoding model is selected according to the target parameters in a dictionary searching manner.

In one embodiment, the at least two channel feedback models may be divided into a plurality of sub-classes according to the applicable static characteristic parameters and/or dynamic environment parameters, and the target channel feedback model is selected from the at least two channel feedback models based on the target parameters, or the coding model-decoding model of the corresponding sub-class is selected according to the target parameters in a dictionary searching manner, the coding model-decoding model in the selected sub-class is sequentially used for compressing and recovering the channel information, and recovering precision is calculated, the models in the sub-class are traversed, and then the models with highest recovering precision are selected as the target channel feedback model according to the recovering precision sequence.

In addition, after selecting the target channel feedback model, the terminal may notify the network side device of the decoding model of the selected target channel feedback model, where the notification may be by sending a signaling carrying the decoding model to the network side device, and when the coding model or the decoding model has a mode sharing or even only a single coding model or decoding model exists in a certain sub-category, the decoding model may not be synchronized, and whether the decoding model of the target channel feedback model selected by the opposite terminal is notified is specified by the network side device.

In one embodiment, the transfer of the model information of the decoding model may be implemented by expanding the radio resource control (Radio Resource Control, RRC) or MAC CE signaling, and taking the transfer of the model information of the decoding model implemented by the MAC CE signaling as an example, an LCG ID field may be added in the MAC CE signaling, and the field carries the model identifier of the decoding model.

In this embodiment, the target channel feedback model is selected from at least two channel feedback models based on the static characteristic parameter and/or the dynamic environment parameter, so that a channel feedback model that is relatively matched with the current static characteristic parameter and/or the dynamic environment parameter can be obtained.

Referring to fig. 8, fig. 8 is a flowchart of a channel feedback model determining method provided by an embodiment of the present invention, for a network side device, as shown in fig. 8, where the method includes the following steps:

step 201, a channel state information reference signal is sent to a terminal, so that the terminal performs channel measurement on the channel state information reference signal to obtain channel information.

Step 202, receiving model information of a decoding model in a target channel feedback model sent by the terminal, wherein the target channel feedback model is obtained by performing model training on the basis of the channel information by the terminal, or the target channel feedback model is selected on the basis of the channel information by the terminal.

It should be noted that, as an implementation manner of the network side device corresponding to the embodiment shown in fig. 1, a specific implementation manner of the embodiment may refer to a related description in the embodiment shown in fig. 1, and in order to avoid repeated description, the embodiment is not repeated.

Optionally, the model information includes at least one of:

model identification of the decoding model;

The above optional implementation manner may refer to the related description in the embodiment shown in fig. 1, and in order to avoid repetitive description, this embodiment is not repeated.

acquiring a first channel characteristic based on the channel information;

In this embodiment, a first channel characteristic is obtained based on the channel information; and comparing the first channel characteristics with second channel characteristics, and determining training configuration for model training based on the comparison result, wherein the second channel characteristics are average channel characteristics determined based on the training samples. In this way, the training configuration can be determined by comparing the average channel characteristics determined by training samples of the training basic model with the first channel characteristics, and the proper training configuration can be determined for different first channel characteristics, so that a better model training effect can be obtained.

Optionally, the first channel characteristic includes at least one of:

angle domain information;

time delay domain information;

doppler domain information.

The angle domain information reflects sparsity of beams/antennas, for example, whether a duty ratio of the first 3 beams with the largest path gain indicated by the angle domain information in the overall power gain exceeds a first threshold may determine different training configurations, where the first threshold is determined by a second channel feature, and in an embodiment, the duty ratio of the first 3 beams with the largest path gain indicated by the angle domain information in the second channel feature in the overall power gain is the first threshold, which may be 80% by way of example.

In addition, the delay domain information reflects the sparsity of the delay domain/multipath, non Line of Sight (Non Line of Sight, NLOS) or Line of Sight (LOS) channels, for example, the different training configurations may be determined by whether the K factor indicated by the delay domain information (i.e., the power ratio occupied by the primary path) exceeds a second threshold, which is determined by the second channel characteristics, and in one embodiment, the K factor indicated by the delay domain information in the second channel characteristics is a second threshold, which may be 50% by way of example.

In addition, the doppler domain information reflects sparsity in a time domain, for example, whether the frequency offset power ratio of the first 5 main paths indicated by the doppler domain information exceeds a third threshold may determine a different training configuration, where the third threshold is determined by the second channel characteristic, and in one embodiment, the frequency offset power ratio of the first 5 main paths indicated by the doppler domain information in the third channel characteristic is the third threshold, which may be 60% by way of example.

In one embodiment, the ratio of the first 3 beams with the largest path gains indicated by the angle domain information in the overall power gain is a, the K factor indicated by the delay domain information is B, and the ratio of the frequency offset power of the first 5 main paths indicated by the doppler domain information is C. As shown in fig. 2, the coding model is composed of four fully connected layers, and the decoding model is composed of four fully connected layers. The model structure is divided into a frozen layer and a fine tuning layer, for the frozen layer, the terminal does not update the weight coefficient when the model training is carried out counter propagation, and for the fine tuning layer, the terminal updates the weight coefficient through gradient descent when the model training is carried out counter propagation.

At a >80% or B >50%, the training configuration for model training may be: the number of layers of the frozen layer is more, and the fine tuning layer is a fourth layer of the decoding model;

at a >50% or B >50%, the training configuration for model training may be: the number of layers of the frozen layer is less, and the fine tuning layer is a third layer and a fourth layer of the decoding model;

at C >60%, the training configuration for model training may be: the number of continuous samples is small, the number of samples is 100, and the duration is 500ms;

At C <60%, the training configuration for model training may be: the number of consecutive samples is large, the number of samples is 1000, and the duration is 1s.

The method for determining the channel feedback model according to the embodiment of the present invention is described below by two specific embodiments:

example 1:

taking network side equipment as a base station as an example, the coding model and the decoding model are subjected to online training by the terminal to determine a channel feedback model, and specifically, as shown in fig. 9, the method comprises the following steps:

(1) Deploying a channel feedback model, the channel feedback model including an encoding model and a decoding model, which may be, for example, a base station transmitting the channel feedback model to a terminal, or may be preconfigured with the channel feedback model at the terminal;

(2) The base station sends the CSI-RS configuration through the RRC reconfiguration message;

(3) The base station periodically transmits the CSI-RS to the terminal according to the CSI-RS configuration;

(4) The terminal receives the CSI-RS and measures the CSI-RS to acquire channel state information;

the terminal sends a CSI report to the base station and sends an SRS to the base station;

(5) The base station selects the configuration of online training according to the channel characteristics;

(6) The base station sends the configuration of the online training to the terminal;

(7) The terminal carries out online training on the coding model and the decoding model based on the channel state information;

(8) The terminal sends the model information of the trained decoding model to the base station;

(9) The base station updates a decoding model based on the received model information;

(10) The base station periodically transmits the CSI-RS to the terminal;

(11) The terminal compresses the CSI by using a trained coding model, and carries out quantization processing on the compressed CSI;

(12) The terminal reports the compressed and quantized CSI to the base station;

(13) And the base station decodes the CSI by using the updated decoding model.

Example 2:

taking a network side device as a base station for example, the determination of the channel feedback model is realized by selecting the channel feedback model by the terminal, specifically, as shown in fig. 10, the method comprises the following steps:

(5) The terminal selects a decoding model according to the channel state information obtained by measurement;

(6) The terminal sends the model information of the selected decoding model to the base station;

(7) The base station updates a decoding model based on the received model information;

(8) The base station periodically transmits the CSI-RS to the terminal;

(9) The terminal compresses the CSI by using the coding model, and carries out quantization processing on the compressed CSI;

(10) The terminal reports the compressed and quantized CSI to the base station;

(11) And the base station decodes the CSI by using the updated decoding model.

Referring to fig. 11, fig. 11 is a schematic structural diagram of a terminal according to an embodiment of the present invention, and as shown in fig. 11, a terminal 300 includes:

a receiving module 301, configured to receive a channel state information reference signal sent by a network side device, and perform channel measurement on the channel state information reference signal to obtain channel information;

a processing module 302, configured to perform model training on a target channel feedback model based on the channel information, or select a target channel feedback model based on the channel information;

and a first sending module 303, configured to send, to the network side device, model information of a decoding model in the target channel feedback model.

Optionally, the model information includes at least one of:

model identification of the decoding model;

Optionally, the processing module is specifically configured to:

Optionally, the apparatus further comprises:

the processing module is specifically configured to:

Optionally, the apparatus further comprises:

Optionally, the processing module is specifically configured to:

The terminal can implement each process implemented by the method embodiment shown in fig. 1, and can achieve the same beneficial effects, and in order to avoid repetition, a detailed description is omitted here.

Referring to fig. 12, fig. 12 is a schematic structural diagram of a network side device according to an embodiment of the present invention, and as shown in fig. 12, a network side device 400 includes:

a first sending module 401, configured to send a channel state information reference signal to a terminal, so that the terminal performs channel measurement on the channel state information reference signal to obtain channel information;

and the receiving module 402 is configured to receive model information of a decoding model in a target channel feedback model sent by the terminal, where the target channel feedback model is obtained by performing model training on the basis of the channel information for the terminal, or the target channel feedback model is selected on the basis of the channel information for the terminal.

Optionally, the model information includes at least one of:

model identification of the decoding model;

Optionally, the apparatus further comprises:

the apparatus further comprises:

Optionally, the first channel characteristic includes at least one of:

angle domain information;

time delay domain information;

doppler domain information.

The network side device can implement each process implemented by the method embodiment shown in fig. 8, and can achieve the same beneficial effects, so that repetition is avoided, and details are not repeated here.

The embodiment of the invention also provides a terminal, which comprises: the system comprises a processor, a memory and a program stored in the memory and capable of running on the processor, wherein the program realizes each process of the channel feedback model determining method embodiment when being executed by the processor, and can achieve the same technical effect, and the repetition is avoided, and the description is omitted here.

Specifically, referring to fig. 13, the embodiment of the present invention further provides a terminal, which includes a bus 501, a transceiver 502, an antenna 503, a bus interface 504, a processor 505, and a memory 506.

The transceiver 502 is configured to receive a channel state information reference signal sent by a network side device, and perform channel measurement on the channel state information reference signal to obtain channel information;

the processor 505 is configured to perform model training on a target channel feedback model based on the channel information, or select a target channel feedback model based on the channel information;

the transceiver 502 is further configured to send, to the network side device, model information of a decoding model in the target channel feedback model.

Optionally, the model information includes at least one of:

model identification of the decoding model;

Optionally, the processor 505 is configured to:

Optionally, the transceiver 502 is further configured to: transmitting the channel information to the network side equipment, and receiving training configuration for model training, which is transmitted by the network side equipment based on the channel information;

the processor 505 is further configured to:

Optionally, the processor 505 is further configured to: and selecting a target channel feedback model based on the channel information.

Optionally, the processor 505 is further configured to:

In fig. 13, a bus architecture (represented by bus 501), the bus 501 may include any number of interconnected buses and bridges, with the bus 501 linking together various circuits, including one or more processors, represented by processor 505, and memory, represented by memory 506. The bus 501 may also link together various other circuits such as peripheral devices, voltage regulators, power management circuits, etc., which are well known in the art and, therefore, will not be described further herein. Bus interface 504 provides an interface between bus 501 and transceiver 502. The transceiver 502 may be one element or may be multiple elements, such as multiple receivers and transmitters, providing a means for communicating with various other apparatus over a transmission medium. The data processed by the processor 505 is transmitted over a wireless medium via the antenna 503, and further, the antenna 503 receives the data and transmits the data to the processor 505.

The processor 505 is responsible for managing the bus 501 and general processing, and may also provide various functions including timing, peripheral interfaces, voltage regulation, power management, and other control functions. And memory 506 may be used to store data used by processor 505 in performing operations.

Alternatively, the processor 505 may be CPU, ASIC, FPGA or a CPLD.

The embodiment of the invention also provides network side equipment, which comprises: the system comprises a processor, a memory and a program stored in the memory and capable of running on the processor, wherein the program realizes each process of the channel feedback model determining method embodiment when being executed by the processor, and can achieve the same technical effect, and the repetition is avoided, and the description is omitted here.

Specifically, referring to fig. 14, the embodiment of the present invention further provides a network side device, which includes a bus 601, a transceiver 602, an antenna 603, a bus interface 604, a processor 605 and a memory 606.

The transceiver 602 is configured to send a channel state information reference signal to a terminal, so that the terminal performs channel measurement on the channel state information reference signal to obtain channel information;

the transceiver 602 is further configured to receive model information of a decoding model in a target channel feedback model sent by the terminal, where the target channel feedback model is obtained by performing model training for the terminal based on the channel information, or the target channel feedback model is selected for the terminal based on the channel information.

Optionally, the model information includes at least one of:

model identification of the decoding model;

Optionally, the transceiver 602 is further configured to:

the processor 605 is configured to:

acquiring a first channel characteristic based on the channel information;

Optionally, the first channel characteristic includes at least one of:

angle domain information;

time delay domain information;

doppler domain information.

In fig. 14, a bus architecture (represented by bus 601), the bus 601 may include any number of interconnected buses and bridges, with the bus 601 linking together various circuits, including memory, represented by one or more processors 605 and memory 606, represented by processor 605605. The bus 601 may also link together various other circuits such as peripheral devices, voltage regulators, power management circuits, etc., which are well known in the art and, therefore, will not be described further herein. Bus interface 604 provides an interface between bus 601 and transceiver 602. The transceiver 602 may be one element or may be multiple elements, such as multiple receivers and transmitters, providing a means for communicating with various other apparatus over a transmission medium. The data processed by the processor 605 is transmitted over a wireless medium via an antenna 603, and further, the antenna 603 also receives data and transmits the data to the processor 605.

The processor 605 is responsible for managing the bus 601 and general processing, and may also provide various functions including timing, peripheral interfaces, voltage regulation, power management, and other control functions. And memory 606 may be used to store data used by processor 605 in performing operations.

Alternatively, the processor 605 may be CPU, ASIC, FPGA or a CPLD.

The embodiment of the invention also provides a computer readable storage medium, on which a computer program is stored, which when executed by a processor, implements the processes of the above-mentioned channel feedback model determining method embodiment, and can achieve the same technical effects, so that repetition is avoided, and no further description is given here. Wherein the computer readable storage medium is such as ROM, RAM, magnetic or optical disk.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or apparatus that comprises the element.

From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) comprising instructions for causing a terminal (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present invention.

The embodiments of the present invention have been described above with reference to the accompanying drawings, but the present invention is not limited to the above-described embodiments, which are merely illustrative and not restrictive, and many forms may be made by those having ordinary skill in the art without departing from the spirit of the present invention and the scope of the claims, which are to be protected by the present invention.

Claims

1. A method for determining a channel feedback model, applied to a terminal, the method comprising:

2. The method of claim 1, wherein the model information comprises at least one of:

model identification of the decoding model;

3. The method of claim 1, wherein said model training a target channel feedback model based on said channel information comprises:

4. The method according to claim 1, wherein the method further comprises:

5. The method of claim 1, wherein prior to model training the target channel feedback model based on the channel information, the method further comprises:

and selecting a target channel feedback model based on the channel information.

6. The method according to claim 1 or 5, wherein said selecting a target channel feedback model based on said channel information comprises:

7. A method for determining a channel feedback model, applied to a network side device, the method comprising:

8. The method of claim 7, wherein prior to receiving the model information of the decoding model in the target channel feedback model transmitted by the terminal, the method further comprises:

9. The method according to claim 8, wherein the target channel feedback model is obtained by training a basic model for the network side device by using training samples;

acquiring a first channel characteristic based on the channel information;

10. The method of claim 9, wherein the first channel characteristics comprise at least one of:

angle domain information;

time delay domain information;

doppler domain information.

11. A terminal, the terminal comprising:

12. A network side device, characterized in that the network side device comprises:

13. A terminal, comprising a transceiver and a processor,

14. A network side device is characterized by comprising a transceiver and a processor,

15. A terminal, comprising: a processor, a memory and a program stored on the memory and executable on the processor, which when executed by the processor implements the steps of the channel feedback model determination method of any of claims 1 to 6.

16. A network side device, comprising: a processor, a memory and a program stored on the memory and executable on the processor, which when executed by the processor implements the steps of the channel feedback model determination method of any of claims 7 to 10.

17. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the steps of the channel feedback model determination method according to any of claims 1 to 6; or the computer program when executed by a processor, implements the steps of the channel feedback model determination method as claimed in any one of claims 7 to 10.