WO2023160336A1

WO2023160336A1 - Channel compression method and apparatus, channel recovery method and apparatus, and device

Info

Publication number: WO2023160336A1
Application number: PCT/CN2023/073892
Authority: WO
Inventors: 李刚; 李宇鹏; 韩双锋
Original assignee: 中国移动通信有限公司研究院; 中国移动通信集团有限公司
Priority date: 2022-02-24
Filing date: 2023-01-30
Publication date: 2023-08-31
Also published as: CN116723066A

Abstract

Provided in the present application are a channel compression method and apparatus, a channel recovery method and apparatus, and a device. The channel compression method is applied to a terminal, and comprises: measuring a reference signal, which is sent by a network device, so as to obtain channel state information; preprocessing the channel state information to obtain a preprocessing result; according to the preprocessing result, performing compression processing on the channel state information by using a target channel compression model, so as to obtain compressed channel information; and sending the compressed channel information and/or the pre-processing result to the network device. In the solution of the present application, channel state information is preprocessed, and a trained target channel compression model is used to compress the preprocessed channel state information and perform subsequent channel recovery, such that the channel compression feedback performance is improved.

Description

A channel compression and recovery method, device and equipment

Cross References to Related Applications

This application is based on a Chinese patent application with application number 202210171751.4 and a filing date of February 24, 2022, and claims the priority of this Chinese patent application. The entire content of this Chinese patent application is hereby incorporated into this application in its entirety .

technical field

The present application relates to the technical field of communication, in particular to a channel compression and recovery method, device and equipment.

Background technique

In a multi-antenna (Multiple-input Multiple-output, MIMO) system, obtaining channel state information (channel state information, CSI) is a key condition for beamforming to improve transmission performance. For an FDD (Frequency Division Duplex) system, since there is no complete reciprocity between uplink and downlink channels, the base station needs to obtain complete downlink CSI through terminal feedback. In the NR system, the CSI fed back by the terminal mainly depends on the codebook. Currently, it supports codebook types such as CSI type I, type II, and type II enhanced, which are used to feed back information such as RI (rank indicator), PMI (Precoding matrix indicator), and CQI (Channel quality indicator).

Traditional channel feedback schemes based on machine learning treat the channel matrix as a whole and directly input the model. Although this method is simple to implement, its performance is often limited due to the lack of design that combines channel characteristics.

Contents of the invention

The technical problem to be solved in this application is to provide a channel compression and recovery method, device and equipment to improve channel processing performance and reduce channel feedback overhead.

In order to solve the problems of the technologies described above, the technical scheme of the present application is as follows:

An embodiment of the present application provides a channel compression method, which is applied to a terminal, and the method includes:

Measure the reference signal sent by the network equipment to obtain channel state information;

Preprocessing the channel state information to obtain a preprocessing result;

According to the preprocessing result, the channel state information is compressed using the target channel compression model to obtain compressed channel information;

Send the compressed channel information and/or the preprocessing result to the network device.

Optionally, the channel state information is preprocessed to obtain a preprocessing result, including one of the following:

Transforming the channel state information into an angle domain, a delay domain or a Doppler domain to obtain a first preprocessing result;

performing cutting preprocessing on the channel state information or the first preprocessing result to obtain a second preprocessing result;

Perform cyclic shift preprocessing on the channel state information or the first preprocessing result to obtain a third preprocessing result.

Optionally, transforming the channel state information into an angle domain, a delay domain or a Doppler domain includes:

The channel state information is transformed into the angle domain through the fast inverse Fourier transform IFFT in the antenna dimension, transformed into the delay domain through the fast inverse inverse Fourier transform IFFT in the frequency dimension, and transformed into the delay domain through the sparse Fourier transform SFFT in the time dimension Transform into the Doppler domain.

Optionally, performing clipping preprocessing on the channel state information to obtain a second preprocessing result, including:

When performing the first pruning preprocessing on the channel state information, retain a first number of elements, and pruning the remaining elements to obtain a first pruning preprocessing result, the first number is a first preset value or The number of elements to be retained obtained by filtering according to the first preset threshold; or

When the second clipping preprocessing is performed on the channel state information, the position of the reserved second number of elements is marked in a bitmap manner, and the remaining elements are clipped to obtain a second clipping preprocessing result, the second The quantity is the second preset value or the number of elements to be kept obtained by filtering according to the second preset threshold.

Optionally, the first trimming preprocessing result includes: the position of the first number of elements and the remaining trimmed elements;

The second clipping preprocessing result includes: marking the positions of the second number of reserved elements and the rest of the clipped elements in a bitmap manner.

Optionally, the first number and the second number are pre-configured by the network device or determined by the terminal;

When the first number and the second number are pre-configured by network equipment, they are pre-configured to the terminal through high-layer signaling;

When the first quantity and the second quantity are determined by the terminal, report the first quantity and the second quantity determined by the terminal to the network device.

Optionally, perform cyclic shift preprocessing on the channel state information to obtain a third preprocessing result, including:

If the first element of the channel satisfies the preset condition, the target element of the channel is cyclically shifted to the position of the first element, and the corresponding position of the target element is recorded to obtain a third preprocessing result.

Optionally, the third preprocessing result includes: a channel obtained after the cyclic shift, and a position corresponding to the target element.

Optionally, transforming the channel state information into an angle domain, a delay domain or a Doppler domain through Fourier transform, including:

The channel state information is transformed into the angle domain through the fast Fourier transform FFT, transformed into the delay domain through the inverse fast Fourier transform IFFT, and transformed into the Doppler domain through the sparse Fourier transform SFFT.

Optionally, perform clipping preprocessing on the first preprocessing result to obtain a second preprocessing result, including:

Based on the first preprocessing result, the first number of elements around the channel delay and/or the maximum value of the angle are reserved, and the remaining elements are cut to obtain a second preprocessing result, and the first number is the first A preset value or the number of elements to be retained obtained by filtering according to the first preset threshold.

Optionally, perform cyclic shift preprocessing on the first preprocessing result to obtain a third preprocessing result, including:

Based on the first preprocessing result, if it is judged that if the first element of the channel is not the channel delay and/or the maximum value of the retention angle, the sequential cyclic shift is started from the element of the channel delay and/or the maximum value of the retention angle, The channel delay and/or the maximum value of the retention angle are cyclically shifted to the position of the first element, and the position corresponding to the channel delay and/or the maximum value of the retention angle is recorded to obtain a third preprocessing result.

Optionally, the channel state information includes at least one of the following:

channel matrix;

channel feature vector;

Channel precoding matrix.

Optionally, the target channel compression model is to select a target machine learning model from a plurality of pre-trained machine learning models with different crop ratios; the multiple machine learning models with different crop ratios are performed through the following process train:

Obtain the downlink channel data set generated by the simulation model or formed by the frequency domain multiplexing downlink channel reported by the terminal;

performing a plurality of trimming processes with different preset trimming ratios on the downlink channels in the downlink channel data set, and obtaining trimming processing results with a plurality of different preset trimming ratios;

According to the cutting processing results of a plurality of different preset cutting ratios, they are respectively input into the preset machine learning models for training, and a plurality of machine learning models of different cutting ratios are obtained.

Embodiments of the present application also provide a method for channel recovery, which is applied to network equipment, and the method includes:

sending a reference signal to the terminal;

receiving the compressed channel information fed back by the terminal and/or the preprocessing result obtained by preprocessing the estimated channel state information when compressing the channel; the compressed channel information is the channel state information obtained by the terminal according to the channel measurement of the reference signal, and performing preprocessing on the channel state information to obtain a preprocessing result; and performing compression processing on the channel state information by using a target channel compression model according to the preprocessing result;

Restoring the compressed channel information according to the target channel decompression model to obtain the restored channel information.

Optionally, the preprocessing result includes one of the following:

A first preprocessing result obtained by transforming the channel state information into an angle domain, a delay domain or a Doppler domain;

A second preprocessing result obtained by cutting and preprocessing the channel state information or the first preprocessing result;

A third preprocessing result obtained by performing cyclic shift preprocessing on the channel state information or the first preprocessing result.

Optionally, the compressed channel information is recovered according to the target channel decompression model, and the recovered channel information includes:

According to the target channel decompression model and the preprocessing result, perform compression on the compressed channel information Restoring to obtain the restored channel information.

Optionally, the target channel decompression model is to select a target machine learning model from a plurality of pre-trained machine learning models with different cropping ratios; the multiple machine learning models with different cropping ratios are performed through the following process train:

Embodiments of the present application also provide a terminal, including:

An estimation module, configured to measure a reference signal sent by a network device to obtain channel state information;

A processing module, configured to preprocess the channel state information to obtain a preprocessing result; according to the preprocessing result, use a target channel compression model to compress the channel state information to obtain compressed channel information;

A transceiver module, configured to send the compressed channel information and/or the preprocessing result to a network device.

Embodiments of the present application also provide a network device, including:

The transceiver module is configured to send a reference signal to the terminal; and receive the compressed channel information fed back by the terminal and/or the preprocessing result obtained by preprocessing the channel state information when the channel is compressed; the compressed channel information is the terminal pair performing channel measurement on a reference signal to obtain channel state information, and performing preprocessing on the channel state information to obtain a preprocessing result; and performing compression processing on the channel by using a target channel compression model according to the preprocessing result;

The processing module is used to restore the compressed channel information according to the target channel decompression model, and obtain the restored channel information.

An embodiment of the present application further provides a communication device, including: a processor, and a memory storing a computer program, and when the computer program is run by the processor, the method as described above is executed.

An embodiment of the present application is also a computer-readable storage medium, which stores instructions, and when the instructions are run on a computer, the computer executes the method as described above

The above-mentioned scheme of the present application at least includes the following beneficial effects:

In this application, the channel state information is obtained by measuring the reference signal sent by the network equipment; the channel state information is preprocessed to obtain the preprocessing result; according to the preprocessing result, the channel state information is processed by using the target channel compression model Compression processing to obtain compressed channel information; sending the compressed channel information and/or the preprocessing result to the network device; thereby improving the performance of channel compression and the accuracy of subsequent channel recovery.

Description of drawings

FIG. 1 is a flowchart of a channel method on the terminal side provided in an embodiment of the present application;

FIG. 2 is an implementation schematic diagram of channel compression and restoration provided by an embodiment of the present application;

Fig. 3 is the implementation flow diagram of the channel compression and restoration method provided by the embodiment of the present application;

FIG. 4 is a schematic block diagram of a terminal module provided by an embodiment of the present application.

Detailed ways

Exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the present application can be more thoroughly understood, and the scope of the present application can be fully conveyed to those skilled in the art.

As shown in FIG. 1, the embodiment of the present application provides a channel compression method, which is applied to a terminal, and the method includes:

Step 11, measuring the reference signal sent by the network equipment to obtain channel state information;

Step 12, performing preprocessing on the channel state information to obtain a preprocessing result;

Step 13, according to the preprocessing result, use the target channel compression model to compress the channel state information, and obtain the compressed channel information;

Step 14, sending the compressed channel information and/or the preprocessing result to the network device.

In this embodiment, the reference signal is periodically sent by the network device to the terminal and is used for channel information measurement. The reference signal may be a CSI-RS channel state information acquisition signal, etc.; signal to obtain channel state information; the terminal preprocesses the channel state information according to the characteristics of the data elements in the channel state information, so as to remove unnecessary data element characteristics in the channel state information and reduce the impact of unnecessary data characteristics on The impact of subsequent processing steps; the preprocessing The results include but are not limited to the preprocessed channel information, the data element information in the preprocessed channel information, and the data element information in the channel information removed during the preprocessing process;

According to the preprocessing result, the terminal selects a target channel compression model that is compatible with the preprocessed channel state information from the set of pre-trained target channel compression models, and uses it as the target channel compression model for the preprocessed Compress and quantize the channel state information to improve the performance of channel compression, and the information in the compressed channel can exist in the form of bit stream; the target channel compression model set is based on the data of multiple historical channels, according to different The clipping ratio is obtained by clipping and training multiple historical channels. The target channel compression model can be obtained by processing and training the data of the historical channels by the terminal, or obtained by processing and training the data of the historical channels by the network equipment. of;

The terminal sends the compressed channel information and/or the preprocessing result to the network device, so that the network device restores the compressed channel.

In an optional embodiment of the present application, the preprocessing result described in step 12 above may include one of the following:

In this embodiment, the channel state information may be transformed into an angle domain, a delay domain or a Doppler domain;

Cutting preprocessing may also be performed on the channel state information;

Cyclic shift preprocessing may also be performed on the channel state information;

It is also possible to perform Fourier transform on the channel state information in the antenna dimension to the angle domain, delay domain or Doppler domain, and then perform clipping preprocessing and/or cyclic shift based on the result of the Fourier transform. bit preprocessing;

Wherein, performing clipping preprocessing and cyclic shift preprocessing on the channel state information is to perform clipping preprocessing and cyclic shift preprocessing on the near-zero values in the estimated channel;

Cutting and cyclically shifting the near-zero value to filter out interference items in the channel, transforming the channel state information into the angle domain, time delay domain or Doppler domain through Fourier transform, avoiding the subsequent channel processing The influence of unnecessary features improves the accuracy of subsequent channel recovery processing.

In an optional embodiment of the present application, Fourier transforming the channel state information into the angle domain, delay domain or Doppler domain may include:

The channel state information is transformed to the angle domain by fast inverse Fourier transform (IFFT, Inverse Fast Fourier Transform) in the antenna dimension, and transformed to the time domain by fast inverse Fourier transform (IFFT, Inverse Fast Fourier Transform) in the frequency dimension. The extended domain is transformed into the Doppler domain by Sparse Fast Fourier Transform (SFFT, Sparse Fast Fourier Transform) in the time dimension.

In this embodiment, when the channel state information is subjected to Fourier transform, according to the configuration of the network device, IDFT is performed on the channel state information to the angle domain according to the row N1 point and the column N2 point of the antenna dimension; Perform IDFT transformation on the channel state information to the delay domain according to the frequency domain dimension N3 point; perform SFFT transformation on the channel state information to the Doppler domain according to the time dimension N4 point, where N1, N2, N3 and N4 are also Can be configured by network devices.

In an optional embodiment of the present application, performing cutting preprocessing on the channel state information to obtain a second preprocessing result may include:

When performing the first pruning preprocessing on the channel state information, retaining a first number of elements, and pruning the remaining elements to obtain a first pruning preprocessing result, the first number is a first preset value or The number of elements to be retained obtained by filtering according to the first preset threshold; or,

In this embodiment, the result of the first trimming preprocessing or the second trimming preprocessing result is the second preprocessing result; the selection of the first trimming preprocessing and the second trimming preprocessing can be based on The distribution characteristics of the data elements in the channel state information are performed;

When the data elements in the channel state information appear in clusters, that is, when some data elements are distributed around the highest value data element and form a fixed distribution area, the channel delay and/or the maximum value of the reserved angle The first number of elements of is used to perform the first trimming on the channel state information Preprocessing, cutting out near-zero values outside the fixed area, and obtaining the first cutting preprocessing result, the highest value may be the highest value in the angle domain or the highest value in the delay domain of the data element;

When the data elements in the channel state information are discretely distributed, and the near-zero value cannot be cut out through the fixed area, then mark the position of the second number of reserved elements in a graph, cut off other unmarked elements, and obtaining the second cutting preprocessing result;

The distribution of the data elements in the channel state information is comprehensively considered for cutting, so as to improve the efficiency and accuracy of cutting; the first quantity is the first preset value or the elements to be retained obtained by screening according to the first preset threshold The second number is the second preset value or the number of elements that need to be retained obtained by screening according to the second preset threshold, the first number, the second number, the first preset The set value or the first preset value and the second preset value or the second preset threshold may be preconfigured according to channel state information or actual application conditions.

In an optional embodiment of the present application, the first trimming preprocessing result includes: the position of the first number of elements and the remaining trimmed elements;

In this embodiment, the terminal sends the first pruning preprocessing result and/or the second pruning preprocessing result after the channel state information pruning to the network device, so that the network device The processing result and/or the second cutting preprocessing result are processed in subsequent channels;

The first trimming preprocessing result may include: when the first trimming is performed on the channel state information, the channel data in the first trimming channel state information is obtained, that is, the reserved channel delay and/or angle are the largest the position of the first number of elements around the value, and other elements that are clipped;

The second pruning preprocessing result may include: the channel data in the second pruning channel obtained when the second pruning is performed on the channel state information, that is, the reserved second number of The position of the element, and other elements that are cropped;

Of course, the first clipping preprocessing result may also include the first clipping channel state information obtained by performing the first clipping; the second clipping preprocessing result may also include the channel state information obtained by performing the second clipping The second trimming channel state information.

Further, the first number and the second number are pre-configured by the network device or determined by the terminal; when the first number and the second number are pre-configured by the network device, the network device will Signaling is pre-configured to the terminal; when the first number and the second number are determined by the terminal, reporting the first number and the second number determined by the terminal to a network device.

In an optional embodiment of the present application, performing cyclic shift preprocessing on the channel state information to obtain a third preprocessing result may include:

Further, the third preprocessing result includes: the channel obtained after the cyclic shift, and the corresponding position of the target element.

In this embodiment, by performing cyclic shift preprocessing on the estimated channel, it is compensated that when the estimated channel does not meet the condition of cutting preprocessing, the cyclic shift preprocessing is performed on the estimated channel to remove unnecessary data elements , to ensure the accuracy of subsequent channel compression recovery.

In an optional embodiment of the present application, the first preprocessing result is cut and preprocessed to obtain the second preprocessing result, including:

Based on the first preprocessing result, the first number of elements around the channel delay and/or the maximum value of the angle are reserved, and the remaining elements are cut to obtain a second preprocessing result, and the first number is the first A preset value or the number of elements to be retained obtained by filtering according to the first preset threshold. The first number is the first preset value or the number of elements that need to be kept after filtering according to the first preset threshold; the third trimming preprocessing result here may include: channel delay and/or angle maximum The positions of the first number of elements around and the rest of the cropped elements;

In an optional embodiment of the present application, cyclic shift preprocessing is performed on the first preprocessing result to obtain a third preprocessing result, including:

Based on the first preprocessing result, if it is judged that if the first element of the channel is not the channel delay and/or the maximum value of the retention angle, the sequential cyclic shift is started from the element of the channel delay and/or the maximum value of the retention angle, The channel delay and/or the maximum value of the reserved angle are cyclically shifted to the position of the first element, and the position corresponding to the channel delay and/or the maximum value of the reserved angle is recorded to obtain the third preprocessing result, that is, for The channel state information is first transformed into the angle domain, time delay domain or Doppler domain through Fourier transform, and then performs cyclic shift preprocessing based on the result of Fourier transform;

During specific implementation, if the first element of the channel is not the channel delay and/or the maximum value of the reserved angle, the sequential cyclic shift starts from the element of the channel delay and/or the maximum value of the reserved angle, and the channel delay and/or Or the maximum value of the reserved angle is cyclically shifted to the position of the first element, and the position corresponding to the channel delay and/or the maximum value of the reserved angle is recorded to obtain the third preprocessing result. Here, the third preprocessing result may include: the channel obtained after the cyclic shift, and the position corresponding to the target element of the maximum value of the channel delay and/or reserved angle.

Of course, when the preprocessing results include the first preprocessing result, the second preprocessing result and the third preprocessing result, that is to say, the channel state information is first transformed into the angle domain by Fourier transform, time delay Domain or Doppler domain, then crop the result of Fourier transform, and then perform cyclic shift preprocessing on the cropped result;

It is also possible to perform Fourier transform on the channel state information to the angle domain, delay domain or Doppler domain first, then preprocess the cyclic shift of the result of the Fourier transform, and then trim the cyclic shift Cutting; in specific implementation, it is carried out according to the above-mentioned cutting and cyclic shift processing.

It should be noted that, the manner in which the terminal side preprocesses the channel state information may be configured by the network device through high-level signaling or agreed upon in advance.

In an optional embodiment of the present application, the channel state information is described, and the channel state information includes at least one of the following: a channel matrix; a channel eigenvector; and a channel precoding matrix.

In an optional embodiment of the present application, the target channel compression model is described, and the target channel compression model is to select a target machine learning model from a plurality of pre-trained machine learning models with different cropping ratios; The multiple machine learning models of different compression ratios are trained through the following process:

Step 01, obtaining the downlink channel data set generated by the simulation model or formed by the frequency domain multiplexed downlink channel reported by the terminal;

Step 02, performing a plurality of trimming processes with different preset trimming ratios on the downlink channels in the downlink channel data set, and obtaining trimming processing results with a plurality of different preset trimming ratios;

Step 03: According to the cropping processing results of multiple preset cropping ratios, input them into preset models for training, and obtain multiple machine learning models with different compression ratios.

In this embodiment, it should be known that the multiple machine learning models with different compression ratios can be obtained by training the terminal, or by training the network device; the terminal can use the simulation model to generate a frequency domain multiplexing FDD downlink channel data or notify the terminal of the FDD downlink channel data measured and reported when the traffic is less and the air interface resources are sufficient, and establish the FDD downlink channel data set, or use Complete with any downlink channel data acquisition method that meets the requirements;

The terminal trains the machine learning model used for channel compression and recovery based on the established FDD downlink channel data set. The machine learning model includes a compression network model deployed at the feedback sending end and a restoration network model deployed at the feedback receiving end; during the training process The downlink channel in the downlink channel data set can be cut and preprocessed according to the characteristics of the current channel data set;

Due to the different sparsity of the actual channel delay domain or other dimensions where clipping occurs, it is necessary to train multiple models according to the clipping ratio, for example, for no clipping, clipping the remaining 3/4 elements, clipping The remaining 1/2 of the number of elements, the remaining 1/4 of the number of cropped elements, and other categories that retain the ratio of elements are used to train the model separately. At this time, the trained model processes the cropped channel, so for the same data Different models can be trained by different sets and different cutting ratios, and a variety of models suitable for different cutting situations can be formed into a model set for selection and use in the subsequent process.

In an optional embodiment of the present application, the above-mentioned channel compression method may also include:

Updating the multiple machine learning models with different cropping ratios according to the channel trimming preprocessing method and trimming ratio;

In this embodiment, by detecting the compression performance of the target channel compression model used, when the model performance drops below a preset compression threshold due to channel environment changes, a corresponding model update mechanism is triggered, according to the channel pruning method and pruning ratio , updating the multiple machine learning models with different cropping ratios, so as to ensure the performance of channel feedback.

Step 21, sending a reference signal to the terminal;

Step 22, receiving the compressed channel information fed back by the terminal and/or the preprocessing result obtained by preprocessing the channel state information when the channel is compressed; the compressed channel information is the channel state information obtained by the terminal performing channel measurement on the reference signal , and performing cutting preprocessing on the channel state information to obtain a preprocessing result; and performing compression processing on the channel state information by using a target channel compression model according to the preprocessing result;

Step 23: Restoring the compressed channel according to the target channel decompression model to obtain the restored channel.

In this embodiment, the reference signal sent by the network device to the terminal may be generated by a simulation model The downlink channel data may also be the complete downlink channel data reported by the network device notifying the terminal when the business is not busy; the reference signal is mainly used for the terminal to perform channel measurement to obtain channel state information; after the network device sends the reference signal, further receiving the compressed channel information sent by the terminal and/or the preprocessing result obtained by preprocessing the channel state information when the channel is compressed, and recovering the compressed channel according to the target channel decompression model to improve channel recovery performance; The compressed channel information is the channel state information obtained by the terminal according to the channel measurement of the reference signal, and the channel state information is obtained after cutting and compressing the channel state information; the target decompression model is obtained from multiple pre-trained A model adapted to the compressed channel information selected in the machine learning model;

In an optional embodiment of the present application, the preprocessing results include one of the following:

performing a Fourier transform on the channel state information to a first preprocessing result obtained in an angle domain, a delay domain or a Doppler domain;

In this embodiment, the preprocessing result is obtained by the base station performing clipping preprocessing, cyclic shift preprocessing, or Fourier transform on the channel state information, or after Fourier transform and then clipping processing and /or obtained by cyclic shift processing;

The third preprocessing result is obtained by the base station through Fourier transform on the channel state information;

The second preprocessing result is that the base station reserves a first number of elements around the maximum value of the channel delay and/or angle for the channel state information, and cuts the rest of the elements, or the base station retains the channel state information by The position of the reserved second number of elements is marked in bitmap mode, and the remaining elements are cut to obtain;

The third preprocessing result is that the base station cyclically shifts the channel state information sequentially from the element of the channel delay and/or the maximum value of the reserved angle, and cyclically shifts the channel delay and/or the maximum value of the reserved angle to the first The position of an element is obtained by recording the position corresponding to the channel delay and/or the maximum value of the reserved angle.

In an optional embodiment of the present application, the above step 23 may include:

Step 231, according to the target channel decompression model and the preprocessing result, the compressed signal The channel information is restored to obtain the restored channel information.

In this embodiment, the clipping preprocessing result is obtained by the terminal clipping the channel state with a near-zero value according to different clipping methods; the network device receives the compressed channel information and/or After the preprocessing result, first use the target decompression model to restore the compressed channel information to obtain the first restored channel information, and then cut the first restored channel information according to the cutting preprocessing result The previous information is restored, and the restored second restored channel information is obtained.

In an optional embodiment of the present application, the target channel decompression model is described, and the target channel compression model is to select a target machine learning model from a plurality of pre-trained machine learning models with different cropping ratios; The multiple machine learning models of different compression ratios are trained through the following process:

Step 011, acquiring the downlink channel data set generated by the simulation model or formed by the frequency domain multiplexed downlink channel reported by the terminal;

Step 012, performing a plurality of trimming processes with different preset trimming ratios on the downlink channels in the downlink channel data set, and obtaining trimming processing results with a plurality of different preset trimming ratios;

Step 013: According to the cropping processing results of multiple preset cropping ratios, input them into preset models for training, and obtain multiple machine learning models with different compression ratios.

In this embodiment, it should be known that the multiple machine learning models with different compression ratios may be obtained through training by the terminal or network devices; the multiple deep neural network models with different compression ratios The training method is the same as the training method on the terminal side, and will not be repeated here.

In an optional embodiment of the present application, the channel recovery method applied to network equipment may also include:

Configure multiple deep neural network models with different compression ratios to the terminal.

In this embodiment, when the multiple deep neural network models with different compression ratios are trained on the network device side, the network device configures the trained multiple deep neural network models with different compression ratios to the terminal , so that the terminal can compress the channel. Of course, the multiple machine learning models with different cropping ratios can also be obtained by the terminal through training.

The above-mentioned channel compression and recovery method will be described below with a specific example, as shown in Figure 2 is a schematic diagram of the channel compression and recovery method; as shown in Figure 3, the specific implementation process is as follows:

Step 31, the network device collects data, including but not limited to: using a simulation model to generate downlink channel data; or notifying the terminal to report complete downlink channel data when the business is not busy etc.; the network device performs model training based on the data received from the mobile phone, and builds a model set according to different values of specific parameters. When building a model collection, different models trained for different crop ratios should be included; it should be understood that the models can also be obtained by side-by-side training on the terminal

Step 32, the base station sends CSI channel state information to the terminal through the downlink control channel, and sends the downlink reference signal for channel information measurement to the terminal. The CSI channel state information includes channel resources occupied by feedback, feedback overhead, and cut Reserved energy ratio threshold, etc.

Step 33, the terminal estimates the downlink channel based on the reference signal, and selects a clipping method according to the distribution characteristics of the data elements in the estimated channel, and clips the estimated channel; An appropriate target compression model is selected to perform compression processing on the cropped channel to obtain a compressed channel, and the information in the compressed channel exists in the form of a bit stream.

Step 34, the terminal sends feedback information to the network device, and the feedback information may include: cropping mode information; cropping result information; target compression model or selection information of the target compression model and compression channel information.

Step 35: According to the feedback information, the network device uses the target compression model and the compressed channel information to first restore the trimmed channel; then restores the original channel before trimming based on the trimming method information and trimming result information , and use it for subsequent transmission signal processing.

Step 36, the network device and the terminal exchange information, and update the compression model according to the channel cutting method and the channel cutting ratio.

In the above-mentioned embodiments of the present application, through the interaction between the network device and the terminal, the terminal performs trimming and preprocessing on the estimated channel to filter out interference items in the channel and effectively improve subsequent recovery accuracy; and then use the compression model of the corresponding trimming ratio to The trimmed estimated channel is compressed, and the result is fed back to the network device. The network device restores the trimmed channel first, and then obtains the original channel according to the trimmed information, thereby improving the performance of the recovery method based on channel compression.

The embodiment of the present application also provides a terminal. As shown in FIG. 4, the terminal 40 includes:

An estimation module 41, configured to measure a reference signal sent by a network device to obtain channel state information;

The processing module 42 is configured to preprocess the channel state information to obtain a preprocessing result; according to the preprocessing result, use a target channel compression model to perform compression processing on the channel state information to obtain compressed channel information;

A transceiver module 43, configured to send the compressed channel information and/or the preprocessing result to the network network device.

Optionally, the processing module 42 is configured to preprocess the channel state information to obtain a preprocessing result, including one of the following:

The channel state information is transformed into the angle domain by fast inverse Fourier transform IFFT in the antenna dimension, transformed into the delay domain by fast inverse Fourier transform IFFT in the frequency dimension, and transformed by sparse Fourier transform SFFT in the time dimension into the Doppler domain.

Optionally, the processing module 42 is specifically configured to: when performing the first trimming preprocessing on the channel state information, retain a first number of elements, and trim the remaining elements to obtain the first trimming preprocessing result , the first number is a first preset value or the number of elements to be retained obtained by filtering according to a first preset threshold; or

When the first quantity and the second quantity are determined by the terminal, the first quantity determined by the terminal and report the second quantity to the network device.

Optionally, the processing module 42 is specifically configured to cyclically shift the target element of the channel to the position of the first element if the first element of the channel satisfies the preset condition, and record the position corresponding to the target element, to obtain the first element Three preprocessing results.

Optionally, the channel state information includes at least one of the following: a channel matrix; a channel eigenvector; and a channel precoding matrix.

It should be noted that the terminal is a terminal corresponding to the above-mentioned channel compression method applied to the terminal, and all the implementation methods in the above-mentioned method embodiments are applicable to this embodiment of the terminal, and can also achieve to the same technical effect.

Embodiments of the present application also provide a network device, where the network device includes:

Optionally, the preprocessing result includes one of the following:

Optionally, the processing module is configured to restore the compressed channel information according to the target channel decompression model, and the obtained restored channel information includes:

According to the target channel decompression model and the preprocessing result, the compressed channel information is restored to obtain the restored channel information.

It should be noted that the network device is a network device corresponding to the above-mentioned channel recovery method applied to the network device, and all the implementation methods in the above method embodiments are applicable to the embodiments of the network device, and can also achieve the same technical effect.

All the implementation manners in the foregoing method embodiments are applicable to the embodiments of the terminal, and can also achieve the same technical effect.

An embodiment of the present application further provides a communication device, including: a processor, and a memory storing a computer program, and when the computer program is run by the processor, the above-mentioned method is executed. All the implementation manners in the foregoing method embodiments are applicable to this embodiment, and can also achieve the same technical effect.

Embodiments of the present application also provide a computer-readable storage medium, including instructions, which, when executed on a computer, cause the computer to execute the method as described above. All the implementation manners in the foregoing method embodiments are applicable to this embodiment, and can also achieve the same technical effect.

Those skilled in the art can appreciate that the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software depends on the specific application and design constraints of the technical solution. Skilled artisans may use different methods to implement the described functions for each specific application, but such implementation should not be regarded as exceeding the scope of the present application.

Those skilled in the art can clearly understand that for the convenience and brevity of the description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiment, which will not be repeated here.

In the embodiments provided in this application, it should be understood that the disclosed devices and methods may be implemented in other ways. For example, the device embodiments described above are only illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or May be integrated into another system, or some features may be ignored, or not implemented. In another point, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection of devices or units may be in electrical, mechanical or other forms.

The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of them can be selected according to actual needs unit to realize the purpose of the solution of this embodiment.

In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, each unit may exist separately physically, or two or more units may be integrated into one unit.

If the functions described above are realized in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including Several instructions are used to make a computer device (which may be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: various media capable of storing program codes such as U disk, mobile hard disk, ROM, RAM, magnetic disk or optical disk.

In addition, it should be pointed out that, in the device and method of the present application, obviously, each component or each step can be decomposed and/or reassembled. These decompositions and/or recombinations should be considered equivalents of this application. Also, the steps for executing the above series of processes can naturally be executed in chronological order according to the illustrated order, but it is not necessary to be executed in chronological order, and some steps can be executed in parallel or independently of each other. For those of ordinary skill in the art, it can be understood that all or any steps or components of the method and device of the present application can be implemented in any computing device (including processors, storage media, etc.) or networks of computing devices with hardware or firmware , software or their combination, which can be realized by those skilled in the art by using their basic programming skills after reading the description of this application.

Therefore, the object of the present application can also be achieved by running a program or a group of programs on any computing device. The computing device may be a known general-purpose device. Therefore, the purpose of the present application can also be achieved only by providing a program product including program codes for realizing the method or device. That is, such a program product also constitutes this application, and a storage medium storing such a program product also constitutes this application. Obviously, the storage medium may be any known storage medium or any storage medium developed in the future. It should also be pointed out that in the device and method of the present application, obviously, each component or each step can be decomposed and/or reassembled. These decompositions and/or recombinations should be considered equivalents of this application. And, the steps of executing the above-mentioned series of processes can naturally follow the order of description. The sequence is executed in chronological order, but it does not need to be executed in chronological order. Certain steps may be performed in parallel or independently of each other.

The above description is the preferred implementation mode of the present application. It should be pointed out that for those of ordinary skill in the art, some improvements and modifications can also be made without departing from the principles described in the application. These improvements and modifications are also It should be regarded as the protection scope of this application.

Claims

A channel compression method applied to a terminal, the method comprising:

Measure the reference signal sent by the network equipment to obtain channel state information;

Preprocessing the channel state information to obtain a preprocessing result;

According to the preprocessing result, the channel state information is compressed using the target channel compression model to obtain compressed channel information;

Send the compressed channel information and/or the preprocessing result to the network device.
The channel compression method according to claim 1, wherein the channel state information is preprocessed to obtain a preprocessing result, including one of the following:

Transforming the channel state information into an angle domain, a delay domain or a Doppler domain to obtain a first preprocessing result;

performing cutting preprocessing on the channel state information or the first preprocessing result to obtain a second preprocessing result;

Perform cyclic shift preprocessing on the channel state information or the first preprocessing result to obtain a third preprocessing result.
The channel compression method according to claim 2, wherein transforming the channel state information into an angle domain, a delay domain or a Doppler domain comprises:

The channel state information is transformed into the angle domain through the fast inverse Fourier transform IFFT in the antenna dimension, transformed into the delay domain through the fast inverse inverse Fourier transform IFFT in the frequency dimension, and transformed into the delay domain through the sparse Fourier transform SFFT in the time dimension Transform into the Doppler domain.
The channel compression method according to claim 2, wherein the channel state information is cut and preprocessed to obtain a second preprocessing result, including:

When performing the first pruning preprocessing on the channel state information, retain a first number of elements, and pruning the remaining elements to obtain a first pruning preprocessing result, the first number is a first preset value or The number of elements to be retained obtained by filtering according to the first preset threshold; or

When the second clipping preprocessing is performed on the channel state information, the position of the reserved second number of elements is marked in a bitmap manner, and the remaining elements are clipped to obtain a second clipping preprocessing result, the second The quantity is the second preset value or the number of elements to be kept obtained by filtering according to the second preset threshold.
The channel compression method according to claim 4, wherein,

The first trimming preprocessing result includes: the position of the first number of elements and the remaining trimmed elements;

The second clipping preprocessing result includes: marking the positions of the second number of reserved elements and the rest of the clipped elements in a bitmap manner.
The channel compression method according to claim 4, wherein the first number and the second number are preconfigured by network equipment or determined by a terminal;

When the first number and the second number are pre-configured by network equipment, they are pre-configured to the terminal through high-layer signaling;

When the first quantity and the second quantity are determined by the terminal, report the first quantity and the second quantity determined by the terminal to the network device.
The channel compression method according to claim 2, wherein performing cyclic shift preprocessing on the channel state information to obtain a third preprocessing result, comprising:

If the first element of the channel satisfies the preset condition, the target element of the channel is cyclically shifted to the position of the first element, and the corresponding position of the target element is recorded to obtain a third preprocessing result.
The channel compression method according to claim 7, wherein the third preprocessing result includes: the channel obtained after cyclic shift, and the position corresponding to the target element.
The channel compression method according to claim 2, wherein the cutting preprocessing is performed on the first preprocessing result to obtain the second preprocessing result, including:

Based on the first preprocessing result, the first number of elements around the channel delay and/or the maximum value of the angle are reserved, and the remaining elements are cut to obtain a second preprocessing result, and the first number is the first A preset value or the number of elements to be retained obtained by filtering according to the first preset threshold.
The channel compression method according to claim 2, wherein performing cyclic shift preprocessing on the first preprocessing result to obtain a third preprocessing result, comprising:

Based on the first preprocessing result, if it is judged that if the first element of the channel is not the channel delay and/or the maximum value of the retention angle, the sequential cyclic shift is started from the element of the channel delay and/or the maximum value of the retention angle, The channel delay and/or the maximum value of the retention angle are cyclically shifted to the position of the first element, and the position corresponding to the channel delay and/or the maximum value of the retention angle is recorded to obtain a third preprocessing result.
The channel compression method according to claim 1, wherein the channel state information includes At least one of the following:

channel matrix;

channel feature vector;

Channel precoding matrix.
The channel compression method according to claim 1, wherein the target channel compression model is a target machine learning model selected from a plurality of pre-trained machine learning models with different cropping ratios; the multiple different cropping ratios The scaled machine learning model is trained through the following process:

Obtain the downlink channel data set generated by the simulation model or formed by the frequency domain multiplexing downlink channel reported by the terminal;

performing a plurality of trimming processes with different preset trimming ratios on the downlink channels in the downlink channel data set, and obtaining trimming processing results with a plurality of different preset trimming ratios;

According to the cutting processing results of a plurality of different preset cutting ratios, they are respectively input into the preset machine learning models for training, and a plurality of machine learning models of different cutting ratios are obtained.
A channel recovery method applied to network equipment, the method comprising:

sending a reference signal for channel estimation to the terminal;

receiving the compressed channel information fed back by the terminal and/or the preprocessing result obtained by preprocessing the channel state information when compressing the channel; the compressed channel information is the channel state information obtained by the terminal according to the channel measurement of the reference signal, and the performing preprocessing on the channel state information to obtain a preprocessing result; and performing compression processing on the channel state information by using a target channel compression model according to the preprocessing result;

Restoring the compressed channel according to the target channel decompression model to obtain the restored channel.
The channel recovery method according to claim 13, wherein the preprocessing result includes one of the following:

A first preprocessing result obtained by transforming the channel state information into an angle domain, a delay domain or a Doppler domain;

A second preprocessing result obtained by cutting and preprocessing the channel state information or the first preprocessing result;

A third preprocessing result obtained by performing cyclic shift preprocessing on the channel state information or the first preprocessing result.
The channel recovery method according to claim 14, wherein the compressed channel information is recovered according to the target channel decompression model, and the recovered channel information comprises:

According to the target channel decompression model and the preprocessing result, the compressed channel information is restored to obtain the restored channel information.
The channel restoration method according to claim 13, wherein the target channel decompression model is to select a target machine learning model from a plurality of pre-trained machine learning models with different cropping ratios; the multiple different cropping ratios The scaled machine learning model is trained through the following process:

Obtain the downlink channel data set generated by the simulation model or formed by the frequency domain multiplexing downlink channel reported by the terminal;

performing a plurality of trimming processes with different preset trimming ratios on the downlink channels in the downlink channel data set, and obtaining trimming processing results with a plurality of different preset trimming ratios;

According to the cutting processing results of a plurality of different preset cutting ratios, they are respectively input into the preset machine learning models for training, and a plurality of machine learning models of different cutting ratios are obtained.
A terminal comprising:

An estimation module, configured to measure a reference signal sent by a network device to obtain channel state information;

A processing module, configured to preprocess the channel state information to obtain a preprocessing result; according to the preprocessing result, use a target channel compression model to compress the channel state information to obtain compressed channel information;

A transceiver module, configured to send the compressed channel information and/or the preprocessing result to a network device.
A network device comprising:

The transceiver module is configured to send a reference signal to the terminal; and receive the compressed channel information fed back by the terminal and/or the preprocessing result obtained by preprocessing the channel state information when the channel is compressed; the compressed channel information is the terminal pair performing channel measurement on a reference signal to obtain channel state information, and performing preprocessing on the channel state information to obtain a preprocessing result; and performing compression processing on the channel by using a target channel compression model according to the preprocessing result;

The processing module is used to restore the compressed channel information according to the target channel decompression model, and obtain the restored channel information.
A communication device, comprising: a processor, a memory storing a computer program, the computer When the computer program is run by the processor, the method according to any one of claims 1 to 12 or the method according to any one of claims 13 to 16 is executed.
A computer-readable storage medium storing instructions, which, when run on a computer, cause the computer to execute the method according to any one of claims 1 to 12 or the method according to any one of claims 13 to 16 .