Channel matrix compression feedback method, network side equipment and user side equipment

Abstract

The invention provides a channel matrix compression feedback method, network side equipment and user side equipment. The method comprises the following steps: receiving a channel detection reference signal sent by the user side equipment; obtaining the sparsity of a channel matrix according to the channel detection reference signal, obtaining a target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and obtaining configuration information of the target channel matrix compression model according to the target channel matrix compression model; the preset channel matrix sparsity threshold value is obtained according to the sparsity distribution of the acquired downlink signal estimation data; and sending configuration information of the target channel matrix compression model to user side equipment. According to the invention, different models are selected according to the sparsity of different channel matrixes, feedback channel information after channel matrix compression is reduced according to the channel matrix with high sparsity, and channel matrix preprocessing and related calculation are carried out on network side equipment, so that the calculation pressure of user side equipment is effectively reduced.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of wireless technology, and in particular to a method for channel matrix compression feedback, a network side device, and a user side device. Background Art

[0002] In a multiple-input multiple-output (MIMO) system, obtaining channel state information (CSI) is a key condition for beamforming to improve transmission performance. For a frequency division duplexing (FDD) system, since there is no complete reciprocity of uplink and downlink channels, network-side devices need to obtain complete downlink CSI with the help of user-side device feedback. In the New Radio (NR) system, user-side devices mainly rely on codebooks to feedback CSI. Currently, codebook types such as CSI type I, type II, and type II enhanced are supported for feedback of rank indicator (RI), precoding matrix indicator (PMI), and channel quality indicator (CQI).

[0003] At present, the main method is to extract low-dimensional features from a large amount of channel data using deep neural networks, and then compress and restore the channel. This method can be optimized for specific scenarios. The data-driven channel feedback method can often achieve higher channel feedback accuracy under the same feedback overhead, but the existing encoder / decoder is only suitable for fixed compression rate feedback. Since the communication environment is constantly changing, the use of a fixed compression rate may result in the transmission of excess bits, resulting in a waste of resources.

[0004] Although the downlink channel compression feedback scheme based on deep learning can effectively reduce the transmission overhead, in order to obtain a higher accuracy of channel matrix compression / decompression, if a unified model is used, a larger number of compressed variables need to be selected. If the channel state information (CSI) matrix is ​​truncated directly by using a fixed matrix dimension after transformation, considering the sparseness of the channel matrix, the sparse channel matrix can be further processed to reduce the compression rate, such as by using a two-dimensional discrete Fourier transform (DFT). Fourier Transform, 2D-DFT) is used to process the CSI matrix to obtain a CSI matrix that is sparse in the angle delay domain; the CSI matrix that is sparse in the angle delay domain is truncated to obtain a truncated matrix, thereby reducing the dimension of the channel matrix. However, useful information may be deleted due to improper setting of the matrix dimension, thereby reducing the accuracy of channel compression feedback; if a two-stage series neural network is used, the first network is used to process the sparsity of the channel matrix, and the second neural network is used to adjust the number of output variables of the neural network, thereby reducing the number of output variables of the sparse channel matrix, which not only increases the difficulty of collecting data labels for the first-stage neural network, but also increases the complexity of training and reasoning by using a two-stage network. Summary of the invention

[0005] The embodiment of the present invention provides a method for channel matrix compression feedback, a network side device and a user side device to solve the problem that the accuracy and efficiency of the existing channel compression feedback cannot be guaranteed.

[0006] In order to solve the above-mentioned technical problems, the present invention is achieved as follows:

[0007] In a first aspect, an embodiment of the present invention provides a method for channel matrix compression feedback, which is applied to a network side device, including:

[0008] Receiving a channel sounding reference signal sent by a user side device;

[0009] The sparsity of the channel matrix is ​​obtained according to the channel sounding reference signal, and the target channel matrix compression model is obtained according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and the configuration information of the target channel matrix compression model is obtained according to the target channel matrix compression model; the preset channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data;

[0010] The configuration information of the target channel matrix compression model is sent to the user side device.

[0011] Optionally, also include:

[0012] Acquire a plurality of downlink channel estimation data collected in different domains as channel matrix training samples;

[0013] Presetting a channel matrix sparsity threshold according to the channel matrix sparsity distribution of the channel matrix training samples, and grouping the channel matrix training samples according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples;

[0014] Training the at least one group of channel matrix training samples to obtain at least one channel matrix decompression model and a corresponding channel matrix compression model;

[0015] The at least one channel matrix compression model and the channel matrix sparsity threshold are sent to the user side device.

[0016] Optionally, obtaining configuration information of a target channel matrix compression model according to the target channel matrix compression model includes:

[0017] Determining a modulation and coding mode of channel state information compression feedback information according to the quality of the channel sounding reference signal;

[0018] According to the modulation and coding mode and the reporting resource configuration of the channel state information compression feedback information, the configuration information of the target channel matrix compression model is obtained.

[0019] Optionally, the configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model and a feedback mode of the target channel matrix compression model, wherein the feedback mode of the target channel matrix compression model includes at least one of the following: a modulation and coding method of channel state information compression feedback information and a reporting resource configuration of the channel state information compression feedback information.

[0020] Optionally, also include:

[0021] Sending a channel state information reference number to a user side device according to the configuration information of the target channel matrix compression model;

[0022] Receiving channel matrix compression information completed by a channel matrix compression model selected by a user-side device according to configuration information of the target channel matrix compression model;

[0023] The channel matrix compression information is decompressed according to a channel matrix decompression model corresponding to the target channel matrix compression model.

[0024] Optionally, also include:

[0025] Monitoring the channel sounding reference signal according to a preset time period;

[0026] Obtaining the latest sparsity of the channel matrix according to the channel sounding reference signal obtained within the preset time period, and comparing the sparsity of the latest channel matrix with the channel matrix sparsity threshold to obtain the latest configuration information of the channel matrix compression model;

[0027] Send the latest channel matrix compression model configuration information to the user side device.

[0028] In a second aspect, an embodiment of the present invention provides a method for channel matrix compression feedback, which is applied to a user side device, including:

[0029] Sending a channel sounding reference signal to a network side device;

[0030] Receive configuration information of a target channel matrix compression model sent by a network side device, wherein the configuration information of the target channel matrix compression model is that the network side device obtains the sparsity of the channel matrix according to the channel detection reference signal, and obtains the target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and the target channel matrix compression model is obtained according to the target channel matrix compression model; the preset channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data.

[0031] Optionally, also include:

[0032] Receiving at least one channel matrix compression model and the channel matrix sparsity threshold sent by a network side device;

[0033] Among them, the at least one channel matrix compression model and the channel matrix sparsity threshold are for the network side device to obtain multiple downlink channel estimation data from different domains as channel matrix training samples; the channel matrix sparsity threshold is preset according to the channel matrix sparsity distribution of the channel matrix training samples, and the channel matrix training samples are grouped according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples; and the at least one group of the channel matrix training samples is trained.

[0034] Optionally, the configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model, a modulation coding method, and time-frequency resources required for compression feedback information.

[0035] Optionally, also include:

[0036] Receiving a channel state information reference number sent by a network side device according to configuration information of the target channel matrix compression model;

[0037] Channel matrix compression information completed by the channel matrix compression model selected according to the configuration information of the target channel matrix compression model is sent to the network side device.

[0038] In a third aspect, an embodiment of the present invention provides a network side device, including:

[0039] A first receiving module, used to receive a channel sounding reference signal sent by a user side device;

[0040] A first processing module is used to obtain the sparsity of the channel matrix according to the channel sounding reference signal, and obtain a target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and obtain configuration information of the target channel matrix compression model according to the target channel matrix compression model; the preset channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data;

[0041] The first sending module is used to send configuration information of the target channel matrix compression model to the user side device.

[0042] Optionally, also include:

[0043] The second processing module is used to obtain a plurality of downlink channel estimation data collected in different domains as channel matrix training samples;

[0044] A third processing module, configured to preset a channel matrix sparsity threshold according to the channel matrix sparsity distribution of the channel matrix training samples, and group the channel matrix training samples according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples;

[0045] A fourth processing module, used for training the at least one group of channel matrix training samples to obtain at least one channel matrix decompression model and a corresponding channel matrix compression model;

[0046] The third sending module is used to send the at least one channel matrix compression model and the channel matrix sparsity threshold to the user side device.

[0047] Optionally, the first processing module includes:

[0048] A first processing submodule, configured to determine a modulation and coding scheme of channel state information compression feedback information according to the quality of the channel sounding reference signal;

[0049] The second processing submodule is used to obtain the configuration information of the target channel matrix compression model according to the modulation and coding mode and the reporting resource configuration of the channel state information compression feedback information.

[0050] Optionally, the configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model and a feedback mode of the target channel matrix compression model, wherein the feedback mode of the target channel matrix compression model includes at least one of the following: a modulation and coding method of channel state information compression feedback information and a reporting resource configuration of the channel state information compression feedback information.

[0051] Optionally, also include:

[0052] A fourth sending module, used to send a channel state information reference number to a user side device according to the configuration information of the target channel matrix compression model;

[0053] A third receiving module is used to receive channel matrix compression information completed by the channel matrix compression model selected by the user side device according to the configuration information of the target channel matrix compression model;

[0054] A decompression module is used to decompress the channel matrix compression information according to a channel matrix decompression model corresponding to the target channel matrix compression model.

[0055] Optionally, also include:

[0056] A monitoring module, used for monitoring the channel sounding reference signal according to a preset time period;

[0057] A fifth processing module, configured to obtain the latest sparsity of the channel matrix according to the channel sounding reference signal obtained within the preset time period, and compare the sparsity of the latest channel matrix with the channel matrix sparsity threshold to obtain the latest configuration information of the channel matrix compression model;

[0058] The fifth sending module is used to send the latest configuration information of the channel matrix compression model to the user side device.

[0059] In a fourth aspect, an embodiment of the present invention provides a user side device, including:

[0060] A second sending module, used to send a channel sounding reference signal to a network side device;

[0061] The second receiving module is used to receive the configuration information of the target channel matrix compression model sent by the network side device, wherein the configuration information of the target channel matrix compression model is that the network side device obtains the sparsity of the channel matrix according to the channel detection reference signal, and obtains the target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and is obtained according to the target channel matrix compression model; the channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data.

[0062] Optionally, also include:

[0063] A fourth receiving module, configured to receive at least one channel matrix compression model and the channel matrix sparsity threshold sent by a network side device;

[0064] Among them, the at least one channel matrix compression model and the channel matrix sparsity threshold are for the network side device to obtain multiple downlink channel estimation data from different domains as channel matrix training samples; the channel matrix sparsity threshold is preset according to the channel matrix sparsity distribution of the channel matrix training samples, and the channel matrix training samples are grouped according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples; and the at least one group of the channel matrix training samples is trained.

[0065] Optionally, the configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model, a modulation coding method, and time-frequency resources required for compression feedback information.

[0066] Optionally, also include:

[0067] A fifth receiving module, configured to receive a channel state information reference number sent by a network side device according to configuration information of the target channel matrix compression model;

[0068] The fifth sending module is used to send channel matrix compression information completed by the channel matrix compression model selected according to the configuration information of the target channel matrix compression model to the network side device.

[0069] In a fifth aspect, an embodiment of the present invention provides an electronic device, characterized in that it includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein when the program or instruction is executed by the processor, the method for channel matrix compression feedback as described in any one of the first aspects or the steps in the method for channel matrix compression feedback as described in any one of the second aspects is implemented.

[0070] In the sixth aspect, an embodiment of the present invention provides a readable storage medium, characterized in that: the readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, it implements the method of channel matrix compression feedback as described in any one of the first aspect, or the steps in the method of channel matrix compression feedback as described in any one of the second aspect.

[0071] In the present invention, by selecting different models according to the sparsity of different channel matrices, while maintaining the compression or decompression accuracy, the feedback channel information after the channel matrix compression can be effectively reduced for the channel matrix with high sparsity, and the channel matrix preprocessing and related calculations are placed on the network side equipment, which effectively reduces the computing pressure of the user side equipment and solves the existing problem of not being able to guarantee the accuracy and efficiency of channel compression feedback. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0073] Figure 1 It is a flow chart of a method for channel matrix compression feedback provided by an embodiment of the present invention applied to a network side device;

[0074] Figure 2 It is a schematic diagram of channel matrix sparsity of channel matrix compression feedback provided by an embodiment of the present invention;

[0075] Figure 3 It is a flow chart of a method for channel matrix compression feedback provided by an embodiment of the present invention applied to a user side device;

[0076] Figure 4 It is an overall flow chart of a method for channel matrix compression feedback provided by an embodiment of the present invention;

[0077] Figure 5 It is a structural diagram of a network side device provided by an embodiment of the present invention;

[0078] Figure 6 is a structural diagram of a user-side device provided by an embodiment of the present invention;

[0079] Figure 7 It is a structural schematic diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0080] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0081] Please refer to Figure 1The embodiment of the present invention provides a method for channel matrix compression feedback, which is applied to a network side device, including:

[0082] Step 11: receiving a channel sounding reference signal sent by a user side device;

[0083] In the embodiment of the present invention, the user side device accesses the network, enters the Radio Resource Control (RRC-CONNECTED) state, and sends a channel sounding reference signal (Sounding Reference Signal, SRS) to the network side device according to the configuration.

[0084] Step 12: obtaining the sparsity of the channel matrix according to the channel sounding reference signal, obtaining a target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and obtaining configuration information of the target channel matrix compression model according to the target channel matrix compression model; the preset channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data;

[0085] In an embodiment of the present invention, a network side device calculates the sparsity of a channel matrix according to the channel sounding reference signal, and compares it with a preset channel matrix sparsity threshold, wherein the preset channel matrix sparsity threshold is pre-processed according to downlink signal estimation data collected in different domains (such as space-frequency domain, time-frequency domain, and space-frequency-time domain, etc.) as channel matrix training samples, and pre-sets a matrix sparsity threshold (such as δ1, δ2…δ N-1 , where δ1<…δ N-1 ); select a model according to the comparison result, determine the modulation and coding method of the channel state information compression feedback information according to the quality of the channel detection reference signal, and obtain the configuration information of the target channel matrix compression model according to the modulation and coding method and the reporting resource configuration of the channel state information compression feedback information. For a highly sparse channel matrix, the feedback channel information after the channel matrix compression can be effectively reduced, and the channel matrix preprocessing and related calculations are placed on the network side device, which effectively reduces the computing pressure of the user side device and effectively reduces the complexity during training and reasoning.

[0086] Step 13: Send the configuration information of the target channel matrix compression model to the user-side device.

[0087] In an embodiment of the present invention, the configuration information of the target channel matrix compression model is sent to a user side device via PDCCH / PDSCH; the configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model and a feedback mode of the target channel matrix compression model, wherein the feedback mode of the target channel matrix compression model includes at least one of the following: a modulation and coding method of channel state information compression feedback information and a reporting resource configuration of the channel state information compression feedback information.

[0088] In an embodiment of the present invention, by selecting different models according to the sparsity of different channel matrices, while maintaining the compression or decompression accuracy, the feedback channel information after the channel matrix compression can be effectively reduced for a channel matrix with high sparsity, and the channel matrix preprocessing and related calculations are placed on the network side device, which effectively reduces the computing pressure of the user side device and solves the existing problem of being unable to guarantee the accuracy and efficiency of channel compression feedback.

[0089] In the embodiment of the present invention, optionally, the following further comprises:

[0090] Acquire a plurality of downlink channel estimation data collected in different domains as channel matrix training samples;

[0091] Presetting a channel matrix sparsity threshold according to the channel matrix sparsity distribution of the channel matrix training samples, and grouping the channel matrix training samples according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples;

[0092] Training the at least one group of channel matrix training samples to obtain at least one channel matrix decompression model and a corresponding channel matrix compression model;

[0093] The at least one channel matrix compression model and the channel matrix sparsity threshold are sent to the user side device.

[0094] In an embodiment of the present invention, a system for training an artificial intelligence model for a wireless access network is deployed on a centralized unit (CU) and / or a distributed unit (DU) of the network side device, or on a logical entity across CUs, by collecting downlink channel estimation data in different domains (such as space-frequency domain, time-frequency domain, and space-frequency-time domain, etc.) as channel matrix training samples, preprocessing the channel matrix training samples, and presetting channel matrix sparsity thresholds (such as δ1, δ2…δ N-1 , where δ1<…δ N-1), for example: after channel analysis, a channel matrix sparsity threshold is selected to be set to δ1=0.17, and according to the sparsity of the channel matrix corresponding to the downlink signal estimation data, the downlink signal estimation data is grouped according to the channel matrix sparsity threshold, and for each group of data, a compression model of the channel matrix and a decompression model of the channel matrix based on an artificial intelligence autoencoder (such as: autoencoder) are designed and trained offline, wherein the compression model of the channel matrix is ​​recorded as E1, E2, ...E M , the decompression model of the channel matrix is ​​recorded as D1, D2, ... D M For example, the compression model of the channel matrix is ​​numbered as E1 and E2, and the decompression model of the channel matrix is ​​numbered as D1 and D2. Offline training is based on two pairs of compression models and decompression models of the channel matrix based on an artificial intelligence autoencoder (such as an autoencoder). The training method is as follows: Models E1 and D1 are trained using corresponding data in the channel matrix sparsity interval [0, δ1), and E2 and D2 are trained using corresponding data in the channel matrix sparsity interval [δ1, 1]. The number of output variables of E1 and E2 and the number of input variables of D1 and D2 can be designed as multiple groups of output / The number of input variables is selected, and the number of output / input variables with compression feedback accuracy higher than the preset accuracy threshold η=0.92 and the smallest number of feedback bits is selected as the optimal number of output / input variables; under the same matrix dimension, the higher the matrix sparsity, the smaller the matrix dimension after compression can be set, that is, the smaller the number of output variables of the compression model of the channel matrix and the number of input variables of the decompression model of the channel matrix can be set, and the compression model of the channel matrix and the corresponding number are synchronized to the user side device, and the decompression model of the channel matrix, the corresponding number and the sparsity threshold of the channel matrix are synchronized to the network side device.

[0095] In the embodiment of the present invention, channel matrix preprocessing and related calculations are placed on the network side device through channel reciprocity, which effectively reduces the computing pressure of the user side device and the complexity of training and reasoning.

[0096] In the embodiment of the present invention, the method for calculating the sparsity of the channel matrix is:

[0097] At time t, x(t) represents the transmitted signal, y(t) represents the received signal, and a mathematical model is established based on the relationship between the transmitted signal, the received signal and the channel matrix H(t): y(t)=H(t)x(t)+n(t); wherein n(t) represents noise.

[0098] Preprocessing the channel matrix may perform a two-dimensional discrete Fourier transform on the channel matrix H(t) in the original space-frequency domain to obtain a sparse channel matrix in the angle delay domain; or perform a two-dimensional discrete Fourier transform on the channel matrix in the original space-time domain to obtain a sparse channel matrix in the angle Doppler domain; or perform a three-dimensional discrete Fourier transform on the channel matrix in the original space-frequency domain-time domain to obtain a sparse channel matrix in the angle delay Doppler domain;

[0099] The preprocessed channel matrix is ​​y(t)=H(t)x(t)+n(t), which is denoted as H′.

[0100] The sparsity of the channel matrix refers to the ratio of the number of elements in the channel matrix that are smaller than a certain threshold value to the number of all elements, expressed as a relative percentage.

[0101] Specifically, the calculation method of the sparsity S of the channel matrix is:

[0102] Calculate the modulus of a single element in each sample H′ in the sampled channel matrix data set, and record the corresponding modulus matrix as a new sample matrix, that is,

[0103] Calculate the sum of all elements in each new sample matrix, denoted as sum;

[0104] Sort each element in the new sample matrix from large to small to obtain a vector h, where the elements of the vector h are denoted as h i ;

[0105] Add up the sorted elements and divide by sum, that is

[0106] Record the first h′ k The subscript k of ≥80% is denoted as K; where the total number of channel matrix elements is -K, which is divided by the total number of channel matrix elements to obtain the sparsity S of the channel matrix.

[0107] The sparsity S of the channel matrix can distinguish the channels, please refer to Figure 2 ,, where the horizontal axis N is the number of points in the channel matrix, and the vertical axis Cumulative rate is the accumulation rate. After the channel matrices of non-line of sight (NLOS) and line of sight (LOS) are pre-processed and transformed into the angle delay domain, the sparsity of the channel matrix of the NLOS channel is calculated as (3200-785) / 3200=74.47%, and the sparsity of the channel matrix of the LOS channel is (3200-294) / 3200=90.8%, as shown in Figure 2 There is a clear difference between the two;

[0108] In an embodiment of the present invention, by introducing a channel matrix sparsity threshold and grouping training models according to the channel matrix sparsity, different models are selected for the sparsity of different channel matrices. While maintaining the compression / decompression accuracy, the feedback channel information after the channel matrix is ​​compressed can be effectively reduced for a channel matrix with high sparsity.

[0109] In the embodiment of the present invention, optionally, the following further comprises:

[0110] Sending a channel state information reference number to a user side device according to the configuration information of the target channel matrix compression model;

[0111] Receiving channel matrix compression information completed by a channel matrix compression model selected by a user-side device according to configuration information of the target channel matrix compression model;

[0112] The channel matrix compression information is decompressed according to a channel matrix decompression model corresponding to the target channel matrix compression model.

[0113] In an embodiment of the present invention, the network side device sends the channel state information parameter number to the user side device according to the configuration; the user side device selects the corresponding channel matrix compression model according to the model number in the configuration information of the received target channel matrix compression model and completes the compression of the channel matrix, and feeds back the compressed channel information to the network side device according to the modulation coding mode and resource configuration set by the network side device; the network side device selects the channel matrix decompression model paired with the channel matrix compression model to decompress the channel information, and through channel reciprocity, places the channel matrix preprocessing and related calculations on the network side device, effectively reducing the calculation pressure of the user side device, and adopts a first-level network, which effectively reduces the complexity during training and reasoning.

[0114] In the embodiment of the present invention, optionally, the following further comprises:

[0115] Monitoring the channel sounding reference signal according to a preset time period;

[0116] Obtaining the latest sparsity of the channel matrix according to the channel sounding reference signal obtained within the preset time period, and comparing the sparsity of the latest channel matrix with the channel matrix sparsity threshold to obtain the latest configuration information of the channel matrix compression model;

[0117] Send the latest channel matrix compression model configuration information to the user side device.

[0118] In an embodiment of the present invention, subsequent network-side devices continue to monitor the channel sounding reference signal according to a preset time period, and calculate the sparsity of the channel matrix according to the channel sounding reference signal. When the threshold interval corresponding to the sparsity of the channel matrix changes, the configuration information of the channel matrix compression model is reconfigured, that is, the model number and the resources required for feedback information are sent to the user-side device, and the user-side device adjusts the corresponding channel matrix compression model and reports the compressed channel information according to the latest configuration information of the channel matrix compression model; and when the uplink channel quality changes, the coding and modulation mode of the model compressed feedback channel information and the resources required for feedback information are reconfigured and sent to the user-side device, and the user-side device adjusts the coding and modulation mode of the compressed feedback channel information and reports the compressed channel information according to the configuration.

[0119] In an embodiment of the present invention, by training corresponding channel matrix compression / decompression models in groups according to the sparsity of the channel matrices, different models are selected according to the sparsity of different channel matrices. While maintaining the compression / decompression accuracy, the feedback channel information after the channel matrix compression can be effectively reduced for the channel matrix with high sparsity; and through channel reciprocity, the channel matrix preprocessing and related calculations are placed on the network side equipment, effectively reducing the computing pressure of the user side equipment; at the same time, a primary network is adopted, which effectively reduces the complexity during training and reasoning.

[0120] Please refer to Figure 3 The embodiment of the present invention provides a method for channel matrix compression feedback, which is applied to a user side device, including:

[0121] Step 31: Send a channel sounding reference signal to a network side device;

[0122] In the embodiment of the present invention, the user side device accesses the network, enters the Radio Resource Control (RRC-CONNECTED) state, and sends a channel sounding reference signal (Sounding Reference Signal, SRS) to the network side device according to the configuration.

[0123] Step 32: Receive configuration information of the target channel matrix compression model sent by the network side device, wherein the configuration information of the target channel matrix compression model is that the network side device obtains the sparsity of the channel matrix according to the channel detection reference signal, and obtains the target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and the preset channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data.

[0124] In an embodiment of the present invention, a network side device calculates the sparsity of a channel matrix according to the channel sounding reference signal, and compares it with a preset channel matrix sparsity threshold, wherein the preset channel matrix sparsity threshold is pre-processed according to downlink signal estimation data collected in different domains (such as space-frequency domain, time-frequency domain, and space-frequency-time domain, etc.) as channel matrix training samples, and pre-sets a matrix sparsity threshold (such as δ1, δ2…δ N-1 , where δ1<…δ N-1 ); select a model according to the comparison result, determine the modulation and coding method of the channel state information compression feedback information according to the quality of the channel detection reference signal, and obtain the configuration information of the target channel matrix compression model according to the modulation and coding method and the reporting resource configuration of the channel state information compression feedback information. For a highly sparse channel matrix, the feedback channel information after the channel matrix compression can be effectively reduced, and the channel matrix preprocessing and related calculations are placed on the network side device, which effectively reduces the computing pressure of the user side device and effectively reduces the complexity during training and reasoning.

[0125] In an embodiment of the present invention, the configuration information of the target channel matrix compression model is sent to a user side device via PDCCH / PDSCH; the configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model and a feedback mode of the target channel matrix compression model, wherein the feedback mode of the target channel matrix compression model includes at least one of the following: a modulation and coding method of channel state information compression feedback information and a reporting resource configuration of the channel state information compression feedback information.

[0126] In the embodiment of the present invention, optionally, the following further comprises:

[0127] Receiving at least one channel matrix compression model and the channel matrix sparsity threshold sent by a network side device;

[0128] Among them, the at least one channel matrix compression model and the channel matrix sparsity threshold are for the network side device to obtain multiple downlink channel estimation data from different domains as channel matrix training samples; the channel matrix sparsity threshold is preset according to the channel matrix sparsity distribution of the channel matrix training samples, and the channel matrix training samples are grouped according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples; and the at least one group of the channel matrix training samples is trained.

[0129] In an embodiment of the present invention, a system for training an artificial intelligence model for a wireless access network is deployed on a centralized unit (CU) and / or a distributed unit (DU) of the network side device, or on a logical entity across CUs, by collecting downlink channel estimation data in different domains (such as space-frequency domain, time-frequency domain, and space-frequency-time domain, etc.) as channel matrix training samples, preprocessing the channel matrix training samples, and presetting channel matrix sparsity thresholds (such as δ1, δ2…δ N-1 , where δ1<…δ N-1 ), for example: after channel analysis, a channel matrix sparsity threshold is selected to be set to δ1=0.17, and according to the sparsity of the channel matrix corresponding to the downlink signal estimation data, the downlink signal estimation data is grouped according to the channel matrix sparsity threshold, and for each group of data, a compression model of the channel matrix and a decompression model of the channel matrix based on an artificial intelligence autoencoder (such as: autoencoder) are designed and trained offline, wherein the compression model of the channel matrix is ​​recorded as E1, E2, ...E M , the decompression model of the channel matrix is ​​recorded as D1, D2, ... D M For example, the compression model of the channel matrix is ​​numbered as E1 and E2, and the decompression model of the channel matrix is ​​numbered as D1 and D2. Offline training is based on two pairs of compression models and decompression models of the channel matrix based on an artificial intelligence autoencoder (such as an autoencoder). The training method is as follows: Models E1 and D1 are trained using corresponding data in the channel matrix sparsity interval [0, δ1), and E2 and D2 are trained using corresponding data in the channel matrix sparsity interval [δ1, 1]. The number of output variables of E1 and E2 and the number of input variables of D1 and D2 can be designed as multiple groups of output / The number of input variables is selected, and the number of output / input variables with compression feedback accuracy higher than the preset accuracy threshold η=0.92 and the smallest number of feedback bits is selected as the optimal number of output / input variables; under the same matrix dimension, the higher the matrix sparsity, the smaller the matrix dimension after compression can be set, that is, the smaller the number of output variables of the compression model of the channel matrix and the number of input variables of the decompression model of the channel matrix can be set, and the compression model of the channel matrix and the corresponding number are synchronized to the user side device, and the decompression model of the channel matrix, the corresponding number and the sparsity threshold of the channel matrix are synchronized to the network side device.

[0130] In an embodiment of the present invention, optionally, the configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model, a modulation coding method, and time-frequency resources required for compression feedback information.

[0131] In the embodiment of the present invention, optionally, the following further comprises:

[0132] Receiving a channel state information reference number sent by a network side device according to configuration information of the target channel matrix compression model;

[0133] Channel matrix compression information completed by the channel matrix compression model selected according to the configuration information of the target channel matrix compression model is sent to the network side device.

[0134] In an embodiment of the present invention, the network side device sends the channel state information parameter number to the user side device according to the configuration; the user side device selects the corresponding channel matrix compression model according to the model number in the configuration information of the received target channel matrix compression model and completes the compression of the channel matrix, and feeds back the compressed channel information to the network side device according to the modulation coding mode and resource configuration set by the network side device; the network side device selects the channel matrix decompression model paired with the channel matrix compression model to decompress the channel information, and through channel reciprocity, places the channel matrix preprocessing and related calculations on the network side device, effectively reducing the calculation pressure of the user side device, and adopts a first-level network, which effectively reduces the complexity during training and reasoning.

[0135] Please refer to Figure 4 In the embodiment of the present invention, the overall process of the channel matrix compression feedback method is as follows:

[0136] The network side device collects multiple downlink channel estimation data from different domains as channel matrix training samples, sets a channel matrix sparsity threshold according to the channel matrix sparsity distribution of the channel matrix training samples, and groups the channel matrix training samples according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples; trains the at least one group of the channel matrix training samples to obtain at least one channel matrix decompression model and a corresponding channel matrix compression model; sends the at least one channel matrix compression model and the channel matrix sparsity threshold to the user side device;

[0137] The user-side device accesses the network, enters the RRC-CONNECTED state, and sends SRS to the network-side device as configured;

[0138] The network side device selects a target channel matrix compression model according to the sparsity of the channel matrix, obtains configuration information of the target channel matrix compression model according to the target channel matrix compression model, and sends the configuration information of the target channel matrix compression model to the user side device;

[0139] The network side device sends a channel state information reference signal to the user side device according to the configuration information of the target channel matrix compression model;

[0140] The user side device selects a channel matrix compression model according to the configuration information of the target channel matrix compression model, compresses the channel information, and feeds back the compressed channel information to the network side device;

[0141] The network side device decompresses the channel information;

[0142] The network side device listens to the channel detection reference signal according to a preset time period, obtains the latest sparsity of the channel matrix according to the channel detection reference signal obtained within the preset time period, and compares the sparsity of the latest channel matrix with the channel matrix sparsity threshold to obtain the latest configuration information of the channel matrix compression model; sends the latest configuration information of the channel matrix compression model to the user side device; the user side device reports the latest compressed channel information according to the configuration.

[0143] In an embodiment of the present invention, the network side device sends the channel state information parameter number to the user side device according to the configuration; the user side device selects the corresponding channel matrix compression model according to the model number in the configuration information of the received target channel matrix compression model and completes the compression of the channel matrix, and feeds back the compressed channel information to the network side device according to the modulation coding mode and resource configuration set by the network side device; the network side device selects the channel matrix decompression model paired with the channel matrix compression model to decompress the channel information, and through channel reciprocity, places the channel matrix preprocessing and related calculations on the network side device, effectively reducing the calculation pressure of the user side device, and adopts a first-level network, which effectively reduces the complexity during training and reasoning.

[0144] Please refer to Figure 5 , an embodiment of the present invention provides a network side device, including:

[0145] A first receiving module 51 is used to receive a channel sounding reference signal sent by a user side device;

[0146] A first processing module 52 is used to obtain the sparsity of the channel matrix according to the channel sounding reference signal, and obtain a target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and obtain configuration information of the target channel matrix compression model according to the target channel matrix compression model; the preset channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data;

[0147] The first sending module 53 is used to send configuration information of the target channel matrix compression model to the user side device.

[0148] In the embodiment of the present invention, optionally, the following further comprises:

[0149] The second processing module is used to obtain a plurality of downlink channel estimation data collected in different domains as channel matrix training samples;

[0150] A third processing module, configured to preset a channel matrix sparsity threshold according to the channel matrix sparsity distribution of the channel matrix training samples, and group the channel matrix training samples according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples;

[0151] A fourth processing module, used for training the at least one group of channel matrix training samples to obtain at least one channel matrix decompression model and a corresponding channel matrix compression model;

[0152] The third sending module is used to send the at least one channel matrix compression model and the channel matrix sparsity threshold to the user side device.

[0153] In the embodiment of the present invention, optionally, the first processing module includes:

[0154] A first processing submodule, configured to determine a modulation and coding scheme of channel state information compression feedback information according to the quality of the channel sounding reference signal;

[0155] The second processing submodule is used to obtain the configuration information of the target channel matrix compression model according to the modulation and coding mode and the reporting resource configuration of the channel state information compression feedback information.

[0156] In an embodiment of the present invention, optionally, the configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model and a feedback mode of the target channel matrix compression model, wherein the feedback mode of the target channel matrix compression model includes at least one of the following: a modulation and coding method of channel state information compression feedback information and a reporting resource configuration of the channel state information compression feedback information.

[0157] In the embodiment of the present invention, optionally, the following further comprises:

[0158] A fourth sending module, used to send a channel state information reference number to a user side device according to the configuration information of the target channel matrix compression model;

[0159] A third receiving module is used to receive channel matrix compression information completed by the channel matrix compression model selected by the user side device according to the configuration information of the target channel matrix compression model;

[0160] A decompression module is used to decompress the channel matrix compression information according to a channel matrix decompression model corresponding to the target channel matrix compression model.

[0161] In the embodiment of the present invention, optionally, the following further comprises:

[0162] A monitoring module, used for monitoring the channel sounding reference signal according to a preset time period;

[0163] A fifth processing module, configured to obtain the latest sparsity of the channel matrix according to the channel sounding reference signal obtained within the preset time period, and compare the sparsity of the latest channel matrix with the channel matrix sparsity threshold to obtain the latest configuration information of the channel matrix compression model;

[0164] The fifth sending module is used to send the latest configuration information of the channel matrix compression model to the user side device.

[0165] The network side device provided by the embodiment of the present invention can realize Figure 1 The various processes implemented by the method embodiment and achieving the same technical effect are not described here to avoid repetition.

[0166] Please refer to Figure 6 , an embodiment of the present invention provides a user side device, including:

[0167] The second sending mode, 61, is used to send a channel sounding reference signal to the network side device;

[0168] The second receiving module 62 is used to receive the configuration information of the target channel matrix compression model sent by the network side device, wherein the configuration information of the target channel matrix compression model is that the network side device obtains the sparsity of the channel matrix according to the channel detection reference signal, and obtains the target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and is obtained according to the target channel matrix compression model; the channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data.

[0169] In the embodiment of the present invention, optionally, the following further comprises:

[0170] A fourth receiving module, configured to receive at least one channel matrix compression model and the channel matrix sparsity threshold sent by a network side device;

[0171] Among them, the at least one channel matrix compression model and the channel matrix sparsity threshold are for the network side device to obtain multiple downlink channel estimation data from different domains as channel matrix training samples; the channel matrix sparsity threshold is preset according to the channel matrix sparsity distribution of the channel matrix training samples, and the channel matrix training samples are grouped according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples; and the at least one group of the channel matrix training samples is trained.

[0172] In an embodiment of the present invention, optionally, the configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model, a modulation coding method, and time-frequency resources required for compression feedback information.

[0173] In the embodiment of the present invention, optionally, the following further comprises:

[0174] A fifth receiving module, configured to receive a channel state information reference number sent by a network side device according to configuration information of the target channel matrix compression model;

[0175] The fifth sending module is used to send channel matrix compression information completed by the channel matrix compression model selected according to the configuration information of the target channel matrix compression model to the network side device.

[0176] The user side device provided by the embodiment of the present invention can realize Figure 3 The various processes implemented by the method embodiment and achieving the same technical effect are not described here to avoid repetition.

[0177] An embodiment of the present invention provides an electronic device 70, see Figure 7 As shown, Figure 7 This is a principle block diagram of an electronic device 70 according to an embodiment of the present invention, which includes a processor 71, a memory 72, and a program or instruction stored in the memory 72 and executable on the processor 71. When the program or instruction is executed by the processor, the steps in any one of the channel matrix compression feedback methods of the present invention are implemented.

[0178] An embodiment of the present invention provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the various processes of an embodiment of a method for channel matrix compression feedback such as any of the above items are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0179] Computer readable media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0180] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0181] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0182] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a service classification device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0183] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A method for channel matrix compression feedback, It is characterized in that Applied to network-side devices, including: Receiving a channel sounding reference signal sent by a user side device; The sparsity of the channel matrix is ​​obtained according to the channel sounding reference signal, and the target channel matrix compression model is obtained according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and the configuration information of the target channel matrix compression model is obtained according to the target channel matrix compression model; the preset channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data; The configuration information of the target channel matrix compression model is sent to the user side device.

2. The method for channel matrix compression feedback according to claim 1, It is characterized in that Also includes: Acquire a plurality of downlink channel estimation data collected in different domains as channel matrix training samples; Presetting a channel matrix sparsity threshold according to the channel matrix sparsity distribution of the channel matrix training samples, and grouping the channel matrix training samples according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples; Training the at least one group of channel matrix training samples to obtain at least one channel matrix decompression model and a corresponding channel matrix compression model; The at least one channel matrix compression model and the channel matrix sparsity threshold are sent to the user side device.

3. The method for channel matrix compression feedback according to claim 1, It is characterized in that The obtaining configuration information of the target channel matrix compression model according to the target channel matrix compression model includes: Determining a modulation and coding mode of channel state information compression feedback information according to the quality of the channel sounding reference signal; According to the modulation and coding mode and the reporting resource configuration of the channel state information compression feedback information, the configuration information of the target channel matrix compression model is obtained.

4. The method for channel matrix compression feedback according to claim 3, It is characterized in that The configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model and a feedback mode of the target channel matrix compression model, wherein the feedback mode of the target channel matrix compression model includes at least one of the following: a modulation and coding method of channel state information compression feedback information and a reporting resource configuration of channel state information compression feedback information.

5. The method for channel matrix compression feedback according to claim 1, It is characterized in that Also includes: Sending a channel state information reference number to a user side device according to the configuration information of the target channel matrix compression model; Receiving channel matrix compression information completed by a channel matrix compression model selected by a user-side device according to configuration information of the target channel matrix compression model; The channel matrix compression information is decompressed according to a channel matrix decompression model corresponding to the target channel matrix compression model.

6. The method for channel matrix compression feedback according to claim 1, It is characterized in that Also includes: Monitoring the channel sounding reference signal according to a preset time period; Obtaining the latest sparsity of the channel matrix according to the channel sounding reference signal obtained within the preset time period, and comparing the sparsity of the latest channel matrix with the channel matrix sparsity threshold to obtain the latest configuration information of the channel matrix compression model; Send the latest channel matrix compression model configuration information to the user side device.

7. A method for channel matrix compression feedback, It is characterized in that Applied to user-side devices, including: Sending a channel sounding reference signal to a network side device; Receive configuration information of a target channel matrix compression model sent by a network side device, wherein the configuration information of the target channel matrix compression model is that the network side device obtains the sparsity of the channel matrix according to the channel detection reference signal, and obtains the target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and the target channel matrix compression model is obtained according to the target channel matrix compression model; the preset channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data.

8. The method for channel matrix compression feedback according to claim 7, It is characterized in that Also includes: Receiving at least one channel matrix compression model and the channel matrix sparsity threshold sent by a network side device; The at least one channel matrix compression model and the channel matrix sparsity threshold are multiple downlink channel estimation data of different domains acquired and collected by the network side device as channel matrix training samples; A channel matrix sparsity threshold is preset according to the channel matrix sparsity distribution of the channel matrix training samples, and the channel matrix training samples are grouped according to the channel matrix sparsity threshold to obtain at least one group of the channel matrix training samples; and the at least one group of the channel matrix training samples is trained.

9. The method for channel matrix compression feedback according to claim 7, It is characterized in that The configuration information of the target channel matrix compression model includes at least one of the following: a model number of the target channel matrix compression model, a modulation coding mode, and time-frequency resources required for compression feedback information.

10. The method for channel matrix compression feedback according to claim 7, It is characterized in that Also includes: Receiving a channel state information reference number sent by a network side device according to configuration information of the target channel matrix compression model; Channel matrix compression information completed by the channel matrix compression model selected according to the configuration information of the target channel matrix compression model is sent to the network side device.

11. A network side device, It is characterized in that include: A first receiving module, used to receive a channel sounding reference signal sent by a user side device; A first processing module is used to obtain the sparsity of the channel matrix according to the channel sounding reference signal, and obtain a target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and obtain configuration information of the target channel matrix compression model according to the target channel matrix compression model; the preset channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data; The first sending module is used to send configuration information of the target channel matrix compression model to the user side device.

12. A user side device, It is characterized in that include: A second sending module, used to send a channel sounding reference signal to a network side device; The second receiving module is used to receive the configuration information of the target channel matrix compression model sent by the network side device, wherein the configuration information of the target channel matrix compression model is that the network side device obtains the sparsity of the channel matrix according to the channel detection reference signal, and obtains the target channel matrix compression model according to the sparsity of the channel matrix and a preset channel matrix sparsity threshold, and is obtained according to the target channel matrix compression model; the channel matrix sparsity threshold is obtained according to the sparsity distribution of the collected downlink signal estimation data.

13. An electronic device, Features: The method comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the method for channel matrix compression feedback as described in any one of claims 1 to 6, or the steps in the method for channel matrix compression feedback as described in any one of claims 7 to 10.

14. A readable storage medium, Features: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, it implements the method for channel matrix compression feedback as described in any one of claims 1 to 7, or the steps in the method for channel matrix compression feedback as described in any one of claims 7 to 10.

Images