Channel state information (CSI) prediction method and device, and storage medium

Abstract

The invention relates to a CSI prediction method and device and a storage medium, and belongs to the technical field of communication. The method is applied to UE, and the method comprises the following steps: receiving a downlink reference signal sent by a base station, and estimating to obtain a first downlink CSI and a downlink communication environment parameter based on the downlink reference signal; according to the first downlink CSI, the downlink communication environment parameter and a pre-trained CSI prediction model, predicting to obtain a second downlink CSI corresponding to the time when the base station sends the PDSCH; and sending the second downlink CSI to the base station. According to the invention, the timeliness of the downlink CSI can be improved.

Description

Technical Field

[0001] The present application relates to the field of communication technologies, and in particular, to a method, apparatus, and storage medium for predicting CSI. Background Art

[0002] Currently, in the scenario of high-speed movement of a UE (User Equipment), there is a problem of poor timeliness in the downlink CSI (Channel State Information) fed back by the UE. Specifically, when the base station performs downlink scheduling based on the downlink CSI fed back by the UE, since the UE is moving at high speed, the downlink communication environment where the UE is located has changed greatly at this time, and the downlink CSI fed back by the UE can no longer reflect the current downlink communication environment of the UE.

[0003] The poor timeliness of the downlink CSI fed back by the UE will lead to a series of problems such as inaccurate base station scheduling and decreased data rate. Therefore, there is an urgent need for a method for predicting downlink CSI, which can improve the timeliness of downlink CSI. Summary of the Invention

[0004] Based on this, in view of the above technical problems, a method, apparatus, and storage medium for predicting CSI are provided.

[0005] In a first aspect, a method for predicting channel state information CSI is provided. The method is applied to a user equipment UE, and the method includes:

[0006] Receiving a downlink reference signal sent by a base station, and estimating a first downlink CSI and downlink communication environment parameters based on the downlink reference signal;

[0007] Predicting a second downlink CSI corresponding to the transmission time of the physical downlink shared channel PDSCH of the base station according to the first downlink CSI, the downlink communication environment parameters, and a pre-trained CSI prediction model;

[0008] Sending the second downlink CSI to the base station.

[0009] As an optional implementation manner, the predicting a second downlink CSI corresponding to the transmission time of the PDSCH of the base station according to the first downlink CSI, the downlink communication environment parameters, and a pre-trained CSI prediction model includes:

[0010] Determining a prediction time interval according to the downlink communication environment parameters, a first time parameter provided by the UE stored in advance, and a second time parameter provided by the base station;

[0011] Predict a second downlink CSI corresponding to the transmission time of the PDSCH of the base station according to the first downlink CSI, the predicted time interval, and a pre-trained CSI prediction model.

[0012] As an alternative implementation, the determining the predicted time interval according to the downlink communication environment parameters, the first time parameter provided by the UE and stored in advance, and the second time parameter provided by the base station includes:

[0013] Determine the transmission time of the downlink reference signal according to the downlink communication environment parameters;

[0014] Determine the transmission time of the second downlink CSI according to the downlink communication environment parameters and the first time parameter;

[0015] Determine the predicted time interval according to the transmission time of the downlink reference signal, the transmission time of the second downlink CSI, the first time parameter, and the second time parameter.

[0016] As an alternative implementation, the downlink communication environment parameters include at least one or more of the moving speed v of the UE, the moving azimuth angle θ of the UE relative to the base station, and the relative distance between the UE and the base station;

[0017] The determining the transmission time T1 of the downlink reference signal according to the downlink communication environment parameters:

[0018] T1 = S0 / c;

[0019] where T1 represents the transmission time of the downlink reference signal, S0 represents the relative distance between the UE and the base station when the base station transmits the downlink reference signal, and c represents the speed of light; or,

[0020] The determining the transmission time T2 of the second downlink CSI according to the downlink communication environment parameters and the first time parameter:

[0021]

[0022] where T2 represents the transmission time of the second downlink CSI, v represents the moving speed of the UE, θ represents the moving azimuth angle of the UE relative to the base station, and t1 represents the first time parameter.

[0023] As an alternative implementation, the sending the second downlink CSI to the base station includes:

[0024] Obtain a compressed second downlink CSI according to the second downlink CSI, a preset compression ratio, and a pre-trained CSI compression model;

[0025] Send the compressed second downlink CSI and the compression ratio to the base station.

[0026] As an optional implementation manner, the first time parameter is the time interval between the last symbol of the physical downlink shared channel PDCCH that triggers the CSI report and the reporting of the second downlink CSI; the second time parameter includes the time for scheduling the PDSCH and preparing the PDSCH; or,

[0027] The first time parameter is the time interval between the last symbol of the PDCCH that triggers the CSI report and the reporting of the compressed second downlink CSI; the second time parameter includes the decompression time of the compressed second downlink CSI and the time for scheduling the PDSCH and preparing the PDSCH.

[0028] As an optional implementation manner, the method further includes:

[0029] Determine the time interval level corresponding to the predicted time interval according to the correspondence between the time interval sent by the base station and the time interval level.

[0030] As an optional implementation manner, the method further includes:

[0031] Send time indication information to the base station; the time indication information is the predicted time interval, or the time indication information is the time interval level corresponding to the predicted time interval.

[0032] As an optional implementation manner, the predicting, according to the first downlink CSI, the predicted time interval, and a pre-trained CSI prediction model, the second downlink CSI corresponding to the PDSCH transmission time of the base station includes:

[0033] Obtain a preset number of historical downlink CSIs;

[0034] Predict, according to the first downlink CSI, each of the historical downlink CSIs, the predicted time interval, and a pre-trained CSI prediction model, the second downlink CSI corresponding to the PDSCH transmission time of the base station.

[0035] As an optional implementation manner, obtaining the CSI prediction model by the following method includes:

[0036] Obtain a basic data set, where the basic data set contains the historical downlink CSIs of the UE;

[0037] Respectively determine the training data set and the label data set corresponding to each predicted time interval in the basic data set;

[0038] Based on each of the training data sets, each of the label data sets, and each of the prediction time intervals, train the initial CSI prediction model to obtain a trained CSI prediction model.

[0039] As an alternative implementation, the determining the training data set and the label data set corresponding to each prediction time interval includes:

[0040] Determine a plurality of first historical downlink CSI groups corresponding to each prediction time interval, where each of the first historical downlink CSI groups includes a first historical downlink CSI and a second historical downlink CSI, and the difference in the channel estimation time between the first historical downlink CSI and the second historical downlink CSI is the prediction time interval;

[0041] Determine a plurality of the first historical downlink CSIs as the training data set, and determine a plurality of the second historical downlink CSIs as the label data set.

[0042] As an alternative implementation, the determining the training data set and the label data set corresponding to each prediction time interval includes:

[0043] Determine a plurality of second historical downlink CSI groups corresponding to each prediction time interval, where each of the second historical downlink CSI groups includes a third historical downlink CSI, a fourth historical downlink CSI, and a preset number of fifth historical downlink CSIs, the difference in the channel estimation time between the third historical downlink CSI and the fourth historical downlink CSI is the prediction time interval, and the channel estimation times of the preset number of fifth historical downlink CSIs are earlier than the channel estimation time of the third historical downlink CSI;

[0044] Determine a plurality of the third historical downlink CSIs and each of the fifth historical downlink CSIs corresponding to the third historical downlink CSI as the training data set, and determine a plurality of the fourth historical downlink CSIs as the label data set.

[0045] As an alternative implementation, training to obtain the CSI compression model is performed by the following method, including:

[0046] Determine the historical downlink CSI of the UE as the third training data set;

[0047] Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the third label data set;

[0048] Based on the third training data set and the third label data set, train the initial CSI compression model to obtain a trained CSI compression model.

[0049] Second aspect, a method for predicting channel state information (CSI) is provided. The method is applied to a base station and includes:

[0050] Sending a downlink reference signal to a user equipment (UE);

[0051] Obtaining the predicted downlink CSI of the UE;

[0052] Performing scheduling according to the downlink CSI and sending a physical downlink shared channel (PDSCH) to the UE.

[0053] As an optional implementation manner, the obtaining the predicted downlink CSI of the UE includes:

[0054] Receiving the compressed downlink CSI and the compression ratio sent by the UE;

[0055] Obtaining the downlink CSI according to the compressed downlink CSI, the compression ratio, and a pre-trained CSI decompression model.

[0056] As an optional implementation manner, the method further includes:

[0057] Receiving the time indication information sent by the UE; the time indication information is the predicted time interval determined by the UE, or the time interval level corresponding to the predicted time interval determined by the UE.

[0058] As an optional implementation manner, when the time indication information is the time interval level corresponding to the predicted time interval determined by the UE, the method further includes:

[0059] Determining the time interval corresponding to the time indication information in the correspondence between the time interval and the time interval level, and determining the predicted time interval by randomly selecting a value or taking the median in the time interval corresponding to the time indication information.

[0060] As an optional implementation manner, the method further includes:

[0061] Determining a predicted time interval range according to the range of the downlink communication environment parameters of the UE stored in advance, the range of the first time parameters provided by the UE, and the range of the second time parameters provided by the base station;

[0062] Dividing the predicted time interval range according to a preset level division rule to obtain the correspondence between the time interval and the time interval level;

[0063] Sending the correspondence between the time interval and the time interval level to the UE.

[0064] As an alternative implementation, the CSI decompression model is trained in the following manner, including:

[0065] Obtain the historical downlink CSI of the UE;

[0066] Determine the compressed and quantized historical downlink CSI using the measurement matrix of compressive sensing as the training dataset;

[0067] Determine the historical downlink CSI after decompression using the compressive sensing recovery algorithm as the label dataset;

[0068] Train the initial CSI decompression model based on the training dataset and the label dataset to obtain the trained CSI decompression model.

[0069] In a third aspect, a user equipment (UE) is provided, including a memory, a transceiver, and a processor;

[0070] Among them, the memory is used to store computer programs; the transceiver is used to transmit and receive data under the control of the processor, and the processor is used to read the computer programs in the memory and perform the following operations:

[0071] Receive the downlink reference signal sent by the base station, and estimate the first downlink CSI and the downlink communication environment parameters based on the downlink reference signal;

[0072] According to the first downlink CSI, the downlink communication environment parameters, and the pre-trained CSI prediction model, predict the second downlink CSI corresponding to the time when the base station sends the physical downlink shared channel (PDSCH);

[0073] Send the second downlink CSI to the base station.

[0074] As an alternative implementation, the step of predicting the second downlink CSI corresponding to the time when the base station sends the PDSCH according to the first downlink CSI, the downlink communication environment parameters, and the pre-trained CSI prediction model includes:

[0075] Determine the prediction time interval according to the downlink communication environment parameters, the first time parameter provided by the UE stored in advance, and the second time parameter provided by the base station;

[0076] According to the first downlink CSI, the prediction time interval, and the pre-trained CSI prediction model, predict the second downlink CSI corresponding to the time when the base station sends the PDSCH.

[0077] As an alternative implementation, determining the predicted time interval according to the downlink communication environment parameters, the first time parameter provided by the UE stored in advance, and the second time parameter provided by the base station includes:

[0078] Determine the transmission time of the downlink reference signal according to the downlink communication environment parameters;

[0079] Determine the transmission time of the second downlink CSI according to the downlink communication environment parameters and the first time parameter;

[0080] Determine the predicted time interval according to the transmission time of the downlink reference signal, the transmission time of the second downlink CSI, the first time parameter, and the second time parameter.

[0081] As an alternative implementation, the downlink communication environment parameters at least include at least one or more of the moving speed v of the UE, the moving azimuth angle θ of the UE relative to the base station, and the relative distance between the UE and the base station;

[0082] Determine the transmission time T1 of the downlink reference signal according to the downlink communication environment parameters:

[0083] T1 = S0 / c;

[0084] where T1 represents the transmission time of the downlink reference signal, S0 represents the relative distance between the UE and the base station when the base station sends the downlink reference signal, and c represents the speed of light; or,

[0085] Determine the transmission time T2 of the second downlink CSI according to the downlink communication environment parameters and the first time parameter:

[0086]

[0087] where T2 represents the transmission time of the second downlink CSI, v represents the moving speed of the UE, θ represents the moving azimuth angle of the UE relative to the base station, and t1 represents the first time parameter.

[0088] As an alternative implementation, sending the second downlink CSI to the base station includes:

[0089] Obtain the compressed second downlink CSI according to the second downlink CSI, a preset compression ratio, and a pre-trained CSI compression model;

[0090] Send the compressed second downlink CSI and the compression ratio to the base station.

[0091] As an alternative implementation, the first time parameter is the time interval between the last symbol of the physical downlink shared channel PDCCH that triggers the CSI report and the reporting of the second downlink CSI; the second time parameter includes the time for scheduling the PDSCH and preparing the PDSCH; or,

[0092] The first time parameter is the time interval between the last symbol of the PDCCH that triggers the CSI report and the reporting of the compressed second downlink CSI; the second time parameter includes the decompression time of the compressed second downlink CSI and the time for scheduling the PDSCH and preparing the PDSCH.

[0093] As an alternative implementation, the processor is further configured to read the computer program in the memory and perform the following operations:

[0094] Determine the time interval level corresponding to the predicted time interval according to the correspondence between the time interval and the time interval level sent by the base station.

[0095] As an alternative implementation, the processor is further configured to read the computer program in the memory and perform the following operations:

[0096] Send time indication information to the base station; the time indication information is the predicted time interval, or the time indication information is the time interval level corresponding to the predicted time interval.

[0097] As an alternative implementation, the predicting the second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the predicted time interval, and a pre-trained CSI prediction model includes:

[0098] Obtain a preset number of historical downlink CSIs;

[0099] Predict the second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, each of the historical downlink CSIs, the predicted time interval, and a pre-trained CSI prediction model.

[0100] As an alternative implementation, obtaining the CSI prediction model by the following method includes:

[0101] Obtain a basic data set, where the basic data set contains the historical downlink CSIs of the UE;

[0102] Respectively determine the training data set and the label data set corresponding to each predicted time interval in the basic data set;

[0103] Train the initial CSI prediction model based on each of the training data sets, each of the label data sets, and each of the prediction time intervals to obtain the trained CSI prediction model.

[0104] As an alternative implementation, the determining the training data set and the label data set corresponding to each prediction time interval includes:

[0105] Determine multiple first historical downlink CSI groups corresponding to each prediction time interval, where each of the first historical downlink CSI groups includes a first historical downlink CSI and a second historical downlink CSI, and the difference in the channel estimation time between the first historical downlink CSI and the second historical downlink CSI is the prediction time interval;

[0106] Determine multiple of the first historical downlink CSIs as the training data set, and determine multiple of the second historical downlink CSIs as the label data set.

[0107] As an alternative implementation, the determining the training data set and the label data set corresponding to each prediction time interval includes:

[0108] Determine multiple second historical downlink CSI groups corresponding to each prediction time interval, where each of the second historical downlink CSI groups includes a third historical downlink CSI, a fourth historical downlink CSI, and a preset number of fifth historical downlink CSIs, and the difference in the channel estimation time between the third historical downlink CSI and the fourth historical downlink CSI is the prediction time interval, and the channel estimation times of the preset number of fifth historical downlink CSIs are earlier than the channel estimation time of the third historical downlink CSI;

[0109] Determine multiple of the third historical downlink CSIs and each of the fifth historical downlink CSIs corresponding to the third historical downlink CSI as the training data set, and determine multiple of the fourth historical downlink CSIs as the label data set.

[0110] As an alternative implementation, the CSI compression model is trained in the following manner, including:

[0111] Determine the historical downlink CSI of the UE as the third training data set;

[0112] Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the third label data set;

[0113] Train the initial CSI compression model based on the third training data set and the third label data set to obtain the trained CSI compression model.

[0114] In a fourth aspect, a base station is provided, including a memory, a transceiver, and a processor;

[0115] wherein, the memory is used for storing a computer program; the transceiver is used for transceiving data under the control of the processor, and the processor is used for reading the computer program in the memory and performing the following operations:

[0116] Sending a downlink reference signal to a user equipment UE;

[0117] Obtaining the downlink CSI predicted by the UE;

[0118] Performing scheduling according to the downlink CSI and sending a physical downlink shared channel PDSCH to the UE.

[0119] As an optional implementation manner, the obtaining the downlink CSI predicted by the UE includes:

[0120] Receiving the compressed downlink CSI and the compression ratio sent by the UE;

[0121] Obtaining the downlink CSI according to the compressed downlink CSI, the compression ratio, and a pre-trained CSI decompression model.

[0122] As an optional implementation manner, the processor is further used for reading the computer program in the memory and performing the following operations:

[0123] Receiving the time indication information sent by the UE; the time indication information is the predicted time interval determined by the UE, or the time indication information is the time interval level corresponding to the predicted time interval determined by the UE.

[0124] As an optional implementation manner, in the case that the time indication information is the time interval level corresponding to the predicted time interval determined by the UE, the processor is further used for reading the computer program in the memory and performing the following operations:

[0125] In the correspondence between the time interval and the time interval level, determining the time interval corresponding to the time indication information, and in the time interval corresponding to the time indication information, determining the predicted time interval by randomly selecting a value or taking the median.

[0126] As an optional implementation manner, the processor is further used for reading the computer program in the memory and performing the following operations:

[0127] Determining a predicted time interval range according to the range of the downlink communication environment parameters of the UE stored in advance, the range of the first time parameters provided by the UE, and the range of the second time parameters provided by the base station;

[0128] Divide the predicted time interval range according to a preset level division rule to obtain the corresponding relationship between the time interval and the time interval level;

[0129] Send the corresponding relationship between the time interval and the time interval level to the UE.

[0130] As an optional implementation manner, the CSI decompression model is trained through the following manner, including:

[0131] Obtain the historical downlink CSI of the UE;

[0132] Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the training data set;

[0133] Determine the historical downlink CSI after decompression using the compressive sensing recovery algorithm as the label data set;

[0134] Train the initial CSI decompression model based on the training data set and the label data set to obtain the trained CSI decompression model.

[0135] In a fifth aspect, a prediction device for channel state information CSI is provided. The device is applied to a user equipment UE, and the device includes:

[0136] An estimation unit, configured to receive a downlink reference signal sent by a base station, and estimate a first downlink CSI and downlink communication environment parameters based on the downlink reference signal;

[0137] A prediction unit, configured to predict a second downlink CSI corresponding to a transmission physical downlink shared channel PDSCH time of the base station according to the first downlink CSI, the downlink communication environment parameters, and a pre-trained CSI prediction model;

[0138] A first sending unit, configured to send the second downlink CSI to the base station.

[0139] As an optional implementation manner, the prediction unit is specifically configured to:

[0140] Determine a prediction time interval according to the downlink communication environment parameters, a first time parameter provided by the UE stored in advance, and a second time parameter provided by the base station;

[0141] Predict a second downlink CSI corresponding to a transmission PDSCH time of the base station according to the first downlink CSI, the prediction time interval, and a pre-trained CSI prediction model.

[0142] As an optional implementation manner, the prediction unit is specifically configured to:

[0143] Determine the transmission time of the downlink reference signal according to the downlink communication environment parameters;

[0144] Determine the transmission time of the second downlink CSI according to the downlink communication environment parameters and the first time parameter;

[0145] Determine the predicted time interval according to the transmission time of the downlink reference signal, the transmission time of the second downlink CSI, the first time parameter, and the second time parameter.

[0146] As an optional implementation manner, the downlink communication environment parameters at least include at least one or more of the moving speed v of the UE, the moving azimuth angle θ of the UE relative to the base station, and the relative distance between the UE and the base station;

[0147] The determining the transmission time T1 of the downlink reference signal according to the downlink communication environment parameters:

[0148] T1 = S0 / c;

[0149] wherein, T1 represents the transmission time of the downlink reference signal, S0 represents the relative distance between the UE and the base station when the base station sends the downlink reference signal, and c represents the speed of light; or,

[0150] The determining the transmission time T2 of the second downlink CSI according to the downlink communication environment parameters and the first time parameter:

[0151]

[0152] wherein, T2 represents the transmission time of the second downlink CSI, v represents the moving speed of the UE, θ represents the moving azimuth angle of the UE relative to the base station, and t1 represents the first time parameter.

[0153] As an optional implementation manner, the first sending unit is specifically configured to:

[0154] Obtain the compressed second downlink CSI according to the second downlink CSI, a preset compression ratio, and a pre-trained CSI compression model;

[0155] Send the compressed second downlink CSI and the compression ratio to the base station.

[0156] As an optional implementation manner, the first time parameter is the time interval from the last symbol of the physical downlink shared channel PDCCH that triggers the CSI report to the reporting of the second downlink CSI; the second time parameter includes the time for scheduling and preparing the PDSCH; or,

[0157] The first time parameter is the time interval from the last symbol of the PDCCH that triggers CSI reporting to the reporting of the compressed second downlink CSI; the second time parameter includes the decompression time of the compressed second downlink CSI and the time for scheduling and preparing the PDSCH.

[0158] As an optional implementation manner, the device further includes:

[0159] A determination unit, configured to determine the time interval level corresponding to the predicted time interval according to the correspondence between the time interval and the time interval level sent by the base station.

[0160] As an optional implementation manner, the device further includes:

[0161] A second sending unit, configured to send time indication information to the base station; the time indication information is the predicted time interval, or the time indication information is the time interval level corresponding to the predicted time interval.

[0162] As an optional implementation manner, the prediction unit is specifically configured to:

[0163] Obtain a preset number of historical downlink CSIs;

[0164] According to the first downlink CSI, each of the historical downlink CSIs, the predicted time interval, and a pre-trained CSI prediction model, predict the second downlink CSI corresponding to the PDSCH transmission time of the base station.

[0165] As an optional implementation manner, the first model training unit is specifically configured to:

[0166] Obtain a basic data set, where the basic data set includes the historical downlink CSIs of the UE;

[0167] Respectively determine the training data set and the label data set corresponding to each predicted time interval in the basic data set;

[0168] Based on each of the training data sets, each of the label data sets, and each of the predicted time intervals, train an initial CSI prediction model to obtain a trained CSI prediction model.

[0169] As an optional implementation manner, the first model training unit is specifically configured to:

[0170] Determine multiple first historical downlink CSI groups corresponding to each prediction time interval, where each of the first historical downlink CSI groups includes a first historical downlink CSI and a second historical downlink CSI, and the difference in the channel estimation time between the first historical downlink CSI and the second historical downlink CSI is the prediction time interval;

[0171] Determine multiple of the first historical downlink CSIs as the training data set, and determine multiple of the second historical downlink CSIs as the label data set.

[0172] As an alternative implementation, the first model training unit is specifically configured to:

[0173] Determine multiple second historical downlink CSI groups corresponding to each prediction time interval, where each of the second historical downlink CSI groups includes a third historical downlink CSI, a fourth historical downlink CSI, and a preset number of fifth historical downlink CSIs, the difference in the channel estimation time between the third historical downlink CSI and the fourth historical downlink CSI is the prediction time interval, and the channel estimation times of the preset number of fifth historical downlink CSIs are earlier than the channel estimation time of the third historical downlink CSI;

[0174] Determine multiple of the third historical downlink CSIs and each of the fifth historical downlink CSIs corresponding to the third historical downlink CSI as the training data set, and determine multiple of the fourth historical downlink CSIs as the label data set.

[0175] As an alternative implementation, the second model training unit is specifically configured to:

[0176] Determine the historical downlink CSI of the UE as the third training data set;

[0177] Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the third label data set;

[0178] Train the initial CSI compression model based on the third training data set and the third label data set to obtain the trained CSI compression model.

[0179] In a sixth aspect, there is provided a prediction apparatus for channel state information (CSI), the apparatus is applied to a base station, and the apparatus includes:

[0180] A first sending unit, configured to send a downlink reference signal to a user equipment (UE);

[0181] An obtaining unit, configured to obtain the predicted downlink CSI of the UE;

[0182] A second transmission unit, configured to perform scheduling according to the downlink CSI and transmit a Physical Downlink Shared Channel (PDSCH) to the UE.

[0183] As an alternative implementation, the obtaining unit is specifically configured to:

[0184] Receive the compressed downlink CSI and the compression ratio sent by the UE;

[0185] Obtain the downlink CSI according to the compressed downlink CSI, the compression ratio, and a pre-trained CSI decompression model.

[0186] As an alternative implementation, the apparatus further includes:

[0187] A receiving unit, configured to receive the time indication information sent by the UE; the time indication information is a predicted time interval determined by the UE, or the time indication information is a time interval level corresponding to the predicted time interval determined by the UE.

[0188] As an alternative implementation, when the time indication information is a time interval level corresponding to the predicted time interval determined by the UE, the apparatus further includes:

[0189] A first determination unit, configured to determine, in the correspondence between time intervals and time interval levels, the time interval corresponding to the time indication information, and determine a predicted time interval by randomly selecting a value or taking a median in the time interval corresponding to the time indication information.

[0190] As an alternative implementation, the apparatus further includes:

[0191] A second determination unit, configured to determine a predicted time interval range according to a range of downlink communication environment parameters of the UE stored in advance, a first time parameter range provided by the UE, and a second time parameter range provided by the base station;

[0192] A partitioning unit, configured to partition the predicted time interval range according to a preset level partitioning rule to obtain a correspondence between time intervals and time interval levels;

[0193] A third transmission unit, configured to transmit the correspondence between the time intervals and the time interval levels to the UE.

[0194] As an alternative implementation, the model training unit is specifically configured to:

[0195] Obtain the historical downlink CSI of the UE;

[0196] Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the training data set;

[0197] Determine the historical downlink CSI after decompression using the compressive sensing recovery algorithm as the label data set;

[0198] Based on the training data set and the label data set, train the initial CSI decompression model to obtain the trained CSI decompression model.

[0199] In a seventh aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the method steps described in the first aspect or the second aspect are implemented.

[0200] This application provides a method, apparatus, and storage medium for predicting CSI. The technical solutions provided by the embodiments of this application at least bring the following beneficial effects: The UE receives the downlink reference signal sent by the base station, and based on the downlink reference signal, estimates the first downlink CSI and the downlink communication environment parameters. Then, the UE predicts the second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the downlink communication environment parameters, and the pre-trained CSI prediction model, and sends the second downlink CSI to the base station. In this way, the second downlink CSI used by the base station to send the PDSCH to the UE at the PDSCH transmission time can better match the downlink communication environment where the UE is located at the PDSCH transmission time, thus ensuring the timeliness of the downlink CSI.

[0201] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit this application. Description of the Drawings

[0202] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0203] Figure 1 It is a flowchart of a method for predicting CSI provided by an embodiment of this application;

[0204] Figure 2 It is a flowchart of a method for predicting CSI provided by an embodiment of this application;

[0205] Figure 3 It is a flowchart of a method for predicting CSI provided by an embodiment of this application;

[0206] Figure 4 A schematic structural diagram of a UE provided by an embodiment of the present application;

[0207] Figure 5 A schematic structural diagram of a base station provided by an embodiment of the present application;

[0208] Figure 6 A schematic structural diagram of a CSI prediction device provided by an embodiment of the present application;

[0209] Figure 7 A schematic structural diagram of a CSI prediction device provided by an embodiment of the present application. Detailed implementation manners

[0210] In the embodiments of the present invention, the term "and / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects before and after.

[0211] In the embodiments of the present application, the term "plurality" refers to two or more, and other quantifiers are similar thereto.

[0212] Next, the technical solutions in the embodiments of the present application will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0213] Next, a CSI prediction method provided by an embodiment of the present application will be described in detail in combination with the specific implementation manners. Figure 1 A flowchart of a CSI prediction method provided by an embodiment of the present application. This CSI prediction method is applied to a UE, as Figure 1 shown, and the specific steps are as follows:

[0214] Step 101: Receive the downlink reference signal sent by the base station, and estimate the first downlink CSI and the downlink communication environment parameters based on the downlink reference signal.

[0215] In implementation, the base station sends a PDCCH (Physical Downlink Control Channel) for triggering CSI reporting and downlink reference information to the UE. Correspondingly, the UE receives the PDCCH for triggering CSI reporting and the downlink reference signal sent by the base station. The UE can perform channel estimation based on the downlink reference signal to obtain a first downlink CSI corresponding to the downlink reference signal. At the same time, the UE can also perform communication environment estimation based on the downlink reference signal to obtain a downlink communication environment parameter corresponding to the downlink reference signal.

[0216] Step 102: Predict a second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the downlink communication environment parameter, and a pre-trained CSI prediction model.

[0217] In implementation, in order to make the downlink CSI used by the base station better match the downlink communication environment where the UE is located at the PDSCH transmission time, the UE can predict a second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the downlink communication environment parameter, and a pre-trained CSI prediction model.

[0218] As an optional implementation manner, in the above step 102, as Figure 2 shown, the processing procedure for the UE to predict a second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the downlink communication environment parameter, and a pre-trained CSI prediction model includes but is not limited to the following manner.

[0219] Step 201: Determine a prediction time interval according to the downlink communication environment parameter, a first time parameter provided by the UE stored in advance, and a second time parameter provided by the base station.

[0220] In implementation, when the base station sends a PDSCH to the UE, in order to make the downlink CSI used by the base station better match the downlink communication environment where the UE is located at the PDSCH transmission time. The UE can determine a prediction time interval according to the downlink communication environment parameter, a first time parameter provided by the UE stored in advance, and a second time parameter provided by the base station.

[0221] It should be noted that the first time parameter and the second time parameter in the embodiments of the present application can be obtained in different ways.

[0222] The first way:

[0223] The first time parameter is the time interval from the last symbol of the PDCCH that triggers the CSI report to the reporting of the second downlink CSI. The second time parameter includes at least the time for scheduling the PDSCH and preparing the PDSCH. The first time parameter includes the estimation time of the first downlink CSI, the estimation time of the downlink communication environment parameters, the calculation time of the prediction time interval, the prediction time of the second downlink CSI, etc.

[0224] The second method:

[0225] The first time parameter is the time interval from the last symbol of the PDCCH that triggers the CSI report to the reporting of the compressed second downlink CSI. The second time parameter includes at least the decompression time of the compressed second downlink CSI and the time for scheduling the PDSCH and preparing the PDSCH. The first time parameter includes the estimation time of the first downlink CSI, the estimation time of the downlink communication environment parameters, the calculation time of the prediction time interval, the prediction time of the second downlink CSI, the compression time of the second downlink CSI, etc.

[0226] To improve the accuracy of predicting the second CSI, when using the current downlink CSI (i.e., the first downlink CSI) as the input parameter, the first parameter and the second parameter obtained by the first method can be preferentially selected as the input parameters to predict the second CSI.

[0227] When using the CSI compression model to determine the predicted second CSI, the first parameter and the second parameter obtained by the second method can be preferentially selected as the input parameters.

[0228] Optionally, the downlink communication environment parameters include at least one or more of the UE's moving speed, the moving azimuth angle relative to the base station, and the relative distance from the base station. Among them, the moving azimuth angle of the UE relative to the base station is the angle formed by the line connecting the UE and the base station and the moving direction of the UE.

[0229] In the above step 201, the specific processing procedure for the UE to determine the prediction time interval according to the downlink communication environment parameters, the first time parameter provided by the UE stored in advance, and the second time parameter provided by the base station is as follows:

[0230] Step 1, determine the transmission time of the downlink reference signal according to the downlink communication environment parameters.

[0231] In implementation, the UE can determine the transmission time of the downlink reference signal according to the downlink communication environment parameters.

[0232] As an optional implementation method, the UE determines the transmission time T1 of the downlink reference signal according to the downlink communication environment parameters:

[0233] T1 = S0 / c

[0234] Wherein, T1 represents the transmission time of the downlink reference signal, S0 represents the relative distance between the UE and the base station when the base station transmits the downlink reference signal, and c represents the speed of light.

[0235] Step 2: Determine the transmission time of the second downlink CSI according to the downlink communication environment parameters and the first time parameter.

[0236] In implementation, in the high-speed moving scenario of the UE, the UE moves at a high speed over time, the distance between the UE and the base station also changes over time, and the transmission time taken for the UE to transmit the second downlink CSI to the base station also changes over time. Based on this, the UE can determine the transmission time of the second downlink CSI according to the downlink communication environment parameters and the first time parameter. Wherein, the predicted transmission time is the transmission time taken for the UE to transmit the second downlink CSI to the base station at the target position after moving from the current position to the target position after the first time parameter.

[0237] As an optional implementation manner, the UE determines the transmission time T2 of the compressed second downlink CSI according to the downlink communication environment parameters and the first time parameter:

[0238]

[0239] Wherein, T2 represents the transmission time of the second downlink CSI, v represents the moving speed of the UE, θ represents the moving azimuth angle of the UE relative to the base station, and t1 represents the first time parameter.

[0240] Step 3: Determine the prediction time interval according to the transmission time of the downlink reference signal, the transmission time of the second downlink CSI, and the first time parameter.

[0241] In implementation, the UE determines the sum value of the transmission time of the downlink reference signal, the transmission time of the second downlink CSI, the first time parameter provided by the UE, and the second time parameter provided by the base station as the prediction time interval. Wherein, the prediction time interval is the time elapsed from the last symbol of the PDCCH that triggers the CSI report until the base station sends the PDSCH to the UE (excluding the transmission time taken for the PDSCH to be transmitted from the base station to the UE).

[0242] Step 202: Predict the second downlink CSI corresponding to the PDSCH transmission moment of the base station according to the first downlink CSI, the prediction time interval, and the pre-trained CSI prediction model.

[0243] In implementation, the UE inputs the first downlink CSI and the prediction time interval into a pre-trained CSI prediction model. Correspondingly, the CSI prediction model outputs the second downlink CSI corresponding to the time when the base station transmits the PDSCH. Since the current time delayed by the prediction time interval is the time when the base station transmits the PDSCH. Therefore, based on the first downlink CSI and the prediction time interval, the second downlink CSI obtained through the CSI prediction model is the second downlink CSI corresponding to the time when the base station transmits the PDSCH. In this way, the second downlink CSI used by the base station to transmit the PDSCH to the UE at the time of transmitting the PDSCH can better match the downlink communication environment in which the UE is located at the time of transmitting the PDSCH, thereby ensuring the timeliness of the second downlink CSI. Optionally, the CSI prediction model may adopt an LSTM (Long-Short Term Memory) neural network model, or a GRU (Gated Recurrent Unit) neural network model, or other types of neural network models, which are not limited in the embodiments of the present application.

[0244] As an optional implementation manner, in order to further improve the accuracy of the second downlink CSI, the UE may obtain a preset number of historical downlink CSIs, and predict the second downlink CSI corresponding to the time when the base station transmits the PDSCH according to the first downlink CSI, each historical downlink CSI, the prediction time interval, and a pre-trained CSI prediction model.

[0245] In implementation, the UE may obtain a preset number of historical downlink CSIs. Among them, the preset number of historical downlink CSIs are the preset number of historical downlink CSIs whose channel estimation time is earlier than the channel estimation time of the first downlink CSI and closest to the channel estimation time of the first downlink CSI. Then, the UE inputs the first downlink CSI, each historical downlink CSI, and the prediction time interval into a pre-trained CSI prediction model. Correspondingly, the CSI prediction model outputs the second downlink CSI corresponding to the time when the base station transmits the PDSCH.

[0246] Step 103: Transmit the second downlink CSI to the base station.

[0247] In implementation, after the UE obtains the second downlink CSI, it may transmit the second downlink CSI to the base station.

[0248] As an optional implementation manner, the UE may compress the second downlink CSI and then transmit it to the base station. The specific processing process is: obtaining the compressed second downlink CSI according to the second downlink CSI, a preset compression ratio, and a pre-trained CSI compression model; transmitting the compressed second downlink CSI and the compression ratio to the base station.

[0249] In implementation, the UE may input the second downlink CSI and a preset compression ratio into a pre-trained CSI compression model. Correspondingly, the CSI compression model outputs the compressed second downlink CSI. The CSI compression model may adopt an RNN (Recurrent Neural Network) neural network model or other types of neural network models, which are not limited in the embodiments of the present application.

[0250] To ensure that the base station can decompress properly, the UE may send the compressed second downlink CSI and the compression ratio to the base station through the PUSCH (Physical Uplink Shared Channel). In this way, the second downlink CSI used by the base station when sending the PDSCH to the UE at the PDSCH sending moment can better match the downlink communication environment where the UE is located at the PDSCH sending moment, thereby solving the problem of poor timeliness of the second downlink CSI and ensuring the accuracy of the second downlink CSI.

[0251] As an alternative implementation, to reduce signaling overhead, the time indication information sent by the UE to the base station may be the time interval level corresponding to the predicted time interval. Correspondingly, the UE may determine the time interval level corresponding to the predicted time interval according to the correspondence between the time interval sent by the base station and the time interval level.

[0252] As an alternative implementation, the UE may also send time indication information to the base station. The time indication information is the predicted time interval, or the time indication information is the time interval level corresponding to the predicted time interval.

[0253] In implementation, directly reporting the predicted time interval will result in relatively large signaling overhead. Therefore, the UE may determine the number index of the time interval level corresponding to the predicted time interval in the correspondence between the time interval range sent by the base station and the time interval level. In this way, by sending the time interval level corresponding to the predicted time interval to the base station, the UE can reduce the signaling overhead. The processing procedure for the base station to determine the correspondence between the time interval range and the time interval level will be introduced in detail later and will not be elaborated here.

[0254] As an alternative implementation, in step 201 above, the UE may obtain the second downlink CSI through the CSI prediction model in two ways: Way 1, input the first downlink CSI and the predicted time interval into a pre-trained CSI prediction model to obtain the second downlink CSI. Way 2, input the first downlink CSI, the historical downlink CSI, and the predicted time interval into a pre-trained CSI prediction model to obtain the second downlink CSI. Correspondingly, the UE obtains the CSI prediction model through the following method.

[0255] Step 1: Obtain a basic dataset. The basic dataset contains the historical downlink CSI of the UE.

[0256] In implementation, each time the UE performs channel estimation based on the downlink reference information sent by the base station and obtains the downlink CSI (i.e., the historical downlink CSI), the UE can record the historical downlink CSI and the channel estimation time to obtain the basic dataset. When the UE needs to train the initial CSI prediction model, the UE can obtain this basic dataset.

[0257] Step 2: Respectively determine the training dataset and the label dataset corresponding to each prediction time interval in the basic dataset.

[0258] In implementation, the UE can respectively determine the training dataset and the label dataset corresponding to each prediction time interval in the basic dataset.

[0259] For the above-mentioned method 1, the UE determines multiple first historical downlink CSI groups corresponding to each prediction time interval. Among them, each first historical downlink CSI group includes a first historical downlink CSI and a second historical downlink CSI. The difference between the channel estimation times of the first historical downlink CSI and the second historical downlink CSI is the prediction time interval. The UE determines the multiple first historical downlink CSIs as the training dataset and determines the multiple second historical downlink CSIs as the label dataset.

[0260] In implementation, for different prediction time intervals, the UE can determine multiple first historical downlink CSI groups corresponding to each prediction time interval in the basic dataset. Among them, each first historical downlink CSI group includes a first historical downlink CSI and a second historical downlink CSI. The difference between the channel estimation time of the first historical downlink CSI and the channel estimation time of the second historical downlink CSI is the prediction time interval. For example, the prediction time interval is Δt, the channel estimation time of the first historical downlink CSI is t, then the channel estimation time of the second historical downlink CSI is t + Δt. Optionally, a prediction time interval can be selected from the time interval range corresponding to a prediction time interval level. Then, the UE can determine the first historical downlink CSI as the training dataset and determine the second historical downlink CSI as the label dataset.

[0261] For the above-mentioned second method, the UE determines multiple second historical downlink CSI groups corresponding to each prediction time interval. Each second historical downlink CSI group includes a third historical downlink CSI, a fourth historical downlink CSI, and a preset number of fifth historical downlink CSIs. The time difference between the channel estimation times of the third historical downlink CSI and the fourth historical downlink CSI is the prediction time interval, and the channel estimation times of the preset number of fifth historical downlink CSIs are earlier than the channel estimation time of the third historical downlink CSI; the UE determines multiple third historical downlink CSIs and each fifth historical downlink CSI corresponding to the third historical downlink CSI as the training data set, and determines multiple fourth historical downlink CSIs as the label data set.

[0262] In implementation, for different prediction time intervals, the UE can determine multiple second historical downlink CSI groups corresponding to each prediction time interval in the basic data set. Each second historical downlink CSI group includes a third historical downlink CSI, a fourth historical downlink CSI, and a preset number of fifth historical downlink CSIs. The time difference between the channel estimation time of the third historical downlink CSI and the channel estimation time of the fourth historical downlink CSI is the prediction time interval, and the channel estimation times of the preset number of fifth historical downlink CSIs are earlier than and close to the channel estimation time of the third historical downlink CSI. For example, if the prediction time interval is Δt and the channel estimation time of the third historical downlink CSI is t, then the channel estimation time of the fourth historical downlink CSI is t + Δt, and the channel estimation times of the fifth historical downlink CSIs are earlier than t. Optionally, a prediction time interval can be selected from the time interval range corresponding to a prediction time interval level. Then, the UE can determine multiple third historical downlink CSIs and each fifth historical downlink CSI corresponding to the third historical downlink CSI as the training data set, and determine multiple fourth historical downlink CSIs as the label data set.

[0263] Step 3: Based on each training data set, each label data set, and each prediction time interval, train the initial CSI prediction model to obtain the trained CSI prediction model.

[0264] In implementation, the UE trains the initial CSI prediction model based on the training data set, the label data set, and each prediction time interval until the model converges to obtain the trained CSI prediction model. In this way, the trained CSI prediction model can output different second downlink CSIs for different prediction time intervals and different first downlink CSIs.

[0265] As an optional implementation method, the UE trains to obtain a CSI compression model through the following method:

[0266] Step 1: Determine the historical downlink CSI of the UE as the third training data set.

[0267] In implementation, each time the UE performs channel estimation based on the downlink reference information sent by the base station and obtains the downlink CSI (i.e., historical downlink CSI), the UE can determine the historical downlink CSI as the third training data set.

[0268] Step 2: Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the third labeled data set.

[0269] In implementation, the UE can determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the third labeled data set. Among them, the measurement matrix is randomly selected from some rows of the orthogonal Fourier transform matrix; the quantization can be uniform quantization or non-uniform quantization, which is not limited in the embodiments of the present application.

[0270] Step 3: Train the initial CSI compression model based on the third training data set and the third labeled data set to obtain the trained CSI compression model.

[0271] In implementation, the UE can train the initial CSI compression model based on the third training data set and the third labeled data set until the model converges to obtain the trained CSI compression model. Optionally, the UE can train the CSI compression model corresponding to different compression ratios for different compression ratios, or design the CSI compression model as an extensible model, that is, connect the corresponding upsampling module or downsampling module according to different compression ratios to adapt to different compression ratios.

[0272] It should be noted that the training processes of the above CSI prediction model and CSI compression model can be trained by the terminal manufacturer and directly deployed on the UE at the factory after training, or can be trained by different UEs themselves respectively, but this will increase the computing burden and energy consumption of the UE.

[0273] The embodiments of the present application also provide a method for predicting CSI, Figure 3 which is a flowchart of a method for predicting CSI provided by the embodiments of the present application. This method for predicting CSI is applied to a base station, as Figure 3 shown, and the specific steps are as follows:

[0274] Step 301: Send a downlink reference signal to the UE.

[0275] In implementation, when the base station needs the UE to perform downlink CSI prediction, the base station can send a PDCCH for triggering CSI reporting and downlink reference information to the UE.

[0276] Step 302: Obtain the downlink CSI predicted by the UE.

[0277] In implementation, the UE may send the predicted downlink CSI to the base station through the PUSCH channel. Correspondingly, the base station may receive the second downlink CSI sent by the UE.

[0278] Step 303: Schedule according to the downlink CSI and send a PDSCH to the UE.

[0279] In implementation, the base station obtains the downlink CSI, schedules according to the predicted downlink CSI, and sends a PDSCH to the UE.

[0280] As an optional implementation manner, the base station receives the compressed downlink CSI and the compression ratio sent by the UE; and obtains the downlink CSI according to the compressed downlink CSI, the compression ratio, and a pre-trained CSI decompression model.

[0281] In implementation, the UE may send the compressed downlink CSI and the compression ratio to the base station through the PUSCH channel. Correspondingly, the base station may receive the compressed downlink CSI and the compression ratio sent by the UE. After receiving the compressed downlink CSI and the compression ratio sent by the UE, the base station may input the compressed downlink CSI and the compression ratio into a pre-trained CSI decompression model. Correspondingly, the CSI decompression model outputs the downlink CSI. Among them, the CSI decompression model may adopt a CNN (Convolutional Neural Network) neural network model, or an MLP (Multi-Layer Perceptron) neural network model, or other types of neural network models, which are not limited in the embodiments of the present application.

[0282] As an optional implementation manner, the base station receives the time indication information sent by the UE. Wherein, the time indication information is the predicted time interval determined by the UE, or the time interval level corresponding to the predicted time interval determined by the UE.

[0283] As an optional implementation manner, in order to reduce signaling overhead, the time indication information sent by the UE is the time interval level corresponding to the predicted time interval. Correspondingly, the base station may determine the time interval corresponding to the time indication information in the correspondence between the time interval and the time interval level, and determine the predicted time interval by randomly selecting a value or taking the median in the time interval corresponding to the time indication information.

[0284] In implementation, after receiving the time indication information, the base station can determine the time interval corresponding to the time indication information in the correspondence between the time interval and the time interval level. Then, the base station determines the predicted time interval within the time interval corresponding to the time indication information. Optionally, the base station can randomly select a time interval within the time interval corresponding to the time indication information as the predicted time interval, or can select the median within the time interval corresponding to the time indication information as the predicted time interval. This application embodiment does not make a limitation.

[0285] As an alternative implementation manner, the processing procedure for the base station to determine the correspondence between the time interval and the time interval level is as follows:

[0286] Step 1: Determine the predicted time interval range according to the pre-stored downlink communication environment parameter range of the UE, the first time parameter range provided by the UE, and the second time parameter range provided by the base station.

[0287] In implementation, the base station can determine the predicted time interval range according to the pre-stored downlink communication environment parameter range of the UE, the first time parameter range provided by the UE, and the second time parameter range provided by the base station. Among them, the processing procedure for the base station to determine the predicted time interval range is similar to the processing procedure for the UE to determine the predicted time interval, and will not be elaborated here.

[0288] Step 2: Divide the predicted time interval range according to the preset level division rule to obtain the correspondence between the time interval and the time interval level.

[0289] In implementation, the base station can divide the predicted time interval range according to the preset level division rule to obtain the correspondence between the time interval and the time interval level. Among them, the time intervals corresponding to each time interval level can be uniform or non-uniform. For example, the base station can divide the predicted time interval range into time interval levels with non-uniform time intervals based on the moving speed of the UE. For example, the greater the moving speed of the UE, the smaller the time interval corresponding to the time interval level.

[0290] Step 3: Send the correspondence between the time interval and the time interval level to the UE.

[0291] In implementation, after obtaining the correspondence between the time interval and the time interval level, the base station can send the correspondence between the time interval and the time interval level to the UE by means of broadcasting, so that after the UE determines the predicted time interval, it can determine the time interval level corresponding to the predicted time interval in the correspondence between the time interval and the time interval level.

[0292] As an alternative implementation manner, the base station trains the CSI decompression model through the following method:

[0293] Step 1: Obtain the historical downlink CSI of the UE.

[0294] In implementation, each time the UE performs channel estimation based on the downlink reference information sent by the base station and obtains the downlink CSI (i.e., the historical downlink CSI), the base station can record the historical downlink CSI.

[0295] Step 2: Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the training data set.

[0296] In implementation, the base station can determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the training data set. Among them, the measurement matrix is randomly selected from some rows of the orthogonal Fourier transform matrix; the quantization can be uniform quantization or non-uniform quantization, which is not limited in the embodiments of the present application.

[0297] Step 3: Determine the historical downlink CSI after decompression using the compressive sensing recovery algorithm as the label data set.

[0298] In implementation, the base station can determine the historical downlink CSI after decompression using the compressive sensing recovery algorithm for the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the label data set. Among them, the compressive sensing recovery algorithm can adopt OMP (Orthogonal Matching Pursuit), or other types of compressive sensing recovery algorithms, which is not limited in the embodiments of the present application.

[0299] Step 4: Train the initial CSI decompression model based on the training data set and the label data set to obtain the trained CSI decompression model.

[0300] In implementation, the base station can train the initial CSI decompression model based on the training data set and the label data set until the model converges to obtain the trained CSI decompression model. Optionally, the base station can train the CSI decompression models corresponding to different compression ratios for different compression ratios, or design the CSI decompression model as an extensible model, that is, connect the corresponding upsampling module or downsampling module according to different compression ratios to adapt to different compression ratios.

[0301] It should be noted that the compression and decompression of downlink CSI can also be implemented using an Auto-Encoder bilateral model (Auto-Encoder is an unsupervised learning model). To perform the compression and decompression of downlink CSI using an Auto-Encoder bilateral model, an encoder needs to be deployed on the UE side and a decoder needs to be deployed on the base station side. Correspondingly, the training and deployment methods of the encoder and decoder can include the following several methods. Method 1: The UE trains the encoder and decoder and sends the trained decoder to the base station. The advantage of Method 1 is that the machine learning models trained by each UE can better adapt to their respective UEs. The disadvantage is that the computational burden on the UE is relatively heavy, and the base station needs to maintain a relatively large number of decoders. Method 2: The terminal manufacturer trains the encoder and decoder, deploys the trained encoder to the UE, and sends the trained decoder to the base station. When the UE uses the encoder for compression, it also needs to send an identifier of the decoder corresponding to the encoder to the base station for the base station to perform decompression. The advantage of Method 2 is that it reduces the training burden on the UE. The disadvantage is that the terminal manufacturer cannot train different machine learning models according to the different environments of the UE. Method 3: The base station trains the encoder and decoder and broadcasts the trained encoder to each UE. The advantage of Method 3 is that it reduces the training burden on the UE. The disadvantage is that the base station cannot train different machine learning models according to the different environments of the UE. Method 4: The base station and the UE jointly train the encoder and decoder, that is, in each training loop of the training, a forward propagation and backward propagation training process needs to be carried out between the UE side and the base station side, and it is ensured that the data sets used by both are consistent. The disadvantage of Method 4 is that the overhead is very large, and due to quantization during the transmission process, the training performance may be affected. Regardless of which of the above training and deployment methods is adopted, since the training data of the Auto-Encoder is the same as the label data, therefore, to collect data, it only requires the UE, the terminal manufacturer, or the base station to obtain the historical downlink CSI within a certain period of time. Common training data is the downlink CSI channel matrix or the eigenvector of the precoding matrix, and the commonly used machine learning model is based on the transformer learning model.

[0302] The present application provides a method for predicting CSI. The UE receives the downlink reference signal sent by the base station, and based on the downlink reference signal, estimates the first downlink CSI and the downlink communication environment parameters. Then, the UE predicts the second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the downlink communication environment parameters, and the pre-trained CSI prediction model, and sends the second downlink CSI to the base station. In this way, the second downlink CSI used by the base station to send the PDSCH to the UE at the PDSCH transmission time can better match the downlink communication environment where the UE is located at the PDSCH transmission time, thereby ensuring the timeliness of the downlink CSI.

[0303] It can be understood that the same / similar parts among the various embodiments of the above method in this specification can be referred to each other. Each embodiment focuses on the differences from other embodiments. For the related parts, refer to the descriptions of other method embodiments.

[0304] The embodiment of the present application also provides a UE, as Figure 4 shown, including a memory 410, a transceiver 420, and a processor 430;

[0305] Among them, the memory 410 is used to store computer programs; the transceiver 420 is used to transmit and receive data under the control of the processor 430, and the processor 430 is used to read the computer programs in the memory 410 and perform the following operations:

[0306] Receive the downlink reference signal sent by the base station, and based on the downlink reference signal, estimate the first downlink CSI and the downlink communication environment parameters;

[0307] Predict the second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the downlink communication environment parameters, and the pre-trained CSI prediction model;

[0308] Send the second downlink CSI to the base station.

[0309] As an optional implementation manner, predicting the second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the downlink communication environment parameters, and the pre-trained CSI prediction model includes:

[0310] Determine the prediction time interval according to the downlink communication environment parameters, the first time parameter provided by the pre-stored UE, and the second time parameter provided by the base station;

[0311] Predict the second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the prediction time interval, and the pre-trained CSI prediction model.

[0312] As an alternative implementation, determining a prediction time interval according to downlink communication environment parameters, a first time parameter provided by a pre-stored UE, and a second time parameter provided by a base station includes:

[0313] Determining the transmission time of a downlink reference signal according to the downlink communication environment parameters;

[0314] Determining the transmission time of a second downlink CSI according to the downlink communication environment parameters and the first time parameter;

[0315] Determining the prediction time interval according to the transmission time of the downlink reference signal, the transmission time of the second downlink CSI, the first time parameter, and the second time parameter.

[0316] As an alternative implementation, the downlink communication environment parameters include at least one or more of the moving speed v of the UE, the moving azimuth angle θ of the UE relative to the base station, and the relative distance between the UE and the base station;

[0317] Determining the transmission time T1 of the downlink reference signal according to the downlink communication environment parameters:

[0318] T1 = S0 / c;

[0319] wherein, T1 represents the transmission time of the downlink reference signal, S0 represents the relative distance between the UE and the base station when the base station sends the downlink reference signal, and c represents the speed of light; or,

[0320] Determining the transmission time T2 of the second downlink CSI according to the downlink communication environment parameters and the first time parameter:

[0321]

[0322] wherein, T2 represents the transmission time of the second downlink CSI, v represents the moving speed of the UE, θ represents the moving azimuth angle of the UE relative to the base station, and t1 represents the first time parameter.

[0323] As an alternative implementation, sending the second downlink CSI to the base station includes:

[0324] Obtaining the compressed second downlink CSI according to the second downlink CSI, a preset compression ratio, and a pre-trained CSI compression model;

[0325] Sending the compressed second downlink CSI and the compression ratio to the base station.

[0326] As an alternative implementation, the first time parameter is the time interval from the last symbol of the PDCCH that triggers the CSI report to the reporting of the second downlink CSI; the second time parameter includes the time for scheduling and preparing the PDSCH; or,

[0327] The first time parameter is the time interval between the last symbol of the PDCCH that triggers CSI reporting and the reporting of the compressed second downlink CSI; the second time parameter includes the decompression time of the compressed second downlink CSI and the time for scheduling and preparing the PDSCH.

[0328] As an optional implementation manner, the processor 430 is further configured to read the computer program in the memory 410 and perform the following operations:

[0329] Determine the time interval level corresponding to the predicted time interval according to the correspondence between the time interval and the time interval level sent by the base station.

[0330] As an optional implementation manner, the processor 430 is further configured to read the computer program in the memory 410 and perform the following operations:

[0331] Send time indication information to the base station; the time indication information is the predicted time interval, or the time indication information is the time interval level corresponding to the predicted time interval.

[0332] As an optional implementation manner, predicting the second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, the predicted time interval, and the pre-trained CSI prediction model includes:

[0333] Obtain a preset number of historical downlink CSIs;

[0334] Predict the second downlink CSI corresponding to the PDSCH transmission time of the base station according to the first downlink CSI, each historical downlink CSI, the predicted time interval, and the pre-trained CSI prediction model.

[0335] As an optional implementation manner, obtaining the CSI prediction model through the following method includes:

[0336] Obtain a basic data set, and the basic data set includes the historical downlink CSI of the UE;

[0337] Respectively determine the training data set and the label data set corresponding to each predicted time interval in the basic data set;

[0338] Train the initial CSI prediction model based on each training data set, each label data set, and each predicted time interval to obtain the trained CSI prediction model.

[0339] As an optional implementation manner, determining the training data set and the label data set corresponding to each predicted time interval includes:

[0340] Determine multiple first historical downlink CSI groups corresponding to each prediction time interval, where each first historical downlink CSI group includes a first historical downlink CSI and a second historical downlink CSI, and the difference in the channel estimation time between the first historical downlink CSI and the second historical downlink CSI is the prediction time interval;

[0341] Determine the multiple first historical downlink CSIs as the training data set, and determine the multiple second historical downlink CSIs as the label data set.

[0342] As an alternative implementation manner, determining the training data set and the label data set corresponding to each prediction time interval includes:

[0343] Determine multiple second historical downlink CSI groups corresponding to each prediction time interval, where each second historical downlink CSI group includes a third historical downlink CSI, a fourth historical downlink CSI, and a preset number of fifth historical downlink CSIs. The difference in the channel estimation time between the third historical downlink CSI and the fourth historical downlink CSI is the prediction time interval, and the channel estimation times of the preset number of fifth historical downlink CSIs are earlier than the channel estimation time of the third historical downlink CSI;

[0344] Determine the multiple third historical downlink CSIs and each fifth historical downlink CSI corresponding to the third historical downlink CSI as the training data set, and determine the multiple fourth historical downlink CSIs as the label data set.

[0345] As an alternative implementation manner, training to obtain a CSI compression model through the following method includes:

[0346] Determine the historical downlink CSI of the UE as the third training data set;

[0347] Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the third label data set;

[0348] Based on the third training data set and the third label data set, train the initial CSI compression model to obtain the trained CSI compression model.

[0349] The embodiment of the present application also provides a base station, as Figure 5 shown, including a memory 510, a transceiver 520, and a processor 530;

[0350] Among them, the memory 510 is used to store computer programs; the transceiver 520 is used to send and receive data under the control of the processor 530, and the processor 530 is used to read the computer programs in the memory 510 and perform the following operations:

[0351] Send a downlink reference signal to the UE;

[0352] Obtain the downlink CSI predicted by the UE;

[0353] Perform scheduling based on the downlink CSI and send the PDSCH to the UE.

[0354] As an alternative implementation, obtaining the downlink CSI predicted by the UE includes:

[0355] Receive the compressed downlink CSI and the compression ratio sent by the UE;

[0356] Obtain the downlink CSI based on the compressed downlink CSI, the compression ratio, and a pre-trained CSI decompression model.

[0357] As an alternative implementation, the processor 530 is further configured to read the computer program in the memory 510 and perform the following operations:

[0358] Receive the time indication information sent by the UE; the time indication information is the predicted time interval determined by the UE, or the time interval level corresponding to the predicted time interval determined by the UE.

[0359] As an alternative implementation, in the case where the time indication information is the time interval level corresponding to the predicted time interval determined by the UE, the processor 530 is further configured to read the computer program in the memory 510 and perform the following operations:

[0360] In the correspondence relationship between the time interval and the time interval level, determine the time interval corresponding to the time indication information, and within the time interval corresponding to the time indication information, determine the predicted time interval by randomly selecting a value or taking the median.

[0361] As an alternative implementation, the processor 530 is further configured to read the computer program in the memory 510 and perform the following operations:

[0362] Determine the predicted time interval range according to the pre-stored downlink communication environment parameter range of the UE, the first time parameter range provided by the UE, and the second time parameter range provided by the base station;

[0363] Divide the predicted time interval range according to the preset level division rule to obtain the correspondence relationship between the time interval and the time interval level;

[0364] Send the correspondence relationship between the time interval and the time interval level to the UE.

[0365] As an alternative implementation, the CSI decompression model is trained in the following manner, including:

[0366] Obtain the historical downlink CSI of the UE;

[0367] Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the training data set;

[0368] Determine the historical downlink CSI after decompression using the compressive sensing recovery algorithm as the label data set;

[0369] Based on the training data set and the label data set, train the initial CSI decompression model to obtain the trained CSI decompression model.

[0370] An embodiment of the present application further provides a prediction device for CSI, as Figure 6 shown. This device is applied to the UE, and this device includes:

[0371] An estimation unit 610, configured to receive the downlink reference signal sent by the base station, and estimate the first downlink CSI and the downlink communication environment parameters based on the downlink reference signal;

[0372] A prediction unit 620, configured to predict the second downlink CSI corresponding to the transmission time of the base station's transmitted PDSCH according to the first downlink CSI, the downlink communication environment parameters, and the pre-trained CSI prediction model;

[0373] A first transmission unit 630, configured to send the second downlink CSI to the base station.

[0374] As an optional implementation manner, the prediction unit 620 is specifically configured to:

[0375] Determine the prediction time interval according to the downlink communication environment parameters, the first time parameter provided by the pre-stored UE, and the second time parameter provided by the base station;

[0376] Predict the second downlink CSI corresponding to the transmission time of the base station's transmitted PDSCH according to the first downlink CSI, the prediction time interval, and the pre-trained CSI prediction model.

[0377] As an optional implementation manner, the prediction unit 620 is specifically configured to:

[0378] Determine the transmission time of the downlink reference signal according to the downlink communication environment parameters;

[0379] Determine the transmission time of the second downlink CSI according to the downlink communication environment parameters and the first time parameter;

[0380] Determine the prediction time interval according to the transmission time of the downlink reference signal, the transmission time of the second downlink CSI, the first time parameter, and the second time parameter.

[0381] As an alternative implementation, the downlink communication environment parameters include at least one or more of the moving speed v of the UE, the moving azimuth angle θ of the UE relative to the base station, and the relative distance between the UE and the base station;

[0382] Determine the transmission time T1 of the downlink reference signal according to the downlink communication environment parameters:

[0383] T1 = S0 / c;

[0384] where T1 represents the transmission time of the downlink reference signal, S0 represents the relative distance between the UE and the base station when the base station sends the downlink reference signal, and c represents the speed of light; or,

[0385] Determine the transmission time T2 of the second downlink CSI according to the downlink communication environment parameters and the first time parameter:

[0386]

[0387] where T2 represents the transmission time of the second downlink CSI, v represents the moving speed of the UE, θ represents the moving azimuth angle of the UE relative to the base station, and t1 represents the first time parameter.

[0388] As an alternative implementation, the first sending unit 630 is specifically configured to:

[0389] Obtain the compressed second downlink CSI according to the second downlink CSI, the preset compression ratio, and the pre-trained CSI compression model;

[0390] Send the compressed second downlink CSI and the compression ratio to the base station.

[0391] As an alternative implementation, the first time parameter is the time interval from the last symbol of the PDCCH that triggers the CSI report to the reporting of the second downlink CSI; the second time parameter includes the time for scheduling the PDSCH and preparing the PDSCH; or,

[0392] The first time parameter is the time interval from the last symbol of the PDCCH that triggers the CSI report to the reporting of the compressed second downlink CSI; the second time parameter includes the decompression time of the compressed second downlink CSI and the time for scheduling the PDSCH and preparing the PDSCH.

[0393] As an alternative implementation, the device further includes:

[0394] A determination unit, configured to determine the time interval level corresponding to the predicted time interval according to the correspondence between the time interval sent by the base station and the time interval level.

[0395] As an alternative implementation, the device further includes:

[0396] A second sending unit, configured to send time indication information to a base station; the time indication information is a predicted time interval, or the time indication information is a time interval level corresponding to the predicted time interval.

[0397] As an optional implementation manner, the prediction unit 620 is specifically configured to:

[0398] Obtain a preset number of historical downlink CSIs;

[0399] According to the first downlink CSI, each historical downlink CSI, the predicted time interval, and a pre-trained CSI prediction model, predict a second downlink CSI corresponding to the time when the base station sends the PDSCH.

[0400] As an optional implementation manner, the first model training unit is specifically configured to:

[0401] Obtain a basic data set, where the basic data set includes the historical downlink CSIs of the UE;

[0402] Respectively determine a training data set and a label data set corresponding to each predicted time interval in the basic data set;

[0403] Based on each training data set, each label data set, and each predicted time interval, train an initial CSI prediction model to obtain a trained CSI prediction model.

[0404] As an optional implementation manner, the first model training unit is specifically configured to:

[0405] Determine multiple first historical downlink CSI groups corresponding to each predicted time interval, where each first historical downlink CSI group includes a first historical downlink CSI and a second historical downlink CSI, and the difference between the channel estimation times of the first historical downlink CSI and the second historical downlink CSI is the predicted time interval;

[0406] Determine multiple first historical downlink CSIs as the training data set, and determine multiple second historical downlink CSIs as the label data set.

[0407] As an optional implementation manner, the first model training unit is specifically configured to:

[0408] Determine multiple second historical downlink CSI groups corresponding to each predicted time interval, where each second historical downlink CSI group includes a third historical downlink CSI, a fourth historical downlink CSI, and a preset number of fifth historical downlink CSIs, the difference between the channel estimation times of the third historical downlink CSI and the fourth historical downlink CSI is the predicted time interval, and the channel estimation times of the preset number of fifth historical downlink CSIs are earlier than the channel estimation time of the third historical downlink CSI;

[0409] Determine multiple third historical downlink CSIs and each fifth historical downlink CSI corresponding to the third historical downlink CSI respectively as the training data set, and determine multiple fourth historical downlink CSIs as the label data set.

[0410] As an alternative implementation, the second model training unit is specifically configured to:

[0411] Determine the historical downlink CSI of the UE as the third training data set;

[0412] Determine the historical downlink CSI after compression and quantization using the measurement matrix of compressive sensing as the third label data set;

[0413] Train the initial CSI compression model based on the third training data set and the third label data set to obtain the trained CSI compression model.

[0414] The embodiment of the present application further provides a CSI prediction device, as Figure 7 shown. This device is applied to a base station, and this device includes:

[0415] A first sending unit 710, configured to send a downlink reference signal to the UE;

[0416] An obtaining unit 720, configured to obtain the predicted downlink CSI of the UE;

[0417] A second sending unit 730, configured to perform scheduling according to the downlink CSI and send a PDSCH to the UE.

[0418] As an alternative implementation, the obtaining unit 720 is specifically configured to:

[0419] Receive the compressed downlink CSI and the compression ratio sent by the UE;

[0420] Obtain the downlink CSI according to the compressed downlink CSI, the compression ratio, and the pre-trained CSI decompression model.

[0421] As an alternative implementation, this device further includes:

[0422] A receiving unit, configured to receive the time indication information sent by the UE; the time indication information is the predicted time interval determined by the UE, or the time interval level corresponding to the predicted time interval determined by the UE.

[0423] As an alternative implementation, in the case where the time indication information is the time interval level corresponding to the predicted time interval determined by the UE, this device further includes:

[0424] A first determination unit, configured to determine a time interval corresponding to time indication information in a correspondence between a time interval and a time interval level, and determine a predicted time interval by randomly selecting a value within the interval or taking the median within the time interval corresponding to the time indication information.

[0425] As an optional implementation manner, the apparatus further includes:

[0426] A second determination unit, configured to determine a predicted time interval range according to a range of downlink communication environment parameters of a UE stored in advance, a first time parameter range provided by the UE, and a second time parameter range provided by a base station;

[0427] A division unit, configured to divide the predicted time interval range according to a preset level division rule to obtain a correspondence between a time interval and a time interval level;

[0428] A third sending unit, configured to send the correspondence between the time interval and the time interval level to the UE.

[0429] As an optional implementation manner, the model training unit is specifically configured to:

[0430] Obtain the historical downlink CSI of the UE;

[0431] Determine the historical downlink CSI after being compressed and quantized by using a measurement matrix of compressive sensing as a training data set;

[0432] Determine the historical downlink CSI after being decompressed by using a compressive sensing recovery algorithm as a label data set;

[0433] Train an initial CSI decompression model based on the training data set and the label data set to obtain a trained CSI decompression model.

[0434] It should be noted that the division of units in the embodiments of the present application is illustrative, and is only a logical function division. In actual implementation, there may be other division methods. In addition, in each embodiment of the present application, each functional unit may be integrated into a processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit. The above integrated units may be implemented in the form of hardware or in the form of software functional units.

[0435] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a processor-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor to execute all or part of the steps of the methods described in various embodiments of this application.

[0436] It should be noted here that the above-mentioned device provided in the embodiments of the present invention can implement all the method steps implemented in the above-mentioned method embodiments and can achieve the same technical effects. The same parts and beneficial effects as those in the method embodiments will not be specifically described in this embodiment.

[0437] The embodiments of this application provide a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the steps of the above-mentioned CSI prediction method.

[0438] The processor-readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to magnetic memories (such as floppy disks, hard disks, magnetic tapes, magneto-optical disks (MO), etc.), optical memories (such as CDs, DVDs, BDs, HVDs, etc.), and semiconductor memories (such as ROM, EPROM, EEPROM, non-volatile memories (NANDFLASH), solid-state drives (SSD)).

[0439] Those skilled in the art should understand that the embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories and optical memories, etc.) that contain computer-usable program code.

[0440] This application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as the combination of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer-executable instructions. These computer-executable instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in the Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0441] These processor-executable instructions can also be stored in a processor-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the processor-readable memory generate a manufactured article including instruction means that implement the functions specified in the Figure 1 one or more flows and / or blocks Figure 1 one or more blocks.

[0442] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalent technologies, this application is also intended to include these modifications and variations.

Claims

1. A method for predicting channel state information (CSI), characterized in that The method is applied to a user equipment (UE), and the method includes: Receiving a downlink reference signal sent by a base station, and estimating a first downlink CSI and downlink communication environment parameters based on the downlink reference signal; Predicting a second downlink CSI corresponding to a physical downlink shared channel (PDSCH) transmission time of the base station according to the first downlink CSI, the downlink communication environment parameters, and a pre-trained CSI prediction model; Sending the second downlink CSI to the base station.

2. The method according to claim 1, characterized in that The predicting a second downlink CSI corresponding to a PDSCH transmission time of the base station according to the first downlink CSI, the downlink communication environment parameters, and a pre-trained CSI prediction model includes: Determining a prediction time interval according to the downlink communication environment parameters, a first time parameter provided by the UE stored in advance, and a second time parameter provided by the base station; Predicting a second downlink CSI corresponding to a PDSCH transmission time of the base station according to the first downlink CSI, the prediction time interval, and a pre-trained CSI prediction model.

3. The method according to claim 1, characterized in that The determining a prediction time interval according to the downlink communication environment parameters, a first time parameter provided by the UE stored in advance, and a second time parameter provided by the base station includes: Determining a transmission time of the downlink reference signal according to the downlink communication environment parameters; Determining a transmission time of the second downlink CSI according to the downlink communication environment parameters and the first time parameter; Determining the prediction time interval according to the transmission time of the downlink reference signal, the transmission time of the second downlink CSI, the first time parameter, and the second time parameter.

4. The method according to claim 3, wherein The downlink communication environment parameters at least include at least one or more of a moving speed v of the UE, a moving azimuth angle θ of the UE relative to the base station, and a relative distance S0 between the UE and the base station; The determining a transmission time T1 of the downlink reference signal according to the downlink communication environment parameters: T1 = S0 / c; where T1 represents the transmission time of the downlink reference signal, S0 represents the relative distance between the UE and the base station when the base station sends the downlink reference signal, and c represents the speed of light; or, The determining a transmission time T2 of the second downlink CSI according to the downlink communication environment parameters and the first time parameter: where T2 represents the transmission time of the second downlink CSI, v represents the moving speed of the UE, θ represents the moving azimuth angle of the UE relative to the base station, and t1 represents the first time parameter.

5. The method according to claim 1, wherein The sending the second downlink CSI to the base station includes: Obtaining a compressed second downlink CSI according to the second downlink CSI, a preset compression ratio, and a pre-trained CSI compression model; Sending the compressed second downlink CSI and the compression ratio to the base station.

6. The method according to claim 1 or 5, characterized in that The first time parameter is the time interval from the last symbol of the physical downlink shared channel PDCCH that triggers CSI reporting to the reporting of the second downlink CSI; the second time parameter includes the time for scheduling the PDSCH and preparing the PDSCH. Or, The first time parameter is the time interval from the last symbol of the PDCCH that triggers CSI reporting to the reporting of the compressed second downlink CSI; the second time parameter includes the decompression time of the compressed second downlink CSI and the time for scheduling the PDSCH and preparing the PDSCH.

7. The method according to claim 2, wherein The method further includes: Determining the time interval level corresponding to the predicted time interval according to the correspondence between the time interval and the time interval level sent by the base station.

8. The method according to claim 2 or 7, characterized in that The method further includes: Sending time indication information to the base station; the time indication information is the predicted time interval, or the time indication information is the time interval level corresponding to the predicted time interval.

9. The method according to claim 2, characterized in that: The predicting the second downlink CSI corresponding to the PDSCH transmission moment of the base station according to the first downlink CSI, the predicted time interval, and a pre-trained CSI prediction model includes: Obtaining a preset number of historical downlink CSIs; Predicting the second downlink CSI corresponding to the PDSCH transmission moment of the base station according to the first downlink CSI, each of the historical downlink CSIs, the predicted time interval, and a pre-trained CSI prediction model.

10. The method according to claim 1, characterized in that Obtaining the CSI prediction model by the following method, including: Obtaining a basic data set, where the basic data set includes the historical downlink CSIs of the UE; Respectively determining a training data set and a label data set corresponding to each predicted time interval in the basic data set; Training an initial CSI prediction model based on each of the training data sets, each of the label data sets, and each of the predicted time intervals to obtain a trained CSI prediction model.

11. The method according to claim 10, characterized in that, The determining the training data set and the label data set corresponding to each predicted time interval includes: Determining a plurality of first historical downlink CSI groups corresponding to each predicted time interval, where each of the first historical downlink CSI groups includes a first historical downlink CSI and a second historical downlink CSI, and the difference in the channel estimation moments between the first historical downlink CSI and the second historical downlink CSI is the predicted time interval; Determining the plurality of first historical downlink CSIs as the training data set, and determining the plurality of second historical downlink CSIs as the label data set.

12. The method according to claim 10, characterized in that The determining the training data set and the label data set corresponding to each predicted time interval includes: Determining a plurality of second historical downlink CSI groups corresponding to each predicted time interval, where each of the second historical downlink CSI groups includes a third historical downlink CSI, a fourth historical downlink CSI, and a preset number of fifth historical downlink CSIs, the difference in the channel estimation moments between the third historical downlink CSI and the fourth historical downlink CSI is the predicted time interval, and the channel estimation moments of the preset number of fifth historical downlink CSIs are earlier than the channel estimation moment of the third historical downlink CSI; Determine multiple of the third historical downlink CSIs and each of the fifth historical downlink CSIs corresponding to the respective third historical downlink CSIs as a training data set, and determine multiple of the fourth historical downlink CSIs as a label data set.

13. The method according to claim 5, characterized in that Train to obtain the CSI compression model through the following means, including: Determine the historical downlink CSI of the UE as a third training data set; Determine the historical downlink CSI after compression and quantization using a measurement matrix of compressive sensing as a third label data set; Based on the third training data set and the third label data set, train an initial CSI compression model to obtain a trained CSI compression model.

14. A method for predicting channel state information (CSI), characterized in that The method is applied to a base station, and the method includes: Send a downlink reference signal to a user equipment UE; Obtain the downlink CSI predicted by the UE; Perform scheduling according to the downlink CSI, and send a physical downlink shared channel PDSCH to the UE.

15. The method according to claim 14, characterized in that, The obtaining the downlink CSI predicted by the UE includes: Receive the compressed downlink CSI and the compression ratio sent by the UE; Obtain the downlink CSI according to the compressed downlink CSI, the compression ratio, and a pre-trained CSI decompression model.

16. The method according to claim 14, characterized in that The method further includes: Receive the time indication information sent by the UE; the time indication information is a predicted time interval determined by the UE, or the time indication information is a time interval level corresponding to the predicted time interval determined by the UE.

17. The method according to claim 16, characterized in that In the case where the time indication information is a time interval level corresponding to the predicted time interval determined by the UE, the method further includes: In the correspondence between time intervals and time interval levels, determine the time interval corresponding to the time indication information, and in the time interval corresponding to the time indication information, determine the predicted time interval by randomly selecting a value or taking the median.

18. The method according to claim 14, wherein The method further includes: Determine a predicted time interval range according to a pre-stored range of downlink communication environment parameters of the UE, a first time parameter range provided by the UE, and a second time parameter range provided by the base station; Divide the predicted time interval range according to a preset level division rule to obtain a correspondence between time intervals and time interval levels; Send the correspondence between the time intervals and the time interval levels to the UE.

19. A user equipment UE, characterized in that, Includes a memory, a transceiver, and a processor; Wherein, the memory is used to store a computer program; the transceiver is used to transmit and receive data under the control of the processor, and the processor is used to read the computer program in the memory and perform the following operations: Receive a downlink reference signal sent by a base station, and based on the downlink reference signal, estimate a first downlink CSI and downlink communication environment parameters; According to the first downlink CSI, the downlink communication environment parameters, and a pre-trained CSI prediction model, predict a second downlink CSI corresponding to the time when the base station sends a physical downlink shared channel PDSCH; Send the second downlink CSI to the base station.

20. The UE according to claim 19, wherein: The predicting, based on the first downlink CSI, the downlink communication environment parameter, and a pre-trained CSI prediction model, to obtain a second downlink CSI corresponding to a PDSCH transmission time of the base station includes: Determining a predicted time interval according to the downlink communication environment parameter, a pre-stored first time parameter provided by the UE, and a second time parameter provided by the base station; According to the first downlink CSI, the prediction time interval and the pre-trained CSI prediction model, a second downlink CSI corresponding to the PDSCH sending moment of the base station is predicted.

21. The UE according to claim 19, wherein: The determining the predicted time interval according to the downlink communication environment parameter, the pre-stored first time parameter provided by the UE, and the second time parameter provided by the base station includes: determining a transmission time of the downlink reference signal according to the downlink communication environment parameter; determining a transmission time of the second downlink CSI according to the downlink communication environment parameter and the first time parameter; The prediction time interval is determined according to the transmission time of the downlink reference signal, the transmission time of the second downlink CSI, the first time parameter, and the second time parameter.

22. The UE according to claim 21, wherein The downlink communication environment parameters include at least one or more of a moving speed v of the UE, a moving azimuth angle θ of the UE relative to the base station, and a relative distance between the UE and the base station; The step of determining a transmission time T1 of the downlink reference signal according to the downlink communication environment parameter: T1=S0 / c; Wherein, T1 represents the transmission time of the downlink reference signal, S0 represents the relative distance between the UE and the base station when the base station sends the downlink reference signal, and c represents the speed of light; or, determining, according to the downlink communication environment parameter and the first time parameter, a transmission time T2 of the second downlink CSI: Wherein, T2 represents the transmission time of the second downlink CSI, v represents the moving speed of the UE, θ represents the moving azimuth angle of the UE relative to the base station, and t1 represents the first time parameter.

23. The UE according to claim 19, characterized in that, The sending the second downlink CSI to the base station includes: Obtaining compressed second downlink CSI according to the second downlink CSI, a preset compression ratio, and a pre-trained CSI compression model; Sending the compressed second downlink CSI and the compression ratio to the base station.

24. The UE according to claim 19 or 23, characterized in that, The first time parameter is the time interval from receiving the last symbol of the physical downlink shared channel PDCCH that triggers CSI reporting to reporting the second downlink CSI; the second time parameter includes the time for scheduling PDSCH and preparing PDSCH; or, The first time parameter is the time interval from receiving the last symbol of the PDCCH that triggers CSI reporting to reporting the compressed second downlink CSI; the second time parameter includes the decompression time of the compressed second downlink CSI and the time for scheduling PDSCH and preparing PDSCH.

25. The UE according to claim 20, wherein The processor is further configured to read the computer program in the memory and perform the following operations: Determine the time interval level corresponding to the predicted time interval according to the correspondence between the time intervals and the time interval levels sent by the base station.

26. The UE according to claim 20 or 25, characterized in that, The processor is further configured to read the computer program in the memory and perform the following operations: Send time indication information to the base station; the time indication information is the predicted time interval, or the time indication information is the time interval level corresponding to the predicted time interval.

27. The UE according to claim 20, wherein, The predicting, according to the first downlink CSI, the predicted time interval, and a pre-trained CSI prediction model, a second downlink CSI corresponding to the time when the base station transmits the PDSCH includes: Obtain a preset number of historical downlink CSIs; Predict, according to the first downlink CSI, each of the historical downlink CSIs, the predicted time interval, and a pre-trained CSI prediction model, a second downlink CSI corresponding to the time when the base station transmits the PDSCH.

28. A base station, characterized in that, Comprising a memory, a transceiver, and a processor; Wherein, the memory is used for storing a computer program; the transceiver is used for transceiving data under the control of the processor, and the processor is used for reading the computer program in the memory and performing the following operations: Send a downlink reference signal to the user equipment UE; Obtain the downlink CSI predicted by the UE; Schedule according to the downlink CSI, and send a physical downlink shared channel PDSCH to the UE.

29. A prediction device for channel state information (CSI), characterized in that, The apparatus is applied to a user equipment UE, and the apparatus includes: An estimation unit, configured to receive a downlink reference signal sent by a base station, and estimate a first downlink CSI and downlink communication environment parameters based on the downlink reference signal; A prediction unit, configured to predict, according to the first downlink CSI, the downlink communication environment parameters, and a pre-trained CSI prediction model, a second downlink CSI corresponding to the time when the base station transmits a physical downlink shared channel PDSCH; A first sending unit, configured to send the second downlink CSI to the base station.

30. A device for predicting channel state information (CSI), characterized in that: The apparatus is applied to a base station, and the apparatus includes: A first sending unit, configured to send a downlink reference signal to a user equipment UE; An obtaining unit, configured to obtain the downlink CSI predicted by the UE; A second sending unit, configured to schedule according to the downlink CSI, and send a physical downlink shared channel PDSCH to the UE.

31. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, the steps of the method according to any one of claims 1 to 13, or 14 to 19 are implemented.

Images