Precoding matrix determination method, apparatus, device, and storage medium

Abstract

The application discloses a precoding matrix determination method, device and equipment and a storage medium. The method comprises the following steps: sending a first instruction to a terminal; the first instruction is used for instructing the terminal to report at least a first matrix; the first matrix is used for determining a precoding matrix; receiving a first uplink reference signal sent by the terminal; obtaining a prediction result by using the first uplink reference signal and a pre-trained neural network model; the prediction result comprises a first downlink channel matrix between a network device and the terminal; and determining the precoding matrix based on the first downlink channel matrix.

Description

Technical Field

[0001] This invention relates to the field of wireless technology, and in particular to a method, apparatus, device, and storage medium for determining a precoding matrix. Background Technology

[0002] Currently, the process of network devices determining the precoding matrix can include: the network device sending a downlink reference signal to the terminal; the terminal performing channel estimation based on the downlink reference signal sent by the network device to obtain downlink channel state information, and feeding it back to the network device through the uplink channel; and the network device selecting a precoding matrix based on the received downlink channel state information. However, while ensuring high feedback accuracy, this feedback method may introduce significant feedback overhead to the uplink channel. Summary of the Invention

[0003] In view of this, embodiments of the present invention aim to provide a method, apparatus, device, and storage medium for determining a precoding matrix.

[0004] The technical solution of this invention is implemented as follows:

[0005] At least one embodiment of the present invention provides a precoding matrix determination method, applied to a network device, the method comprising:

[0006] Send a first instruction to the terminal; the first instruction is used to instruct the terminal not to report a first matrix at least; the first matrix is ​​used to determine a precoding matrix;

[0007] Receive the first uplink reference signal sent by the terminal;

[0008] Using the first uplink reference signal and a pre-trained neural network model, a prediction result is obtained; the prediction result includes the first downlink channel matrix between the network device and the terminal.

[0009] Based on the first downlink channel matrix, the precoding matrix is ​​determined.

[0010] Furthermore, according to at least one embodiment of the present invention, obtaining the prediction result using the first uplink reference signal and a pre-trained neural network model includes:

[0011] Using the first uplink reference signal, the uplink channel is estimated to obtain the first uplink channel matrix;

[0012] The first uplink channel matrix is ​​input into a pre-trained neural network model to obtain the prediction result.

[0013] Furthermore, according to at least one embodiment of the present invention, the first instruction is used to instruct the terminal not to report the first matrix; the step of determining the precoding matrix based on the first downlink channel matrix includes:

[0014] The first downlink channel matrix included in the prediction result is decomposed to obtain the first matrix;

[0015] Use the first matrix as the downlink precoding matrix.

[0016] Furthermore, according to at least one embodiment of the present invention, the first instruction is used to instruct the terminal not to report the first matrix but to report the second matrix; the second matrix is ​​used to determine the precoding matrix in conjunction with the first matrix; the step of determining the precoding matrix based on the first downlink channel matrix includes:

[0017] The first downlink channel matrix included in the prediction result is decomposed to obtain the first matrix;

[0018] Based on the first matrix and the second matrix, the precoding matrix is ​​determined.

[0019] Furthermore, according to at least one embodiment of the present invention, the method further includes:

[0020] Send a second instruction to the terminal; the second instruction is used to instruct the terminal to report the second downlink channel matrix between the network device and the terminal;

[0021] Receive the second downlink channel matrix sent by the terminal; and receive the second uplink reference signal sent by the terminal;

[0022] The uplink channel is estimated using the second uplink reference signal to obtain the second uplink channel matrix;

[0023] The second uplink channel matrix is ​​used as the input to the prediction model, and the second downlink channel matrix is ​​used as the target output of the prediction model to train the prediction model and obtain a trained neural network model.

[0024] This invention provides a precoding matrix determination method applied to a terminal, the method comprising:

[0025] The terminal receives a first instruction sent by a network device; the first instruction is used to instruct the terminal not to report a first matrix; the first matrix is ​​used to determine a precoding matrix.

[0026] Send a first uplink reference signal to the network device;

[0027] The first uplink reference signal is used by the network device to obtain a prediction result using a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and a precoding matrix is ​​determined based on the first downlink channel matrix.

[0028] Furthermore, according to at least one embodiment of the present invention, the first instruction is used to instruct the terminal not to report the first matrix.

[0029] Furthermore, according to at least one embodiment of the present invention, the first instruction is used to instruct the terminal not to report the first matrix but to report the second matrix; the second matrix is ​​used to determine the precoding matrix in conjunction with the first matrix.

[0030] Furthermore, according to at least one embodiment of the present invention, the method further includes:

[0031] The terminal receives a second instruction sent by the network device; the second instruction is used to instruct the terminal to report the second downlink channel matrix between the network device and the terminal.

[0032] Send a second downlink channel matrix to the network device; and send a second uplink reference signal to the network device;

[0033] The second uplink reference signal is used by the network device to estimate the uplink channel to obtain a second uplink channel matrix; and the second uplink channel matrix is ​​used as the input of the prediction model, and the second downlink channel matrix is ​​used as the target output of the prediction model to train the prediction model and obtain a trained neural network model.

[0034] At least one embodiment of the present invention provides a precoding matrix determination apparatus, comprising:

[0035] A first sending unit is configured to send a first instruction to a terminal; the first instruction is configured to instruct the terminal not to report a first matrix at least; the first matrix is ​​configured to determine a precoding matrix.

[0036] The first receiving unit is configured to receive the first uplink reference signal sent by the terminal;

[0037] The processing unit is configured to obtain a prediction result using the first uplink reference signal and a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and to determine a precoding matrix based on the first downlink channel matrix.

[0038] At least one embodiment of the present invention provides a precoding determination apparatus, comprising:

[0039] The second receiving unit is configured to receive a first instruction sent by the network device; the first instruction is configured to instruct the terminal not to report the first matrix at least; the first matrix is ​​configured to determine the precoding matrix.

[0040] The second transmitting unit is used to transmit a first uplink reference signal to the network device;

[0041] The first uplink reference signal is used by the network device to obtain a prediction result using a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and a precoding matrix is ​​determined based on the first downlink channel matrix.

[0042] At least one embodiment of the present invention provides a network device, comprising:

[0043] A first communication interface is used to send a first instruction to a terminal; the first instruction is used to instruct the terminal not to report a first matrix at least; the first matrix is ​​used to determine a precoding matrix; and to receive a first uplink reference signal sent by the terminal.

[0044] A first processor is configured to use the first uplink reference signal and a pre-trained neural network model to obtain a prediction result; the prediction result includes a first downlink channel matrix between the network device and the terminal; and to determine a precoding matrix based on the first downlink channel matrix.

[0045] At least one embodiment of the present invention provides a terminal, comprising:

[0046] Second processor,

[0047] The second communication interface is used to receive a first instruction sent by a network device; the first instruction is used to instruct the terminal not to report a first matrix at least; the first matrix is ​​used to determine a precoding matrix; and to send a first uplink reference signal to the network device.

[0048] The first uplink reference signal is used by the network device to obtain a prediction result using a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and a precoding matrix is ​​determined based on the first downlink channel matrix.

[0049] At least one embodiment of the present invention provides a network device, including a processor and a memory for storing a computer program capable of running on the processor.

[0050] Wherein, when the processor is used to run the computer program, it executes the steps of any of the methods described above on the network device side.

[0051] At least one embodiment of the present invention provides a terminal, including a processor and a memory for storing a computer program capable of running on the processor.

[0052] Wherein, when the processor is running the computer program, it executes the steps of any of the above-described terminal-side methods.

[0053] At least one embodiment of the present invention provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the above methods.

[0054] The precoding matrix determination method, apparatus, device, and storage medium provided in this embodiment of the invention send a first instruction to a terminal; the first instruction instructs the terminal not to report a first matrix; the first matrix is ​​used to determine the precoding matrix; a first uplink reference signal sent by the terminal is received; a prediction result is obtained using the first uplink reference signal and a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and the precoding matrix is ​​determined based on the first downlink channel matrix. Using the technical solution provided in this embodiment of the invention, when the terminal does not report the first matrix used to determine the precoding matrix, the network device uses the first uplink reference signal reported by the terminal and a pre-trained neural network model to obtain the first downlink channel matrix, and uses the first downlink channel matrix to determine the precoding matrix. Since the terminal does not need to feed back the first matrix used to determine the precoding matrix to the network device, the problem of increasing feedback overhead to ensure high accuracy under this feedback method can be avoided. Furthermore, determining the precoding matrix in conjunction with the neural network model can avoid quantization errors fed back by the terminal, thereby improving the determination accuracy of the precoding matrix. Attached Figure Description

[0055] Figure 1 This is a schematic diagram illustrating the process by which the terminal feeds back downlink channel quality to the base station in related technologies.

[0056] Figure 2 This is a schematic diagram of Channel State Information (CSI) in related technologies;

[0057] Figure 3 This is a schematic diagram of the implementation flow of the precoding matrix determination method according to an embodiment of the present invention. Figure 1 ;

[0058] Figure 4 This is a schematic diagram of instructions sent from a network device to a terminal according to an embodiment of the present invention;

[0059] Figure 5 This is a schematic diagram of the implementation flow of the precoding matrix determination method according to an embodiment of the present invention. Figure 2 ;

[0060] Figure 6 This is a schematic diagram of the specific implementation process of the precoding matrix determination method according to an embodiment of the present invention. Figure 1 ;

[0061] Figure 7 This is a schematic diagram of the specific implementation process of the precoding matrix determination method according to an embodiment of the present invention. Figure 2 ;

[0062] Figure 8 This is a schematic diagram illustrating the implementation process of training a neural network model according to an embodiment of the present invention;

[0063] Figure 9 This is a schematic diagram of the composition structure of the precoding matrix determination device according to an embodiment of the present invention. Figure 1 ;

[0064] Figure 10 This is a schematic diagram of the composition structure of the precoding matrix determination device according to an embodiment of the present invention. Figure 2 ;

[0065] Figure 11 This is a schematic diagram of the composition structure of the network device according to an embodiment of the present invention;

[0066] Figure 12 This is a schematic diagram of the component structure of the terminal according to an embodiment of the present invention. Detailed Implementation

[0067] Before introducing the technical solutions of the embodiments of the present invention, the relevant technologies will be explained first.

[0068] In related technologies, in frequency division duplex (FDD) massive multiple input multiple output (MIMO) wireless communication systems, the uplink and downlink channels use different frequency bands, resulting in only partial heterogeneity between them. That is, the uplink and downlink channels are only similar in terms of angle and time delay, but their channel attenuation values ​​differ significantly.

[0069] Because the uplink and downlink channels are mutually exclusive, the uplink channel can be used to provide feedback on the downlink channel quality. Figure 1 This is a schematic diagram illustrating the process by which the terminal feeds back downlink channel quality to the base station in related technologies, such as... Figure 1 As shown, the specific implementation process includes: the base station sends a parameter signal (RS, Reference Signal) to the terminal; the terminal performs CSI estimation based on the reference signal to obtain the CSI; and feeds it back to the base station through the uplink channel; the base station selects a precoding / transmission scheme based on the received CSI.

[0070] In related technologies, CSI can be fed back to the base station based on two codebook types, specifically including:

[0071] The first method involves feeding back the CSI to the base station based on the Type I codebook.

[0072] Type I codebooks use a two-level codebook design structure, i.e., W = w1 × w2.

[0073] Among them, w1 is used to report the beam. w2 is used to report the beam selected from the beam group, the weighting coefficients between beams, and the polarization direction, etc. B = [b0, b1, ..., b L-1 ], corresponding to L oversampled DFT beams, therefore, matrix w1 in the Type I codebook represents the beam direction.

[0074] For channel transmission modes rank1 or rank2, matrix W1 can define one beam; or, it can define four adjacent beams.

[0075] When matrix w1 defines only one beam, the corresponding B is a single-column matrix. In this case, matrix w2 is used to adjust the phase between polarizations.

[0076] When matrix w1 corresponds to four adjacent beams, the four adjacent beams correspond to the four columns of matrix B respectively. In this case, matrix w2 is used to select the specific beam to use.

[0077] The second method involves feeding back the CSI to the base station based on the Type II codebook.

[0078] The Type II codebook provides channel information with finer spatial granularity than the Type I codebook. Similar to Type I, Type II is also based on wideband selection and selects beams from a set of possible beams. The difference is that Type I ultimately reports only one beam, while Type II reports up to four beams. For each beam, and its two corresponding polarization directions, the reported PMI provides a corresponding amplitude value (wideband and subband) and a phase value (subband). In other words, the Type II codebook captures the main transmission path and its corresponding amplitude and phase, thus providing detailed information about the channel.

[0079] In related technologies, the process of feeding CSI back to the base station based on the codebook generally adopts a quantization feedback method. That is, the amplitude and phase of the information contained in the CSI are quantized, and then the quantized CSI is fed back to the base station based on the codebook.

[0080] The amplitude is fed back through quantization in the following two ways:

[0081] The first type is the broadband amplitude reporting mode, in which the terminal does not report the differential amplitude on the subband.

[0082] The second method is broadband + sub-band amplitude reporting, where the terminal reports the differential amplitude on the sub-band.

[0083] The subband differential amplitude is quantized using 1 bit, i.e. The bandwidth amplitude uses 3-bit quantization, that is...

[0084] Phase is also fed back after quantization, and the specific quantization method is determined by the high-level configuration parameters.

[0085] Figure 2 This is a schematic diagram of CSI in related technologies. Table 1 shows the content of CSI and its application scenarios.

[0086]

[0087] Table 1

[0088] In related technologies, using codebook feedback CSI to the base station has the following technical drawbacks:

[0089] (1) Using Type I or Type II codebooks to feed back CSI to the base station, since the w1 matrix in Type I and Type II codebooks is a mandatory feedback quantity and the w2 matrix is ​​an optional feedback quantity, this feedback will bring a large feedback overhead to the uplink communication link, thereby reducing spectrum efficiency.

[0090] The w1 matrix is ​​used to report beams. The w2 matrix is ​​used to report the beams selected from the beam group, the weighting coefficients between beams, and the polarization direction.

[0091] (2) Using quantization feedback, the more bits quantized, the greater the feedback accuracy. However, if the feedback accuracy is too high, it will bring a large feedback overhead to the uplink communication link.

[0092] Based on this, in this embodiment of the invention, the network device sends a first instruction to the terminal; the first instruction is used to instruct the terminal not to report a first matrix at least; the first matrix is ​​used to determine a precoding matrix; a first uplink reference signal sent by the terminal is received; a prediction result is obtained using the first uplink reference signal and a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and a precoding matrix is ​​determined based on the first downlink channel matrix.

[0093] Figure 3 This is a schematic diagram illustrating the implementation flow of the precoding matrix determination method according to an embodiment of the present invention, applied to network devices, such as... Figure 3As shown, the method includes steps 301 to 304:

[0094] Step 301: Send a first instruction to the terminal; the first instruction is used to instruct the terminal not to report the first matrix at least; the first matrix is ​​used to determine the precoding matrix.

[0095] It is understood that the embodiments of the present invention can be applied to Type I codebooks or Type II codebooks, and the first matrix can specifically refer to the w1 matrix in the Type I codebook or Type II codebook.

[0096] Understandably, in some applications of Type I codebooks, the precoding matrix can be determined using only the first matrix. Therefore, the network device can instruct the terminal not to report the first matrix. Subsequently, the network device combines the trained neural network model to predict the first downlink channel matrix, and obtains the first matrix based on the predicted first downlink channel matrix. Thus, the precoding matrix is ​​determined based on the first matrix.

[0097] In other words, in this case, the first matrix can specifically refer to the w1 matrix in the Type I codebook.

[0098] Understandably, in some cases applied to Type I codebooks, and in cases applied to Type II codebooks, the precoding matrix can be determined using the first matrix and the second matrix. Therefore, the network device can instruct the terminal not to report the first matrix but to report the second matrix. Subsequently, the network device combines the trained neural network model to predict the first downlink channel matrix, and obtains the first matrix based on the predicted first downlink channel matrix. Thus, based on the first matrix and the second matrix reported by the terminal, the precoding matrix is ​​determined.

[0099] In other words, in this case, the first matrix can specifically refer to the w1 matrix in the Type I codebook or the Type II codebook.

[0100] Step 302: Receive the first uplink reference signal sent by the terminal.

[0101] It is understood that the first uplink reference signal sent by the terminal may specifically refer to the channel sounding reference signal (SRS).

[0102] Step 303: Using the first uplink reference signal and the pre-trained neural network model, obtain the prediction result; the prediction result includes the first downlink channel matrix between the network device and the terminal.

[0103] In practical applications, in FDD systems, although the uplink and downlink transmission frequencies will cause differences in the uplink and downlink channel matrices, the uplink and downlink channel matrices will still have significant dissimilarity components. For example, the uplink and downlink multipath angles and multipath delays are relatively similar and have a certain degree of correlation.

[0104] Meanwhile, with the rapid development of artificial intelligence (AI) technology, AI's predictive capabilities in image recognition and speech recognition have been greatly enhanced, making it possible to predict downlink channel matrices by reasonably deploying AI prediction models on the base station side and collecting uplink channel matrices.

[0105] Based on this, in one embodiment, obtaining the prediction result using the first uplink reference signal and a pre-trained neural network model includes:

[0106] Using the first uplink reference signal, the uplink channel is estimated to obtain the first uplink channel matrix;

[0107] The first uplink channel matrix is ​​input into a pre-trained neural network model to obtain the prediction result.

[0108] It is understandable that the process of training a neural network model may include:

[0109] Send a second instruction to the terminal; the second instruction is used to instruct the terminal to report the second downlink channel matrix between the network device and the terminal;

[0110] Receive the second downlink channel matrix sent by the terminal; and receive the second uplink reference signal sent by the terminal;

[0111] The uplink channel is estimated using the second uplink reference signal to obtain the second uplink channel matrix;

[0112] The second uplink channel matrix is ​​used as the input to the prediction model, and the second downlink channel matrix is ​​used as the target output of the prediction model to train the prediction model and obtain a trained neural network model.

[0113] It is understood that the second downlink channel matrix reported by the terminal can be obtained by channel estimation of the downlink reference signal sent by the network device.

[0114] Step 304: Determine the precoding matrix based on the first downlink channel matrix.

[0115] In some applications using Type I codebooks, the precoding matrix can be determined using only the first matrix. Therefore, the network device can instruct the terminal not to report the first matrix. Subsequently, the network device combines a trained neural network model to predict the first downlink channel matrix, and obtains the first matrix based on the predicted first downlink channel matrix. Thus, the precoding matrix is ​​determined based on the first matrix.

[0116] Based on this, as one implementation, the first instruction is used to instruct the terminal not to report the first matrix; the step of determining the precoding matrix based on the first downlink channel matrix includes:

[0117] The first downlink channel matrix included in the prediction result is decomposed to obtain the first matrix;

[0118] Use the first matrix as the downlink precoding matrix.

[0119] It is understandable that the first matrix may specifically refer to the w1 matrix in the Type I codebook.

[0120] In other cases applied to Type I codebooks, and in cases applied to Type II codebooks, the precoding matrix can be determined using the first matrix and the second matrix. Therefore, the network device can instruct the terminal not to report the first matrix but to report the second matrix. Subsequently, the network device combines a trained neural network model to predict the first downlink channel matrix, and obtains the first matrix based on the predicted first downlink channel matrix. Thus, based on the first matrix and the second matrix reported by the terminal, the precoding matrix is ​​determined.

[0121] In another implementation, the first instruction is used to instruct the terminal not to report the first matrix but to report the second matrix; the second matrix is ​​used to determine the precoding matrix in conjunction with the first matrix; the step of determining the precoding matrix based on the first downlink channel matrix includes:

[0122] Receive the second matrix fed back by the terminal;

[0123] The first downlink channel matrix included in the prediction result is decomposed to obtain the first matrix;

[0124] Based on the first matrix and the second matrix, the precoding matrix is ​​determined.

[0125] It is understood that the first matrix can specifically refer to matrix w1 in the Type I codebook or the Type II codebook. The second matrix can specifically refer to matrix w2 in the Type I codebook or the Type II codebook.

[0126] Figure 4 This is a diagram illustrating the instructions sent from a network device to a terminal, such as... Figure 4 As shown, this can specifically include the following three situations:

[0127] In the first scenario, the network device sends the cri-RI-i1-modified instruction to the terminal, instructing the terminal not to provide a precoding matrix (PMI), i.e., not to provide the first matrix. This applies to Type I codebooks.

[0128] In the second scenario, the network device sends the cri-RI-PMI-CQI-modified instruction to the terminal, instructing the terminal to only provide the second matrix and not the first matrix. This applies to Type I and Type II codebooks.

[0129] In the third scenario, the network device sends a cri-full instruction to the terminal, which instructs the terminal to provide the full (non-quantized) second downlink channel matrix for training the neural network model.

[0130] In this embodiment of the invention, the network device instructs the terminal to report the first matrix, which has the following advantages:

[0131] (1) Not reporting the first matrix can effectively reduce the overhead of uplink channel feedback.

[0132] (2) Without reporting the first matrix, the first downlink channel matrix is ​​obtained using the first uplink reference signal reported by the terminal and the pre-trained neural network model, and the precoding matrix is ​​determined using the first downlink channel matrix. Since the first downlink channel matrix is ​​predicted using a neural network model, the quantization error fed back by the terminal can be avoided, thus effectively improving the accuracy of downlink precoding.

[0133] Figure 5 This is a schematic diagram illustrating the implementation flow of the precoding matrix determination method according to an embodiment of the present invention, applied to a terminal, such as... Figure 5 As shown, the method includes steps 501 to 502:

[0134] Step 501: Receive a first instruction sent by the network device; the first instruction is used to instruct the terminal not to report the first matrix at least; the first matrix is ​​used to determine the precoding matrix.

[0135] It is understood that the embodiments of the present invention can be applied to Type I codebooks or Type II codebooks, and the first matrix can specifically refer to the w1 matrix in the Type I codebook or Type II codebook.

[0136] Understandably, in some applications of Type I codebooks, the precoding matrix can be determined using only the first matrix. Therefore, the network device can instruct the terminal not to report the first matrix. Subsequently, the network device combines the trained neural network model to predict the first downlink channel matrix, and obtains the first matrix based on the predicted first downlink channel matrix. Thus, the precoding matrix is ​​determined based on the first matrix.

[0137] In other words, in this case, the first matrix can specifically refer to the w1 matrix in the Type I codebook.

[0138] Understandably, in some cases applied to Type I codebooks, and in cases applied to Type II codebooks, the precoding matrix can be determined using the first matrix and the second matrix. Therefore, the network device can instruct the terminal not to report the first matrix but to report the second matrix. Subsequently, the network device combines the trained neural network model to predict the first downlink channel matrix, and obtains the first matrix based on the predicted first downlink channel matrix. Thus, based on the first matrix and the second matrix reported by the terminal, the precoding matrix is ​​determined.

[0139] In other words, in this case, the first matrix can specifically refer to the w1 matrix in the Type I codebook or the Type II codebook.

[0140] Step 502: Send a first uplink reference signal to the network device.

[0141] The first uplink reference signal is used by the network device to obtain a prediction result using a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and a precoding matrix is ​​determined based on the first downlink channel matrix.

[0142] It is understood that the first uplink reference signal sent by the terminal may specifically refer to the SRS.

[0143] In some applications using Type I codebooks, the precoding matrix can be determined using only the first matrix. Therefore, the network device can instruct the terminal not to report the first matrix. Subsequently, the network device combines a trained neural network model to predict the first downlink channel matrix, and obtains the first matrix based on the predicted first downlink channel matrix. Thus, the precoding matrix is ​​determined based on the first matrix.

[0144] Based on this, as one implementation method, the first instruction is used to instruct the terminal not to report the first matrix.

[0145] In other cases applied to Type I codebooks, and in cases applied to Type II codebooks, the precoding matrix can be determined using the first matrix and the second matrix. Therefore, the network device can instruct the terminal not to report the first matrix but to report the second matrix. Subsequently, the network device combines a trained neural network model to predict the first downlink channel matrix, and obtains the first matrix based on the predicted first downlink channel matrix. Thus, based on the first matrix and the second matrix reported by the terminal, the precoding matrix is ​​determined.

[0146] Based on this, as another implementation, the first instruction is used to instruct the terminal not to report the first matrix but to report the second matrix; the second matrix is ​​used to determine the precoding matrix in combination with the first matrix.

[0147] In practical applications, network devices can instruct terminals to report the second downlink channel matrix. In this way, network devices can train neural network models based on the second downlink channel matrix reported by the terminals.

[0148] Based on this, in one embodiment, the method further includes:

[0149] The terminal receives a second instruction sent by the network device; the second instruction is used to instruct the terminal to report the second downlink channel matrix between the network device and the terminal.

[0150] Send a second downlink channel matrix to the network device; and send a second uplink reference signal to the network device;

[0151] The second uplink reference signal is used by the network device to estimate the uplink channel to obtain a second uplink channel matrix; and the second uplink channel matrix is ​​used as the input of the prediction model, and the second downlink channel matrix is ​​used as the target output of the prediction model to train the prediction model and obtain a trained neural network model.

[0152] In this embodiment of the invention, the network device instructs the terminal not to report the first matrix, which has the following advantages:

[0153] (1) Not reporting the first matrix can effectively reduce the overhead of uplink channel feedback.

[0154] (2) Without reporting the first matrix, the terminal reports the first uplink reference signal to the network device. The network device can then use the first uplink reference signal and a pre-trained neural network model to obtain the first downlink channel matrix and determine the precoding matrix. Since using a neural network model to predict the first downlink channel matrix avoids quantization errors fed back by the terminal, it effectively improves the accuracy of downlink precoding.

[0155] Figure 6 This is a schematic diagram illustrating the specific implementation flow of the precoding matrix determination method according to an embodiment of the present invention, as shown below. Figure 6 As shown, taking a network device as a base station as an example, the base station can be a gNB; the method includes steps 601 to 606:

[0156] Step 601: The terminal receives the cri-RI-i1-modified instruction sent by the base station; the cri-RI-i1-modified instruction is used to indicate that the terminal does not need to feed back the w1 matrix.

[0157] Understandably, the base station can send the higher-layer configuration parameter `reportQuantity` to the terminal. This higher-layer configuration parameter `reportQuantity` is configured as the `cri-RI-i1-modified` instruction, which instructs the terminal not to feed back any precoding matrix (PMI), i.e., the w1 matrix.

[0158] It is understandable that the first matrix may specifically refer to the w1 matrix in the Type I codebook.

[0159] Step 602: The terminal sends CSI to the base station; the CSI includes Channel Quality Indicator (CQI) and Rank Indicator (RI), but does not include the w1 matrix.

[0160] Step 603: The terminal sends an SRS to the base station.

[0161] Step 604: The base station receives the SRS sent by the terminal and estimates the first uplink channel matrix H1' based on the SRS.

[0162] Step 605: The base station inputs the first uplink channel matrix H1' into the trained neural network to obtain the first downlink channel matrix H1.

[0163] Step 606: The base station decomposes the first downlink channel matrix H1 to obtain the first matrix w1; the first matrix w1 is used as the downlink precoding matrix W.

[0164] In other words, the downlink precoding matrix W = w1.

[0165] In this example, the terminal does not report the w1 matrix, which has the following advantages:

[0166] (1) Applicable to Type I codebook.

[0167] Taking a Type I codebook from related technologies as an example, the CSI content reported by this Type I codebook is 'cri-R1-i1-CQI'. The purpose is quasi-open-loop CSI reporting, where the terminal only reports w1 and not w2. The terminal randomly selects one of the multiple w2 values ​​corresponding to w1 for each PRG to calculate the CQI. Compared to the method in related technologies that determines the precoding matrix based on this Type I codebook-reported CSI, in this example, because the terminal does not report the w1 matrix, the uplink channel feedback overhead can be reduced.

[0168] (2) While reducing overhead, the use of a neural network model to predict the first downlink channel matrix can avoid quantization errors fed back by the terminal, thus effectively improving the accuracy of downlink precoding.

[0169] Figure 7 This is a schematic diagram illustrating the specific implementation flow of the precoding matrix determination method according to an embodiment of the present invention, as shown below. Figure 7 As shown, taking a network device as a base station as an example, the method includes steps 701 to 706:

[0170] Step 701: The terminal receives the cri-RI-PMI-CQI-modified instruction sent by the base station; the cri-RI-PMI-CQI-modified instruction is used to instruct the terminal not to feed back the w1 matrix but to feed back the w2 matrix.

[0171] Understandably, the base station can send the higher-layer configuration parameter `reportQuantity` to the terminal. When this parameter is configured with the `cri-RI-PMI-CQI-modified` directive, it can instruct the terminal not to report the `w1` matrix but to report the `w2` matrix.

[0172] It is understood that the first matrix can specifically refer to matrix w1 in the Type I codebook or the Type II codebook. The second matrix can specifically refer to matrix w2 in the Type I codebook or the Type II codebook.

[0173] Step 702: The terminal sends CSI to the base station; the CSI includes CQI, RI and w2 matrix, but does not include w1 matrix.

[0174] Step 703: The terminal sends SRS to the base station.

[0175] Step 704: The base station receives the SRS sent by the terminal and estimates the first uplink channel matrix H1' based on the SRS.

[0176] Step 705: The base station inputs the first uplink channel matrix H1' into the trained neural network to obtain the first downlink channel matrix H1.

[0177] Step 706: The base station decomposes the first downlink channel matrix H1 to obtain the first matrix w1.

[0178] Step 707: The base station uses the first matrix w1 and the w2 matrix sent by the terminal to determine the precoding matrix W.

[0179] It is understandable that the downlink precoding matrix W = w1 × w2.

[0180] In this example, the terminal does not report matrix w1, but reports matrix w2, which has the following advantages:

[0181] (1) Applicable to Type I codebook and Type II codebook.

[0182] Taking a Type I or Type II codebook from related technologies as an example, the CSI content reported by this Type I or Type II codebook is 'cri-RI-PMI-CQI'. The purpose is to report CSI based on PMI, with the terminal reporting w1 and w2, and to support beam selection and CSI reporting based on CSI-RS. Compared to the method in related technologies that uses this Type I or Type II codebook to report CSI to determine the precoding matrix, in this example, since there is no need to report the w1 matrix, the uplink channel feedback overhead is reduced.

[0183] (2) While reducing overhead, the accuracy of downlink precoding can be effectively improved by using a neural network model to predict the first downlink channel matrix and avoiding quantization errors fed back by the terminal.

[0184] Figure 8 This is a schematic diagram illustrating the implementation process of training a neural network model according to an embodiment of the present invention, as shown below. Figure 8 As shown, taking a network device as a base station as an example, the method includes steps 801 to 805:

[0185] Step 801: The terminal receives the cri-full instruction sent by the base station; the cri-full instruction is used to instruct the terminal to feed back the second downlink channel matrix for training the neural network model.

[0186] Understandably, the base station can send the higher-layer configuration parameter `reportQuantity` to the terminal. This higher-layer configuration parameter `reportQuantity` is configured to use the `cri-full` command.

[0187] Step 802: The terminal determines the second downlink channel matrix H1 and feeds back the second downlink channel matrix H1 to the base station through the uplink.

[0188] Step 803: The terminal sends a second uplink reference signal to the base station.

[0189] Step 804: The base station estimates based on the second uplink reference signal to determine the second uplink channel matrix H1'.

[0190] Step 805: The base station uses the second uplink channel matrix H1' as the input to the prediction model and the second downlink channel matrix H1 as the target output of the prediction model to train the prediction model and obtain a trained neural network model.

[0191] In this example, the reasonable deployment of the neural network model on the network device side has the following advantages:

[0192] (1) The AI ​​model is trained offline by collecting uplink and downlink channel data and applying the collected data. Subsequently, the trained neural network model can be used for online real-time prediction of the first downlink channel matrix.

[0193] (2) When the second uplink reference signal is the full amount, the accuracy of the trained neural network model can be improved.

[0194] To implement the precoding matrix determination method of the present invention, the present invention also provides a precoding matrix determination device, which is installed on a network device. Figure 9 This is a schematic diagram of the composition structure of the precoding matrix determination device according to an embodiment of the present invention, as shown below. Figure 9 As shown, the device includes:

[0195] The first sending unit 91 is configured to send a first instruction to the terminal; the first instruction is configured to instruct the terminal not to report the first matrix at least; the first matrix is ​​configured to determine the precoding matrix.

[0196] The first receiving unit 92 is used to receive the first uplink reference signal sent by the terminal;

[0197] Processing unit 93 is configured to obtain a prediction result using the first uplink reference signal and a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and to determine a precoding matrix based on the first downlink channel matrix.

[0198] In one embodiment, the processing unit 93 is specifically used for:

[0199] Using the first uplink reference signal, the uplink channel is estimated to obtain the first uplink channel matrix; the first uplink channel matrix is ​​input into a pre-trained neural network model to obtain the prediction result.

[0200] In one embodiment, the processing unit 93 is specifically used for:

[0201] The first instruction is used to instruct the terminal not to report the first matrix; to decompose the first downlink channel matrix included in the prediction result to obtain the first matrix; and to use the first matrix as the downlink precoding matrix.

[0202] In one embodiment, the processing unit 93 is specifically used for:

[0203] The first instruction is used to instruct the terminal not to report the first matrix but to report the second matrix; the second matrix is ​​used to determine the precoding matrix in combination with the first matrix; the second matrix sent by the terminal is received; the first downlink channel matrix included in the prediction result is decomposed to obtain the first matrix; and the precoding matrix is ​​determined based on the first matrix and the second matrix.

[0204] In one embodiment, the first sending unit 91 is further configured to send a second instruction to the terminal; the second instruction is configured to instruct the terminal to report the second downlink channel matrix between the network device and the terminal;

[0205] Accordingly, the first receiving unit 92 is also configured to receive the second downlink channel matrix sent by the terminal; and to receive the second uplink reference signal sent by the terminal;

[0206] Accordingly, the processing unit 93 is further configured to estimate the uplink channel using the second uplink reference signal to obtain a second uplink channel matrix; use the second uplink channel matrix as the input of the prediction model and the second downlink channel matrix as the target output of the prediction model to train the prediction model and obtain a trained neural network model.

[0207] In practical applications, the first sending unit 91 and the first receiving unit 92 can be implemented by the communication interface in the precoding determination device; the processing unit 93 can be implemented by the processor in the precoding matrix determination device.

[0208] It should be noted that the precoding matrix determination device provided in the above embodiments is only illustrated by the division of the above program modules when performing information processing. In practical applications, the above processing can be assigned to different program modules as needed, that is, the internal structure of the device can be divided into different program modules to complete all or part of the processing described above. In addition, the precoding matrix determination device and the information processing method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0209] To implement the precoding matrix determination method of the present invention, the present invention also provides a precoding matrix determination device, which is installed on a terminal. Figure 10This is a schematic diagram of the composition structure of the precoding matrix determination device according to an embodiment of the present invention, as shown below. Figure 10 As shown, the device includes:

[0210] The second receiving unit 101 is configured to receive a first instruction sent by the network device; the first instruction is configured to instruct the terminal not to report the first matrix at least; the first matrix is ​​configured to determine the precoding matrix.

[0211] The second transmitting unit 102 is used to transmit a first uplink reference signal to the network device;

[0212] The first uplink reference signal is used by the network device to obtain a prediction result using a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and a precoding matrix is ​​determined based on the first downlink channel matrix.

[0213] In one embodiment, the first instruction is used to instruct the terminal not to report the first matrix.

[0214] In one embodiment, the first instruction is used to instruct the terminal not to report the first matrix but to report the second matrix; the second matrix is ​​used to determine the precoding matrix in conjunction with the first matrix.

[0215] In one embodiment, the second receiving unit 101 is further configured to:

[0216] The terminal receives a second instruction sent by the network device; the second instruction is used to instruct the terminal to report the second downlink channel matrix between the network device and the terminal.

[0217] Send a second downlink channel matrix to the network device; and send a second uplink reference signal to the network device;

[0218] The second uplink reference signal is used by the network device to estimate the uplink channel to obtain a second uplink channel matrix; and the second uplink channel matrix is ​​used as the input of the prediction model, and the second downlink channel matrix is ​​used as the target output of the prediction model to train the prediction model and obtain a trained neural network model.

[0219] In practical applications, the second receiving unit 101 and the second sending unit 102 can be connected via the communication interface in the precoding determination device.

[0220] It should be noted that the precoding matrix determination device provided in the above embodiments is only illustrated by the division of the above program modules when performing information processing. In practical applications, the above processing can be assigned to different program modules as needed, that is, the internal structure of the device can be divided into different program modules to complete all or part of the processing described above. In addition, the precoding matrix determination device and the information processing method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.

[0221] This invention also provides a network device, such as... Figure 11 As shown, it includes:

[0222] The first communication interface 111 is capable of exchanging information with other devices;

[0223] The first processor 112, connected to the first communication interface 111, is used to execute the methods provided by one or more technical solutions on the network device side when running a computer program. The computer program is stored in the first memory 113.

[0224] It should be noted that the specific processing procedures of the first processor 112 and the first communication interface 111 are detailed in the method embodiment and will not be repeated here.

[0225] Of course, in practical applications, the various components in network device 110 are coupled together through bus system 114. It can be understood that bus system 114 is used to implement communication between these components. In addition to a data bus, bus system 114 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in... Figure 11 The general labeled all buses as Bus System 114.

[0226] The first memory 113 in this embodiment is used to store various types of data to support the operation of the network device 110. Examples of such data include any computer program used to operate on the network device 110.

[0227] The methods disclosed in the above embodiments of this application can be applied to the first processor 112, or implemented by the first processor 112. The first processor 112 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware or by instructions in the form of software in the first processor 112. The first processor 112 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The first processor 112 can implement or execute the methods, steps and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly reflected as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, which is located in the first memory 113. The first processor 112 reads the information in the first memory 113 and completes the steps of the aforementioned method in combination with its hardware.

[0228] This invention also provides a terminal, such as... Figure 12 As shown, it includes:

[0229] The second communication interface 121 is capable of exchanging information with other devices;

[0230] The second processor 122, connected to the second communication interface 121, is used to execute the methods provided by one or more of the aforementioned terminal-side technical solutions when running a computer program. The computer program is stored in the second memory 123.

[0231] It should be noted that the specific processing procedures of the second processor 122 and the second communication interface 121 are detailed in the method embodiment and will not be repeated here.

[0232] Of course, in practical applications, the various components in terminal 120 are coupled together through bus system 124. It can be understood that bus system 124 is used to implement communication between these components. In addition to a data bus, bus system 124 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in... Figure 12 The general labeled all buses as Bus System 124.

[0233] The second memory 123 in this embodiment is used to store various types of data to support the operation of the terminal 120. Examples of such data include any computer program used to operate on the terminal 120.

[0234] The methods disclosed in the above embodiments of this application can be applied to the second processor 122, or implemented by the second processor 122. The second processor 122 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware or by instructions in the form of software in the second processor 122. The second processor 122 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The second processor 122 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, which is located in the second memory 123. The second processor 122 reads the information in the second memory 123 and completes the steps of the aforementioned method in conjunction with its hardware.

[0235] In an exemplary embodiment, the network device 110 and the terminal 120 may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components to perform the aforementioned method.

[0236] It is understood that the memories (first memory 113, second memory 123) in the embodiments of this application can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM); magnetic surface memory can be disk storage or magnetic tape storage. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), SyncLink Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM).The memories described in the embodiments of this application are intended to include, but are not limited to, these and any other suitable types of memories.

[0237] In an exemplary embodiment, the present invention also provides a storage medium, namely a computer storage medium, specifically a computer-readable storage medium, such as a memory that stores a computer program. This computer program can be executed by the second processor 122 of the terminal 120 to complete the steps described in the aforementioned terminal-side method. The computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface memory, optical disc, or CD-ROM.

[0238] It should be noted that , , etc. are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0239] Furthermore, the technical solutions described in the embodiments of the present invention can be combined arbitrarily without conflict.

[0240] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention.

Claims

1. A method for determining a precoding matrix, characterized in that, Applied to network devices, the method includes: Send a first instruction to the terminal; the first instruction is used to instruct the terminal not to report a first matrix at least; the first matrix is ​​used to determine a precoding matrix; wherein, when the first instruction instructs the terminal to report a second matrix, the second matrix is ​​used to determine a precoding matrix in conjunction with the first matrix; Receive the first uplink reference signal sent by the terminal; Using the first uplink reference signal and a pre-trained neural network model, a prediction result is obtained; the prediction result includes the first downlink channel matrix between the network device and the terminal. Based on the first downlink channel matrix, the precoding matrix is ​​determined.

2. The method according to claim 1, characterized in that, The step of obtaining the prediction result using the first uplink reference signal and the pre-trained neural network model includes: Using the first uplink reference signal, the uplink channel is estimated to obtain the first uplink channel matrix; The first uplink channel matrix is ​​input into a pre-trained neural network model to obtain the prediction result.

3. The method according to claim 1 or 2, characterized in that, The first instruction is used to instruct the terminal not to report the first matrix; the step of determining the precoding matrix based on the first downlink channel matrix includes: The first downlink channel matrix included in the prediction result is decomposed to obtain the first matrix; Use the first matrix as the downlink precoding matrix.

4. The method according to claim 1 or 2, characterized in that, The first instruction is used to instruct the terminal not to report the first matrix but to report the second matrix; the second matrix is ​​used to determine the precoding matrix in conjunction with the first matrix; the step of determining the precoding matrix based on the first downlink channel matrix includes: Receive the second matrix sent by the terminal; The first downlink channel matrix included in the prediction result is decomposed to obtain the first matrix; Based on the first matrix and the second matrix, the precoding matrix is ​​determined.

5. The method according to claim 1 or 2, characterized in that, The method further includes: Send a second instruction to the terminal; the second instruction is used to instruct the terminal to report the second downlink channel matrix between the network device and the terminal; Receive the second downlink channel matrix sent by the terminal; and receive the second uplink reference signal sent by the terminal; The uplink channel is estimated using the second uplink reference signal to obtain the second uplink channel matrix; The second uplink channel matrix is ​​used as the input to the prediction model, and the second downlink channel matrix is ​​used as the target output of the prediction model to train the prediction model and obtain a trained neural network model.

6. A method for determining a precoding matrix, characterized in that, Applied to a terminal, the method includes: The terminal receives a first instruction sent by a network device; the first instruction is used to instruct the terminal not to report a first matrix; the first matrix is ​​used to determine a precoding matrix; wherein, when the first instruction instructs the terminal to report a second matrix, the second matrix is ​​used to determine a precoding matrix in conjunction with the first matrix. Send a first uplink reference signal to the network device; The first uplink reference signal is used by the network device to obtain a prediction result using a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and a precoding matrix is ​​determined based on the first downlink channel matrix.

7. The method according to claim 6, characterized in that, The first instruction is used to instruct the terminal not to report the first matrix.

8. The method according to claim 6, characterized in that, The first instruction is used to instruct the terminal not to report the first matrix but to report the second matrix; the second matrix is ​​used to determine the precoding matrix in combination with the first matrix.

9. The method according to claim 6, characterized in that, The method further includes: The terminal receives a second instruction sent by the network device; the second instruction is used to instruct the terminal to report the second downlink channel matrix between the network device and the terminal. Send a second downlink channel matrix to the network device; and send a second uplink reference signal to the network device; The second uplink reference signal is used by the network device to estimate the uplink channel to obtain a second uplink channel matrix; and the second uplink channel matrix is ​​used as the input of the prediction model, and the second downlink channel matrix is ​​used as the target output of the prediction model to train the prediction model and obtain a trained neural network model.

10. A precoding matrix determination device, characterized in that, Applied to network devices, the device includes: A first sending unit is configured to send a first instruction to a terminal; the first instruction is configured to instruct the terminal not to report a first matrix at least; the first matrix is ​​configured to determine a precoding matrix; wherein, when the first instruction instructs the terminal to report a second matrix, the second matrix is ​​configured to combine with the first matrix to determine a precoding matrix; The first receiving unit is configured to receive the first uplink reference signal sent by the terminal; The processing unit is configured to obtain a prediction result using the first uplink reference signal and a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and to determine a precoding matrix based on the first downlink channel matrix.

11. A precoding determination device, characterized in that, Applied to a terminal, the device includes: The second receiving unit is configured to receive a first instruction sent by the network device; the first instruction is configured to instruct the terminal not to report a first matrix at least; the first matrix is ​​configured to determine a precoding matrix; wherein, when the first instruction instructs the terminal to report a second matrix, the second matrix is ​​configured to combine with the first matrix to determine a precoding matrix; The second transmitting unit is used to transmit a first uplink reference signal to the network device; The first uplink reference signal is used by the network device to obtain a prediction result using a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and a precoding matrix is ​​determined based on the first downlink channel matrix.

12. A network device, characterized in that, include: The first communication interface is used to send the first instruction to the terminal; The first instruction is used to instruct the terminal not to report the first matrix at least; The first matrix is ​​used to determine the precoding matrix; And receive a first uplink reference signal sent by the terminal; wherein, when the first instruction is used to instruct the terminal to report the second matrix, the second matrix is ​​used to determine the precoding matrix in conjunction with the first matrix; A first processor is configured to use the first uplink reference signal and a pre-trained neural network model to obtain a prediction result; the prediction result includes a first downlink channel matrix between the network device and the terminal; and to determine a precoding matrix based on the first downlink channel matrix.

13. A terminal, characterized in that, include: Second processor, The second communication interface is used to receive the first instruction sent by the network device; The first instruction is used to instruct the terminal not to report the first matrix at least; The first matrix is ​​used to determine the precoding matrix; And send a first uplink reference signal to the network device; wherein, when the terminal reports the second matrix, the second matrix is ​​used to determine the precoding matrix in conjunction with the first matrix; The first uplink reference signal is used by the network device to obtain a prediction result using a pre-trained neural network model; the prediction result includes a first downlink channel matrix between the network device and the terminal; and a precoding matrix is ​​determined based on the first downlink channel matrix.

14. A network device, characterized in that, This includes a processor and memory for storing computer programs that can run on the processor. When the processor is used to run the computer program, it performs the steps of the method according to any one of claims 1 to 5.

15. A terminal, characterized in that, This includes a processor and memory for storing computer programs that can run on the processor. When the processor is used to run the computer program, it performs the steps of the method according to any one of claims 6 to 9.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 9.

Images