Base station and terminal

Abstract

The present disclosure provides a base station and a terminal. The base station comprises: a receiving unit configured to receive precoding matrix indication information of a first granularity from a terminal; a processing unit configured to perform channel reconstruction according to the precoding matrix indication information to obtain a first channel, perform interpolation and denoising processing on the channel of the first granularity using a super-resolution network to obtain a second channel, and perform downlink precoding on the second channel, wherein the first channel has the first granularity, the second channel has a second granularity, and the second granularity is finer than the first granularity.

Description

Technical Field

[0001] This disclosure relates to the field of wireless communication, and more specifically to a base station and a corresponding terminal for channel reconstruction based on feedback from a terminal. Background Technology

[0002] In a communication system, the base station sends a downlink reference signal to the UE. The UE performs channel estimation based on the downlink reference signal and sends Type I or Type II precoding matrix indication (PMI) information to the base station. The base station can determine the corresponding codewords and codeword-related combination coefficients based on the PMI information sent by the UE to reconstruct the channel, and use the reconstructed channel for downlink precoding. Currently, the Type II PMI information sent by the UE is subband-level PMI information, which results in large quantization granularity in both the spatial and frequency domains, as well as large quantization granularity of the combination coefficients. Accordingly, the base station reconstructs the subband-level channel based on the subband-level PMI information sent by the UE. On the other hand, in a 5G NR system, the base station can use the reconstructed channel for precoding on a physical resource block bundling basis, and the granularity of physical resource block bundling is usually much smaller than the granularity of the PMI information.

[0003] For example, with a communication bandwidth of 100MHz between the base station and the UE and a subcarrier spacing (SCS) of 30kHz, a subband can include 16 resource blocks (RBs). The UE can transmit Type II PMI information at the subband (i.e., 16 RBs) level. On the other hand, the minimum size of physical resource block binding can be 2 RBs, which is much smaller than the granularity of PMI information.

[0004] Furthermore, when reconstructing the subband-level channel based on PMI information, it introduces significant errors in the quantization noise of the spatial domain and coefficients. This reduces the ability of various multiple-input multiple-output (MIMO) precoding methods to eliminate inter-user interference. Therefore, it is necessary to reduce quantization noise. Summary of the Invention

[0005] According to one aspect of this disclosure, a base station is provided. The base station includes: a receiving unit configured to receive precoding matrix indication information of a first granularity from a terminal; and a processing unit configured to perform channel reconstruction based on the precoding matrix indication information to obtain a first channel, perform interpolation and denoising processing on the first granularity channel using a super-resolution network to obtain a second channel, and perform downlink precoding on the second channel, wherein the first channel has the first granularity, the second channel has a second granularity, and the second granularity is finer than the first granularity.

[0006] According to another aspect of this disclosure, another base station is provided. The base station includes: a receiving unit configured to receive precoding matrix indication information of a first granularity from a terminal; and a processing unit configured to perform channel reconstruction, interpolation, and denoising processing based on the precoding matrix indication information of the first granularity through a first sub-network to obtain a second channel, and to perform downlink precoding on the channel of the second granularity, wherein the second granularity is finer than the first granularity.

[0007] According to another aspect of this disclosure, another base station is provided. The base station includes: a transmitting unit configured to transmit first channel state information reference information of a first density to a terminal; and a receiving unit configured to receive first feedback information from the terminal regarding the first channel state information reference information; wherein the first feedback information includes first channel response information obtained by the terminal performing a first channel estimation based on the first channel state information reference information, and precoding matrix indication information determined by the terminal based on the downsampled first channel state information reference information.

[0008] According to another aspect of this disclosure, a terminal is provided. The terminal includes: a receiving unit configured to receive first channel state information reference information of a first density; a processing unit configured to perform a first channel estimation based on the first channel state information reference information to obtain first channel response information, and to perform downsampling processing on the first channel state information reference information, and determine precoding matrix indication information based on the downsampling channel state information reference information; and a transmitting unit configured to transmit the first channel response information and the precoding matrix indication information to a base station. Attached Figure Description

[0009] The above and other objects, features, and advantages of this disclosure will become more apparent from the more detailed description of the embodiments thereof in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this disclosure and form part of the specification. They are used together with the embodiments of this disclosure to explain the disclosure and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0010] Figure 1 This is a schematic diagram illustrating how a base station reconstructs a channel based on feedback from a terminal in a communication system.

[0011] Figure 2 This is a schematic diagram illustrating the Channel State Information (CSI) and the ideal CSI determined based on the PMI information fed back by the terminal in an existing communication system.

[0012] Figure 3 This is a schematic block diagram illustrating a base station according to an embodiment of the present disclosure.

[0013] Figure 4 This is a schematic block diagram illustrating a base station according to another embodiment of the present disclosure.

[0014] Figure 5 This is a schematic block diagram illustrating a base station according to another embodiment of the present disclosure.

[0015] Figure 6 This is a schematic diagram illustrating an example of training a neural network using a PMI training dataset and a corresponding channel response dataset, according to the present disclosure.

[0016] Figure 7 This is a schematic block diagram illustrating a terminal according to an embodiment of the present disclosure.

[0017] Figure 8 This is a schematic diagram illustrating, according to an example of the present disclosure, the processing unit processes the first channel state information reference information.

[0018] Figure 9 This is a flowchart of a channel processing method according to an embodiment of the present disclosure.

[0019] Figure 10 This is a flowchart of a channel processing method according to another embodiment of the present disclosure.

[0020] Figure 11 This is a flowchart of a reference signal transmission method according to an embodiment of the present disclosure.

[0021] Figure 12 This is a flowchart of an information sending method according to an embodiment of the present disclosure.

[0022] Figure 13 This is a schematic diagram of the hardware structure of a device according to an embodiment of the present disclosure.

[0023] Figures 14A-14C This is a schematic diagram illustrating a first sub-network structure according to an embodiment of the present disclosure. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this disclosure more apparent, exemplary embodiments according to this disclosure will now be described in detail with reference to the accompanying drawings. In the drawings, the same reference numerals denote the same elements throughout. It should be understood that the embodiments described herein are merely illustrative and should not be construed as limiting the scope of this disclosure. Furthermore, the terminals described herein may include various types of terminals, such as user equipment (UE), mobile terminals (or mobile stations), or fixed terminals; however, for convenience, the terms "terminal" and "UE" are sometimes used interchangeably below.

[0025] Figure 1This is a schematic diagram illustrating how a base station reconstructs a channel based on feedback from a terminal in a communication system. For example... Figure 1 As shown, terminal 110 performs channel estimation based on the downlink reference signal and obtains Type I or Type II precoding matrix indication (PMI) information based on the channel estimation result to send to base station 120. Base station 120 reconstructs the channel based on the PMI information sent by the UE and uses the reconstructed channel for downlink precoding. The ideal spatial-temporal channel H that the base station expects to obtain can be expressed by the following formula (1):

[0026]

[0027] The spatial-temporal channel H can be viewed as the superposition of N multipath components, each with an amplitude of α. i It can be written as a delay τ i Horizontal angle of arrival Vertical angle of arrival θ i Phase φ i The function F. Currently, it is very difficult to estimate each parameter in the above formula one by one based on the downlink reference signal, thus making it difficult to accurately recover the channel.

[0028] Specifically, in the current communication system, terminal 110 transmits subband-level Type II PMI information. This results in large quantization granularity in both the spatial and frequency domains, as well as large quantization granularity in the combination coefficients. Correspondingly, the channel reconstructed by base station 120 based on the PMI information is also a subband-level channel, which makes the granularity of the reconstructed channel coarse. Figure 2 This diagram illustrates the Channel State Information (CSI) determined based on PMI information fed back from the terminal in an existing communication system, and the ideal CSI. Figure 2 In the example shown, a diamond represents the CSI determined based on the PMI information fed back by the terminal, and a circle represents the ideal channel state information. For example... Figure 2 As shown, the granularity of CSI determined based on PMI information fed back by the terminal is much coarser than that of ideal channel state information. When the base station uses CSI directly reconstructed from PMI, the spectral efficiency is reduced by more than 50% compared to using ideal CSI.

[0029] On the other hand, in 5G NR systems, base stations can use reconstructed channels for precoding on a physical resource block binding basis, and the granularity of physical resource block binding is usually much smaller than the granularity of PMI information.

[0030] Furthermore, when reconstructing the subband-level channel based on the sub-PMI information, it introduces significant errors in the quantization noise of the spatial domain and coefficients. This reduces the ability of various multiple-input multiple-output (MIMO) precoding methods to eliminate inter-user interference. Therefore, it is necessary to reduce the quantization noise.

[0031] The following is for reference. Figure 3 To illustrate a base station according to an embodiment of the present disclosure. Figure 3 This is a schematic block diagram illustrating a base station according to one embodiment of the present disclosure. Figure 3 As shown, a base station 300 according to one embodiment of this disclosure may include a receiving unit 310 and a processing unit 320. In addition to the receiving unit and the processing unit, the base station 300 may also include other components; however, since these components are not relevant to the content of this disclosure embodiment, their illustrations and descriptions are omitted here.

[0032] like Figure 3 As shown, the receiving unit 310 of the base station 300 receives precoding matrix indication information of the first granularity from the terminal. The processing unit 320 performs channel reconstruction based on the precoding matrix indication information to obtain a first channel, wherein the first channel has a first granularity. For example, the receiving unit 310 may receive sub-band level type II PMI information from the terminal. Specifically, the type II PMI information may include spatial codeword selection information, amplitude and phase information of the wideband and sub-band level codeword combination coefficients, and may also include frequency domain codeword selection information and spatial-frequency domain codeword combination coefficients. The processing unit 320 may use the amplitude and phase information in the PMI information received by the receiving unit 310 to perform amplitude and phase weighting on the spatial (also called "beam domain")-frequency domain channel codewords of multiple beams, and merge the weighted vectors to obtain a first channel with sub-band level. The first channel of the sub-band level obtained after merging can be represented in the form of a spatial-frequency channel matrix. The spatial value of the spatial-frequency channel matrix can be the number of antennas through which the base station 300 communicates with the terminal, while the frequency value can be determined according to the number of sub-bands through which the base station 300 communicates with the terminal.

[0033] Then, the processing unit 320 uses a super-resolution network to interpolate and denoise the channel at the first granularity to obtain a second channel, wherein the second channel has a second granularity, and the second granularity is finer than the first granularity. For example, as described above, the channel at the first granularity can be a subband-level channel. In this case, the channel at the second granularity can be a subcarrier-level or resource block (RB)-level channel.

[0034] According to an example of this disclosure, before inputting the first channel of the first granularity into the super-resolution network, the processing unit 320 can preprocess the first channel to facilitate subsequent operation of the super-resolution network. For example, if the first channel is a spatial-frequency domain channel, the processing unit 320 can perform a Fourier transform on the first channel to convert the spatial-frequency domain channel into a beam-delay domain channel. Furthermore, since the channel delay component is mainly concentrated at the beginning of the delay domain channel matrix in the beam-delay domain, the processing unit 320 can truncate the delay domain channel, retaining the beginning, and divide the truncated data into two channels, real and imaginary, as input to the super-resolution network. By transforming the first channel to be processed to the delay domain and truncating the data, the computational complexity of the super-resolution network can be reduced.

[0035] Furthermore, to further simplify the operation of the super-resolution network, before transforming it to the beam-delay domain channel, the processing unit 320 can also pre-interpolate the first channel using existing interpolation methods such as zero-filling, linear interpolation, or nearest-neighbor interpolation to obtain the desired frequency domain accuracy (e.g., RB-level or subcarrier-level). It should be understood that this pre-interpolation operation only formally increases the channel dimension to the output dimension for easier processing by the subsequent network, without substantially improving the channel accuracy. Super-resolution interpolation of the channel matrix is ​​performed by the super-resolution network.

[0036] According to one example of this disclosure, processing unit 320 can use various super-resolution networks. The first channel at a first granularity can be interpolated and denoised using a method similar to that used for image interpolation and denoising with a super-resolution network. Furthermore, the super-resolution network can be pre-trained using a high-density reference signal. For example, base station 300 may also include a transmitting unit to transmit a high-density reference signal to at least one of user equipment and data acquisition device, and to receive feedback information from at least one of user equipment and data acquisition device regarding the first channel state information reference information. Processing unit 320 can use the feedback information regarding the first channel state information reference information to train the super-resolution network so that the super-resolution network learns to obtain a candidate set of parameters and functions F of the channel as shown in the above formula (1) corresponding to specific feedback information through interpolation and denoising processing. Thus, in actual deployment, processing unit 320 can input the first channel obtained according to the precoding matrix indication information fed back by the terminal into the trained super-resolution network for accurate channel recovery.

[0037] For example, processing unit 320 can use at least one of a Very Deep Super Resolution (VDSR) network and a Cascading Residual Network (CARN) to interpolate and denoise the channel at the first granularity to obtain a second channel. Specifically, processing unit 320 can use a VDSR with 16-20 layers and a kernel size of 3 to interpolate and denoise the channel at the first granularity to obtain a second channel. Since deep networks are beneficial for learning features in the channel, using deep networks can improve channel recovery. Furthermore, according to an example of this disclosure, a residual network structure can be applied in a deep super resolution network. Specifically, the input can be superimposed before the output layer to enhance the correspondence between the output and the input.

[0038] Alternatively, the processing unit 320 may also use a cascaded residual convolutional network formed by introducing multiple small convolutional networks into the residual structure. In the cascaded residual network, each small convolutional network can be a 3-layer convolutional network with a kernel size of 3. Compared to deep super-resolution networks, cascaded residual networks can achieve better performance with less complexity.

[0039] Finally, the processing unit 320 performs downlink precoding on the second channel for transmission to the terminal. In combination with... Figure 3 In the described embodiment, after the base station's processing unit performs conventional channel reconstruction based on the precoding matrix indication information, it uses a super-resolution network to interpolate and denoise the first-granularity channel to obtain a second channel with finer granularity for downlink precoding.

[0040] According to another embodiment of this disclosure, in addition to interpolation and denoising, a neural network can also be used to perform channel reconstruction based on precoding matrix indication information of a first granularity received from the terminal. Hereinafter, references... Figure 4 To illustrate another embodiment of the base station according to this disclosure. Figure 4 This is a schematic block diagram illustrating a base station according to another embodiment of the present disclosure. Figure 4 As shown, a base station 300 according to one embodiment of this disclosure may include a receiving unit 410 and a processing unit 420. In addition to the receiving unit and the processing unit, the base station 400 may also include other components; however, since these components are not relevant to the content of this disclosure embodiment, their illustrations and descriptions are omitted here.

[0041] like Figure 4As shown, the receiving unit 410 of the base station 400 receives precoding matrix indication information of the first granularity from the terminal. For example, the receiving unit 410 may receive sub-band level type II PMI information from the terminal. Specifically, the type II PMI information may include sub-band level amplitude information and phase information.

[0042] The processing unit 420 performs channel reconstruction, interpolation, and denoising processing based on the precoding matrix indication information of the first granularity through a first sub-network to obtain a second channel, and performs downlink precoding on the channel of the second granularity, wherein the second granularity is finer than the first granularity. Specifically, the first sub-network may be a first sub-neural network.

[0043] According to one example of this disclosure, the input dimension of the first sub-network may be higher than the output dimension of the first sub-network. In other words, the first sub-network adopts a high-dimensional input and low-dimensional output design. Due to the use of high-dimensional input, the original information from the pre-encoded matrix indication information can be preserved, and due to the use of low-dimensional output, dimensionality reduction during network processing can reduce network complexity and training difficulty.

[0044] Specifically, the input to the first sub-network may be precoding matrix indication information from the terminal or preprocessed precoding matrix indication information, and the first sub-network may reconstruct the input from the precoding matrix indication information. According to an example of this disclosure, the first sub-network may weightedly combine the amplitude and phase of the input data. For example, the precoding matrix indication information may include amplitude information and phase information. Further, the amplitude information may include broadband beam information, broadband amplitude information, and sub-band amplitude information of each beam through which the base station 400 communicates with the terminal, as well as sub-band phase information of each sub-band. The first sub-network may combine the broadband beam information, broadband amplitude information, and sub-band amplitude information of each beam to obtain a channel amplitude matrix. Specifically, the first sub-network may obtain the channel amplitude matrix in each polarization direction according to the element polarization direction of the antenna array. Furthermore, the first sub-network may obtain the real and imaginary parts of the matrix based on the broadband beam information of each beam and the phase information of each sub-band. Similarly, the first sub-network may obtain the real and imaginary parts of the matrix in each polarization direction according to the polarization direction of the beam. Then, the first sub-network can multiply the channel amplitude matrix in each polarization direction with the real part matrix and the imaginary part matrix respectively to obtain the beam-frequency domain channel matrix (also simply referred to as "beam-frequency domain channel").

[0045] For example, base station 400 communicates with the terminal using beams in two polarization directions. The first sub-network can obtain the channel amplitude matrices A1 and A2 in the two polarization directions, and the channel phase matrices (i.e., the real and imaginary parts) Pr1, Pr2, Pi1, and Pi2 in the two polarization directions. The first sub-network can obtain the beam-frequency domain channel matrices Hr1, Hr2, Hi1, and Hi2 using the following formula (2):

[0046] Hr1 = A1 * Pr1, Hr2 = A2 * Pr2;

[0047] Hi1=A1*Pi1, Hi2=A2*Pi2…… (2)

[0048] In formula (2), "*" indicates the multiplication of corresponding matrix elements.

[0049] Next, the first sub-network can perform a Fourier transform on the beam-frequency domain channel to transform it into a beam-delay domain channel. Furthermore, since the channel delay component is mainly concentrated at the beginning of the delay domain channel matrix in the beam-delay domain, the first sub-network can truncate the delay domain channel to reduce the dimensionality of the output.

[0050] According to one example of this disclosure, the first sub-network may include a fully connected layer (dense layer) and is used to reconstruct the input precoded matrix indication information. Specifically, the fully connected layer weightedly combines the amplitude and phase of the input data to obtain a beam-frequency domain channel, converts the beam-frequency domain channel into a beam-delay domain channel, and truncates the beam-delay domain channel to reduce the dimensionality of the network output. Alternatively, since the operation required in the weighted combining process is only the element-wise multiplication of the individual matrices, the corresponding fully connected layer can be replaced with a partially connected layer that only connects the elements that need to be directly multiplied.

[0051] According to another example of this disclosure, the first sub-network may further include one or more super-resolution networks for interpolation and denoising. The one or more super-resolution networks may be positioned before or after the aforementioned fully connected or partially connected layers. Furthermore, the aforementioned fully connected or partially connected layers may be positioned between multiple super-resolution networks. The above has been combined with... Figure 3 The examples shown provide a detailed description of super-resolution networks according to embodiments of this disclosure, and therefore will not be repeated here.

[0052] Figures 14A-14C This is a schematic diagram illustrating a first sub-network structure according to an embodiment of the present disclosure. For example, as... Figure 14AAs shown, a super-resolution network can be placed before the fully connected layer or partially connected layer to interpolate and denoise the precoding matrix indication information from the terminal. The interpolated and denoised data is then input into the fully connected layer or partially connected layer for channel reconstruction and dimensionality reduction. For example, as... Figure 14B As shown, the precoding matrix indication information from the terminal can first be input into a fully connected layer or a partially connected layer for channel reconstruction and dimensionality reduction, and then the resulting channel can be input into a super-resolution network for interpolation and denoising. For example, ... Figure 14C As shown, precoding matrix indication information from the terminal can be input into the first super-resolution network for denoising. Then, the denoised data is input into a fully connected layer or a partially connected layer for channel reconstruction and dimensionality reduction. Finally, the dimensionality-reduced channel is input into the second super-resolution network for interpolation.

[0053] Alternatively, according to another example of this disclosure, the processing unit 400 may also perform at least one of temporal channel estimation enhancement and temporal prediction on multiple second channels obtained from precoding matrix indication information transmitted multiple times from the same terminal via a second sub-network. For example, the first sub-network may process precoding matrix indication information transmitted once by the terminal (e.g., transmitted within a single time slot) and perform channel reconstruction, interpolation, and denoising based on the precoding matrix indication information transmitted once by the terminal. The second sub-network includes at least one of a recurrent neural network (RNN) and a long and short-term memory (LSTM) network, and the second sub-network may input precoding matrix indication information transmitted multiple times by the same terminal into the RNN / LSTM network to achieve at least one of temporal channel estimation enhancement and temporal prediction.

[0054] The following will combine Figure 5 According to one embodiment of this disclosure, the channel state information reference information transmitted by the base station will be described. The following description, in conjunction with... Figure 5 The described channel state information reference information can be applied to the combination Figure 3 and Figure 4 The base station described.

[0055] Figure 5 This is a schematic block diagram illustrating a base station according to another embodiment of the present disclosure. Figure 5 As shown, a base station 500 according to another embodiment of this disclosure may include a transmitting unit 510 and a receiving unit 520. In addition to the transmitting and receiving units, the base station 500 may also include other components; however, since these components are not relevant to the content of this disclosure, their illustrations and descriptions are omitted here.

[0056] like Figure 5 As shown, the transmitting unit 510 transmits first channel state information reference information of a first density to the terminal. The first density can be a high density. Furthermore, according to an example of this disclosure, the transmitting unit 510 transmits the first channel state information reference information at a first density over the entire communication bandwidth of the base station 500. In other words, the transmitting unit 510 can transmit high-density first channel state information reference information over the entire communication bandwidth. For example, the resource blocks or resource elements occupied by ports 1-12 can be used to transmit the first channel state information reference information.

[0057] According to one embodiment of this disclosure, the first channel state information reference information can occupy all subcarriers in the frequency domain, that is, the density of the first channel state information reference information can reach 12 resource elements (REs) per resource block (RB) per port, and the reference signal for each port uses one OFDM symbol. Furthermore, multiple ports can be multiplexed on the same OFDM symbol using cyclic shift, or transmitted on different OFDM symbols using TDM. Alternatively, the density of the first channel state information reference information can reach 6 resource elements (REs) per resource block (RB) per port, and every two ports can be multiplexed in the frequency domain in an interleaved comb configuration, occupying one OFDM symbol. Similarly, multiplexing can also be performed using cyclic shift and TDM.

[0058] According to another disclosed example, the transmitting unit 510 also transmits channel state information reference information configuration information indicating the first channel state information reference information, and uses the resources indicated by the channel state information reference information configuration information to transmit the first channel state information reference information during the training data acquisition period. For example, the training data acquisition period may include multiple time slots.

[0059] Then, the receiving unit 520 receives the first feedback information regarding the first channel state information reference information sent by the terminal. According to one example of this disclosure, the terminal may be at least one of a user equipment and a data acquisition device. When the terminal is a data acquisition device, the data acquisition device can send the first feedback information regarding the first channel state information reference information to the base station 500 through a dedicated interface. On the other hand, when the terminal is a user equipment (UE), the sending unit 510 may also send control signaling to the UE to schedule the UE to send the first feedback information regarding the first channel state information reference information to the base station 500 through a data channel.

[0060] For example, the first feedback information may include the terminal performing a first channel estimation based on high-density first channel state information reference information to obtain first channel response information (hereinafter referred to as "channel response dataset"), and the terminal determining precoding matrix indication information based on channel state information reference information after downsampling the first channel state information reference information (hereinafter referred to as "PMI training dataset").

[0061] According to one example of this disclosure, Figure 5 The base station shown may further include a processing unit 530. The processing unit 530 can process the above-mentioned combinations based on the first feedback information. Figure 3 and Figure 4 The super-resolution network, first sub-network, second sub-network, and other neural networks in the described base station are trained so that the neural networks can perform high-precision channel estimation based on the state reference signal fed back by the terminal, that is, perform channel reconstruction, denoising, and interpolation.

[0062] The processing unit 530 can train the neural network using the PMI training dataset and the corresponding channel response dataset.

[0063] Figure 6 This is a schematic diagram illustrating an example of training a neural network using a PMI training dataset and a corresponding channel response dataset, according to this disclosure. Figure 6 As shown, the base station's processing unit uses the PMI training dataset as the first input to the neural network to perform channel reconstruction, denoising, and interpolation based on this first input. The PMI training dataset can be used to simulate the precoded matrix indication information sent by the terminal during actual deployment. On the other hand, the base station's processing unit can use a high-density channel response dataset of the first channel state information reference information as the second input to the neural network for network optimization, thereby, for example, providing the network with a target channel response for a specific PMI training dataset.

[0064] Base station 500 can be, for example, a dedicated base station for neural network training. After training is completed, the dedicated base station 500 can provide the trained neural network to the base station in the communication network that actually communicates with the UE, so that the trained neural network can be used for channel reconstruction, denoising and / or interpolation processing during actual communication.

[0065] For example, after training, base station 500 can be used to communicate with the UE in practice. During actual deployment, transmitting unit 510 sends second channel state information reference information with a second density to the UE. Receiving unit 520 receives second feedback information from the UE regarding the second channel state information reference information. The second channel state information reference information can be existing channel state information reference information used for channel estimation in actual communication, and the first density is greater than the second density. That is, the second channel state information reference information is sparser than the first channel state information reference information. Processing unit 530 can use a pre-trained network to perform channel reconstruction, denoising, and interpolation based on the second channel state information reference information to obtain a high-precision channel for downlink precoding in actual communication. Furthermore, during actual deployment, processing unit 530 can perform operations similar to those described above for processing units 320 and 420, which will not be repeated here.

[0066] The following will combine Figure 7 According to one embodiment of this disclosure, and Figure 5 The terminal corresponding to the base station mentioned above will be described. Figure 7 This is a schematic block diagram illustrating a terminal according to an embodiment of the present disclosure. Figure 7 As shown, a terminal 700 according to another embodiment of this disclosure may include a receiving unit 710, a processing unit 720, and a transmitting unit 730. In addition to the transmitting unit, receiving unit, and processing unit, the terminal 700 may also include other components; however, since these components are not relevant to the content of this disclosure, their illustrations and descriptions are omitted here.

[0067] like Figure 7 As shown, the receiving unit 710 receives first channel state information reference information of a first density. As described above, the first channel state information reference information can be used to train the neural network of the base station. According to an example of this disclosure, the first channel state information reference information may be transmitted over the entire communication bandwidth of the base station, and the first channel state information reference information has a higher density than the channel state information reference information used for channel measurement during the actual deployment phase.

[0068] The processing unit 720 performs a first channel estimation based on the first channel state information reference information to obtain the first channel response information, performs downsampling processing on the first channel state information reference information, and determines the precoding matrix indication information based on the downsampled channel state information reference information. Figure 8 This is a schematic diagram illustrating, according to an example of the present disclosure, the processing unit 720 processes the first channel state information reference information. For example... Figure 8As shown, on one hand, the processing unit performs a first channel estimation on the high-density first channel state information reference information to obtain high-precision first channel response information. On the other hand, the processing unit downsamples the first channel state information reference information to simulate the channel state information reference information density received during the actual deployment phase and obtains a low-density channel. Then, the processing unit calculates the PMI feedback amount based on the downsampled channel state information reference information to determine the precoding matrix indication information.

[0069] return Figure 7 The transmitting unit 730 sends the first channel response information and precoding matrix indication information to the base station. The base station can then train its neural network based on the received first channel response information and precoding matrix indication information. According to an example of this disclosure, the terminal 700 can be at least one of a user equipment (UE) and a data acquisition device. When the terminal 700 is a data acquisition device used to train the base station's neural network, the transmitting unit 730 can send the first channel response information and precoding matrix indication information to the base station offline or via an air interface. When the terminal 700 is a UE, the transmitting unit 730 can use a precoding matrix indication feedback channel to send the precoding matrix indication information to the base station, and can transmit the first channel response information to the base station after modulation and coding using an uplink data channel or an uplink control channel. The base station can train its neural network using the first channel response information and precoding matrix indication information, performing channel reconstruction, interpolation, denoising, etc.

[0070] Below, refer to Figure 9 This describes a channel processing method according to embodiments of the present disclosure. Figure 9 This is a flowchart of a channel processing method 900 according to an embodiment of the present disclosure. Since the steps of the channel processing method 900 correspond to the operation of the base station 300 described above with reference to the figures, detailed descriptions of the same content are omitted here for simplicity.

[0071] like Figure 9As shown, in step S901, precoding matrix indication information of the first granularity is received from the terminal. In step S902, channel reconstruction is performed according to the precoding matrix indication information to obtain a first channel, wherein the first channel has a first granularity. For example, in step S902, the amplitude and phase information in the received PMI information can be used to perform amplitude and phase weighting on the spatial (also called "beam domain")-frequency domain channel codewords of multiple beams, and the weighted vectors are merged to obtain a first channel with sub-band level. The sub-band level first channel obtained after merging can be represented in the form of a spatial-frequency domain channel matrix, wherein the spatial value of the spatial-frequency domain channel matrix can be the number of antennas for communication between the base station and the terminal, and the frequency value can be determined according to the number of sub-bands for communication between the base station and the terminal.

[0072] Then, in step S903, a super-resolution network is used to interpolate and denoise the channel of the first granularity to obtain a second channel, wherein the second channel has a second granularity and the second granularity is finer than the first granularity.

[0073] According to one example of this disclosure, Figure 9 The method shown may further include preprocessing the first channel before inputting it into the super-resolution network to facilitate subsequent super-resolution network operations. For example, if the first channel is a spatial-frequency domain channel, a Fourier transform can be performed on it to convert it into a beam-delay domain channel. Furthermore, since the channel delay component is mainly concentrated at the beginning of the delay domain channel matrix in the beam-delay domain, the delay domain channel can be truncated, retaining the beginning, and the truncated data can be divided into two channels, real and imaginary, as input to the super-resolution network. By transforming the first channel to be processed to the delay domain and truncating the data, the computational complexity of the super-resolution network can be reduced.

[0074] Furthermore, to further simplify the operation of super-resolution networks... Figure 9 The method shown may also include pre-interpolating the first channel using existing interpolation methods such as zero-filling or linear interpolation, nearest neighbor interpolation, etc., before transforming it to the beam-delay domain channel, to obtain a channel with the desired frequency domain accuracy (e.g., RB level or subcarrier level).

[0075] Finally, in step S904, downlink precoding is performed on the second channel for transmission to the terminal. (In combination with...) Figure 9 In the described channel processing method, after performing conventional channel reconstruction based on the precoding matrix indication information, a super-resolution network is used to interpolate and denoise the first-granularity channel to obtain a second channel with finer granularity for downlink precoding.

[0076] According to another embodiment of this disclosure, in addition to interpolation and denoising processing, a neural network can also be used to perform channel reconstruction based on precoding matrix indication information of the first granularity received from the terminal. Referring hereafter... Figure 10 To describe a channel processing method according to another embodiment of the present disclosure. Figure 10 This is a flowchart of a channel processing method 1000 according to another embodiment of the present disclosure. Since the steps of the channel processing method 1000 correspond to the operation of the base station 400 described above with reference to the figures, detailed descriptions of the same content are omitted here for simplicity.

[0077] like Figure 10 As shown, in step S1001, precoding matrix indication information of the first granularity is received from the terminal. For example, type II PMI information at the sub-band level can be received from the terminal. Specifically, the type II PMI information may include amplitude information and phase information at the sub-band level.

[0078] In step S1002, a second channel is obtained by performing channel reconstruction, interpolation, and denoising processing based on the precoding matrix indication information of the first granularity through a first sub-network, and downlink precoding is performed on the channel of the second granularity, wherein the second granularity is finer than the first granularity. Specifically, the first sub-network may be a first sub-neural network.

[0079] According to one example of this disclosure, the input dimension of the first sub-network may be higher than the output dimension of the first sub-network. In other words, the first sub-network adopts a high-dimensional input and low-dimensional output design. Due to the use of high-dimensional input, the original information from the pre-encoded matrix indication information can be preserved, and due to the use of low-dimensional output, dimensionality reduction during network processing can reduce network complexity and training difficulty.

[0080] Specifically, the input to the first sub-network may be precoding matrix indication information from the terminal or preprocessed precoding matrix indication information, and the first sub-network may reconstruct the input from the precoding matrix indication information. According to an example of this disclosure, in step S1002, the first sub-network may weightedly combine the amplitude and phase of the input data. For example, the precoding matrix indication information may include amplitude information and phase information. Further, the amplitude information may include broadband beam information, broadband amplitude information, and sub-band amplitude information of each beam through which the base station communicates with the terminal, as well as sub-band phase information of each sub-band. In step S1002, the first sub-network may combine the broadband beam information, broadband amplitude information, and sub-band amplitude information of each beam to obtain a channel amplitude matrix. Specifically, the first sub-network may obtain the channel amplitude matrix in each polarization direction according to the beam polarization direction. Furthermore, in step S1002, the first sub-network may also obtain the real part matrix and the imaginary part matrix based on the broadband beam information of each beam and the phase information of each sub-band. Similarly, the first sub-network can obtain the real and imaginary matrices in each polarization direction according to the polarization direction of the beam. Then, the first sub-network can multiply the channel amplitude matrix in each polarization direction with the real and imaginary matrices respectively to obtain the beam-frequency domain channel matrix (also referred to as "beam-frequency domain channel").

[0081] Next, in step S1002, the first sub-network can perform a Fourier transform on the beam-frequency domain channel to convert it into a beam-delay domain channel. Furthermore, since the channel delay component is mainly concentrated at the beginning of the delay domain channel matrix in the beam-delay domain, the first sub-network can truncate the delay domain channel to reduce the dimensionality of the output.

[0082] According to one example of this disclosure, the first sub-network may include a fully connected layer (dense layer) and is used to reconstruct the input precoded matrix indication information. Specifically, the fully connected layer weightedly combines the amplitude and phase of the input data to obtain a beam-frequency domain channel, converts the beam-frequency domain channel into a beam-delay domain channel, and truncates the beam-delay domain channel to reduce the dimensionality of the network output. Alternatively, since the operation required in the weighted combining process is only the element-wise multiplication of the individual matrices, the corresponding fully connected layer can be replaced with a partially connected layer that only connects the elements that need to be directly multiplied.

[0083] According to another example of this disclosure, the first sub-network may further include one or more super-resolution networks for interpolation and denoising. The one or more super-resolution networks may be positioned before or after the aforementioned fully connected or partially connected layers. Furthermore, the aforementioned fully connected or partially connected layers may be positioned between multiple super-resolution networks.

[0084] For example, a super-resolution network can be set up before the fully connected layer or partially connected layer to interpolate and denoise the precoding matrix indication information from the terminal. The interpolated and denoised data is then input into the fully connected layer or partially connected layer for channel reconstruction and dimensionality reduction. Alternatively, the precoding matrix indication information from the terminal can first be input into the fully connected layer or partially connected layer for channel reconstruction and dimensionality reduction, and then the resulting channel can be input into the super-resolution network for interpolation and denoising. Yet another example is that the precoding matrix indication information from the terminal can be input into a first super-resolution network for denoising. The denoised data is then input into the fully connected layer or partially connected layer for channel reconstruction and dimensionality reduction. Finally, the dimensionality-reduced channel is input into a second super-resolution network for interpolation.

[0085] Optionally, Figure 10 The method shown may further include performing at least one of temporal channel estimation enhancement and temporal prediction on multiple second channels obtained from precoding matrix indication information transmitted multiple times from the same terminal via a second sub-network. For example, the first sub-network may process precoding matrix indication information transmitted once by the terminal (e.g., transmitted within a single time slot) and perform channel reconstruction, interpolation, and denoising based on the precoding matrix indication information transmitted once by the terminal. The second sub-network includes at least one of an RNN and an LSTM network, and the second sub-network may input precoding matrix indication information transmitted multiple times by the same terminal into the RNN / LSTM network to achieve at least one of temporal channel estimation enhancement and temporal prediction.

[0086] The following will combine Figure 11 A reference signal transmission method according to an embodiment of the present disclosure will be described. Figure 11 This is a flowchart of a reference signal transmission method 1100 according to an embodiment of the present disclosure. Since the steps of the reference signal transmission method 1100 correspond to the operation of the base station 500 described above with reference to the figures, detailed descriptions of the same content are omitted here for simplicity.

[0087] like Figure 11 As shown, in step S1101, first channel state information reference information of a first density is sent to the terminal. The first density can be a high density. Furthermore, according to an example of this disclosure, in step S1101, the first channel state information reference information is sent at a first density across the entire communication bandwidth of the base station. In other words, in step S1101, high-density first channel state information reference information can be sent across the entire communication bandwidth.

[0088] According to another disclosed example, method 1100 may further include sending channel state information reference information configuration information for indicating the first channel state information reference information, and using the resources indicated by the channel state information reference information configuration information to send the first channel state information reference information during the training data acquisition period. For example, the training data acquisition period may include multiple time slots.

[0089] Then, in step S1102, the first feedback information regarding the first channel state information reference information sent by the terminal is received. According to an example of this disclosure, the terminal may be at least one of a user equipment (UE) and a data acquisition device. When the terminal is a data acquisition device, the data acquisition device can send the first feedback information regarding the first channel state information reference information to the base station through a dedicated interface. On the other hand, when the terminal is a user equipment (UE), method 1100 may further include sending control signaling to the UE to schedule the UE to send the first feedback information regarding the first channel state information reference information to the base station 500 through a data channel.

[0090] For example, the first feedback information may include the terminal performing a first channel estimation based on high-density first channel state information reference information to obtain first channel response information (hereinafter referred to as "channel response dataset"), and the terminal determining precoding matrix indication information based on channel state information reference information after downsampling the first channel state information reference information (hereinafter referred to as "PMI training dataset"). The processing unit 530 may use the PMI training dataset and the corresponding channel response dataset to train the neural network.

[0091] Furthermore, according to another example of this disclosure, method 1100 may also include adjusting the above combination based on first feedback information. Figure 3 and Figure 4 The super-resolution network, first sub-network, second sub-network, and other neural networks in the described base station are trained so that the neural networks can perform high-precision channel estimation based on the state reference signal fed back by the terminal, that is, perform channel reconstruction, denoising, and interpolation.

[0092] The base station can be, for example, a dedicated base station for training the neural network. After training is completed, the dedicated base station can provide the trained neural network to the base station in the communication network that actually communicates with the UE, so that the trained neural network can be used for operations such as channel reconstruction, denoising and / or interpolation during actual communication.

[0093] For example, after training, the base station can be used to communicate with the UE. In this case, during actual deployment, such as... Figure 11As shown, in step S1103, second channel state information reference information with a second density is sent to the UE. And in step S1104, second feedback information from the UE regarding the second channel state information reference information is received. The second channel state information reference information can be existing channel state information reference information used for channel estimation in actual communication, and the first density is greater than the second density. That is, the second channel state information reference information is sparser than the first channel state information reference information. Then, in step S1105, a pre-trained network is used to perform channel reconstruction, denoising, and interpolation based on the second channel state information reference information to obtain a high-precision channel that will be used for downlink precoding in actual communication.

[0094] The following will combine Figure 12 An information transmission method executed by a terminal according to an embodiment of the present disclosure will be described. Figure 12 This is a flowchart of an information transmission method 1200 according to an embodiment of the present disclosure. Since the steps of the reference signal transmission method 1200 correspond to the operation of the terminal 700 described above with reference to the figures, detailed descriptions of the same content are omitted here for simplicity.

[0095] like Figure 12 As shown, in step S1201, first channel state information reference information with a first density is received. As described above, the first channel state information reference information can be used to train the neural network of the base station. According to an example of this disclosure, the first channel state information reference information may be transmitted over the entire communication bandwidth of the base station, and the first channel state information reference information has a higher density than the channel state information reference information used for channel measurement during the actual deployment phase.

[0096] In step S1202, a first channel estimation is performed based on the first channel state information reference information to obtain the first channel response information. Then, in step S1203, the first channel state information reference information is downsampled, and precoding matrix indication information is determined based on the downsampled channel state information reference information. Although in Figure 12 The example given is that step S1202 is executed first, followed by step S1203; however, this disclosure is not limited to this. For example, step S1203 may be executed first, followed by step S1201, or steps S1202 and S1203 may be executed simultaneously.

[0097] Then, in step S1204, the first channel response information and precoding matrix indication information are sent to the base station. Thus, the base station can train its neural network based on the received first channel response information and precoding matrix indication information.

[0098] <Hardware Structure>

[0099] Furthermore, the block diagrams used in the above description of the embodiments illustrate blocks based on function. These functional blocks (structural units) are implemented through any combination of hardware and / or software. Moreover, the means of implementing each functional block are not particularly limited. That is, each functional block can be implemented using a single device that is physically and / or logically combined, or it can be implemented using multiple devices by directly and / or indirectly (e.g., via wired and / or wireless) connecting two or more physically and / or logically separate devices.

[0100] For example, a first network element in one embodiment of this disclosure can function as a computer performing the wireless communication method of this disclosure. Figure 13 This is a schematic diagram of the hardware structure of the device 1300 (base station, terminal) according to an embodiment of the present disclosure. The device 1300 (base station, terminal) described above can be configured as a computer device that physically includes a processor 1310, a memory 1320, a storage device 1330, a communication device 1340, an input device 1350, an output device 1360, a bus 1370, etc.

[0101] Additionally, in the following description, the word "device" can be replaced with circuit, device, unit, etc. The hardware structure of the first network element may include one or more of the devices shown in the figure, or may not include some of the devices.

[0102] For example, only one processor 1310 is shown, but there can be multiple processors. Furthermore, processing can be performed by a single processor, or by more than one processor simultaneously, sequentially, or using other methods. Additionally, processor 1310 can be mounted on more than one chip.

[0103] The functions of device 1300 are implemented, for example, by reading the specified software (program) into hardware such as processor 1310 and memory 1320, thereby enabling processor 1310 to perform calculations, control the communication performed by communication device 1340, and control the reading and / or writing of data in memory 1320 and storage 330.

[0104] Processor 1310 enables the operating system to function, thereby controlling the computer as a whole. Processor 1310 may be a central processing unit (CPU) that includes interfaces with peripheral devices, control units, arithmetic units, registers, etc. For example, the aforementioned processing units can be implemented by processor 1310.

[0105] Furthermore, the processor 1310 reads programs (program code), software modules, data, etc., from the memory 1330 and / or communication device 1340 into the memory 1320, and performs various processes accordingly. As a program, a program that causes the computer to perform at least a portion of the actions described in the above embodiments can be employed. For example, the processing unit of the first network element can be implemented by a control program stored in the memory 1320 and operated by the processor 1310; similarly, other functional blocks can be implemented.

[0106] Memory 1320 is a computer-readable recording medium, and may be constituted, for example, at least one of read-only memory (ROM), erasable programmable read-only memory (EPROM), electrically programmable read-only memory (EEPROM), random access memory (RAM), or other suitable storage media. Memory 1320 may also be referred to as a register, cache, main memory (main storage device), etc. Memory 1320 may store executable programs (program code), software modules, etc., for implementing the methods involved in one embodiment of this disclosure.

[0107] The memory 1330 is a computer-readable recording medium, and may be constituted by at least one of the following: a flexible disk, a floppy disk, a magneto-optical disk (e.g., a CD-ROM (Compact Disc ROM), a Digital Universal Optical Disc, a Blu-ray disc), a removable disk, a hard disk, a smart card, a flash memory device (e.g., a card, a stick, a key driver), a magnetic stripe, a database, a server, or other suitable storage media. The memory 1330 may also be referred to as an auxiliary storage device.

[0108] Communication device 1340 is hardware (transmitting and receiving device) used for communication between computers via wired and / or wireless networks, and is also referred to as a network device, network controller, network interface card (NIC), communication module, etc. To implement, for example, frequency division duplex (FDD) and / or time division duplex (TDD), communication device 1340 may include high-frequency switches, duplexers, filters, frequency synthesizers, etc. For example, the aforementioned transmitting unit and receiving unit can be implemented using communication device 1340.

[0109] Input device 1350 is an input device that accepts input from external sources (e.g., keyboard, mouse, microphone, switch, button, sensor, etc.). Output device 1360 is an output device that performs output to external sources (e.g., display, speaker, light-emitting diode (LED) lamp, etc.). Alternatively, input device 1350 and output device 1360 can also be integrated into a single structure (e.g., a touch panel).

[0110] Furthermore, the processor 1310, memory 1320, and other devices are connected via a bus 1370 for communication of information. The bus 1370 can consist of a single bus or different buses between devices.

[0111] Furthermore, electronic devices may include hardware such as microprocessors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), programmable logic devices (PLDs), and field-programmable gate arrays (FPGAs), which can be used to implement some or all of the functional blocks. For example, processor 710 can be installed using at least one of these hardware components.

[0112] (Modified Example)

[0113] Furthermore, the terms used in this specification and / or those necessary for understanding this specification may be used interchangeably with terms that have the same or similar meanings. For example, a channel and / or symbol may also be a signal (signaling). Additionally, a signal may also be a message. A reference signal may also be simply referred to as RS (Reference Signal), and depending on the applicable standard, may also be called a pilot, pilot signal, etc. Furthermore, a component carrier (CC) may also be referred to as a cell, frequency carrier, carrier frequency, etc.

[0114] Furthermore, the information and parameters described in this specification may be expressed in absolute values, relative values ​​to specified values, or other corresponding information. For example, wireless resources may be indicated by a specified index. Moreover, the formulas used to apply these parameters may differ from those explicitly disclosed in this specification.

[0115] The names used for parameters, etc., in this specification are not limiting in any way. For example, various channels (Physical Uplink Control Channel (PUCCH), Physical Downlink Control Channel (PDCCH), etc.) and information elements can be identified by any appropriate name, and therefore the various names assigned to these various channels and information elements are not limiting in any way.

[0116] The information, signals, etc., described in this specification can be represented using any of a wide variety of different technologies. For example, data, commands, instructions, information, signals, bits, symbols, chips, etc., that may be mentioned in all of the above descriptions can be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, light fields or photons, or any combination thereof.

[0117] Furthermore, information and signals can be output from upper layers to lower layers and / or from lower layers to upper layers. Information and signals can be input or output via multiple network nodes.

[0118] Input or output information and signals can be stored in a specific location (such as memory) or managed through a management table. Input or output information and signals can be overwritten, updated, or supplemented. Output information and signals can be deleted. Input information and signals can be sent to other devices.

[0119] The notification of information is not limited to the methods / implementations described in this specification, and may also be carried out by other methods. For example, the notification of information may be implemented by physical layer signaling (e.g., downlink control information (DCI), uplink control information (UCI)), upper layer signaling (e.g., radio resource control (RRC) signaling, broadcast information (master information block (MIB), system information block (SIB) etc.), media access control (MAC) signaling), other signals, or combinations thereof.

[0120] In addition, physical layer signaling can also be referred to as L1 / L2 (Layer 1 / Layer 2) control information (L1 / L2 control signals), L1 control information (L1 control signals), etc. Furthermore, RRC signaling can also be referred to as RRC messages, such as RRC connection setup messages, RRC connection reconfiguration messages, etc. Additionally, MAC signaling can be communicated, for example, through a MAC control unit (MAC CE (Control Element)).

[0121] Furthermore, notification of specified information (e.g., notification of “for X”) is not limited to being made explicitly, but can also be made implicitly (e.g., by not making notification of the specified information, or by notifying other information).

[0122] The determination can be made by a value represented by 1 bit (0 or 1), by a true or false Boolean value, or by a numerical comparison (e.g., a comparison with a specified value).

[0123] Whether it is called software, firmware, middleware, microcode, hardware description language, or any other name, software should be broadly interpreted as commands, command sets, code, code segments, program code, programs, subroutines, software modules, application programs, software applications, software packages, routines, subroutines, objects, executable files, execution threads, steps, functions, etc.

[0124] Furthermore, software, commands, information, etc., can be sent or received via a transmission medium. For example, when software is sent from a website, server, or other remote resource using wired technologies (coaxial cable, optical fiber, twisted pair, digital subscriber line (DSL), etc.) and / or wireless technologies (infrared, microwave, etc.), these wired and / or wireless technologies are included within the definition of transmission medium.

[0125] The terms “system” and “network” used in this manual are interchangeable.

[0126] In this manual, the terms "base station (BS)," "wireless base station," "eNB," "gNB," "cell," "sector," "cell group," "carrier," and "component carrier" are used interchangeably. Base stations are sometimes also referred to as fixed stations, NodeBs, eNodeBs (eNBs), access points, transmitting points, receiving points, femtocells, small cells, etc.

[0127] A base station can accommodate one or more (e.g., three) cells (also called sectors). When a base station accommodates multiple cells, the entire coverage area of ​​the base station can be divided into multiple smaller areas, each of which can also provide communication services through a base station subsystem (e.g., an indoor small cell (remote radio head (RRH))). Terms such as "cell" or "sector" refer to a portion or the entire coverage area of ​​the base station and / or base station subsystem that provides communication services within that coverage area.

[0128] In this specification, the terms "Mobile Station (MS)," "user terminal," "User Equipment (UE)," and "terminal" are used interchangeably. A mobile station is sometimes also referred to by those skilled in the art as a user station, mobile unit, user unit, radio unit, remote unit, mobile device, radio device, wireless communication device, remote device, mobile user station, access terminal, mobile terminal, wireless terminal, remote terminal, handheld device, user agent, mobile client, client, or several other appropriate terms.

[0129] Furthermore, the wireless base station in this specification can also be replaced by a user terminal. For example, various methods / implementations of this disclosure can also be applied to a structure that replaces the communication between the wireless base station and the user terminal with communication between multiple user terminals (D2D, Device-to-Device). In this case, the functions of the electronic device described above can be regarded as the functions of the user terminal. In addition, terms such as "uplink" and "downlink" can also be replaced with "side". For example, the uplink channel can also be replaced with the side channel.

[0130] Similarly, the user terminal in this specification can also be replaced by a wireless base station. In this case, the functions of the user terminal described above can be regarded as the functions of the first communication device or the second communication device.

[0131] In this specification, specific actions performed via a base station may sometimes also be performed via its upper node, depending on the circumstances. Clearly, in a network consisting of one or more network nodes with a base station, various actions performed for communication with a terminal can be performed via the base station, one or more network nodes other than the base station (considering, but not limited to, Mobility Management Entities (MMEs), Serving Gateways (S-GWs), etc.), or combinations thereof.

[0132] The various methods / implementations described in this specification can be used individually or in combination, and can be switched during execution. Furthermore, the processing steps, sequences, flowcharts, etc., of the various methods / implementations described in this specification can be rearranged as long as there are no contradictions. For example, regarding the methods described in this specification, various step units are given in an exemplary order, but the method is not limited to the specific order given.

[0133] The methods / implementations described in this specification can be applied to systems utilizing Long Term Evolution (LTE), LTE-Advanced (LTE-A), LTE-Beyond (LTE-B), Super 3G, IMT-Advanced, 4G, 5G, Future Radio Access (FRA), New-RAT (Radio Access Technology), New Radio (NR), New Radio Access (NX), Future Generation Radio Access (FX), Global System for Mobile Communications (GSM), CDMA3000, Ultra Mobile Broadband (UMB), IEEE 920.11 (Wi-Fi), and IEEE... Systems based on and / or extended from WiMAX (registered trademark), IEEE 920.16, Ultra-Wideband (UWB), Bluetooth (registered trademark), other suitable wireless communication methods, and / or next-generation systems based on them.

[0134] The use of the word "based on" in this specification, unless explicitly stated elsewhere, does not imply "based on only". In other words, the use of "based on" refers to both "based on only" and "based on at least".

[0135] Any reference to units using the names "first," "second," etc., as used in this specification is not intended to fully define the number or order of these units. These names may be used in this specification as a convenient method of distinguishing two or more units. Therefore, reference to a first unit and a second unit does not imply that only two units may be used, or that the first unit must take precedence over the second unit in some form.

[0136] The term "determining" as used in this specification sometimes encompasses a variety of actions. For example, "determining" can refer to actions such as calculating, computing, processing, deriving, investigating, looking up (e.g., searching in tables, databases, or other data structures), and ascertaining. Furthermore, "determining" can also refer to actions such as receiving (e.g., receiving information), transmitting (e.g., sending information), inputting, outputting, and accessing (e.g., accessing data in memory). Additionally, "determining" can refer to actions such as resolving, selecting, choosing, establishing, and comparing. In other words, "determining" can encompass several actions.

[0137] As used in this specification, the terms "connected," "coupled," or any variations thereof refer to any direct or indirect connection or combination between two or more units, including situations where one or more intermediate units exist between two mutually "connected" or "coupled" units. The combination or connection between units can be physical, logical, or a combination of both. For example, "connected" can also be replaced by "accessed." In this specification, it can be understood that two units are "connected" or "coupled" by using one or more wires, cables, and / or printed electrical connections, and, as several non-limiting and non-exhaustive examples, by using electromagnetic energy with wavelengths in the radio frequency region, microwave region, and / or light (both visible and invisible light) region.

[0138] When the terms "including," "comprising," and variations thereof are used in this specification or claims, these terms are open-ended, just like the term "possess." Furthermore, the term "or" as used in this specification or claims is not an XOR expression.

[0139] The present disclosure has been described in detail above; however, it will be apparent to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented in modified and altered ways without departing from the spirit and scope defined by the claims. Therefore, the description herein is for illustrative purposes only and is not intended to be restrictive.

Claims

1. A base station, comprising: The receiving unit is configured to receive precoding matrix indication information of the first granularity from the terminal; The processing unit is configured to perform channel reconstruction based on the precoding matrix indication information to obtain a first channel, perform interpolation and denoising processing on the first channel using a super-resolution network to obtain a second channel, and perform downlink precoding on the second channel, wherein... The first channel has the first granularity. The second channel has a second granularity. The second particle size is finer than the first particle size. The super-resolution network is a trained neural network.

2. The base station as described in claim 1, wherein The first channel is a spatial-frequency channel; and The processing unit is further configured to transform the reconstructed first channel to the delay domain to obtain a delay domain channel, and to divide the delay domain channel into a real part and an imaginary part for input to the super-resolution network.

3. A base station, comprising: The receiving unit is configured to receive precoding matrix indication information of the first granularity from the terminal; The processing unit is configured to perform channel reconstruction, interpolation, and denoising processing based on the precoding matrix indication information at the first granularity through a first sub-network to obtain a second channel at the second granularity, and to perform downlink precoding on the second channel at the second granularity, wherein... The second particle size is finer than the first particle size. The first sub-network is a trained neural network.

4. The base station as described in claim 3, wherein The first sub-network performs amplitude and phase weighted merging according to the precoding matrix indication information to obtain a beam-frequency domain channel, transforms the beam-frequency domain channel into a beam-delay domain channel, and truncates the beam-delay domain channel.

5. The base station as described in claim 3 or 4, wherein The processing unit is further configured to perform at least one of time-domain channel estimation enhancement and time-domain prediction on multiple second channels obtained from precoding matrix indication information sent multiple times from the same terminal via a second sub-network.

6. A base station, comprising: The transmitting unit is configured to transmit first channel state information reference information of first density to the terminal. The receiving unit is configured to receive first feedback information from the terminal regarding the first channel state information reference information; in The first feedback information includes first channel response information obtained by the terminal performing first channel estimation based on the first channel state information reference information, and precoding matrix indication information determined by the terminal based on the downsampled first channel state information reference information. The base station also uses a trained neural network to perform channel reconstruction, interpolation, and denoising based on the first feedback information received from the terminal.

7. The base station as described in claim 6, wherein The transmitting unit transmits the first channel state information reference information over the entire communication bandwidth of the base station at the first density.

8. The base station as described in claim 6 or 7, wherein The transmitting unit is further configured to transmit channel state information reference information configuration information for indicating the first channel state information reference information; and The transmitting unit is further configured to trigger the measurement of the first channel state information reference information at a specific training data acquisition time.

9. A terminal, comprising: The receiving unit is configured to receive first channel state information reference information of the first density; The processing unit is configured to perform a first channel estimation based on the first channel state information reference information to obtain first channel response information, and to perform downsampling processing on the first channel state information reference information, and to determine precoding matrix indication information based on the downsampled channel state information reference information. as well as The transmitting unit is configured to transmit the first channel response information and precoding matrix indication information to the base station. The first channel response information and the precoding matrix indication information are used by the base station to perform channel reconstruction, interpolation, and denoising processing using a trained neural network.

10. The terminal as claimed in claim 9, wherein The terminal is at least one of a user equipment and a data acquisition device.

Images