Method, transmitter device and receiver device for channel information feedback

Abstract

A method for channel information feedback, a transmitting device, and a receiving device are disclosed, which helps to balance the feedback accuracy of channel information and the CSI feedback overhead. The method includes: the transmitting device receiving a reference signal transmitted by the receiving device; performing channel estimation based on the reference signal to obtain channel information between the transmitting device and the receiving device; performing feature decomposition on the channel information to obtain at least one first feature vector; encoding the at least one first feature vector using a neural network to obtain a target bit stream; and transmitting the target bit stream to the receiving device.

Description

Technical Field

[0001] This application relates to the field of communications, specifically to a method for channel information feedback, a transmitting device, and a receiving device. Background Technology

[0002] In New Radio (NR) systems, the Channel State Information Reference Signal (CSI-RS) feedback design primarily utilizes a codebook-based scheme to extract and feedback channel features. Specifically, after channel estimation at the transmitter, the precoding matrix that best matches the channel estimation structure is selected from a pre-defined precoding codebook based on specific optimization criteria. The index information of this precoding matrix is ​​then fed back to the receiver via an air interface feedback link for precoding. This mapping process from channel information to channel information in the precoding codebook is lossy in quantization, reducing the accuracy of the fed-back channel information and consequently lowering precoding performance. Furthermore, if the entire channel estimation information is fed back, the CSI feedback overhead is significant. Therefore, balancing the accuracy of channel information feedback with the CSI feedback overhead is a pressing issue that needs to be addressed. Summary of the Invention

[0003] This application provides a method for channel information feedback, a transmitting device, and a receiving device, which can balance the feedback accuracy of channel information and the CSI feedback overhead.

[0004] In a first aspect, a method for channel information feedback is provided, comprising: a transmitting device receiving a reference signal transmitted by a receiving device; performing channel estimation based on the reference signal to obtain channel information between the transmitting device and the receiving device; performing feature decomposition on the channel information to obtain at least one first feature vector; encoding the at least one first feature vector through a neural network to obtain a target bit stream; and transmitting the target bit stream to the receiving device.

[0005] Secondly, a method for channel information feedback is provided, comprising: a receiving device receiving a target bit stream transmitted by a transmitting device, the target bit stream being encoded by the transmitting device into at least one first feature vector, the at least one first feature vector being obtained by the transmitting device through feature decomposition of the channel estimation result; and decoding the target bit stream through a neural network to obtain at least one target feature vector.

[0006] Thirdly, a transmitting device is provided for executing the methods described in the first aspect or its various implementations.

[0007] Specifically, the transmitting device includes a functional module for performing the methods described in the first aspect or its various implementations.

[0008] Fourthly, a receiving device is provided for performing the methods in the second aspect or their respective implementations described above.

[0009] Specifically, the receiving device includes a functional module for performing the methods described in the second aspect or its various implementations.

[0010] Fifthly, a transmitting device is provided, including a processor and a memory. The memory is used to store a computer program, and the processor is used to call and run the computer program stored in the memory to perform the methods described in the first aspect or its various implementations.

[0011] Sixthly, a receiving device is provided, including a processor and a memory. The memory is used to store a computer program, and the processor is used to call and run the computer program stored in the memory to perform the methods in the second aspect or its implementations described above.

[0012] In a seventh aspect, a chip is provided for implementing the methods of any one of the first to second aspects or their respective implementations.

[0013] Specifically, the chip includes a processor for calling and running a computer program from memory, causing a device equipped with the device to perform the method as described in any of the first to second aspects above or in their respective implementations.

[0014] Eighthly, a computer-readable storage medium is provided for storing a computer program that causes a computer to perform the methods of any one of the first to second aspects or their respective implementations.

[0015] Ninthly, a computer program product is provided, including computer program instructions that cause a computer to perform the methods of any one of the first to second aspects or their respective implementations.

[0016] In a tenth aspect, a computer program is provided that, when run on a computer, causes the computer to perform the methods of any one of the first to second aspects or their respective implementations.

[0017] Through the above technical solution, the transmitting device obtains at least one feature vector by performing feature decomposition on the full channel information of the channel estimation, and further encodes the feature vector using a neural network to obtain the target bit stream, which is then sent to the receiving end. Correspondingly, the receiving end decodes the target bit stream to obtain the target feature vector. On the one hand, when transmitting channel information, only the target bit stream encoded by the feature vector obtained by feature decomposition of the full channel information needs to be sent, which helps to reduce CSI overhead. On the other hand, encoding the feature vector of the full channel information instead of directly encoding the full channel information takes into account the correlation characteristics between feature vectors, which helps to avoid compressing too much redundant information, reduce compression efficiency, and improve coding performance. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of a communication system architecture provided in an embodiment of this application.

[0019] Figure 2 This is a schematic interactive diagram of a channel information feedback method provided according to an embodiment of this application.

[0020] Figure 3 This is a system architecture diagram of a channel information feedback method according to an embodiment of this application.

[0021] Figure 4 This is a schematic diagram of a neural network structure.

[0022] Figure 5 This is a schematic structural diagram of an encoder according to an embodiment of this application.

[0023] Figure 6 This is a schematic structural diagram of a decoder according to an embodiment of this application.

[0024] Figure 7 yes Figure 6 A schematic diagram of the common modules in the system.

[0025] Figure 8 This is a schematic diagram of a recurrent neural network.

[0026] Figure 9 This is a schematic structural diagram of an encoder according to another embodiment of this application.

[0027] Figure 10 This is a schematic structural diagram of a decoder according to another embodiment of this application.

[0028] Figure 11 This is a schematic structural diagram of an encoder according to yet another embodiment of this application.

[0029] Figure 12 yes Figure 11 An exemplary structural diagram of the self-attention module in [the system].

[0030] Figure 13 This is a schematic structural diagram of a decoder according to yet another embodiment of this application.

[0031] Figure 14 yes Figure 13 A schematic diagram of the mask block structure.

[0032] Figure 15 yes Figure 13 A schematic structural diagram of the residual block in the diagram.

[0033] Figure 16 This is a schematic block diagram of a transmitting device according to an embodiment of this application.

[0034] Figure 17 This is a schematic block diagram of a receiving device according to an embodiment of this application.

[0035] Figure 18 This is a schematic block diagram of a communication device provided according to an embodiment of this application.

[0036] Figure 19 This is a schematic block diagram of a chip provided according to an embodiment of this application.

[0037] Figure 20 This is a schematic block diagram of a communication system provided according to an embodiment of this application. Detailed Implementation

[0038] The technical solutions of the embodiments of this application will now be described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art without creative effort regarding the embodiments of this application are within the scope of protection of this application.

[0039] The technical solutions of this application embodiment can be applied to various communication systems, such as: Global System for Mobile Communication (GSM) system, Code Division Multiple Access (CDMA) system, Wideband Code Division Multiple Access (WCDMA) system, General Packet Radio Service (GPRS), Long Term Evolution (LTE) system, Advanced Long Term Evolution (LTE-A) system, New Radio (NR) system, evolution of NR system, LTE-based access to unlicensed spectrum (LTE-U) system, NR-based access to unlicensed spectrum (NR-U) system, Non-Terrestrial Networks (NTN) system, Universal Mobile Telecommunication System (UMTS), and Wireless Local Area Network (WLAN). Area Networks (WLAN), Wireless Fidelity (WiFi), 5th Generation (5G) systems, or other communication systems.

[0040] Traditional communication systems typically support a limited number of connections and are easy to implement. However, with the development of communication technology, mobile communication systems will not only support traditional communication but also, for example, device-to-device (D2D) communication, machine-to-machine (M2M) communication, machine-type communication (MTC) communication, vehicle-to-vehicle (V2V) communication, or vehicle-to-everything (V2X) communication. The embodiments of this application can also be applied to these communication systems.

[0041] Optionally, the communication system in this application embodiment can be applied to a carrier aggregation (CA) scenario, a dual connectivity (DC) scenario, or a standalone (SA) network deployment scenario.

[0042] Optionally, the communication system in this application embodiment can be applied to unlicensed spectrum, wherein unlicensed spectrum can also be considered as shared spectrum; or, the communication system in this application embodiment can also be applied to licensed spectrum, wherein licensed spectrum can also be considered as non-shared spectrum.

[0043] This application describes various embodiments in conjunction with network devices and terminal devices. The terminal device may also be referred to as user equipment (UE), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent, or user device, etc.

[0044] Terminal devices can be stations (STs) in WLANs, cellular phones, cordless phones, Session Initiation Protocol (SIP) phones, Wireless Local Loop (WLL) stations, Personal Digital Assistant (PDA) devices, handheld devices with wireless communication capabilities, computing devices or other processing devices connected to a wireless modem, in-vehicle devices, wearable devices, terminal devices in next-generation communication systems such as NR networks, or terminal devices in future evolved Public Land Mobile Network (PLMN) networks, etc.

[0045] In the embodiments of this application, the terminal device can be deployed on land, including indoor or outdoor, handheld, wearable or vehicle-mounted; it can also be deployed on water (such as ships); and it can also be deployed in the air (such as airplanes, balloons and satellites).

[0046] In the embodiments of this application, the terminal device may be a mobile phone, a tablet computer, a computer with wireless transceiver capabilities, a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal device in industrial control, a wireless terminal device in self-driving, a wireless terminal device in remote medical care, a wireless terminal device in a smart grid, a wireless terminal device in transportation safety, a wireless terminal device in a smart city, or a wireless terminal device in a smart home, etc.

[0047] By way of example and not limitation, in this embodiment, the terminal device can also be a wearable device. Wearable devices, also known as wearable smart devices, are a general term for devices that utilize wearable technology to intelligently design and develop everyday wearables, such as glasses, gloves, watches, clothing, and shoes. Wearable devices are portable devices that are worn directly on the body or integrated into the user's clothing or accessories. Wearable devices are not merely hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are feature-rich, large in size, and can achieve complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses, as well as those that focus on a specific type of application function and require the use of other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0048] In the embodiments of this application, the network device can be a device for communicating with mobile devices. The network device can be an access point (AP) in WLAN, a base station (BTS) in GSM or CDMA, a base station (NodeB, NB) in WCDMA, an evolved Node B (eNB or eNodeB) in LTE, a relay station or access point, or a vehicle-mounted device, wearable device, or a network device (gNB) in an NR network, or a network device in a future evolved PLMN network or an NTN network, etc.

[0049] By way of example and not limitation, in this embodiment, the network device may have mobility characteristics; for example, the network device may be a mobile device. Optionally, the network device may be a satellite or a balloon station. For example, the satellite may be a low Earth orbit (LEO) satellite, a medium Earth orbit (MEO) satellite, a geostationary earth orbit (GEO) satellite, a high elliptical orbit (HEO) satellite, etc. Optionally, the network device may also be a base station located on land, water, or other similar locations.

[0050] In this embodiment, the network device can provide services to a cell. The terminal device communicates with the network device through the transmission resources (e.g., frequency domain resources, or spectrum resources) used by the cell. The cell can be the cell corresponding to the network device (e.g., a base station). The cell can belong to a macro base station or to a base station corresponding to a small cell. The small cell can include: metro cell, micro cell, pico cell, femto cell, etc. These small cells have the characteristics of small coverage area and low transmission power, and are suitable for providing high-speed data transmission services.

[0051] For example, the communication system 100 used in the embodiments of this application is as follows: Figure 1 As shown. The communication system 100 may include a network device 110, which may be a device that communicates with a terminal device 120 (or a communication terminal, terminal). The network device 110 can provide communication coverage for a specific geographical area and can communicate with terminal devices located within that coverage area.

[0052] Figure 1 An exemplary embodiment shows a network device and two terminal devices. Optionally, the communication system 100 may include multiple network devices and each network device may include other numbers of terminal devices within its coverage area. This application embodiment does not limit this.

[0053] Optionally, the communication system 100 may also include other network entities such as a network controller and a mobility management entity, which is not limited in this embodiment.

[0054] It should be understood that devices with communication functions in the network / system of this application embodiment can be referred to as communication devices. Figure 1Taking the communication system 100 shown as an example, the communication equipment may include a network device 110 and a terminal device 120 with communication functions. The network device 110 and the terminal device 120 may be the specific devices described above, which will not be repeated here. The communication equipment may also include other devices in the communication system 100, such as network controllers, mobility management entities and other network entities. This application embodiment does not limit this.

[0055] It should be understood that the terms "system" and "network" are often used interchangeably in this document. The term "and / or" in this document merely describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Furthermore, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0056] It should be understood that the term "instruction" mentioned in the embodiments of this application can be a direct instruction, an indirect instruction, or an indication of a relationship. For example, A instructing B can mean that A directly instructs B, such as B being able to obtain information through A; it can also mean that A indirectly instructs B, such as A instructing C, so B can obtain information through C; or it can mean that there is a relationship between A and B.

[0057] In the description of the embodiments of this application, the term "correspondence" may indicate that there is a direct or indirect correspondence between two things, or that there is an association between two things, or that there is a relationship of instruction and being instructed, configuration and being configured, etc.

[0058] In this application embodiment, "predefined" can be implemented by pre-storing corresponding codes, tables, or other means that can be used to indicate relevant information in the device (e.g., including terminal devices and network devices). This application does not limit the specific implementation method. For example, predefined can refer to what is defined in the protocol.

[0059] In this application embodiment, the "protocol" may refer to a standard protocol in the field of communication, such as the LTE protocol, the NR protocol, and related protocols applied to future communication systems. This application does not limit this.

[0060] To facilitate understanding of the technical solutions of the embodiments of this application, the technical solutions of this application are described in detail below through specific embodiments. The following related technologies are optional solutions and can be arbitrarily combined with the technical solutions of the embodiments of this application, all of which fall within the protection scope of the embodiments of this application. The embodiments of this application include at least some of the following contents.

[0061] In NR systems, channel information feedback employs a codebook-based feedback scheme. This scheme selects the optimal channel information feature vector from the precoding codebook based on the channel estimation results. However, due to the finite nature of the precoding codebook, the mapping process from the channel estimation results to the channel information in the precoding codebook is lossy in terms of quantization. This reduces the accuracy of the feedback channel information, thereby lowering the performance of the precoding. If the feedback is of the entire channel information, it increases the CSI feedback overhead. Therefore, balancing the feedback accuracy of channel information with the CSI feedback overhead is a problem that urgently needs to be solved.

[0062] Figure 2 This is a schematic interactive diagram of a channel information feedback method 200 according to an embodiment of this application, such as... Figure 2 As shown, the method 200 includes at least the following:

[0063] S201, The transmitting device receives the reference signal sent by the receiving device;

[0064] S202, the transmitting device performs channel estimation based on the reference signal to obtain channel information between the transmitting device and the receiving device;

[0065] S203, The transmitting device performs feature decomposition on the channel information to obtain at least one first feature vector;

[0066] S204, the transmitting device encodes the at least one first feature vector through a neural network to obtain the target bit stream;

[0067] S205, the sending device sends the target bit stream to the receiving device.

[0068] S206, the receiving device decodes the target bit stream to obtain at least one target feature vector.

[0069] In some embodiments, the transmitting device is a terminal device, and the receiving device is a network device.

[0070] In other embodiments, the transmitting device is a network device and the receiving device is a terminal device.

[0071] In some other embodiments, the transmitting device is a terminal device and the receiving device is another terminal device.

[0072] In some other embodiments, the transmitting device is a network device and the receiving device is another network device.

[0073] It should be understood that the reference signal will vary depending on the transmitting and receiving devices. For example, if the transmitting device is a terminal device and the receiving device is a network device, the reference signal may be a demodulation reference signal (DMRS).

[0074] In this embodiment of the application, the transmitting device is equipped with an encoder and the receiving device is equipped with a decoder. In this embodiment of the application, the encoder in the transmitting device and the decoder in the receiving device can be implemented by a neural network.

[0075] It should be noted that the embodiments of this application do not limit the specific method by which the transmitting device performs feature decomposition on the channel information. As an example, the transmitting device can perform Singular Value Decomposition (SVD) on the channel information to obtain the at least one first feature vector.

[0076] In some embodiments of this application, the channel information is the full channel information obtained by channel estimation of the reference signal, that is, the unquantized channel information.

[0077] Therefore, in this embodiment, the transmitting device performs feature decomposition on the full channel information obtained from channel estimation to obtain at least one feature vector, further encodes the feature vector using a neural network to obtain the target bit stream, and then transmits the target bit stream to the receiving device. Correspondingly, the receiving device decodes the target bit stream to obtain a target feature vector, and further performs precoding based on the target feature vector.

[0078] Based on the channel information feedback scheme of this application embodiment, on the one hand, when the transmitting device transmits channel information, it only needs to transmit the target bit stream obtained by encoding the feature vector obtained by feature decomposition of the entire channel information, which is beneficial to reduce CSI overhead. On the other hand, encoding the feature vector of the entire channel information instead of directly compressing and encoding the entire channel information can take into account the correlation characteristics between channel information, which is beneficial to avoid compressing too much redundant information, improve compression efficiency, and thus improve coding performance.

[0079] In some embodiments of this application, the receiving device may perform precoding based on at least one target feature vector obtained by decoding, for example, it may perform beamforming based on the at least one target feature vector.

[0080] In some scenarios, the scheduling bandwidth of the transmitting device may include multiple subcarrier groups, each corresponding to multiple subbands. In some embodiments of the application, the transmitting device may feed back the channel information on the multiple subbands of the transmitting device separately.

[0081] In some embodiments, the transmitting device can perform channel estimation based on a reference signal to obtain channel information on multiple sub-bands, and further perform feature decomposition on the channel information on the multiple sub-bands to obtain feature vectors corresponding to the multiple sub-bands respectively.

[0082] In other words, the full channel information for channel estimation includes the channel information corresponding to each subcarrier group among the multiple subcarrier groups of the transmitting device. Or, the full channel information can include the channel information corresponding to each subband among the multiple subbands of the transmitting device.

[0083] Correspondingly, the at least one first feature vector includes the feature vector corresponding to each of the plurality of subcarrier groups. The at least one first feature vector also includes the feature vector corresponding to each of the plurality of subbands.

[0084] Since the feature vectors of channel information on multiple sub-bands are correlated, for example, there is a lot of correlation information between channel information on adjacent sub-bands, joint compression coding of the feature vectors of multiple sub-bands takes into account this correlation information, which is beneficial to improve compression efficiency and enhance feedback recovery performance.

[0085] Figure 3 This is a schematic system architecture diagram of a channel feedback method according to an embodiment of this application.

[0086] The transmitting device can perform feature decomposition on the channel information of n sub-bands to obtain the feature vectors corresponding to the n sub-bands, denoted as W_1, W_2, ..., W_n. These feature vectors are then input into the encoder, which uses a neural network to jointly encode the feature vectors of the n sub-bands to obtain the target bitstream, denoted as B. The transmitting device then sends the target bitstream B to the receiving device, where the decoder decodes and recovers the target bitstream B to obtain n target feature vectors, denoted as W`_1, W`_2, ..., W`_n.

[0087] Therefore, in this embodiment of the application, the transmitting device performs joint compression feedback on the feature vectors of channel information of multiple sub-bands on the scheduling bandwidth. Correspondingly, the receiving device can decompress and reconstruct the feature vectors of multiple sub-bands. Compared with utilizing the correlation information between feature vectors, this is beneficial to improving compression efficiency, thereby enhancing the compression, feedback and decompression performance of the entire system.

[0088] Optionally, in some embodiments, the scheduling bandwidth may be at least one BWP, or at least one carrier, or at least one frequency band, etc., and this application does not limit it in this way.

[0089] It should be understood that the embodiments of this application do not specifically limit the specific implementation of the encoder in the transmitting device and the decoder in the receiving device. The implementation of the encoder in the transmitting device and the decoder in the receiving device will be described below with reference to specific embodiments.

[0090] In some embodiments of this application, the encoder in the transmitting end and the decoder in the receiving end can be employed as follows: Figure 4 The neural network structure is implemented in [the document]. This neural network includes an input layer, hidden layers, and an output layer. The input layer receives data, the hidden layers process the received data, and the results are generated in the output layer. In this neural network, each node represents a processing module, which can be considered as simulating a neuron. Multiple neurons form a layer of the neural network, and the information transmission and processing across multiple layers construct a complete neural network.

[0091] In some embodiments of this application, the encoder of the transmitting device can use computer vision technology to process at least one first feature vector of the input, for example, by using a neural network to compress and encode the at least one first feature vector as an image to be compressed to obtain the target bitstream.

[0092] Correspondingly, the decoder of the receiving device can decode and recover the target bitstream as information obtained from image compression to obtain the target image, wherein the target image includes at least one target feature vector.

[0093] In some embodiments, the encoder of the transmitting device may employ a neural network for image processing to compress and encode the at least one first feature vector. For example, the neural network for image processing may be a convolutional neural network or other neural networks with superior image processing performance. This application does not limit this to any particular type.

[0094] In some embodiments, the encoder of the receiving device may employ a neural network for image processing to decompress the target bitstream. For example, the neural network for image processing may be a convolutional neural network or other neural networks with superior image processing performance. This application does not limit this to any particular type.

[0095] Optionally, the convolutional neural network includes an input layer, at least one convolutional layer, at least one pooling layer, a fully connected layer, and an output layer. By introducing convolutional and pooling layers, relative to... Figure 4 The neural network architecture effectively controls the rapid increase in network parameters, limits the number of parameters, and facilitates the discovery of local structural features, thereby improving the robustness of the algorithm.

[0096] In some embodiments, encoding the at least one first feature vector using a neural network to obtain a target bitstream includes:

[0097] The at least one first feature vector is concatenated into a feature vector matrix, which is then input into the neural network.

[0098] The neural network encodes the feature vector matrix as an image to be compressed to obtain the target bitstream.

[0099] For example, the feature vectors W_1, W_2, ..., W_n corresponding to the n sub-bands are concatenated into a feature vector matrix W = [W_1, W_2, ..., W_n]. T The feature vector matrix W is input into the neural network as the image to be compressed. The neural network then compresses and encodes the feature vector matrix W as the image to be compressed, resulting in the target bitstream B.

[0100] Correspondingly, at the decoding end, the decoding of the target bitstream using a neural network to obtain at least one target feature vector includes:

[0101] The target bitstream is input into the neural network;

[0102] The target bitstream is decoded using the neural network as information obtained from encoding the image to obtain a target image, the target image including the at least one target feature vector.

[0103] That is, the decoder at the receiving end uses the received target bitstream B as information obtained from image compression, decompresses the target bitstream B, and obtains a target image including at least one target feature vector.

[0104] In some embodiments, the encoder in the transmitting device and the decoder in the receiving device can respectively employ Figure 5 and Figure 6 The network structure implementation in [the project / project].

[0105] In some embodiments, the encoder in the transmitting device may include: a feature extraction module, configured to receive an input feature vector matrix W, extract features from the feature vector matrix W, and obtain a feature map corresponding to the feature vector matrix W.

[0106] In some embodiments, the feature extraction module extracts features from the feature vector matrix W using convolution kernels of different sizes to obtain feature maps of different views of the feature vector matrix W, thereby increasing the nonlinearity in the convolution process and improving the expressive power of the convolutional neural network.

[0107] As an example, such as Figure 5As shown, the feature extraction module may include 3×3 convolutional layers, 5×5 convolutional layers, and 7×7 convolutional layers. Specifically, the 3×3 convolutional layers use 3x3 kernels, the 5×5 convolutional layers use 5x5 kernels, and the 7×7 convolutional layers use 7x7 kernels. In practical applications, other kernel sizes can be used instead, and this application does not limit this.

[0108] Furthermore, the multiple feature maps output by the feature extraction module are input to the stitching module for feature map merging, for example, by stitching the multiple feature maps along the channel dimension.

[0109] As an example, such as Figure 5 As shown, the splicing module can be implemented using a 1×1 convolutional layer. The 1×1 convolutional layer uses a 1×1 convolutional kernel. The number of channels can be controlled by controlling the number of convolutional kernels in the 1×1 convolutional layer. The convolution process using a 1×1 convolutional kernel is larger than the calculation process of a fully connected layer. In addition, a non-linear activation function is added, which is beneficial to increase the non-linearity of the neural network and make the features expressed by the neural network more complex.

[0110] Furthermore, such as Figure 5 As shown, the feature map output from the splicing module is converted into the target bit stream B after being processed by a fully connected layer and a quantization layer.

[0111] It should be understood that Figure 5 The convolutional neural network structure in the example is only an example. In practical applications, it can be flexibly designed according to information such as the number of subbands and encoding / decoding performance requirements. For example, activation layers, normalization layers and other network layers can be added between convolutional layers. This application is not limited to this.

[0112] In some embodiments, such as Figure 6 As shown, the decoder may include a fully connected layer, a dimension adjustment module, and a residual module. The target bitstream B received from the transmitting device is first input to the fully connected layer and the dimension adjustment module to be converted into the dimension of the feature vector matrix W, and then input to the residual module to output the feature vector matrix W' composed of the at least one target feature vector.

[0113] In some embodiments, such as Figure 6 As shown, the residual module may include a convolutional layer, an L-fold common module, a convolutional layer, and a summing module, where L is a positive integer. First, the output of the dimension adjustment module is sampled and copied, with one path input to the summing module and the other to the convolutional layer. The convolutional kernels of the convolutional layer amplify the number of channels, and feature information is further extracted through the L-fold common module. The output of the L-fold common module is further reduced in the number of channels by the convolutional layer. The output of this convolutional layer is then input to the summing module and summed with the sampled and copied output of the dimension adjustment module to obtain the feature vector matrix W'.

[0114] It should be understood that the embodiments of this application do not specifically limit the number or structural composition of the common modules. As an example, the common modules may adopt... Figure 7 The structure shown is not limited to this.

[0115] It should be understood that Figure 6 The convolutional neural network structure in the example is only an example. In practical applications, it can be flexibly configured according to information such as the number of subbands and encoding / decoding performance requirements. This application is not limited to this.

[0116] In some embodiments of this application, the model parameters of the neural networks of the encoder and decoder are obtained through joint training. For example, the model parameters of the neural networks of the encoder and decoder are first initialized, and multiple sets of feature vector matrix samples are input into the neural network of the encoder for encoding to obtain multiple target bit streams. The multiple target bit streams are then input into the decoder for decoding. The model parameters of the encoder and decoder are adjusted according to the decoding results until the feature vector output by the decoder and the feature vector input into the neural network of the encoder satisfy the convergence condition.

[0117] Therefore, in this embodiment, the transmitting device uses a convolutional neural network to compress and encode the feature vectors corresponding to the channel information of multiple sub-bands as an image to obtain the target bitstream. Correspondingly, the receiving device uses a convolutional neural network to decompress the target bitstream and recover the target image, thereby obtaining at least one target feature vector. On the one hand, compared to directly encoding the full channel information of the channel estimation, encoding the feature vectors of the full channel information helps to avoid compressing too much redundant information and reduces CSI feedback overhead. On the other hand, joint compression feedback based on the cross-correlation information between the feature vectors of multiple sub-bands in the frequency domain helps to improve compression feedback performance.

[0118] In other embodiments of this application, the encoder of the transmitting device may use a recurrent neural network (RNN) to process at least one first feature vector of the input. For example, the at least one first feature vector may be compressed as an element of a sequence through a neural network to obtain an encoding result, namely the target bit stream.

[0119] Correspondingly, the encoder of the receiving device can use a recurrent neural network to decompress and recover the target bit stream by using the information obtained by compressing and encoding the sequence, thereby obtaining a target sequence composed of at least one target bit vector.

[0120] As shown earlier, a neural network consists of an input layer, hidden layers, and an output layer. The output is controlled by activation functions, and layers are connected by weights. The activation functions are predetermined, and the neural network model learns through training, which is then embedded in the weights. Basic neural networks only establish weighted connections between layers; the biggest difference between RNNs and basic neural networks is that RNNs also establish weighted connections between neurons within layers.

[0121] Figure 8 This is a typical RNN structure diagram. Figure 8 Each arrow represents a transformation, meaning that the arrow connections all have weights. Figure 8 The left side of the diagram is a folded diagram of the RNN structure, and the right side is an unfolded diagram of the RNN structure. The arrow next to 'h' in the left side diagram indicates that the "recurrence" in this structure is reflected in the hidden layer.

[0122] As can be seen from the unfolded RRN structure diagram, the neurons in the hidden layers also have weights. That is to say, as the sequence progresses, the earlier hidden layers will influence the later hidden layers. Figure 8 In this diagram, x represents the input, h represents the hidden unit, O represents the output, y represents the label of the training set, L represents the loss function, K, V, and U represent the weights, t represents time t, t-1 represents time t-1, and t+1 represents time t+1. This shows that the "loss" accumulates continuously as the sequence is recommended. Based on this structure, RNNs can perform well in processing sequence data; that is, an RNN is a recursive neural network that recurs in the direction of sequence evolution and where all nodes (recurrent units) are connected in a chain-like manner.

[0123] Long Short-Term Memory (LSTM) is an evolved form of RNN. Unlike typical RNN structures, LSTM introduces the concept of cell states. Unlike RNNs, which only consider the most recent state, LSTM's cell states determine which states should be retained and which should be forgotten, thus addressing the shortcomings of traditional RNNs in long-term memory.

[0124] In some embodiments of this application, the at least one first feature vector can be processed by a basic RNN to obtain the target bitstream B, or the at least one first feature vector can be processed by an LSTM to obtain the target bitstream B.

[0125] In some embodiments, encoding the at least one first feature vector using a neural network to obtain a target bitstream includes:

[0126] Each feature vector in the at least one first feature vector is sequentially input into the recurrent neural network;

[0127] The target bitstream is obtained by encoding each feature vector as an element of the sequence using the recurrent neural network.

[0128] For example, each feature vector in the at least one first feature vector is sequentially input into the LSTM, and the LSTM encodes each feature vector as an element of the sequence to obtain the target bitstream.

[0129] As an example, the feature vectors W_1, W_2, ..., W_n corresponding to the n sub-bands are input into the LSTM as different elements of the sequence, and the sequence is processed by the LSTM to obtain the target bitstream.

[0130] Correspondingly, at the decoding end, the decoding of the target bitstream using a neural network to obtain at least one target feature vector includes:

[0131] The target bitstream is input into the recurrent neural network;

[0132] The target bitstream is decoded using the recurrent neural network as information obtained by encoding the sequence to obtain a target sequence, the target sequence including the at least one target feature vector.

[0133] That is, the decoder at the receiving end uses the received target bitstream B as information obtained by sequence compression, decompresses the target bitstream B, and obtains a target sequence including at least one target feature vector.

[0134] The following describes the encoder in the transmitting device and the decoder in the receiving device, respectively. Figure 9 and Figure 10 Taking the network structure in the example, the specific encoding and decoding process is explained. It should be understood that... Figure 9 and Figure 10 The neural network structure shown is merely an example. In practical applications, it can be flexibly configured according to information such as the number of sub-bands and encoding / decoding performance requirements. This application is not limited to this.

[0135] In some embodiments, such as Figure 9 As shown, the encoder may include an LSTM module, which is used to sequentially receive each feature vector in the at least one first feature vector and process each feature vector as an element of the sequence.

[0136] Furthermore, the encoder may include a fully connected layer and a quantization layer for converting the result processed by the LSTM module to obtain the target bitstream B.

[0137] In some embodiments, such as Figure 10As shown, the decoder may include: a fully connected layer, multiple LSTM modules, and a fully connected layer connected to each of the multiple LSTM modules. The first fully connected layer performs a dimension transformation on the target bitstream B, and its output serves as the input to the first LSTM module. The output of each LSTM module serves as the input to the next LSTM module. The outputs of the multiple LSTM modules, after passing through their respective fully connected layers, output the corresponding feature vector matrix W` = {W`_1, W`_2, ..., W`_n}.

[0138] Therefore, in this embodiment, the transmitting device compresses the feature vectors of multiple sub-bands as elements of a sequence using a recurrent neural network to obtain the target bitstream. The receiving device decompresses the target bitstream using a recurrent neural network to recover the elements of the sequence, thereby obtaining at least one target feature vector. On the one hand, compared to directly encoding the full channel information of the channel estimation, encoding the feature vectors of the full channel information helps avoid compressing too much redundant information and reduces CSI feedback overhead. On the other hand, joint compression feedback based on the cross-correlation information between the feature vectors of multiple sub-bands in the frequency domain helps improve compression feedback performance.

[0139] In some embodiments of this application, encoding the at least one first feature vector using a neural network to obtain the target bitstream includes:

[0140] The target bitstream is obtained by encoding the at least one first feature vector based on an attention mechanism.

[0141] Optionally, in the embodiments of this application, the attention mechanism can be a self-attention mechanism or other attention mechanisms, and this application does not limit it.

[0142] The self-attention mechanism adopts a "query-key-value" pattern.

[0143] The calculation of attention mainly involves three steps:

[0144] Step 1: Calculate the similarity between the query and each key to obtain the weight. Common similarity functions include dot product, concatenation, and perceptron.

[0145] Step 2: Normalize these weights using a softmax function;

[0146] Finally, the weights and their corresponding key values ​​are weighted and summed to obtain the final attention.

[0147] In some embodiments of this application, the transmitting device extracts correlation features between feature vectors of multiple sub-bands and correlation features between elements of the feature vectors based on an attention mechanism, thereby further improving encoding performance and decompression performance. For example, the transmitting device extracts correlation features of elements in the at least one first feature vector based on an attention mechanism to obtain at least one second feature vector, and further compresses the at least one second feature vector to obtain the target bitstream B.

[0148] Correspondingly, at the decoding end, the receiving device can decode the target bitstream based on an attention mechanism to obtain at least one target feature vector.

[0149] Optionally, in the embodiments of this application, the attention mechanism can be a self-attention mechanism or other attention mechanisms, and this application does not limit it.

[0150] In some embodiments of this application, the decoder first extracts features from the target bitstream to obtain a first feature map of the target bitstream; further, it determines the weights of the elements in the first feature map of the target bitstream based on an attention mechanism; it performs a dot product between the first feature map of the target bitstream and the weights of the elements in the first feature map of the target bitstream to obtain a second feature map of the target bitstream; and it decompresses the second feature map of the target bitstream to obtain the at least one target feature vector.

[0151] The following describes the encoder in the transmitting device and the decoder in the receiving device, respectively. Figure 11 and Figure 13 Taking the network structure in the example, the specific encoding and decoding process is explained. It should be understood that... Figure 11 and Figure 13 The neural network structure shown is merely an example. In practical applications, it can be flexibly configured according to information such as the number of sub-bands and encoding / decoding performance requirements. This application is not limited to this.

[0152] In some embodiments, such as Figure 11 As shown, the encoder may include: s-th self-attention modules, a concatenation module, a fully connected layer, and a quantization layer, where s is a positive integer. Each self-attention module is used to extract the correlation features of elements in multiple input first feature vectors using a self-attention mechanism. After passing through these s cascaded sub-attention modules, multiple second feature vectors are output. Further, these multiple second feature vectors are input to the concatenation module and concatenated into a single feature vector. After processing by the fully connected layer and the quantization layer, the target bitstream B is obtained.

[0153] In some embodiments, the attention module may employ Figure 12 The structure shown is implemented, but this application is not limited to it.

[0154] As an example, such as Figure 12 As shown, the attention module includes: n fully connected layers, an attention layer, and n sets of two fully connected layers. Each of the n fully connected layers receives a feature vector, which can be either the first feature vector mentioned earlier or the feature vector output by the previous attention module. The attention layer receives multiple feature vectors output from the n fully connected layers and extracts the correlation features between the elements of these multiple feature vectors. Each output of the self-attention layer is first copied and sampled, and then each output is summed with the sampled output through the corresponding set of two fully connected layers to obtain the output of the self-attention module.

[0155] In some embodiments, such as Figure 13 As shown, the decoder may include a fully connected layer, a dimension adjustment module, a feature extraction module, and a self-attention module. The target bitstream B is first input to the fully connected layer and the dimension adjustment module to be converted into the dimension of the feature vector matrix W, and then input to the feature extraction module for feature extraction to obtain the first feature map. The output of the dimension adjustment module is sampled and copied.

[0156] Optionally, such as Figure 13 As shown, the feature extraction module may include a convolutional layer, an activation function, a q-th order residual block, and a t-th order residual block, where q and t are positive integers.

[0157] This convolutional layer is used to adjust the number of channels in the output of the dimension adjustment module. The output of the convolutional layer is input to the q-th order residual block through an activation function, and the output of the activation function is sampled and copied.

[0158] The output of the activation function is split into two paths after passing through q residual blocks. One path passes through t residual blocks to obtain the second feature map, while the other path is input to the self-attention module to extract the attention weights of the elements in the feature vector matrix.

[0159] In some embodiments, the self-attention module can be implemented using a mask block, or it can also be implemented using... Figure 11 The attention module implementation in the application is not limited in this application.

[0160] For example, the output of the qth residual block can be input into a mask block to extract an attention mask for elements in the feature map.

[0161] Furthermore, the output of the mask block and the output of the t-th residual block are multiplied by a dot product, and the result of the dot product is added to the sampled copy result of the output of the activation function.

[0162] Then, the summed result is input into the q-th residual block for result correction. Then, the channel dimension is reduced through a convolutional layer and added to the sampling copy result of the second feature vector matrix to output the recovered feature vector matrix W` = [W`_1, W`_2, ..., W`_n].

[0163] Figure 14 This is an exemplary structural diagram of the mask block, but the application is not limited thereto.

[0164] like Figure 14 As shown, the mask block can include an m-th residual block, a downsampling module, a 2m-th residual block, an upsampling module, an m-th residual block, a convolutional layer, and a sigmoid activation function. That is, the mask block can be implemented using a 4m-th residual block. Upsampling and downsampling are performed after the m-th and 3m-th residual blocks, respectively. The intermediate 2m-th residual blocks can be used to extract features from a wider range of global information. Finally, the channel dimension is matched through the convolutional layer, and the output of the convolutional layer is mapped to between 0 and 1 through the sigmoid activation function to obtain the attention weights.

[0165] It should be understood that the residual block in the embodiments of this application can be adopted. Figure 15 The structure shown can be used to implement the functionality, or other equivalent structures can be used; this application does not limit the implementation in this regard.

[0166] As an example, such as Figure 15 As shown, the residual block may include: a convolutional layer, a normalization layer, an activation function, and a summing module. The input to the residual block is sampled and copied to the summing module in one path, and then fed into the summing module through the outputs of the convolutional layers. The summing module sums the input of the residual block and the output of the convolutional layers, resulting in the output of the residual block. Designing appropriate residual blocks in a neural network is beneficial for solving gradient problems in neural networks.

[0167] Therefore, in this embodiment, the transmitting device uses an attention mechanism to compress the feature vectors of multiple sub-bands as elements of a sequence to obtain the target bitstream. Correspondingly, the receiving device uses an attention mechanism to decompress and recover the target bitstream, thereby obtaining at least one target feature vector. On the one hand, compared to directly encoding the full channel information of the channel estimation, encoding the feature vectors of the full channel information helps avoid compressing too much redundant information and reduces CSI feedback overhead. On the other hand, using an attention mechanism to jointly compress and feedback the mutual correlation information between the elements of the feature vectors of multiple sub-bands considers the correlation characteristics between the feature vectors of multiple sub-bands and the correlation characteristics between the elements of the feature vectors, which helps improve the compression feedback performance.

[0168] The above text combined Figures 2 to 15 The method embodiments of this application are described in detail below, in conjunction with... Figures 16 to 20 The present application describes the device embodiments in detail. It should be understood that the device embodiments correspond to the method embodiments, and similar descriptions can be referred to the method embodiments.

[0169] Figure 16 A schematic block diagram of a transmitting device 400 according to an embodiment of this application is shown. Figure 16 As shown, the transmitting device 400 includes:

[0170] The communication module 410 is used to receive reference signals sent by the receiving device;

[0171] The processing module 420 is used to perform channel estimation based on the reference signal to obtain channel information between the transmitting end device and the receiving end device, and to perform feature decomposition on the channel information to obtain at least one first feature vector.

[0172] Encoding module 430 is used to encode the at least one first feature vector through a neural network to obtain a target bit stream;

[0173] The communication module 410 is further configured to: send the target bit stream to the receiving device.

[0174] In some embodiments of this application, the channel information includes channel information corresponding to each of the plurality of subcarrier groups in the transmitting device, and the at least one first feature vector includes feature vectors corresponding to each of the plurality of subcarrier groups.

[0175] In some embodiments of this application, the encoding module 430 is specifically used for:

[0176] The at least one first feature vector is concatenated into a feature vector matrix, which is then input into the neural network.

[0177] The neural network encodes the feature vector matrix as an image to be compressed, thereby obtaining the target bitstream.

[0178] In some embodiments of this application, the neural network is a convolutional neural network.

[0179] In some embodiments of this application, the neural network is a recurrent neural network, and the encoding module 430 is further configured to:

[0180] Each feature vector in the at least one first feature vector is sequentially input into the recurrent neural network;

[0181] The target bitstream is obtained by encoding each feature vector as an element of the sequence using the recurrent neural network.

[0182] In some embodiments of this application, the recurrent neural network includes a long short-term memory (LSTM) neural network.

[0183] In some embodiments of this application, the encoding module 430 is further configured to:

[0184] The target bitstream is obtained by encoding the at least one first feature vector based on an attention mechanism.

[0185] In some embodiments of this application, the encoding module 430 further includes:

[0186] An attention module is used to extract correlation features between elements in the at least one first feature vector based on an attention mechanism to obtain at least one second feature vector.

[0187] The feature vector compression module is used to perform feature compression on the at least one second feature vector to obtain the target bit stream.

[0188] In some embodiments of this application, the transmitting device is a terminal device, and the receiving device is a network device.

[0189] Optionally, in some embodiments, the communication module described above may be a communication interface or transceiver, or an input / output interface of a communication chip or system-on-a-chip. The processing module and encoding module described above may be one or more processors.

[0190] It should be understood that the transmitting device 400 according to the embodiments of this application may correspond to the transmitting device in the method embodiments of this application, and the above and other operations and / or functions of each unit in the transmitting device 400 are respectively for implementing Figures 2 to 15 The corresponding process of the sending device in method 200 shown will not be described in detail here for the sake of brevity.

[0191] Figure 17 This is a schematic block diagram of a receiving device according to an embodiment of this application. Figure 17 The receiving device 500 includes:

[0192] Communication module 510 is used to receive a target bit stream sent by a transmitting device, wherein the target bit stream is obtained by the transmitting device encoding at least one first feature vector, and the at least one first feature vector is obtained by the transmitting device performing feature decomposition on the channel estimation result;

[0193] The decoding module 520 is used to decode the target bitstream through a neural network to obtain at least one target feature vector.

[0194] In some embodiments of this application, the channel information includes channel information corresponding to each of the plurality of subcarrier groups in the transmitting device, and the at least one first feature vector includes feature vectors corresponding to each of the plurality of subcarrier groups.

[0195] In some embodiments of this application, the decoding module 520 is used for:

[0196] The target bitstream is input into the neural network;

[0197] The target bitstream is decoded using the neural network as information obtained from encoding the image to obtain a target image, the target image including the at least one target feature vector.

[0198] In some embodiments of this application, the neural network is a convolutional neural network.

[0199] In some embodiments of this application, the neural network is a recurrent neural network, and the decoding module 520 is further configured to:

[0200] The target bitstream is input into the recurrent neural network;

[0201] The target bitstream is decoded using the recurrent neural network as information obtained by encoding the sequence to obtain a target sequence, the target sequence including the at least one target feature vector.

[0202] In some embodiments of this application, the recurrent neural network includes a long short-term memory (LSTM) neural network.

[0203] In some embodiments of this application, the decoding module 520 is further configured to:

[0204] The target bitstream is decoded based on an attention mechanism to obtain at least one target feature vector.

[0205] In some embodiments of this application, the encoding module 520 includes:

[0206] The feature extraction module is used to extract features from the target bitstream to obtain a first feature map;

[0207] An attention module is used to determine the weights of elements in the first feature map based on an attention mechanism;

[0208] The dot product module is used to perform a dot product between the first feature map and the weights of the elements in the first feature map to obtain the second feature map;

[0209] The decompression module is used to decompress the second feature map to obtain the at least one target feature vector.

[0210] In some embodiments of this application, the receiving device 500 further includes:

[0211] A processing module is used to pre-encode based on the at least one target feature vector.

[0212] Optionally, in some embodiments, the communication module described above may be a communication interface or transceiver, or an input / output interface of a communication chip or system-on-a-chip. The processing module and encoding module described above may be one or more processors.

[0213] It should be understood that the receiving device 500 according to the embodiments of this application may correspond to the receiving device in the method embodiments of this application, and the above and other operations and / or functions of each unit in the receiving device 500 are respectively for implementing Figures 2 to 15 The corresponding process of the receiving device in method 200 shown will not be described in detail here for the sake of brevity.

[0214] In summary, the transmitting device performs feature decomposition on the full channel information obtained from channel estimation to obtain at least one feature vector. It then uses a neural network to encode this feature vector to obtain the target bitstream, which it then transmits to the receiving device. Correspondingly, the receiving device decodes the target bitstream to obtain a target feature vector, and further performs precoding based on this target feature vector. On the one hand, when transmitting channel information, the transmitting device only needs to send the target bitstream obtained by encoding the feature vector obtained from feature decomposition of the full channel information, which helps reduce CSI overhead. On the other hand, encoding the feature vector of the full channel information, rather than directly compressing and encoding the full channel information, can consider the correlation characteristics between channel information, which helps avoid compressing too much redundant information, improves compression efficiency, and thus improves coding performance.

[0215] Figure 18 This is a schematic structural diagram of a communication device 600 provided in an embodiment of this application. Figure 18 The communication device 600 shown includes a processor 610, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0216] Optionally, such as Figure 18 As shown, the communication device 600 may further include a memory 620. The processor 610 can retrieve and run computer programs from the memory 620 to implement the methods described in this embodiment.

[0217] The memory 620 can be a separate device independent of the processor 610, or it can be integrated into the processor 610.

[0218] Optionally, such as Figure 6 As shown, the communication device 600 may also include a transceiver 630, and the processor 610 may control the transceiver 630 to communicate with other devices. Specifically, it may send information or data to other devices or receive information or data sent by other devices.

[0219] The transceiver 630 may include a transmitter and a receiver. The transceiver 630 may further include antennas, and the number of antennas may be one or more.

[0220] Optionally, the communication device 600 may specifically be a transmitting device in the embodiments of this application, and the communication device 600 may implement the corresponding processes implemented by the transmitting device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0221] In some embodiments, the transceiver 630 in the communication device 600 can be used to implement Figure 16 The relevant operations of the communication module 410 in the transmitting device 400 shown are not described here for the sake of brevity.

[0222] In some embodiments, the processor 610 in the communication device 600 can be used to implement Figure 16 The operations of the processing module 420 and the encoding module 430 in the transmitting device 400 shown are not described in detail here for the sake of brevity.

[0223] Optionally, the communication device 600 may specifically be a receiving device in the embodiments of this application, and the communication device 600 may implement the corresponding processes implemented by the receiving device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0224] In some embodiments, the transceiver 630 in the communication device 600 can be used to implement Figure 17 The relevant operations of the communication module 510 in the receiving device 500 shown are not described here for the sake of brevity.

[0225] In some embodiments, the processor 610 in the communication device 600 can be used to implement Figure 17 The operations of the decoding module 520 and the processing module in the receiver device 500 shown are not described in detail here for the sake of simplicity.

[0226] Figure 19 This is a schematic structural diagram of the chip according to an embodiment of this application. Figure 19The chip 700 shown includes a processor 710, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0227] Optionally, such as Figure 19 As shown, chip 700 may further include memory 720. Processor 710 can retrieve and run computer programs from memory 720 to implement the methods described in this embodiment.

[0228] The memory 720 can be a separate device independent of the processor 710, or it can be integrated into the processor 710.

[0229] Optionally, the chip 700 may also include an input interface 730. The processor 710 can control the input interface 730 to communicate with other devices or chips; specifically, it can acquire information or data sent by other devices or chips.

[0230] Optionally, the chip 700 may also include an output interface 740. The processor 710 can control the output interface 740 to communicate with other devices or chips, specifically, to output information or data to other devices or chips.

[0231] Optionally, the chip can be applied to the transmitting device in the embodiments of this application, and the chip can implement the corresponding processes implemented by the transmitting device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0232] In some embodiments, the input interface 730 and output interface 740 in the chip 700 can be used to implement Figure 16 The relevant operations of the communication module 410 in the transmitting device 400 shown are not described here for the sake of brevity.

[0233] In some embodiments, the processor 710 in the chip 700 can be used to implement Figure 16 The operations of the processing module 420 and the encoding module 430 in the transmitting device 400 shown are not described in detail here for the sake of brevity.

[0234] Optionally, the chip can be applied to the receiving device in the embodiments of this application, and the chip can implement the corresponding processes implemented by the receiving device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0235] In some embodiments, the input interface 730 and output interface 740 in the chip 700 can be used to implement Figure 17 The relevant operations of the communication module 510 in the receiving device 500 shown are not described here for the sake of brevity.

[0236] In some embodiments, the processor 710 in the chip 700 can be used to implement Figure 17 The operations of the decoding module 520 and the processing module in the receiver device 500 shown are not described in detail here for the sake of simplicity.

[0237] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0238] Figure 20 This is a schematic block diagram of a communication system 900 provided in an embodiment of this application. Figure 20 As shown, the communication system 900 includes a transmitting device 910 and a receiving device 920.

[0239] The transmitting device 910 can be used to implement the corresponding functions implemented by the transmitting device in the above method, and the receiving device 920 can be used to implement the corresponding functions implemented by the receiving device in the above method. For the sake of brevity, these will not be elaborated here.

[0240] It should be understood that the processor in the embodiments of this application may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method embodiments can be completed by integrated logic circuits in the processor's hardware or by instructions in software form. The processor described above can be a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules can be located in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. The storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0241] It is understood that the memory in the embodiments of this application can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The 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), or flash memory. The 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), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct Rambus RAM (DR RAM). It should be noted that the memory used in the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0242] It should be understood that the above-described memory is exemplary and not a limiting description. For example, the memory in the embodiments of this application may also be static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DR RAM), etc. That is to say, the memory in the embodiments of this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0243] This application also provides a computer-readable storage medium for storing computer programs.

[0244] Optionally, the computer-readable storage medium can be applied to the transmitting device in the embodiments of this application, and the computer program causes the computer to execute the corresponding processes implemented by the transmitting device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0245] Optionally, the computer-readable storage medium can be applied to the receiving device in the embodiments of this application, and the computer program causes the computer to execute the corresponding processes implemented by the receiving device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0246] This application also provides a computer program product, including computer program instructions.

[0247] Optionally, the computer program product can be applied to the transmitting device in the embodiments of this application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the transmitting device in the various methods of the embodiments of this application. For the sake of brevity, they will not be described in detail here.

[0248] Optionally, the computer program product can be applied to the receiving device in the embodiments of this application, and the computer program instructions cause the computer to execute the corresponding processes implemented by the receiving device in the various methods of the embodiments of this application. For the sake of brevity, they will not be described in detail here.

[0249] This application also provides a computer program.

[0250] Optionally, the computer program can be applied to the sending device in the embodiments of this application. When the computer program is run on a computer, it causes the computer to execute the corresponding processes implemented by the sending device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0251] Optionally, the computer program can be applied to the receiving device in the embodiments of this application. When the computer program is run on a computer, it causes the computer to execute the corresponding processes implemented by the receiving device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0252] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0253] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0254] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0255] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0256] In addition, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0257] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0258] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for channel information feedback, characterized in that, include: The transmitting device receives the reference signal sent by the receiving device; Channel estimation is performed based on the reference signal to obtain channel information between the transmitting device and the receiving device. Performing feature decomposition on the channel information to obtain at least one first feature vector includes: performing Singular Value Decomposition (SVD) on the channel information to obtain the at least one first feature vector; The target bitstream is obtained by encoding the at least one first feature vector using a neural network; Send the target bit stream to the receiving device; Wherein, the channel information includes the channel information corresponding to each of the multiple subcarrier groups in the transmitting device, and the at least one first feature vector includes the feature vector corresponding to each of the multiple subcarrier groups. Encoding the at least one first feature vector using a neural network to obtain a target bitstream includes: The target bitstream is obtained by jointly encoding at least one first feature vector, which includes the feature vector corresponding to each of the plurality of subcarrier groups, using a neural network.

2. The method according to claim 1, characterized in that, The step of jointly encoding the at least one first feature vector through a neural network to obtain the target bitstream includes: The at least one first feature vector is concatenated into a feature vector matrix, which is then input into the neural network. The neural network encodes the feature vector matrix as an image to be compressed, thereby obtaining the target bitstream.

3. The method according to claim 2, characterized in that, The neural network is a convolutional neural network.

4. The method according to claim 1, characterized in that, The neural network is a recurrent neural network, and the step of jointly encoding the at least one first feature vector through the neural network to obtain the target bitstream includes: Each feature vector in the at least one first feature vector is sequentially input into the recurrent neural network; The target bitstream is obtained by encoding each feature vector as an element of the sequence using the recurrent neural network.

5. The method according to claim 4, characterized in that, The recurrent neural network includes a long short-term memory (LSTM) neural network.

6. The method according to any one of claims 1-5, characterized in that, The step of jointly encoding the at least one first feature vector through a neural network to obtain the target bitstream includes: The target bitstream is obtained by jointly encoding the at least one first feature vector based on an attention mechanism.

7. The method according to claim 6, characterized in that, The process of encoding the at least one first feature vector based on an attention mechanism to obtain the target bitstream includes: Based on the attention mechanism, the correlation features between elements in the at least one first feature vector are extracted to obtain at least one second feature vector; The target bitstream is obtained by performing feature compression on the at least one second feature vector.

8. The method according to any one of claims 1-5, characterized in that, The transmitting device is a terminal device, and the receiving device is a network device.

9. A method for channel information feedback, characterized in that, include: The receiving device receives a target bit stream sent by the transmitting device. The target bit stream is obtained by the transmitting device encoding at least one first feature vector. The at least one first feature vector is obtained by the transmitting device performing feature decomposition on channel information obtained from the channel estimation result, and is obtained by performing Singular Value Decomposition (SVD) on the channel information. The target bitstream is decoded using a neural network to obtain at least one target feature vector; The channel information includes channel information corresponding to each of the multiple subcarrier groups in the transmitting device, and the at least one first feature vector includes feature vectors corresponding to each of the multiple subcarrier groups. The target bit stream is obtained by the transmitting device through joint encoding of at least one first feature vector, which includes feature vectors corresponding to each of the multiple subcarrier groups.

10. The method according to claim 9, characterized in that, The step of decoding the target bitstream using a neural network to obtain at least one target feature vector includes: The target bitstream is input into the neural network; The target bitstream is decoded using the neural network as information obtained from encoding the image to obtain a target image, the target image including the at least one target feature vector.

11. The method according to claim 10, characterized in that, The neural network is a convolutional neural network.

12. The method according to claim 9, characterized in that, The neural network is a recurrent neural network. Decoding the target bitstream through the neural network to obtain at least one target feature vector includes: The target bitstream is input into the recurrent neural network; The target bitstream is decoded using the recurrent neural network as information obtained by encoding the sequence to obtain a target sequence, the target sequence including the at least one target feature vector.

13. The method according to claim 12, characterized in that, The recurrent neural network includes a long short-term memory (LSTM) neural network.

14. The method according to claim 9, characterized in that, The step of decoding the target bitstream using a neural network to obtain at least one target feature vector includes: The target bitstream is decoded based on an attention mechanism to obtain at least one target feature vector.

15. The method according to claim 14, characterized in that, The decoding of the target bitstream based on the attention mechanism to obtain at least one target feature vector includes: The target bitstream is subjected to feature extraction to obtain a first feature map; The weights of elements in the first feature map are determined based on an attention mechanism; The second feature map is obtained by multiplying the first feature map by the weights of the elements in the first feature map. The second feature map is decompressed to obtain the at least one target feature vector.

16. The method according to any one of claims 9-15, characterized in that, The method further includes: The receiving device performs precoding based on the at least one target feature vector.

17. A transmitting device, characterized in that, include: The communication module is used to receive reference signals sent by the receiving device. The processing module is configured to perform channel estimation based on the reference signal to obtain channel information between the transmitting device and the receiving device, and to perform feature decomposition on the channel information to obtain at least one first feature vector, wherein the processing module performs Singular Value Decomposition (SVD) on the channel information to obtain the at least one first feature vector. An encoding module is used to encode the at least one first feature vector through a neural network to obtain a target bitstream; The communication module is also used to: send the target bit stream to the receiving device; Wherein, the channel information includes the channel information corresponding to each of the multiple subcarrier groups in the transmitting device, and the at least one first feature vector includes the feature vector corresponding to each of the multiple subcarrier groups. The encoding module is specifically used to jointly encode at least one first feature vector, including the feature vector corresponding to each of the plurality of subcarrier groups, through a neural network to obtain the target bit stream.

18. The transmitting device according to claim 17, characterized in that, The encoding module is specifically used for: The at least one first feature vector is concatenated into a feature vector matrix, which is then input into the neural network. The neural network encodes the feature vector matrix as an image to be compressed, thereby obtaining the target bitstream.

19. The transmitting device according to claim 18, characterized in that, The neural network is a convolutional neural network.

20. The transmitting device according to claim 17, characterized in that, The neural network is a recurrent neural network, and the encoding module is further used for: Each feature vector in the at least one first feature vector is sequentially input into the recurrent neural network; The target bitstream is obtained by encoding each feature vector as an element of the sequence using the recurrent neural network.

21. The transmitting device according to claim 20, characterized in that, The recurrent neural network includes a long short-term memory (LSTM) neural network.

22. The transmitting end device according to any one of claims 17-21, characterized in that, The encoding module is also used for: The target bitstream is obtained by jointly encoding the at least one first feature vector based on an attention mechanism.

23. The transmitting device according to claim 22, characterized in that, The encoding module further includes: An attention module is used to extract correlation features between elements in the at least one first feature vector based on an attention mechanism to obtain at least one second feature vector. The feature vector compression module is used to perform feature compression on the at least one second feature vector to obtain the target bit stream.

24. The transmitting end device according to any one of claims 17-21, characterized in that, The transmitting device is a terminal device, and the receiving device is a network device.

25. A receiving device, characterized in that, include: A communication module is used to receive a target bit stream sent by a transmitting device. The target bit stream is obtained by the transmitting device encoding at least one first feature vector. The at least one first feature vector is obtained by the transmitting device performing feature decomposition on channel information obtained from the channel estimation result, and is obtained by performing Singular Value Decomposition (SVD) on the channel information. A decoding module is used to decode the target bitstream through a neural network to obtain at least one target feature vector; The channel information includes channel information corresponding to each of the multiple subcarrier groups in the transmitting device, and the at least one first feature vector includes feature vectors corresponding to each of the multiple subcarrier groups. The target bit stream is obtained by the transmitting device through joint encoding of at least one first feature vector, which includes feature vectors corresponding to each of the multiple subcarrier groups.

26. The receiving end device according to claim 25, characterized in that, The decoding module is used for: The target bitstream is input into the neural network; The target bitstream is decoded using the neural network as information obtained from encoding the image to obtain a target image, the target image including the at least one target feature vector.

27. The receiving device according to claim 26, characterized in that, The neural network is a convolutional neural network.

28. The receiving device according to claim 25, characterized in that, The neural network is a recurrent neural network, and the decoding module is further used for: The target bitstream is input into the recurrent neural network; The target bitstream is decoded using the recurrent neural network as information obtained by encoding the sequence to obtain a target sequence, the target sequence including the at least one target feature vector.

29. The receiving end device according to claim 28, characterized in that, The recurrent neural network includes a long short-term memory (LSTM) neural network.

30. The receiving end device according to any one of claims 25-29, characterized in that, The decoding module is also used for: The target bitstream is decoded based on an attention mechanism to obtain at least one target feature vector.

31. The receiving end device according to claim 30, characterized in that, The decoding module includes: The feature extraction module is used to extract features from the target bitstream to obtain a first feature map; An attention module is used to determine the weights of elements in the first feature map based on an attention mechanism; The dot product module is used to perform a dot product between the first feature map and the weights of the elements in the first feature map to obtain the second feature map; The decompression module is used to decompress the second feature map to obtain the at least one target feature vector.

32. The receiving end device according to any one of claims 25-29, characterized in that, The receiving device also includes: A processing module is used to pre-encode based on the at least one target feature vector.

33. A transmitting device, characterized in that, include: A processor and a memory for storing a computer program, the processor for calling and running the computer program stored in the memory to perform the method as described in any one of claims 1 to 8.

34. A chip, characterized in that, include: A processor for retrieving and running a computer program from memory, causing a device on which the chip is mounted to perform the method as described in any one of claims 1 to 8.

35. A computer-readable storage medium, characterized in that, Used to store a computer program that causes a computer to perform the method as described in any one of claims 1 to 8.

36. A computer program product, characterized in that, It includes computer program instructions that cause a computer to perform the method as described in any one of claims 1 to 8.

37. A network device, characterized in that, include: A processor and a memory for storing a computer program, the processor for calling and running the computer program stored in the memory to perform the method as described in any one of claims 9 to 16.

38. A chip, characterized in that, include: A processor for retrieving and running a computer program from memory, causing a device on which the chip is mounted to perform the method as described in any one of claims 9 to 16.

39. A computer-readable storage medium, characterized in that, Used to store a computer program that causes a computer to perform the method as described in any one of claims 9 to 16.

40. A computer program product, characterized in that, It includes computer program instructions that cause a computer to perform the method as described in any one of claims 9 to 16.

Images