Channel information feedback method, transmitting end device and receiving end device

Abstract

The embodiment of the invention provides a channel information feedback method, a transmitting end device and a receiving end device, channel information (such as CSI) feedback can adapt to different channel information input and output dimensions and different feedback overhead configurations, the feedback performance of the channel information is improved, and the feedback performance of the channel information is improved. And the flexibility and the expandability of the coding network and the decoding network in actual deployment are also improved. The channel information feedback method comprises the following steps: a sending end device encodes first channel information through an encoding network to obtain a first feedback bit stream; the sending end device sends the first feedback bit stream to the receiving end device; the first channel information is pre-processed channel information aligned with the target channel information in a first dimension, and / or the first feedback bit stream is a feedback bit stream after a feedback bit stream output by the coding network is post-processed and aligned with the target feedback bit stream; the first dimension is at least one of the following: the number of transmitting antenna ports, the number of sub-bands, the number of RBs, the number of delay paths, the number of symbols and the number of time slots.

Description

Channel information feedback method, transmitting device and receiving device Technical Field

[0001] The embodiments of the present application relate to the field of communications, and more specifically, to a method, a transmitting device, and a receiving device for channel information feedback. Background Art

[0002] In New Radio (NR) systems, channel state information (CSI) can be fed back based on a codebook. Specifically, based on high-layer signaling configuration, the estimated channel is used to select the optimal feedback matrix and corresponding feedback coefficient from the codebook on a periodic, aperiodic, or semi-continuous basis. However, since the codebook itself is a pre-defined finite set, the mapping process from the estimated channel to the channel in the codebook is quantization-lossy. Furthermore, a fixed codebook design cannot be dynamically adjusted based on channel changes, which reduces the accuracy of the feedback channel information and, in turn, degrades precoding performance.

[0003] Summary of the Invention

[0004] An embodiment of the present application provides a method for channel information feedback, a transmitting device, and a receiving device. The transmitting device can feedback channel information through a coding network, and the receiving device can obtain the channel information fed back by the transmitting device through a decoding network. Channel information (such as CSI) feedback can adapt to different channel information input and output dimensions and different feedback overhead configurations, thereby improving the feedback performance of channel information (such as CSI) and improving the flexibility and scalability of the coding network and the decoding network in actual deployment.

[0005] In a first aspect, a method for channel information feedback is provided, the method comprising:

[0006] The transmitting device encodes the first channel information through a coding network to obtain a first feedback bit stream; and

[0007] The transmitting device sends the first feedback bit stream to the receiving device;

[0008] The first channel information is channel information pre-processed to be aligned with the target channel information in the first dimension, and / or the first feedback bit stream is a feedback bit stream output by the coding network post-processed to be aligned with the target feedback bit stream;

[0009] The first dimension is at least one of the following: the number of transmitting antenna ports, the number of subbands, the number of resource blocks (RBs), the number of delay paths, the number of symbols, and the number of time slots.

[0010] In a second aspect, a method for channel information feedback is provided, the method comprising:

[0011] The receiving device receives the first feedback bit stream sent by the transmitting device;

[0012] The receiving device decodes the first feedback bit stream through a decoding network to obtain first channel information;

[0013] The first feedback bit stream is a feedback bit stream output by the encoding network corresponding to the decoding network after post-processing to align it with the target feedback bit stream, and / or the first channel information is channel information output by the decoding network that is aligned with the target channel information in the first dimension and is post-processed to be different from the target channel information in the first dimension;

[0014] The first dimension is at least one of the following: the number of transmitting antenna ports, the number of subbands, the number of resource blocks (RBs), the number of delay paths, the number of symbols, and the number of time slots.

[0015] In a third aspect, a transmitting device is provided for executing the method in the first aspect.

[0016] Specifically, the originating device includes a functional module for executing the method in the above-mentioned first aspect.

[0017] In a fourth aspect, a receiving device is provided for executing the method in the second aspect.

[0018] Specifically, the receiving device includes a functional module for executing the method in the above-mentioned second aspect.

[0019] In a fifth aspect, a sending device is provided, comprising a processor and a memory; the memory is used to store computer programs, and the processor is used to call and run the computer programs stored in the memory, so that the sending device executes the method in the above-mentioned first aspect.

[0020] In the sixth aspect, a receiving device is provided, comprising a processor and a memory; the memory is used to store computer programs, and the processor is used to call and run the computer programs stored in the memory, so that the receiving device executes the method in the above-mentioned second aspect.

[0021] In a seventh aspect, a device is provided for implementing the method in any one of the first to second aspects above.

[0022] Specifically, the apparatus includes: a processor, configured to call and run a computer program from a memory, so that a device equipped with the apparatus executes the method in any one of the first to second aspects described above.

[0023] In an eighth aspect, a computer-readable storage medium is provided for storing a computer program, wherein the computer program enables a computer to execute the method in any one of the first to second aspects above.

[0024] In a ninth aspect, a computer program product is provided, comprising computer program instructions, wherein the computer program instructions enable a computer to execute the method in any one of the first to second aspects above.

[0025] In a tenth aspect, a computer program is provided, which, when executed on a computer, enables the computer to execute the method in any one of the first to second aspects above.

[0026] Through the above technical solution, the transmitting device encodes the first channel information through the coding network to obtain a first feedback bit stream, and the transmitting device sends the first feedback bit stream to the receiving device, wherein the first channel information is the channel information that has been pre-processed to align with the target channel information in the first dimension, and / or the first feedback bit stream is the feedback bit stream output by the coding network that has been post-processed to align with the target feedback bit stream. That is, the coding network supports multiple types of first-dimensional channel information that have been pre-processed to align with the target channel information in the first dimension as input, and / or the coding network supports multiple types of feedback overhead feedback bit streams as output, and the transmitting device sends the feedback bit stream output by the coding network to the receiving device after it has been post-processed to align with the target feedback bit stream. The receiving device receives the first feedback bit stream sent by the transmitting device, and the receiving device decodes the first feedback bit stream through the decoding network to obtain first channel information, wherein the first feedback bit stream is the feedback bit stream output by the encoding network corresponding to the decoding network after post-processing and alignment with the target feedback bit stream, and / or the first channel information is the channel information output by the decoding network that is aligned with the target channel information in the first dimension and is different from the target channel information in the first dimension after post-processing. That is, the decoding network supports the feedback bit stream of multiple feedback overheads after pre-processing and alignment with the target feedback bit stream as input, and / or the output of the decoding network corresponds to multiple first-dimensional channel information after post-processing. That is, through the above technical solution, channel information (such as CSI) feedback can adapt to different channel information input and output dimensions and different feedback overhead configurations, thereby improving the feedback performance of channel information (such as CSI) and improving the flexibility and scalability of the encoding network and decoding network in actual deployment. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] FIG1 is a schematic diagram of a communication system architecture applied in an embodiment of the present application.

[0028] FIG2 is a schematic diagram of a neuron provided in the present application.

[0029] FIG3 is a schematic diagram of a neural network provided by the present application.

[0030] FIG4 is a schematic diagram of a convolutional neural network provided in this application.

[0031] FIG5 is a schematic diagram of an LSTM unit provided in this application.

[0032] FIG6 is a schematic framework diagram of an AI-based CSI autoencoder provided in this application.

[0033] FIG7 is a schematic flowchart of a method for channel information feedback provided according to an embodiment of the present application.

[0034] FIG8 is a schematic diagram of a hybrid iterative training for the same Config but different Payloads provided in an embodiment of the present application.

[0035] FIG9 is a schematic diagram of an embodiment of the present application in which the encoding network is the same and the decoding network is different under different Configs.

[0036] Figure 10 is a schematic diagram of hybrid iterative training with the same encoding network and different decoding networks under different Configs provided by an embodiment of the present application.

[0037] FIG11 is a schematic diagram of different encoding networks and the same decoding network under different Configs provided by an embodiment of the present application.

[0038] Figure 12 is a schematic diagram of hybrid iterative training with different encoding networks and the same decoding networks under different Configs provided by an embodiment of the present application.

[0039] FIG13 is a schematic diagram of a hybrid iterative training for different Configs and different Payloads provided in an embodiment of the present application.

[0040] FIG14 is a schematic diagram of an embodiment of the present application in which the encoding network is the same and the decoding network is different under different Configs and different Payloads.

[0041] Figure 15 is a schematic diagram of hybrid iterative training with the same encoding network and different decoding networks under different Configs and different Payloads provided by an embodiment of the present application.

[0042] FIG16 is a schematic diagram of different encoding networks and the same decoding network under different Configs and different Payloads provided by an embodiment of the present application.

[0043] Figure 17 is a schematic diagram of hybrid iterative training with different encoding networks and the same decoding network under different Configs and different Payloads provided by an embodiment of the present application.

[0044] Figure 18 is a schematic block diagram of a transmitting device provided according to an embodiment of the present application.

[0045] Figure 19 is a schematic block diagram of a receiving device provided according to an embodiment of the present application.

[0046] Figure 20 is a schematic block diagram of a communication device provided according to an embodiment of the present application.

[0047] Figure 21 is a schematic block diagram of a device provided according to an embodiment of the present application.

[0048] Figure 22 is a schematic block diagram of a communication system provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0049] The following will describe the technical solutions in the embodiments of this application in conjunction with the drawings in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of the embodiments. With respect to the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0050] The technical solutions of the embodiments of the present application can be applied to various communication systems, such as: Global System of 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, NR system evolution system, LTE-based access to unlicensed spectrum (LTE-U) system on unlicensed spectrum, NR-based access to unlicensed spectrum (NR-U) system on unlicensed spectrum, Non-Terrestrial Networks (NTN) system, Universal Mobile Telecommunication System (UMTS), Wireless Local Area Networks (WLAN), Internet of Things (IoT), Wireless Fidelity (WFI) system. Fidelity, WiFi), fifth-generation communication (5th-Generation, 5G) system, sixth-generation communication (6G) system or other communication systems.

[0051] Generally speaking, traditional communication systems 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 communications, but will also support, for example, device-to-device (D2D) communication, machine-to-machine (M2M) communication, machine type communication (MTC), vehicle-to-vehicle (V2V) communication, sidelink (SL) communication, vehicle-to-everything (V2X) communication, etc. The embodiments of the present application can also be applied to these communication systems.

[0052] In some embodiments, the communication system in the embodiments of the present application can be applied to a carrier aggregation (CA) scenario, a dual connectivity (DC) scenario, an independent (SA) networking scenario, or a non-standalone (NSA) networking scenario.

[0053] In some embodiments, the communication system in the embodiments of the present application can be applied to an unlicensed spectrum, where the unlicensed spectrum can also be considered as a shared spectrum; or, the communication system in the embodiments of the present application can also be applied to an authorized spectrum, where the authorized spectrum can also be considered as an unshared spectrum.

[0054] In some embodiments, the communication system in the embodiments of the present application can be applied to the FR1 frequency band (corresponding to the frequency band range of 410MHz to 7.125GHz), can also be applied to the FR2 frequency band (corresponding to the frequency band range of 24.25GHz to 52.6GHz), and can also be applied to new frequency bands such as high-frequency bands corresponding to the frequency band range of 52.6GHz to 71GHz or the frequency band range of 71GHz to 114.25GHz.

[0055] The embodiments of the present application describe various embodiments in conjunction with network devices and terminal devices, wherein 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.

[0056] The terminal device can be a station (ST) in a WLAN, a cellular phone, a cordless phone, a Session Initiation Protocol (SIP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA) device, a handheld device with wireless communication capabilities, a computing device or other processing device connected to a wireless modem, a vehicle-mounted device, a wearable device, a terminal device in a next-generation communication system such as an NR network, or a terminal device in a future evolved Public Land Mobile Network (PLMN) network, etc.

[0057] In an embodiment of the present application, the terminal device can be deployed on land, including indoors or outdoors, handheld, wearable or vehicle-mounted; it can also be deployed on the water surface (such as ships, etc.); it can also be deployed in the air (such as airplanes, balloons and satellites, etc.).

[0058] In an embodiment of the present application, the terminal device may be a mobile phone, a tablet computer, a computer with wireless transceiver function, 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, 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, an in-vehicle communication device, a wireless communication chip / application specific integrated circuit (ASIC) / system on chip (SoC), etc.

[0059] As an example and not a limitation, in the embodiment of the present application, the terminal device may also be a wearable device. Wearable devices may also be called wearable smart devices, which are a general term for wearable devices that are intelligently designed and developed using wearable technology for daily wear, such as glasses, gloves, watches, clothing, and shoes. A wearable device is a portable device that is worn directly on the body or integrated into the user's clothes or accessories. Wearable devices are not only hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are fully functional, large in size, and can achieve complete or partial functions without relying on smartphones, such as smart watches or smart glasses, as well as those that only focus on a certain type of application function and need to be used in conjunction with other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0060] In an embodiment of the present application, the network device may be a device for communicating with a mobile device. The network device may be an access point (AP) in WLAN, a base station (BTS) in GSM or CDMA, a base station (NodeB, NB) in WCDMA, an evolved base station (eNB or eNodeB) in LTE, or a relay station or access point, or a network device or base station (gNB) or a transmission reception point (TRP) in a vehicle-mounted device, a wearable device, and an NR network, or a network device in a future evolved PLMN network or a network device in an NTN network, etc.

[0061] As an example and not a limitation, in an embodiment of the present application, the network device may have a mobile feature, for example, the network device may be a mobile device. In some embodiments, 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. In some embodiments, the network device may also be a base station set up in a location such as land or water.

[0062] In an embodiment of the present application, the network device can provide services for a cell, and the terminal device communicates with the network device through the transmission resources used by the cell (for example, frequency domain resources, or spectrum resources). The cell can be a cell corresponding to the network device (for example, a base station). The cell can belong to a macro base station or a base station corresponding to a small cell. The small cells here may include: metro cells, micro cells, pico cells, femto cells, etc. These small cells have the characteristics of small coverage and low transmission power, and are suitable for providing high-speed data transmission services.

[0063] For example, a communication system 100 used in an embodiment of the present application is shown in FIG1 . 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 or terminal). The network device 110 may provide communication coverage for a specific geographic area and may communicate with terminal devices within the coverage area.

[0064] FIG1 exemplarily shows a network device and two terminal devices. In some embodiments, the communication system 100 may include multiple network devices and each network device may include other numbers of terminal devices within its coverage area, which is not limited in the embodiments of the present application.

[0065] In some embodiments, the communication system 100 may further include other network entities such as a network controller and a mobility management entity, which is not limited in the embodiments of the present application.

[0066] It should be understood that in the embodiments of the present application, a device having a communication function in a network / system may be referred to as a communication device. Taking the communication system 100 shown in FIG1 as an example, the communication device may include a network device 110 and a terminal device 120 having a communication function. The network device 110 and the terminal device 120 may be the specific devices described above and will not be described in detail here. The communication device may also include other devices in the communication system 100, such as a network controller, a mobility management entity, and other network entities, which are not limited in the embodiments of the present application.

[0067] It should be understood that the terms "system" and "network" are often used interchangeably herein. The term "and / or" is simply a description of an association between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " generally indicates that the related objects are in an "or" relationship.

[0068] It should be understood that this document relates to both transmitting and receiving devices. A transmitting device may be a terminal device, such as a mobile phone, machine facility, customer premises equipment (CPE), industrial equipment, or a vehicle. A receiving device may be a communication device on the other end of the transmitting device, such as a network device, mobile phone, industrial equipment, or a vehicle. In the embodiments of the present application, the transmitting device may be a terminal device, and the receiving device may be a network device (i.e., for uplink or downlink communication); alternatively, the transmitting device may be a first terminal, and the receiving device may be a second terminal (i.e., for sideline communication).

[0069] The terms used in the embodiments of this application are intended only to explain the specific embodiments of this application and are not intended to limit this application. The terms "first," "second," "third," and "fourth," etc. in the specification and claims of this application and the accompanying drawings are used to distinguish different objects, not to describe a specific order. In addition, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusions.

[0070] It should be understood that the "indication" mentioned in the embodiments of this application can be a direct indication, an indirect indication, or an indication of an association. For example, "A indicates B" can mean that A directly indicates B, for example, B can be obtained through A; it can also mean that A indirectly indicates B, for example, A indicates C, and B can be obtained through C; it can also mean that there is an association between A and B.

[0071] In the description of the embodiments of the present application, the term "corresponding" may indicate a direct or indirect correspondence between the two, or an association relationship between the two, or a relationship between indication and being indicated, configuration and being configured, etc.

[0072] In the embodiments of the present application, "pre-definition" or "pre-configuration" may be implemented by pre-storing corresponding codes, tables, or other methods that can be used to indicate relevant information in a device (e.g., a terminal device and a network device). The present application does not limit the specific implementation method. For example, pre-definition may refer to information defined in a protocol.

[0073] In the embodiments of the present application, the “protocol” may refer to a standard protocol in the communications field, for example, it may be an evolution of an existing LTE protocol, NR protocol, Wi-Fi protocol, or a protocol related to other communication systems. The present application does not limit the protocol type.

[0074] To facilitate understanding of the technical solutions of the embodiments of the present application, the technical solutions of the present application are described in detail below through specific embodiments. The following related technologies can be combined with the technical solutions of the embodiments of the present application as optional solutions, and they all fall within the scope of protection of the embodiments of the present application. The embodiments of the present application include at least part of the following contents.

[0075] To facilitate a better understanding of the embodiments of the present application, the neural network and machine learning (ML) related to the present application are explained.

[0076] A neural network is a computational model composed of multiple interconnected neuron nodes. The connections between nodes represent weighted values, called weights, from input signals to output signals. Each node performs a weighted summation (SUM) of different input signals and outputs them using a specific activation function (f). An example of a neuron structure is shown in Figure 2. A simple neural network, shown in Figure 3, consists of an input layer, a hidden layer, and an output layer. By using different connections between multiple neurons, weights, and activation functions, different outputs can be generated, thereby fitting the mapping relationship from input to output.

[0077] Deep learning utilizes deep neural networks with multiple hidden layers, significantly improving the network's ability to learn features and fitting complex, nonlinear mappings from input to output. Consequently, it has found widespread application in speech and image processing. In addition to deep neural networks, deep learning also includes other commonly used basic structures such as convolutional neural networks (CNNs) and recurrent neural networks (RNNs) for different tasks.

[0078] The basic structure of a convolutional neural network includes an input layer, multiple convolutional layers, multiple pooling layers, a fully connected layer, and an output layer, as shown in Figure 4. Each neuron in the convolution kernel of the convolutional layer is locally connected to its input, and the introduction of the pooling layer extracts the local maximum or average features of a certain layer, effectively reducing the network parameters and mining local features, enabling the convolutional neural network to converge quickly and achieve excellent performance.

[0079] RNNs are neural networks that model sequential data and have achieved remarkable success in natural language processing applications such as machine translation and speech recognition. Specifically, the network memorizes information from past moments and uses it in the calculation of current outputs. This means that nodes in the hidden layers are no longer disconnected but connected, and the input to a hidden layer includes not only the input layer but also the output of the previous hidden layer. Common RNN structures include long short-term memory (LSTM) and gated recurrent unit (GRU). Figure 5 shows a basic LSTM cell structure, which can include a tanh activation function. Unlike RNNs, which only consider the most recent state, the LSTM cell state determines which states should be retained and which should be forgotten, addressing the shortcomings of traditional RNNs in long-term memory.

[0080] To facilitate a better understanding of the embodiments of the present application, the codebook-based CSI feedback scheme in the NR system related to the present application is described.

[0081] In the current NR system, for the CSI feedback scheme, a codebook-based eigenvector feedback is usually used to enable the base station to obtain downlink CSI. Specifically, the base station sends a downlink channel state information reference signal (Channel State Information Reference Signal, CSI-RS) to the terminal, and the terminal uses the CSI-RS to estimate the CSI of the downlink channel, and performs eigenvalue decomposition on the estimated downlink channel to obtain the eigenvector corresponding to the downlink channel. Furthermore, the NR system provides two codebook design schemes, Type 1 and Type 2. The Type 1 codebook is used for CSI feedback with conventional accuracy, mainly for transmission in single-user multiple input multiple output (SU-MIMO) scenarios, and the Type 2 codebook is mainly used to improve the transmission performance of multi-user multiple input multiple output (MU-MIMO). Both Type 1 and Type 2 codebooks utilize a two-level codebook feedback mechanism, where W = W1W2. W1 describes the wideband, long-period characteristics of the channel, determining a set of L Discrete Fourier Transform (DFT) beams; W2 describes the subband, short-term characteristics of the channel. Specifically, for the Type 1 codebook, W2 selects a beam from the L DFT beams; for the Type 2 codebook, W2 linearly combines the L DFT beams in W1 and provides feedback in the form of amplitude and phase. Generally, the Type 2 codebook utilizes a higher number of feedback bits, achieving higher-precision CSI feedback performance.

[0082] To facilitate a better understanding of the embodiments of the present application, the CSI feedback method based on artificial intelligence (AI) related to the present application is described.

[0083] Given the tremendous success of AI technology, especially deep learning, in computer vision and natural language processing, the communications field has begun to explore the use of deep learning to address technical challenges that are difficult to solve with traditional communications methods, such as deep learning. The neural network architecture commonly used in deep learning is nonlinear and data-driven. It can extract features from actual channel matrix data and restore the channel matrix information compressed and fed back by the terminal side as much as possible on the base station side. This ensures the restoration of channel information while also providing the possibility of reducing CSI feedback overhead on the terminal side. Deep learning-based CSI feedback treats channel information as an image to be compressed, uses a deep learning autoencoder to compress and feed back the channel information, and reconstructs the compressed channel image at the transmitter, which can preserve the channel information to a greater extent.

[0084] Using an AI-based CSI autoencoder method, the entire feedback system is divided into an encoder and a decoder, which are deployed at the terminal transmitter and the base station receiver, respectively. After the terminal obtains the channel information through channel estimation, it compresses and encodes the channel information matrix through the encoder's neural network and feeds the compressed bit stream back to the base station through the air interface feedback link. The base station recovers the channel information based on the feedback bit stream through the decoder to obtain complete feedback channel information. The backbone network of the encoder and decoder shown in Figure 6 can adopt a deep neural network (DNN) composed of multiple layers of fully connected layers, a CNN composed of multiple layers of convolutional layers, or an RNN with structures such as LSTM and GRU. Various neural network architectures such as residual and self-attention mechanisms can also be used to improve the performance of the encoder and decoder.

[0085] The above-mentioned CSI input and CSI output can both be full channel information or feature vector information obtained based on full channel information. Therefore, the current deep learning-based channel information feedback methods are mainly divided into full channel information feedback and feature vector feedback. Although the former can achieve compression and feedback of full channel information, the feedback bit stream overhead is high. At the same time, this feedback method is not supported in existing NR systems. The feature vector-based feedback method is the feedback architecture currently supported by NR systems, and the AI-based feature vector feedback method can achieve higher CSI feedback accuracy with the same feedback bit overhead, or significantly reduce the feedback overhead while achieving the same CSI feedback accuracy.

[0086] In order to facilitate a better understanding of the embodiments of the present application, the problems solved by the present application are explained.

[0087] 1. Disadvantages of Codebook-Based CSI Feedback in NR

[0088] Currently, the CSI feedback in the 5G NR standard uses codebook-based feedback for Type 1 and Type 2. This codebook-based CSI feedback method has good generalization capabilities for different users and various channel scenarios, and can be flexibly configured to adapt to different numbers of transmit antenna ports, different numbers of subbands, and different feedback overheads. However, because the codebook is pre-set, it does not effectively utilize the correlation between different antenna ports and subbands, resulting in high feedback overhead and poor feedback performance.

[0089] 2. Disadvantages of AI-based CSI feedback

[0090] AI-based CSI feedback methods can extract the correlation of feature vectors in the time and frequency domains, thereby achieving good feedback performance with low feedback overhead. However, the encoder and decoder used in this solution both utilize a neural network architecture and require training on large-scale datasets. Therefore, during deployment, the actual number of antenna ports and subbands configured must be consistent with the dataset used during training, making it very inflexible.

[0091] For example, when constructing a dataset with 32 transmit antenna ports and 12 subbands, each sample in the dataset is 32*12*2=768 in size (where 2 represents the split between the real and imaginary parts). The CSI encoder and decoder models trained on this dataset use vectors or matrices with input and output lengths of 768, and are unable to effectively adapt to feature vector inputs for configurations with other numbers of transmit antenna ports and subbands. For example, in actual deployment, a configuration with 16 antenna ports and 12 subbands is used, and the input vector is 384 in length. The spatial and frequency domain features it contains are inconsistent with those in the training set, so the model obtained on the above training set is not applicable to this configuration.

[0092] At the same time, for feedback bit overhead, for example, for a channel eigenvector configured with 32 transmit antenna ports and 12 subbands, the feedback overhead configuration can have different configuration requirements from high to low (for example, from 300 bits to approximately 60 bits). For different feedback overheads, the CSI encoder and decoder models will also have different model parameters.

[0093] In actual deployment, different configurations may be generated for different UEs. If a different CSI autoencoder model is provided for each antenna port, subband configuration, and feedback overhead, the model storage overhead required on the network and UE sides will be too high. For the same UE, when a configuration switch occurs, the UE may need to download the encoder model from the network side, and the model download overhead generated by this process is also too high.

[0094] Therefore, how to solve the scalability problem of AI-based CSI autoencoders under different configurations during actual deployment is of great significance to the research on AI-based CSI feedback.

[0095] Based on the above problems, this application proposes a channel information feedback solution, in which the transmitting device can feedback channel information through the coding network, and the receiving device can obtain the channel information fed back by the transmitting device through the decoding network. The channel information (such as CSI) feedback can adapt to different channel information input and output dimensions and different feedback overhead configurations, thereby improving the feedback performance of channel information (such as CSI), and also improving the flexibility and scalability of the coding network and decoding network in actual deployment.

[0096] The technical solution of this application is described in detail below through specific embodiments.

[0097] FIG7 is a schematic flowchart of a method 200 for channel information feedback according to an embodiment of the present application. As shown in FIG7 , the method 200 for channel information feedback may include at least part of the following contents:

[0098] S210: The transmitting device encodes the first channel information through a coding network to obtain a first feedback bit stream;

[0099] S220, the transmitting device sends the first feedback bit stream to the receiving device; wherein the first channel information is channel information that has been pre-processed to be aligned with the target channel information in a first dimension, and / or the first feedback bit stream is a feedback bit stream output by the coding network that has been post-processed to be aligned with the target feedback bit stream; wherein the first dimension is at least one of the following: the number of transmit antenna ports, the number of subbands, the number of RBs, the number of delay paths, the number of symbols, and the number of time slots;

[0100] S230, the receiving device receives the first feedback bit stream sent by the transmitting device;

[0101] S240. The receiving device decodes the first feedback bit stream through a decoding network to obtain first channel information. The first feedback bit stream is a feedback bit stream output by an encoding network corresponding to the decoding network after post-processing to align it with a target feedback bit stream, and / or the first channel information is channel information output by the decoding network that is aligned with the target channel information in a first dimension and is post-processed to be different from the target channel information in the first dimension.

[0102] In an embodiment of the present application, a transmitting device may feed back channel information through a coding network; wherein the coding network supports multiple types of first-dimensional channel information as input after pre-processing and padding in the first dimension, and / or the coding network supports multiple types of feedback overhead feedback bit streams as output, and the transmitting device sends the feedback bit stream output by the coding network to the receiving device after post-processing and alignment with the target feedback bit stream.

[0103] In an embodiment of the present application, a receiving device obtains channel information fed back by a transmitting device through a decoding network; wherein the decoding network supports feedback bit streams of multiple feedback overheads that are pre-processed and aligned with a target feedback bit stream as input, and / or the output of the decoding network corresponds to multiple first-dimensional channel information after post-processing.

[0104] In some embodiments, when the first channel information is preprocessed to be aligned with the target channel information in a first dimension, the length of the feedback bit stream output by the coding network is the same for different physical resource configurations associated with the channel information feedback.

[0105] In some embodiments, when the first feedback bit stream is a feedback bit stream output by the coding network and is aligned with the target feedback bit stream after post-processing, for different channel information feedback overhead configurations, no preprocessing is required, and the channel information input to the coding network is the same in the first dimension.

[0106] In some embodiments, the channel information described in the embodiments of the present application may be CSI. Of course, it may also be other channel information, and the present application is not limited to this.

[0107] In an embodiment of the present application, the coding network can support multiple first-dimensional channel information as input after pre-processing and alignment with the target channel information in the first dimension, and / or the coding network supports multiple feedback overhead feedback bit streams as output, and the transmitting device sends the feedback bit stream output by the coding network to the receiving device after post-processing and alignment with the target feedback bit stream. That is, through the embodiment of the present application, channel information (such as CSI) feedback can adapt to different channel information input and output dimensions and different feedback overhead configurations, thereby improving the feedback performance of channel information (such as CSI) and also improving the flexibility and scalability of the coding network and decoding network in actual deployment.

[0108] The encoding network described in the embodiment of the present application is an AI / ML neural network, which can also be called an encoder or an autoencoder, or similar names, and the embodiment of the present application is not limited to this; the decoding network described in the embodiment of the present application is an AI / ML neural network corresponding to the encoding network, which can also be called a decoder or an autodecoder, or similar names, and the embodiment of the present application is not limited to this.

[0109] In some embodiments, the receiving device is a terminal device, and the emitting device is a network device; or, the receiving device is a network device, and the emitting device is a terminal device.

[0110] In some other embodiments, the receiving device is a terminal device, and the transmitting device is another terminal device. For example, the embodiments of the present application are applied to sidelink (SL) communication.

[0111] In some further embodiments, the transmitting device is a network device, and the receiving device is another network device. For example, the embodiments of the present application are applied to backhaul link communication.

[0112] In some embodiments, in the frequency domain, the first dimension may be not only the number of resource blocks (RBs), but also the number of subcarriers, the number of bandwidth parts (BWPs), or other frequency domain resource granularities.

[0113] In some embodiments, in the time domain, the first dimension can be not only the number of delay paths, the number of symbols, the number of time slots, etc., but also the number of microslots or minislots, the number of subframes, etc., or other time domain resource granularity.

[0114] The transmitting antenna port described in the embodiment of the present application can be a CSI-RS antenna port or other antenna port, and the present application does not limit this.

[0115] It should be noted that the first dimension may also be other resource granularities, which is not limited in this application.

[0116] In some embodiments, when the encoding network supports multiple types of first-dimensional channel information as input, the decoding network supports multiple types of first-dimensional channel information as output. Specifically, the encoding network corresponds to the decoding network, that is, the encoding network and the decoding network support the same type of first dimension.

[0117] In some embodiments, when the encoding network supports multiple feedback overhead feedback bitstreams as output, the decoding network supports multiple feedback overhead feedback bitstreams as input. Specifically, the encoding network corresponds to the decoding network, that is, the encoding network and the decoding network support the same type of feedback overhead.

[0118] In some embodiments, when the encoding network supports multiple first-dimensional channel information as input and the encoding network supports multiple feedback bit streams of feedback overhead as output, the decoding network supports multiple feedback bit streams of feedback overhead as input and the decoding network supports multiple first-dimensional channel information as output.

[0119] In some embodiments, when the transmitting device is a terminal device and the receiving device is a network device, the feedback bit stream output by the coding network can be carried by at least one of the following:

[0120] Physical Uplink Control Channel (PUCCH), Physical Uplink Shared Channel (PUSCH).

[0121] In some embodiments, when the transmitting device is a network device and the receiving device is a terminal device, the feedback bit stream output by the coding network can be carried by at least one of the following:

[0122] Physical Downlink Control Channel (PDCCH), Physical Downlink Shared Channel (PDSCH).

[0123] In some embodiments, when the transmitting device is a terminal device and the receiving device is another terminal device, the feedback bit stream output by the coding network can be carried by at least one of the following:

[0124] Physical Sidelink Control Channel (PSCCH), Physical Sidelink Shared Channel (PSSCH), Physical Sidelink Feedback Channel (PSFCH).

[0125] In some embodiments, the target channel information is the maximum channel information on the first dimension among the channel information corresponding to M physical resource configurations associated with the channel information feedback, where different physical resource configurations in the M physical resource configurations are different on the first dimension, M is a positive integer, and M ≥ 2. Optionally, the M physical resource configurations may be preconfigured.

[0126] In some embodiments, the target channel information is specific channel information among the channel information corresponding to the M physical resource configurations associated with the channel information feedback, which is not limited in the embodiments of the present application.

[0127] In some embodiments, the target channel information is configured by a network device, or the target channel information is agreed upon by a protocol.

[0128] In some embodiments, the target feedback bit stream is the maximum or minimum feedback bit stream among the feedback bit streams corresponding to the N channel information feedback overhead configurations, where N is a positive integer and N ≥ 2. Optionally, the N channel information feedback overhead configurations may be preconfigured.

[0129] In some embodiments, the target feedback bit stream is configured by a network device, or the target feedback bit stream is agreed upon by a protocol.

[0130] In some embodiments, the channel information input to the coding network may be a CSI matrix or a CSI vector. That is, the first channel information may be a CSI matrix or a CSI vector.

[0131] For example, the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmitting antenna ports, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of subbands, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of RBs, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of delay paths, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of symbols, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of time slots.

[0132] For another example, the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports and multiple numbers of subbands, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports and multiple numbers of RBs, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports and multiple numbers of delay paths, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports and multiple numbers of symbols, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports and multiple numbers of time slots.

[0133] For another example, the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of subbands and multiple numbers of RBs, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of subbands and multiple numbers of delay paths, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of subbands and multiple numbers of symbols, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of subbands and multiple numbers of time slots.

[0134] For another example, the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of RBs and multiple numbers of delay paths, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of RBs and multiple numbers of symbols, or the channel information of the input coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of RBs and multiple numbers of time slots.

[0135] For another example, the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple delay paths and multiple symbol numbers, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple delay paths and multiple time slot numbers.

[0136] For another example, the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports, multiple numbers of subbands, and multiple numbers of RBs, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports, multiple numbers of subbands, and multiple numbers of delay paths, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports, multiple numbers of subbands, and multiple numbers of time slots, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports, multiple numbers of subbands, and multiple numbers of symbols, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of subbands, multiple numbers of RBs, and multiple numbers of delay paths, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of subbands, multiple numbers of RBs, and multiple numbers of time slots, or the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of subbands, multiple numbers of RBs, and multiple numbers of symbols.

[0137] For another example, the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports, multiple numbers of subbands, multiple numbers of RBs, and multiple numbers of delay paths; or, the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports, multiple numbers of subbands, multiple numbers of RBs, and multiple numbers of symbols; or, the channel information input to the coding network may be CSI matrices or CSI vectors corresponding to multiple numbers of transmit antenna ports, multiple numbers of subbands, multiple numbers of RBs, and multiple numbers of time slots.

[0138] In the embodiment of the present application, different physical resource configurations associated with channel information feedback affect the CSI matrix or CSI vector input to the coding network and the CSI matrix or CSI vector output by the decoding network.

[0139] In the embodiment of the present application, the channel information feedback overhead configuration affects the feedback bit stream output by the encoding network and the feedback bit stream input by the decoding network.

[0140] In some embodiments, under different physical resource configurations associated with channel information feedback and / or under different channel information feedback overhead configurations, the coding network is the same, or the model weight parameters of the coding network are the same;

[0141] The different physical resource configurations associated with the channel information feedback are different in the first dimension.

[0142] In some embodiments, under different physical resource configurations associated with channel information feedback and / or under different channel information feedback overhead configurations, the decoding network is the same, or the model weight parameters of the decoding network are the same;

[0143] The different physical resource configurations associated with the channel information feedback are different in the first dimension.

[0144] In some embodiments, under different physical resource configurations associated with channel information feedback, and / or under different channel information feedback overhead configurations, the encoding network and the decoding network are the same, or the model weight parameters of the encoding network and the decoding network are the same.

[0145] For example, under different antenna ports, different numbers of subbands, and different feedback overhead configurations, the encoding network and the decoding network are the same.

[0146] In some embodiments, under different physical resource configurations associated with channel information feedback, and / or under different channel information feedback overhead configurations, the encoding networks are the same, or the model weight parameters of the encoding networks are the same, and the decoding networks are different, or the model weight parameters of the decoding networks are different.

[0147] For example, under different antenna ports, different numbers of subbands, and different feedback overhead configurations, the encoding network is the same but the decoding network is different.

[0148] In some embodiments, under different physical resource configurations associated with channel information feedback, and / or under different channel information feedback overhead configurations, the encoding networks are different, or the model weight parameters of the encoding networks are different, and the decoding networks are the same, or the model weight parameters of the decoding networks are the same.

[0149] For example, under different antenna ports, different numbers of subbands, and different feedback overhead configurations, the encoding network is different, but the decoding network is the same.

[0150] In some embodiments, the encoding network and the decoding network may be trained by a transmitting device, or the encoding network and the decoding network may be trained by a receiving device, or the encoding network and the decoding network may be trained by a specific server or network element, or the encoding network may be trained by a transmitting device and the decoding network may be trained by a receiving device.

[0151] In some embodiments, for M physical resource configurations associated with channel information feedback, the transmitting device preprocesses the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, thereby obtaining inputs of the coding network corresponding to the M physical resource configurations, respectively, where M is a positive integer and M ≥ 2. In this case, the first channel information may be the channel information that has been preprocessed to align with the target channel information in the first dimension.

[0152] In some embodiments, for the channel information corresponding to the M physical resource configurations, the channel information is aligned with the target channel information by adding first placeholder information in the first dimension during preprocessing. In postprocessing, the channel information is deleting the first placeholder information in the first dimension to obtain channel information that is different from the target channel information in the first dimension. Optionally, the first placeholder information is 0. Of course, the first placeholder information may also be other information, and this is not limited in the embodiments of the present application.

[0153] Specifically, for example, for channel information, the transmitting device aligns with the target channel information by continuously padding 0s backward in the first dimension, or the transmitting device aligns with the target channel information by continuously padding 0s forward in the first dimension, or the transmitting device aligns with the target channel information by interpolating and padding 0s in the first dimension.

[0154] In some embodiments, for the M physical resource configurations associated with channel information feedback, the receiving device post-processes the channel information output by the decoding network corresponding to the M physical resource configurations and aligned with the target channel information on the first dimension, so as to delete the first placeholder information in the channel information output by the decoding network corresponding to the M physical resource configurations, and obtain the channel information corresponding to the M physical resource configurations and different from the target channel information on the first dimension; wherein, the input of the coding network corresponding to the M physical resource configurations is obtained by filling the channel information corresponding to the M physical resource configurations with the first placeholder information on the first dimension, M is a positive integer, and M≥2.

[0155] For example, taking the eigenvectors on multiple subbands of the feedback CSI as an example, consider data sets with two different physical resource configurations associated with channel information feedback, denoted as Config 1 and Config 2, respectively, where Config 1 and Config 2 differ in the first dimension. Furthermore, consider two different feedback overhead configurations, Payload 1 and Payload 2, where the feedback bit stream length of Payload 1 is L1, the feedback bit stream length of Payload 2 is L2, and L1>L2. For data sets with two types of physical resource configurations and two different feedback overhead configurations, four combinations of physical resource configurations and feedback overhead configurations are generated, such as {Config x-Payload y}, where x and y can take the values ​​of 1 or 2, respectively. Specifically, for different Configs and the same Payload, the method of pre-processing the input of the encoding network and post-processing the output of the decoding network is adopted: for the input of the encoding network, Config 1 and Config 2 are padded in the first dimension. That is, if Config 1 is longer than Config 2 in the first dimension (for example, Config 1 is configured with 32 transmit antenna ports and Config 2 is configured with 16 transmit antenna ports), the input of Config 2 is padded with 0 in the first dimension to align with Config 1. For the output of the decoding network, corresponding post-processing is performed. That is, if Config 2 is padded with 0 in the first dimension, the decoding network is cut at the corresponding position, retaining only the part that is not padded with 0. Based on the above pre-processing and post-processing methods, the encoding network and decoding network can be directly trained by mixing two different types of physical resource configurations to obtain models adapted to Config 1 and Config 2.

[0156] In some embodiments, when the first feedback bit stream is the feedback bit stream output by the encoding network after post-processing and alignment with the target feedback bit stream (that is, the encoding network supports feedback bit streams of multiple feedback overheads as output), for N channel information feedback overhead configurations, the transmitting device can train the encoding network and the decoding network. Specifically, the transmitting device constructs N coding models and N decoding models respectively, wherein the N coding models correspond to the N decoding models respectively, the N coding models have the same model architecture, and the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; the transmitting device post-processes the feedback bit stream output by the i-th coding model and the feedback bit stream output by the i+k-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the i+k-th coding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream after alignment with the target feedback bit stream into the corresponding decoding model; within the j-th training cycle of model training, the transmitting device completes the training on the i-th coding model and its corresponding decoding model. Complete one training, and then copy the weight parameters of the i-th coding model and its corresponding decoding model to the i+k-th coding model and its corresponding decoding model respectively; in the j+1-th training cycle of model training, the transmitting device completes one training on the i+k-th coding model and its corresponding decoding model based on the copied weight parameters, and copies the weight parameters of the i+k-th coding model and its corresponding decoding model to the i-th coding model and its corresponding decoding model respectively; wherein, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

[0157] In some embodiments, when the first feedback bit stream is the feedback bit stream output by the coding network after post-processing and alignment with the target feedback bit stream (that is, the coding network supports feedback bit streams of multiple feedback overheads as output), for N channel information feedback overhead configurations, the receiving device can train the coding network and the decoding network. Specifically, for N channel information feedback overhead configurations, the receiving device constructs N coding models and N decoding models respectively, wherein the N coding models correspond to the N decoding models respectively, the N coding models have the same model architecture, and the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; the receiving device post-processes the feedback bit stream output by the i-th coding model and the feedback bit stream output by the i+k-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the i+k-th coding model with the target feedback bit stream, and the receiving device inputs the feedback bit stream after alignment with the target feedback bit stream into the corresponding decoding model; within the j-th training cycle of model training, the receiving device performs post-processing on the feedback bit stream output by the i-th coding model and its corresponding decoding model model, and then copy the weight parameters of the i-th coding model and its corresponding decoding model to the i+k-th coding model and its corresponding decoding model respectively; in the j+1-th training cycle of model training, the receiving device completes a training on the i+k-th coding model and its corresponding decoding model based on the copied weight parameters, and copies the weight parameters of the i+k-th coding model and its corresponding decoding model to the i-th coding model and its corresponding decoding model respectively; wherein, the i-th coding model or the i+k-th coding model after S training cycles is the coding network corresponding to the decoding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

[0158] In some embodiments, the feedback bit stream is aligned with the target feedback bit stream by adding second placeholder information during post-processing, or the feedback bit stream is aligned with the target feedback bit stream by truncating part of the bit stream during post-processing, or the feedback bit stream is aligned with the target feedback bit stream by deleting part of the bit stream during post-processing.

[0159] In some embodiments, the second placeholder information is 0 or 1.

[0160] It should be noted that the i-th coding model and its corresponding decoding model are mixed and trained iteratively with the i+k-th coding model and its corresponding decoding model. That is, in this embodiment, the N coding models are mixed and trained iteratively in pairs, and the N decoding models are mixed and trained iteratively in pairs, or, the N coding models perform serial mixed training iterations, and the N decoding models perform serial mixed training iterations (such as the first coding model and its corresponding decoding model and the second coding model and its corresponding decoding model perform mixed training iterations, the second coding model and its corresponding decoding model and the third coding model and its corresponding decoding model perform mixed training iterations, the third coding model and its corresponding decoding model and the fourth coding model and its corresponding decoding model perform mixed training iterations, and so on; of course, it can also be other serial methods, which are not limited to this in the embodiment of the present application), ensuring that each coding model and its corresponding decoding model undergo at least one mixed training iteration, thereby ensuring that the trained coding network and decoding network are adapted to the N channel information feedback overhead configurations.

[0161] For example, taking the feedback of eigenvectors on multiple CSI subbands as an example, consider data sets with two different physical resource configurations associated with channel information feedback, denoted as Configuration 1 (Config 1) and Configuration 2 (Config 2), with Config 1 and Config 2 differing in the first dimension. Furthermore, consider two different feedback overhead configurations, Payload 1 and Payload 2, where the feedback bit stream length of Payload 1 is L1 and the feedback bit stream length of Payload 2 is L2, with L1 > L2. For data sets with two types of physical resource configurations and two different feedback overhead configurations, four combinations of physical resource configurations and feedback overhead configurations are generated, such as {Config x - Payload y}, where x and y can take values ​​of 1 or 2, respectively. Specifically, for the same Config but different Payloads, bitstream truncation and hybrid iterative training methods are used to achieve scalable encoding and decoding networks. Taking Config 1 as an example, both Payload 1 and Payload 2 feedback bit overheads need to be adapted. A straightforward approach is to train the model on (Config1-Payload1). During deployment, if the feedback overhead for Payload2 is configured, a bitstream of length L2 is intercepted from the L1 bitstream for feedback. The remaining bits not fed back are directly set to 0 at the network-side CSI decoder input. However, this direct interception discards a significant amount of information because all CSI input information has already been distributed across the L1 bitstream by the encoder. This discarded information cannot be effectively retrieved by the CSI decoder, resulting in poor CSI performance recovered by the CSI decoder.To ensure that the intercepted L2 bitstream can also carry sufficient CSI information, this case proposes a bitstream interception and hybrid iterative training method, as shown in Figure 8. The hybrid training iterative method is as follows: For Payload1 and Payload2, two encoding models + decoding models are constructed, denoted as Model1 (encoding model + decoding model) and Model2 (encoding model + decoding model), respectively. Both use the same encoding model architecture and decoding model architecture. However, for Payload2 model Model2, only the first L2 bits are intercepted as valid input when bits are fed back. The unfed L1-L2 bits default to 0 when the CSI decoder input. In the jth training cycle of the training process, training is first completed on (Config1-Payload1), and the encoding model weights and decoding model weights of Model1 are copied to Model2 respectively. In the j+1th training cycle of the training process, Model2 is trained on (Config1-Payload2) based on the copied weights, and the encoding model weights and decoding model weights of Model2 are copied to Model1 respectively. The above process is iterated through multiple rounds of training. The weights of Model 1 and Model 2 are consistent, and this weight model can be retained as a configuration adapted to different payloads. This hybrid iterative training method enables the trained encoding and decoding networks to flexibly adapt to the feedback overhead configuration of L1 length and L2 length.

[0162] In an embodiment of the present application, when the first feedback bit stream is the feedback bit stream output by the coding network and is aligned with the target feedback bit stream after post-processing (that is, the coding network supports feedback bit streams of multiple feedback overheads as output), the coding network and the decoding network share the same model weight parameters. Taking into account the scalability of the coding network and the decoding network to adapt to multiple channel information feedback overhead configurations, there may be a certain performance loss in the recovery accuracy of the channel information (such as CSI). Therefore, it is considered to design different channel information feedback overhead configurations, with the coding network being the same (such as using the same model weight parameters) and the decoding network being different (such as using different model weight parameters). Specifically, as shown in Figure 9, still taking two different Payloads as an example, different Payloads share the same coding network, and since the decoding network is deployed on the receiving device, different model weight parameters can be used to further retrain the recovery accuracy of the channel information (such as CSI) under different Payloads.

[0163] In some embodiments, under different channel information feedback overhead configurations (that is, the coding network supports feedback bit streams of multiple feedback overheads as output), the coding networks are the same (such as using the same model weight parameters) and the decoding networks are different (such as using different model weight parameters). In this case, the transmitting device can, based on the coding network (that is, the coding network is the same under different channel information feedback overhead configurations, such as using the same model weight parameters) and the decoding network (that is, the decoding network is the same under different channel information feedback overhead configurations, such as using the same model weight parameters) obtained based on the hybrid iterative training method shown in 8 above, keep the coding network unchanged, and the transmitting device or the receiving device can re-train the decoding network corresponding to the coding network based on the N channel information feedback overhead configurations.

[0164] For example, hybrid iterative retraining: Based on the encoding network (i.e., the encoding network is the same under different payloads, such as using the same model weight parameters) and decoding network (i.e., the decoding network is the same under different payloads, such as using the same model weight parameters) obtained using the hybrid iterative training method shown in Figure 8, the decoding network can be retrained on different payloads by keeping the encoding network part unchanged. It should be noted that the different decoding networks only differ in the model weight parameters, but maintain the same model structure.

[0165] In some embodiments, under different channel information feedback overhead configurations (that is, the coding network supports feedback bit streams of multiple feedback overheads as output), the coding networks are the same (such as using the same model weight parameters), and the decoding networks are different (such as using different model weight parameters). In this case, the transmitting device can train the coding network based on hybrid iteration. Specifically, for N channel information feedback overhead configurations, the transmitting device constructs N coding models respectively, wherein the N coding models have the same model architecture, N is a positive integer, and N≥2; the transmitting device post-processes the feedback bit stream output by the i-th coding model and the feedback bit stream output by the i+k-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the i+k-th coding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into the decoding network; in the model In the j-th training cycle of model training, the transmitting device completes a training on the i-th coding model, and then copies the weight parameters of the i-th coding model to the i+k-th coding model; in the j+1-th training cycle of model training, the transmitting device completes a training on the i+k-th coding model based on the copied weight parameters, and copies the weight parameters of the i+k-th coding model to the i-th coding model; wherein, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, i, j, k, S are all positive integers, and i+k≤N.

[0166] It should be noted that the i-th coding model and the i+k-th coding model are mixed and trained iteratively, that is, in this embodiment, the N coding models are mixed and trained iteratively in pairs, or the N coding models perform serial mixed training iterations (such as the first coding model and the second coding model perform mixed training iterations, the second coding model and the third coding model perform mixed training iterations, the third coding model and the fourth coding model perform mixed training iterations, and so on; of course, it can also be other serial methods, which are not limited to this embodiment of the present application), ensuring that each coding model undergoes at least one two-by-two mixed training iteration, thereby ensuring that the trained coding network adapts to the N channel information feedback overhead configurations.

[0167] For example, taking the feedback of eigenvectors on multiple subbands of CSI as an example, consider data sets with two different physical resource configurations associated with channel information feedback, denoted as Configuration 1 (Config 1) and Configuration 2 (Config 2), with Config 1 and Config 2 differing in the first dimension. Furthermore, consider two different feedback overhead configurations, Payload 1 and Payload 2, where the feedback bit stream length of Payload 1 is L1 and the feedback bit stream length of Payload 2 is L2, with L1 > L2. For data sets with two types of physical resource configurations and two different feedback overhead configurations, four combinations of physical resource configurations and feedback overhead configurations are generated, such as {Config x - Payload y}, where x and y can take values ​​of 1 or 2, respectively. Specifically, for the same Config but different Payloads, bitstream truncation and hybrid iterative training methods are used to achieve a scalable coding network. Taking Config 1 as an example, it is necessary to adapt to both feedback bit overheads of Payload 1 and Payload 2. Specifically, as shown in Figure 10, the hybrid training iterative method is as follows: For Payload 1 and Payload 2, two encoding models are constructed, denoted as Model 1 (encoding model) and Model 2 (encoding model). Both use the same encoding model architecture, but for Payload 2, Model 2 only uses the first L2 bits as valid input when feeding back bits. The unfed L1-L2 bits are defaulted to 0 when inputting the decoding network. During the jth training cycle, training is first completed on (Config 1 - Payload 1), and the encoding model weights of Model 1 are copied to Model 2. During the j+1th training cycle, Model 2 is trained on (Config 1 - Payload 2) based on the copied weights, and the encoding model weights of Model 2 are copied to Model 1. This process is iterated through multiple rounds of training, and the weights of Model 1 and Model 2 are consistent. This weight model is retained as the configuration for different payloads. The receiving device can use different decoding networks for different payloads, where the model structures of different decoding networks can vary. Under this hybrid iterative training method, the trained coding network can flexibly adapt to the feedback overhead configuration of L1 length and L2 length.

[0168] Specifically, since the encoding network is deployed on the transmitting side (such as the UE side) and does not change with changes in the Payload configuration, there is no need for additional encoding network model storage and download overhead; while the decoding network is deployed on the receiving side (such as the network side), the adjustment of its model weights and model selection can be directly adjusted by the receiving device (such as the network device) according to the configuration of the currently served transmitting device (UE). This process is transparent to the transmitting device (UE) and can be decided by the receiving device (such as the network device). Under this working method, the receiving device (such as the network device) can still instruct the transmitting device (UE) on the Config and Payload and locally adapt the decoding network.

[0169] In an embodiment of the present application, when the first feedback bit stream is the feedback bit stream output by the coding network and is aligned with the target feedback bit stream after post-processing (that is, the coding network supports feedback bit streams of multiple feedback overheads as output), the coding network and the decoding network share the same model weight parameters. Taking into account the scalability of the coding network and the decoding network to adapt to multiple channel information feedback overhead configurations, there may be a certain performance loss in the recovery accuracy of the channel information (such as CSI). Therefore, it is possible to consider designing different channel information feedback overhead configurations, with different coding networks (such as using different model weight parameters) and the same decoding networks (such as using the same model weight parameters). Specifically, as shown in Figure 11, still taking two different Payloads as an example, different Payloads share the same decoding network, and since the coding network is deployed on the transmitting device, different model weight parameters can be used to further retrain the recovery accuracy of the channel information (such as CSI) under different Payloads.

[0170] In some embodiments, under different channel information feedback overhead configurations (that is, the coding network supports feedback bit streams of multiple feedback overheads as output), the coding networks are different (such as using different model weight parameters) and the decoding networks are the same (such as using the same model weight parameters). In this case, the receiving device can, based on the coding network (that is, the coding network is the same under different channel information feedback overhead configurations, such as using the same model weight parameters) and the decoding network (that is, the decoding network is the same under different channel information feedback overhead configurations, such as using the same model weight parameters) obtained based on the hybrid iterative training method shown in 8 above, while keeping the decoding network unchanged, the transmitting device or the receiving device can re-train the coding network based on the N channel information feedback overhead configurations.

[0171] For example, hybrid iterative retraining: Based on the encoding network (i.e., the encoding network is the same under different payloads, such as using the same model weight parameters) and decoding network (i.e., the decoding network is the same under different payloads, such as using the same model weight parameters) obtained using the hybrid iterative training method shown in Figure 8, the encoding network can be retrained on different payloads by keeping the decoding network part unchanged. It should be noted that the different encoding networks only differ in the model weight parameters, but maintain the same model structure.

[0172] In some embodiments, under different channel information feedback overhead configurations (that is, the coding network supports feedback bit streams of multiple feedback overheads as output), the coding networks are different (such as using different model weight parameters), and the decoding networks are the same (such as using the same model weight parameters). In this case, the receiving device can train the decoding network based on hybrid iteration. Specifically, for N channel information feedback overhead configurations, the receiving device constructs N decoding models respectively, wherein the N decoding models have the same model architecture, N is a positive integer, and N≥2; the receiving device aligns the feedback bit stream corresponding to the i-th decoding model with the target feedback bit stream, aligns the feedback bit stream corresponding to the i+k-th decoding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream after alignment with the target feedback bit stream into the corresponding decoding model; in the j-th training cycle of the model training, the receiving device A training is completed on the i-th decoding model, and then the weight parameters of the i-th decoding model are copied to the i+k-th decoding model; within the j+1-th training cycle of model training, the receiving device completes a training on the i+k-th decoding model based on the copied weight parameters, and copies the weight parameters of the i+k-th decoding model to the i-th decoding model; wherein, the i-th decoding model or the i+k-th decoding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

[0173] It should be noted that the i-th decoding model and the i+k-th decoding model are mixed and trained iteratively, that is, in this embodiment, the N decoding models are mixed and trained iteratively in pairs, or the N decoding models perform serial mixed training iterations (such as the first decoding model and the second decoding model perform mixed training iterations, the second decoding model and the third decoding model perform mixed training iterations, the third decoding model and the fourth decoding model perform mixed training iterations, and so on; of course, it can also be other serial methods, which are not limited to this embodiment of the present application), ensuring that each decoding model undergoes at least one two-by-two mixed training iteration, thereby ensuring that the trained decoding network adapts to the N channel information feedback overhead configurations.

[0174] For example, taking the feedback of eigenvectors on multiple subbands of CSI as an example, consider data sets with two different physical resource configurations associated with channel information feedback, denoted as Configuration 1 (Config 1) and Configuration 2 (Config 2), with Config 1 and Config 2 differing in the first dimension. Furthermore, consider two different feedback overhead configurations, Payload 1 and Payload 2, where the feedback bit stream length of Payload 1 is L1 and the feedback bit stream length of Payload 2 is L2, with L1 > L2. For data sets with two types of physical resource configurations and two different feedback overhead configurations, four combinations of physical resource configurations and feedback overhead configurations are generated, such as {Config x - Payload y}, where x and y can take values ​​of 1 or 2, respectively. Specifically, for the same Config but different Payloads, bitstream truncation and hybrid iterative training methods are used to achieve a scalable coding network. Taking Config 1 as an example, it is necessary to adapt to both feedback bit overheads of Payload 1 and Payload 2. Specifically, as shown in Figure 12, the hybrid training iterative method is as follows: For Payload1 and Payload2, two decoding models are constructed, denoted as Model1 (decoding model) and Model2 (decoding model). Both use the same decoding model architecture, but for Payload2's Model2, only the first L2 bits are intercepted as valid input when bits are fed back. The unfed L1-L2 bits are defaulted to 0 when the decoding model input is used. During the jth training cycle of the training process, training is first completed on (Config1-Payload1), and the decoding model weights of Model1 are copied to Model2. During the j+1th training cycle of the training process, Model2 is trained on (Config1-Payload2) based on the copied weights, and the decoding model weights of Model2 are copied to Model1. The above process is iterated for multiple rounds of training. The weights of Model1 and Model2 are consistent. The weight model is retained as the configuration adapted to different payloads. The transmitting device can use different coding networks for different payloads, where the model structures of different coding networks can be different. Under this hybrid iterative training method, the trained decoding network can flexibly adapt to the feedback overhead configuration of L1 length and L2 length.

[0175] Specifically, by utilizing a scalable decoding network, the receiving side (such as the network side) can support the transmitting side (such as the UE side) to implement different coding network designs, that is, the coding network can be UE-specific, and the coding network model can be transparent to the network side. Therefore, when the UE accesses a cell, the network can transmit the pre-trained coding network and decoding network models to different UE sides. When the UE subsequently updates the payload, different UEs can implement model updates and switching of the coding network through local retraining on the UE side. During this process, the decoding network on the network side remains unchanged and can adapt to the coding networks of different payloads of different UEs. When the UE side stores the coding network model weights corresponding to different payloads, when the network side indicates the UE with a joint or segmented payload through MAC CE and / or DCI, the UE will automatically switch the coding network model based on the correspondence between the payload and the coding network model to achieve better CSI feedback performance.

[0176] In some embodiments, when the first channel information is channel information that has been pre-processed to be aligned with the target channel information in a first dimension, and the first feedback bit stream is the feedback bit stream output by the encoding network that has been post-processed to be aligned with the target feedback bit stream, the transmitting device can train the encoding network and the decoding network. Specifically, for N channel information feedback overhead configurations, the transmitting device constructs N coding models and N decoding models respectively, wherein the N coding models correspond to the N decoding models respectively, the N coding models have the same model architecture, and the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; for M physical resource configurations associated with channel information feedback, the transmitting device preprocesses the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtains the input of the N coding models corresponding to the M physical resource configurations, wherein M is a positive integer, and M ≥ 2; the transmitting device post-processes the feedback bit stream output by the i-th coding model and the feedback bit stream output by the i+k-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the i+k-th coding model with the target feedback bit stream, and the transmitting device aligns the feedback bit stream output by the target The feedback bit stream after feedback bit stream alignment is input into the corresponding decoding model; within the j-th training cycle of model training, the transmitting device completes a training on the i-th coding model and its corresponding decoding model, and then copies the weight parameters of the i-th coding model and its corresponding decoding model to the i+k-th coding model and its corresponding decoding model respectively; within the j+1-th training cycle of model training, the transmitting device completes a training on the i+k-th coding model and its corresponding decoding model based on the copied weight parameters, and copies the weight parameters of the i+k-th coding model and its corresponding decoding model to the i-th coding model and its corresponding decoding model respectively; wherein, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network corresponding to the coding network, i, j, k, S are all positive integers, and i+k≤N.

[0177] In some embodiments, when the first channel information is channel information that has been pre-processed to be aligned with the target channel information in a first dimension, and the first feedback bit stream is the feedback bit stream output by the encoding network that has been post-processed to be aligned with the target feedback bit stream, the receiving device can train the encoding network and the decoding network. Specifically, for N channel information feedback overhead configurations, the receiving device constructs N coding models and N decoding models respectively, wherein the N coding models correspond to the N decoding models respectively, the N coding models have the same model architecture, and the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; for M physical resource configurations associated with channel information feedback, the receiving device aligns the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtains the input of the N coding models corresponding to the M physical resource configurations, wherein M is a positive integer, and M ≥ 2; the receiving device post-processes the feedback bit stream output by the i-th coding model and the feedback bit stream output by the i+k-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the i+k-th coding model with the target feedback bit stream, and the transmitting device aligns the feedback bit stream output by the target feedback bit stream. The feedback bit stream after the training is input into the corresponding decoding model; in the jth training cycle of model training, the receiving device completes a training on the i-th coding model and its corresponding decoding model, and then copies the weight parameters of the i-th coding model and its corresponding decoding model to the i+k-th coding model and its corresponding decoding model respectively; in the j+k-th training cycle of model training, the receiving device completes a training on the i+k-th coding model and its corresponding decoding model based on the copied weight parameters, and copies the weight parameters of the i+k-th coding model and its corresponding decoding model to the i-th coding model and its corresponding decoding model respectively; wherein, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

[0178] It should be noted that the i-th coding model and its corresponding decoding model are mixed and trained iteratively with the i+k-th coding model and its corresponding decoding model. That is, in this embodiment, the N coding models are mixed and trained iteratively in pairs, and the N decoding models are mixed and trained iteratively in pairs, or, the N coding models perform serial mixed training iterations, and the N decoding models perform serial mixed training iterations (such as the first coding model and its corresponding decoding model and the second coding model and its corresponding decoding model perform mixed training iterations, the second coding model and its corresponding decoding model and the third coding model and its corresponding decoding model perform mixed training iterations, the third coding model and its corresponding decoding model and the fourth coding model and its corresponding decoding model perform mixed training iterations, and so on; of course, it can also be other serial methods, which are not limited to this in the embodiment of the present application), ensuring that each coding model and its corresponding decoding model undergo at least one mixed training iteration, thereby ensuring that the trained coding network and decoding network are adapted to the N channel information feedback overhead configurations.

[0179] For example, taking the feedback of eigenvectors on multiple subbands of CSI as an example, consider data sets with two different physical resource configurations associated with channel information feedback, denoted as Configuration 1 (Config 1) and Configuration 2 (Config 2), with Config 1 and Config 2 differing in the first dimension. Furthermore, consider two different feedback overhead configurations, Payload 1 and Payload 2, where the feedback bit stream length of Payload 1 is L1 and the feedback bit stream length of Payload 2 is L2, with L1 > L2. For data sets with two types of physical resource configurations and two different feedback overhead configurations, four combinations of physical resource configurations and feedback overhead configurations are generated, such as {Config x - Payload y}, where x and y can take values ​​of 1 or 2, respectively. Specifically, for different Configs and different Payloads, bitstream truncation and hybrid iterative training methods are used to achieve scalable encoding and decoding networks. Config 1 and Config 2 need to adapt to the two feedback bit overheads of Payload 1 and Payload 2. As shown in FIG13 , the hybrid training iteration method is as follows: if Config 1 is longer than Config 2 in the first dimension (for example, Config 1 is configured with 32 transmit antenna ports and Config 2 is configured with 16 transmit antenna ports), Config 2 is pre-processed by adding 0 (for example, the input of Config 2 is added with 0 in the first dimension so that it is equal to Config 2). 1 alignment). Furthermore, two encoding models + decoding models are constructed for Config1+Payload1 and Config2+Payload2, respectively, denoted as Model1 (encoding model + decoding model) and Model2 (encoding model + decoding model). Both use the same encoding model architecture and decoding model architecture. However, for the Config2+Payload2 model Model2, when feeding back bits, only the first L2 bits are intercepted as valid input. The unfed L1-L2 defaults to 0 when input to the CSI decoder. In the jth training cycle of the training process, training is first completed on (Config1-Payload1), and the encoding model weights and decoding model weights of Model1 are copied to Model2. In the j+1th training cycle of the training process, Model2 is trained on (Config2-Payload2) based on the copied weights, and the encoding model weights and decoding model weights of Model2 are copied to Model1. The above process is iterated for multiple rounds of training. The weights of Model1 and Model2 are consistent, and this weight model can be retained as the configuration adapted to different payloads.Under this hybrid iterative training method, the trained encoding network and decoding network can flexibly adapt to the feedback overhead configuration of L1 length and L2 length.

[0180] In an embodiment of the present application, when the first channel information is the channel information after being pre-processed and aligned with the target channel information in the first dimension, and the first feedback bit stream is the feedback bit stream output by the coding network after being post-processed and aligned with the target feedback bit stream, the coding network and the decoding network share the same model weight parameters. Taking into account the scalability of the coding network and the decoding network to adapt to multiple channel information feedback overhead configurations, there may be a certain performance loss in the recovery accuracy of the channel information (such as CSI). Therefore, it is possible to consider designing different channel information feedback overhead configurations and different physical resource configurations associated with the channel information feedback, with the coding network being the same (such as using the same model weight parameters) and the decoding network being different (such as using different model weight parameters). Specifically, as shown in Figure 14, still taking two different Configs and two different Payloads as an example, a total of 4 groups, different Configs and Payloads share the same coding network, and the decoding network, since it is deployed on the receiving device, can use different model weight parameters to further retrain the recovery accuracy of the channel information (such as CSI) under different Configs and Payloads.

[0181] In some embodiments, under different channel information feedback overhead configurations and different physical resource configurations associated with channel information feedback (that is, the coding network supports feedback bit streams of multiple feedback overheads as output, and the coding network supports multiple first-dimensional channel information that is preprocessed and aligned with the target channel information in the first dimension as input), the coding networks are the same (such as using the same model weight parameters), and the decoding networks are different (such as using different model weight parameters). In this case, the transmitting device can, based on the coding network (that is, the coding network is the same under different channel information feedback overhead configurations, such as using the same model weight parameters) and the decoding network (that is, the decoding network is the same under different channel information feedback overhead configurations, such as using the same model weight parameters) obtained based on the hybrid iterative training method shown in 13 above, while keeping the coding network unchanged, the transmitting device or the receiving device can re-train the decoding network corresponding to the coding network based on N channel information feedback overhead configurations and M physical resource configurations associated with channel information feedback.

[0182] For example, retraining based on hybrid iteration: Based on the encoding network (i.e., the encoding network is the same under different Configs and Payloads, such as using the same model weight parameters) and decoding network (i.e., the decoding network is the same under different Configs and Payloads, such as using the same model weight parameters) obtained by the hybrid iterative training method shown in 13 above, the decoding network can be retrained on different Configs and Payloads by keeping the encoding network part unchanged. It should be noted that different decoding networks only have different model weight parameters, but maintain the same model structure.

[0183] In some embodiments, under different channel information feedback overhead configurations and different physical resource configurations associated with channel information feedback (that is, the coding network supports feedback bit streams of multiple feedback overheads as output, and the coding network supports multiple first-dimensional channel information pre-processed and aligned with the target channel information in the first dimension as input), the coding networks are the same (such as using the same model weight parameters) and the decoding networks are different (such as using different model weight parameters). In this case, the transmitting device can train the coding network based on hybrid iteration. Specifically,

[0184] In some embodiments, for N channel information feedback overhead configurations, the transmitting device constructs N coding models respectively, wherein the N coding models have the same model architecture, N is a positive integer, and N ≥ 2; for M physical resource configurations associated with channel information feedback, the transmitting device aligns the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtains the inputs of the N coding models corresponding to the M physical resource configurations, wherein M is a positive integer, and M ≥ 2; the transmitting device post-processes the feedback bit stream output by the i-th coding model and the feedback bit stream output by the i+k-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, and aligns the feedback bit stream output by the i+k-th coding model with the target feedback bit stream. The feedback bit stream is aligned with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into the decoding network; within the jth training cycle of model training, the transmitting device completes a training on the i-th coding model, and then copies the weight parameters of the i-th coding model to the i+k-th coding model; within the j+1th training cycle of model training, the transmitting device completes a training on the i+k-th coding model based on the copied weight parameters, and copies the weight parameters of the i+k-th coding model to the i-th coding model; wherein, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, i, j, k, S are all positive integers, and i+k≤N.

[0185] It should be noted that the i-th coding model and the i+k-th coding model are mixed and trained iteratively, that is, in this embodiment, the N coding models are mixed and trained iteratively in pairs, or the N coding models perform serial mixed training iterations (such as the first coding model and the second coding model perform mixed training iterations, the second coding model and the third coding model perform mixed training iterations, the third coding model and the fourth coding model perform mixed training iterations, and so on; of course, it can also be other serial methods, which are not limited to this embodiment of the present application), ensuring that each coding model undergoes at least one two-by-two mixed training iteration, thereby ensuring that the trained coding network adapts to the N channel information feedback overhead configurations.

[0186] For example, taking the feedback of eigenvectors on multiple subbands of CSI as an example, consider data sets with two different physical resource configurations associated with channel information feedback, denoted as Configuration 1 (Config 1) and Configuration 2 (Config 2), and Config 1 and Config 2 differ in the first dimension. Also, consider two different feedback overhead configurations, Payload 1 and Payload 2, where the feedback bit stream length of Payload 1 is L1 and the feedback bit stream length of Payload 2 is L2, and L1>L2. For data sets with two types of physical resource configurations and two different feedback overhead configurations, four combinations of physical resource configurations and feedback overhead configurations are generated, such as {Config x-Payload y}, where x and y can take values ​​of 1 or 2, respectively. Specifically, for different Configs and different Payloads, bitstream truncation and hybrid iterative training methods are used to achieve a scalable coding network. Config1 and Config2 need to adapt to the two feedback bit overheads of Payload 1 and Payload 2. Specifically, as shown in FIG15 , the hybrid training iteration method is as follows: If Config 1 is longer than Config 2 in the first dimension (for example, Config 1 is configured with 32 transmit antenna ports and Config 2 is configured with 16 transmit antenna ports), Config 2 is pre-processed by adding 0 (for example, the input of Config 2 is added with 0 in the first dimension so that it is equal to Config 2). 1 alignment), further, for Config1-Payload1 and Config2-Payload2, two coding models are constructed respectively, denoted as Model1 (coding model) and Model2 (coding model), and the two use the same coding model architecture, but for the model Model2 of Config2-Payload2, when feeding back bits, only the first L2 bits are intercepted as valid input, and the L1-L2 that is not fed back defaults to 0 when inputting the decoding network; in the j-th training cycle of the training process, first complete a training on (Config1-Payload1), and copy the coding model weights of Model1 to Model2; in the j+1-th training cycle of the training process, Model2 completes a training on (Config2-Payload2) based on the copied weights, and copy the coding model weights of Model2 to Model1. The above process is iterated for multiple rounds of training. The weights of Model 1 and Model 2 are consistent. The weight model can be retained as a configuration to adapt to different Configs and Payloads. The receiving device can use different decoding networks for different Configs and Payloads. The model structures of different decoding networks can be different.Under this hybrid iterative training method, the trained coding network can flexibly adapt to the feedback overhead configuration of L1 length and L2 length.

[0187] Specifically, since the encoding network is deployed on the transmitting side (such as the UE side) and does not change with changes in the Config and Payload configurations, no additional encoding network model storage and download overhead is required; while the decoding network is deployed on the receiving side (such as the network side), the adjustment of its model weights and model selection can be directly adjusted by the receiving device (such as the network device) according to the configuration of the currently served transmitting device (UE). This process is transparent to the transmitting device (UE) and can be decided by the receiving device (such as the network device). Under this working method, the receiving device (such as the network device) can still instruct the transmitting device (UE) on the Config and Payload and locally adapt the decoding network.

[0188] In an embodiment of the present application, when the coding network supports feedback bit streams of multiple feedback overheads as output, the decoding network supports feedback bit streams of multiple feedback overheads after pre-processing and padding as input, and the coding network supports multiple first-dimensional channel information after pre-processing and padding on the first dimension as input, and the output of the decoding network corresponds to multiple first-dimensional channel information after post-processing, the coding network and the decoding network share the same model weight parameters. Taking into account the scalability of the coding network and the decoding network to adapt to multiple channel information feedback overhead configurations, there may be a certain performance loss in the recovery accuracy of the channel information (such as CSI), therefore, it can be considered to design different channel information feedback overhead configurations and different physical resource configurations associated with the channel information feedback, with different coding networks (such as using different model weight parameters) and the same decoding networks (such as using the same model weight parameters). Specifically, as shown in Figure 16, taking four groups with two different configurations and two different payloads as an example, different configurations and payloads share the same decoding network. However, since the encoding network is deployed on the transmitting device, different model weight parameters can be used to further retrain the channel information (such as CSI) recovery accuracy under different configurations and payloads.

[0189] In some embodiments, under different channel information feedback overhead configurations and different physical resource configurations associated with channel information feedback (that is, the coding network supports feedback bit streams of multiple feedback overheads as output, and the output of the decoding network corresponds to multiple first-dimensional channel information after post-processing), the coding networks are different (such as using different model weight parameters) and the decoding networks are the same (such as using the same model weight parameters). In this case, the receiving device can, based on the coding network (that is, the coding network is the same under different channel information feedback overhead configurations, such as using the same model weight parameters) and the decoding network (that is, the decoding network is the same under different channel information feedback overhead configurations, such as using the same model weight parameters) obtained based on the hybrid iterative training method shown in 13 above, while keeping the decoding network unchanged, the transmitting device or the receiving device can re-train the coding network based on N channel information feedback overhead configurations and M physical resource configurations associated with the channel information feedback.

[0190] For example, retraining based on hybrid iteration: Based on the encoding network (i.e., the encoding network is the same under different Configs and Payloads, such as using the same model weight parameters) and decoding network (i.e., the decoding network is the same under different Configs and Payloads, such as using the same model weight parameters) obtained by the hybrid iterative training method shown in 13 above, the encoding network can be retrained on different Configs and Payloads by keeping the decoding network part unchanged. It should be noted that different encoding networks only have different model weight parameters, but maintain the same model structure.

[0191] In some embodiments, under different channel information feedback overhead configurations and different physical resource configurations associated with channel information feedback (that is, the coding network supports feedback bit streams of multiple feedback overheads as output, the decoding network supports feedback bit streams of multiple feedback overheads after pre-processing and padding as input, and the output of the decoding network corresponds to multiple first-dimensional channel information after post-processing), the coding networks are different (such as using different model weight parameters) and the decoding networks are the same (such as using the same model weight parameters). In this case, the receiving device can train the decoding network based on hybrid iteration. Specifically, for N channel information feedback overhead configurations, the receiving device constructs N decoding models respectively, wherein the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; for M physical resource configurations associated with channel information feedback, the receiving device preprocesses the channel information corresponding to the M physical resource configurations on the first dimension to align the channel information corresponding to the M physical resource configurations with the target channel information on the first dimension, and obtains the input of the coding network corresponding to the M physical resource configurations, wherein M is a positive integer, and M ≥ 2; the receiving device aligns the feedback bit stream corresponding to the i-th decoding model with the target feedback bit stream, and aligns the feedback bit stream corresponding to the i+k-th decoding model with the target feedback bit stream. The target feedback bit stream is aligned, and the transmitting device inputs the feedback bit stream after being aligned with the target feedback bit stream into the corresponding decoding model; within the j-th training cycle of model training, the receiving device completes a training on the i-th decoding model, and then copies the weight parameters of the i-th decoding model to the i+k-th decoding model; within the j+1-th training cycle of model training, the receiving device completes a training on the i+k-th decoding model based on the copied weight parameters, and copies the weight parameters of the i+k-th decoding model to the i-th decoding model; wherein, the i-th decoding model or the i+k-th decoding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

[0192] It should be noted that the i-th decoding model and the i+k-th decoding model are mixed and trained iteratively, that is, in this embodiment, the N decoding models are mixed and trained iteratively in pairs, or the N decoding models perform serial mixed training iterations (such as the first decoding model and the second decoding model perform mixed training iterations, the second decoding model and the third decoding model perform mixed training iterations, the third decoding model and the fourth decoding model perform mixed training iterations, and so on; of course, it can also be other serial methods, which are not limited to this embodiment of the present application), ensuring that each decoding model undergoes at least one two-by-two mixed training iteration, thereby ensuring that the trained decoding network adapts to the N channel information feedback overhead configurations.

[0193] For example, taking the feedback of eigenvectors on multiple subbands of CSI as an example, consider data sets with two different physical resource configurations associated with channel information feedback, denoted as Configuration 1 (Config 1) and Configuration 2 (Config 2), where Config 1 and Config 2 differ in the first dimension. Furthermore, consider two different feedback overhead configurations, Payload 1 and Payload 2, where the feedback bit stream length of Payload 1 is L1 and the feedback bit stream length of Payload 2 is L2, with L1 > L2. For data sets with two types of physical resource configurations and two different feedback overhead configurations, four combinations of physical resource configurations and feedback overhead configurations are generated, such as {Config x - Payload y}, where x and y can take values ​​of 1 or 2, respectively. Specifically, for different Configs and different Payloads, a scalable coding network is implemented using bitstream truncation and hybrid iterative training methods. Config 1 and Config 2 need to adapt to the two feedback bit overheads of Payload 1 and Payload 2. Specifically, as shown in FIG17 , the hybrid training iteration method is as follows: If Config 1 is longer than Config 2 in the first dimension (for example, Config 1 is configured with 32 transmit antenna ports and Config 2 is configured with 16 transmit antenna ports), Config 2 is pre-processed by adding 0 (for example, the input of Config 2 is added with 0 in the first dimension so that it is equal to Config 2). 1 alignment), further, for Config1-Payload1 and Config2-Payload2, two decoding models are constructed, denoted as Model1 (decoding model) and Model2 (decoding model), respectively. The two use the same decoding model architecture, but for the model Model2 of Config2-Payload2, when feeding back bits, only the first L2 bits are intercepted as valid input, and the L1-L2 that is not fed back defaults to 0 when inputting the decoding model; in the j-th training cycle of the training process, first complete a training on (Config1-Payload1), and at the same time, copy the decoding model weights of Model1 to Model2; in the j+1-th training cycle of the training process, Model2 completes a training on (Config2-Payload2) based on the copied weights, and copy the decoding model weights of Model2 to Model1. The above process is iterated for multiple rounds of training. The weights of Model 1 and Model 2 are consistent. The weight model can be retained as a configuration adapted to different Configs and Payloads. The transmitting device can use different encoding networks for different Configs and Payloads, where the model structures of different encoding networks can be different.Under this hybrid iterative training method, the trained decoding network can flexibly adapt to the feedback overhead configuration of L1 length and L2 length.

[0194] Specifically, by utilizing a scalable decoding network, the receiving side (such as the network side) can support the transmitting side (such as the UE side) to implement different coding network designs, that is, the coding network can be UE-specific, and the coding network model can be transparent to the network side. Therefore, when the UE accesses a cell, the network can transmit the pre-trained coding network and decoding network models to different UE sides. When the UE subsequently updates the Config and Payload, different UEs can update and switch the coding network model through local retraining on the UE side. During this process, the decoding network on the network side remains unchanged and can adapt to the coding networks of different Configs and Payloads of different UEs. When the UE side stores the coding network model weights corresponding to different Configs and Payloads, when the network side indicates the Config and Payload to the UE in a joint or segmented manner through MAC CE and / or DCI, the UE will automatically switch the coding network model based on the correspondence between the (Config-Payload) combination and the coding network model to achieve better CSI feedback performance.

[0195] It should be noted that the operations of interception and 0-filling in the above embodiments can all be implemented in different ways. For example, 0 can be continuously padded backward in the first dimension, or padded forward, or padded by interpolation. Different implementation methods require corresponding different models, but their performance will not be different. Therefore, the embodiment of the present application mainly uses continuous backward 0-filling as the implementation method, but other interception and 0-filling implementation methods are all within the protection scope of this application. At the same time, for the L1-L2 bit streams that are not fed back, the default is 0 on the decoding network side, but it can also be implemented as other methods such as defaulting to 1. The embodiment of the present application mainly supports the implementation method of defaulting to 0, but other default completion methods for non-feedback bit streams are also within the protection scope of this application.

[0196] In some embodiments, based on the above embodiments, a scalable coding network model and decoding network model can be obtained to adapt to different channel information feedback overhead configurations and different physical resource configurations associated with channel information feedback. The coding network is shared by all transmitting devices (such as UE) in the cell, that is, when different transmitting devices (such as UE) have different configurations, they share the model weight parameters of the coding network; at the same time, when the channel environments of different cells are similar, the coding network model can also be shared between cells. When the transmitting device (such as UE) enters a new cell, there is no need to re-download the coding network model. For the same coding network model, since different pre-processing, post-processing and feedback bit interception operations are required when adapting to different Configs and Payloads, the receiving device (such as a network device) needs to indicate the Config and Payload configurations to the transmitting device (such as a UE) through MAC CE or DCI signaling. The transmitting device (such as a UE) pre-processes the CSI input according to the indication and intercepts the corresponding bit stream for feedback.

[0197] In some embodiments, the transmitting device receives first information, wherein the first information is used to indicate at least one of the following: physical resource configuration information associated with channel information feedback, and channel information feedback overhead configuration information. Optionally, different physical resource configurations associated with the channel information feedback differ in a first dimension.

[0198] For example, the transmitting device receives the first information sent by the receiving device.

[0199] For another specific example, the transmitting device receives the first information sent by other devices except the receiving device.

[0200] Optionally, the physical resource configuration information associated with the channel information feedback includes M physical resource configurations associated with the channel information feedback, or the physical resource configuration information associated with the channel information feedback is an identifier or index of a physical resource configuration among the M physical resource configurations associated with the pre-configured channel information feedback, where M is a positive integer and M≥2.

[0201] Optionally, the channel information feedback overhead configuration information includes N channel information feedback overhead configurations, or the channel information feedback overhead configuration information is an identifier or index of a channel information feedback overhead configuration among the N pre-configured channel information feedback overhead configurations, where N is a positive integer and N≥2.

[0202] In some embodiments, the first information is carried by at least one of the following signaling:

[0203] Radio Resource Control (RRC) signaling, Media Access Control Control Element (MAC CE), Downlink Control Information (DCI), Sidelink Control Information (SCI).

[0204] In some embodiments, the first information includes a first information field; wherein the first information field is used to jointly indicate physical resource configuration information associated with the channel information feedback and the channel information feedback overhead configuration information.

[0205] For example, the first information field is used for joint indication (Config x-Payload y), where Config represents the physical resource configuration associated with the channel information feedback, and Payload represents the channel information feedback overhead configuration. That is, an indication field indicating (x, y) is added to the first information, and the configuration table of the pre-configured common W group (Config x-Payload y) is adopted. bits to indicate (here ( indicates rounding up.) It should be noted that the advantage of joint indication is that different numbers of payload configurations can be implemented for different configurations. For example, for configurations with a smaller first dimension, the allowable payload is lower, and larger payload feedback is not supported. Conversely, for configurations with a larger first dimension, the allowable payload is larger, and smaller payload feedback is not supported.

[0206] In some embodiments, the first information includes a second information field and a third information field; wherein the second information field is used to indicate physical resource configuration information associated with the channel information feedback, and the third information field is used to indicate the channel information feedback overhead configuration information.

[0207] For example, the second information field indicates Config x, where Config represents the physical resource configuration associated with the channel information feedback; the third information field indicates Payload y, where Payload represents the channel information feedback overhead configuration. That is, the indication fields indicating x and y are added to the first information, where Config has pre-configured X groups and Payload has pre-configured Y groups, respectively. and The advantage of segmented indication is that it can achieve more flexible Config and Payload signaling scheduling.

[0208] In some embodiments, the transmitting device receives second information and third information; wherein the second information is used to indicate physical resource configuration information associated with the channel information feedback, and the third information is used to indicate channel information feedback overhead configuration information. Optionally, different physical resource configurations associated with the channel information feedback differ in a first dimension.

[0209] For example, the transmitting device receives the second information and the third information sent by the receiving device.

[0210] For another specific example, the transmitting device receives the second information and the third information sent by other devices except the receiving device.

[0211] In some embodiments, the second information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI; and / or, the third information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI.

[0212] Therefore, in an embodiment of the present application, a transmitting device encodes the first channel information through a coding network to obtain a first feedback bit stream, and the transmitting device sends the first feedback bit stream to a receiving device, wherein the first channel information is channel information that has been pre-processed to be aligned with the target channel information in the first dimension, and / or the first feedback bit stream is a feedback bit stream that has been post-processed and aligned with the target feedback bit stream output by the coding network. That is, the coding network supports multiple types of first-dimensional channel information that have been pre-processed and aligned with the target channel information in the first dimension as input, and / or the coding network supports multiple types of feedback overhead feedback bit streams as output, and the transmitting device sends the feedback bit stream that has been post-processed and aligned with the target feedback bit stream output by the coding network to the receiving device. The receiving device receives the first feedback bit stream sent by the transmitting device, and the receiving device decodes the first feedback bit stream through the decoding network to obtain first channel information, wherein the first feedback bit stream is the feedback bit stream output by the encoding network corresponding to the decoding network after post-processing and alignment with the target feedback bit stream, and / or the first channel information is the channel information output by the decoding network that is aligned with the target channel information in the first dimension and is different from the target channel information in the first dimension after post-processing. That is, the decoding network supports the feedback bit stream of multiple feedback overheads after pre-processing and alignment with the target feedback bit stream as input, and / or the output of the decoding network corresponds to multiple first-dimensional channel information after post-processing. That is, through the above technical solution, channel information (such as CSI) feedback can adapt to different channel information input and output dimensions and different feedback overhead configurations, thereby improving the feedback performance of channel information (such as CSI) and improving the flexibility and scalability of the encoding network and decoding network in actual deployment.

[0213] The above text, in combination with Figures 7 to 17, describes in detail the method embodiment of the present application. The following text, in combination with Figures 18 to 22, describes in detail the device embodiment of the present application. It should be understood that the device embodiment and the method embodiment correspond to each other, and similar descriptions can refer to the method embodiment.

[0214] Figure 18 shows a schematic block diagram of a transmitting device 300 according to an embodiment of the present application. As shown in Figure 18, the transmitting device 300 includes:

[0215] The processing unit 310 is configured to encode the first channel information through a coding network to obtain a first feedback bit stream;

[0216] The communication unit 320 is configured to send the first feedback bit stream to a receiving device;

[0217] The first channel information is channel information pre-processed to be aligned with the target channel information in the first dimension, and / or the first feedback bit stream is a feedback bit stream output by the coding network post-processed to be aligned with the target feedback bit stream;

[0218] The first dimension is at least one of the following: the number of transmitting antenna ports, the number of subbands, the number of resource blocks (RBs), the number of delay paths, the number of symbols, and the number of time slots.

[0219] In some embodiments, under different physical resource configurations associated with channel information feedback and / or under different channel information feedback overhead configurations, the coding network is the same, or the model weight parameters of the coding network are the same;

[0220] The different physical resource configurations associated with the channel information feedback are different in the first dimension.

[0221] In some embodiments, when the coding network supports multiple types of first-dimensional channel information and uses the pre-processed channel information padded in the first dimension as input, for M physical resource configurations associated with the channel information feedback, the processing unit 310 is used to pre-process the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtain the inputs of the coding network corresponding to the M physical resource configurations, where M is a positive integer and M≥2.

[0222] In some embodiments, when the first feedback bit stream is a feedback bit stream output by the coding network and aligned with a target feedback bit stream after post-processing, for N channel information feedback overhead configurations, the processing unit 310 is further configured to respectively construct N coding models and N decoding models, wherein the N coding models respectively correspond to the N decoding models, the N coding models have the same model architecture, and the N decoding models have the same model architecture, and N is a positive integer, and N ≥ 2;

[0223] The processing unit 310 is further configured to perform post-processing on the feedback bit stream output by the i-th coding model and the feedback bit stream output by the (i+k)-th coding model, so as to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and input the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model;

[0224] During the j-th training cycle of the model training, the processing unit 310 is further configured to complete a training on the i-th encoding model and its corresponding decoding model, and then copy the weight parameters of the i-th encoding model and its corresponding decoding model to the i+k-th encoding model and its corresponding decoding model respectively;

[0225] In the j+1th training cycle of the model training, the processing unit 310 is further configured to complete a training on the i+kth encoding model and its corresponding decoding model based on the copied weight parameters, and copy the weight parameters of the i+kth encoding model and its corresponding decoding model to the ith encoding model and its corresponding decoding model respectively;

[0226] Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network corresponding to the coding network, i, j, k, S are all positive integers, and i+k≤N.

[0227] In some embodiments, while keeping the encoding network unchanged, the processing unit 310 is further configured to retrain the decoding network corresponding to the encoding network based on the N channel information feedback overhead configurations.

[0228] In some embodiments, when the first feedback bit stream is a feedback bit stream output by the coding network that is aligned with a target feedback bit stream after post-processing, the processing unit 310 is further configured to construct N coding models for N channel information feedback overhead configurations, wherein the N coding models have the same model architecture, N is a positive integer, and N ≥ 2;

[0229] The processing unit 310 is further configured to post-process the feedback bit stream output by the i-th coding model and the feedback bit stream output by the (i+k)-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and input the feedback bit stream aligned with the target feedback bit stream into the decoding network.

[0230] In the jth training cycle of the model training, the processing unit 310 is further configured to complete a training on the i-th coding model, and then copy the weight parameters of the i-th coding model to the (i+k)-th coding model;

[0231] In the j+1th training cycle of the model training, the processing unit 310 is further configured to complete a training on the i+kth coding model based on the copied weight parameters, and copy the weight parameters of the i+kth coding model to the i-th coding model;

[0232] Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, i, j, k, S are all positive integers, and i+k≤N.

[0233] In some embodiments, when the first channel information is channel information that is pre-processed to be aligned with the target channel information in the first dimension, and the first feedback bit stream is a feedback bit stream that is post-processed and aligned with the target feedback bit stream output by the coding network, for N channel information feedback overhead configurations, the processing unit 310 is further configured to respectively construct N coding models and N decoding models, wherein the N coding models respectively correspond to the N decoding models, the N coding models have the same model architecture, the N decoding models have the same model architecture, and N is a positive integer, and N ≥ 2;

[0234] For M physical resource configurations associated with channel information feedback, the processing unit 310 is further configured to align channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtain inputs of the N coding models corresponding to the M physical resource configurations, respectively, where M is a positive integer and M ≥ 2;

[0235] The processing unit 310 is further configured to perform post-processing on the feedback bit stream output by the i-th coding model and the feedback bit stream output by the (i+k)-th coding model, so as to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and input the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model;

[0236] During the j-th training cycle of the model training, the processing unit 310 is further configured to complete a training on the i-th encoding model and its corresponding decoding model, and then copy the weight parameters of the i-th encoding model and its corresponding decoding model to the i+k-th encoding model and its corresponding decoding model respectively;

[0237] In the j+1th training cycle of the model training, the processing unit 310 is further configured to complete a training on the i+kth encoding model and its corresponding decoding model based on the copied weight parameters, and copy the weight parameters of the i+kth encoding model and its corresponding decoding model to the ith encoding model and its corresponding decoding model respectively;

[0238] Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network corresponding to the coding network, i, j, k, S are all positive integers, and i+k≤N.

[0239] In some embodiments, while keeping the encoding network unchanged, the processing unit 310 is further configured to retrain the decoding network corresponding to the encoding network based on the N channel information feedback overhead configurations and the M physical resource configurations.

[0240] In some embodiments, when the first channel information is channel information pre-processed to be aligned with the target channel information in the first dimension, and the first feedback bit stream is a feedback bit stream output by the coding network that is post-processed to be aligned with the target feedback bit stream, for N channel information feedback overhead configurations, the processing unit 310 is further configured to respectively construct N coding models, wherein the N coding models have the same model architecture, N is a positive integer, and N ≥ 2;

[0241] For M physical resource configurations associated with the channel information feedback, the processing unit 310 is further configured to preprocess the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtain inputs of the N coding models corresponding to the M physical resource configurations, respectively, where M is a positive integer and M ≥ 2;

[0242] The processing unit 310 is further configured to post-process the feedback bit stream output by the i-th coding model and the feedback bit stream output by the (i+k)-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and input the feedback bit stream aligned with the target feedback bit stream into the decoding network.

[0243] In the jth training cycle of the model training, the processing unit 310 is further configured to complete a training on the i-th coding model, and then copy the weight parameters of the i-th coding model to the (i+k)-th coding model;

[0244] In the j+1th training cycle of the model training, the processing unit 310 is further configured to complete a training on the i+kth coding model based on the copied weight parameters, and copy the weight parameters of the i+kth coding model to the i-th coding model;

[0245] Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, i, j, k, S are all positive integers, and i+k≤N.

[0246] In some embodiments, the channel information is aligned with the target channel information by adding first placeholder information in the first dimension during pre-processing.

[0247] In some embodiments, the first placeholder information is 0.

[0248] In some embodiments, the feedback bit stream is aligned with the target feedback bit stream by adding second placeholder information during post-processing, or the feedback bit stream is aligned with the target feedback bit stream by truncating part of the bit stream during post-processing, or the feedback bit stream is aligned with the target feedback bit stream by deleting part of the bit stream during post-processing.

[0249] In some embodiments, the second placeholder information is 0 or 1.

[0250] In some embodiments, the communication unit 320 is further configured to receive first information;

[0251] The first information is used to indicate at least one of the following: physical resource configuration information associated with channel information feedback, and channel information feedback overhead configuration information.

[0252] In some embodiments, the first information includes a first information field; wherein the first information field is used to jointly indicate physical resource configuration information associated with the channel information feedback and the channel information feedback overhead configuration information.

[0253] In some embodiments, the first information includes a second information field and a third information field; wherein the second information field is used to indicate physical resource configuration information associated with the channel information feedback, and the third information field is used to indicate the channel information feedback overhead configuration information.

[0254] In some embodiments, the first information is carried by at least one of the following signaling: radio resource control RRC signaling, media access control element MAC CE, downlink control information DCI, and sidelink control information SCI.

[0255] In some embodiments, the communication unit 320 is further configured to receive second information and third information;

[0256] The second information is used to indicate physical resource configuration information associated with channel information feedback, and the third information is used to indicate channel information feedback overhead configuration information.

[0257] In some embodiments, the second information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI; and / or, the third information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI.

[0258] In some embodiments, the target channel information is the maximum channel information in the first dimension among the channel information corresponding to the M physical resource configurations associated with the channel information feedback, and / or the target feedback bit stream is the maximum or minimum feedback bit stream among the feedback bit streams corresponding to the N channel information feedback overhead configurations;

[0259] Different physical resource configurations in the M physical resource configurations are different in the first dimension, M and N are both positive integers, and M≥2, N≥2.

[0260] In some embodiments, the communication unit may be a communication interface or a transceiver, or an input / output interface of a communication chip or a system on chip. The processing unit may be one or more processors.

[0261] It should be understood that the transmitting device 300 according to the embodiment of the present application may correspond to the transmitting device in the embodiment of the method of the present application, and the above-mentioned and other operations and / or functions of each unit in the transmitting device 300 are respectively for realizing the corresponding processes of the transmitting device in the method 200 shown in Figure 7. For the sake of brevity, they will not be repeated here.

[0262] FIG19 shows a schematic block diagram of a receiving device 400 according to an embodiment of the present application. As shown in FIG19 , the receiving device 400 includes:

[0263] The communication unit 410 is configured to receive a first feedback bit stream sent by a transmitting device;

[0264] The processing unit 420 is configured to decode the first feedback bit stream through a decoding network to obtain first channel information;

[0265] The first feedback bit stream is a feedback bit stream output by the encoding network corresponding to the decoding network after post-processing to align it with the target feedback bit stream, and / or the first channel information is channel information output by the decoding network that is aligned with the target channel information in the first dimension and is post-processed to be different from the target channel information in the first dimension;

[0266] The first dimension is at least one of the following: the number of transmitting antenna ports, the number of subbands, the number of resource blocks (RBs), the number of delay paths, the number of symbols, and the number of time slots.

[0267] In some embodiments, under different physical resource configurations associated with channel information feedback and / or under different channel information feedback overhead configurations, the decoding network is the same, or the model weight parameters of the decoding network are the same;

[0268] The different physical resource configurations associated with the channel information feedback are different in the first dimension.

[0269] In some embodiments, when the first channel information is channel information output by the decoding network that is aligned with the target channel information in the first dimension and is obtained after post-processing, the processing unit 420 is further configured to perform post-processing on the channel information feedback-associated M physical resource configurations, the channel information output by the decoding network that is aligned with the target channel information in the first dimension and corresponds to the M physical resource configurations, to delete the first placeholder information in the channel information output by the decoding network that corresponds to the M physical resource configurations, and obtain the channel information that is different from the target channel information in the first dimension and corresponds to the M physical resource configurations;

[0270] The inputs of the coding networks corresponding to the M physical resource configurations are obtained by completing the channel information corresponding to the M physical resource configurations with the first placeholder information in the first dimension, where M is a positive integer and M≥2.

[0271] In some embodiments, when the first feedback bit stream is a feedback bit stream output by the encoding network corresponding to the decoding network and aligned with the target feedback bit stream after post-processing, for N channel information feedback overhead configurations, the processing unit 420 is further configured to respectively construct N encoding models and N decoding models, wherein the N encoding models respectively correspond to the N decoding models, the N encoding models have the same model architecture, and the N decoding models have the same model architecture, and N is a positive integer, and N ≥ 2;

[0272] The processing unit 420 is further configured to perform post-processing on the feedback bit stream output by the i-th coding model and the feedback bit stream output by the (i+k)-th coding model, so as to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and input the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model;

[0273] In the j-th training cycle of the model training, the processing unit 420 is further configured to complete a training on the i-th encoding model and its corresponding decoding model, and then copy the weight parameters of the i-th encoding model and its corresponding decoding model to the i+k-th encoding model and its corresponding decoding model respectively;

[0274] In the j+1th training cycle of the model training, the processing unit 420 is further configured to complete a training on the i+kth encoding model and its corresponding decoding model based on the copied weight parameters, and copy the weight parameters of the i+kth encoding model and its corresponding decoding model to the ith encoding model and its corresponding decoding model respectively;

[0275] Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network corresponding to the decoding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

[0276] In some embodiments, while keeping the decoding network unchanged, the processing unit 420 is further configured to retrain the encoding network corresponding to the decoding network based on the N channel information feedback overhead configurations.

[0277] In some embodiments, when the first feedback bit stream is a feedback bit stream output by the encoding network corresponding to the decoding network and aligned with the target feedback bit stream after post-processing, for N channel information feedback overhead configurations, the processing unit 420 is further configured to respectively construct N decoding models, wherein the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2;

[0278] The processing unit 420 is further configured to align the feedback bit stream corresponding to the i-th decoding model with the target feedback bit stream, align the feedback bit stream corresponding to the i+k-th decoding model with the target feedback bit stream, and input the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model by the transmitting device;

[0279] In the jth training cycle of the model training, the processing unit 420 is further configured to complete a training on the i-th decoding model, and then copy the weight parameters of the i-th decoding model to the i+k-th decoding model;

[0280] In the j+1th training cycle of the model training, the processing unit 420 is further configured to complete a training on the i+kth decoding model based on the copied weight parameters, and copy the weight parameters of the i+kth decoding model to the i-th decoding model;

[0281] Among them, the i-th decoding model or the i+k-th decoding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

[0282] In some embodiments, when the first feedback bit stream is a feedback bit stream output by the encoding network corresponding to the decoding network and aligned with the target feedback bit stream after post-processing, and the first channel information is channel information output by the decoding network that is aligned with the target channel information in a first dimension and is different from the target channel information in the first dimension after post-processing, for N channel information feedback overhead configurations, the processing unit 420 is further configured to respectively construct N encoding models and N decoding models, wherein the N encoding models respectively correspond to the N decoding models, the N encoding models have the same model architecture, the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2;

[0283] For M physical resource configurations associated with the channel information feedback, the processing unit 420 is further configured to preprocess the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtain inputs of the N coding models corresponding to the M physical resource configurations, respectively, where M is a positive integer and M≥2;

[0284] The processing unit 420 is further configured to perform post-processing on the feedback bit stream output by the i-th coding model and the feedback bit stream output by the (i+k)-th coding model, so as to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, align the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and input the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model;

[0285] In the j-th training cycle of the model training, the processing unit 420 is further configured to complete a training on the i-th encoding model and its corresponding decoding model, and then copy the weight parameters of the i-th encoding model and its corresponding decoding model to the i+k-th encoding model and its corresponding decoding model respectively;

[0286] In the j+kth training cycle of the model training, the processing unit 420 is further configured to complete a training on the i+kth encoding model and its corresponding decoding model based on the copied weight parameters, and copy the weight parameters of the i+kth encoding model and its corresponding decoding model to the ith encoding model and its corresponding decoding model respectively;

[0287] Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network corresponding to the decoding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

[0288] In some embodiments, while keeping the decoding network unchanged, the processing unit 420 is further configured to retrain the encoding network corresponding to the decoding network based on the N channel information feedback overhead configurations and the M physical resource configurations.

[0289] In some embodiments, when the first feedback bit stream is a feedback bit stream output by the encoding network corresponding to the decoding network after post-processing to align it with the target feedback bit stream, and the first channel information is channel information output by the decoding network that is aligned with the target channel information in a first dimension and is different from the target channel information in the first dimension after post-processing, for N channel information feedback overhead configurations, the processing unit 420 is further configured to respectively construct N decoding models, wherein the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2;

[0290] For M physical resource configurations associated with the channel information feedback, the processing unit 420 is further configured to preprocess the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtain inputs of the coding networks corresponding to the M physical resource configurations, respectively, where M is a positive integer and M≥2;

[0291] The processing unit 420 is further configured to align the feedback bit stream corresponding to the i-th decoding model with the target feedback bit stream, align the feedback bit stream corresponding to the i+k-th decoding model with the target feedback bit stream, and input the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model;

[0292] In the jth training cycle of the model training, the processing unit 420 is further configured to complete a training on the i-th decoding model, and then copy the weight parameters of the i-th decoding model to the i+k-th decoding model;

[0293] In the j+1th training cycle of the model training, the processing unit 420 is further configured to complete a training on the i+kth decoding model based on the copied weight parameters, and copy the weight parameters of the i+kth decoding model to the i-th decoding model;

[0294] Among them, the i-th decoding model or the i+k-th decoding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

[0295] In some embodiments, the channel information is aligned with the target channel information in preprocessing by adding first placeholder information in the first dimension; and the channel information is post-processed by deleting the first placeholder information in the first dimension to obtain channel information different from the target channel information in the first dimension.

[0296] In some embodiments, the first placeholder information is 0.

[0297] In some embodiments, the feedback bit stream is aligned with the target feedback bit stream by adding second placeholder information during post-processing, or the feedback bit stream is aligned with the target feedback bit stream by truncating part of the bit stream during post-processing, or the feedback bit stream is aligned with the target feedback bit stream by deleting part of the bit stream during post-processing.

[0298] In some embodiments, the second placeholder information is 0 or 1.

[0299] In some embodiments, the communication unit 410 is further configured to send the first information;

[0300] The first information is used to indicate at least one of the following: physical resource configuration information associated with channel information feedback, and channel information feedback overhead configuration information.

[0301] In some embodiments, the first information includes a first information field; wherein the first information field is used to jointly indicate physical resource configuration information associated with the channel information feedback and the channel information feedback overhead configuration information.

[0302] In some embodiments, the first information includes a second information field and a third information field; wherein the second information field is used to indicate physical resource configuration information associated with the channel information feedback, and the third information field is used to indicate the channel information feedback overhead configuration information.

[0303] In some embodiments, the first information is carried by at least one of the following signaling: radio resource control RRC signaling, media access control element MAC CE, downlink control information DCI, and sidelink control information SCI.

[0304] In some embodiments, the communication unit 410 further sends second information and third information;

[0305] The second information is used to indicate physical resource configuration information associated with channel information feedback, and the third information is used to indicate channel information feedback overhead configuration information.

[0306] In some embodiments, the second information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI; and / or, the third information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI.

[0307] In some embodiments, the target channel information is the maximum channel information in the first dimension among the channel information corresponding to the M physical resource configurations associated with the channel information feedback, and / or the target feedback bit stream is the maximum or minimum feedback bit stream among the feedback bit streams corresponding to the N channel information feedback overhead configurations;

[0308] Different physical resource configurations in the M physical resource configurations are different in the first dimension, M and N are both positive integers, and M≥2, N≥2.

[0309] In some embodiments, the communication unit may be a communication interface or a transceiver, or an input / output interface of a communication chip or a system on chip. The processing unit may be one or more processors.

[0310] It should be understood that the receiving device 400 according to the embodiment of the present application may correspond to the receiving device in the embodiment of the method of the present application, and the above-mentioned and other operations and / or functions of each unit in the receiving device 400 are respectively for realizing the corresponding processes of the receiving device in the method 200 shown in Figure 7. For the sake of brevity, they will not be repeated here.

[0311] Figure 20 is a schematic structural diagram of a communication device 500 provided in an embodiment of the present application. The communication device 500 shown in Figure 20 includes a processor 510, which can call and run a computer program from a memory to implement the method in the embodiment of the present application.

[0312] In some embodiments, as shown in FIG20 , the communication device 500 may further include a memory 520. The processor 510 may call and execute a computer program from the memory 520 to implement the method in the embodiment of the present application.

[0313] The memory 520 may be a separate device independent of the processor 510 , or may be integrated into the processor 510 .

[0314] In some embodiments, as shown in FIG. 20 , the communication device 500 may further include a transceiver 530 , and the processor 510 may control the transceiver 530 to communicate with other devices. Specifically, the transceiver 530 may send information or data to other devices, or receive information or data sent by other devices.

[0315] The transceiver 530 may include a transmitter and a receiver. The transceiver 530 may further include an antenna, and the number of antennas may be one or more.

[0316] In some embodiments, the processor 510 may implement the functions of a processing unit in a transmitting device, or the processor 510 may implement the functions of a processing unit in a receiving device, which will not be described in detail here for the sake of brevity.

[0317] In some embodiments, the transceiver 530 may implement the functions of a communication unit in a transmitting device, which will not be described in detail here for the sake of brevity.

[0318] In some embodiments, the transceiver 530 may implement the functions of a communication unit in a receiving device, which will not be described in detail here for the sake of brevity.

[0319] In some embodiments, the communication device 500 may specifically be a receiving device of an embodiment of the present application, and the communication device 500 may implement the corresponding processes implemented by the receiving device in each method of the embodiment of the present application. For the sake of brevity, they will not be repeated here.

[0320] In some embodiments, the communication device 500 may specifically be the transmitting device of the embodiment of the present application, and the communication device 500 may implement the corresponding processes implemented by the transmitting device in each method of the embodiment of the present application. For the sake of brevity, they will not be repeated here.

[0321] Figure 21 is a schematic structural diagram of an apparatus according to an embodiment of the present application. The apparatus 600 shown in Figure 21 includes a processor 610, which can call and execute a computer program from a memory to implement the method according to the embodiment of the present application.

[0322] In some embodiments, as shown in FIG21 , the apparatus 600 may further include a memory 620. The processor 610 may call and execute a computer program from the memory 620 to implement the method in the embodiment of the present application.

[0323] The memory 620 may be a separate device independent of the processor 610 , or may be integrated into the processor 610 .

[0324] In some embodiments, the apparatus 600 may further include an input interface 630. The processor 610 may control the input interface 630 to communicate with other devices or chips, and specifically, may obtain information or data sent by other devices or chips. Optionally, the processor 610 may be located inside or outside the chip.

[0325] In some embodiments, the processor 610 may implement the functions of a processing unit in a transmitting device, or the processor 610 may implement the functions of a processing unit in a receiving device, which will not be described in detail here for the sake of brevity.

[0326] In some embodiments, the input interface 630 may implement the function of a communication unit in a transmitting device, or the input interface 630 may implement the function of a communication unit in a receiving device.

[0327] In some embodiments, the apparatus 600 may further include an output interface 640. The processor 610 may control the output interface 640 to communicate with other devices or chips, and specifically, may output information or data to other devices or chips. Optionally, the processor 610 may be located inside or outside the chip.

[0328] In some embodiments, the output interface 640 may implement the function of a communication unit in a transmitting device, or the output interface 640 may implement the function of a communication unit in a receiving device.

[0329] In some embodiments, the apparatus can be applied to the receiving device in the embodiments of the present application, and the apparatus can implement the corresponding processes implemented by the receiving device in each method of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0330] In some embodiments, the apparatus can be applied to the originating device in the embodiments of the present application, and the apparatus can implement the corresponding processes implemented by the originating device in each method of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0331] In some embodiments, the device mentioned in the embodiments of the present application may also be a chip, such as a system-on-chip, a system-on-chip, a chip system, or a system-on-chip chip.

[0332] FIG22 is a schematic block diagram of a communication system 700 provided in an embodiment of the present application. As shown in FIG22 , the communication system 700 includes a transmitting device 710 and a receiving device 720 .

[0333] Among them, the transmitting device 710 can be used to implement the corresponding functions implemented by the transmitting device in the above method, and the receiving device 720 can be used to implement the corresponding functions implemented by the receiving device in the above method. For the sake of brevity, they will not be repeated here.

[0334] It should be understood that the processor of the embodiments of the present application may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method embodiment can be completed by hardware integrated logic circuits in the processor or software instructions. The above processor 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. The various methods, steps, and logic block diagrams disclosed in the embodiments of the present application can be implemented or executed. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of the present application can be directly embodied as being executed by a hardware decoding processor, or can be executed by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium mature in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware.

[0335] It is understood that the memory in the embodiments of the present application may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of RAM are available, such as 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 RAM bus random access memory (DR RAM). It should be noted that the memory of the systems and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0336] It should be understood that the above-mentioned memories are exemplary but not restrictive. For example, the memories in the embodiments of the present 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 RAM RAM (DR RAM), etc. In other words, the memories in the embodiments of the present application are intended to include, but are not limited to, these and any other suitable types of memories.

[0337] An embodiment of the present application also provides a computer-readable storage medium for storing a computer program.

[0338] In some embodiments, the computer-readable storage medium can be applied to the receiving device in the embodiments of the present application, and the computer program enables the computer to execute the corresponding processes implemented by the receiving device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0339] In some embodiments, the computer-readable storage medium can be applied to the originating device in the embodiments of the present application, and the computer program enables the computer to execute the corresponding processes implemented by the originating device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0340] An embodiment of the present application also provides a computer program product, including computer program instructions.

[0341] In some embodiments, the computer program product can be applied to the receiving device in the embodiments of the present application, and the computer program instructions enable the computer to execute the corresponding processes implemented by the receiving device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0342] In some embodiments, the computer program product can be applied to the originating device in the embodiments of the present application, and the computer program instructions enable the computer to execute the corresponding processes implemented by the originating device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0343] The embodiment of the present application also provides a computer program.

[0344] In some embodiments, the computer program can be applied to the receiving device in the embodiments of the present application. When the computer program runs on the computer, the computer executes the corresponding processes implemented by the receiving device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0345] In some embodiments, the computer program can be applied to the originating device in the embodiments of the present application. When the computer program runs on a computer, the computer executes the corresponding processes implemented by the originating device in the various methods of the embodiments of the present application. For the sake of brevity, they will not be repeated here.

[0346] Those skilled in the art will appreciate that the units and algorithm steps of each example 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 performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel 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.

[0347] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0348] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0349] The units described as separate components may or may not be physically separate, and 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 these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0350] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0351] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. In view of this understanding, the technical solution of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0352] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A method for channel information feedback, characterized in that: include: The transmitting device encodes the first channel information through a coding network to obtain a first feedback bit stream; as well as The transmitting device sends the first feedback bit stream to the receiving device; The first channel information is channel information that has been pre-processed to be aligned with target channel information in a first dimension, and / or the first feedback bit stream is a feedback bit stream that has been post-processed and aligned with a target feedback bit stream output by the coding network; The first dimension is at least one of the following: the number of transmitting antenna ports, the number of subbands, the number of resource blocks (RBs), the number of delay paths, the number of symbols, and the number of time slots.

2. The method according to claim 1, wherein Under different physical resource configurations associated with channel information feedback and / or under different channel information feedback overhead configurations, the coding network is the same, or the model weight parameters of the coding network are the same; The different physical resource configurations associated with the channel information feedback are different in the first dimension.

3. The method according to claim 2, wherein In a case where the first channel information is channel information pre-processed and aligned with target channel information in a first dimension, the method further includes: For M physical resource configurations associated with channel information feedback, the transmitting device preprocesses the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtains the input of the coding network corresponding to the M physical resource configurations, where M is a positive integer and M≥2.

4. The method according to claim 2, wherein In a case where the first feedback bit stream is a feedback bit stream output by the coding network and is aligned with a target feedback bit stream after post-processing, the method further includes: For N channel information feedback overhead configurations, the transmitting device constructs N coding models and N decoding models, respectively, wherein the N coding models correspond to the N decoding models, respectively, the N coding models have the same model architecture, the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; The transmitting device post-processes a feedback bit stream output by the i-th coding model and a feedback bit stream output by the (i+k)-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, and aligns the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into a corresponding decoding model; In the jth training cycle of model training, the transmitting device completes a training on the i-th encoding model and its corresponding decoding model, and then copies the weight parameters of the i-th encoding model and its corresponding decoding model to the i+k-th encoding model and its corresponding decoding model respectively; In the j+1th training cycle of the model training, the transmitting device completes a training on the i+kth encoding model and its corresponding decoding model based on the copied weight parameters, and copies the weight parameters of the i+kth encoding model and its corresponding decoding model to the ith encoding model and its corresponding decoding model respectively; Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network corresponding to the coding network, i, j, k, S are all positive integers, and i+k≤N.

5. The method according to claim 4, wherein The method further comprises: While keeping the encoding network unchanged, the transmitting device retrains the decoding network corresponding to the encoding network based on the N channel information feedback overhead configurations.

6. The method according to claim 2, wherein In a case where the first feedback bit stream is a feedback bit stream output by the coding network and is aligned with a target feedback bit stream after post-processing, the method further includes: For N channel information feedback overhead configurations, the transmitting device constructs N coding models respectively, wherein the N coding models have the same model architecture, N is a positive integer, and N ≥ 2; The transmitting device post-processes a feedback bit stream output by the i-th coding model and a feedback bit stream output by the (i+k)-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, and aligns the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into a decoding network; In the jth training cycle of the model training, the transmitting device completes a training on the i-th coding model, and then copies the weight parameters of the i-th coding model to the i+k-th coding model; In the j+1th training cycle of the model training, the transmitting device completes a training on the i+kth coding model based on the copied weight parameters, and copies the weight parameters of the i+kth coding model to the i-th coding model; The i-th coding model or the i+k-th coding model after S training cycles is the coding network, i, j, k, S are all positive integers, and i+k≤N.

7. The method according to claim 2, wherein In a case where the first channel information is channel information that is pre-processed to be aligned with target channel information in a first dimension, and the first feedback bit stream is a feedback bit stream that is post-processed and aligned with a target feedback bit stream output by the coding network, the method further includes: For N channel information feedback overhead configurations, the transmitting device constructs N coding models and N decoding models, respectively, wherein the N coding models correspond to the N decoding models, respectively, the N coding models have the same model architecture, the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; For M physical resource configurations associated with channel information feedback, the transmitting device preprocesses the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtains inputs of the N coding models corresponding to the M physical resource configurations, respectively, where M is a positive integer and M≥2; The transmitting device post-processes a feedback bit stream output by the i-th coding model and a feedback bit stream output by the (i+k)-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, and aligns the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into a corresponding decoding model; In the jth training cycle of model training, the transmitting device completes a training on the i-th encoding model and its corresponding decoding model, and then copies the weight parameters of the i-th encoding model and its corresponding decoding model to the i+k-th encoding model and its corresponding decoding model respectively; In the j+1th training cycle of the model training, the transmitting device completes a training on the i+kth encoding model and its corresponding decoding model based on the copied weight parameters, and copies the weight parameters of the i+kth encoding model and its corresponding decoding model to the ith encoding model and its corresponding decoding model respectively; Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network corresponding to the coding network, i, j, k, S are all positive integers, and i+k≤N.

8. The method according to claim 7, wherein The method further comprises: While keeping the coding network unchanged, the transmitting device retrains the decoding network corresponding to the coding network based on the N channel information feedback overhead configurations and the M physical resource configurations.

9. The method according to claim 2, wherein In a case where the first channel information is channel information that is pre-processed to be aligned with target channel information in a first dimension, and the first feedback bit stream is a feedback bit stream that is post-processed and aligned with a target feedback bit stream output by the coding network, the method further includes: For N channel information feedback overhead configurations, the transmitting device constructs N coding models respectively, wherein the N coding models have the same model architecture, N is a positive integer, and N ≥ 2; For M physical resource configurations associated with channel information feedback, the transmitting device preprocesses the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtains inputs of the N coding models corresponding to the M physical resource configurations, respectively, where M is a positive integer and M≥2; The transmitting device post-processes a feedback bit stream output by the i-th coding model and a feedback bit stream output by the (i+k)-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, and aligns the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into a decoding network; In the jth training cycle of the model training, the transmitting device completes a training on the i-th coding model, and then copies the weight parameters of the i-th coding model to the i+k-th coding model; In the j+1th training cycle of the model training, the transmitting device completes a training on the i+kth coding model based on the copied weight parameters, and copies the weight parameters of the i+kth coding model to the i-th coding model; Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network, i, j, k, S are all positive integers, and i+k≤N.

10. The method according to any one of claims 1 to 9, characterized in that During preprocessing, the channel information is aligned with the target channel information by adding first placeholder information in the first dimension.

11. The method according to claim 10, wherein The first placeholder information is 0.

12. The method according to any one of claims 1 to 11, characterized in that The feedback bit stream is aligned with the target feedback bit stream by adding the second placeholder information in post-processing, or the feedback bit stream is aligned with the target feedback bit stream by cutting off part of the bit stream in post-processing, or the feedback bit stream is aligned with the target feedback bit stream by deleting part of the bit stream in post-processing.

13. The method according to claim 12, wherein: The second placeholder information is 0 or 1.

14. The method according to any one of claims 1 to 13, characterized in that The method further comprises: The originating device receives first information; The first information is used to indicate at least one of the following: physical resource configuration information associated with channel information feedback, and channel information feedback overhead configuration information.

15. The method according to claim 14, wherein The first information includes a first information field; wherein, the first information field is used to jointly indicate the physical resource configuration information associated with the channel information feedback and the channel information feedback overhead configuration information.

16. The method according to claim 14, wherein The first information includes a second information field and a third information field; wherein the second information field is used to indicate physical resource configuration information associated with the channel information feedback, and the third information field is used to indicate the channel information feedback overhead configuration information.

17. The method according to any one of claims 14 to 16, characterized in that The first information is carried by at least one of the following signaling: radio resource control RRC signaling, media access control element MAC CE, downlink control information DCI, and sidelink control information SCI.

18. The method according to any one of claims 1 to 13, characterized in that The method further comprises: The originating device receives the second information and the third information; The second information is used to indicate physical resource configuration information associated with channel information feedback, and the third information is used to indicate channel information feedback overhead configuration information.

19. The method according to claim 18, wherein The second information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI; and / or, The third information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI.

20. The method according to any one of claims 1 to 19, characterized in that The target channel information is the maximum channel information in the first dimension among the channel information corresponding to the M physical resource configurations associated with the channel information feedback, and / or the target feedback bit stream is the maximum or minimum feedback bit stream among the feedback bit streams corresponding to the N channel information feedback overhead configurations; Among them, different physical resource configurations in the M physical resource configurations are different in the first dimension, M and N are both positive integers, and M≥2, N≥2.

21. A method for channel information feedback, characterized in that: include: The receiving device receives the first feedback bit stream sent by the transmitting device; The receiving device decodes the first feedback bit stream through a decoding network to obtain first channel information; The first feedback bit stream is a feedback bit stream output by the encoding network corresponding to the decoding network, which is aligned with the target feedback bit stream after post-processing, and / or the first channel information is channel information output by the decoding network, which is aligned with the target channel information in the first dimension, which is different from the target channel information in the first dimension after post-processing; The first dimension is at least one of the following: the number of transmitting antenna ports, the number of subbands, the number of resource blocks (RBs), the number of delay paths, the number of symbols, and the number of time slots.

22. The method according to claim 21, wherein Under different physical resource configurations associated with channel information feedback and / or under different channel information feedback overhead configurations, the decoding networks are the same, or the model weight parameters of the decoding networks are the same; The different physical resource configurations associated with the channel information feedback are different in the first dimension.

23. The method according to claim 22, wherein In a case where the first channel information is channel information output by the decoding network and aligned with target channel information in a first dimension and is channel information different from the target channel information in the first dimension after post-processing, the method further includes: For M physical resource configurations associated with channel information feedback, the receiving device post-processes the channel information output by the decoding network corresponding to the M physical resource configurations and aligned with the target channel information in the first dimension, so as to delete the first placeholder information in the channel information output by the decoding network corresponding to the M physical resource configurations, thereby obtaining channel information corresponding to the M physical resource configurations and different from the target channel information in the first dimension; The inputs of the coding networks corresponding to the M physical resource configurations are obtained by filling the channel information corresponding to the M physical resource configurations with the first placeholder information in the first dimension, where M is a positive integer and M≥2.

24. The method of claim 22, wherein: In a case where the first feedback bit stream is a feedback bit stream output by an encoding network corresponding to the decoding network and aligned with a target feedback bit stream after post-processing, the method further includes: For N channel information feedback overhead configurations, the receiving device constructs N coding models and N decoding models, respectively, wherein the N coding models correspond to the N decoding models, respectively, the N coding models have the same model architecture, the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; The receiving device post-processes the feedback bit stream output by the i-th coding model and the feedback bit stream output by the (i+k)-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, and aligns the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model; In the jth training cycle of model training, the receiving device completes a training on the i-th encoding model and its corresponding decoding model, and then copies the weight parameters of the i-th encoding model and its corresponding decoding model to the i+k-th encoding model and its corresponding decoding model respectively; In the j+1th training cycle of the model training, the receiving device completes a training on the i+kth encoding model and its corresponding decoding model based on the copied weight parameters, and copies the weight parameters of the i+kth encoding model and its corresponding decoding model to the ith encoding model and its corresponding decoding model respectively; Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network corresponding to the decoding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

25. The method of claim 24, wherein: The method further comprises: While keeping the decoding network unchanged, the receiving device retrains the encoding network corresponding to the decoding network based on the N channel information feedback overhead configurations.

26. The method of claim 22, wherein: In a case where the first feedback bit stream is a feedback bit stream output by an encoding network corresponding to the decoding network and aligned with a target feedback bit stream after post-processing, the method further includes: For N channel information feedback overhead configurations, the receiving device constructs N decoding models respectively, wherein the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; The receiving device aligns the feedback bit stream corresponding to the i-th decoding model with the target feedback bit stream, aligns the feedback bit stream corresponding to the i+k-th decoding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model; In the jth training cycle of the model training, the receiving device completes a training on the i-th decoding model, and then copies the weight parameters of the i-th decoding model to the i+k-th decoding model; In the j+1th training cycle of the model training, the receiving device completes a training on the i+kth decoding model based on the copied weight parameters, and copies the weight parameters of the i+kth decoding model to the i-th decoding model; Among them, the i-th decoding model or the i+k-th decoding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

27. The method of claim 22, wherein: In a case where the first feedback bit stream is a feedback bit stream output by an encoding network corresponding to the decoding network and is aligned with a target feedback bit stream after post-processing, and the first channel information is channel information output by the decoding network and aligned with the target channel information in a first dimension and is different from the target channel information in the first dimension after post-processing, the method further includes: For N channel information feedback overhead configurations, the receiving device constructs N coding models and N decoding models, respectively, wherein the N coding models correspond to the N decoding models, respectively, the N coding models have the same model architecture, the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; For M physical resource configurations associated with channel information feedback, the receiving device preprocesses the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtains inputs of the N coding models corresponding to the M physical resource configurations, respectively, where M is a positive integer and M≥2; The receiving device post-processes the feedback bit stream output by the i-th coding model and the feedback bit stream output by the (i+k)-th coding model to align the feedback bit stream output by the i-th coding model with the target feedback bit stream, and aligns the feedback bit stream output by the (i+k)-th coding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model; In the jth training cycle of model training, the receiving device completes a training on the i-th encoding model and its corresponding decoding model, and then copies the weight parameters of the i-th encoding model and its corresponding decoding model to the i+k-th encoding model and its corresponding decoding model respectively; In the j+kth training cycle of model training, the receiving device completes a training on the i+kth encoding model and its corresponding decoding model based on the copied weight parameters, and copies the weight parameters of the i+kth encoding model and its corresponding decoding model to the ith encoding model and its corresponding decoding model respectively; Among them, the i-th coding model or the i+k-th coding model after S training cycles is the coding network corresponding to the decoding network, and the decoding model corresponding to the i-th coding model or the decoding model corresponding to the i+k-th coding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

28. The method of claim 27, wherein: The method further comprises: While keeping the decoding network unchanged, the receiving device retrains the encoding network corresponding to the decoding network based on the N channel information feedback overhead configurations and the M physical resource configurations.

29. The method of claim 22, wherein: In a case where the first feedback bit stream is a feedback bit stream output by an encoding network corresponding to the decoding network and is aligned with a target feedback bit stream after post-processing, and the first channel information is channel information output by the decoding network and aligned with the target channel information in a first dimension and is different from the target channel information in the first dimension after post-processing, the method further includes: For N channel information feedback overhead configurations, the receiving device constructs N decoding models respectively, wherein the N decoding models have the same model architecture, N is a positive integer, and N ≥ 2; For M physical resource configurations associated with channel information feedback, the receiving device preprocesses the channel information corresponding to the M physical resource configurations to align the channel information corresponding to the M physical resource configurations with the target channel information in the first dimension, and obtains inputs of the coding networks corresponding to the M physical resource configurations, respectively, where M is a positive integer and M≥2; The receiving device aligns the feedback bit stream corresponding to the i-th decoding model with the target feedback bit stream, aligns the feedback bit stream corresponding to the i+k-th decoding model with the target feedback bit stream, and the transmitting device inputs the feedback bit stream aligned with the target feedback bit stream into the corresponding decoding model; In the jth training cycle of the model training, the receiving device completes a training on the i-th decoding model, and then copies the weight parameters of the i-th decoding model to the i+k-th decoding model; In the j+1th training cycle of the model training, the receiving device completes a training on the i+kth decoding model based on the copied weight parameters, and copies the weight parameters of the i+kth decoding model to the i-th decoding model; Among them, the i-th decoding model or the i+k-th decoding model after S training cycles is the decoding network, i, j, k, S are all positive integers, and i+k≤N.

30. The method according to any one of claims 21 to 29, wherein The channel information is aligned with the target channel information by adding first placeholder information in the first dimension during preprocessing; In post-processing, the channel information is post-processed by deleting the first placeholder information in the first dimension to obtain channel information that is different from the target channel information in the first dimension.

31. The method according to claim 23 or 30, wherein: The first placeholder information is 0.

32. The method according to any one of claims 21 to 31, wherein The feedback bit stream is aligned with the target feedback bit stream by adding the second placeholder information in post-processing, or the feedback bit stream is aligned with the target feedback bit stream by cutting off part of the bit stream in post-processing, or the feedback bit stream is aligned with the target feedback bit stream by deleting part of the bit stream in post-processing.

33. The method of claim 32, wherein: The second placeholder information is 0 or 1.

34. The method according to any one of claims 21 to 33, wherein The method further comprises: The receiving device sends the first information; The first information is used to indicate at least one of the following: physical resource configuration information associated with channel information feedback, and channel information feedback overhead configuration information.

35. The method of claim 34, wherein: The first information includes a first information field; wherein, the first information field is used to jointly indicate the physical resource configuration information associated with the channel information feedback and the channel information feedback overhead configuration information.

36. The method of claim 34, wherein: The first information includes a second information field and a third information field; wherein the second information field is used to indicate physical resource configuration information associated with the channel information feedback, and the third information field is used to indicate the channel information feedback overhead configuration information.

37. The method according to any one of claims 34 to 36, wherein The first information is carried by at least one of the following signaling: radio resource control RRC signaling, media access control element MAC CE, downlink control information DCI, and sidelink control information SCI.

38. The method according to any one of claims 21 to 33, wherein The method further comprises: The receiving device sends the second information and the third information; The second information is used to indicate physical resource configuration information associated with channel information feedback, and the third information is used to indicate channel information feedback overhead configuration information.

39. The method of claim 38, wherein The second information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI; and / or, The third information is carried by at least one of the following signaling: RRC signaling, MAC CE, DCI, SCI.

40. The method according to any one of claims 21 to 39, wherein The target channel information is the maximum channel information in the first dimension among the channel information corresponding to the M physical resource configurations associated with the channel information feedback, and / or the target feedback bit stream is the maximum or minimum feedback bit stream among the feedback bit streams corresponding to the N channel information feedback overhead configurations; Among them, different physical resource configurations in the M physical resource configurations are different in the first dimension, M and N are both positive integers, and M≥2, N≥2.

41. A terminating device, characterized in that: include: a processing unit, configured to encode the first channel information through a coding network to obtain a first feedback bit stream; a communication unit, configured to send the first feedback bit stream to a receiving device; The first channel information is channel information that has been pre-processed to be aligned with target channel information in a first dimension, and / or the first feedback bit stream is a feedback bit stream that has been post-processed and aligned with a target feedback bit stream output by the coding network; The first dimension is at least one of the following: the number of transmitting antenna ports, the number of subbands, the number of resource blocks (RBs), the number of delay paths, the number of symbols, and the number of time slots.

42. A receiving device, characterized in that: include: A communication unit, configured to receive a first feedback bit stream sent by a transmitting device; a processing unit, configured to decode the first feedback bit stream through a decoding network to obtain first channel information; The first feedback bit stream is a feedback bit stream output by the encoding network corresponding to the decoding network, which is aligned with the target feedback bit stream after post-processing, and / or the first channel information is channel information output by the decoding network, which is aligned with the target channel information in the first dimension, which is different from the target channel information in the first dimension after post-processing; The first dimension is at least one of the following: the number of transmitting antenna ports, the number of subbands, the number of resource blocks (RBs), the number of delay paths, the number of symbols, and the number of time slots.

43. A terminating device, characterized in that: include: A processor and a memory, the memory being used to store a computer program, the processor being used to call and run the computer program stored in the memory, so that the originating device executes the method according to any one of claims 1 to 20.

44. A receiving device, characterized in that: include: A processor and a memory, the memory being used to store a computer program, the processor being used to call and run the computer program stored in the memory, so that the receiving device executes the method as described in any one of claims 21 to 40.

45. A chip, characterized in that include: A processor, configured to call and run a computer program from a memory, so that a device equipped with the chip executes the method according to any one of claims 1 to 20.

46. ​​A chip, characterized in that include: A processor, configured to call and execute a computer program from a memory, so that a device equipped with the chip executes the method according to any one of claims 21 to 40.

47. A computer-readable storage medium, characterized in that For storing a computer program, when the computer program is executed, the method according to any one of claims 1 to 20 is implemented.

48. A computer-readable storage medium, characterized in that For storing a computer program, when said computer program is executed, the method according to any one of claims 21 to 40 is implemented.

49. A computer program product, characterized in that The method comprises computer program instructions, which, when executed, implement the method according to any one of claims 1 to 20.

50. A computer program product, characterized in that The method comprises computer program instructions which, when executed, implement the method according to any one of claims 21 to 40.

51. A computer program, characterized in that When the computer program is executed, the method according to any one of claims 1 to 20 is implemented.

52. A computer program, characterized in that When the computer program is executed, the method according to any one of claims 21 to 40 is implemented.

Images