Communication method and communication apparatus

Abstract

The application provides a communication method and device, in which a terminal side determines a type of data included in a data set of the terminal side, the type belonging to a first type or a second type, the data of the first type including labels corresponding to inputs of an encoder in a first dual-end model and outputs of a decoder in the first dual-end model, the data of the second type including the inputs of the encoder, the inputs of the decoder and labels corresponding to the outputs of the decoder, the first dual-end model being a dual-end model of the terminal side; the encoder is trained or updated based on the data set of the terminal side of the type; wherein the encoder is used for CSI compression of CSI, and the encoder is used for CSI reconstruction of the compressed CSI. By determining that the type of the data set for the dual-end model belongs to the first type or the second type, and training the UE side model which can be docked with the network side model based on the data set of the type, the feedback performance of the CSI is improved.

Description

Technical Field

[0001] This application relates to the field of wireless communication, and more particularly to a communication method and a communication device. Background Technology

[0002] In channel state information (CSI) feedback based on artificial intelligence (AI), when the model is deployed at the base station, the base station obtains the estimated results of the channel state information-reference signal (CSI-RS) at the user equipment (UE) side as labels for model training. CSI feedback can be implemented based on an auto-encoder (AE) model. An AE model can generally refer to a network structure consisting of two sub-models, such as an encoder and a decoder. The encoder and decoder of the AE model are usually trained together and can be used in a matched manner. In one implementation, the base station locally trains the encoder and decoder based on historically collected CSI data. The network side can send the trained encoder to the UE side, enabling pairing of the two-end models at the base station and UE sides. Since the dataset used by the network side to train the two-end models is not fixed and may have various types with different data, how the terminal side constructs the dataset for model training or updating needs to be considered to improve the model integration effect. Summary of the Invention

[0003] This application provides a communication method and a communication device, which facilitates model training on the terminal side based on a dataset constructed on the network side, thereby improving the feedback performance of CSI.

[0004] Firstly, a communication method is provided, which can be applied to the terminal side, for example, it can be executed by a terminal device; or it can be executed by a component deployed in the terminal device, such as circuits or chips inside the terminal device (such as modem chips, also known as baseband chips, or system-on-chip (SoC) chips or system-in-package (SIP) chips containing modem cores, etc.); or it can be executed by a device deployed outside the terminal device (such as the host of an over-the-top (OTT) system or a cloud server) or a component in the device (such as chips, processors, or circuits inside the device, etc.); or it can be implemented by a logic module or software capable of realizing all or part of the functions of the terminal device, etc. Alternatively, the method can be executed by a first device, which can be a terminal device or a component within the terminal device, such as internal circuitry or a chip (e.g., a modem chip), or a SoC chip or SIP chip containing a modem core; the first device can also be a device outside the terminal device (e.g., the host of an OTT system or a cloud server) or a component within a device (e.g., internal chips, processors, or circuitry); the first device can also be a logic module or software capable of implementing some or all of the functions on the terminal side, etc., and this application does not limit this. For ease of understanding and explanation, the method provided in the first aspect will be described below using the terminal side as an example.

[0005] The method includes: determining the type of data included in the terminal-side dataset, wherein the type belongs to a first type or a second type, the first type of data includes the input of the encoder in the first dual-ended model and the label corresponding to the output of the decoder in the first dual-ended model, the second type of data includes the input of the encoder, the input of the decoder and the label corresponding to the output of the decoder, and the first dual-ended model is a dual-ended model on the terminal side; the terminal-side dataset of this type is used to train or update the encoder; wherein the encoder is used to perform a first processing on channel state information (CSI), the first processing including at least CSI compression, and the encoder is used to perform a second processing on the compressed CSI, the second processing including at least CSI reconstruction.

[0006] Based on the above scheme, the terminal side can determine whether the dataset used for the dual-end model belongs to the first type or the second type, and train a UE-side model that can interface with the network-side model based on the dataset of this type. This is beneficial to improve the feedback performance of CSI. In other words, this scheme can avoid the terminal side from selecting other types of datasets for model training, which would cause the terminal-side model and the network-side model to fail to meet the performance requirements.

[0007] In conjunction with the first aspect, in some implementations of the first aspect, when the data type belongs to the first type, the input of the encoder is a first CSI, and the tag is the first CSI or a second CSI; wherein the first CSI represents the channel state of a first resource set, the second CSI represents the channel state of a second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set.

[0008] In conjunction with the first aspect, in certain implementations of the first aspect, when the data type belongs to the second type, the input of the encoder is a first CSI, the input of the decoder includes a third CSI, and the tag is a second CSI; wherein the first CSI characterizes the channel state of a first resource set, the second CSI characterizes the channel state of a second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set, the third CSI characterizes the channel state of a third resource set, the third resource set is different from the first resource set, the third resource set is a part or all of the second resource set, and the second resource set includes at least one of frequency domain resources or spatial domain resources.

[0009] In conjunction with the first aspect, in some implementations of the first aspect, the second resource set corresponds to the first time domain resource, the first resource set corresponds to the first time domain resource, the third resource set corresponds to the second time domain resource and / or the first time domain resource, and the second time domain resource is earlier than the first time domain resource.

[0010] Based on the above scheme, the terminal side can determine that the dataset used for the dual-end model includes multiple different CSIs, such as CSIs with different time-domain resources and / or frequency-domain resources. Alternatively, these multiple CSIs can also include historical CSIs, such as a third CSI corresponding to the second time-frequency resource. This can increase the diversity of samples, allowing the model to obtain accurate CSIs based on different CSIs. Training the model based on this first dataset can improve the model's generalization ability, thereby improving the CSI feedback performance.

[0011] In conjunction with the first aspect, in some implementations of the first aspect, the third resource set includes a first part and a second part, the first part corresponding to the second time-domain resource and the second part corresponding to the first time-domain resource.

[0012] In conjunction with the first aspect, in some implementations of the first aspect, an instruction message is received, which is used to determine whether the type is the first type or the second type.

[0013] Based on the above scheme, the terminal side can determine the type of data included in the dataset based on the instructions from the network side, and thus train a UE-side model that can interface with the network-side model based on the dataset of that type.

[0014] In conjunction with the first aspect, in some implementations of the first aspect, the indication information indicates N types, which are N types of data included in the dataset on the terminal side, or N types of data included in the dataset on the network side, which include the first type or the second type, where N is a positive integer.

[0015] For example, the data type included in the determined dataset on the terminal side is either the first type or the second type among the N types.

[0016] In conjunction with the first aspect, in some implementations of the first aspect, the indication information is used to determine M types, which are M types of data included in the dataset on the terminal side, or M types of data included in the dataset on the network side, one of which belongs to the first type or the second type, and M is a positive integer.

[0017] In conjunction with the first aspect, in certain implementations of the first aspect, the M types include at least one of the following: the encoder input is a first CSI, and the tag is the first CSI; the encoder input is a first CSI, and the tag is a second CSI; the encoder input is a first CSI, the decoder input includes a third CSI, and the tag is the second CSI; wherein the first CSI characterizes the channel state of a first resource set, the second CSI characterizes the channel state of a second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set, the third CSI characterizes the channel state of a third resource set, the third resource set is different from the first resource set, the third resource set is a part of the second resource set, and the second resource set includes at least one of frequency domain resources or spatial domain resources.

[0018] For example, the data type included in the determined terminal-side dataset is one of the M types.

[0019] In conjunction with the first aspect, in certain implementations of the first aspect, capability information is sent, which indicates the computing capabilities and / or types of data supported on the terminal side.

[0020] Based on the above scheme, by sending capability information to the network side, the network side can indicate to the terminal side the types of data included in the dataset that are compatible with the terminal side's capability information. This can avoid the situation where the types of data included in the dataset used for model training are incompatible with the terminal side's capabilities, resulting in the terminal side's model training failing to meet performance requirements.

[0021] In conjunction with the first aspect, in some implementations of the first aspect, before training or updating the encoder based on the terminal-side dataset of this type, a network-side encoder is obtained. The network-side encoder is obtained by training a second dual-end model based on the network-side dataset. The second dual-end model is the network-side dual-end model, and the data included in the network-side dataset belongs to the first type or the second type.

[0022] Secondly, a communication method is provided, which can be applied to the network side. For example, it can be executed by a network device; or it can be executed by a component deployed in the network device, such as circuits or chips inside the network device (e.g., modem chips, also known as baseband chips, or SoC chips or SIP chips containing modem cores, etc.); or it can be executed by a device outside the network device (e.g., intelligent network elements on the network side) or a component within the device (e.g., chips, processors, or circuits inside the device); or it can be implemented by a logic module or software capable of implementing all or part of the functions of the network device, etc. Alternatively, the method can be executed by a second device, which can be a network device or a component within the network device, such as circuits or chips inside the network device (e.g., modem chips), or SoC chips or SIP chips containing modem cores, etc.; the second device can also be a device outside the network device (e.g., intelligent network elements) or a component within the device (e.g., chips, processors, or circuits inside the device); the second device can also be a logic module or software capable of implementing some or all of the functions of the network side, etc., and this application does not limit this. For ease of understanding and explanation, the following text uses the network side as an example to illustrate the method provided in the second aspect.

[0023] The method includes: sending indication information, which is used to determine whether the data included in the terminal-side dataset is of a first type or a second type, the terminal-side dataset being used to train or update the encoder in a first dual-end model, the first type of data including the input of the encoder and the label corresponding to the output of the decoder in the first dual-end model, the second type of data including the input of the encoder, the input of the decoder and the label corresponding to the output of the decoder, the first dual-end model being a dual-end model on the terminal side; wherein, the encoder is used to perform a first processing on channel state information (CSI), the first processing including at least CSI compression, and the encoder is used to perform a second processing on the compressed CSI, the second processing including at least CSI reconstruction.

[0024] Based on the above scheme, the network side can send indication information to the terminal side, so that the terminal side can determine the type of dataset used for the dual-end model based on the indication information. This type belongs to the first type or the second type, and a UE-side model that can be interfacing with the network-side model can be trained based on the dataset of this type. This is beneficial to improving the feedback performance of CSI. In other words, this scheme can avoid the terminal side selecting other types of datasets for model training, which would cause the terminal-side model and the network-side model to fail to meet the performance requirements.

[0025] In conjunction with the second aspect, in some implementations of the second aspect, when the data type belongs to the first type, the input of the encoder is the first CSI, and the tag is the first CSI or the second CSI; wherein the first CSI represents the channel state of the first resource set, the second CSI represents the channel state of the second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set.

[0026] In conjunction with the second aspect, in some implementations of the second aspect, when the data type belongs to the second type, the input of the encoder is a first CSI, the input of the decoder includes a third CSI, and the tag is a second CSI; wherein the first CSI characterizes the channel state of a first resource set, the second CSI characterizes the channel state of a second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set, the third CSI characterizes the channel state of a third resource set, the third resource set is different from the first resource set, the third resource set is a part or all of the second resource set, and the second resource set includes at least one of frequency domain resources or spatial domain resources.

[0027] In conjunction with the second aspect, in some implementations of the second aspect, the second resource set corresponds to the first time domain resource, the first resource set corresponds to the first time domain resource, the third resource set corresponds to the second time domain resource and / or the first time domain resource, and the second time domain resource is earlier than the first time domain resource.

[0028] In conjunction with the second aspect, in some implementations of the second aspect, the indication information indicates N types, which are N types of data included in the dataset on the terminal side, or N types of data included in the dataset on the network side, and the N types include the first type or the second type, where N is a positive integer.

[0029] In conjunction with the second aspect, in some implementations of the second aspect, the indication information is used to determine M types, which are M types of data included in the dataset on the terminal side, or M types of data included in the dataset on the network side, one of which belongs to the first type or the second type, where M is a positive integer.

[0030] In conjunction with the second aspect, in certain implementations of the second aspect, the M types include at least one of the following: the encoder input is a first CSI, and the tag is the first CSI; the encoder input is a first CSI, and the tag is a second CSI; the encoder input is a first CSI, the decoder input includes a third CSI, and the tag is the second CSI; wherein the first CSI characterizes the channel state of a first resource set, the second CSI characterizes the channel state of a second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set, the third CSI characterizes the channel state of a third resource set, the third resource set is different from the first resource set, the third resource set is a part of the second resource set, and the second resource set includes at least one of frequency domain resources or spatial domain resources.

[0031] In conjunction with the second aspect, in some implementations of the second aspect, capability information is received, which indicates the computing capabilities and / or types of data supported on the terminal side; and indication information is sent according to the capability information, which indicates one of the N types or the M types.

[0032] In conjunction with the second aspect, in some implementations of the second aspect, the encoder on the network side is sent. The encoder on the network side is obtained by training the second dual-end model on the network side dataset. The second dual-end model is the dual-end model on the network side. The data included in the network side dataset belongs to the first type or the second type.

[0033] Thirdly, an apparatus is provided. This apparatus may include functional modules corresponding to each of the methods / operations / steps / actions described in any possible implementation of the first aspect, or may include functional modules corresponding to each of the methods / operations / steps / actions described in any of the second aspects. The module may be a hardware circuit, software, or a combination of hardware circuitry and software implementation.

[0034] In one design, the device may include a processing module and a communication module. The communication module is used to perform the sending and receiving actions performed by the terminal side in the method described in the first aspect above, while the processing module is used to perform processing-related actions performed by the terminal side in the method described in the first aspect above.

[0035] In one design, the device can be a terminal device, or a device, module, circuit, or chip configured in the terminal device, or a device that can be used in conjunction with the terminal device, such as an OTT host or cloud server.

[0036] In one design, the device may include a processing module and a communication module. The communication module is used to perform the sending and receiving actions performed by the network side in the method described in the second aspect above, while the processing module is used to perform processing-related actions performed by the network side in the method described in the second aspect above.

[0037] In one design, the device can be a network device, or a device, module, circuit, or chip configured in the network device, or a device that can be used in conjunction with the network device, such as an intelligent network element with a radio access network (RAN) intelligent controller (RIC) deployed thereon.

[0038] Fourthly, an apparatus is provided, comprising a processor and a storage medium storing instructions that, when executed by the processor, cause a method as described in the first aspect or any possible implementation thereof to be implemented, or cause a method as described in the second aspect or any possible implementation thereof to be implemented.

[0039] Fifthly, an apparatus is provided, comprising a processing circuit for processing data and / or information such that a method as in the first aspect or any possible implementation thereof is implemented, or a method as in the second aspect or any possible implementation thereof is implemented.

[0040] The processing circuit may include one or more processors, or all or part of the circuitry in one or more processors used for control or processing functions.

[0041] Optionally, the apparatus may further include a memory for storing programs or instructions, and the processor for running the programs or instructions to implement the methods as described in the first aspect or any possible implementation thereof, or to implement the methods as described in the second aspect or any possible implementation thereof.

[0042] Optionally, the device may also include the transceiver circuit, or an input / output interface.

[0043] In a sixth aspect, a chip is provided, including processing circuitry for running a program or instructions to cause the method as described in the first aspect or any possible implementation thereof to be implemented, or to cause the method as described in the second aspect or any possible implementation thereof to be implemented.

[0044] Optionally, the chip may further include a memory for storing programs or instructions.

[0045] Optionally, the chip may also include transceiver circuitry, or input / output interfaces.

[0046] A seventh aspect provides a computer-readable storage medium comprising instructions that, when executed by a processor, cause the method as described in the first aspect or any possible implementation thereof to be implemented, or cause the method as described in the second aspect or any possible implementation thereof to be implemented.

[0047] Eighthly, a computer program product is provided, the computer program product comprising computer program code or instructions, which, when executed, cause the method as described in the first aspect and any possible implementation thereof to be implemented, or cause the method as described in the second aspect and any possible implementation thereof to be implemented.

[0048] Ninth aspect, a communication system is provided, the communication system including means for performing the first aspect and any possible implementation thereof, or including means for performing the second aspect and any possible implementation thereof.

[0049] It should be understood that the third to ninth aspects of this application correspond to the technical solutions of the first to second aspects of this application, and the beneficial effects obtained by each aspect and the corresponding feasible implementation are similar, and will not be repeated here. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of a communication system applicable to the communication method in the embodiments of this application.

[0051] Figure 2 This is a schematic diagram of another communication system applicable to the communication method of the embodiments of this application.

[0052] Figure 3 This is a schematic diagram of a possible application framework in a communication system.

[0053] Figure 4 This is a schematic diagram of another possible application framework in a communication system.

[0054] Figure 5 This is a schematic diagram of a frequency-hopping SRS transmission.

[0055] Figure 6 This is a schematic diagram of CSI feedback using an AE model.

[0056] Figure 7 This is an example of a neuron structure.

[0057] Figure 8 It is an example of a deep neural network (DNN) structure.

[0058] Figure 9 This is a schematic diagram illustrating the data set connection between the network side and the terminal side.

[0059] Figure 10 This is a schematic diagram illustrating the model interoperability between the network side and the terminal side.

[0060] Figure 11 This is a schematic flowchart of the communication method provided in this application.

[0061] Figure 12 This is a schematic diagram of the joint training process on the network side.

[0062] Figure 13 This is a schematic diagram of the model training process on the terminal side.

[0063] Figure 14 and Figure 15 This is a schematic block diagram of the communication device provided in the embodiments of this application;

[0064] Figure 16 This is a schematic block diagram of the AI ​​processor provided in the embodiments of this application.

[0065] Figure 17 This is a schematic diagram illustrating the application of a dual-ended model encoder and decoder in a channel information processing scenario, as provided in an embodiment of this application. Detailed Implementation

[0066] The technical solutions in this application will now be described with reference to the accompanying drawings.

[0067] To facilitate understanding of the embodiments of this application, the following points will be explained first:

[0068] First, in this application, the terminal side can also be referred to as the UE (user equipment) side, including: terminal equipment, components deployed in the terminal equipment (such as circuits or chips inside the terminal equipment), equipment deployed outside the terminal equipment (such as over-the-top (OTT) hosts or cloud servers), or components deployed in equipment outside the terminal equipment (such as circuits or chips inside the equipment). The network (NW) side includes: network equipment communicating with the terminal equipment, components deployed in the network equipment (such as circuits or chips with near real-time RAN intelligent control functions inside the network equipment), equipment deployed outside the network equipment (such as intelligent network elements, for example, intelligent network elements with near real-time RAN intelligent control functions), or components deployed in the intelligent network element (such as circuits or chips inside the intelligent network element). The network equipment may include: access network equipment, core network equipment, or operation administration and maintenance (OAM) equipment.

[0069] Second, in this application, the indication includes direct indication (also known as explicit indication) and indirect indication (also known as implicit indication). Directly indicating information A means including information A; indirectly indicating information A can mean indicating information A through the correspondence between information A and information B and by directly indicating information B; or by indicating information A through a preset rule that can be used to determine A based on B and by directly indicating information B. The correspondence between information A and information B, and the preset rule, can be predefined, pre-stored, pre-burned, or pre-configured.

[0070] Third, in this application, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates an "or" relationship between the preceding and following related objects, but it does not exclude the possibility of indicating an "and" relationship; the specific meaning can be understood in context. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can mean: a, b, c; a and b; a and c; b and c; or a and b and c. Here, a, b, and c can be single or multiple.

[0071] Fourth, in this application, the use of prefixes such as "first" and "second" is merely for the purpose of distinguishing and describing different things belonging to the same category, and does not constrain the order, size, or quantity of things. For example, "first information" and "second information" are simply different information, and do not limit the quantity of information, the order of transmission, or the relationship of priority.

[0072] Fifth, in this application, "send" and "receive" indicate the direction of signal transmission. For example, "send information to XX" can be understood as the destination of the information being XX, which can include direct transmission via the air interface or indirect transmission by other units or modules via the air interface. "Receive information from YY" can be understood as the source of the information being YY, which can include direct reception from YY via the air interface or indirect reception from YY by other units or modules via the air interface. "Send" can also be understood as the "output" of the chip interface, and "receive" can also be understood as the "input" of the chip interface. In other words, sending and receiving can occur between devices, such as between a terminal device and a computing node, or within a device, such as between components, modules, chips, software modules, or hardware modules within the device via a bus, wiring, or interface.

[0073] Sixth, in the embodiments of this application, "when," "if," and "if" all refer to the device making corresponding processing under certain objective circumstances, and are not limited to a time, nor do they require the device to make a judgment action when it is implemented, nor do they mean that there are other limitations.

[0074] Seventh, in this application, the words "example," "exemplarily," "for example," or "such as" are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "example," "exemplarily," "for example," or "such as" in this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "example," "exemplarily," "for example," or "such as" is intended to present the relevant concepts in a specific manner.

[0075] The technical solutions provided in this application can be applied to various communication systems, such as: 5th generation (5G) or new radio (NR) systems, long term evolution (LTE) systems, LTE frequency division duplex (FDD) systems, LTE time division duplex (TDD) systems, wireless local area network (WLAN) systems, satellite communication systems, future communication systems, or integrated systems of multiple systems. The technical solutions provided in this application can also be applied to device-to-device (D2D) communication, vehicle-to-everything (V2X) communication, machine-to-machine (M2M) communication, machine-type communication (MTC), and Internet of Things (IoT) communication systems or other communication systems.

[0076] In a communication system, one network element can send signals to or receive signals from another network element. These signals can include information, signaling, or data. The term "network element" can also be replaced by an entity, network entity, device, communication equipment, communication module, node, communication node, etc. This disclosure uses a network element as an example. For instance, a communication system can include at least one terminal device and at least one network device. The network device can send downlink signals to the terminal device, and / or the terminal device can send uplink signals to the network device. It is understood that the terminal device in this disclosure can be replaced by a first network element, and the network device can be replaced by a second network element, both performing the corresponding communication methods described in this disclosure.

[0077] Figure 1 This is a schematic diagram of a communication system applicable to the communication method in the embodiments of this application. For example... Figure 1 As shown, the communication system 100A may include at least one access network device, such as Figure 1The access network device 110 shown; the communication system 100A may also include at least one terminal device, such as Figure 1 Terminal devices 120 and 130 are shown. Access network device 110 can communicate with terminal devices (such as terminal devices 120 and 130) via a wireless link. Communication devices in this communication system, for example, between access network device 110 and terminal device 120, can communicate via multi-antenna technology.

[0078] In wireless communication networks, such as mobile communication networks, the services supported by the networks are becoming increasingly diverse, thus requiring increasingly diverse demands. For example, networks need to support ultra-high speeds, ultra-low latency, and / or massive connectivity. This characteristic makes network planning, network configuration, and / or resource scheduling increasingly complex. Furthermore, as network functions become more powerful, such as supporting higher spectrum levels, supporting higher-order multiple-input multiple-output (MIMO) technologies, supporting beamforming (BF), and supporting beam management, network energy efficiency has become a hot research topic. These new demands, new scenarios, and new characteristics bring unprecedented challenges to network planning, operation, and efficient operation. To meet these challenges, artificial intelligence (AI) technology can be introduced into wireless communication networks to achieve network intelligence. To support AI technology in wireless networks, AI nodes may also be introduced.

[0079] Figure 2 This is a schematic diagram of another communication system applicable to the communication method in the embodiments of this application. Compared to Figure 1 Regarding the communication system 100A shown, Figure 2 The communication system 100B shown also includes an AI network element 140. The AI ​​network element 140 is used to perform AI-related operations, such as building training datasets or training AI models. The AI ​​network element can also be referred to as an intelligent network element.

[0080] In one possible implementation, access network device 110 can send data related to the training of the AI ​​model to AI network element 140, which then constructs a training dataset and trains the AI ​​model. For example, the data related to the training of the AI ​​model may include data reported by terminal devices. AI network element 140 can send the results of operations related to the AI ​​model to access network device 110, which then forwards them to the terminal devices. For example, the results of operations related to the AI ​​model may include at least one of the following: a trained AI model, model evaluation results, or test results. Exemplarily, a portion of the trained AI model may be deployed on access network device 110, and another portion on terminal devices 120 and / or 130. Alternatively, the trained AI model may be deployed on access network device 110. Or, the trained AI model may be deployed on terminal devices 120 and / or 130.

[0081] It should be understood that Figure 2 This explanation only uses the direct connection between AI network element 140 and access network device 110 as an example. In other scenarios, AI network element 140 can also be connected to terminal devices. Alternatively, AI network element 140 can be connected to both access network device 110 and terminal devices simultaneously. Alternatively, AI network element 140 can also be connected to access network device 110 through a third-party network element. This application embodiment does not limit the connection relationship between AI network element and other network elements. For example, AI network element 140 can also be configured as a module in access network device and / or terminal device, for example, configured in... Figure 1 In the access network device 110 or terminal device shown.

[0082] It should be noted that, Figure 1 and Figure 2 This is a simplified illustration for ease of understanding only. For example, the communication system may also include other devices, such as wireless relay devices and / or wireless backhaul devices. Figure 1 and Figure 2 The figures are not shown. In practical applications, this communication system may include multiple access network devices or multiple terminal devices. This application does not limit the number of access network devices and terminal devices included in the communication system.

[0083] In the embodiments of this application, the terminal device may also be referred to as 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 equipment.

[0084] Terminal devices can be devices that provide voice / data, such as handheld devices with wireless connectivity, in-vehicle devices, etc. Currently, examples of terminals include: mobile phones, tablets, laptops, PDAs, mobile internet devices (MIDs), wearable devices, virtual reality (VR) devices, augmented reality (AR) devices, wireless terminals in industrial control, wireless terminals in self-driving, wireless terminals in remote medical surgery, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, wireless terminals in smart homes, cellular phones, cordless phones, session initiation protocol (SIP) phones, wireless local loop (WLL) stations, personal digital assistants (PDAs), handheld devices with wireless communication capabilities, computing devices or other processing devices connected to a wireless modem, wearable devices, terminal devices in 5G networks, or future public land mobile communication networks. Terminal devices in a network (PLMN), etc., are not limited to this in the embodiments of this application.

[0085] By way of example and not limitation, in this embodiment, the terminal device can also be a wearable device. Wearable devices, also known as wearable smart devices, are a general term for devices that utilize wearable technology to intelligently design and develop everyday wearables, such as glasses, gloves, watches, clothing, and shoes. Wearable devices are portable devices that are worn directly on the body or integrated into the user's clothing or accessories. Wearable devices are not merely hardware devices, but also achieve powerful functions through software support, data interaction, and cloud interaction. Broadly speaking, wearable smart devices include those that are feature-rich, large in size, and can achieve complete or partial functions without relying on a smartphone, such as smartwatches or smart glasses, as well as those that focus on a specific type of application function and require the use of other devices such as smartphones, such as various smart bracelets and smart jewelry for vital sign monitoring.

[0086] In this embodiment, the device for implementing the functions of the terminal device can be the terminal device itself, or it can be any device capable of supporting the terminal device in implementing those functions, such as a chip system. This device can be installed in or used in conjunction with the terminal device. In this embodiment, the chip system can be composed of chips or may include chips and other discrete components. This embodiment only uses the terminal device as an example to illustrate the device for implementing the functions of the terminal device, and does not constitute a limitation on the solution of this embodiment.

[0087] The access network device in this application embodiment can be a device used to communicate with terminal devices. This access network device can also be called a network device, such as a base station. In this application embodiment, the access network device can refer to a RAN node (or device) that connects the terminal device to the wireless network. A base station can broadly encompass, or be replaced by, various names such as: NodeB, evolved NodeB (eNB), next-generation NodeB (gNB), relay station, access point, transmitting and receiving point (TRP), transmitting point (TP), master station, auxiliary station, motor slide retainer (MSR) node, home base station, network controller, access node, wireless node, access point (AP), transmission node, transceiver node, baseband unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), central unit (CU), distributed unit (DU), radio unit (RU), positioning node, etc. A base station can be a macro base station, micro base station, relay node, donor node, or similar entities, or combinations thereof. A base station can also refer to a communication module, modem, or chip installed within the aforementioned equipment or apparatus. A base station can also be a mobile switching center, a device that performs base station functions in D2D, V2X, and M2M communications, or a device that performs base station functions in future communication systems. A base station can support networks using the same or different access technologies. Optionally, a RAN node can also be a server, wearable device, vehicle, or in-vehicle equipment. For example, the access network equipment in V2X technology can be a roadside unit (RSU). The embodiments of this application do not limit the specific technology or equipment form used in the access network equipment.

[0088] Base stations can be fixed or mobile. For example, a helicopter or drone can be configured to act as a mobile base station, and one or more cells can move depending on the location of the mobile base station. In other examples, a helicopter or drone can be configured as a device to communicate with another base station.

[0089] In some deployments, the access network equipment mentioned in this application may be equipment including a CU, or a DU, or equipment including both CU and DU, or equipment with a control plane CU node (central unit-control plane (CU-CP)) and a user plane CU node (central unit-user plane (CU-UP)) and a DU node. For example, the access network equipment may include gNB-CU-CP, gNB-CU-UP, and gNB-DU.

[0090] In some deployments, multiple RAN nodes collaborate to assist terminals in achieving wireless access, with different RAN nodes each implementing some of the base station's functions. For example, RAN nodes can be CUs, DUs, CU-CPs, CU-UPs, or RUs. CUs and DUs can be configured separately or included in the same network element, such as a BBU. RUs can be included in radio frequency equipment or radio frequency units, such as RRUs, AAUs, or RRHs.

[0091] RAN nodes can support one or more types of fronthaul interfaces, each corresponding to a DU and RU with different functions. If the fronthaul interface between the DU and RU is a Common Public Radio Interface (CPRI), the DU is configured to implement one or more baseband functions, and the RU is configured to implement one or more radio frequency functions. If the fronthaul interface between the DU and RU is another type of interface, relative to CPRI, it moves some downlink and / or uplink baseband functions—for example, for downlink, precoding, digital beamforming, or one or more of inverse fast Fourier transform (IFFT) / cyclic prefix addition (CP)—from the DU to the RU; and for uplink, digital beamforming, or one or more of fast Fourier transform (FFT) / cyclic prefix removal (CP)—from the DU to the RU. In one possible implementation, this interface can be an enhanced common public radio interface (eCPRI). Under the eCPRI architecture, the splitting methods between DU and RU are different, corresponding to different types (category, Cat) of eCPRI, such as eCPRI Cat A, B, C, D, E, F.

[0092] Taking eCPRI Cat A as an example, for downlink transmission, layer mapping is used as the dividing line. The DU is configured to implement one or more functions preceding layer mapping (i.e., coding, rate matching, scrambling, modulation, and layer mapping itself), while other functions following layer mapping (e.g., resource element (RE) mapping, digital beamforming, or one or more of IFFT / CP addition) are implemented in the RU. For uplink transmission, de-RE mapping is used as the dividing line. The DU is configured to implement one or more functions preceding de-mapping (i.e., decoding, rate matching de-matching, descrambling, demodulation, inverse discrete Fourier transform (IDFT), channel equalization, and de-RE mapping itself), while other functions following de-mapping (e.g., digital BF or FFT / CP removal) are implemented in the RU. It is understood that descriptions of the functions of the DU and RU corresponding to various types of eCPRI can be found in the eCPRI protocol and will not be elaborated upon here.

[0093] In one possible design, the processing unit in the BBU used to implement baseband functions is called the baseband high (BBH) unit, and the processing unit in the RRU / AAU / RRH used to implement baseband functions is called the baseband low (BBL) unit.

[0094] In different systems, CU (or CU-CP and CU-UP), DU, or RU may have different names, but those skilled in the art will understand their meaning. For example, in an open RAN (ORAN) architecture, CU can also be called open CU (open-CU, O-CU), DU can also be called open DU (open-DU, O-DU), CU-CP can also be called open CU-CP (open-CU-CP) O-CU-CP, CU-UP can also be called open CU-UP (open-CU-UP, O-CU-UP), and RU can also be called open RU (open-RU, O-RU). Any of the units among CU (or CU-CP, CU-UP), DU, and RU in this application can be implemented through software modules, hardware modules, or a combination of software modules and hardware modules.

[0095] In this embodiment, the apparatus for implementing the functions of a network device can be a network device itself; it can also be an apparatus capable of supporting the network device in implementing those functions, such as a chip system, hardware circuit, software module, or a hardware circuit plus a software module. This apparatus can be installed in the network device or used in conjunction with the network device. In this embodiment, the example of a network device being used to implement the functions of a network device is provided only and does not constitute a limitation on the solutions described in this embodiment.

[0096] Network devices and / or terminal devices can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted; they can also be deployed on water; and they can also be deployed in the air on airplanes, balloons, and satellites. This application does not limit the scenario in which the network devices and terminal devices are located. Furthermore, terminal devices and network devices can be hardware devices, or software functions running on dedicated hardware or general-purpose hardware, such as virtualization functions instantiated on a platform (e.g., a cloud platform), or entities that include dedicated or general-purpose hardware devices and software functions. This application does not limit the specific form of the terminal devices and network devices.

[0097] Optionally, the AI ​​node can be deployed in one or more of the following locations within the communication system: access network equipment, terminal equipment, or core network elements. Alternatively, the AI ​​node can also be deployed independently, for example, in a location other than any of the aforementioned devices, such as in the host or cloud server of an OTT system. The AI ​​node can communicate with other devices in the communication system, which can be one or more of the following: access network equipment, terminal equipment, or core network elements.

[0098] It is understood that this application does not limit the number of AI nodes. For example, when there are multiple AI nodes, they can be divided based on function, such as different AI nodes being responsible for different functions.

[0099] It can also be understood that AI nodes can be independent devices, or they can be integrated into the same device to achieve different functions. Alternatively, they can be network elements in hardware devices, software functions running on dedicated hardware, or virtualization functions instantiated on a platform (e.g., a cloud platform). This application does not limit the specific form of the aforementioned AI nodes.

[0100] AI nodes can be AI network elements or AI modules.

[0101] Figure 3 This is a schematic diagram of a possible application framework in a communication system. For example... Figure 3 As shown, network elements in a communication system are connected via interfaces (such as next-generation (NG) interfaces or Xn interfaces) or air interfaces. The NG interface is the interface between the radio access network and the 5G core network. The Xn interface is the interface between access network devices, and the air interface is the interface between access network devices and terminal devices. These network element nodes, such as core network devices, access network nodes (RAN nodes), terminals, or one or more devices in the OAM, are equipped with one or more AI modules (for clarity, ...). Figure 3 (Only one is shown in the image). The access network node can be a single RAN node or can include multiple RAN nodes, such as RU, CU, and DU. One or more of the RU, CU, or DU can also be equipped with one or more AI modules. Optionally, the CU can also be split into CU-CP and CU-UP. One or more AI modules are provided in CU-CP and / or CU-UP.

[0102] The AI ​​module is used to implement corresponding AI functions. AI modules deployed in different network elements can be the same or different. Depending on the parameter configuration, the AI ​​module can implement different functions. The AI ​​module model can be configured based on one or more of the following parameters: structural parameters (e.g., at least one of the following: number of neural network layers, neural network width, inter-layer connections, neuron weights, neuron activation function, or bias in the activation function), input parameters (e.g., type and / or dimension of input parameters), or output parameters (e.g., type and / or dimension of output parameters). The bias in the activation function can also be referred to as the neural network bias.

[0103] An AI module can have one or more models. A model can infer an output, which includes one or more parameters. The learning, training, or inference processes of different models can be deployed on different nodes or devices, or they can be deployed on the same node or device.

[0104] The network device can be a network device equipped with one or more AI modules. The network device may include... Figure 3 The core network equipment, access network node (RAN node), or one or more devices in the OAM are shown. For example, the AI ​​module can be... Figure 4 The RAN intelligent controller (RIC) shown can be a near-real-time RIC (near-RT RIC) or a non-real-time RIC (non-RT RIC). For example, a near-real-time RIC is located in a RAN node (e.g., in a CU or DU), while a non-real-time RIC is located in an OAM, a cloud server, a core network device, or other access network devices. The RIC can obtain subsets from multiple terminal devices from RAN nodes (e.g., CU, CU-CP, CU-UP, DU, and / or RU), reassemble them into a training dataset, and then be trained based on this training dataset. Exemplarily, near-real-time RICs and non-real-time RICs can also be set up as separate network elements, and access network devices can be either near-real-time RICs or non-real-time RICs.

[0105] Figure 4 This is a schematic diagram of another possible application framework in a communication system. Figure 4 The communication system shown includes access network nodes (CU, DU, and RU shown in the figure) and terminals, as well as a RIC. For example, the RIC could be... Figure 3The AI ​​module shown can be used to implement AI-related functions. The RIC includes near real-time RIC and non-real-time RIC. Non-real-time RIC primarily processes non-real-time information, such as data that is not sensitive to latency, with latency on the order of seconds. Real-time RIC primarily processes near real-time information, such as data that is relatively sensitive to latency, with latency on the order of tens of milliseconds.

[0106] The near real-time RIC is used for model training and inference. For example, it can be used to train an AI model and then use that AI model for inference. The near real-time RIC can obtain network-side and / or terminal-side information from RAN nodes (e.g., CU, CU-CP, CU-UP, DU, and / or RU) and / or terminals. This information can be used as training data or inference data. Optionally, the near real-time RIC can deliver inference results to RAN nodes and / or terminals. Optionally, inference results can be exchanged between CU and DU, and / or between DU and RU. For example, the near real-time RIC delivers the inference result to the DU, and the DU sends it to the RU.

[0107] The non-real-time RIC is also used for model training and inference. For example, it can be used to train an AI model and then use that model for inference. The non-real-time RIC can obtain network-side and / or terminal-side information from RAN nodes (e.g., one or more of CU, CU-CP, CU-UP, DU, or RU) and / or terminals. This information can be used as training data or inference data, and the inference results can be delivered to the RAN nodes and / or terminals. Optionally, inference results can be exchanged between CU and DU, and / or between DU and RU; for example, the non-real-time RIC delivers the inference results to the DU, which then forwards them to the RU.

[0108] The near real-time RIC and non-real-time RIC can also be set up as separate network elements. Optionally, the near real-time RIC and non-real-time RIC can also be part of other devices. For example, the near real-time RIC can be set in the RAN node (e.g., in CU, DU), while the non-real-time RIC can be set in the OAM, cloud server, core network device, or other network device.

[0109] With the development of wireless communication technology and the increasing number of supported services, higher demands are being placed on communication systems in terms of system capacity, communication latency, and other indicators. Among these advancements, massive MIMO (Multi-User MIMO) systems, by configuring large-scale antenna arrays at the transceiver end, can achieve spatial diversity gain, thereby significantly increasing system capacity. For example, access network equipment can simultaneously transmit data to multiple terminal devices using the same time-frequency resources, i.e., multi-user MIMO (MU-MIMO); or, access network equipment can simultaneously transmit multiple data streams to the same terminal device, i.e., single-user MIMO (SU-MIMO). The data between these multiple terminal devices or the multiple data streams within the same terminal device are spatially multiplexed, thus becoming a key direction in the evolution of communication systems.

[0110] Access network equipment can obtain the CSI of the downlink channel, which is used to determine one or more of the following configurations of the downlink data channel of the scheduling terminal equipment: resources, modulation and coding scheme (MCS), precoding, etc.

[0111] Taking precoding as an example, in massive MIMO, access network equipment needs to precode downlink data using a precoding matrix. By using precoding techniques, access network equipment can achieve spatial multiplexing between terminal devices or data streams. This means that data between different terminal devices or between different data streams of the same terminal device is spatially isolated, thereby reducing interference between different terminal devices or different data streams and improving the received signal-to-interference-plus-noise ratio (SINR) of the terminal devices. To calculate the precoding matrix, the access network equipment needs to obtain the CSI of the downlink channel and determine the precoding matrix based on the CSI.

[0112] In TDD systems, due to the reciprocity of uplink and downlink channels, access network devices can obtain the uplink CSI by measuring the uplink reference signal and then infer a relatively accurate downlink CSI, for example, using the uplink CSI as the downlink CSI. However, in FDD systems, uplink and downlink reciprocity cannot be guaranteed. The downlink CSI is obtained by the terminal device measuring the downlink reference signal, such as the channel state information reference signal (CSI-RS) or the synchronizing signal block (SSB). Therefore, the terminal device needs to generate a CSI report according to the protocol predefined method or the access network device configuration, and then feed the generated CSI report back to the access network device so that it can obtain the downlink CSI.

[0113] In FDD systems, a crucial part of CSI feedback is the precoding matrix indicator (PMI), which uses 0-1 bits in the CSI to quantize the channel matrix or precoding matrix. PMI design (also known as codebook design) is a fundamental issue in mobile communication systems. Traditional codebook design methods predefine (or agree upon) a series of precoding matrices and their corresponding numbers in the protocol; these precoding matrices are called codewords. The channel matrix or precoding matrix can be approximated using predefined codewords or linear combinations of multiple predefined codewords. Therefore, the terminal device can feed back the corresponding codeword numbers and one or more weighting coefficients to the access network device via the PMI, which is used by the access network device to reconstruct the channel matrix or precoding matrix.

[0114] As the antenna array size of MIMO systems continues to increase, the number of supported antenna ports also increases, leading to a growth in the dimensionality of the corresponding channel matrix and precoding matrix. To enable terminal devices to estimate (or measure) the downlink channel, the overhead of the access network equipment transmitting reference signals increases. Simultaneously, the error of approximating large-scale channel matrices and precoding matrices with a finite number of predefined codewords increases. One method to improve channel reconstruction accuracy is to increase the number of codewords in the codebook, but this simultaneously increases the overhead of CSI feedback (including the corresponding codeword number and one or more weighting coefficients), thereby reducing the available resources for data transmission and causing system capacity loss. Therefore, it is necessary to investigate how to more effectively compress and represent channel information without increasing the overhead of transmitting reference signals and CSI feedback, and how to more effectively reconstruct the channel based on feedback information. Correlation exists between different elements in the downlink channel matrix between access network equipment and terminal equipment; furthermore, correlation exists between downlink channel matrices in different time slots. For example, the correlation between different elements in the channel matrix implies the existence of a basis (which can be represented by matrices U1 and U2). Projecting the channel matrix H onto this basis yields a sparse equivalent channel, i.e., H' = U1H. H U2 is a sparse matrix, where the superscript H denotes the conjugate transpose operation. Theoretically, the channel matrix H can be reconstructed simply by estimating the non-zero elements in H' using the reference signal and feeding them back. Therefore, the overhead of the reference signal transmission and CSI feedback has room for compression. However, traditional CSI feedback schemes, such as the codebook-based feedback method mentioned above, do not fully utilize the channel compression space, and the channel compression process may cause significant information loss. Machine learning methods (such as deep learning (DL)) have stronger nonlinear feature extraction capabilities, thus enabling more effective extraction of correlations between channel matrices. Consequently, compared to traditional schemes, they can more effectively compress and represent channel information, and more effectively reconstruct channel information based on the feedback information.

[0115] In some implementations, neural network models can be used for compressed feedback of channel information. After the terminal measures the reference signal to obtain channel information, it inputs this information into its neural network model to obtain compressed channel information. The terminal then feeds back the compressed channel information to the network device. The network device inputs this compressed information into its neural network model to reconstruct the channel information measured by the terminal. This leverages the nonlinear feature extraction capabilities of neural networks to improve the accuracy of channel measurements.

[0116] As an example, such as Figure 17The diagram illustrates the application of a dual-ended model encoder and decoder in a channel information processing scenario. The terminal model is the encoder, used to compress CSI-RS measurement data. The terminal sends the compressed CSI-RS measurement data to the network device. Upon receiving the compressed CSI-RS measurement data, the network device uses its decoder to reconstruct the CSI-RS measurement data, obtaining the final CSI-RS measurement data. This allows for large-port, full-band channel measurement between the terminal and network device with limited CSI-RS measurement overhead, based on a dual-ended model. During CSI-RS measurement, multiple resource sets can be used to achieve multi-port channel measurement. Currently, one resource set supports a maximum of 32 antenna ports for channel measurement; adding multiple resource sets can support channel measurement with larger arrays. In related technologies, to support high-bandwidth channel measurement, the number of resource blocks (RBs) that the terminal needs to measure can be indicated. The density indicates the number of channels measured within one RB, and the value can be 0.5, 1, or 3.

[0117] In the embodiments of this application, combined with Figure 17 In the CSI-RS measurement scenario shown, the terminal can measure only a portion of the CSI-RS and compress that portion of the CSI-RS measurement data; correspondingly, the network device recovers the compressed portion of the CSI-RS measurement data to obtain the full CSI-RS measurement data. Alternatively, the terminal can measure only a portion of the CSI-RS and compress that portion of the CSI-RS measurement data; however, the network device not only recovers the compressed portion of the CSI-RS measurement data to obtain the full CSI-RS measurement data, but it can also predict the unmeasured CSI-RS measurement data to obtain the full CSI-RS measurement data. Alternatively, the terminal can measure only a portion of the CSI-RS and compress that portion of the CSI-RS measurement data; after the terminal sends the compressed portion of the CSI-RS measurement data multiple times, the network device predicts the full CSI-RS measurement data based on these multiple transmissions of the compressed portion of the CSI-RS measurement data.

[0118] To facilitate understanding of the embodiments of this application, the terms involved in this application will be briefly explained below.

[0119] 1. Artificial Intelligence (AI): This refers to enabling machines to learn, accumulate experience, and solve problems that humans can solve through experience, such as natural language understanding, image recognition, and chess. AI can be understood as the intelligence exhibited by machines created by humans. Generally, AI refers to the technology of using computer programs to represent human intelligence. The goals of AI include understanding intelligence by constructing computer programs that demonstrate symbolic reasoning or logical reasoning.

[0120] 2. Machine Learning (ML): This is an implementation method of artificial intelligence. Machine learning is a method that endows machines with the ability to perform functions that cannot be directly programmed. In practical terms, machine learning is a method of training a model using data and then using the model to make predictions. There are many machine learning methods, such as neural networks (NN), decision trees, and support vector machines (SVM). Machine learning theory mainly involves designing and analyzing algorithms that allow computers to learn automatically. Machine learning algorithms are a class of algorithms that automatically analyze data to obtain patterns and use these patterns to predict unknown data. Machine learning can be divided into supervised learning, unsupervised learning, and reinforcement learning.

[0121] Supervised learning, based on collected sample values ​​and labels, uses machine learning algorithms to learn the mapping relationship between sample values ​​and labels, and expresses this learned mapping relationship using a machine learning model. The process of training the machine learning model is the process of learning this mapping relationship. For example, in signal detection, the noisy received signal is the sample, and the corresponding real constellation point is the label. Machine learning aims to learn the mapping relationship between samples and labels through training, that is, to enable the machine learning model to learn a signal detector. During training, the model parameters are optimized by calculating the error between the model's predicted values ​​and the real labels. Once the mapping relationship is learned, it can be used to predict the sample label of each new sample. The mapping relationship learned in supervised learning can include linear mappings and nonlinear mappings. Based on the type of label, the learning task can be divided into classification tasks and regression tasks.

[0122] Unsupervised learning relies solely on collected sample values, using algorithms to discover inherent patterns within the samples. One type of unsupervised learning algorithm uses the samples themselves as supervisory signals; that is, the model learns the mapping relationship from sample to sample, which is called self-supervised learning. During training, model parameters are optimized by calculating the error between the model's predictions and the samples themselves. Self-supervised learning can be used for signal compression and decompression recovery applications; common algorithms include autoencoders and generative adversarial networks.

[0123] Reinforcement learning, unlike supervised learning, is a type of algorithm that learns problem-solving strategies through interaction with the environment. Unlike supervised and unsupervised learning, reinforcement learning problems do not have explicit "correct" action labels. The algorithm needs to interact with the environment to obtain reward signals from the environment, and then adjust its decision actions to obtain a larger reward signal value. For example, in downlink power control, the reinforcement learning model adjusts the downlink transmission power of each user based on the total system throughput feedback from the wireless network, aiming to achieve a higher system throughput. The goal of reinforcement learning is also to learn the mapping relationship between the environment state and the optimal / near-optimal decision actions. However, because the label of the "correct action" cannot be obtained in advance, the network cannot be optimized by calculating the error between the action and the "correct action." Reinforcement learning training is achieved through iterative interaction with the environment.

[0124] 3. AI Model: This can be an algorithm or computer program that implements AI functions. An AI model represents the mapping relationship between the model's input and output. An AI model can be understood as a function model that maps an input of a certain dimension to an output of a certain dimension, and its parameters are obtained through machine learning training. For example, f(x) = ax 2 +b is a quadratic function model, which can be viewed as an AI model. a and b correspond to the parameters of this AI model, and can be obtained through machine learning training. An AI model can also be called a model, an AI function, or a feature. One AI function can correspond to one or more AI models.

[0125] AI models can be neural networks, linear regression models, decision tree models, SVMs, Bayesian networks, Q-learning models, or other machine learning models.

[0126] The implementation of an AI model can be a hardware circuit, software, or a combination of both; there are no restrictions. Non-restrictive examples of software include: program code, program, subroutine, instruction, instruction set, code, code segment, software module, application program, or software application, etc.

[0127] 4. Neural Network (NN): A specific implementation of AI or machine learning. According to the general approximation theorem, neural networks can theoretically approximate any continuous function, thus enabling them to learn arbitrary mappings.

[0128] Neural networks can be composed of neural units, which can refer to units represented by x. sThe neural network is a computational unit that takes an intercept of 1 as input. A neural network is a network formed by connecting many of these individual neural units together; that is, the output of one neural unit can be the input of another. The input of each neural unit can be connected to the local receptive field of the previous layer to extract features from the local receptive field, which can be a region composed of several neural units.

[0129] Taking neural networks as an example of AI models, the AI ​​model involved in this application can be a deep neural network (DNN). The idea behind DNNs originates from the neuronal structure of the brain. Each neuron performs a weighted summation of its input values ​​and generates an output through a non-linear function. DNNs generally have a multi-layered structure, with each layer containing multiple neurons. The input layer processes the received values ​​through neurons and then passes them to the hidden layers. Similarly, the hidden layers pass the calculation results to the final output layer, producing the final output of the DNN. DNNs typically have more than one hidden layer, which often directly affects the ability to extract information and fit functions. Increasing the number of hidden layers or widening the width of each layer can improve the function fitting ability of the DNN. The weighted values ​​in each neuron are the parameters of the DNN network model. The model parameters are optimized through the training process, enabling the DNN network to extract data features and express mapping relationships.

[0130] Depending on how the network is constructed, DNNs can include feedforward neural networks (FNNs), convolutional neural networks (CNNs), and recurrent neural networks (RNNs), etc.

[0131] CNNs are neural networks specifically designed to process data with a grid-like structure. For example, time-series data (discrete sampling along the time axis) and image data (two-dimensional discrete sampling) can both be considered grid-like data. CNNs do not use all the input information at once for computation; instead, they use a fixed-size window to extract a portion of the information for convolution operations, which significantly reduces the computational cost of model parameters. Furthermore, depending on the type of information extracted by the window (such as people and objects in an image representing different types of information), each window can use different convolution kernels, allowing CNNs to better extract features from the input data.

[0132] Recurrent Neural Networks (RNNs) are a type of distributed neural network (DNN) that utilizes feedback time-series information. Their input includes the current input value and their own output value from the previous time step. RNNs are well-suited for acquiring temporally correlated sequence features, and are particularly applicable to applications such as speech recognition and channel coding / decoding.

[0133] The characteristic of FNN networks is that neurons in adjacent layers are completely connected to each other, which makes FNNs typically require a large amount of storage space and result in high computational complexity.

[0134] The FNN, CNN, and RNN mentioned above are all constructed based on neurons. As mentioned earlier, each neuron performs a weighted summation operation on its respective input value and generates an output through a nonlinear function. The weights of the weighted summation operation of neurons in a neural network and the nonlinear function are called the parameters of the neural network. The parameters of all neurons in a neural network constitute the parameters of that neural network.

[0135] 5. Dataset: This refers to the data used in machine learning for model training, validation, and testing. The quantity and quality of the data will affect the effectiveness of machine learning.

[0136] In the field of machine learning, ground truth usually refers to data that is considered accurate or real.

[0137] Training datasets can be used to train AI models. A training dataset can include the input to the AI ​​model, or it can include both the input and the target output of the AI ​​model. Specifically, a training dataset includes one or more training data points, which can include training samples input to the AI ​​model, or the target output of the AI ​​model. The target output can also be referred to as a label, sample label, or labeled sample. The label is the ground truth value.

[0138] In the field of communications, training datasets can include simulation data collected through simulation platforms, experimental data collected from experimental scenarios, or measured data collected in actual communication networks. Because the geographical environment and channel conditions where the data is generated vary—for example, indoor / outdoor conditions, movement speed, frequency bands, or antenna configurations—the collected data can be categorized during acquisition. For instance, data with the same channel propagation environment and antenna configuration can be grouped together.

[0139] Model training essentially involves learning certain features from training data. In training AI models (such as neural network models), the goal is to make the model's output as close as possible to the desired predicted value. This is achieved by comparing the network's current predictions with the target value and updating the weight vector of each layer based on the difference. (Of course, there's usually an initialization process before the first update, where parameters are pre-configured for each layer.) For example, if the network's prediction is too high, the weight vector is adjusted to predict a lower value. This adjustment continues until the AI ​​model can predict the target value or a value very close to it. Therefore, it's necessary to predefine "how to compare the difference between the predicted and target values," which is the loss function or objective function. These are important equations used to measure the difference between the predicted and target values. Taking the loss function as an example, a higher output value (loss) indicates a greater difference. Therefore, training the AI ​​model becomes a process of minimizing this loss, making the loss function value less than a threshold, or making the loss function value meet the target requirements. For example, assuming the AI ​​model is a neural network, adjusting the model parameters of the neural network includes adjusting at least one of the following parameters: the number of layers of the neural network, the width, the weights of the neurons, or the parameters in the activation function of the neurons.

[0140] Inference data can be used as input to a trained AI model for inference. During the model's inference process, the inference data is input into the AI ​​model, and the corresponding output, which is the inference result, is obtained.

[0141] 6. CSI Feedback: In LTE and NR communication systems, network equipment needs to obtain downlink CSI to determine the resources, modulation and coding scheme (MCS), precoding, and other configurations for scheduling downlink data channels of terminal equipment. CSI can be understood as a type of channel information, reflecting channel characteristics and quality.

[0142] CSI measurement refers to the process by which the receiver deciphers channel information based on a reference signal transmitted by the transmitter, i.e., estimating channel information using channel estimation methods. For example, the reference signal may include one or more of the following: channel state information reference signal (CSI-RS), synchronizing signal / physical broadcast channel block (SSB), sounding reference signal (SRS), or demodulation reference signal (DMRS). CSI-RS, SSB, and DMRS can be used to measure downlink CSI. SRS and DMRS can be used to measure uplink CSI.

[0143] In time-division duplex (TDD) systems, due to the reciprocity of uplink and downlink channels, network devices can obtain the uplink CSI by measuring the uplink reference signal (such as the SRS) and then infer a more accurate downlink CSI, for example, using the uplink CSI as the downlink CSI. In frequency-division duplex (FDD) systems, uplink and downlink reciprocity cannot be guaranteed. The downlink CSI is obtained by the terminal device measuring the downlink reference signal. Therefore, the terminal device needs to be able to generate a CSI report according to a predefined protocol or base station configuration and feed the CSI back to the base station so that it can obtain the downlink CSI. This application will subsequently describe the uplink reference signal using the SRS as an example. It is understood that the SRS in this application can be replaced by the uplink reference signal, or other uplink reference signals used for channel measurement, which are not limited here.

[0144] In the NR protocol, the configuration and reporting process of downlink CSI is roughly as follows: The network device sends a CSI reporting configuration (CSI-ReportConfig) to the terminal device, specifying the reporting type (reportConfigType), reporting quantity (reportQuantity), etc. The reporting type can be periodic, semi-persistent, or aperiodic; the reporting quantity can be a rank indicator (RI), channel quality indicator (CQI), precoding matrix indicator (PMI), reference signal received power (RSRP), etc.

[0145] For example, the network device sends a CSI-RS to the terminal device, which performs channel and interference measurements based on the CSI-RS to obtain measurement results. Based on the measurement results, the terminal device determines the reported quantities for configuration and reports downlink CSI to the network device. This downlink CSI includes one or more of the RI, CQI, PMI, RSRP, etc., measured by the terminal device.

[0146] In this context, RI indicates the number of downlink transmission layers recommended by the terminal device; CQI indicates the modulation and coding schemes that the terminal device determines the current channel conditions can support; and PMI indicates the precoding recommended by the receiver of the reference signal, such as the terminal device. The number of precoding layers indicated by PMI corresponds to RI.

[0147] It should be understood that the RI, CQI, and PMI values ​​indicated in the aforementioned CSI report are recommended values ​​for the terminal equipment. The network equipment may perform downlink transmission according to some or all of the information indicated in the CSI report. Alternatively, the network equipment may also perform downlink transmission without following the information indicated in the CSI report.

[0148] In this embodiment, the meaning of CSI is broader than that in traditional schemes. It is not limited to CQI, PMI, RI, or CSI-RS resource indicator (CRI). It can also be one or more of the following: channel response (such as channel response matrix), channel matrix, channel feature matrix, precoding matrix, RSRP, signal to interference plus noise ratio (SINR) (also known as signal-to-interference-plus-noise ratio), the identity (ID) of the optimal beam, or the IDs of the top K beams. For example, the optimal beam can be the beam with the highest channel quality (e.g., RSRP, SINR) in the beam set. The top K beams can be the K beams in the beam set whose channel quality (e.g., RSRP, SINR) is greater than or equal to a certain threshold, or the K beams ranked first when sorted by channel quality from largest to smallest, where K is a positive integer.

[0149] Among them, the channel response and channel matrix represent the channel itself, while the channel feature matrix and precoding matrix are matrices composed of features extracted from the channel.

[0150] The following is a description of several CSI-related terms used in this application:

[0151] Target CSI: Also known as the full information of CSI or ground-truth CSI. For ease of distinction and explanation, target CSI will be represented by H4 in the following text.

[0152] CSI feedback information: also known as CSI feedback information, channel measurement result feedback information, channel information feedback information, compressed information, compressed channel information, compressed CSI information, compressed channel information, or compressed CSI, etc. For ease of distinction and explanation, CSI feedback information will be represented by C2 in the following text.

[0153] Input CSI: Also known as measurement CSI, raw CSI, uncompressed CSI information, or raw CSI information. Input CSI can be understood as the CSI input to the AI ​​model (such as the first model mentioned below), meaning the input CSI can be the input to the AI ​​model. For ease of distinction and explanation, input CSI will be represented by H1 in the following text.

[0154] It should be understood that in this application, during the training phase, CSI feedback information can refer to the information obtained after compressing the input CSI, which can be simply referred to as compressed CSI. CSI feedback information can also refer to the information obtained by compressing and quantizing the input CSI, which can be simply referred to as quantized CSI. This application does not limit the specific terminology. During the model deployment phase, CSI feedback information can be information sent from the terminal to the network side. To save feedback overhead, CSI feedback information can refer to the information obtained by compressing and quantizing the measured CSI (also known as input CSI), i.e., quantized CSI.

[0155] It should also be understood that quantizing compressed CSI yields quantized CSI; dequantizing quantized CSI yields compressed CSI. Compressed CSI obtained by dequantizing quantized CSI is compressed CSI with quantization loss.

[0156] Reconstructed CSI: Also known as recovered CSI, reconstructed CSI, reconstructed channel information, decompressed CSI, decompressed channel information, or output CSI, etc. Reconstructed CSI can be understood as the CSI output by the AI ​​model; that is, the reconstructed CSI can be the output of the AI ​​model (such as the second model mentioned below). For ease of distinction and explanation, the reconstructed CSI will be referred to as H3 or... express.

[0157] For example, the relationship between the reconstructed CSI and the input CSI can be referenced. Figure 17 The corresponding description.

[0158] 7. Sounding Reference Signal (SRS): This is an uplink channel sounding signal, sent by the terminal equipment and received by the network equipment. The transmission method of the SRS includes the time-frequency resources, transmission beam, transmission power, etc., which are generally configured by the network equipment for the terminal equipment. In the 3rd Generation Partnership Project (3GPP)... rd In the Generation Partnership Project (3GPP) related protocol framework, network devices can configure one or more SRS resource sets for terminal devices, and each SRS resource set contains one or more SRS resources.

[0159] Furthermore, in 3GPP related protocols, different SRS resource sets can perform different functions. SRS resource sets typically support four functions: {beam management (BM), codebook (CB), non-codebook (NCB), and antenna switching (AS)}. Antenna switching can also be called antenna selection. Network devices can configure the usage of each SRS resource set through radio resource control (RRC) signaling, thereby informing terminal devices of the function of the corresponding SRS resource set. For example, when the usage of an SRS resource set is antenna switching, the SRS corresponding to that SRS resource set is generally used to obtain complete uplink channel information. Assuming that in a time division duplex (TDD) system, the channel has uplink and downlink reciprocity (i.e., the uplink and downlink channels are consistent), the SRS corresponding to that SRS resource can also obtain the downlink transmission channel (or downlink transmission precoding) through uplink channel measurement.

[0160] In one possible implementation, the terminal device can transmit SRS via frequency hopping, meaning that multiple SRS transmissions by a single terminal device can switch between different frequency bands within the frequency domain resources. Taking two SRS transmissions by the terminal device as an example, the terminal device can transmit SRS on subband 1 within a specific bandwidth, then switch from subband 1 to subband 2 within the same bandwidth, and then transmit SRS again on subband 2.

[0161] In some cases, such as in the NR protocol, the uplink power of the SRS transmitted by the terminal device to the network device is limited, resulting in low accuracy of the channel state information obtained by the network device based on the received SRS reference signal. To improve the accuracy of channel estimation obtained by the network device based on SRS, the bandwidth of a single SRS transmission by the terminal device can be reduced, and the frequency power spectral density of the SRS can be increased, thereby ensuring the uplink power of a single SRS transmission and improving the accuracy of the channel state information obtained by the network device.

[0162] Figure 5 This is a schematic diagram of a frequency-hopping SRS transmission. Figure 5The diagram illustrates single-bandwidth SRS transmission, two-subband frequency-hopping SRS transmission, and four-subband frequency-hopping SRS transmission. Specifically, when the terminal device transmits SRS in a non-frequency-hopping manner (i.e., single-bandwidth SRS transmission), the SRS transmitted in a single transmission can cover the configured bandwidth of the SRS resource. When the terminal device transmits SRS in a frequency-hopping manner (i.e., multiple subband frequency-hopping SRS transmissions), each SRS transmission can cover a portion of the configured bandwidth of the SRS resource (i.e., one frequency-hopping subband). Multiple SRS transmissions within one frequency-hopping cycle can cover the configured bandwidth of the SRS resource.

[0163] The frequency hopping characteristics of SRS can be determined by parameters in both the time domain and the frequency domain.

[0164] For example, the process of determining the time domain location of the SRS is as follows:

[0165] In the time domain, SRS occupies N slots within a time slot. S There are 1, 2, 4 symbols, and the repetition factor R ∈ {1, 2, 4}, satisfying R ≤ N. S That is, repeating each symbol R times.

[0166] According to the repetitionFactor:

[0167] When R = N S At that time, SRS transmission in frequency hopping mode within a time slot is not supported;

[0168] When R = 1, N S When the frequency hopping frequency is 2 or 4, SRS can be transmitted in a frequency hopping manner within a time slot, specifically in units of one OFDM symbol.

[0169] When R = 2, N S When the value is 4, SRS can be transmitted in a frequency-hopping manner within a time slot, specifically in units of a pair of OFDM symbols (i.e., 2 OFDM symbols).

[0170] For periodic SRS and semi-static SRS, corresponding period and time-domain offset parameters need to be configured. Periodic SRS and semi-static SRS can be transmitted in frequency hopping mode within a time slot or in frequency hopping mode between time slots (i.e., according to the SRS period). Aperiodic SRS frequency hopping is performed within a time slot (i.e., all hopping is completed after one trigger).

[0171] Network devices can configure SRS resources for terminal devices via RRC signaling. The RRC signaling can indicate the number of ports included in the SRS resource, its frequency and time domain location, usage period, comb teeth, cyclic shift value, sequence ID, and other information. The frequency domain location of the SRS resource is determined by a set of frequency domain parameters in the RRC signaling (for example, in existing 3GPP protocols, the frequency domain parameters include n...). RRC n shift B SRS C SRS b hop For example, the terminal device determines the bandwidth occupied by the SRS (i.e., the configured bandwidth of the SRS resource), the starting position of the frequency domain, and the bandwidth occupied by the SRS in each symbol (or the bandwidth occupied by the frequency hopping subband) through these frequency domain parameters and the rules predetermined by the protocol.

[0172] 8. Autoencoder (AE) model: This generally refers to a network structure consisting of two AI models, such as an encoder and a decoder. Each model can be an independent AI model. AE models are also called bilateral models, two-end models, collaborative models, etc. The encoder and decoder of an AE are usually trained together and can be used in a coordinated manner.

[0173] Introducing AI technology into wireless communication networks has led to a CSI feedback method based on AI models. Terminal devices can use AI models to compress and feed back CSI, and network devices can use AI models to recover the compressed CSI. For example, AI model-based CSI feedback can be implemented using an AE model.

[0174] Figure 6 This is a schematic diagram illustrating CSI feedback using an AE (Advanced Effects) model. For example... Figure 6 As shown, the terminal side compresses the target CSI using an encoder, while the network side reconstructs the CSI using a decoder. For example, the terminal side can use the target CSI (i.e., V) as input to the encoder, which compresses the target CSI to obtain the compressed CSI (i.e., C). The terminal side can quantize the compressed CSI to obtain CSI feedback information. The terminal side then sends this CSI feedback information to the network side, for example, through a CSI report.

[0175] The network side can first dequantize the CSI feedback information to obtain compressed CSI with quantization loss (i.e., The network side uses the compressed CSI as input to the decoder, and the decoder decompresses the CSI based on the compressed CSI to obtain the reconstructed CSI (i.e., ).

[0176] The quantizer used to perform quantization can be predefined, such as protocol predefined, or it can be indicated by the network side, without limitation.

[0177] In another implementation, the encoder on the terminal side can also compress and quantize the target CSI, outputting CSI feedback information. The decoder on the network side can also dequantize and decompress the CSI feedback information to obtain the reconstructed CSI.

[0178] It should be understood that the encoder on the terminal side can be deployed inside the terminal device or in other devices outside the terminal device, such as the aforementioned OTT host or cloud server; the decoder on the network side can be deployed inside the network device or in other devices outside the network device, such as the aforementioned smart network element.

[0179] It should also be understood that although the figure shows an encoder and a decoder, this is only a model division from a functional perspective. The encoder can also be called the first model, and the decoder can also be called the second model. Furthermore, this application does not limit the number of models included in the AE model.

[0180] It should also be understood that the AE model is only one possible model for achieving the above functions and should not constitute any limitation on this application. The AE model can also be replaced by other AI models that can achieve the same or similar functions.

[0181] 9. Model Files and Model Parameters: Model files and / or model parameters can be used to define the model. Model files can indicate the model structure, which may include, but is not limited to, FNN, CNN, and RNN. The model file may have a fixed format, such as a standard predefined format, or a format pre-negotiated by both ends of the connection. Model parameters may refer to parameters in the neural network model, such as, but not limited to, the number of layers in the neural network, the type and weights of neurons in each layer, etc. This application does not limit the method of distributing the parameters of the reference model.

[0182] Take DNN as an example. The idea behind DNN comes from the neuronal structure of the brain. Each neuron can perform a weighted summation operation on its inputs and then use the result of the weighted summation operation to generate the output through a nonlinear function. Figure 7 This is an example of a neuron structure. Figure 7 The input to the neuron shown is x = [x0 x1 … x N-1 The weights corresponding to the inputs are w = [w0 w1 … w] N-1The bias of the weighted summation is b. The nonlinear function f() can take many forms; for example, the nonlinear function f() can be the maximum value function max{0, x}. Then the effect of a neuron's execution is... Where N is a positive integer; n is a positive integer greater than or equal to 0 and less than or equal to (N-1).

[0183] A typical DNN has multiple neural network layers, including an input layer, one or more hidden layers, and an output layer. Generally, the first layer is the input layer, the last layer is the output layer, and the layers in between are hidden layers. Each layer contains multiple neurons. Layers are fully connected; that is, any neuron in the i-th layer is connected to any neuron in the (i+1)-th layer. The input layer processes the received values ​​(i.e., the DNN's input) through neurons and then passes them to the hidden layers. Similarly, the hidden layers pass the computation results to the final output layer, producing the DNN's output. Figure 8 An example of a DNN is shown. Figure 8 The DNN model shown has three neural network layers: an input layer, a hidden layer, and an output layer.

[0184] It should be understood that the above text, in combination with... Figure 7 and Figure 8 The examples provided are for illustrative purposes only and should not be construed as limiting the scope of this application. This application does not limit the structure and parameters used in the AI ​​model.

[0185] One of the model structure or model parameters can be predefined, while the other can be sent by the sender (e.g., the network side). Alternatively, both the model structure and model parameters can be sent by the sender (e.g., the network side). This application does not impose any restrictions on this.

[0186] In this embodiment of the application, sending a model may refer to sending a model file and / or model parameters, and receiving a model may refer to receiving a model file and / or model parameters.

[0187] 10. Two-way connection: This refers to the connection between the sender (e.g., the network side) and the receiver (e.g., the terminal side). It can be based on a dataset or on a model. The following uses a CSI compressed inference task as an example to explain both dataset and model connections.

[0188] Among them, dataset-based docking mainly refers to the sender providing a dataset to the receiver for the receiver's model training.

[0189] Figure 9 This is a diagram illustrating the data set connection between the network side and the terminal side. For example... Figure 9As shown, the network can first acquire the dataset through joint training. For example, the network employs a CSI compression model (e.g., an encoder) to compress the target CSIs in the dataset, and a CSI reconstruction model (e.g., a decoder) to decompress the target CSIs. That is, the input to the CSI compression model can be the target CSIs, and the output can be the compressed CSIs; the input to the CSI reconstruction model can be the compressed CSIs, and the output can be the reconstructed CSIs. Thus, the dataset obtained by the network includes both the target CSIs and the compressed CSIs.

[0190] Optionally, the network side further quantizes the compressed CSI to obtain CSI feedback information. Accordingly, the dataset may also include: the target CSI and the CSI feedback information. The dataset may also include a quantizer. The quantizer can also be replaced by a quantization method. The quantizer can be included in the CSI compression model; in other words, the CSI compression model can also have the function of quantizing the compressed CSI.

[0191] Optionally, if the compressed CSI has been quantized to obtain CSI feedback information, the network side can dequantize the CSI feedback information before inputting it into the CSI reconstruction model to obtain the input for the CSI reconstruction model. The dequantizer used for dequantizing the CSI feedback information can be predefined, such as protocol-predefined, or it can be issued by the sender. This dequantizer can be included in the CSI reconstruction model; in other words, the CSI reconstruction model can also have the function of dequantizing the CSI feedback information.

[0192] Furthermore, the network side can distribute this dataset to the terminal side. The terminal side then trains a model based on the received dataset to obtain the terminal's CSI compressed model.

[0193] Model-based interoperability primarily refers to the sender providing the receiver with model files and / or model parameters for the receiver's model deployment, or for the receiver's model development (e.g., model training) and deployment. In this application, the sender can provide the receiver with model files and / or model parameters for the receiver to train the model.

[0194] Figure 10 This is a schematic diagram illustrating the model interfacing between the network side and the terminal side. For example... Figure 10 As shown, the network side can first obtain the CSI compressed model through joint training. The specific process of joint training on the network side can be found in the above text. Figure 9The explanation will not be repeated here. The network side can distribute the CSI compressed model or CSI reconstructed model obtained from joint training to the terminal side. For example, the network side can send the model file and / or model parameters of the CSI compressed model to the terminal side, or send the model file and / or model parameters of the CSI reconstructed model to the terminal side. The terminal side can then train the model based on the received model file and / or model parameters.

[0195] It should be understood that the above text, in combination with... Figure 9 and Figure 10 The CSI compression model shown can be an encoder, and the CSI reconstruction model can be a decoder.

[0196] As mentioned earlier, to support the integration and development / deployment of dual-end models, datasets or models need to be provided for one or both sides of the dual-end model. For example, the network side can provide a model to the terminal side, such as the encoder in the dual-end model. Since the dataset used by the network side to train the dual-end model is not fixed and can be of various types, with different types containing different data, how the terminal side constructs the dataset for model training or updates is a crucial consideration to improve the performance of the integration between the network and terminal models.

[0197] In view of this, this application provides a communication method in which the terminal side can determine that the dataset used for model training includes one of multiple types of datasets, which can improve the performance of model docking and thus improve the performance of CSI feedback.

[0198] To facilitate understanding, before introducing the methods provided in this application, the resources involved in the following embodiments will be described first.

[0199] The resources involved in this application may include one or more of the following: time-domain resources, frequency-domain resources, or spatial-domain resources.

[0200] Time-domain resources may include one or more time-domain units. A time-domain unit may be any of the following: one or more slots, one or more subframes, one or more frames, one or more orthogonal frequency division multiplexing (OFDM) symbols, 1 second (s) or several seconds (s), 1 millisecond (ms) or several milliseconds (ms), etc.

[0201] The term "same or different time-domain resources" as used in this application can refer to whether the time-domain units included in the time-domain resources are the same or different. Taking time-domain resources A and B as examples, if time-domain resources A and B are different, then the time-domain units included in time-domain resources A are different from those included in time-domain resources B. For example, time-domain resources A includes time slot #a, and time-domain resources A also includes time slot #b. If time-domain resources A and B are the same, then time-domain resources A and B include the same time-domain units. For example, both time-domain resources A and B include time slot #a. If time-domain resources A include time-domain resources B, then time-domain resources A includes at least the time-domain units belonging to time-domain resources B. For example, if time-domain resources B includes time slot #a, then time-domain resources A includes at least time slot #a.

[0202] Frequency domain resources may include one or more frequency domain units. A frequency domain unit may be any of the following: subcarrier, component carrier (CC), resource block (RB), subchannel, resource pool, bandwidth, subband, bandwidthpart (BWP), channel, physical resource block (PRB), resource block group (RBG), PRB bundling, or an interlaced RB, etc.

[0203] The term "same or different frequency domain resources" as used in this application can refer to whether the frequency domain units included in the frequency domain resources are the same or different. "One frequency domain resource includes another frequency domain resource" can mean that the one frequency domain resource includes at least the frequency domain units belonging to the other frequency domain resource. The other frequency domain resource can also be referred to as a portion of the frequency domain resources within the one frequency domain resource. Taking frequency domain resources A and B as examples, if frequency domain resources A and B are different, then the frequency domain units included in frequency domain resources A are different from those included in frequency domain resources B. For example, frequency domain resources A includes subband #a, and frequency domain resources B includes subband #b. If frequency domain resources A and B are the same, then frequency domain resources A and B include the same frequency domain units. For example, both frequency domain resources A and B include subband #a. If frequency domain resource A includes frequency domain resource B, then frequency domain resource A must include at least the frequency domain elements belonging to frequency domain resource B. In other words, frequency domain resource B is a subset of frequency domain resources in frequency domain resource A. For example, if frequency domain resource B includes subband #a, and frequency domain resource A includes frequency domain resource B, then frequency domain resource A must include at least subband #a. Frequency domain resource A including frequency domain resource B can also be referred to as resource B being a subset of resource A.

[0204] A spatial resource is associated with one or more of the following: a sending port, a receiving port, a sending stream, or a receiving stream. For example, if spatial resource A is different from spatial resource B, then spatial resource A differs from spatial resource B in at least one of the following: a sending port, a receiving port, a sending stream, or a receiving stream. For example, if spatial resource A includes spatial resource B, then the sending port / receiving port corresponding to spatial resource A includes the sending port / receiving port corresponding to spatial resource B, and / or, the sending stream / receiving stream corresponding to spatial resource A includes the sending stream / receiving stream corresponding to spatial resource B. If spatial resource A includes spatial resource B, then each of the above-mentioned items corresponding to spatial resource A includes at least the corresponding item corresponding to spatial resource B. For example, if spatial resource B corresponds to sending port A and sending stream A, then the sending port corresponding to spatial resource A includes at least sending port A, and the sending stream corresponding to spatial resource A includes at least sending stream A.

[0205] If a resource includes multiple resources in the time domain, frequency domain, or spatial domain, then if resource M is different from resource N, it means that resource M and resource N differ in at least one of the aforementioned multiple resources; if resource M and resource N are the same, it means that resource M and resource N are identical in every single one of the aforementioned multiple resources; if resource M includes resource N, it means that every single one of the aforementioned multiple resources included in resource M includes at least the corresponding single one included in resource N. For example, resource M includes time domain resource M and frequency domain resource M, and resource N includes time domain resource N and frequency domain resource N. If resource M is different from resource N, it means that time domain resource M and time domain resource N are different, and / or, frequency domain resource M and frequency domain resource N are different. If resource M and resource N are the same, it means that time domain resource M and time domain resource N are the same, and frequency domain resource M and frequency domain resource N are the same. If resource M includes resource N, it means that time domain resource M includes at least time domain resource N, and frequency domain resource M includes at least frequency domain resource N.

[0206] In this application, a resource may also be referred to as a resource set, which may include at least one of the above-described time-domain resources, and / or at least one of the above-described frequency-domain resources, and / or at least one of the above-described spatial-domain resources.

[0207] It should also be noted that in the embodiments below, "model" can be replaced by names such as encoder, decoder, compressor, decompressor, quantizer, or dequantizer, and this application does not limit this. For example, if the inference task corresponding to the model (such as the first model) is to compress data, then the model can be replaced by encoder, compressor, or quantizer, etc. As another example, if the inference task corresponding to the model (such as the second model) is to decompress data, then the model can be replaced by decoder, decompressor, or dequantizer, etc.

[0208] It should also be noted that in the embodiments below, "model" can be replaced by names such as encoder, decoder, compressor, decompressor, quantizer, or dequantizer, and this application does not limit this. For example, if the inference task corresponding to the model (such as the first model) is to compress data, then the model can be replaced by encoder, compressor, or quantizer, etc. As another example, if the inference task corresponding to the model (such as the second model) is to decompress data, then the model can be replaced by decoder, decompressor, or dequantizer, etc.

[0209] Figure 11 This is a schematic flowchart of a communication method 1100 provided in an embodiment of this application. The method may include the following steps.

[0210] S1110, The terminal side determines the type of data included in the dataset on the terminal side.

[0211] The dataset on the terminal side (denoted as the first dataset) can be used for model training or updating on the terminal side.

[0212] In this application, the device that performs model training on the terminal side can be referred to as a training device. The training device can be a terminal device or a device other than a terminal device, such as the aforementioned OTT system server, and there is no limitation in this regard.

[0213] For example, the data included in the first dataset is of type (denoted as type #1) belonging to either type 1 or type 2.

[0214] The first type of data includes the labels corresponding to the input of encoder 1 and the output of decoder 1 in dual-end model 1 (an example of the first dual-end model); or, the first type of data includes two parts: the input of encoder 1 and the labels corresponding to the output of decoder 1; or, when the data in the first dataset belongs to the first type, a sample in the first dataset may include the input of encoder 1 (denoted as the first input) and the labels corresponding to the output of decoder 1. The second type of data includes the input of encoder 1, the input of decoder 1, and the labels corresponding to the output of decoder 1; or, the first type of data includes three parts: the first part is the input of encoder 1, the second part is the input of decoder 1, and the third part is the labels corresponding to the output of decoder 1; or, when the data in the first dataset belongs to the second type, a sample in the first dataset may include the input of encoder 1, the input of decoder 1, and the labels corresponding to the output of decoder 1. Dual-end model 1 can be understood as a dual-end model on the terminal side.

[0215] Optionally, the first and second types of data also include a second input to decoder 1, which can be the output of encoder 1, for example, the output of encoder 1 is CSI feedback information. This CSI feedback information is obtained by compressing (or compressing and quantizing) the first CSI. Unless otherwise specified, the input to decoder 1 below refers to this first input.

[0216] For example, if the data in the first dataset belongs to the first type, the input of the encoder 1 is the first CSI, and the label corresponding to the output of the decoder 1 is the first CSI or the second CSI.

[0217] The first CSI can characterize the channel state of the first resource set, or in other words, the first CSI characterizes the channel information of the first channel, which corresponds to the first resource set (e.g., the channel information of the first channel is obtained by measuring the reference signal transmitted on the first resource set), and the first channel can be a channel between the terminal side and the network side.

[0218] The second CSI can characterize the channel state of the second resource set, or in other words, the second CSI characterizes the channel information of the second channel, which corresponds to the second resource set (e.g., the channel information of the second channel is obtained by measuring the reference signal transmitted on the second resource set). The first resource set can be a part of the second resource set, or the second resource set can be different from the first resource set.

[0219] For example, the second resource set includes at least one of frequency domain resources or spatial domain resources. It can be understood that the type of resources included in the first resource set (i.e., frequency domain resources or spatial domain resources) is the same as the type of resources included in the second resource set, that is, the first resource set also includes at least one of frequency domain resources or spatial domain resources.

[0220] Taking the resources included in the resource set as frequency domain resources as an example, the first resource set being a part of the second resource set can be understood as the frequency domain resources (or frequency domain units) included in the first resource set being a subset of the frequency domain resources (or frequency domain units) included in the first resource set. The second resource set being different from the first resource set can be understood as the frequency domain resources (or frequency domain units) included in the first resource set not overlapping or partially overlapping with the frequency domain resources (or frequency domain units) included in the first resource set. It should be understood that when the first resource set and the second resource set are frequency domain resources and / or spatial domain resources, the relationship between the resources in the first resource set and the second resource set is not limited in describing the relationship between the time domain resources corresponding to the first resource set and the time domain resources corresponding to the second resource set.

[0221] Optionally, the first resource set and the second resource set correspond to the same time-domain resource. For example, the second resource set is the full-band corresponding to the first time-domain resource, and the first resource set can be a partial sub-band of the full-band. That is, the second CSI can be the full-band CSI on the first time-domain resource, and the first CSI can be the CSI corresponding to a partial sub-band of the full-band on the first time-domain resource.

[0222] If the data in the first dataset belongs to the second type, the input of encoder 1 is the first CSI, the input of decoder 1 includes the third CSI, and the output of decoder 1 corresponds to the second CSI.

[0223] The first and second CSIs can be referred to in the description above. The third CSI can be used to characterize the channel state of the third resource set, or in other words, the third CSI characterizes the channel information of the third channel, which corresponds to the third resource set (e.g., the channel information of the third channel is obtained by measuring the reference signal transmitted on the third resource set). The third resource set may include at least one of frequency domain resources or spatial domain resources.

[0224] For example, the third resource set is different from the first resource set, and the third resource set is part or all of the second resource set.

[0225] Taking the resources included in the resource set as frequency domain resources as an example, the difference between the third resource set and the first resource set can be understood as the third resource set partially overlapping or not overlapping with the first resource set. For example, the first resource set may include a first part of the sub-band in the full band, and the third resource set may include a second part of the sub-band in the full band. The first part of the sub-band and the second part of the sub-band do not overlap, or in other words, the first part of the sub-band and the second part of the sub-band include different sub-bands; or, the first part of the sub-band and the second part of the sub-band include some of the same sub-bands, while other parts are different.

[0226] The statement that a third resource set comprises all of the second resource set can be understood as meaning that the frequency domain resources (or frequency domain units) included in the third resource set are the same as those included in the second resource set. The statement that a third resource set comprises only a portion of the second resource set can be understood as meaning that the frequency domain resources included in the third resource set are a subset of those included in the second resource set. It should be understood that when a resource set comprises both frequency domain resources and / or spatial domain resources, the time domain resources corresponding to that resource set are not limited when describing the relationships between resource sets.

[0227] Optionally, the second resource set corresponds to the first time-domain resource, and the third resource set corresponds to the second time-domain resource and / or the first time-domain resource, wherein the second time-domain resource is earlier than the first time-domain resource. In other words, the time-domain resource corresponding to the third resource set is earlier than the time-domain resource corresponding to the second resource set, or the third resource set corresponds to the same time-domain resource as the second resource set, or the third resource set includes frequency-domain resources and / or spatial-domain resources corresponding to two parts of time-domain resources, wherein one part of the time-domain resources is the same as the time-domain resource corresponding to the second resource set, and the other part of the time-domain resources is earlier than the time-domain resource corresponding to the second resource set.

[0228] For example, when the third resource set is a portion of the second resource set, the time-domain resources corresponding to the third resource set can be the same as those corresponding to the second resource set. If the second resource set is the full-band corresponding to the first time-domain resource, the third resource set can be a sub-band within that full-band; that is, the second CSI can be the full-band CSI on the first time-domain resource, and the third CSI can be the CSI corresponding to a sub-band within the full-band of the first time-domain resource. The third CSI can also be channel information obtained by measuring the uplink reference signal (such as SRS); that is, the third CSI can also be SRS measurements, such as those obtained by measuring the SRS on that sub-band.

[0229] For example, when the third resource set is a part of the second resource set, the time-domain resources corresponding to the third resource set can be earlier than the time-domain resources corresponding to the second resource set. If the second resource set is the full band corresponding to the first time-domain resource, the second resource set can be a partial sub-band within that full band; that is, the second CSI can be the full-band CSI on the first time-domain resource, and the third CSI can be the CSI corresponding to that partial sub-band on the second time-domain resource. For instance, the third CSI can also be the SRS measurement result corresponding to that partial sub-band. This partial sub-band can include one or more sub-bands. When the partial sub-band includes multiple sub-bands, the time-domain resources corresponding to each of these multiple sub-bands can be different; that is, the second time-domain resource can include the time-domain resources corresponding to each sub-band. For example, each sub-band's corresponding time-domain resource includes a time-domain unit, and the time-domain position of each sub-band's corresponding time-domain unit can be different. The time-domain positions of the multiple time-domain resources corresponding to these multiple sub-bands can be the time-domain positions of frequency-hopping SRS transmissions.

[0230] For example, when the third resource set is all of the second resource set, the third resource set may include the first part of the resource set and the second part of the resource set. The first part of the resource set corresponds to the first time-domain resource, and the second part of the resource set corresponds to the second time-domain resource. The first part of the resource set and the second part of the resource set do not overlap, or only partially overlap. If the second resource set is the full band corresponding to the first time-domain resource, the third resource set may be that full band. The first part of the resource set in the third resource set may be a first sub-band of the full band, and the time-domain resource corresponding to the first sub-band is the first time-domain resource. The second part of the resource set may be a second sub-band of the full band. The second sub-band may be any other sub-band besides the first sub-band, or the second sub-band may partially overlap with the first sub-band, and the time-domain resource corresponding to the second sub-band is the second time-domain resource. Or, the second CSI may be the full-band CSI on the first time-domain resource, and the third CSI includes the CSI corresponding to the first sub-band of the full band on the first time-domain resource, and the CSI corresponding to the second sub-band of the second time-domain resource. The third CSI can also be the SRS measurement corresponding to the full band. This SRS measurement can characterize the channel information of multiple sub-bands. A portion of these multiple sub-bands (i.e., the first portion of the sub-bands) corresponds to the first time-domain resource, and other portions of these sub-bands correspond to the second time-domain resource. The frequency domain location of these multiple sub-bands can be referenced to the frequency domain resource of frequency-hopping SRS. The first and second time-domain resources can each include one or more time-domain units, and the time-domain location of these one or more time-domain units can be referenced to the time-domain location of frequency-hopping SRS.

[0231] For example, when the third resource set comprises all of the second resource set, the time-domain resources corresponding to the third resource set are earlier than those corresponding to the second resource set. If the second resource set is the full band corresponding to the first time-domain resource, the third resource set can be that full band, and this third resource set consists of multiple sub-bands, each of which corresponds to a time-domain resource earlier than the time-domain resource corresponding to the second resource set. The multiple time-domain units corresponding to these multiple sub-bands can be located in the time domain of the frequency-hopping SRS transmission.

[0232] In this application, the two resource sets corresponding to the same time-domain resources can be understood as the two resource sets corresponding to the same time-domain unit, or the corresponding time-domain units being close to each other. For example, if the third resource set corresponds to the same time-domain resource as the second resource set, and the second resource set is the resource set corresponding to CSI-RS, and the third resource set is the resource set corresponding to SRS, CSI-RS and SRS may not be transmitted in the same time slot, but when their time-domain intervals are close, the corresponding time-domain resources can be considered to be the same. For example, if the second resource set corresponds to the first time-domain unit, the third resource set that is closest to the first time-domain unit in the time domain can be considered to have the same time-domain resource as the second resource set. The closest time-domain unit can be located before, after, or both before and after the first time-domain unit, without limitation.

[0233] In this application, the time-domain resource may include one or more time-domain units. For example, the time-domain unit may be a time slot or a frame. The time-domain unit may also be other granularities, as described above.

[0234] For example, the execution device used on the terminal side to determine the type of data included in the dataset on the terminal side can be a terminal device or an OTT system server, without limitation. Furthermore, the device used on the terminal side to perform model training (i.e., the training device) can be the same as or different from the execution device, without limitation.

[0235] In one possible implementation, the terminal receives indication information from the network side (S1110a) and determines, based on the indication information, whether the data included in the first dataset is of a first type or a second type. Alternatively, the indication information is used to determine whether the data included in the first dataset is of the first type or the second type.

[0236] As an example, the indication information indicates N types, which are either the N types of data included in the first dataset, or the N types of data included in the network-side dataset (denoted as the second dataset). Alternatively, the indication information indicates that the data in the first dataset can be constructed according to one of the N types, or that the data in the second dataset is constructed according to one of the N types, where N is a positive integer. The N types include both the first and second types.

[0237] For example, the instruction information can directly indicate the N types, or in other words, the instruction information includes N types.

[0238] For example, the indication information may include k bits, and different values ​​of these k bits can indicate N types, where k is a positive integer, and the value of k is related to the value of N. For example, the value of k can be 2, and these 2 bits can take different values ​​("00", "01", "10", and "11") to represent one of the N types respectively. The meaning represented by the different values ​​of these k bits can be predetermined or configured by the network side.

[0239] As another example, the indication information is used to determine M types, or in other words, the indication information indicates M types. These M types are either M types of data included in the first dataset, or M types of data included in the second dataset. Alternatively, the indication information indicates that the data in the first dataset can be constructed according to one of the M types, or that the indication information indicates that the data in the second dataset is constructed according to one of the M types. Each of the M types belongs to either the first type or the second type, where M is a positive integer.

[0240] For example, the M types include at least one of the following types:

[0241] Type 1: The input of encoder 1 is the first CSI, and the output of decoder 1 corresponds to the tag of the first CSI;

[0242] Type 2: The input of encoder 1 is the first CSI, and the corresponding tag of the output of decoder 1 is the second CSI;

[0243] Type 3: The input of encoder 1 is the first CSI, the input of decoder 1 includes the third CSI, and the output of decoder 1 corresponds to the second CSI.

[0244] Optionally, corresponding to different relationships between the frequency domain positions and / or time domain positions of the resource sets, type 3 above may further include multiple types. For example, type 3 may further include multiple types from type a to type c: type a: the second CSI is a full-band CSI on the first time domain resource, the third CSI is a CSI corresponding to a portion of the sub-bands in the full-band of the first time domain resource, and the first CSI is a CSI corresponding to other portions of the sub-bands in the full-band of the first time domain resource; type b: the second CSI is a full-band CSI on the first time domain resource, the third CSI is a CSI corresponding to a portion of the sub-bands in the full-band of the second time domain resource, and the first CSI is a CSI corresponding to other portions of the sub-bands in the full-band of the first time domain resource; type c: the second CSI is a full-band CSI on the first time domain resource, the third CSI includes the CSI corresponding to the first portion of the sub-bands in the full-band of the first time domain resource, and the CSI corresponding to the second portion of the sub-bands in the second time domain resource, and the first CSI is a CSI corresponding to the third portion of the sub-bands in the first time domain resource.

[0245] It should be understood that the use of the resource set including sub-bands to illustrate the relationship of the resource set in this application is merely an example and does not constitute a limitation on this application. The resources included in the resource set can also be of other granularities, such as RB, as detailed in the description of resources above.

[0246] For example, the way a network device indicates M types through indication information is similar to the way it indicates N types through indication information. For details, please refer to the description above, which will not be repeated here.

[0247] When the training devices used to perform model training on the terminal side are different devices, the way this instruction information is received can be different.

[0248] As an example, the device that performs model training on the terminal side is a terminal device. The terminal device receives the instruction information from the network device, or the terminal device receives the instruction information from the OTT server, which may be received by the OTT system server from the intelligent network element.

[0249] As another example, if the device performing model training on the terminal side is an OTT system server, then the OTT system server can receive the instruction information from the intelligent network element, or it can receive the instruction information from the terminal device, which may be received by the terminal device from the network device.

[0250] Optionally, the value of M or N can be 1, meaning the network side can indicate a type of data included in the first or second dataset. In other words, the network side indicates to the terminal side a type of data in the first dataset used by the terminal side. This type could be the type of data included in the first dataset that the network side expects the terminal side to use, or the type of data included in the first dataset that the network side determines is applicable to the terminal side. Alternatively, the network side can indicate to the terminal side a type of data included in the dataset used for training the dual-end model (denoted as dual-end model 2).

[0251] Optionally, before the network side indicates the N or M types, the network side receives the terminal's capability information and determines the N or M types based on the capability information. This capability information may indicate the terminal's computing capabilities and / or the types of data it supports.

[0252] For example, the computing power on the terminal side can specifically refer to the computing power of the device (hereinafter referred to as the training device) that performs model training on the terminal side. This computing power can be characterized by parameters such as the device's computing speed and processing time per unit of data, which are not limited in this application. The types of data supported by the terminal side can specifically refer to the types of training data supported by the terminal side training device. For example, the types of training data supported by the training device can include one or more of the above N or M types.

[0253] In one example, the training device is a terminal device. Sending capability information from the terminal to the network side can include: the terminal device sending capability information to the network device, or the terminal device sending capability information to the OTT system server, so that the OTT system server can send the capability information to the intelligent network element.

[0254] In another example, the training device is an OTT system server. Sending capability information from the terminal side to the network side can include: the OTT system server sending the capability information to the intelligent network element, or the OTT system sending the capability information to the terminal device, so that the terminal device can send the capability information to the network device.

[0255] The network side can determine the types of data included in the dataset that are compatible with the capabilities of the terminal side based on the capability information of the terminal side. This can avoid the situation where the types of data included in the dataset used for model training are incompatible with the capabilities of the terminal side, resulting in the model training on the terminal side failing to meet performance requirements.

[0256] Optionally, when determining the type of data included in the first dataset based on the indication information, the terminal can also determine it based on its own capability information. For details on how to determine the type of data included in a dataset based on capability information, refer to the method used by the network side, which will not be elaborated further.

[0257] S1120, the terminal side trains or updates encoder 1 based on the first dataset of type #1.

[0258] Alternatively, the terminal side trains or updates encoder 1 based on the first dataset, which includes data of type #1.

[0259] The encoder 1 is used to perform a first processing on the CSI to be fed back. The CSI to be fed back can be the currently measured CSI on the terminal side or the predicted CSI; the first processing includes at least CSI compression, for example, the first processing is compression, or the first processing includes compression and quantization, or the first processing may also include prediction.

[0260] Optionally, before training or updating encoder 1 based on the first dataset of type #1, the terminal side obtains encoder 2 from the network side. Encoder 2 is obtained by training the dual-end model 2 (an example of the second dual-end model) based on the training dataset of the network side (i.e., the second dataset). The data type included in the second dataset is either the first type or the second type. More specifically, the data type included in the second dataset can be one of the N types or M types.

[0261] Taking the second dataset as an example where the data type is the second type, the process of training the dual-end model 2 on the network side is as follows: Figure 12 As shown. The network side will use the first CSI (which can be represented as H1) as an example. NW The first input CSI is used as the input to encoder 2. Encoder 2 can compress the first input CSI to obtain first CSI feedback information (which can be represented as C2). This first CSI feedback information can be used as the input to decoder 2. The input to decoder 2 also includes SRS measurement values ​​(which can be represented as H2). NW (An example of the third CSI). Decoder 2 decompresses the first CSI feedback information based on the SRS measurement to obtain the reconstructed CSI (which can be represented as H3). NW Furthermore, the network side can target the CSI (which can be represented by H4). NW Using an example of the second CSI as a label, encoder 2 and decoder 2 are trained. By training encoder 2 and decoder 2, the reconstructed CSI output by decoder 1 becomes closer to the target CSI.

[0262] Optionally, encoder 2 can compress and quantize the first CSI to obtain first CSI feedback information. Correspondingly, decoder 2 can decompress and dequantize the first CSI feedback information to obtain reconstructed CSI.

[0263] Optionally, encoder 2 can compress the first CSI to obtain compressed CSI, and the compressed CSI is quantized to obtain the first CSI feedback information. Correspondingly, decoder 2 can decompress the dequantized information of the first CSI feedback information to obtain the reconstructed CSI. The dequantized information of the first CSI feedback information corresponds to the compressed CSI.

[0264] After training is completed, the network side sends encoder 2 to the terminal side, and the terminal side receives encoder 2 accordingly.

[0265] In one example, the training device on the terminal side is a terminal device. The terminal side receiving the encoder 2 can specifically include: the terminal device receiving the model from the network device, such as receiving it through the air interface; or, the terminal device obtaining the model from the OTT system server, which is received by the OTT system from the intelligent network element on the network side, such as receiving it through the wired network.

[0266] In another example, the training device on the terminal side is an OTT system server. The terminal side receiving the encoder 2 may specifically include: the OTT system server receiving the model from a smart network element, such as through a wired network; or the OTT system server obtaining the model from the terminal device, which is received by the terminal device from a network device, such as through an air interface.

[0267] Sending the model can refer to sending the model file and / or model parameters via a message carrier. The model file can indicate the model structure (e.g., encoder 2), such as the aforementioned CNN or RNN structure; the model parameters can indicate parameters such as the weights of each neuron in the model. Based on the model file and / or model parameters, the model can be determined. The network side can send the model to the terminal side through one or more messages, and this application does not limit this.

[0268] In one possible implementation, after receiving encoder 2, the terminal side trains or updates encoder 1 based on the first dataset of type #1 and encoder 2.

[0269] Taking type #1 as the second type as an example, Figure 13 A schematic diagram illustrating the process of model training on the terminal side. (Example) Figure 13 As shown, the process of model training on the terminal side can include two stages, namely stage 1 and stage 2.

[0270] Phase 1: The terminal can fix the parameters of decoder 2 (i.e., the model parameters received from the network side) (i.e., use a frozen decoder 2; in other words, do not change the parameters of decoder 2 during Phase 1 training). Encoder 2 received by the terminal can also be understood as encoder 1 before model training on the terminal side, and encoder 2 can also be understood as the encoder 1 on the terminal side to be trained (or updated). The terminal side will use the first CSI (i.e., H1) UE Using the third CSI as the input to encoder 2, and the third CSI as the input to decoder 1, end-to-end training is performed on encoder 2 and decoder 1, so that the output of decoder 1 (i.e., ) and tags (i.e., H4) UE They tend to approach each other. At the output of decoder 1 with H4 UE When the performance indicators are met, the training of the encoder 1 model on the terminal side can be stopped, thereby obtaining the trained decoder 1.

[0271] Phase 2: The terminal side can fix the parameters of decoder 1 (i.e., the model parameters of decoder 1 obtained through phase 1 training) (i.e., use the fixed decoder 1; in other words, do not change the parameters of decoder 1 during the training process of phase 2), and set H1... UE '(An example of the first CSI, which can be compared with H1)' UE (Same or different) as input to encoder 1, H2 UE '(An example of the third CSI, which can be compared with H2)' UE (Same or different) is used as input to decoder 1 to train encoder 1 and decoder 1 end-to-end, so that the output of decoder 1 (i.e., ) and tags (i.e., H4) UE The outputs of decoder 1 tend to approach each other. with H4 UE When the performance indicators are met, the terminal can stop training the encoder 1 model, thereby obtaining the trained encoder 1.

[0272] For example, the end-to-end training process is as follows: A first CSI is input to encoder 1, which can be used to compress the first CSI and output CSI feedback information; the output of encoder 1 is used as the input to decoder 1, which can be used to decompress the CSI feedback information output by encoder 1 and output the reconstructed CSI. End-to-end training of encoder 1 and decoder 1 can also be referred to as joint training of encoder 1 and decoder 1.

[0273] It should be understood that if encoder 1 is used to compress and quantize the input CSI, the CSI feedback information output by encoder 1 is quantized CSI. In this case, decoder 1 can be used to dequantize and decompress the CSI feedback information.

[0274] During the training phase described above, the labels corresponding to the inputs of encoder 1, decoder 1, and output of decoder 1 used on the terminal side constitute the data in the training dataset (i.e., the first dataset) on the terminal side. The data in the first dataset can be CSI data acquired by the terminal side itself. The methods by which the terminal side acquires the data in the first dataset include, but are not limited to, measurement, simulation, training generation, etc.

[0275] It should be understood that Figure 13The first dataset includes the input of encoder 1, the input of decoder 1, and the label corresponding to the output of decoder 1. This is only an example of the type of data included in the first dataset and does not constitute a limitation on this application. The type of data included in the first dataset can also be other types, such as the first type.

[0276] Based on the above technical solution, the terminal side can determine the type of dataset (type #1) used for training the dual-end model, and train the dual-end model on the terminal side based on the dataset of type #1. Type #1 is either a first type or a second type, and the network side can also train the dual-end model on the network side based on the dataset of either the first type or the second type. Thus, the terminal side selects one type from the first type or the second type to construct the dataset for model training, which is beneficial to improving the decoding performance of the decoder in the dual-end model, thereby improving the feedback performance of CSI.

[0277] It should be understood that the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0278] It should also be understood that, in the various embodiments of this application, unless otherwise specified or in case of logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced by each other, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.

[0279] It should also be understood that in the above embodiments, the examples are mainly based on devices in existing network architectures, and it should be understood that the specific form of the device is not limited in the embodiments of this application. For example, any device that can achieve the same function in the future is applicable to the embodiments of this application.

[0280] It is understood that, in the above-described method embodiments, the methods and operations implemented by a device (such as a terminal device or a network device) can also be implemented by components of the device (such as chips or circuits).

[0281] Those skilled in the art will recognize that, based on the units and algorithm steps described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is implemented in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0282] The methods provided in the embodiments of this application have been described in detail above with reference to several accompanying drawings. The apparatus provided in the embodiments of this application will now be described with reference to the accompanying drawings.

[0283] Figure 14 and Figure 15 This is a schematic block diagram of possible apparatuses provided in the embodiments of this application. These apparatuses can be used to implement the functions on the terminal side or network side in the above method embodiments, and thus can also achieve the beneficial effects of the above method embodiments.

[0284] Figure 14 A schematic block diagram of the apparatus provided in the embodiments of this application. Figure 14 The device 1400 shown may include a processing module 1410 and a communication module 1420.

[0285] In one possible design, device 1400 can be used to implement Figure 11 The communication method shown in the embodiment is implemented by the terminal side. For example, the processing module 1410 is used to implement the processing-related steps performed by the terminal device in each method embodiment, such as determining the type of data included in the dataset; the communication module 1420 is used to implement the sending and / or receiving steps performed by the terminal device in each method embodiment.

[0286] For example, the processing module 1410 is configured to: determine the type of data included in the dataset on the terminal side, wherein the type belongs to a first type or a second type, the first type of data includes the input of the encoder in the first dual-end model and the label corresponding to the output of the decoder in the first dual-end model, the second type of data includes the input of the encoder, the input of the decoder and the label corresponding to the output of the decoder, and the first dual-end model is a dual-end model on the terminal side; the processing module 1410 is further configured to: train or update the encoder based on the dataset on the terminal side of the type; wherein the encoder is configured to perform a first processing on the channel state information (CSI), the first processing including at least CSI compression, and the encoder is configured to perform a second processing on the compressed CSI, the second processing including at least CSI reconstruction.

[0287] The first and second types of datasets can be referred to in the description above, and will not be repeated here.

[0288] The communication module 1420 can be used to: receive indication information, which is used to determine whether the type is the first type or the second type. The specific indication information can be found in the description above.

[0289] Optionally, the communication module 1420 is also used to: send capability information indicating the computing capabilities and / or types of data supported on the terminal side.

[0290] Optionally, the communication module 1420 is further configured to: acquire a network-side encoder, which is obtained by training a second dual-end model on a network-side dataset, wherein the second dual-end model is a network-side dual-end model, and the data included in the network-side dataset belongs to the first type or the second type.

[0291] For a more detailed description of the processing module 1410 and the communication module 1420, please refer to [the relevant documentation]. Figure 11 The relevant descriptions in the method embodiments shown are directly obtained and will not be repeated here.

[0292] In another possible design, device 1400 can be used to achieve Figure 11 The illustrated embodiments demonstrate communication methods implemented by network devices. For example, processing module 1410 is used to implement processing-related steps performed by the network device in each method embodiment; communication module 1420 is used to implement sending and / or receiving steps performed by the network device in each method embodiment.

[0293] For example, the communication module 1420 is configured to: send indication information, the indication information being used to determine whether the data type included in the terminal-side dataset is a first type or a second type, the terminal-side dataset being used to train or update the encoder in a first dual-end model, the first type of data including the input of the encoder and the label corresponding to the output of the decoder in the first dual-end model, the second type of data including the input of the encoder, the input of the decoder and the label corresponding to the output of the decoder, the first dual-end model being a dual-end model on the terminal side; wherein, the encoder is used to perform a first processing on channel state information (CSI), the first processing including at least CSI compression, the encoder is used to perform a second processing on the compressed CSI, the second processing including at least CSI reconstruction.

[0294] For a more detailed description of the processing module 1410 and the communication module 1420, please refer to [the relevant documentation]. Figure 11 The relevant descriptions in the method embodiments shown are directly obtained and will not be repeated here.

[0295] It should be noted that the communication module can also be called a transceiver module, transceiver unit, transceiver, transceiver device, or transceiver apparatus, etc. The processing module can also be called a processor, processing board, processing unit, or processing apparatus, etc. Optionally, the communication module is used to perform the sending and receiving operations of the terminal device or network device in the above method. The device in the communication module that implements the receiving function can be considered as the receiving module, and the device in the communication module that implements the sending function can be considered as the sending module; that is, the communication module can include both a receiving module and a sending module.

[0296] It should also be noted that, in one possible design, the aforementioned processing module and / or communication module can be implemented through virtual modules. For example, the processing module can be implemented through software functional units or virtual devices, and the communication module can be implemented through software functions or virtual devices. In another possible design, the processing module or communication module can also be implemented through physical devices. For example, if the device is implemented using a chip / chip circuit, the communication module can be an input / output circuit and / or a communication interface, performing input operations (corresponding to the aforementioned receiving operation) and output operations (corresponding to the aforementioned sending operation); the processing module can be an integrated processor, a microprocessor, or an integrated circuit.

[0297] The module division in this embodiment is illustrative and represents only one logical functional division; in actual implementation, other division methods may be used. Furthermore, the functional modules in the various examples of this embodiment can be integrated into a single processor, exist as separate physical entities, or be integrated into a single module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0298] Figure 15 This is a schematic diagram of the structure of a communication device provided in yet another embodiment of this application. Figure 15 As shown, the device 1500 includes a processing circuit 1510 and a communication circuit 1520. The processing circuit 1510 and the communication circuit 1520 are coupled to each other.

[0299] It can be understood that the processing circuit 1510 can be one or more processors, or it can be all or part of the circuits with processing functions in one or more processors.

[0300] Understandably, the communication circuit 1520 can be a transceiver or an input / output interface.

[0301] Optionally, the device 2000 may further include a memory 1530 for storing instructions executed by the processing circuit 1510, or storing input data required for the running instructions of the processing circuit 1510, or storing data generated after the running instructions of the processing circuit 1510.

[0302] It is understood that the memory 1530 may be located outside the processing circuit 1510, or inside the processing circuit 1510.

[0303] As an example, the processing circuit 1510 is used to implement the functions of the processing module 1410, and the communication circuit 1520 is used to implement the functions of the communication module 1420.

[0304] As an example, device 1500 can be a communication device (such as a terminal device or network device) or a chip used in a communication device.

[0305] When device 1500 is a communication device, the communication circuit can be a transceiver; when device 1500 is a chip, the communication circuit can be an input / output circuit, a bus, pins, or other types of communication interfaces. The input circuit in the input / output circuit can be used for receiving, and the output interface can be used for transmitting.

[0306] In one possible implementation, the apparatus 1500 is used to implement the various processes and steps corresponding to the terminal device in the above method embodiments. In another possible implementation, the apparatus 1500 is used to implement the various processes and steps corresponding to the network device in the above method embodiments.

[0307] It is understood that the device 1500 can specifically be the terminal device or network device in the above embodiments, or it can be a chip or chip system. Correspondingly, the communication circuit can be the interface circuit of the chip, or an input / output circuit, which is not limited here. Specifically, the device 1500 can be used to execute the various steps and / or processes corresponding to the terminal device or network device in the above method embodiments.

[0308] When the aforementioned communication device is a chip or OTT device applied to a terminal device, the chip or OTT device of the terminal device implements the functions of the terminal device in the above method embodiments, for example, implementing the processing functions of the terminal device. The terminal device chip or OTT device receiving information from the network device can be understood as the information being first received by other modules (such as radio frequency modules or antennas) in the terminal device, and then sent by these modules to the terminal device chip or OTT device. The terminal device chip or OTT device sending information to the network device can be understood as the information being first sent by the terminal device chip or OTT device to other modules (such as radio frequency modules or antennas) in the network device, and then sent by these modules to the network device.

[0309] When the aforementioned communication device is a chip or OTT device applied to a network device, the chip or OTT device of the network device implements the functions of the network device in the above method embodiments, for example, implementing the processing functions of the network device. The chip or OTT device of the network device receiving information from the terminal device can be understood as the information being first received by other modules (such as radio frequency modules or antennas) in the network device, and then sent by these modules to the chip or OTT device. The chip or OTT device of the network device sending information to the terminal device can be understood as the information being first sent by the chip or OTT device to other modules (such as radio frequency modules or antennas) in the network device, and then sent by these modules to the terminal device.

[0310] This application also provides a computer program product that, when run on a processor, can implement the communication method executed by the terminal device or the communication method executed by the network device in the above method embodiments.

[0311] This application also provides a computer-readable storage medium containing computer instructions that, when executed on a processor, can implement the communication method executed by a terminal device or a network device in the above method embodiments.

[0312] This application also provides a communication system, including the aforementioned terminal equipment and network equipment.

[0313] It is understood that the processor in the embodiments of this application may be any of the following devices or all or part of the circuitry used for processing functions: a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSPs), processors for AI, field-programmable gate arrays (FPGAs), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. A general-purpose processor may be a microprocessor or any conventional processor.

[0314] For example, the processor used for AI can be one or more of the following: graphics processing unit (GPU), neural processing unit (NPU), tensor processing unit (TPU), and data processing unit (DPU).

[0315] For example, one possible implementation of a processor for AI could be... Figure 16 The AI ​​processor shown is 1600.

[0316] like Figure 16As shown, the AI ​​processor 1600 may include one or more of the following: AI core, digital vision pre-processing (DVPP) module, task scheduler (TS), L3 cache, AI CPU (central processing unit), control CPU (central processing unit), L2 cache, universal serial bus (USB) interface, network card, peripheral component interconnect express (PCIe) interface (PCIe is a high-speed serial computer expansion bus standard), double data rate (DDR) / high bandwidth memory (HBM) interface, generation purpose input / output (GPIO) / inter-integrated circuit (I2C) bus, etc. It is understood that the specific meanings of these terms are well known to those skilled in the art and will not be elaborated upon here.

[0317] The terms “unit”, “module”, etc., used in this specification may be used to refer to computer-related entities, hardware, firmware, combinations of hardware and software, software, or software in execution.

[0318] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0319] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0320] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0321] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0322] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0323] In the above embodiments, the functions of each functional unit can be implemented entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. This computer program product includes one or more computer instructions (programs). When the computer program instructions (programs) are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., digital video discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0324] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0325] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A communication method, characterized in that, Applied to the terminal side, including: The data type of the dataset on the terminal side is determined, and the type belongs to a first type or a second type. The data of the first type includes the labels corresponding to the input of the encoder and the output of the decoder in the first dual-end model. The data of the second type includes the input of the encoder, the input of the decoder and the labels corresponding to the output of the decoder. The first dual-end model is the dual-end model on the terminal side. The terminal-side dataset of this type is used to train or update the encoder; The encoder is used to perform a first process on the channel state information (CSI), the first process including at least CSI compression, and the encoder is used to perform a second process on the compressed CSI, the second process including at least CSI reconstruction.

2. The method according to claim 1, characterized in that, When the data type belongs to the first type, the encoder input is the first CSI, and the tag is either the first CSI or the second CSI; Wherein, the first CSI represents the channel state of the first resource set, the second CSI represents the channel state of the second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set.

3. The method according to claim 1, characterized in that, When the data type belongs to the second type, the encoder input is the first CSI, the decoder input includes the third CSI, and the tag is the second CSI; Wherein, the first CSI represents the channel state of the first resource set, the second CSI represents the channel state of the second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set, the third CSI represents the channel state of the third resource set, the third resource set is different from the first resource set, the third resource set is a part or all of the second resource set, and the second resource set includes at least one of frequency domain resources or spatial domain resources.

4. The method according to claim 3, characterized in that, The second resource set corresponds to the first time domain resource, the first resource set corresponds to the first time domain resource, and the third resource set corresponds to the second time domain resource and / or the first time domain resource, wherein the second time domain resource is earlier than the first time domain resource.

5. The method according to any one of claims 1 to 4, characterized in that, The method further includes: Receive indication information, the indication information being used to determine whether the type is the first type or the second type.

6. The method according to claim 5, characterized in that, The indication information indicates N types, which are N types of data included in the dataset on the terminal side, or N types of data included in the dataset on the network side. The N types include the first type or the second type, and N is a positive integer.

7. The method according to any one of claims 1 to 4, characterized in that, The indication information is used to determine M types, which are M types of data included in the dataset on the terminal side, or M types of data included in the dataset on the network side, one of which belongs to the first type or the second type, and M is a positive integer.

8. The method according to claim 7, characterized in that, The M types include at least one of the following: The encoder's input is the first CSI, and the tag is the first CSI; The encoder's input is the first CSI, and the tag is the second CSI; The encoder's input is a first CSI, the decoder's input includes a third CSI, and the tag is a second CSI; Wherein, the first CSI represents the channel state of the first resource set, the second CSI represents the channel state of the second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set, the third CSI represents the channel state of the third resource set, the third resource set is different from the first resource set, the third resource set is a part of the second resource set, and the second resource set includes at least one of frequency domain resources or spatial domain resources.

9. The method according to any one of claims 1 to 8, characterized in that, The method further includes: Send capability information, which indicates the computing capabilities and / or types of data supported on the terminal side.

10. The method according to any one of claims 1 to 9, characterized in that, Before training or updating the encoder using the terminal-side dataset of the type described above, the method further includes: Obtain the network-side encoder, which is obtained by training a second dual-end model based on the network-side dataset. The second dual-end model is the network-side dual-end model, and the data included in the network-side dataset belongs to either the first type or the second type.

11. A communication method, characterized in that, Applied to the network side, including: Sending instruction information, the instruction information is used to determine whether the data type included in the dataset on the terminal side is a first type or a second type, the dataset on the terminal side is used to train or update the encoder in the first dual-end model, the first type of data includes the input of the encoder and the label corresponding to the output of the decoder in the first dual-end model, the second type of data includes the input of the encoder, the input of the decoder and the label corresponding to the output of the decoder, and the first dual-end model is the dual-end model on the terminal side; The encoder is used to perform a first process on the channel state information (CSI), the first process including at least CSI compression, and the encoder is used to perform a second process on the compressed CSI, the second process including at least CSI reconstruction.

12. The method according to claim 11, characterized in that, When the data type belongs to the first type, the encoder input is the first CSI, and the tag is either the first CSI or the second CSI; Wherein, the first CSI represents the channel state of the first resource set, the second CSI represents the channel state of the second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set.

13. The method according to claim 11, characterized in that, When the data type belongs to the second type, the encoder input is the first CSI, the decoder input includes the third CSI, and the tag is the second CSI; Wherein, the first CSI represents the channel state of the first resource set, the second CSI represents the channel state of the second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set, the third CSI represents the channel state of the third resource set, the third resource set is different from the first resource set, the third resource set is a part or all of the second resource set, and the second resource set includes at least one of frequency domain resources or spatial domain resources.

14. The method according to claim 13, characterized in that, The second resource set corresponds to the first time domain resource, the first resource set corresponds to the first time domain resource, and the third resource set corresponds to the second time domain resource and / or the first time domain resource, wherein the second time domain resource is earlier than the first time domain resource.

15. The method according to any one of claims 11 to 14, characterized in that, The indication information indicates N types, which are N types of data included in the dataset on the terminal side, or N types of data included in the dataset on the network side. The N types include the first type or the second type, and N is a positive integer.

16. The method according to any one of claims 11 to 14, characterized in that, The indication information is used to determine M types, which are M types of data included in the dataset on the terminal side, or M types of data included in the dataset on the network side, one of which belongs to the first type or the second type, and M is a positive integer.

17. The method according to claim 16, characterized in that, The M types include at least one of the following: The encoder's input is the first CSI, and the tag is the first CSI; The encoder's input is the first CSI, and the tag is the second CSI; The encoder's input is a first CSI, the decoder's input includes a third CSI, and the tag is a second CSI; Wherein, the first CSI represents the channel state of the first resource set, the second CSI represents the channel state of the second resource set, the first resource set is a part of the second resource set, or the second resource set is different from the first resource set, the third CSI represents the channel state of the third resource set, the third resource set is different from the first resource set, the third resource set is a part of the second resource set, and the second resource set includes at least one of frequency domain resources or spatial domain resources.

18. The method according to any one of claims 11 to 17, characterized in that, The method further includes: Receive capability information, which indicates the computing capabilities and / or types of data supported on the terminal side; The sending instruction information includes: The instruction information is sent based on the capability information.

19. The method according to any one of claims 11 to 18, characterized in that, The method further includes: The encoder on the network side is sent. The encoder on the network side is obtained by training the second dual-end model based on the dataset on the network side. The second dual-end model is the dual-end model on the network side. The data included in the dataset on the network side belongs to the first type or the second type.

20. A communication device, characterized in that, It includes functional modules for implementing the method as described in any one of claims 1 to 19.

Images