Control method and device

Abstract

The present application relates to a control method, and a device. The method comprises: a terminal device receives a reference signal sent by a network device; and the terminal device sends to the network device at least one of first input channel state information (CSI), performance monitoring information, and a performance monitoring result, the first input CSI being determined on the basis of the reference signal.

Description

Control methods and equipment Technical Field

[0001] This application relates to the field of communications, and more specifically, to control methods and devices. Background Technology

[0002] The CSI feedback method based on joint source-channel coding can significantly improve the performance of CSI feedback under non-ideal uplink feedback conditions. In this method, both the terminal device and the network device deploy AI models to send and receive CSI feedback, respectively. How to manage and operate these models to improve feedback accuracy is a technical problem that needs to be solved.

[0003] Summary of the Invention

[0004] This application provides a control method and device.

[0005] This application provides a control method, including:

[0006] The terminal device receives reference signals sent by the network device;

[0007] The terminal device sends at least one of the following to the network device: a first input CSI, performance monitoring information, and performance monitoring results;

[0008] The first input CSI is determined based on the reference signal.

[0009] This application provides a control method, including:

[0010] Network devices send reference signals to terminal devices;

[0011] The network device receives at least one of the first input CSI, performance monitoring information, and performance monitoring results sent by the terminal device;

[0012] The first input CSI is determined based on the reference signal.

[0013] This application provides a terminal device, including:

[0014] The first transceiver module is configured to receive a reference signal sent by a network device and send at least one of a first input CSI, performance monitoring information, and performance monitoring results to the network device; wherein the first input CSI is determined based on the reference signal.

[0015] This application provides a network device, including:

[0016] The second transceiver module is used to send a reference signal to the terminal device and receive at least one of the first input CSI, performance monitoring information, and performance monitoring results sent by the terminal device; wherein the first input CSI is determined based on the reference signal.

[0017] This application provides a terminal device, including a transceiver, a processor, and a memory. The memory stores a computer program, the transceiver communicates with other devices, and the processor calls and runs the computer program stored in the memory to enable the terminal device to perform the control method described above.

[0018] This application provides a network device, including a transceiver, a processor, and a memory. The memory stores a computer program, the transceiver communicates with other devices, and the processor calls and runs the computer program stored in the memory to cause the network device to perform the control method described above.

[0019] This application provides a chip for implementing the control method described above.

[0020] Specifically, the chip includes a processor for retrieving and running a computer program from memory, causing a device equipped with the chip to perform the aforementioned control method.

[0021] This application provides a computer-readable storage medium for storing a computer program, which, when run by a device, causes the device to perform the control method described above.

[0022] This application provides a computer program product, including computer program instructions that cause a computer to execute the control method described above.

[0023] This application provides a computer program that, when run on a computer, causes the computer to execute the control method described above.

[0024] In this embodiment, the terminal device sends the first input CSI and the information required for performance monitoring to the network device, thereby providing the network device with the information needed to manage and operate the AI ​​model, thus realizing the management and operation of the model and improving the feedback accuracy. Attached Figure Description

[0025] Figure 1 illustrates a communication system 100 as an example.

[0026] Figure 2 is a schematic diagram of the AI-based CSI autoencoder framework.

[0027] Figure 3 is a schematic diagram of AI-based spatial-frequency-time joint CSI compressed feedback.

[0028] Figure 4 is a schematic diagram of AI-based joint CSI prediction and compression.

[0029] Figure 5A is a schematic diagram of the first working method of CSI feedback for joint source-channel coding.

[0030] Figure 5B is a schematic diagram of the second working method of CSI feedback for joint source-channel coding.

[0031] Figure 6 is a schematic flowchart of a control method 600 according to an embodiment of this application.

[0032] Figure 7 is a schematic flowchart of a control method 700 according to an embodiment of this application.

[0033] Figure 8 is a schematic diagram of the deployment architecture of the first model and the second model in one embodiment of this application.

[0034] Figure 9A is a schematic diagram of performance monitoring in Embodiment 2-1 of this application.

[0035] Figure 9B is a schematic diagram of performance monitoring in Embodiment 2-1 of this application.

[0036] Figure 10A is a schematic diagram of performance monitoring in Embodiment 3-1 of this application.

[0037] Figure 10B is a second schematic diagram of performance monitoring in Embodiment 3-1 of this application.

[0038] Figure 11 is a schematic block diagram of a terminal device 1100 according to an embodiment of the present application.

[0039] Figure 12 is a schematic block diagram of a terminal device 1200 according to an embodiment of the present application.

[0040] Figure 13 is a schematic block diagram of a network device 1300 according to an embodiment of the present application.

[0041] Figure 14 is a schematic block diagram of a network device 1400 according to an embodiment of the present application.

[0042] Figure 15 is a schematic structural diagram of a communication device 1500 according to an embodiment of this application.

[0043] Figure 16 is a schematic structural diagram of a chip 1600 according to an embodiment of this application.

[0044] Figure 17 is a schematic block diagram of a communication system 1700 according to an embodiment of this application. Detailed Implementation

[0045] The technical solutions in the embodiments of this application will now be described with reference to the accompanying drawings.

[0046] The technical solutions of this application embodiment can be applied to various communication systems, such as: Long Term Evolution (LTE) systems, Advanced Long Term Evolution (LTE-A) systems, New Radio (NR) systems, evolution systems of NR systems, LTE-based access to unlicensed spectrum (LTE-U) systems, NR-based access to unlicensed spectrum (NR-U) systems, Non-Terrestrial Networks (NTN) systems, Universal Mobile Telecommunication System (UMTS), Wireless Local Area Networks (WLAN), Wireless Fidelity (WiFi), 5th-Generation (5G) systems, or other communication systems.

[0047] Traditional communication systems typically support a limited number of connections and are easy to implement. However, with the development of communication technology, mobile communication systems will not only support traditional communication but also, for example, device-to-device (D2D) communication, machine-to-machine (M2M) communication, machine-type communication (MTC), vehicle-to-vehicle (V2V) communication, or vehicle-to-everything (V2X) communication. The embodiments of this application can also be applied to these communication systems.

[0048] In one implementation, the communication system in this application embodiment can be applied to a carrier aggregation (CA) scenario, a dual connectivity (DC) scenario, or a standalone (SA) network deployment scenario.

[0049] In one embodiment, the communication system in this application can be applied to unlicensed spectrum, wherein the unlicensed spectrum can also be considered as shared spectrum; or, the communication system in this application can also be applied to licensed spectrum, wherein the licensed spectrum can also be considered as non-shared spectrum.

[0050] This application describes various embodiments in conjunction with network devices and terminal devices. The terminal device may also be referred to as user equipment (UE), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent, or user device, etc.

[0051] Terminal devices can be stations (STAION, ST) in WLANs, cellular phones, cordless phones, Session Initiation Protocol (SIP) phones, Wireless Local Loop (WLL) stations, Personal Digital Assistant (PDA) devices, handheld devices with wireless communication capabilities, computing devices or other processing devices connected to a wireless modem, in-vehicle devices, wearable devices, terminal devices in next-generation communication systems such as NR networks, or terminal devices in future evolved Public Land Mobile Network (PLMN) networks, etc.

[0052] In the embodiments of this application, the terminal device can be deployed on land, including indoor or outdoor, handheld, wearable or vehicle-mounted; it can also be deployed on water (such as ships); and it can also be deployed in the air (such as airplanes, balloons and satellites).

[0053] In the embodiments of this application, the terminal device may be a mobile phone, a tablet computer, a computer with wireless transceiver capabilities, a virtual reality (VR) terminal device, an augmented reality (AR) terminal device, a wireless terminal device in industrial control, a wireless terminal device in self-driving, a wireless terminal device in remote medical care, a wireless terminal device in a smart grid, a wireless terminal device in transportation safety, a wireless terminal device in a smart city, or a wireless terminal device in a smart home, etc.

[0054] 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.

[0055] In the embodiments of this application, the network device can be a device for communicating with mobile devices, such as an access point (AP) in a WLAN, an evolved Node B (eNB or eNodeB) in LTE, a relay station or access point, or a vehicle-mounted device, a wearable device, a network device (gNB) in an NR network, or a network device in a future evolved PLMN network or an NTN network, etc.

[0056] By way of example and not limitation, in this embodiment, the network device may have mobility characteristics; for example, the network device may be a mobile device. Optionally, the network device may be a satellite or a balloon station. For example, the satellite may be a low Earth orbit (LEO) satellite, a medium Earth orbit (MEO) satellite, a geostationary earth orbit (GEO) satellite, a high elliptical orbit (HEO) satellite, etc. Optionally, the network device may also be a base station located on land, water, or other similar locations.

[0057] In this embodiment, the network device can provide services to a cell. The terminal device communicates with the network device through the transmission resources (e.g., frequency domain resources, or spectrum resources) used by the cell. The cell can be the cell corresponding to the network device (e.g., a base station). The cell can belong to a macro base station or to a base station corresponding to a small cell. The small cell can include: metro cell, micro cell, pico cell, femto cell, etc. These small cells have the characteristics of small coverage area and low transmission power, and are suitable for providing high-speed data transmission services.

[0058] Figure 1 illustrates an exemplary communication system 100. The communication system includes a network device 110 and two terminal devices 120. In one embodiment, the communication system 100 may include multiple network devices 110, and the coverage area of ​​each network device 110 may include other numbers of terminal devices 120; this embodiment does not limit the scope of the present application.

[0059] In one embodiment, the communication system 100 may also include other network entities such as a Mobility Management Entity (MME) and an Access and Mobility Management Function (AMF), which are not limited in this application.

[0060] Network equipment can be further divided into access network equipment and core network equipment. That is, the wireless communication system also includes multiple core networks used to communicate with the access network equipment. Access network equipment can be evolved Node Bs (eNBs or e-NodeBs) in Long-Term Evolution (LTE), Next-Generation Radio (NR) (mobile communication system), or Authorized Auxiliary Access Long-Term Evolution (LAA-LTE) systems, such as macro base stations, micro base stations (also called "small base stations"), pico base stations, access points (APs), transmission points (TPs), or new generation Node Bs (gNodeBs).

[0061] It should be understood that devices with communication functions in the network / system of this application embodiment can be referred to as communication devices. Taking the communication system shown in Figure 1 as an example, the communication device may include network devices and terminal devices with communication functions. The network devices and terminal devices can be specific devices in this application embodiment, which will not be described in detail here. The communication device may also include other devices in the communication system, such as network controllers, mobility management entities, and other network entities. This application embodiment does not limit this.

[0062] It should be understood that the terms "system" and "network" are often used interchangeably in this document. The term "and / or" in this document merely describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Furthermore, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0063] It should be understood that the term "instruction" mentioned in the embodiments of this application can be a direct instruction, an indirect instruction, or an indication of a relationship. For example, A instructing B can mean that A directly instructs B, such as B being able to obtain information through A; it can also mean that A indirectly instructs B, such as A instructing C, so B can obtain information through C; or it can mean that there is a relationship between A and B.

[0064] In the description of the embodiments of this application, the term "correspondence" may indicate that there is a direct or indirect correspondence between two things, or that there is an association between two things, or that there is a relationship of instruction and being instructed, configuration and being configured, etc.

[0065] To facilitate understanding of the technical solutions of the embodiments of this application, the relevant technologies of the embodiments of this application are described below. The following relevant technologies are optional solutions and can be combined with the technical solutions of the embodiments of this application in any way, and they all fall within the protection scope of the embodiments of this application.

[0066] I. Artificial Intelligence and Semantic Communication:

[0067] In recent years, Artificial Intelligence (AI) technology, relying on the development of different types of neural networks and deep learning algorithms, has been widely applied in various fields such as image, speech, and video processing. Drawing on the rapid development of AI technology, the combination of AI and wireless communication technology has also attracted widespread interest from academia and industry. Extensive research, evaluation, and standardization work has been carried out on AI-based Channel State Information (CSI) feedback, beam management, and positioning technologies.

[0068] In future wireless communication systems, semantic communication technology, combined with AI, has received widespread attention. In contrast, the traditional communication paradigm is called "syntactic communication," which uses traditional information theory methods such as Shannon's formula, information entropy, and the Nyquist sampling theorem to compress source information and then achieve efficient and reliable transmission through channel coding. The concept of "semantic communication," however, focuses on the extraction and representation of semantic information from source information, as well as the design of semantic communication systems integrated with AI, thereby coupling various modules together to achieve the transmission of semantic information.

[0069] A key difference between traditional grammatical communication systems and modern semantic communication lies in the design of source and channel coding. In traditional grammatical communication systems, source coding and channel coding are separate. Specifically, at the transmitter, source coding first compresses source information such as images, voice, and video into the original bitstream to be transmitted, which is then sent out via channel coding. At the receiver, the original bitstream is first recovered via channel decoding, and then the original source is reconstructed via source decoding. This separation of source and channel coding facilitates modular optimization and design. Source coding compresses information redundancy, saving transmission bandwidth, while channel decoding enhances redundancy, improving the information's ability to resist noise and interference, and enabling error detection and correction.

[0070] In semantic communication systems, source coding and channel coding are jointly designed. Specifically, on the transmitter side, a joint source-channel encoder is designed to directly encode the original source information into the bit stream to be transmitted; on the receiver side, a joint source-channel decoder is designed to directly recover the original source information from the received bit stream. This method of coupling source and channel can achieve joint optimization in source compression redundancy and channel redundancy. Furthermore, since both the encoder and decoder are designed using AI models, end-to-end training and deployment can be achieved, allowing the AI ​​model to be more adapted to the current wireless channel characteristics and noise conditions, thus optimizing system performance. Simultaneously, joint source-channel coding can further incorporate modulation and demodulation functions, further optimizing system performance.

[0071] II. CSI Feedback in NR:

[0072] In current NR systems, CSI feedback schemes typically employ codebook-based feature vector feedback to enable the base station to acquire downlink CSI. Specifically, the base station sends a downlink channel state information-reference signal (CSI-RS) to the user. The user uses the CSI-RS to estimate the downlink channel's CSI and performs eigenvalue decomposition on the estimated downlink channel to obtain its corresponding feature vector. NR provides two codebook design schemes: Type I and Type II.

[0073] III. AI-based CSI feedback (discussed in R18):

[0074] Given the tremendous success of AI technology, especially deep learning, in computer vision and natural language processing, the communications field has begun to explore its application to solve technical challenges that traditional communication methods struggle with. The neural network architectures commonly used in deep learning are non-linear and data-driven, enabling feature extraction from actual channel matrix data and, at the base station, reconstructing the compressed channel matrix information from the UE feedback as accurately as possible. This not only ensures accurate channel information reconstruction but also reduces CSI feedback overhead at the UE. Deep learning-based CSI feedback treats channel information as an image to be compressed, using a deep learning autoencoder to compress the input channel information and then reconstructing the compressed channel image at the transmitting end, thus preserving channel information to a greater extent.

[0075] In the R18 discussion of the Third Generation Partnership Project (3GPP), AI-based CSI feedback, as one of the main use cases of the AI ​​project, underwent multiple rounds of simulation results and discussions on its potential standard impact. Its main implementation framework is shown in Figure 2. Figure 2 is a schematic diagram of the AI-based CSI autoencoder framework. As shown in Figure 2, using the AI-based CSI autoencoder method, the entire feedback system is divided into encoder and decoder parts, deployed at the user transmitter and base station receiver, respectively. After the user obtains channel information through channel estimation, it serves as the input to the encoder. The encoder's neural network compresses and encodes the channel information matrix, and the compressed bitstream is fed back to the base station through the air interface feedback link. The base station uses the decoder to recover the channel information based on the feedback bitstream and outputs the complete feedback channel information.

[0076] IV. AI-based CSI feedback (discussed in R19):

[0077] In the R19 discussion, to further enhance the performance of AI-based CSI feedback, the temporal correlation between CSI at different times was utilized to improve the compressed feedback performance of CSI. Specifically, different implementation methods mainly fall into the following two categories:

[0078] 1) AI-based joint spatial-frequency-time domain CSI compression feedback, as shown in Figure 3: The encoder input on the user side includes not only the CSI measurement information at the current time but also historical CSI information from past time points. This historical information is characterized by the feature output of the latent vector space from the encoder output at past time points and does not have explicit physical meaning. Similarly, the decoder input on the network side includes not only the feedback information at the current time but also historical CSI information from past time points. This historical information is characterized by the feature output of the latent space from the decoder output at past time points and also does not have explicit physical meaning. For the feedback information at time t on the air interface, it not only implicitly represents the features of the CSI at the current time but also includes information features from historical CSI that help in the recovery of the CSI at the current time, which can be used by the network side to better recover the CSI at the current time.

[0079] 2) AI-based joint CSI prediction and compression, as shown in Figure 4: The user-side CSI prediction model takes the CSI measurement information within the measurement window as input and outputs the predicted CSI information at least one time step. This is preprocessed (e.g., if the predicted CSI information is full-channel information, Singular Value Decomposition (SVD) can be used to extract feature vectors from different layers for the CSI compression process) and then used as the encoder input. When the encoder input is CSI information from multiple time steps, the encoder output is joint feedback information, and the decoder can simultaneously recover CSI information from multiple prediction time steps. By extracting the correlation information of CSI from multiple time steps, the performance of CSI feedback can be further enhanced with priority given to air interface feedback overhead. Furthermore, the user-side CSI prediction and CSI encoding compression processes can be either separated as shown in Figure 3 or coupled, i.e., completed through a single AI model, as indicated by the dashed box.

[0080] As mentioned above, both use cases considered in R19 utilize AI models to extract, compress, and leverage the correlations across multiple CSI time points, aiming to further improve CSI feedback performance. The CSI compression and recovery models need to be trained using collected datasets. Currently, the types of models used to train CSI compression and recovery models include the following three:

[0081] The first method involves training a portion of the model (CSI compressed model or CSI restored model) on a single device (such as a terminal device or network device), and then sending the trained portion of the model to another device.

[0082] The second method involves jointly training the CSI compression model and the CSI recovery model on the terminal device and the network device, respectively.

[0083] The third approach involves first training a portion of the model on a single device, then sending the training data or other auxiliary information to another device to train the remaining portion of the model. This training process can be further divided into two forms: the network device first trains the CSI compression model and the CSI recovery model, and then sends the dataset used to train the CSI compression model and / or other auxiliary information to the terminal device; or, the terminal device first trains the CSI compression model and the CSI recovery model, and then sends the dataset used to train the CSI recovery model and / or other auxiliary information to the network device.

[0084] V. CSI Feedback Method Based on Source-Channel Joint Coding:

[0085] Besides the AI-based CSI feedback discussed in 3GPP, there is another type of method that borrows from semantic communication: a joint source-channel CSI feedback approach. Figure 5A illustrates this first method. On the transmitter side, a joint source-channel encoder is designed, treating the measured CSI as the original source information and directly encoding it into the bit stream to be transmitted. This bit stream is then mapped to physical resources using a traditional modulation module. On the receiver side, a joint source-channel decoder is designed, directly recovering the original source information (i.e., recovering the CSI) from the demodulated log-likelihood ratio. This method, which couples source and channel coding, achieves joint optimization in both source compression redundancy and channel redundancy. Furthermore, since both the encoder and decoder are designed using AI models, end-to-end training and deployment are possible. This allows the AI ​​model to better adapt to current wireless channel characteristics and noise conditions, further optimizing the performance of the CSI encoder and decoder against noise and fading channels in non-ideal channels, thus achieving optimal system performance.

[0086] Furthermore, joint source-channel coding can also incorporate modulation and demodulation functions. As shown in Figure 5B, this is the second working method. In this method, the output of the joint source-channel encoder is not a bitstream, but rather complex symbols that can be directly mapped to physical resources. Power constraints are applied to these complex symbols to meet certain conditions. At the receiver, the joint source-channel decoder directly uses the received symbols after channel equalization and symbol detection as input and directly outputs the original information. This end-to-end design approach achieves a balance between source coding, channel coding, and modulation, further optimizing end-to-end link performance.

[0087] In existing discussions, after the two-end model is trained, it is deployed on terminal devices and / or network devices. During application, the performance of the trained two-end model will change due to changes in the channel environment. CSI feedback based on joint source-channel coding can significantly improve the performance of CSI feedback under non-ideal uplink feedback conditions, but how to manage the AI ​​model under such conditions is a technical problem that needs to be solved.

[0088] Figure 6 is a schematic flowchart of a control method 600 according to an embodiment of this application. This method can optionally be applied to the systems or scenarios shown in Figures 1 to 5B, but is not limited thereto. The method includes at least a portion of the following:

[0089] S610, The terminal device receives the reference signal sent by the network device;

[0090] S620, the terminal device sends at least one of a first input channel state information (CSI), performance monitoring information, and performance monitoring results to the network device; wherein the first input CSI is determined based on the reference signal.

[0091] By sending the first input CSI and the information required for performance monitoring to the network device, the terminal device can provide the information needed to manage and operate the AI ​​model, thereby enabling the management and operation of the model and improving the feedback accuracy.

[0092] In some implementations, it also includes:

[0093] The terminal device performs a first management operation, wherein the first management operation includes at least one of the following:

[0094] Switching between AI models;

[0095] Activate the AI ​​model;

[0096] Deactivate the AI ​​model;

[0097] Update the AI ​​model;

[0098] Stop using AI models for CSI feedback;

[0099] Adjust the allocation of upstream resources.

[0100] The embodiments of this application can be applied to CSI feedback scenarios based on joint source channel coding. By performing a first management operation that includes multiple methods, the AI ​​model or uplink resource allocation can be adjusted, thereby improving the feedback accuracy.

[0101] In one implementation, the AI ​​model includes at least one of a first model and a second model; wherein,

[0102] The first model is deployed on the terminal device and is used to perform joint source-channel coding on the CSI feedback sent by the terminal device;

[0103] The second model is deployed on network devices to perform joint source-channel decoding on the CSI feedback received by the network devices.

[0104] In some implementations, the terminal device may perform a first management operation based on instructions from the network device. For example, the terminal device receives model management information sent by the network device; the terminal device then performs the first management operation based on the model management information.

[0105] In this scenario, the network device monitors the performance of the AI ​​model and generates model management information based on the monitoring results. The terminal device can send necessary data to the network device for performance monitoring. For example, before receiving the model management information from the network device, the terminal device also sends a first transmission to the network device, wherein...

[0106] The first input CSI is the input content of the first model when performing joint source-channel feedback;

[0107] The first traditional CSI includes: CSI measured and reported using a preset method.

[0108] In other embodiments, the terminal device may perform performance monitoring and send the performance monitoring results to the network device; the network device generates model management information based on the received performance monitoring results and sends the model management information to the terminal device; the terminal device performs a first management operation based on the received model management information. For example, before the terminal device receives the model management information sent by the network device, the process may include: the terminal device determining the performance monitoring results of the AI ​​model; and the terminal device sending the performance monitoring results of the AI ​​model to the network device for the network device to determine the model management information.

[0109] In other embodiments, the terminal device may perform performance monitoring and, based on the performance monitoring results, perform a first management operation. For example, the terminal device determines the performance monitoring results of the AI ​​model; the terminal device performs the first management operation based on the performance monitoring results of the AI ​​model.

[0110] When a terminal device performs performance monitoring, it needs to use relevant data sent by the network device. For example, in one example, the terminal device determines the performance monitoring result of the AI ​​model, including: the terminal device receiving a second output CSI sent by the network device; and the terminal device determining the performance monitoring result of the AI ​​model based on the second output CSI.

[0111] The second output CSI includes: the result output by the second model after performing joint source-channel decoding on the second input CSI;

[0112] The second input CSI includes: the result of the first output CSI transmitted via the uplink channel;

[0113] The first output CSI includes the result of the first model performing joint source-channel coding on the first input CSI.

[0114] Terminal devices can perform performance monitoring by using the second output CSI sent by network devices. There are two methods for performance monitoring using the second output CSI:

[0115] Method 1:

[0116] The terminal device determines the first monitoring value based on the first input CSI and the second output CSI;

[0117] The terminal device determines the second monitoring value based on the first input CSI and the fourth output CSI;

[0118] The fourth output CSI includes the result of the fourth model performing joint source-channel decoding on the fourth input CSI.

[0119] The fourth model is the same as the second model, or the fourth model and the second model were trained on the same dataset;

[0120] The fourth input CSI is the output data transmitted through the uplink channel within the training set range.

[0121] In one example, the terminal device determines a first monitoring value based on a first input CSI and a second output CSI, including: the terminal device determines the first monitoring value based on the generalized cosine similarity (SGCS), difference, or correlation between the first input CSI and the second output CSI.

[0122] In one example, the terminal device determines a second monitoring value based on a first input CSI and a fourth output CSI, including: the terminal device determines the second monitoring value based on the SGCS, difference, or correlation between the first input CSI and the fourth output CSI.

[0123] Furthermore, the performance monitoring process may also include:

[0124] The terminal device determines the first monitoring result based on the first monitoring value and the first threshold.

[0125] The terminal device determines the second monitoring result based on the second monitoring value and the second threshold.

[0126] Furthermore, the performance monitoring process may also include:

[0127] The terminal device determines whether a first management operation is required and / or the reason for requiring a first management operation based on at least one of the first monitoring results and the second monitoring results.

[0128] Method 2:

[0129] The terminal device determines the third monitoring value based on the second and fourth CSI outputs;

[0130] The fourth output CSI includes the result of the fourth model performing joint source-channel decoding on the fourth input CSI.

[0131] The fourth model is the same as the second model, or the fourth model and the second model were trained on the same dataset;

[0132] The fourth input CSI is the output data transmitted through the uplink channel within the training set range.

[0133] In one example, the terminal device determines a third monitoring value based on the second output CSI and the fourth output CSI, including: the terminal device determines the third monitoring value based on the SGCS, difference, or correlation between the second output CSI and the fourth output CSI.

[0134] Furthermore, the performance monitoring process may also include: the terminal device determining a third monitoring result based on the third monitoring value.

[0135] Furthermore, the performance monitoring process may also include: the terminal device determining, based on the third monitoring results, whether a first management operation is required and / or the reason for requiring the first management operation.

[0136] This application also proposes a control method. Figure 7 is a schematic flowchart of a control method 700 according to an embodiment of this application. This method can optionally be applied to the systems or scenarios shown in Figures 1 to 5B, but is not limited thereto. The method includes at least a portion of the following:

[0137] S710, the network device sends a reference signal to the terminal device;

[0138] S720, the network device receives at least one of a first input CSI, performance monitoring information, and performance monitoring results sent by the terminal device; wherein the first input CSI is determined based on the reference signal.

[0139] By receiving the first input CSI and performance monitoring information sent by the terminal device, the network device can obtain the information needed to manage and operate the AI ​​model, thereby realizing the management and operation of the model and improving the feedback accuracy.

[0140] In some implementations, it also includes:

[0141] Network devices determine model management information;

[0142] The network device sends the model management information to the terminal device, the model management information being used to instruct the terminal device to perform a first management operation, the first management operation including at least one of the following:

[0143] Switching between AI models;

[0144] Activate the AI ​​model;

[0145] Deactivate the AI ​​model;

[0146] Update the AI ​​model;

[0147] Stop using AI models for CSI feedback;

[0148] Adjust the allocation of upstream resources.

[0149] The embodiments of this application can be applied to CSI feedback scenarios based on joint source channel coding. The network device determines and sends model management information, which instructs the terminal device to perform a first management operation including multiple methods, which can adjust the AI ​​model or adjust the uplink resource allocation, thereby improving the feedback accuracy.

[0150] In one implementation, the AI ​​model includes at least one of a first model and a second model; wherein,

[0151] The first model is deployed on the terminal device and is used to perform joint source-channel coding on the CSI feedback sent by the terminal device;

[0152] The second model is deployed on network devices to perform joint source-channel decoding on the CSI feedback received by the network devices.

[0153] In some implementations, the network device can perform performance monitoring of the AI ​​model based on data sent by the terminal device, and generate model management information based on the performance monitoring results. For example, the network device receives a first conventional CSI sent by the terminal device; the network device determines the performance monitoring results of the AI ​​model based on at least one of a first input CSI and a first conventional CSI, and the performance monitoring results of the AI ​​model are used to determine the model management information; wherein,

[0154] The first input CSI includes: the input content of the first model when performing joint source-channel feedback;

[0155] The first traditional CSI includes CSI that is measured and reported using a preset method.

[0156] The following are some ways to monitor the performance of network devices:

[0157] Method 1:

[0158] The network device determines the first monitoring value based on the first input CSI and the second output CSI;

[0159] The network device determines the second monitoring value based on the first input CSI and the third output CSI;

[0160] The second output CSI includes the result of the second model performing joint source-channel decoding on the second input CSI.

[0161] The second input CSI includes: the result of the first output CSI after transmission via the uplink channel;

[0162] The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI;

[0163] The third output CSI includes the result of the second model performing joint source-channel decoding on the third input CSI.

[0164] The third input CSI includes: the result output by the third model after performing joint source-channel decoding on the input content;

[0165] The third model is the same as the first model, or the third model and the first model are trained using the same dataset; the input content is the output data transmitted through the uplink channel within the training set range.

[0166] In one example, the network device determines a first monitoring value based on a first input CSI and a second output CSI, including: the network device determines the first monitoring value based on the SGCS, difference, or correlation between the first input CSI and the second output CSI.

[0167] In one example, the network device determines a second monitoring value based on a first input CSI and a third output CSI, including: the network device determines the second monitoring value based on the SGCS, difference, or correlation between the first input CSI and the third output CSI.

[0168] Furthermore, the performance monitoring process may also include:

[0169] The network device determines the first monitoring result based on the first monitoring value and the first threshold.

[0170] The network device determines the second monitoring result based on the second monitoring value and the second threshold.

[0171] Furthermore, the performance monitoring process may also include:

[0172] The network device determines whether a first management operation is required and / or the reason for requiring a first management operation based on at least one of the first monitoring results and the second monitoring results.

[0173] Method 2:

[0174] The network device determines the third monitoring value based on the second and third output CSIs;

[0175] The second output CSI includes the result of the second model performing joint source-channel decoding on the second input CSI.

[0176] The second input CSI includes: the result of the first output CSI after transmission via the uplink channel;

[0177] The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI;

[0178] The third output CSI includes the result of the second model performing joint source-channel decoding on the third input CSI.

[0179] The third input CSI includes: the result output by the third model after performing joint source-channel decoding on the input content;

[0180] The third model is the same as the first model, or the third model and the first model are trained using the same dataset; the input content is the output data transmitted through the uplink channel within the training set range.

[0181] In one example, the network device determines a third monitoring value based on the second output CSI and the third output CSI, including: the network device determines the third monitoring value based on the SGCS, difference, or correlation between the second output CSI and the third output CSI.

[0182] Furthermore, the performance monitoring process may also include: the network device determining a third monitoring result based on a third monitoring value and a third threshold.

[0183] Furthermore, the performance monitoring process may also include: the network device determining, based on the third monitoring results, whether a first management operation is required and / or the reason for requiring the first management operation.

[0184] Method 3:

[0185] The network device determines a fourth monitoring value based on a first input CSI and a first conventional CSI. In one example, the network device determines the fourth monitoring value based on the SGCS, difference, or correlation between the first input CSI and the first conventional CSI.

[0186] Furthermore, the performance monitoring process may also include: the network device determining a fourth monitoring result based on a fourth monitoring value and a fourth threshold.

[0187] Furthermore, the performance monitoring process may also include: the network device determining, based on the fourth monitoring result, whether a first management operation is required and / or the reason for requiring a first management operation.

[0188] In other implementations, the network device may also send necessary data to the terminal device, which will then perform performance monitoring of the AI ​​model and obtain the performance monitoring results. The network device may receive the performance monitoring results sent by the terminal device, generate model management information based on the performance monitoring results, and send the model management information to the terminal device to instruct the terminal device to perform a first management operation.

[0189] For example, network devices send a second output CSI to terminal devices, and the second output CSI is used by the terminal devices to determine the performance monitoring results of the AI ​​model;

[0190] Network devices receive performance monitoring results of AI models sent by terminal devices;

[0191] The second output CSI includes the result of the second model performing joint source-channel decoding on the second input CSI.

[0192] The second input CSI includes: the result of the first output CSI after transmission via the uplink channel;

[0193] The first output CSI includes the result of the first model performing joint source-channel coding on the first input CSI.

[0194] It should be noted that the three methods for network device performance monitoring involve the first to fourth monitoring values, the first to fourth thresholds, and the first to fourth monitoring results. These contents are unrelated to the first to third monitoring values, the first to third thresholds, and the first to third monitoring results involved in the aforementioned methods for terminal device performance monitoring.

[0195] For a specific example of the network device executing method 700 in this embodiment, please refer to the relevant description of the network device, such as the base station, in the above method 700. For the sake of brevity, it will not be repeated here.

[0196] The following detailed description, in conjunction with the accompanying drawings, provides specific embodiments.

[0197] Example 1:

[0198] In this embodiment, the terminal device deploys a first model, which is used to perform joint source-channel coding on the CSI feedback sent by the terminal device; the network device deploys a second model, which is used to perform joint source-channel decoding on the CSI feedback received by the network device. Figure 8 is a schematic diagram of the deployment architecture of the first and second models in one embodiment of this application. As shown in Figure 8, the first model is deployed on the terminal device, and the data processed by the first model is transmitted to the network device through the uplink channel; the network device deploys the second model to perform joint source-channel decoding on the received data.

[0199] This embodiment describes the training methods for the first model and / or the second model, including at least the following:

[0200] Method 1: The network device trains the first and second models, and transmits the first model to the terminal device. The second model has two inputs: the second input CSI, or the second CSI via the uplink channel. The output of the second model is the second output CSI. The network can store a third model, which can be the same as the first model or have the same dataset as the first model.

[0201] Method 2: The network device trains the first and second models and transmits the dataset related to the first model to the terminal device. The second model has two inputs: a second input CSI or a second CSI transmitted via the uplink channel. The output of the second model is the second output CSI. The network can store a third model, which can be the same as the first model or have the same dataset as the first model.

[0202] Method 3: The terminal device trains the first and second models and transmits the dataset related to the second model to the network device. The terminal device can store a fourth model, which is identical to the second model.

[0203] Method 4: The terminal device trains the first and second models and transmits the dataset related to the second model to the network device. The terminal device can store the fourth model, which has the same dataset as the second model.

[0204] Method 5: The first model on the terminal side and the second model on the network side are trained separately. The network-first (NW-first) training method or the UE-first (UE-first) training method described in the existing technology can be used. After training, the model can still be sent to the other side and stored for model performance monitoring.

[0205] Example 2:

[0206] In this embodiment, the network device performs performance monitoring to obtain the performance monitoring results of the AI ​​model, generates model management information based on the performance monitoring results, and sends the model management information; the terminal device receives the model management information sent by the network device and performs a first management operation based on the model management information. The first management operation includes at least one of the following:

[0207] Switching between AI models;

[0208] Activate the AI ​​model;

[0209] Deactivate the AI ​​model;

[0210] Update the AI ​​model;

[0211] Stop using AI models for CSI feedback;

[0212] Adjust the allocation of upstream resources.

[0213] In some examples, the terminal device can send the content required for performance monitoring to the network device before the network device performs performance monitoring. Additionally, before the terminal device sends the required content, a performance monitoring triggering process may also be included; this can be triggered by the terminal device or instructed by the network device regarding those performance monitoring actions.

[0214] Network devices can be monitored in various ways. The following examples 2-1, 2-2, and 2-3 illustrate three ways to monitor the performance of network devices.

[0215] Example 2-1:

[0216] In this embodiment, a first model is deployed on the terminal device side, a second model is deployed on the network device side, and a third model can also be stored on the network device side. The third model is the same as the first model, or the third model and the first model are trained using the same dataset.

[0217] This embodiment includes the following stages:

[0218] Phase 1: The terminal device sends the first input CSI to the network device. The first input CSI can be measured and reported using a preset method, such as traditional CSI feedback reporting supported by existing protocols, which includes CSI measurement and reporting based on Type I or Type II codebooks. The codebook type or parameters are configured by the network device for the terminal device. Alternatively, the first input CSI can be a higher-precision CSI transmitted to the network device using an optimized method. The first input CSI can be obtained from a reference signal sent by the network device.

[0219] In Phase Two, after receiving the first input CSI, the network device performs performance monitoring of the AI ​​model using the following steps.

[0220] The first step, as shown in Figure 9A, is to compare the received first input CSI with the output value of the second model (i.e., the second output CSI). The second input CSI is the input data of the second model, which is the data transmitted through the uplink channel after the first output CSI. The first output CSI is the output value of the first model deployed on the terminal device, whose input data is the first input CSI.

[0221] One possible approach is to use the SGCS (SGatonic Response Center) between the first input CSI (Concurrent Input System Index) and the second output CSI (Concurrent Output System Index) to obtain the first monitoring value. The network device configures a first threshold and obtains a first monitoring result based on the first monitoring value and the first threshold. For example, if the first monitoring value is greater than or equal to the first threshold, the first monitoring result is that no first management operation is required; if the first monitoring value is less than the first threshold, the first monitoring result is that first management operation is required. In other words, the closer the SGCS is to 1, the better the performance.

[0222] Another possible approach is to use the difference between the first input CSI and the second output CSI to obtain the first monitoring value. The network device is configured with a first threshold, and the first monitoring result is obtained based on the first monitoring value and the first threshold. For example, if the first monitoring value is greater than the first threshold, the first monitoring result indicates that a first management operation is required; if the first monitoring value is less than or equal to the first threshold, the first monitoring result indicates that no first management operation is required. In other words, the larger the difference, the worse the performance.

[0223] Another possible approach is to measure the correlation between the first input CSI and the second output CSI to obtain the first monitoring value. The network device configures a first threshold and obtains a first monitoring result based on the first monitoring value and the first threshold. For example, if the first monitoring value is greater than or equal to the first threshold, the first monitoring result is that no first management operation is required; if the first monitoring value is less than the first threshold, the first monitoring result is that first management operation is required. That is, the higher the correlation, the better the performance.

[0224] If the first step determines that a first management operation is required, it indicates that one or more of the first model, uplink channel, and second model (within the dashed line in Figure 9A) are causing poor performance. Then, the second step can be performed to determine the reason for the need for the first management operation.

[0225] The second step involves the network device performing performance monitoring using the stored third model. This third model can be the same as the first model, or both can be trained on the same dataset. The uplink channel for the third model is the uplink channel within the training set. As shown in Figure 9B, the input data for the second model is the third input CSI; the third input CSI is the output data of the third model after transmission through the uplink channel within the training dataset, meaning it considers the uplink channel within the training set and is the inferred value of the third model. The output data for the second model is the third output CSI. The network device compares the received first input CSI with the third output CSI.

[0226] One possible approach is to use the SGCS (SGatonic Response Center) between the first input CSI (Concurrent Input System) and the third output CSI (Concurrent Output System) to obtain the second monitoring value. The network device is configured with a second threshold, and the second monitoring result is derived based on the second monitoring value and the second threshold. For example, if the second monitoring value is greater than or equal to the second threshold, the second monitoring result indicates that no first management operation is required; if the second monitoring value is less than the second threshold, the second monitoring result indicates that the first management operation is required. In other words, the closer the SGCS is to 1, the better the performance.

[0227] Another possible approach is to use the difference between the first input CSI and the third output CSI to obtain the second monitoring value. The network device is configured with a second threshold, and the second monitoring result is derived based on both the second monitoring value and the second threshold. For example, if the second monitoring value is greater than the second threshold, the second monitoring result indicates that a first management operation is required; if the second monitoring value is less than or equal to the second threshold, the second monitoring result indicates that no first management operation is required. In other words, the larger the difference, the worse the performance.

[0228] Another possible approach is to measure the correlation between the first input CSI and the third output CSI to obtain the second monitoring value. The network device is configured with a second threshold, and the second monitoring result is obtained based on the second monitoring value and the second threshold. For example, if the second monitoring value is greater than or equal to the second threshold, the second monitoring result indicates that no first management operation is required; if the second monitoring value is less than the second threshold, the second monitoring result indicates that the first management operation is required. In other words, a higher correlation indicates better performance.

[0229] The third step is to determine the cause of poor performance based on the performance monitoring results in the first and second steps, that is, to determine the reason for the need for the first management operation.

[0230] In one example, if the first monitoring result indicates that a first management operation is required, and the second monitoring result indicates that a first management operation is not required, then the uplink channel is determined to be the cause of poor performance, i.e., the uplink channel is the cause of requiring a first management operation. If both the first and second monitoring results indicate that a first management operation is required, then both the AI ​​model (including the first model and / or the second model) and the uplink channel are determined to be the causes of poor performance. It should be noted that when making the aforementioned judgment, the first prediction result and the second prediction result need to be measured using the same metric. For example, the first prediction result is determined based on the first monitoring value, and the second prediction result is determined based on the second monitoring value. Both the first and second monitoring values ​​are measured based on the SGCS of the relevant CSIs, or both are measured based on the difference between the relevant CSIs, or both are measured based on the correlation between the relevant CSIs.

[0231] Using a third model for auxiliary performance monitoring can further identify whether poor performance is caused by the uplink channel, enabling more accurate model management and avoiding unnecessary model switching.

[0232] In Phase Three, the network device generates model management information based on the performance monitoring results obtained in Phase Two and sends this information to the terminal device. For example, the model management information may include: No first management operation is required. Alternatively, it may include: The first management operation is required. Yet another example may include: The first management operation is required, and the uplink channel is the reason for requiring the first management operation.

[0233] The terminal device performs corresponding operations based on the model management information. Specifically, the first management operation may include at least one of the following: activating the AI ​​model, deactivating the AI ​​model, updating the AI ​​model, stopping the use of the AI ​​model for CSI feedback, and adjusting uplink resource allocation. In one example, if the uplink channel is the reason for performing the first management operation, then the first management operation may be adjusting uplink resource allocation.

[0234] Example 2-2:

[0235] In this embodiment, a first model is deployed on the terminal device side, a second model is deployed on the network device side, and a third model is also stored on the network device side. The third model is the same as the first model, or the third model and the first model are trained using the same dataset.

[0236] This embodiment includes the following stages:

[0237] In Phase 1, the network device compares the third output CSI with the second output CSI to obtain the monitoring results.

[0238] As shown in Figure 9A above, the second output CSI is the output value of the second model. At this time, the input data of the second model is the second input CSI, which is the data after the first output CSI is transmitted through the uplink channel. The first output CSI is the output value of the first model deployed by the terminal device, and the input data of the first model is the first input CSI. The first input CSI can be obtained through a reference signal sent by the network device.

[0239] As shown in Figure 9B above, the second output CSI is the output value of the second model. At this time, the input data of the second model is the third input CSI. The third input CSI is the output data of the third model after transmission through the uplink channel within the training dataset range. That is, the third input CSI considers the uplink channel within the training set range. The third input CSI is the inferred value of the third model.

[0240] One possible approach is to use the SGCS (SGatactic coefficients) between the second and third output CSIs to obtain the third monitoring value. The network device configures a third threshold and obtains the third monitoring result based on the third monitoring value and the third threshold. For example, if the third monitoring value is greater than or equal to the third threshold, the third monitoring result indicates that no first management operation is required; if the third monitoring value is less than the third threshold, the third monitoring result indicates that a first management operation is required, and the uplink channel is the cause of the requirement. In other words, the closer the SGCS is to 1, the better the performance.

[0241] Another possible approach is to use the difference between the second and third output CSIs to obtain the third monitoring value. The network device configures a third threshold, and the third monitoring result is derived based on the third monitoring value and the third threshold. For example, if the third monitoring value is greater than or equal to the third threshold, the third monitoring result indicates that a first management operation is required, and the uplink channel is the cause of this requirement; if the third monitoring value is less than the third threshold, the third monitoring result indicates that no first management operation is required. In other words, the larger the difference, the worse the performance.

[0242] Another possible approach is to measure the correlation between the second and third output CSIs to obtain the third monitoring value. The network device configures a third threshold, and the third monitoring result is derived based on the third monitoring value and the third threshold. For example, if the third monitoring value is greater than or equal to the third threshold, the third monitoring result indicates that no first management operation is required; if the third monitoring value is less than the third threshold, the third monitoring result indicates that a first management operation is required, and the uplink channel is the cause of this requirement. In other words, a higher correlation indicates better performance.

[0243] Using a third model for auxiliary performance monitoring can further identify whether poor performance is caused by the uplink channel, enabling more accurate model management and avoiding unnecessary model switching. Furthermore, since only one judgment is required, this method is simpler than the method in Example 2-1.

[0244] In Phase Two, the network device generates model management information based on the performance monitoring results obtained in Phase One and sends this information to the terminal device. For example, the model management information might include: No first management operation is required. Alternatively, it might include: First management operation is required. Yet another example might include: First management operation is required, and the uplink channel is the reason for requiring the first management operation.

[0245] The terminal device performs corresponding operations based on the model management information. Specifically, the first management operation may include at least one of the following: activating the AI ​​model, deactivating the AI ​​model, updating the AI ​​model, stopping the use of the AI ​​model for CSI feedback, and adjusting uplink resource allocation. In one example, if the uplink channel is the reason for performing the first management operation, then the first management operation may be adjusting uplink resource allocation.

[0246] Examples 2-3:

[0247] In this embodiment, the first model is deployed on the terminal device side, and the second model is deployed on the network device side.

[0248] This embodiment includes the following stages:

[0249] Phase 1: The terminal device sends a first input CSI to the network device. This first input CSI can be a higher-precision CSI transmitted to the network device using an optimized method. Furthermore, the terminal device sends a first traditional CSI to the network device. This first traditional CSI can be a CSI measured and reported using a preset method, such as traditional CSI feedback reporting supported by existing protocols. This includes CSI measurement and reporting based on Type I or Type II codebooks, where the codebook type or parameters are configured by the network device for the terminal device. The first input CSI can be obtained from a reference signal sent by the network device.

[0250] In Phase Two, after receiving the first input CSI and the first traditional CSI, the network device performs performance monitoring of the AI ​​model using the following steps.

[0251] The first step involves the network device comparing the received first input CSI with the output value of the second model (i.e., the second output CSI), and determining the monitoring result based on the comparison result. The specific method is the same as the first step of the second stage in Example 2-1, and will not be repeated here. If it is determined that a first management operation is required, the following second step of judgment is then executed.

[0252] The second step involves the network device comparing the first conventional CSI with the first input CSI and determining the monitoring result based on the comparison result.

[0253] One possible approach is to use the SGCS (SGatactic coefficients) between the first input CSI (Cybernetic Interface) and the first conventional CSI (Cybernetic Interface) to obtain the fourth monitoring value. The network device configures a fourth threshold, and the fourth monitoring result is derived based on the fourth monitoring value and the fourth threshold. For example, if the fourth monitoring value is greater than or equal to the fourth threshold, the fourth monitoring result indicates that no first management operation is required; if the fourth monitoring value is less than the fourth threshold, the fourth monitoring result indicates that the first management operation is required. In other words, the closer the SGCS is to 1, the better the performance.

[0254] Another possible approach is to use the difference between the first input CSI and the first conventional CSI to obtain the fourth monitoring value. The network device is configured with a fourth threshold, and the fourth monitoring result is derived based on the fourth monitoring value and the fourth threshold. For example, if the fourth monitoring value is greater than the fourth threshold, the fourth monitoring result indicates that a first management operation is required; if the fourth monitoring value is less than or equal to the fourth threshold, the fourth monitoring result indicates that no first management operation is required. In other words, the larger the difference, the worse the performance.

[0255] Another possible approach is to measure the correlation between the first input CSI and the first conventional CSI to obtain the fourth monitoring value. The network device is configured with a fourth threshold, and the fourth monitoring result is derived based on the fourth monitoring value and the fourth threshold. For example, if the fourth monitoring value is greater than or equal to the fourth threshold, the fourth monitoring result indicates that no first management operation is required; if the fourth monitoring value is less than the fourth threshold, the fourth monitoring result indicates that the first management operation is required. In other words, a higher correlation indicates better performance.

[0256] The third step is to determine the cause of poor performance based on the performance monitoring results in the first and second steps, that is, to determine the reason for the need for the first management operation.

[0257] In one example, if the first monitoring result indicates that a first management operation is required, and the fourth monitoring result indicates that a first management operation is not required, then the uplink channel is determined to be the cause of poor performance, i.e., the uplink channel is the cause of the need for a first management operation; if both the first and fourth monitoring results indicate that a first management operation is required, then both the AI ​​model (including the first model and / or the second model) and the uplink channel are determined to be the causes of poor performance.

[0258] It should be noted that when making the aforementioned judgments, the first and fourth prediction results need to be measured using the same indicators. For example, the first prediction result is determined based on the first monitoring value, and the fourth prediction result is determined based on the fourth monitoring value. Both the first and fourth monitoring values ​​are measured based on the SGCS of the relevant CSIs, or both are measured based on the difference between the relevant CSIs, or both are measured based on the correlation between the relevant CSIs.

[0259] In Phase Three, the network device generates model management information based on the performance monitoring results obtained in Phase Two and sends this information to the terminal device. For example, the model management information may include: No first management operation is required. Alternatively, it may include: The first management operation is required. Yet another example may include: The first management operation is required, and the uplink channel is the reason for requiring the first management operation.

[0260] The terminal device performs corresponding operations based on the model management information. Specifically, the first management operation may include at least one of the following: activating the AI ​​model, deactivating the AI ​​model, updating the AI ​​model, stopping the use of the AI ​​model for CSI feedback, and adjusting uplink resource allocation. In one example, if the uplink channel is the reason for performing the first management operation, then the first management operation may be adjusting uplink resource allocation.

[0261] The above embodiments describe how to perform performance monitoring on the network device side and determine model management information based on the monitoring results.

[0262] In this embodiment, performance monitoring can also be performed by the terminal device. After performance monitoring, the terminal device can send the performance monitoring results to the network device. The network device generates model management information based on the performance monitoring results and sends the model management information to the terminal device. The terminal device performs a first management operation based on the model management information. Alternatively, the network device performs performance monitoring and then performs the first management operation based on the performance monitoring results. The following describes the implementation method of performance monitoring performed on the terminal device side using Embodiment 3.

[0263] Example 3:

[0264] In this embodiment, the terminal device performs performance monitoring to obtain the performance monitoring results of the AI ​​model, and sends the performance monitoring results to the network device; the network device generates model management information based on the performance monitoring results and sends the model management information; the terminal device receives the model management information sent by the network device and performs a first management operation based on the model management information. The first management operation includes at least one of the following:

[0265] Switching between AI models;

[0266] Activate the AI ​​model;

[0267] Deactivate the AI ​​model;

[0268] Update the AI ​​model;

[0269] Stop using AI models for CSI feedback;

[0270] Adjust the allocation of upstream resources.

[0271] In some implementations, the terminal device reports performance monitoring information of the first model and / or the second model to the network device. The performance monitoring information includes at least one of the following: performance monitoring values ​​and performance monitoring auxiliary information. The performance monitoring values ​​include intermediate correlation indicators; for example, the performance monitoring values ​​include monitoring values ​​or monitoring results obtained after the terminal device performs performance monitoring, such as at least one of the first to third monitoring values ​​and first to third monitoring results described in Examples 3-1 and 3-2 below. The performance monitoring auxiliary information includes correlation information for performance monitoring; for example, the correlation information may include correlation information between the second output CSI and the first input CSI, where the second output CSI and the first input CSI correspond to the same measurement CSI-RS or the same CSI report.

[0272] The terminal device receives model management information sent by the network device.

[0273] Terminal devices can be monitored in various ways. The following examples 3-1 and 3-2 illustrate two methods for monitoring terminal device performance. It should be noted that examples 3-1 and 3-2 involve first to third monitoring values, first to third thresholds, and first to third monitoring results. These are unrelated to the first to third monitoring values, first to third thresholds, and first to third monitoring results in example 2.

[0274] Example 3-1:

[0275] In this embodiment, the first model is deployed on the terminal device side, the second model is deployed on the network device side, and the terminal device side may also store a fourth model. The fourth model is the same as the second model, or the fourth model and the second model are trained using the same dataset.

[0276] This embodiment includes the following stages:

[0277] Phase 1: The terminal device receives the second output CSI sent by the network device, as shown in Figure 10A. The second output CSI is the output value of the second model, and the input information of the second model is the second input CSI. The second input CSI is the result of the first output CSI being transmitted via the uplink channel, and the first output CSI is the result of the first model performing joint source-channel coding on the first input CSI. The first input CSI can be obtained from a reference signal sent by the network device.

[0278] In Phase Two, after receiving the second output CSI, the terminal device performs performance monitoring of the AI ​​model using the following steps.

[0279] The first step, as shown in Figure 10A, is for the terminal device to compare the first input CSI with the received second output CSI and determine the monitoring result based on the comparison result. Here, the input data of the second model is the second input CSI, which is the data transmitted through the uplink channel after the first output CSI; the first output CSI is the output value of the first model deployed by the terminal device, and the input data of this first model is the first input CSI.

[0280] One possible approach is to use the SGCS (SGatonic Response Center) between the first input CSI (Concurrent Input System Index) and the second output CSI (Concurrent Output System Index) to obtain the first monitoring value. The network device configures a first threshold and obtains a first monitoring result based on the first monitoring value and the first threshold. For example, if the first monitoring value is greater than or equal to the first threshold, the first monitoring result is that no first management operation is required; if the first monitoring value is less than the first threshold, the first monitoring result is that first management operation is required. In other words, the closer the SGCS is to 1, the better the performance.

[0281] Another possible approach is to use the difference between the first input CSI and the second output CSI to obtain the first monitoring value. The network device is configured with a first threshold, and the first monitoring result is obtained based on the first monitoring value and the first threshold. For example, if the first monitoring value is greater than the first threshold, the first monitoring result indicates that a first management operation is required; if the first monitoring value is less than or equal to the first threshold, the first monitoring result indicates that no first management operation is required. In other words, the larger the difference, the worse the performance.

[0282] Another possible approach is to measure the correlation between the first input CSI and the second output CSI to obtain the first monitoring value. The network device configures a first threshold and obtains a first monitoring result based on the first monitoring value and the first threshold. For example, if the first monitoring value is greater than or equal to the first threshold, the first monitoring result is that no first management operation is required; if the first monitoring value is less than the first threshold, the first monitoring result is that first management operation is required. That is, the higher the correlation, the better the performance.

[0283] If it is determined that the first management operation is required, it indicates that one or more of the first model, uplink channel, and second model within the dashed line in Figure 10A are causing poor performance. Then, the following second step can be performed to determine the reason for the need for the first management operation.

[0284] The second step, as shown in Figure 10B, involves the terminal device storing a fourth model for performance monitoring. This fourth model, along with the first model, assists in performance monitoring. The fourth model is the same as the second model, or both models are trained using the same dataset. The input data for the fourth model is the fourth input CSI, and the output data is the fourth output CSI. The fourth output CSI is the output data of the fourth model after transmission through uplink channels within the training dataset range; that is, the fourth output CSI considers the uplink channels within the training set range. The fourth output CSI is the inferred value of the fourth model. The terminal device compares the fourth output CSI with the first input CSI and determines the monitoring result based on the comparison result.

[0285] One possible approach is to use the SGCS (SGatonic Response Center) between the first input CSI and the fourth output CSI to obtain the second monitoring value. A second threshold is configured, and the second monitoring result is obtained based on the second monitoring value and the second threshold. For example, if the second monitoring value is greater than or equal to the second threshold, the second monitoring result indicates that no first management operation is required; if the second monitoring value is less than the second threshold, the second monitoring result indicates that a first management operation is required, and the uplink channel is the cause of the requirement. In other words, the closer the SGCS is to 1, the better the performance.

[0286] Another possible approach is to use the difference between the first input CSI and the fourth output CSI to obtain the second monitoring value. A second threshold is configured, and the second monitoring result is derived based on the second monitoring value and the second threshold. For example, if the second monitoring value is greater than or equal to the second threshold, the second monitoring result indicates that a first management operation is required, and the uplink channel is the cause of this requirement; if the second monitoring value is less than the second threshold, the second monitoring result indicates that no first management operation is required. In other words, the larger the difference, the worse the performance.

[0287] Another possible approach is to use the difference between the first input CSI and the fourth output CSI to obtain the second monitoring value. A second threshold is configured, and the second monitoring result is derived based on the second monitoring value and the second threshold. For example, if the second monitoring value is greater than or equal to the second threshold, the second monitoring result indicates that no first management operation is required; if the second monitoring value is less than the second threshold, the second monitoring result indicates that a first management operation is required, and the uplink channel is the cause of the requirement. In other words, higher correlation indicates better performance.

[0288] Using a fourth model for auxiliary performance monitoring can further identify whether poor performance is caused by the uplink channel, enabling more accurate model management and avoiding unnecessary model switching.

[0289] Phase 3: The terminal device sends the aforementioned monitoring values ​​and / or monitoring results to the network device. For example, the terminal device sends one or more of the aforementioned monitoring values ​​and / or monitoring results to the network device.

[0290] Phase Four: The network device generates model management information based on the monitoring values ​​and / or monitoring results received from the terminal device and sends this model management information to the terminal device. For example, the model management information may include: No first management operation is required. Alternatively, the model management information may include: First management operation is required. Or, the model management information may include: First management operation is required, and the uplink channel is the reason for requiring the first management operation. The model management information is then determined.

[0291] In one example, if the first monitoring result indicates that a first management operation is required, and the second monitoring result indicates that a first management operation is not required, then the uplink channel is determined to be the cause of poor performance, i.e., the uplink channel is the cause of requiring a first management operation. If both the first and second monitoring results indicate that a first management operation is required, then both the AI ​​model (including the first model and / or the second model) and the uplink channel are determined to be the causes of poor performance. It should be noted that when making the aforementioned judgment, the first prediction result and the second prediction result need to be measured using the same metric. For example, the first prediction result is determined based on the first monitoring value, and the second prediction result is determined based on the second monitoring value. Both the first and second monitoring values ​​are measured based on the SGCS of the relevant CSIs, or both are measured based on the difference between the relevant CSIs, or both are measured based on the correlation between the relevant CSIs.

[0292] The terminal device performs corresponding operations based on the model management information. Specifically, the first management operation may include at least one of the following: activating the AI ​​model, deactivating the AI ​​model, updating the AI ​​model, stopping the use of the AI ​​model for CSI feedback, and adjusting uplink resource allocation. In one example, if the uplink channel is the reason for performing the first management operation, then the first management operation may be adjusting uplink resource allocation.

[0293] Example 3-2:

[0294] In this embodiment, the first model is deployed on the terminal device side, the second model is deployed on the network device side, and the terminal device side may also store a fourth model. The fourth model is the same as the second model, or the fourth model and the second model are trained using the same dataset.

[0295] This embodiment includes the following stages:

[0296] In Phase 1, the terminal device compares the fourth output CSI with the second output CSI to obtain the monitoring results.

[0297] As shown in Figure 10A above, the second output CSI is the output value of the second model. The input information of the second model is the second input CSI, which is the result of the first output CSI being transmitted via the uplink channel. The first output CSI is the result of the first model performing joint source-channel coding on the first input CSI. The first input CSI can be obtained from a reference signal transmitted by a network device.

[0298] As shown in Figure 10B above, the input data of the fourth model is the fourth input CSI, and the output data of the fourth model is the fourth output CSI. The fourth output CSI is the output data of the fourth model after transmission through the uplink channel within the training dataset range; that is, the fourth output CSI considers the uplink channel within the training set range. The fourth output CSI is the inferred value of the fourth model.

[0299] One possible approach is to use the SGCS (SGatonic Response Center) between the second and fourth output CSIs to obtain the third monitoring value. A third threshold is configured, and the third monitoring result is derived based on the third monitoring value and the third threshold. For example, if the third monitoring value is greater than or equal to the third threshold, the third monitoring result indicates that no first management operation is required; if the third monitoring value is less than the third threshold, the third monitoring result indicates that a first management operation is required, and the uplink channel is the cause of this requirement. In other words, the closer the SGCS is to 1, the better the performance.

[0300] Another possible approach is to use the difference between the second and fourth output CSIs to obtain the third monitoring value. A third threshold is configured, and the third monitoring result is derived based on the third monitoring value and the third threshold. For example, if the third monitoring value is greater than or equal to the third threshold, the third monitoring result indicates that a first management operation is required, and the uplink channel is the cause of this requirement; if the third monitoring value is less than the third threshold, the third monitoring result indicates that no first management operation is required. In other words, the larger the difference, the worse the performance.

[0301] Another possible approach is to use the correlation between the second and fourth output CSIs to obtain the third monitoring value. A third threshold is configured, and the third monitoring result is derived based on the third monitoring value and the third threshold. For example, if the third monitoring value is greater than or equal to the third threshold, the third monitoring result indicates that no first management operation is required; if the third monitoring value is less than the third threshold, the third monitoring result indicates that a first management operation is required, and the uplink channel is the cause of this requirement. In other words, higher correlation indicates better performance.

[0302] Using a fourth model for auxiliary performance monitoring can further identify whether poor performance is caused by the uplink channel, enabling more accurate model management and avoiding unnecessary model switching. Furthermore, since only one judgment is required, this method is simpler than the method in Example 3-1.

[0303] Phase Two: The terminal device sends the aforementioned monitoring values ​​and / or monitoring results to the network device. For example, the terminal device sends one or more of the aforementioned monitoring values ​​and / or monitoring results to the network device.

[0304] Phase 3: The network device generates model management information based on the monitoring values ​​and / or monitoring results received from the terminal device and sends this model management information to the terminal device. For example, the model management information may include: No first management operation is required. Alternatively, the model management information may include: First management operation is required. Or, the model management information may include: First management operation is required, and the uplink channel is the reason for requiring the first management operation. The model management information is then determined.

[0305] The terminal device performs corresponding operations based on the model management information. Specifically, the first management operation may include at least one of the following: activating the AI ​​model, deactivating the AI ​​model, updating the AI ​​model, stopping the use of the AI ​​model for CSI feedback, and adjusting uplink resource allocation. In one example, if the uplink channel is the reason for performing the first management operation, then the first management operation may be adjusting uplink resource allocation.

[0306] In some implementations, the performance monitoring metrics for CSI feedback based on joint source-channel coding can be at least one of the following:

[0307] Intermediate key performance indicators (KPIs) include, for example, the squared cosine similarity (SGCS). Terminal devices can report target input CSIs, or network devices can send target output CSIs, enabling performance monitoring of either the network or terminal devices.

[0308] The final KPIs, such as: Block Error Rate (BLER), assumed bit error rate, ACK / NACK.

[0309] The monitoring values ​​in Example 2 can be replaced with the description of the final KPI;

[0310] The distribution of input or output data in a two-sided model, such as the data drift between training and output data. In this example, a comparison between outputs is used. For instance, comparing the mathematical distribution between the output of the training dataset and the actual model inference values; if the variance between the two is within a certain threshold, the model performance is considered good; if the variance is greater than a certain threshold, the model performance is considered poor.

[0311] In some implementations, when performance monitoring methods determine that the poor performance is caused by the uplink channel, there are three possibilities: possibility 1 is to collect data and retrain the first model; possibility 2 is to recollect uplink channel data and update the first model; possibility 3 is to adjust the uplink resource allocation and allocate more uplink resources to counteract the impact of the uplink channel.

[0312] In some examples, multiple thresholds can be configured for each monitoring method, such as threshold 1, threshold 2, and threshold 3. The monitored value is compared to different thresholds, resulting in different model management behaviors. For example, if SGCS is less than threshold 1, it falls back to traditional communication without using the AI ​​model; if SGCS is between threshold 1 and threshold 2, the AI ​​model is updated; if SGCS is between threshold 2 and threshold 3, resource allocation is performed; if SGCS is greater than threshold 3, model management is not required. This approach allows for more precise model management.

[0313] This application provides a terminal device. FIG11 is a schematic block diagram of a terminal device 1100 according to an embodiment of this application. The terminal device 1100 may include:

[0314] The first transceiver module 1110 is used to receive a reference signal sent by a network device and send at least one of a first input CSI, performance monitoring information, and performance monitoring results to the network device; wherein the first input CSI is determined based on the reference signal.

[0315] This application provides a terminal device. Figure 12 is a schematic block diagram of a terminal device 1200 according to an embodiment of this application. The terminal device 1200 further includes:

[0316] The first processing module 1220 is configured to perform a first management operation, which includes at least one of the following:

[0317] Switching between AI models;

[0318] Activate the AI ​​model;

[0319] Deactivate the AI ​​model;

[0320] Update the AI ​​model;

[0321] Stop using AI models for CSI feedback;

[0322] Adjust the allocation of upstream resources.

[0323] In some implementations, the AI ​​model includes at least one of a first model and a second model; wherein,

[0324] The first model is deployed on the terminal device and is used to perform joint source-channel coding on the CSI feedback sent by the terminal device;

[0325] The second model is deployed on network devices to perform joint source-channel decoding on the CSI feedback received by the network devices.

[0326] In some implementations, the first processing module 1220 is used for:

[0327] Receive model management information sent by network devices;

[0328] Based on the management information from this model, perform the first management operation.

[0329] In some embodiments, the first transceiver module 1110 is further configured to send at least one of a first input CSI and a first conventional CSI to the network device, wherein,

[0330] The first input CSI is the input content of the first model when performing joint source-channel feedback;

[0331] The first traditional CSI includes: CSI measured and reported using a preset method.

[0332] In some implementations, the first processing module 1220 is further configured to determine the performance monitoring results of the AI ​​model;

[0333] The first transceiver module 1110 is also used to send the performance monitoring results of the AI ​​model to the network device, so that the network device can determine the model management information.

[0334] In some implementations, the first processing module 1220 is used for:

[0335] Determine the performance monitoring results of the AI ​​model;

[0336] Based on the performance monitoring results of the AI ​​model, the first management operation is performed.

[0337] In some implementations, the first transceiver module 1110 is used to receive a second output CSI sent by the network device;

[0338] The first processing module 1110 is used to determine the performance monitoring results of the AI ​​model based on the second output CSI.

[0339] The second output CSI includes the result of the second model performing joint source-channel decoding on the second input CSI.

[0340] The second input CSI includes: the result of the first output CSI after transmission via the uplink channel;

[0341] The first output CSI includes the result of the first model performing joint source-channel coding on the first input CSI.

[0342] In some implementations, the first processing module 1220 is used for at least one of the following:

[0343] The first monitoring value is determined based on the first input CSI and the second output CSI;

[0344] The second monitoring value is determined based on the first input CSI and the fourth output CSI;

[0345] The fourth output CSI includes the result of the fourth model performing joint source-channel decoding on the fourth input CSI.

[0346] The fourth model is the same as the second model, or the fourth model and the second model were trained on the same dataset;

[0347] The fourth input CSI is the output data transmitted through the uplink channel within the training set range.

[0348] In some implementations, the first processing module 1220 is used to determine the first monitoring value based on the SGCS, difference, or correlation between the first input CSI and the second output CSI.

[0349] In some implementations, the first processing module 1220 is used to determine the second monitoring value based on the SGCS, difference, or correlation between the first input CSI and the fourth output CSI.

[0350] In some implementations, the first processing module 1220 is used for at least one of the following:

[0351] Based on the first monitoring value and the first threshold, the first monitoring result is determined;

[0352] The second monitoring result is determined based on the second monitoring value and the second threshold.

[0353] In some implementations, the first processing module 1220 is further configured to determine, based on at least one of the first monitoring result and the second monitoring result, whether a first management operation is required and / or the reason for requiring the first management operation.

[0354] In some implementations, the first processing module 1220 is used for:

[0355] The third monitoring value is determined based on the second and fourth CSI outputs;

[0356] The fourth output CSI includes the result of the fourth model performing joint source-channel decoding on the fourth input CSI.

[0357] The fourth model is the same as the second model, or the fourth model and the second model were trained on the same dataset;

[0358] The fourth input CSI is the output data transmitted through the uplink channel within the training set range.

[0359] In some implementations, the first processing module 1220 is used to determine a third monitoring value based on the SGCS, difference, or correlation between the second output CSI and the fourth output CSI.

[0360] In some implementations, the first processing module 1220 is further configured to determine a third monitoring result based on the third monitoring value.

[0361] In some implementations, the first processing module 1220 is further configured to determine, based on the third monitoring result, whether a first management operation is required and / or the reason for requiring the first management operation.

[0362] In some implementations, the performance monitoring information includes at least one of performance monitoring values ​​and performance monitoring auxiliary information.

[0363] The terminal devices 1100 and 1200 in this application embodiment can implement the corresponding functions of the terminal devices in the aforementioned method embodiments. The processes, functions, implementation methods, and beneficial effects of each module (sub-module, unit, or component, etc.) in terminal devices 1100 and 1200 can be found in the corresponding descriptions in the above method embodiments, and will not be repeated here. It should be noted that the functions described in the various modules (sub-modules, units, or components, etc.) of terminal devices 1100 and 1200 in this application embodiment can be implemented by different modules (sub-modules, units, or components, etc.) or by the same module (sub-module, unit, or component, etc.).

[0364] Figure 13 is a schematic block diagram of a network device 1300 according to an embodiment of the present application. The network device 1300 may include:

[0365] The second transceiver module 1310 is used to send a reference signal to the terminal device and receive at least one of the first input CSI, performance monitoring information and performance monitoring results sent by the terminal device; wherein the first input CSI is determined based on the reference signal.

[0366] This application provides a network device. Figure 14 is a schematic block diagram of a network device 1400 according to an embodiment of this application. The network device 1400 further includes:

[0367] The second processing module 1420 is used to determine model management information;

[0368] The second transceiver module 1310 is also used to send model management information to the terminal device;

[0369] The model management information is used to instruct the terminal device to perform a first management operation, which includes at least one of the following:

[0370] Switching between AI models;

[0371] Activate the AI ​​model;

[0372] Deactivate the AI ​​model;

[0373] Update the AI ​​model;

[0374] Stop using AI models for CSI feedback;

[0375] Adjust the allocation of upstream resources.

[0376] In some implementations, the AI ​​model includes at least one of a first model and a second model; wherein,

[0377] The first model is deployed on the terminal device and is used to perform joint source-channel coding on the CSI feedback sent by the terminal device;

[0378] The second model is deployed on network devices to perform joint source-channel decoding on the CSI feedback received by the network devices.

[0379] In some implementations, the second transceiver module 1310 is further configured to receive a first conventional CSI sent by the terminal device;

[0380] The second processing module 1420 is further configured to determine, based on at least one of the first input CSI and the first conventional CSI, the performance monitoring result of the AI ​​model, wherein the performance monitoring result of the AI ​​model is used to determine model management information; wherein...

[0381] The first input CSI is the input content of the first model when performing joint source-channel feedback;

[0382] The first traditional CSI includes CSI that is measured and reported using a preset method.

[0383] In some implementations, the second processing module 1420 is used for:

[0384] The first monitoring value is determined based on the first input CSI and the second output CSI;

[0385] The second monitoring value is determined based on the first input CSI and the third output CSI;

[0386] The second output CSI includes the result of the second model performing joint source-channel decoding on the second input CSI.

[0387] The second input CSI includes: the result of the first output CSI after transmission via the uplink channel;

[0388] The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI;

[0389] The third output CSI includes the result of the second model performing joint source-channel decoding on the third input CSI.

[0390] The third input CSI includes: the result output by the third model after performing joint source-channel decoding on the input content;

[0391] The third model is the same as the first model, or the third model and the first model are trained using the same dataset; the input content is the output data transmitted through the uplink channel within the training set range.

[0392] In some implementations, the second processing module 1420 is used to determine the first monitoring value based on the SGCS, difference, or correlation between the first input CSI and the second output CSI.

[0393] In some implementations, the second processing module 1420 is used to determine the second monitoring value based on the SGCS, difference, or correlation between the first input CSI and the third output CSI.

[0394] In some implementations, the second processing module 1420 is used for at least one of the following:

[0395] Based on the first monitoring value and the first threshold, the first monitoring result is determined;

[0396] The second monitoring result is determined based on the second monitoring value and the second threshold.

[0397] In some embodiments, the second processing module 1420 is further configured to determine, based on at least one of the first monitoring result and the second monitoring result, whether a first management operation is required and / or the reason for requiring the first management operation.

[0398] In some implementations, the second processing module 1420 is used for:

[0399] The third monitoring value is determined based on the second and third CSI outputs;

[0400] The second output CSI includes the result of the second model performing joint source-channel decoding on the second input CSI.

[0401] The second input CSI includes: the result of the first output CSI after transmission via the uplink channel;

[0402] The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI;

[0403] The third output CSI includes the result of the second model performing joint source-channel decoding on the third input CSI.

[0404] The third input CSI includes: the result output by the third model after performing joint source-channel decoding on the input content;

[0405] The third model is the same as the first model, or the third model and the first model are trained using the same dataset; the input content is the output data transmitted through the uplink channel within the training set range.

[0406] In some implementations, the second processing module 1420 is used to determine a third monitoring value based on the SGCS, difference, or correlation between the second output CSI and the third output CSI.

[0407] In some implementations, the second processing module 1420 is further configured to determine a third monitoring result based on the third monitoring value and the third threshold.

[0408] In some implementations, the second processing module 1420 is further configured to determine, based on the third monitoring result, whether a first management operation is required and / or the reason for requiring the first management operation.

[0409] In some implementations, the second processing module 1420 is used to determine a fourth monitoring value based on the first input CSI and the first conventional CSI.

[0410] In some implementations, the second processing module 1420 is used to determine a fourth monitoring value based on the SGCS, difference, or correlation between the first input CSI and the first conventional CSI.

[0411] In some embodiments, the second processing module 1420 is further configured to:

[0412] The fourth monitoring result is determined based on the fourth monitoring value and the fourth threshold;

[0413] Based on the results of this fourth monitoring, determine whether a first management operation is required and / or the reasons for requiring a first management operation.

[0414] In some embodiments, the second transceiver module 1310 is further configured to:

[0415] Send a second output CSI to the terminal device, which is used by the terminal device to determine the performance monitoring results of the AI ​​model;

[0416] Receive performance monitoring results of the AI ​​model sent by the terminal device;

[0417] The second output CSI includes the result of the second model performing joint source-channel decoding on the second input CSI.

[0418] The second input CSI includes: the result of the first output CSI after transmission via the uplink channel;

[0419] The first output CSI includes the result of the first model performing joint source-channel coding on the first input CSI.

[0420] In some implementations, the performance monitoring information includes at least one of performance monitoring values ​​and performance monitoring auxiliary information.

[0421] Network devices 1300 and 1400 in this application embodiment can implement the corresponding functions of the network devices in the aforementioned method embodiments. The processes, functions, implementation methods, and beneficial effects of each module (sub-module, unit, or component, etc.) in network devices 1300 and 1400 can be found in the corresponding descriptions in the above method embodiments, and will not be repeated here. It should be noted that the functions described for each module (sub-module, unit, or component, etc.) in network devices 1300 and 1400 in this application embodiment can be implemented by different modules (sub-modules, units, or components, etc.) or by the same module (sub-module, unit, or component, etc.).

[0422] Figure 15 is a schematic structural diagram of a communication device 1500 according to an embodiment of this application. The communication device 1500 includes a processor 1510, which can call and run computer programs from memory to enable the communication device 1500 to implement the methods in the embodiments of this application.

[0423] In one embodiment, the communication device 1500 may further include a memory 1520. The processor 1510 can retrieve and run computer programs from the memory 1520 to enable the communication device 1500 to implement the methods described in the embodiments of this application.

[0424] The memory 1520 can be a separate device independent of the processor 1510, or it can be integrated into the processor 1510.

[0425] In one embodiment, the communication device 1500 may further include a transceiver 1530, and the processor 1510 may control the transceiver 1530 to communicate with other devices. Specifically, it may send information or data to other devices or receive information or data sent by other devices.

[0426] The transceiver 1530 may include a transmitter and a receiver. The transceiver 1530 may further include an antenna, and the number of antennas may be one or more.

[0427] In one embodiment, the communication device 1500 may be a network device in the embodiments of this application, and the communication device 1500 may implement the corresponding processes implemented by the network device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0428] In one embodiment, the communication device 1500 may be a terminal device in the embodiments of this application, and the communication device 1500 may implement the corresponding processes implemented by the terminal device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0429] Figure 16 is a schematic structural diagram of a chip 1600 according to an embodiment of this application. The chip 1600 includes a processor 1610, which can call and run computer programs from memory to implement the methods in the embodiments of this application.

[0430] In one embodiment, chip 1600 may further include memory 1620. Processor 1610 can retrieve and run computer programs from memory 1620 to implement the methods executed by a terminal device or network device in this embodiment.

[0431] The memory 1620 can be a separate device independent of the processor 1610, or it can be integrated into the processor 1610.

[0432] In one embodiment, the chip 1600 may further include an input interface 1630. The processor 1610 can control the input interface 1630 to communicate with other devices or chips; specifically, it can acquire information or data sent by other devices or chips.

[0433] In one embodiment, the chip 1600 may further include an output interface 1640. The processor 1610 can control the output interface 1640 to communicate with other devices or chips; specifically, it can output information or data to other devices or chips.

[0434] In one implementation, the chip can be applied to the network device in the embodiments of this application, and the chip can implement the corresponding processes implemented by the network device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0435] In one embodiment, the chip can be applied to the terminal device in the embodiments of this application, and the chip can implement the corresponding processes implemented by the terminal device in the various methods of the embodiments of this application. For the sake of brevity, it will not be described in detail here.

[0436] The chips used in network equipment and terminal equipment can be the same chip or different chips.

[0437] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0438] The processors mentioned above can be general-purpose processors, digital signal processors (DSPs), field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), or other programmable logic devices, transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processors mentioned above can be microprocessors or any conventional processor.

[0439] The aforementioned memory can be volatile memory or non-volatile memory, or a combination of both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM).

[0440] It should be understood that the above-described memory is exemplary and not a limiting description. For example, the memory in the embodiments of this application may also be static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link dynamic random access memory (SLDRAM), and direct memory bus RAM (DR RAM), etc. That is to say, the memory in the embodiments of this application is intended to include, but is not limited to, these and any other suitable types of memory.

[0441] Figure 17 is a schematic block diagram of a communication system 1700 according to an embodiment of this application. The communication system 1700 includes a terminal device 1710 and a network device 1720.

[0442] A terminal device, comprising:

[0443] The first transceiver module is configured to receive a reference signal sent by a network device and send at least one of a first input CSI, performance monitoring information, and performance monitoring results to the network device; wherein the first input CSI is determined based on the reference signal.

[0444] A network device, comprising:

[0445] The second transceiver module is used to send a reference signal to the terminal device and receive at least one of the first input CSI, performance monitoring information, and performance monitoring results sent by the terminal device; wherein the first input CSI is determined based on the reference signal.

[0446] Specifically, the terminal device 1710 can be used to implement the corresponding functions implemented by the terminal device in the above method, and the network device 1720 can be used to implement the corresponding functions implemented by the network device in the above method. For the sake of brevity, further details are omitted here.

[0447] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions 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., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

[0448] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes 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.

[0449] Those skilled in the art will clearly 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.

[0450] 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 control method, comprising: The terminal device receives reference signals sent by the network device; The terminal device sends at least one of the following to the network device: first input channel state information (CSI), performance monitoring information, and performance monitoring results; The first input CSI is determined based on the reference signal.

2. The method according to claim 1, further comprising: The terminal device performs a first management operation, which includes at least one of the following: Switching between artificial intelligence (AI) models; Activate the AI ​​model; Deactivate the AI ​​model; Update the AI ​​model; Stop using AI models for Channel State Information (CSI) feedback; Adjust the allocation of upstream resources.

3. The method according to claim 2, wherein, The AI ​​model includes at least one of the first model and the second model; wherein... The first model is deployed on a terminal device and is used to perform joint source-channel coding on the CSI feedback sent by the terminal device; The second model is deployed on network devices and is used to perform joint source-channel decoding on the CSI feedback received by the network devices.

4. The method according to claim 3, wherein, The terminal device performs a first management operation, including: The terminal device receives model management information sent by the network device; The terminal device performs the first management operation based on the model management information.

5. The method according to claim 4, further comprising, before the terminal device receives the model management information sent by the network device: The terminal device sends a first traditional CSI to the network device, wherein... The first input CSI is the input content of the first model when performing joint source-channel feedback; The first conventional CSI includes: CSI measured and reported using a preset method.

6. The method according to claim 4, wherein, Before the terminal device receives the model management information sent by the network device, it also includes: The terminal device determines the performance monitoring results of the AI ​​model; The terminal device sends the performance monitoring results of the AI ​​model to the network device, which is used by the network device to determine the model management information.

7. The method according to claim 3, wherein, The terminal device performs a first management operation, including: The terminal device determines the performance monitoring results of the AI ​​model; The terminal device performs the first management operation based on the performance monitoring results of the AI ​​model.

8. The method according to claim 6 or 7, wherein, The terminal device determines the performance monitoring results of the AI ​​model, including: The terminal device receives the second output CSI sent by the network device; The terminal device determines the performance monitoring result of the AI ​​model based on the second output CSI. Wherein, the second output CSI includes: the result output by the second model after performing joint source-channel decoding on the second input CSI; The second input CSI includes: the result of the first output CSI transmitted via the uplink channel; The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI.

9. The method according to claim 8, wherein, The terminal device determines the performance monitoring result of the AI ​​model based on the second output CSI, including at least one of the following: The terminal device determines the first monitoring value based on the first input CSI and the second output CSI; The terminal device determines the second monitoring value based on the first input CSI and the fourth output CSI; The fourth output CSI includes: the result output by the fourth model after performing joint source-channel decoding on the fourth input CSI; The fourth model is the same as the second model, or the fourth model and the second model are trained on the same dataset; The fourth input CSI is the output data transmitted through the uplink channel within the training set range.

10. The method according to claim 9, wherein, The terminal device determines a first monitoring value based on the first input CSI and the second output CSI, including: The terminal device determines the first monitoring value based on the generalized cosine similarity (SGCS), difference, or correlation between the first input CSI and the second output CSI.

11. The method according to claim 9, wherein, The terminal device determines a second monitoring value based on the first input CSI and the fourth output CSI, including: The terminal device determines the second monitoring value based on the SGCS, difference, or correlation between the first input CSI and the fourth output CSI.

12. The method according to any one of claims 9-11, wherein, The terminal device determines the performance monitoring result of the AI ​​model based on the second output CSI, and further includes at least one of the following: The terminal device determines a first monitoring result based on the first monitoring value and the first threshold. The terminal device determines the second monitoring result based on the second monitoring value and the second threshold.

13. The method according to claim 12, wherein, The terminal device determines the performance monitoring result of the AI ​​model based on the second output CSI, and also includes: The terminal device determines whether a first management operation is required and / or the reason for requiring a first management operation based on at least one of the first monitoring result and the second monitoring result.

14. The method according to claim 8, wherein, The terminal device determines the performance monitoring result of the AI ​​model based on the second output CSI, including: The terminal device determines the third monitoring value based on the second output CSI and the fourth output CSI; The fourth output CSI includes: the result output by the fourth model after performing joint source-channel decoding on the fourth input CSI; The fourth model is the same as the second model, or the fourth model and the second model are trained on the same dataset; The fourth input CSI is the output data transmitted through the uplink channel within the training set range.

15. The method according to claim 14, wherein, The terminal device determines a third monitoring value based on the second and fourth output CSIs, including: The terminal device determines the third monitoring value based on the SGCS, difference, or correlation between the second output CSI and the fourth output CSI.

16. The method according to claim 14 or 15, wherein, The terminal device determines the performance monitoring result of the AI ​​model based on the second output CSI, and also includes: The terminal device determines the third monitoring result based on the third monitoring value.

17. The method according to claim 16, wherein, The terminal device determines the performance monitoring result of the AI ​​model based on the second output CSI, and also includes: Based on the third monitoring result, the terminal device determines whether a first management operation is required and / or the reason for requiring the first management operation.

18. The method according to any one of claims 1-17, wherein, The performance monitoring information includes at least one of performance monitoring values ​​and performance monitoring auxiliary information.

19. A control method, comprising: Network devices send reference signals to terminal devices; The network device receives at least one of the first input CSI, performance monitoring information, and performance monitoring results sent by the terminal device; The first input CSI is determined based on the reference signal.

20. The method of claim 19, further comprising: Network devices determine model management information; The network device sends the model management information to the terminal device; The model management information is used to instruct the terminal device to perform a first management operation, the first management operation including at least one of the following: Switching between AI models; Activate the AI ​​model; Deactivate the AI ​​model; Update the AI ​​model; Stop using AI models for CSI feedback; Adjust the allocation of upstream resources.

21. The method according to claim 20, wherein, The AI ​​model includes at least one of the first model and the second model; wherein... The first model is deployed on a terminal device and is used to perform joint source-channel coding on the CSI feedback sent by the terminal device; The second model is deployed on network devices and is used to perform joint source-channel decoding on the CSI feedback received by the network devices.

22. The method of claim 21, further comprising: The network device receives a first conventional CSI sent by the terminal device; The network device determines the performance monitoring result of the AI ​​model based on at least one of the first input CSI and the first traditional CSI, wherein the performance monitoring result of the AI ​​model is used to determine the model management information; wherein... The first input CSI is the input content of the first model when performing joint source-channel feedback; The first conventional CSI includes CSI measured and reported using a preset method.

23. The method according to claim 22, wherein, The network device determines the performance monitoring results of the AI ​​model based on at least one of the first input CSI and the first conventional CSI, including: The network device determines a first monitoring value based on the first input CSI and the second output CSI; The network device determines the second monitoring value based on the first input CSI and the third output CSI; Wherein, the second output CSI includes: the result output by the second model after performing joint source-channel decoding on the second input CSI; The second input CSI includes: the result of the first output CSI transmitted via the uplink channel; The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI; The third output CSI includes: the result output by the second model after performing joint source-channel decoding on the third input CSI; The third input CSI includes: the result output by the third model after performing joint source-channel decoding on the input content; The third model is the same as the first model, or the third model and the first model are trained using the same dataset; the input content is the output data transmitted through the uplink channel within the training set range.

24. The method according to claim 23, wherein, The network device determines a first monitoring value based on the first input CSI and the second output CSI, including: The network device determines the first monitoring value based on the SGCS, difference, or correlation between the first input CSI and the second output CSI.

25. The method according to claim 23, wherein, The network device determines a second monitoring value based on the first input CSI and the third output CSI, including: The network device determines the second monitoring value based on the SGCS, difference, or correlation between the first input CSI and the third output CSI.

26. The method according to any one of claims 23-25, wherein, The network device determines the performance monitoring results of the AI ​​model based on at least one of the first input CSI and the first conventional CSI, and further includes at least one of the following: The network device determines a first monitoring result based on the first monitoring value and the first threshold. The network device determines the second monitoring result based on the second monitoring value and the second threshold.

27. The method according to claim 26, wherein, The network device determines the performance monitoring results of the AI ​​model based on at least one of the first input CSI and the first conventional CSI, and further includes: The network device determines whether a first management operation is required and / or the reason for requiring a first management operation based on at least one of the first monitoring result and the second monitoring result.

28. The method according to claim 22, wherein, The network device determines the performance monitoring result of the AI ​​model based on at least one of the first input CSI and the first conventional CSI, including: The network device determines the third monitoring value based on the second output CSI and the third output CSI; Wherein, the second output CSI includes: the result output by the second model after performing joint source-channel decoding on the second input CSI; The second input CSI includes: the result of the first output CSI transmitted via the uplink channel; The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI; The third output CSI includes: the result output by the second model after performing joint source-channel decoding on the third input CSI; The third input CSI includes: the result output by the third model after performing joint source-channel decoding on the input content; The third model is the same as the first model, or the third model and the first model are trained using the same dataset; the input content is the output data transmitted through the uplink channel within the training set range.

29. The method according to claim 28, wherein, The network device determines a third monitoring value based on the second output CSI and the third output CSI, including: The network device determines the third monitoring value based on the SGCS, difference, or correlation between the second output CSI and the third output CSI.

30. The method according to claim 28 or 29, wherein, The network device determines the performance monitoring result of the AI ​​model based on at least one of the first input CSI and the first conventional CSI, and further includes: The network device determines the third monitoring result based on the third monitoring value and the third threshold.

31. The method according to claim 30, wherein, The network device determines the performance monitoring result of the AI ​​model based on at least one of the first input CSI and the first conventional CSI, and further includes: Based on the third monitoring result, the network device determines whether a first management operation is required and / or the reason for requiring a first management operation.

32. The method according to claim 22, wherein, The network device determines the performance monitoring result of the AI ​​model based on at least one of the first input CSI and the first conventional CSI, including: The network device determines a fourth monitoring value based on the first input CSI and the first conventional CSI.

33. The method according to claim 32, wherein, The network device determines a fourth monitoring value based on the first input CSI and the first traditional CSI, including: The network device determines the fourth monitoring value based on the SGCS, difference, or correlation between the first input CSI and the first conventional CSI.

34. The method according to claim 33, wherein, The network device determines the performance monitoring result of the AI ​​model based on at least one of the first input CSI and the first conventional CSI, and further includes: The network device determines the fourth monitoring result based on the fourth monitoring value and the fourth threshold. Based on the fourth monitoring result, the network device determines whether a first management operation is required and / or the reason for requiring a first management operation.

35. The method of claim 21, further comprising: The network device sends a second output CSI to the terminal device, and the second output CSI is used by the terminal device to determine the performance monitoring result of the AI ​​model. The network device receives the performance monitoring results of the AI ​​model sent by the terminal device; in, The second output CSI includes: the result output by the second model after performing joint source-channel decoding on the second input CSI; The second input CSI includes: the result of the first output CSI transmitted via the uplink channel; The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI.

36. The method according to any one of claims 19-35, wherein, The performance monitoring information includes at least one of performance monitoring values ​​and performance monitoring auxiliary information.

37. A terminal device, comprising: The first transceiver module is used to receive reference signals sent by the network device and send a first input CSI and performance monitoring data to the network device. At least one of the information and performance monitoring results; wherein the first input CSI is determined based on the reference signal.

38. The terminal device according to claim 37, further comprising: A first processing module is configured to perform a first management operation, wherein the first management operation includes at least one of the following: Switching between AI models; Activate the AI ​​model; Deactivate the AI ​​model; Update the AI ​​model; Stop using AI models for CSI feedback; Adjust the allocation of upstream resources.

39. The terminal device according to claim 37, wherein, The AI ​​model includes at least one of the first model and the second model; wherein... The first model is deployed on a terminal device and is used to perform joint source-channel coding on the CSI feedback sent by the terminal device; The second model is deployed on network devices and is used to perform joint source-channel decoding on the CSI feedback received by the network devices.

40. The terminal device according to claim 39, wherein, The first processing module is used for: Receive model management information sent by network devices; Based on the model management information, perform the first management operation.

41. The terminal device according to claim 40, wherein, The first transceiver module is further configured to send a first transmission to the network device, wherein, The first input CSI is the input content of the first model when performing joint source-channel feedback; The first conventional CSI includes: CSI measured and reported using a preset method.

42. The terminal device according to claim 40, wherein, The first processing module is also used to determine the performance monitoring results of the AI ​​model; The first transceiver module is further configured to send the performance monitoring results of the AI ​​model to the network device, so that the network device can determine the model management information.

43. The terminal device according to claim 38, wherein the first processing module is configured to: Determine the performance monitoring results of the AI ​​model; Based on the performance monitoring results of the AI ​​model, the first management operation is performed.

44. The terminal device according to claim 42 or 43, wherein, The first transceiver module is used to receive the second output CSI sent by the network device; The first processing module is used to determine the performance monitoring result of the AI ​​model based on the second output CSI; Wherein, the second output CSI includes: the result output by the second model after performing joint source-channel decoding on the second input CSI; The second input CSI includes: the result of the first output CSI transmitted via the uplink channel; The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI.

45. The terminal device according to claim 44, wherein, The first processing module is used for at least one of the following: Based on the first input CSI and the second output CSI, a first monitoring value is determined; The second monitoring value is determined based on the first input CSI and the fourth output CSI; The fourth output CSI includes: the result output by the fourth model after performing joint source-channel decoding on the fourth input CSI; The fourth model is the same as the second model, or the fourth model and the second model are trained on the same dataset; The fourth input CSI is the output data transmitted through the uplink channel within the training set range.

46. ​​The terminal device according to claim 45, wherein, The first processing module is used to determine the first monitoring value based on the SGCS, difference, or correlation between the first input CSI and the second output CSI.

47. The terminal device according to claim 45, wherein, The first processing module is used to determine the second monitoring value based on the SGCS, difference, or correlation between the first input CSI and the fourth output CSI.

48. The terminal device according to any one of claims 45-47, wherein, The first processing module is used for at least one of the following: Based on the first monitoring value and the first threshold, a first monitoring result is determined; Based on the second monitoring value and the second threshold, a second monitoring result is determined.

49. The terminal device according to claim 48, wherein, The first processing module is further configured to determine, based on at least one of the first monitoring result and the second monitoring result, whether a first management operation is required and / or the reason for requiring a first management operation.

50. The terminal device according to claim 44, wherein, The first processing module is used for: The third monitoring value is determined based on the second and fourth output CSIs; The fourth output CSI includes: the result output by the fourth model after performing joint source-channel decoding on the fourth input CSI; The fourth model is the same as the second model, or the fourth model and the second model are trained on the same dataset; The fourth input CSI is the output data transmitted through the uplink channel within the training set range.

51. The terminal device according to claim 50, wherein, The first processing module is used to determine the third monitoring value based on the SGCS, difference, or correlation between the second output CSI and the fourth output CSI.

52. The terminal device according to claim 50 or 51, wherein, The first processing module is further configured to determine a third monitoring result based on the third monitoring value.

53. The terminal device according to claim 52, wherein, The first processing module is further configured to determine, based on the third monitoring result, whether a first management operation is required and / or the reason for requiring a first management operation.

54. The terminal device according to any one of claims 37-53, wherein, The performance monitoring information includes at least one of performance monitoring values ​​and performance monitoring auxiliary information.

55. A network device, comprising: The second transceiver module is used to send a reference signal to the terminal device and receive at least one of the first input CSI, performance monitoring information, and performance monitoring results sent by the terminal device; wherein the first input CSI is determined based on the reference signal.

56. The network device according to claim 55, further comprising: The second processing module is used to determine model management information; The second transceiver module is also used to send the model management information to the terminal device; The model management information is used to instruct the terminal device to perform a first management operation, the first management operation including at least one of the following: Switching between AI models; Activate the AI ​​model; Deactivate the AI ​​model; Update the AI ​​model; Stop using AI models for CSI feedback; Adjust the allocation of upstream resources.

57. The network device according to claim 55, wherein, The AI ​​model includes at least one of the first model and the second model; wherein... The first model is deployed on a terminal device and is used to perform joint source-channel coding on the CSI feedback sent by the terminal device; The second model is deployed on network devices and is used to perform joint source-channel decoding on the CSI feedback received by the network devices.

58. The network device according to claim 57, wherein the second transceiver module is further configured to receive a first conventional CSI sent by the terminal device; The second processing module is further configured to determine, based on at least one of the first input CSI and the first conventional CSI, a performance monitoring result of the AI ​​model, wherein the performance monitoring result of the AI ​​model is used to determine the model management information; wherein, The first input CSI is the input content of the first model when performing joint source-channel feedback; The first conventional CSI includes CSI measured and reported using a preset method.

59. The network device according to claim 58, wherein, The second processing module is used for: The first monitoring value is determined based on the first input CSI and the second output CSI; The second monitoring value is determined based on the first input CSI and the third output CSI; Wherein, the second output CSI includes: the result output by the second model after performing joint source-channel decoding on the second input CSI; The second input CSI includes: the result of the first output CSI transmitted via the uplink channel; The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI; The third output CSI includes: the result output by the second model after performing joint source-channel decoding on the third input CSI; The third input CSI includes: the result output by the third model after performing joint source-channel decoding on the input content; The third model is the same as the first model, or the third model and the first model are trained using the same dataset; the input content is the output data transmitted through the uplink channel within the training set range.

60. The network device according to claim 59, wherein, The second processing module is used to determine the first monitoring value based on the SGCS, difference, or correlation between the first input CSI and the second output CSI.

61. The network device according to claim 59, wherein, The second processing module is used to determine the second monitoring value based on the SGCS, difference, or correlation between the first input CSI and the third output CSI.

62. The network device according to any one of claims 59-61, wherein, The second processing module is used for at least one of the following: Based on the first monitoring value and the first threshold, a first monitoring result is determined; Based on the second monitoring value and the second threshold, a second monitoring result is determined.

63. The network device according to claim 62, wherein, The second processing module is further configured to determine, based on at least one of the first monitoring result and the second monitoring result, whether a first management operation is required and / or the reason for requiring a first management operation.

64. The network device according to claim 58, wherein, The second processing module is used for: Based on the second and third CSI outputs, a third monitoring value is determined; Wherein, the second output CSI includes: the result output by the second model after performing joint source-channel decoding on the second input CSI; The second input CSI includes: the result of the first output CSI transmitted via the uplink channel; The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI; The third output CSI includes: the result output by the second model after performing joint source-channel decoding on the third input CSI; The third input CSI includes: the result output by the third model after performing joint source-channel decoding on the input content; The third model is the same as the first model, or the third model and the first model are trained using the same dataset; the input content is the output data transmitted through the uplink channel within the training set range.

65. The network device according to claim 64, wherein, The second processing module is used to determine the third monitoring value based on the SGCS, difference, or correlation between the second output CSI and the third output CSI.

66. The network device according to claim 64 or 65, wherein, The second processing module is further configured to determine a third monitoring result based on the third monitoring value and the third threshold.

67. The network device according to claim 66, wherein, The second processing module is further configured to determine, based on the third monitoring result, whether a first management operation is required and / or the reason for requiring a first management operation.

68. The network device according to claim 58, wherein, The second processing module is used to determine a fourth monitoring value based on the first input CSI and the first conventional CSI.

69. The network device according to claim 68, wherein, The second processing module is used to determine the fourth monitoring value based on the SGCS, difference, or correlation between the first input CSI and the first conventional CSI.

70. The network device according to claim 69, wherein, The second processing module is also used for: Based on the fourth monitoring value and the fourth threshold, the fourth monitoring result is determined; Based on the fourth monitoring result, determine whether a first management operation is required and / or the reason for requiring a first management operation.

71. The network device according to claim 57, wherein the second transceiver module is further configured to: Send a second output CSI to the terminal device, the second output CSI being used by the terminal device to determine the performance monitoring result of the AI ​​model; Receive the performance monitoring results of the AI ​​model sent by the terminal device; in, The second output CSI includes: the result output by the second model after performing joint source-channel decoding on the second input CSI; The second input CSI includes: the result of the first output CSI transmitted via the uplink channel; The first output CSI includes: the result output by the first model after performing joint source-channel coding on the first input CSI.

72. The network device according to any one of claims 55-71, wherein, The performance monitoring information includes at least one of performance monitoring values ​​and performance monitoring auxiliary information.

73. A terminal device, comprising: A transceiver, a processor, and a memory, wherein the memory is used to store a computer program, the transceiver is used to communicate with other devices, and the processor is used to invoke and run the computer program stored in the memory to cause the terminal device to perform the method as described in any one of claims 1 to 18.

74. A network device, comprising: A transceiver, a processor, and a memory, wherein the memory is used to store a computer program, the transceiver is used to communicate with other devices, and the processor is used to invoke and run the computer program stored in the memory to cause the network device to perform the method as described in any one of claims 19 to 36.

75. A chip, comprising: A processor for retrieving and running a computer program from memory, causing a device on which the chip is mounted to perform the method as described in any one of claims 1 to 36.

76. A computer-readable storage medium for storing a computer program that, when run by a device, causes the device to perform the method as claimed in any one of claims 1 to 36.

77. A computer program product comprising computer program instructions that cause a computer to perform the method as described in any one of claims 1 to 36.

78. A computer program that causes a computer to perform the method as claimed in any one of claims 1 to 36.

79. A communication system, comprising: A terminal device for performing the method as described in any one of claims 1 to 18; A network device for performing the method as described in any one of claims 19 to 36.

Images