Training method, model usage method, and wireless communication method and apparatus

Abstract

Provided are a training method, a model usage method, and a wireless communication method and apparatus. The training method comprises: a first device generating a second data set according to a first data set, wherein data in the second data set is low-dimensional represented data of data in the first data set; and according to the second data set, the first device training a first model for wireless communication.

Description

Training method, method for using model, method and device for wireless communication Technical Field

[0001] The present application relates to the field of communication technology, and more specifically, to a training method, a method for using a model, and a method and apparatus for wireless communication. Background Art

[0002] With the development of artificial intelligence (AI) technology, wireless communication systems have begun to use models for wireless communication to improve communication performance. However, when using datasets to train models, there is a problem of poor training timeliness.

[0003] Summary of the Invention

[0004] This application provides a training method, a method for using a model, and a method and apparatus for wireless communication. The following introduces various aspects of this application.

[0005] In a first aspect, a training method is provided, including: a first device generates a second data set based on a first data set, wherein the data in the second data set is low-dimensional representation data of the data in the first data set; and the first device trains a first model for wireless communication based on the second data set.

[0006] In a second aspect, a method for using a model is provided, including: a first device generates second data based on first data, wherein the second data is low-dimensional representation data of the first data; and the first device obtains a processing result of the first model based on the second data and a first model for wireless communication.

[0007] In a third aspect, a method for wireless communication is provided, including: a terminal device receives a first model and a second model from a network device; wherein the second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than the first data, and the first model is used to process the second data.

[0008] In a fourth aspect, a method for wireless communication is provided, comprising: a network device sends a first model and a second model to a terminal device; wherein the second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than the first data, and the first model is used to process the second data.

[0009] In a fifth aspect, a training device is provided, comprising: a generation unit for generating a second data set based on a first data set, wherein the data in the second data set is low-dimensional representation data of the data in the first data set; and a training unit for training a first model for wireless communication based on the second data set.

[0010] In a sixth aspect, a device for using a model is provided, comprising: a generation unit for generating second data based on first data, wherein the second data is low-dimensional representation data of the first data; and a processing unit for obtaining a processing result of the first model based on the second data and a first model for wireless communication.

[0011] In the seventh aspect, a terminal device is provided, including: a receiving unit for receiving a first model and a second model from a network device; wherein the second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than the first data, and the first model is used to process the second data.

[0012] In the eighth aspect, a network device is provided, including: a sending unit for sending a first model and a second model to a terminal device; wherein the second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than the first data, and the first model is used to process the second data.

[0013] In a ninth aspect, a device is provided, comprising a memory and a processor, wherein the memory is used to store a program, and the processor is used to call the program in the memory to execute the method as described in any one of the first to fourth aspects.

[0014] In a tenth aspect, a device is provided, comprising a processor for calling a program from a memory to execute the method as described in any one of the first to fourth aspects.

[0015] In an eleventh aspect, a chip is provided, comprising a processor for calling a program from a memory so that a device equipped with the chip executes the method described in any one of the first to fourth aspects.

[0016] In a twelfth aspect, a computer-readable storage medium is provided, on which a program is stored, wherein the program enables a computer to execute the method as described in any one of the first to fourth aspects.

[0017] In a thirteenth aspect, a computer program product is provided, comprising a program, wherein the program enables a computer to execute the method as described in any one of the first to fourth aspects.

[0018] In a fourteenth aspect, a computer program is provided, wherein the computer program enables a computer to execute the method as described in any one of the first to fourth aspects.

[0019] This application first generates low-dimensional representation data of the first dataset, that is, generates a second dataset, and then uses the second dataset to train the first model. Because the dimension of the data in the second dataset is lower than the dimension of the data in the first dataset, compared with the solution of directly using the first dataset to train the first model, using the second dataset to train the first model can reduce the parameters in the first model and reduce the size of the first model, thereby improving the timeliness of the first model training. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] FIG1 is a wireless communication system used in an embodiment of the present application.

[0021] FIG2 is a structural diagram of a neural network applicable to an embodiment of the present application.

[0022] FIG3 is a structural diagram of a CNN applicable to an embodiment of the present application.

[0023] FIG4 is a schematic diagram of a CSI feedback system provided in an embodiment of the present application.

[0024] FIG5 is a schematic diagram of the structure of an autoencoder.

[0025] FIG6 is a schematic diagram of an online training method.

[0026] FIG7 is a schematic diagram of an offline training method.

[0027] FIG8 is a flow chart of a training method provided in an embodiment of the present application.

[0028] Figure 9 is a schematic diagram of a VAE encoder provided in an embodiment of the present application.

[0029] FIG10 is a schematic diagram of a second model provided in an embodiment of the present application.

[0030] FIG11 is a schematic diagram of a training method of a first model provided in an embodiment of the present application.

[0031] FIG12 is a flow chart of a method for using a model provided in an embodiment of the present application.

[0032] FIG13 is a flow chart of a wireless communication method provided in an embodiment of the present application.

[0033] FIG14 is a flow chart of a method for performing offline training by a network device provided in an embodiment of the present application.

[0034] FIG15 is a schematic diagram showing an online training method provided in an embodiment of the present application.

[0035] FIG16 is a flow chart of a method for performing online training by a network device provided in an embodiment of the present application.

[0036] FIG17 is a flow chart of a method for performing online training by a terminal device provided in an embodiment of the present application.

[0037] FIG18 is a schematic block diagram of a training device provided in an embodiment of the present application.

[0038] FIG19 is a schematic block diagram of a device using a model provided in an embodiment of the present application.

[0039] Figure 20 is a schematic block diagram of a terminal device provided in an embodiment of the present application.

[0040] Figure 21 is a schematic block diagram of a network device provided in an embodiment of the present application.

[0041] FIG22 is a schematic structural diagram of a device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] The technical solution in this application will be described below with reference to the accompanying drawings.

[0043] Figure 1 illustrates a wireless communication system 100 used in an embodiment of the present application. The wireless communication system 100 may include a network device 110 and a terminal device 120. The network device 110 may be a device that communicates with the terminal device 120. The network device 110 may provide communication coverage for a specific geographic area and may communicate with the terminal device 120 within the coverage area.

[0044] FIG1 exemplarily shows a network device and two terminals. Optionally, the wireless communication system 100 may include multiple network devices and each network device may include other numbers of terminal devices within its coverage area, which is not limited in the embodiments of the present application.

[0045] Optionally, the wireless communication system 100 may further include other network entities such as a network controller and a mobility management entity, which is not limited in the embodiment of the present application.

[0046] It should be understood that the technical solutions of the embodiments of the present application can be applied to various communication systems, such as: fifth generation (5G) system or new radio (NR), long term evolution (LTE) system, LTE frequency division duplex (FDD) system, LTE time division duplex (TDD), etc. The technical solutions provided in this application can also be applied to future communication systems, such as the sixth generation mobile communication system, satellite communication system, etc.

[0047] The terminal device in the embodiments of the present application may also be referred to as user equipment (UE), access terminal, user unit, user station, mobile station, mobile station (MS), mobile terminal (MT), remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent or user device. The terminal device in the embodiments of the present application may refer to a device that provides voice and / or data connectivity to a user and can be used to connect people, objects and machines, such as a handheld device with wireless connection function, a vehicle-mounted device, etc. The terminal device in the embodiments of the present application can be a mobile phone, a tablet computer, a laptop computer, a PDA, a mobile internet device (MID), a wearable device, a virtual reality (VR) device, an augmented reality (AR) device, a wireless terminal in industrial control, a wireless terminal in self-driving, a wireless terminal in remote medical surgery, a wireless terminal in a smart grid, a wireless terminal in transportation safety, a wireless terminal in a smart city, a wireless terminal in a smart home, etc. Optionally, the UE can be used to act as a base station. For example, the UE can act as a scheduling entity that provides sidelink signals between UEs in V2X or D2D, etc. For example, a cellular phone and a car communicate with each other using sidelink signals. The cellular phone and smart home devices communicate without relaying the communication signal through the base station.

[0048] The network device in the embodiments of the present application may be a device for communicating with a terminal device, and may also be referred to as an access network device or a radio access network device. For example, the network device may be a base station. The network device in the embodiments of the present application may refer to a radio access network (RAN) node (or device) that connects a terminal device to a wireless network. A base station can broadly cover various names as follows, or be replaced with the following names, such as: NodeB, evolved NodeB (eNB), next generation NodeB (gNB), relay station, access point, transmission point (TRP), transmission point (TP), master station MeNB, secondary station SeNB, multi-standard radio (MSR) node, home base station, network controller, access node, wireless node, access point (AP), transmission node, transceiver node, baseband unit (BBU), remote radio unit (RRU), active antenna unit (AAU), remote radio head (RRH), central unit (CU), distributed unit (DU), positioning node, etc. A base station can be a macro base station, a micro base station, a relay node, a donor node or the like, or a combination thereof. A base station can also refer to a communication module, a modem or a chip used to be set in the aforementioned device or apparatus. The base station can also be a mobile switching center and a device that performs base station functions in device-to-device D2D, vehicle-to-everything (V2X), and machine-to-machine (M2M) communications, a network-side device in a 6G network, or a device that performs base station functions in future communication systems. The base station can support networks with the same or different access technologies. The embodiments of this application do not limit the specific technology and specific device form used by the network equipment.

[0049] Base stations can be fixed or mobile. For example, a helicopter or drone can be configured to act as a mobile base station, and one or more cells can move based on the location of the mobile base station. In other examples, a helicopter or drone can be configured to act as a device that communicates with another base station.

[0050] In some deployments, the network device in the embodiments of the present application may refer to a CU or a DU, or the network device includes a CU and a DU. The gNB may also include an AAU.

[0051] The network equipment and terminal devices can be deployed on land, including indoors or outdoors, handheld or vehicle-mounted; they can also be deployed on water; they can also be deployed in the air on aircraft, balloons, and satellites. The embodiments of this application do not limit the scenarios in which the network equipment and terminal devices are located.

[0052] It should be understood that all or part of the functions of the communication device in this application can also be implemented through software functions running on hardware, or through virtualization functions instantiated on a platform (such as a cloud platform).

[0053] AI is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. AI is a hot topic in science and a cutting-edge technology in global development, with applications in a wide variety of everyday situations.

[0054] One way to implement AI is through a neural network, which is described below.

[0055] In recent years, artificial intelligence research, exemplified by neural networks, has achieved remarkable success in many fields and will continue to play a vital role in people's lives and production for a long time to come. In particular, machine learning (ML), a key research area in AI, leverages the nonlinear processing capabilities of neural networks (NNs) to successfully solve a range of previously intractable problems. It has even demonstrated superior performance to humans in areas such as image recognition, speech processing, natural language processing, and gaming, and has therefore garnered increasing attention. Common neural networks include convolutional neural networks (CNNs), recurrent neural networks (RNNs), and deep neural networks (DNNs).

[0056] The following describes a neural network applicable to embodiments of the present application in conjunction with FIG2 . The neural network shown in FIG2 can be divided into three categories based on the location of different layers: input layer 210 , hidden layer 220 , and output layer 230 . Generally speaking, the first layer is the input layer 210 , the last layer is the output layer 230 , and the intermediate layers between the first and last layers are all hidden layers 220 .

[0057] The input layer 210 is used to input data. The hidden layer 220 is used to process the input data. The output layer 230 is used to output the processed output data.

[0058] As shown in Figure 2, a neural network consists of multiple layers, each of which contains multiple neurons. The neurons between layers can be fully connected or partially connected. For connected neurons, the output of the neurons in the previous layer can serve as the input of the neurons in the next layer.

[0059] With the continuous advancement of neural network research, deep learning algorithms have been proposed in recent years. These algorithms introduce a large number of hidden layers into neural networks, forming DNNs. More hidden layers allow DNNs to better capture complex real-world situations. Theoretically, a model with more parameters has higher complexity and a greater "capacity," meaning it can handle more complex learning tasks. These neural network models are widely used in pattern recognition, signal processing, optimization and combination, anomaly detection, and other fields.

[0060] CNN is a deep neural network with a convolutional structure. Its structure is shown in Figure 3 and may include an input layer 310, a convolutional layer 320, a pooling layer 330, a fully connected layer 340, and an output layer 350.

[0061] Each convolution layer 320 may include a plurality of convolution operators, which are also called kernels. The convolution operator can be regarded as a filter for extracting specific information from the input signal. The convolution operator can essentially be a weight matrix, which is usually predefined.

[0062] The weight values ​​in these weight matrices need to be obtained through a lot of training in practical applications. The weight matrices formed by the weight values ​​obtained through training can extract information from the input signal, thereby helping CNN to make correct predictions.

[0063] When CNN has multiple convolutional layers, the initial convolutional layer tends to extract more general features, which can also be called low-level features. As the depth of CNN increases, the features extracted by the subsequent convolutional layers become more and more complex.

[0064] Pooling layers 330 are often used periodically after convolutional layers to reduce the number of training parameters. For example, a single convolutional layer can be followed by a single pooling layer, as shown in Figure 3, or multiple convolutional layers can be followed by one or more pooling layers. In signal processing, the sole purpose of a pooling layer is to reduce the spatial size of the extracted information.

[0065] After processing by the convolutional layer 320 and the pooling layer 330, the CNN is still not sufficient to output the required output information. As previously mentioned, the convolutional layer 320 and the pooling layer 330 only extract features and reduce the parameters of the input data. However, to generate the final output information, the CNN also requires the use of the fully connected layer 340. Typically, the fully connected layer 340 can include multiple hidden layers, the parameters of which can be pre-trained based on relevant training data for the specific task type.

[0066] Following the multiple hidden layers in the fully connected layer 340, the final layer of the CNN is the output layer 350, which is used to output the results. Typically, the output layer 350 is configured with a loss function (e.g., a loss function similar to categorical cross entropy) to calculate the prediction error, or to evaluate the degree of difference between the output of the CNN model (also known as the predicted value) and the ideal result (also known as the true value).

[0067] To minimize the loss function, the CNN model needs to be trained. In some implementations, the CNN model can be trained using a backpropagation algorithm (BP). The BP training process consists of a forward propagation process and a backward propagation process. During the forward propagation process (e.g., the propagation from 310 to 350 in Figure 3 is forward propagation), the input data is fed into the aforementioned layers of the CNN model, processed layer by layer, and transmitted to the output layer. If the output result differs significantly from the ideal result, the minimization of the aforementioned loss function is used as the optimization goal, and the backward propagation process is switched to (e.g., the propagation from 350 to 310 in Figure 3 is backward propagation). The partial derivatives of the optimization goal with respect to each neuron weight are calculated layer by layer, forming the gradient of the optimization goal with respect to the weight vector, which serves as the basis for modifying the model weights. The CNN training process is completed during the weight modification process. When the aforementioned error reaches the desired value, the CNN training process ends.

[0068] It should be noted that the CNN shown in Figure 3 is only an example of a convolutional neural network. In specific applications, the convolutional neural network can also exist in the form of other network models, and the embodiments of the present application are not limited to this.

[0069] Given the tremendous success of AI technology, particularly deep learning, in areas such as computer vision and natural language processing, the communications field has begun exploring the use of models to address technical challenges that are difficult to address with traditional communications methods. Existing research demonstrates that AI has significant application potential in many areas, including modeling and learning complex and unknown environments, channel prediction, intelligent signal generation and processing, network status tracking and intelligent scheduling, and network optimization and deployment. It is expected to promote the evolution of future communications paradigms and transform network architectures, and is of great significance and value to the research of communication system technologies, including 6G systems.

[0070] In some embodiments, a communication device may utilize a first model to process data, thereby improving communication performance and reducing data processing complexity. For example, a communication device may utilize a first model to encode and decode data to improve data encoding and decoding performance. The first model may be referred to as an AI model.

[0071] Taking channel state information (CSI) feedback as an example, the terminal device can use the first model to extract features from the actual channel information and generate a bitstream. The network device can then reconstruct the bitstream using the first model to restore the actual channel information as closely as possible. Using the first model can reduce the CSI feedback overhead of the terminal device while ensuring the restoration of the actual channel information.

[0072] The following describes the CSI feedback system in conjunction with FIG4 .

[0073] The network device can send a reference signal to the terminal device. Based on the reference signal, the terminal device can estimate the channel and obtain the CSI data to be fed back. The terminal device uses an encoder to encode the CSI data to be fed back to obtain an encoded bit stream, and sends the bit stream to the network device. After receiving the bit stream, the network device can use a decoder to decode the bit stream to restore the original CSI data. The above-mentioned encoder and decoder can be implemented by a first model. The first model for CSI feedback is also called a CSI feedback model. The CSI feedback model may include an AI encoder and an AI decoder. The network model structure of the AI ​​encoder and AI decoder can be flexibly designed, and the embodiments of the present application do not specifically limit this.

[0074] Taking the first model, a deep learning model, as an example, the neural network architecture commonly used in deep learning is nonlinear and data-driven. Terminal devices can use deep learning models to extract features from actual channel data, and network equipment can use deep learning models to restore the actual channel data as closely as possible. Deep learning-based CSI feedback treats channel information as an image to be compressed, compresses and feeds this information back using a deep learning model, and reconstructs the compressed channel image at the receiver, preserving channel information to a greater extent.

[0075] The architecture of the CSI feedback system shown in Figure 4 is the same as that of an autoencoder (AE). An autoencoder is a type of neural network used in semi-supervised learning and unsupervised learning. Its function is to represent and learn the input information by using the input information as a learning target. Figure 5 shows a schematic diagram of the structure of an autoencoder. As shown in Figure 5, the autoencoder may include an AI encoder and an AI decoder. After the autoencoder training is completed, the AI ​​encoder can be deployed at the transmitting end (such as a terminal device) and the AI ​​decoder can be deployed at the receiving end (such as a network device). The transmitting end can use the AI ​​encoder to encode the data, and the receiving end can use the AI ​​decoder to decode the data.

[0076] Currently, in the field of wireless AI, training the first model is crucial and requires a large amount of computing resources. Most training processes treat the first model as a black box, directly using raw data as input for training and updating.

[0077] The performance of the first model is strongly related to the distribution of data. The distribution of data will be affected by the wireless environment. For example, the data distribution will be affected by factors such as time, environment, and system strategy, which will cause the actual data in the wireless communication system to be different from the simulated data. If the first model is trained using simulated data, the performance of the trained first model will be very poor. Therefore, after the first model is deployed, it is necessary to perform online training. The current training process of the first model can be shown in Figure 6. The first model in the embodiment of the present application can also be called a task model, a business model, etc.

[0078] First, the first model is pre-trained using offline training data. After pre-training is complete, the first model can be deployed. For example, the first model can be deployed on a terminal device or a network device. If the first model is pre-trained locally, the deployment step can be omitted. For example, if offline training is performed by a network device, the network device can send the first model to the terminal device after offline training is complete. For another example, if offline training is performed by a third-party device, the third-party device can send the first model to the terminal device and / or network device after offline training is complete.

[0079] Secondly, after the first model is deployed, it can be trained and updated online using online training data. If the online training of the first model is not performed locally, the first model needs to be deployed online after the online training is completed. For example, if the online training is performed by a network device, the network device can send the trained first model to the terminal device after completing the online training of the first model.

[0080] Finally, after the first model training is completed, the first model can be used for inference (or usage). The terminal device or network device can use the first model to infer data.

[0081] However, in the scenario where the input of the first model is raw data, if the dimension of the raw data is relatively large, the parameters of the first model will be relatively large, resulting in a relatively large first model, which requires more computing resources to complete the training task. In the case of limited computing power, it takes a long time to complete the training of the first model, which makes it difficult to meet the timeliness requirements of the first model training. In addition, the training process of a larger first model also needs to rely on more new data, which further increases the training time of the first model. This is especially true for online training, which has higher requirements for timeliness, and the current training method is difficult to meet the timeliness requirements of online training.

[0082] When the first model is large, traditional methods typically use the training method shown in Figure 7, which skips the online training process and trains the first model offline. However, offline training alone cannot effectively combat the impact of data drift, and the offline trained first model cannot adapt to the current network environment, resulting in poor performance.

[0083] However, whether it is online training or offline training, since the first model is relatively large, there will be problems such as long training time, slow update rhythm, and failure to meet timeliness requirements.

[0084] There are two ways to speed up the training of the first model: reducing the amount of computation required for each iteration, and reducing the number of training iterations. For the first approach, a lightweight first model is needed to reduce computation and speed up training. Current research focuses on reducing the number of training iterations, such as through meta-learning. However, for devices with limited memory and computing power (such as terminal devices), simply reducing the number of iterations cannot effectively solve the problem of timeliness in first model training.

[0085] Based on this, an embodiment of the present application provides a training method. The method of the embodiment of the present application can first generate a low-dimensional representation dataset of the first data set, and then train the first model based on the low-dimensional representation dataset. Since the dimension of the input data of the first model is reduced, the method of the embodiment of the present application can reduce the parameters of the first model and reduce the size of the first model, thereby reducing the training time of the first model, which is conducive to meeting the timeliness requirements. The training process of the embodiment of the present application is introduced below in conjunction with Figure 8.

[0086] 8 , in step S810 , the first device generates a second data set based on the first data set.

[0087] The embodiments of the present application do not specifically limit the type of the first device. The first device can be any computing device. For example, the first device can be a communications device, such as a terminal device or a network device. In another example, the first device can be a non-communication device, that is, a dedicated computing device.

[0088] The first data set may also be referred to as a training data set. The first data set may be an offline data set or an online data set. The offline data set may include historical real data and / or simulated data, etc. The online data set may be data generated in real time by the wireless communication system. Taking CSI feedback as an example, the first data set may include CSI data to be fed back. The embodiments of the present application do not specifically limit the number of samples included in the first data set. For example, the first data set may include a single sample or a batch of samples.

[0089] The data in the second dataset is a low-dimensional representation of the data in the first dataset. In other words, the dimensionality of the data in the second dataset is lower than that of the data in the first dataset. The specific method for generating the second dataset will be described in detail below.

[0090] In step S820, the first device trains a first model for wireless communication according to the second data set.

[0091] The first model in the embodiment of the present application may be any AI model in the wireless communication system. The first model may be, for example, a service model or a task model. The embodiment of the present application does not specifically limit the type of the first model. For example, the first model may be a neural network model, or the first model may be a deep learning model. In some embodiments, the first model may include a coding model, that is, the first model may include an AI encoder and an AI decoder. For example, the first model may include a CSI feedback model, or the first model may include a channel prediction model (or a channel estimation model). Of course, in some embodiments, the first model may also include a coding model, that is, the first model includes an AI encoder. Alternatively, the first model may also include a decoding model, that is, the first model includes an AI decoder.

[0092] The first device training the first model based on the second data set can be understood as the first device using the data in the second data set as the input of the first model to train the first model. In some embodiments, the first device can use the data in the second data set as the input of the first model to obtain the output result of the first model; the first device uses the difference between the output result of the first model and the label data of the first model to train the first model. The label data can be set according to actual needs and is not specifically limited in this embodiment of the present application. Taking the example of the first model including a codec model, the label data can be the data in the first data set.

[0093] For example, assuming the first model is a CSI feedback model, the label data may be the first data set, that is, the label data may be a channel feature vector. assuming the first model is a channel prediction model, the label data may be channel information at a future time.

[0094] Compared with the data in the first data set, the dimension of the data in the second data set is lower. Therefore, using the data in the second data set to train the first model can reduce the parameters in the first model and reduce the size of the first model, which is conducive to improving the timeliness of the first model training.

[0095] The training method of the embodiment of the present application can be applied to online training and offline training. As can be seen from the above, due to the problem of timeliness of online training, the current first model can only be trained offline. The solution of the embodiment of the present application can improve the timeliness of the training of the first model. Therefore, the solution of the embodiment of the present application is conducive to the evolution of the first model from offline training to online training, that is, the solution of the embodiment of the present application can perform online training on the first model. Online training of the first model is also beneficial to combat the impact of data distribution drift. For example, after the first model is deployed, the data generated in real time can be used to train and update the first model online, so that the first model can match the current network environment and improve the performance of the first model.

[0096] If online training is performed, the embodiments of the present application do not specifically limit the execution time of online training. As an example, online training can be performed when new data is generated. As another example, online training can be performed when the number of samples reaches a preset threshold. The preset threshold can be set according to actual needs. For example, the preset threshold can be one or more of the following: 16, 32, 64, 128, 512. As another example, online training can be performed at fixed intervals, that is, online training can be performed periodically. The fixed duration can be set according to actual needs. For example, the fixed duration can be one or more of the following: 5 time slots, 10 time slots, 20 time slots, etc.

[0097] The embodiments of the present application do not limit the method for generating the second data set. For example, the first device may process the data in the first data set using the second model to generate the second data set. For another example, the first device may process the data in the first data set using a specific algorithm to generate the second data set. The specific algorithm may be referred to as feature engineering. The algorithm may be designed based on some experience and prior knowledge. The specific algorithm may be, for example, a dimensionality reduction algorithm and / or a matrix decomposition algorithm, and the process of obtaining the optimal representation of the data may be considered a special training method.

[0098] In some embodiments, embodiments of the present application may use a specific algorithm to generate the second data set when the data volume is small, and use a second model to generate the second data set when the data volume is large or the data is relatively complex. Alternatively, embodiments of the present application may not consider the size of the data volume or the complexity of the data, that is, regardless of whether the data volume is large or the data is complex, the second model is used to generate the second data set.

[0099] The second model of the embodiment of the present application may include a representation learning model. Representation learning is a type of machine learning method that can learn the representation of data to extract useful information from the data. The purpose of representation learning is to simplify complex raw data, eliminate invalid or redundant information in the raw data, and refine the effective information to form features. Therefore, using the representation learning model to reduce the dimensionality of the data in the first data set can retain more useful information in the data, thereby facilitating the training of subsequent models.

[0100] The embodiment of the present application does not specifically limit the specific implementation method of the second model, as long as the data can be reduced in dimension and the useful information of the data can be retained. In some embodiments, the second model may include an encoder in a variational auto-encoder (VAE) model. Since the VAE model has a strong representation ability, that is, it can use a small dimension (or vector) to represent more information and can contain higher-level feature information, the use of the encoder in the VAE model can achieve a greater degree of dimensionality reduction of the data, thereby further reducing the size of the first model and improving the timeliness of the first model training.

[0101] VAE has the same structure as the autoencoder, both including an encoder and a decoder. However, unlike the autoencoder, VAE can add constraints to the encoder part, that is, the output of the encoder can be manually specified. For example, the AI ​​encoder output can be constrained to obey the latent variables of the Gaussian distribution. In other words, the encoder in the VAE model can output a better spatial embedding rather than an uncontrolled distribution space. Therefore, the output of the encoder in the VAE model can be used as a low-dimensional representation of the original data. In this new embedding space, different data form a distribution state with better correlation, and this distribution state is conducive to the learning of downstream models (such as the first model).

[0102] Since the output of the encoder in the VAE model can be manually specified, when the encoder of the VAE model is used to generate the second data set, the dimension of the data in the second data set can be manually specified. In other words, the dimension of the data in the second data set in the embodiment of the present application can be flexibly designed according to actual needs.

[0103] In some embodiments, embodiments of the present application may further train a second model based on the first dataset. For example, the first device may use the first dataset as input to train the second model. After the second model is trained, the first device may use the trained second model to process the first dataset to generate a second dataset.

[0104] The following introduces the training process of the second model by taking the second model including the encoder in the VAE model as an example.

[0105] As shown in FIG9 , the VAE model may include an encoder 1 and a decoder 1. The first device may use the first data set as the input and output of the VAE model to train the VAE model. The dimension N of the output data of encoder 1 is RL It can be set in advance. After the VAE model training is completed, only encoder 1 can be retained, decoder 1 can be deleted, and encoder 1 can be used as the second model. The input of encoder 1 can be the first dataset and the output can be the second dataset. The dimension of the second dataset is N RL The final second model can be shown in FIG10 .

[0106] The specific training method of the second model can be determined based on the representation learning algorithm, and the embodiments of the present application do not specifically limit this. For example, taking the VAE model as an example, the input and output of the VAE model are the same, and the loss function can adopt the VAE standard loss function, such as reconstruction loss and distribution hypothesis loss, to train the VAE model.

[0107] Since the second model is insensitive to data distribution, i.e., different data distributions have little impact on the performance of the second model, after the second model is deployed, it is not necessary to update and train the second model, but only the first model.

[0108] The following two specific examples are used to illustrate the first data set and the second data set. It should be understood that the following examples are only provided for ease of understanding and should not limit the solutions of the embodiments of the present application.

[0109] Example 1: For CSI feedback scenarios, the first data set may be a channel feature vector. For example, if the transmitting terminal device has 32 ports and the subcarriers are divided into 13 subbands, the first data set w may include 13 subband feature vectors:

[0110] w=[w1,w2,…,w 13 ]

[0111] Among them, w k represents the kth subband eigenvector, 1≤k≤13. Each subband eigenvector w k Contains the complex information of each sending port. When training the model, the complex information is generally decomposed into real part information and imaginary part information. For example, the terminal device has 32 sending ports. k It can be expressed as:

[0112] w k =[Re{w k,1},Im{w k,1},Re{w k,2},Im{w k,2},…,Re{w k,32},Im{w k,32}]

[0113] Where Re{} and Im{} represent the real part and imaginary part of the complex number, respectively. Therefore, the sample of the first data set is a vector of 13*32*2 real numbers with a dimension of 832. As the number of ports and the number of subbands into which the subcarriers are divided increase, the dimension of the first data set doubles. The embodiment of the present application can use the first model (such as the representation learning model) to reduce the dimension of the first data set to a target dimension N RL The target dimension can be any integer smaller than the original data dimension 832. For example, the target dimension can be any integer among 256, 128, 100, and 50. It is understood that the target dimension is the dimension of the second data set.

[0114] Example 2, for the channel prediction scenario, the first device can use the past (or historical) measurement reference signal to predict the channel information at a future moment, and the measurement reference signal can be a periodic reference signal. The first data set can be a past measurement reference signal. For example, assume that the network device uses a 4-row 8-column dual-polarization antenna array for transmission and uses two dual-polarization antennas for reception, that is, the network device includes 64 transmitting ports and 4 receiving ports. In this case, the first data set can be a channel slice data set, and each input sample (channel slice data) in the first data set can contain 32256 complex numbers, that is, 126 delay taps × 4 receiving antennas × 64 transmitting antennas. The embodiment of the present application can use a first model (such as a representation learning model) to reduce the dimension of the first data set to a target dimension N RL The target dimension can be any integer smaller than the original data dimension of 32256. For example, the target dimension can be any integer among 4096, 2000, 1024, 500, and 256. It is understood that the target dimension is the dimension of the second data set.

[0115] After the training of the second model is completed, the first model can be trained. Below, in conjunction with Figure 11, the training process of the first model is introduced by taking the first model including the encoding and decoding model as an example. As shown in Figure 11, the first model may include an AI encoder and an AI decoder. In the embodiment of the present application, the second data set can be used as the input of the first model and the first data set as the output of the first model to train the first model. It should be noted that in the embodiment of the present application, using the first data set as the output of the first model can be understood as using the first data set as the training label of the first model.

[0116] The training process of the first model is described in detail above. The reasoning process of the first model is described below in conjunction with Figure 12. It should be noted that the reasoning process of the first model corresponds to some contents of the training process of the first model. For parts not described in detail, please refer to the description above.

[0117] 12 , in step S1210 , the first device generates second data according to the first data.

[0118] The first device may be a device in a wireless communication system, such as a terminal device or a network device.

[0119] The first data may be wireless communication data. In some embodiments, the first data may be data to be encoded. For example, the first data may be CSI data to be fed back.

[0120] The second data is a low-dimensional representation of the first data, that is, the dimensionality of the second data is lower than that of the first data. The embodiments of this application do not specifically limit the method for generating the second data. As an example, the first device may process the first data using a specific algorithm to generate the second data. This specific algorithm may be referred to as feature engineering. The algorithm may be designed based on experience and prior knowledge.

[0121] As another example, the first device can use the second model to process the first data to generate second data. While reducing the dimensionality of the data, the second model can also retain more useful information of the data, which is beneficial to the subsequent data processing. The second model may, for example, include an encoder in a VAE model. Since the VAE model has a strong representation ability, that is, it can use a small dimension (or vector) to represent more information and can contain higher-level feature information, the use of the encoder in the VAE model can achieve a greater degree of dimensionality reduction of the data and reduce the complexity of subsequent data processing.

[0122] In step S1220, the first device obtains a processing result of the first model based on the second data and the first model for wireless communication.

[0123] The first model in the embodiment of the present application can be any AI model in the wireless communication system. The first model can be, for example, a business model or a task model. The embodiment of the present application does not specifically limit the type of the first model. For example, the first model can be a neural network model, or the first model can be a deep learning model. In some embodiments, the first model can include a coding model, that is, the first model can include an AI encoder and an AI decoder. For example, the first model can include a CSI feedback model. Of course, in some embodiments, the first model can also include a coding model, that is, the first model includes an AI encoder. Alternatively, the first model can also include a decoding model, that is, the first model includes an AI decoder.

[0124] The first device can use the second data as input to the first model to obtain the processing results of the first model. The processing results of the first model can be understood as the output of the first model. Because the second data has a lower dimension than the first data, using the second data as input to the first model can reduce the processing time of the first model and increase its processing speed.

[0125] Taking the example that the first model includes a codec model, the first model may include an AI encoder and an AI decoder. Since the AI ​​encoder and the AI ​​decoder have correspondence, that is, the AI ​​decoder can decode the data encoded by the AI ​​encoder, the AI ​​encoder and the AI ​​decoder need to be jointly trained together. After the training of the AI ​​encoder and the AI ​​decoder is completed, the AI ​​encoder and / or the AI ​​decoder need to be sent to the corresponding device. For example, if the AI ​​encoder and the AI ​​decoder are trained on the encoding end, the AI ​​decoder can be sent from the encoding end to the decoding end. If the AI ​​encoder and the AI ​​decoder are trained on the decoding end, the AI ​​encoder can be sent from the decoding end to the encoding end. If the AI ​​encoder and the AI ​​decoder are trained on the decoding end, the AI ​​encoder can be sent from the decoding end to the encoding end. If the AI ​​encoder and the AI ​​decoder are trained on a third-party device, the third-party device can send the AI ​​encoder to the encoding end and send the AI ​​decoder to the decoding end. The above-mentioned encoding end may also be referred to as a sending end, and the decoding end may also be referred to as a receiving end.

[0126] The following describes the solution of the embodiment of the present application from the perspective of communication interaction, using a terminal device as the encoding end and a network device as the decoding end as an example. The communication interaction process between the terminal device and the network device can include the transmission process of the model and the inference process of the model. For any content not described in detail below, please refer to the description above.

[0127] 13 , in step S1310 , the network device sends a first model and a second model to the terminal device.

[0128] The first model in the embodiment of the present application can be any first model in the wireless communication system. The first model can be, for example, a service model or a task model. The embodiment of the present application does not specifically limit the type of the first model. For example, the first model can be a neural network model, or the first model can be a deep learning model. In some embodiments, the first model can include a coding model, that is, the first model can include an AI encoder and an AI decoder. For example, the first model can include a CSI feedback model. Of course, in some embodiments, the first model can also include a coding model, that is, the first model includes an AI encoder. Alternatively, the first model can also include a decoding model, that is, the first model includes an AI decoder.

[0129] The network device of the embodiment of the present application can train the first model and the second model. Since the terminal device has limited memory space and computing power, the first model of the embodiment of the present application can be trained by the network device to save the computing overhead of the terminal device. After training is completed, the network device can send the first model and the second model to the terminal device so that the first model and the second model are deployed on the terminal device. The training of the first model and the second model can be obtained through offline training.

[0130] The training process of the first and second models can be found in the description above. The second model can be used to convert the first data of the terminal device into second data, where the second data is a low-dimensional representation of the first data. The first model can be used to process the second data. After obtaining the first model, the terminal device can use the second data to infer the first model, or it can also use the second data to train the first model.

[0131] In some embodiments, after the terminal device obtains the first model and the second model, it can use the second model to process the first data to generate the second data. The terminal device can also use the first model to process the second data to obtain the processing result of the first model. The first data can be data generated by the terminal device, the first data can be data measured by the terminal device, or the first data can be data to be sent by the terminal device. Taking the first model including an AI encoder as an example, the processing result of the first model is encoded data. The terminal device can send the encoded data to the network device. After receiving the encoded data sent by the terminal device, the network device can use the AI ​​decoder to process the encoded data to generate the first data.

[0132] After the first model is deployed, the terminal device or network device can also update (i.e., train) the first model. The update of the first model can be performed by the network device or the terminal device. The update of the first model can be performed offline or online. If the update is offline, the update of the first model can be performed by the network device to save computing overhead on the terminal device.

[0133] In some embodiments, the terminal device may train the first model. For example, the terminal device may process the first data using the second model to generate the second data. The terminal device may train the first model using the second data. This training process may be online training, i.e., the terminal device may train the first model online using the second data.

[0134] In some embodiments, the network device may train the first model. For example, the network device may process the first data using the second model to generate the second data. The network device may then use the second data to update and train the first model. After training is complete, the network device may send the updated first model to the terminal device.

[0135] Taking the first model including an AI encoder and an AI decoder as an example, the update of the first model may include updating the AI ​​encoder and the AI ​​decoder at the same time, or it may include updating only the AI ​​encoder without updating the AI ​​decoder, or it may include updating only the AI ​​decoder without updating the AI ​​encoder.

[0136] Because the second model is insensitive to data distribution, embodiments of the present application can update only the first model. Taking the example of a network device updating the first model, after updating the first model, the network device can send the updated first model to the terminal device. Because the first model is smaller, the efficiency of updating the first model is also improved. Furthermore, a smaller first model reduces the resource overhead required to transmit the model, thereby reducing air interface overhead.

[0137] The update of the first model may include offline update and online update (or online training). As described above, the offline update of the first model can be performed by the network device to save the computing overhead of the terminal device. In some embodiments, the online training of the first model can also be performed by the network device to further save the computing overhead of the terminal device. In other embodiments, the online training of the first model can also be performed by the terminal device. Since the terminal device is the source of the data, it is more direct for the terminal device to perform online training of the first model.

[0138] The following takes the first model including the AI ​​encoder and the AI ​​decoder as an example to introduce the training process of online training of network equipment and online training of terminal equipment respectively.

[0139] When the network device performs online training, the network device can obtain the first data from the terminal device. As an example, the terminal device can send the first data to the network device. For example, the terminal device can process the first data using the second model to generate second data, and process the second data using the AI ​​encoder to obtain encoded data. The terminal device sends the encoded data to the network device. After receiving the encoded data, the network device decodes the encoded data using the AI ​​decoder to obtain the first data. As another example, the wireless communication system of an embodiment of the present application may further include a data collection module, which can collect the first data from the terminal device and send the first data to the network device.

[0140] After obtaining the first data, the network device can use the second model to process the first data to generate the second data. Furthermore, the network device can also use the second data to update the first model (such as the AI ​​encoder) to obtain the updated first model. For example, the network device can use the second data as the input of the first model and the first data as the output of the first model to perform online training on the first model. After the online training is completed, the network device can send the updated AI encoder to the terminal device. After the terminal device receives the updated AI encoder, it can use the updated AI encoder to process the data. It should be noted that using the first data as the output of the first model can be understood as using the first data as the label data of the first model, that is, using the difference between the output result of the first model and the first data to train the first model.

[0141] In some embodiments, to reduce model transmission overhead, the network device can fix the parameters in the AI ​​encoder when training the first model and only update the parameters in the AI ​​decoder. In this way, after the network device updates the first model, it is no longer necessary to send the AI ​​encoder to the terminal device, and the terminal device can still use the previous AI encoder to process the data. After the network device receives the encoded data sent by the terminal device, it can use the updated AI decoder to decode the encoded data. The encoded data can be the bitstream described above.

[0142] If online training is performed by the terminal device, since the AI ​​encoder and AI decoder need to be trained jointly, the terminal device needs to obtain the AI ​​decoder. In some embodiments, the network device can send the AI ​​decoder in the first model to the terminal device so that the terminal device can train the first model.

[0143] During online training, the terminal device can use the second model to process the first data to generate second data. The terminal device can then use the second data to perform online training on the first model (i.e., the AI ​​encoder and AI decoder). After the online training is completed, the terminal device can send the updated AI decoder to the network device so that the network device can use the updated AI decoder to process the data.

[0144] In some embodiments, in order to reduce the model transmission overhead of the terminal device, the terminal device can fix the parameters of the AI ​​decoder when training the first model, and only update the parameters of the AI ​​encoder. In this way, after the terminal device updates the first model, it is not necessary to send the AI ​​decoder to the network device. During the data transmission process, the terminal device can use the updated AI encoder to encode the data and send the encoded data to the network device. The network device can use the original AI decoder to decode the encoded data to restore the first data. The parameters of the AI ​​decoder can be the same as or different from the parameters of the AI ​​decoders corresponding to other terminal devices, and the embodiments of the present application do not specifically limit this.

[0145] The embodiments of the present application do not specifically limit the execution time of online training. As an example, online training can be performed when new data is generated. As another example, online training can be performed when the number of samples reaches a preset threshold. The preset threshold can be set according to actual needs. For example, the preset threshold can be one or more of the following: 16, 32, 64, 128, 512. As another example, online training can be performed at fixed intervals, that is, online training can be performed periodically. The fixed duration can be set according to actual needs. For example, the fixed duration can be one or more of the following: 5 time slots, 10 time slots, 20 time slots, etc.

[0146] In some embodiments, the network device generally needs to communicate with multiple terminal devices. The AI ​​encoder of each terminal device corresponds to an AI decoder. If the network device saves an AI decoder for each terminal device, that is, the network device saves the AI ​​decoder corresponding to each terminal device, it will greatly increase the storage overhead of the network device and the pressure of model management. Therefore, in the embodiment of the present application, the AI ​​encoders of different terminal devices can correspond to the same AI decoder, that is, the parameters of the AI ​​decoders corresponding to the AI ​​encoders of multiple terminal devices are the same. In this way, the network device can store only one AI decoder, and can decode the encoded data sent by multiple terminal devices to restore the original data, which is conducive to reducing the storage overhead and model management pressure of the network device.

[0147] The scheme in which the parameters of the AI ​​decoders corresponding to the AI ​​encoders of multiple terminal devices are the same can be combined with the other schemes described above. For example, when the network device trains the first model (online training or offline update), it can fix the parameters of the AI ​​decoder in the first model and only train the parameters in the AI ​​encoder. After the training is completed, the network device sends the AI ​​encoder to the terminal device. For another example, the terminal device can fix the parameters of the AI ​​decoder and only train the parameters of the AI ​​encoder. The parameters of the AI ​​decoder can be the same as the parameters of the AI ​​decoders corresponding to other terminal devices. Of course, the parameters of the AI ​​decoder can also be different from the parameters of the AI ​​decoders corresponding to other terminal devices, and the embodiments of the present application do not specifically limit this.

[0148] The above describes the model training process. The following describes the model reasoning process.

[0149] For the model inference process, the terminal device can use the second model to process the first data to generate the second data. The terminal device can then use the AI ​​encoder to process the second data to generate encoded data. The terminal device can send the encoded data to the network device. After the network device receives the encoded data, it can use the AI ​​decoder to process the encoded data to generate the first data. It should be noted that the AI ​​decoder only restores the first data as much as possible, and the output of the AI ​​decoder is not necessarily exactly the same as the first data. In other words, there may be differences between the first data generated by the network device and the first data on the terminal device side.

[0150] The online training process and data inference process of the embodiment of the present application can be performed simultaneously. For example, the terminal device can use the AI ​​encoder to process the second data to generate encoded data, and can also use the second data to train the AI ​​encoder to update the first model.

[0151] The following is a detailed description of the scheme of the embodiments of the present application in combination with three embodiments, taking the example that the first model includes a CSI feedback model and the second model includes a representation learning model. It should be noted that the following embodiments are only for ease of understanding, and the examples provided to illustrate the scheme of the embodiments of the present application do not limit the embodiments of the present application. Example 1 introduces the offline training and update process of the CSI feedback model. Example 2 and Example 3 introduce the online training process of the CSI feedback model. The difference between Example 2 and Example 3 is that Example 2 is that the network device performs online training of the CSI feedback model, and Example 3 is that the terminal device performs online training of the CSI feedback model. Examples 1 to 3 are introduced below.

[0152] Example 1

[0153] 14 , in step S1410 , the network device may train a representation learning model using dataset 1. The representation learning model may be the second model described above. The representation learning model may include, for example, an encoder in a VAE model.

[0154] In step S1420, after the representation learning model is trained, the network device inputs the data in dataset 1 into the trained representation learning model to infer low-dimensional representations of each data point, thereby obtaining dataset 2. Dataset 2 can be understood as a low-dimensional representation of dataset 1. Compared to the data in dataset 1, the data in dataset 2 has a significantly lower dimensionality.

[0155] In step S1430, the network device can use dataset 1 and dataset 2 to train a CSI feedback model. This CSI feedback model can be the first model described above. The network device uses the data in dataset 2 as input and the data in dataset 1 as output to train the CSI feedback model. The CSI feedback model includes an AI encoder and an AI decoder. However, the AI ​​encoder in the CSI feedback model does not directly encode the CSI data to be fed back into encoded data. Instead, it encodes the low-dimensional representation data of the CSI data to be fed back. Since dataset 2 is a low-dimensional representation of dataset 1, the AI ​​encoder of the CSI feedback model is a lightweight model.

[0156] In step S1440, the network device detects that the terminal device is connected to the network device and receives first indication information, and the first indication information is used to instruct the network device to send a model to the terminal device. The first indication information can be, for example, a service indication that triggers CSI feedback.

[0157] In step S1450, the network device sends the AI ​​encoder for the representation learning model and the CSI feedback model to the terminal device, so that the terminal device processes the data using the AI ​​encoder for the representation learning model and the CSI feedback model. Because the representation learning model is not sensitive to data, it does not need to be updated after deployment. Subsequent update strategies only involve updating the CSI feedback model.

[0158] During the model update process, the network device can input new Dataset 3 into the representation learning model to obtain Dataset 4. The network device then updates the CSI feedback model using Datasets 3 and 4, obtaining an updated CSI feedback model. The network device then sends the AI ​​encoder for the updated CSI feedback model to the terminal device. After each update, the network device only sends the AI ​​encoder for the CSI feedback model with fewer parameters to the terminal device, eliminating the need to transmit the representation learning model. Compared to traditional solutions, this reduces the model transmission overhead between the network device and the terminal device during the update process.

[0159] After training or updating, the terminal device and network device can perform the inference process and jointly complete the CSI feedback task.

[0160] In step S1460, the terminal device measures the channel to obtain CSI data to be fed back. The terminal device inputs the CSI data to be fed back into the representation learning model to obtain low-dimensional representation data of the CSI data to be fed back.

[0161] In step S1470, the terminal device inputs the low-dimensional representation data into the AI ​​encoder of the CSI feedback model for inference to obtain encoded data.

[0162] In step S1480, the terminal device sends the encoded data to the network device through the air interface resources.

[0163] In step S1490, the network device uses the AI ​​decoder of the CSI feedback model to infer the encoded data and decode the original CSI data.

[0164] Before introducing Embodiment 2 and Embodiment 3, the online training process applicable to Embodiment 2 and Embodiment 3 is introduced by taking FIG. 15 as an example.

[0165] Referring to Figure 15, the entire flowchart in the embodiment of the present application can be divided into three main working modules from left to right: a data collection module, a representation learning module, and a downstream task module. Compared with the online learning scheme shown in Figure 6, the embodiment of the present application adds a representation learning module between the downstream task module and the data collection module. The representation learning model can process high-dimensional raw data into low-dimensional data, that is, high-dimensional data can be expressed with less information. Compared with traditional solutions, the downstream task model of the embodiment of the present application can be significantly reduced to achieve model compression. For online learning, the amount of computation for each iteration can be reduced, effectively solving the problem of timeliness of online training.

[0166] The data collection module can be a system data platform, used to implement data preprocessing tasks such as data filtering, and provide training data and inference data in the model training and inference stages respectively.

[0167] The representation learning module may include any of the second models introduced above, or it may be a representation learning algorithm based on a specific algorithm. Taking the representation learning model as an example, the input of the representation learning model is the original high-dimensional data, and the output is a low-dimensional representation of the original data. The training method of the representation learning model can be determined in combination with the representation learning algorithm, and the embodiments of the present application do not specifically limit this. Taking the representation learning model based on the VAE model as an example, the input and output of the VAE model are both original data, and the loss function can adopt the VAE standard loss function (reconstruction loss and distribution assumption loss) to train the VAE model. Delete the decoder of the VAE model, and the resulting encoder is an ideal representation learning model. The input of the encoder is the original data, and the output is a low-dimensional expression of the original data. The trained representation learning model can be deployed to an online device. The representation learning model is insensitive to changes in data distribution, so the representation learning model is no longer trained and updated online after deployment.

[0168] Representation learning model inference can be performed after model deployment. During model inference, high-dimensional inference data can be input into the representation learning model to infer a low-dimensional representation of the original high-dimensional data.

[0169] Downstream task models can be, for example, the AI ​​models described above, such as the CSI feedback model. Offline pre-training of downstream task models can be performed based on business needs. The objective function and model structure are designed based on the business needs. This pre-training is done offline using the low-dimensional representation data obtained through inference in the representation learning module. After training is complete, the downstream task model can be deployed to online devices.

[0170] Online training of downstream task models can be based on offline training, continuously utilizing new data to complete online training of downstream task models, resulting in a model that better matches the current data distribution. The online training dataset can be a low-dimensional representation dataset obtained after inference by the representation learning module.

[0171] Downstream task model inference can refer to inputting inference data into a trained model to obtain the model's expected output. This inference data can be low-dimensional representation data obtained by inference of online inference data through the representation learning module.

[0172] Example 2

[0173] Referring to Figure 16, in step S1602, the network device uses dataset 1 (i.e., the offline dataset) to train a representation learning model. The representation learning model may be the second model described above. The representation learning model may, for example, include an encoder in a VAE model.

[0174] In step S1604, after the representation learning model is trained, the network device inputs the data in dataset 1 into the trained representation learning model, inferring low-dimensional representations of each data point, thereby obtaining dataset 2. Dataset 2 can be understood as a low-dimensional representation of dataset 1. Compared to the data in dataset 1, the data in dataset 2 has a significantly lower dimensionality.

[0175] In step S1606, the network device can use Dataset 1 and Dataset 2 to train CSI feedback model 1. This CSI feedback model 1 can be the AI ​​model described above. The network device uses the data in Dataset 2 as input and the data in Dataset 1 as output to train CSI feedback model 1. CSI feedback model 1 includes an AI encoder and an AI decoder. However, the AI ​​encoder in CSI feedback model 1 does not directly encode the CSI data to be fed back into coded data. Instead, it encodes low-dimensional representation data of the CSI data to be fed back. Because Dataset 2 is a low-dimensional representation of Dataset 1, the AI ​​encoder in CSI feedback model 1 is a lightweight model.

[0176] After offline training is completed, network devices can perform online training.

[0177] In step S1608, if the network device detects that the terminal device has connected to the network device and receives the second indication information, the network device may perform online training on the CSI feedback model. The second indication information is used to instruct the network device to perform online training on the CSI feedback model. The second indication information may, for example, be a service indication that triggers CSI feedback. In some embodiments, the network device may perform online training on the CSI feedback model after completing preparatory work such as online data.

[0178] In step S1610, the network device inputs dataset 3 (also called the online dataset) into the representation learning model, and infers low-dimensional representation data for each data point in dataset 3, resulting in dataset 4. Compared to the data in dataset 3, the dimensionality of the data in dataset 4 is significantly reduced.

[0179] In step S1612, similar to step S1606, the network device can use dataset 4 as input and dataset 3 as output to update CSI feedback model 1, resulting in CSI feedback model 2. During the CSI feedback model update process, the structure of the CSI feedback model is not readjusted; only the model parameters are updated. Therefore, the structure and size of CSI feedback model 1 and CSI feedback model 2 are consistent, differing only in the model parameters. Furthermore, since the AI ​​encoder of CSI feedback model 2 encodes the low-dimensional representation of the CSI data, the AI ​​encoder of the CSI feedback model is also a lightweight network model.

[0180] When the network device updates the CSI feedback model, it can continuously update the model as real-time data continues to arrive. In an embodiment of the present application, the data collection model can continuously send CSI data to the network device, and the CSI data will be directly converted into low-dimensional data through step S1610. When the number of samples in the online data set 4 meets the preset number (such as 16, 32, 64 or 128, etc.) or the waiting time meets the preset waiting time (such as 5 time slots, 10 time slots or 20 time slots, etc.), it will trigger the network device to execute step S1612 again to complete the update of the CSI feedback model.

[0181] In step S1614, the network device may send the AI ​​encoder representing the learning model and the CSI feedback model 2 to the terminal device via air interface resources, so that the terminal device completes the deployment of the model.

[0182] After the terminal device completes model deployment, it can start executing model reasoning. The terminal device and network equipment can jointly complete the task of CSI feedback.

[0183] In step S1616, the terminal device performs channel measurement to obtain CSI data to be fed back. The terminal device inputs the CSI data to be fed back into the representation learning model to obtain low-dimensional representation data of the CSI data to be fed back.

[0184] In step S1618, the terminal device infers the low-dimensional representation data through the AI ​​encoder of the CSI feedback model 2 to obtain encoded data.

[0185] In step S1620, the terminal device reports the encoded data to the network device through the air interface resources.

[0186] In step S1622, the network device obtains the AI ​​decoder corresponding to the terminal device, that is, the AI ​​decoder of CSI feedback model 2. The network device uses the AI ​​decoder of CSI feedback model 2 to infer the encoded data and decode the original CSI data.

[0187] It should be noted that the online training and inference processes of the CSI feedback model can be performed simultaneously. The network device can use CSI data to infer the CSI feedback model, and can also use CSI data to update the CSI feedback model. When performing inference, the network device can always use the latest CSI feedback model for inference.

[0188] Example 3

[0189] Referring to Figure 17, in step S1702, the network device uses dataset 1 (i.e., the offline dataset) to train a representation learning model. The representation learning model may be the second model described above. The representation learning model may, for example, include an encoder in a VAE model.

[0190] In step S1704, after the representation learning model is trained, the network device inputs the data in Dataset 1 into the trained representation learning model to infer low-dimensional representations of each data point, thereby obtaining Dataset 2. Dataset 2 can be understood as a low-dimensional representation of Dataset 1. Compared to the data in Dataset 1, the data in Dataset 2 has a significantly lower dimensionality.

[0191] In step S1706, the network device can use Dataset 1 and Dataset 2 to train CSI feedback model 1. This CSI feedback model 1 can be the AI ​​model described above. The network device uses the data in Dataset 2 as input and the data in Dataset 1 as output to train CSI feedback model 1. CSI feedback model 1 includes an AI encoder and an AI decoder. However, the AI ​​encoder in CSI feedback model 1 does not directly encode the CSI data to be fed back into coded data. Instead, it encodes low-dimensional representation data of the CSI data to be fed back. Because Dataset 2 is a low-dimensional representation of Dataset 1, the AI ​​encoder in CSI feedback model 1 is a lightweight model.

[0192] After offline training is completed, online training is performed next.

[0193] In step S1708, the network device recognizes that the terminal device has accessed the network device and receives third indication information. The third indication information is used to instruct online training of the CSI feedback model, or the third indication information is used to instruct the network device to send a representation learning model or a CSI feedback model to the terminal device. The third indication information may be, for example, a service indication that triggers CSI feedback.

[0194] In step S1710, the network device sends the AI ​​encoder representing the learning model and CSI feedback model 1 to the terminal device. The terminal device can collect online data to obtain an online dataset 3. The online dataset 3 can include a single sample or a batch of samples.

[0195] In step S1712, the terminal device inputs dataset 3 into the representation learning model, and can infer low-dimensional representation data of the data in dataset 3, that is, dataset 4. Compared with the data in dataset 3, the dimension of the data in dataset 4 is greatly reduced.

[0196] In step S1714, similar to step S1706, the terminal device can use dataset 4 as input and dataset 3 as output to update CSI feedback model 1, resulting in CSI feedback model 2. During the CSI feedback model update process, the structure of the CSI feedback model is not readjusted; only the model parameters are updated. Therefore, the structure and size of CSI feedback model 1 and CSI feedback model 2 are consistent, differing only in the model parameters. Furthermore, since the AI ​​encoder portion of CSI feedback model 2 encodes the low-dimensional representation of the CSI data, the AI ​​encoder of CSI feedback model 2 is also a lightweight network model.

[0197] To reduce air interface overhead, the terminal device can fix the parameters of the decoder in CSI feedback model 1 during the update training process of CSI feedback model 1. That is, the online training process only updates the parameters of the AI ​​encoder, and does not update the parameters of the AI ​​decoder. In this way, after the terminal device completes the online training of the CSI feedback model, it is no longer necessary to send the updated AI decoder to the terminal device, thereby reducing air interface overhead. In other words, the parameters of the AI ​​decoder in CSI feedback model 1 are the same as those of the AI ​​decoder in CSI feedback model 2.

[0198] Furthermore, the AI ​​decoder in this CSI feedback model can be adapted to the AI ​​encoders in multiple terminal devices. This means that the AI ​​encoders in different terminal devices can correspond to the same AI decoder. This allows network devices to maintain fewer AI decoders, such as just one, which can be used to decode encoded data sent by multiple terminal devices.

[0199] Because it is online learning, as long as the terminal device is online, the terminal device will continuously generate CSI data. The terminal device can convert the newly generated data into low-dimensional representation data through step S1712. When the number of samples in the online dataset 4 meets the preset number (such as 16, 32, 64, or 128, etc.) or the waiting time meets the preset waiting time (such as 5 time slots, 10 time slots, or 20 time slots, etc.), it will trigger the terminal device to execute step S1714 again to complete the update of the CSI feedback model.

[0200] After the model is trained and updated, it can start reasoning. The terminal device and network equipment can jointly complete the task of CSI feedback.

[0201] In step S1716, the terminal device performs channel measurement to obtain CSI data to be fed back. The terminal device inputs the CSI data to be fed back into the representation learning model to obtain low-dimensional representation data of the CSI data to be fed back.

[0202] In step S1718, the terminal device infers the low-dimensional representation data through the AI ​​encoder of the CSI feedback model 2 to obtain encoded data.

[0203] In step S1720, the terminal device reports the encoded data to the network device through the air interface resources.

[0204] In step S1722, the network device obtains the AI ​​decoder corresponding to the terminal device, that is, the AI ​​decoder of CSI feedback model 2. The network device uses the AI ​​decoder of CSI feedback model 2 to infer the encoded data and decode the original CSI data.

[0205] It should be noted that the online training and inference processes of the CSI feedback model can be performed simultaneously. The terminal device can use the CSI data to be fed back to perform inference on the CSI feedback model, or it can use the CSI data to be fed back to update the CSI feedback model. When performing inference, the terminal device can always use the latest CSI feedback model for inference.

[0206] The method embodiment of the present application is described in detail above in conjunction with Figures 1 to 17 , and the device embodiment of the present application is described in detail below in conjunction with Figures 18 to 22 . It should be understood that the description of the method embodiment corresponds to the description of the device embodiment, and therefore, for portions not described in detail, reference can be made to the above method embodiment.

[0207] FIG18 is a schematic structural diagram of a training device provided in an embodiment of the present application. The training device 1800 shown in FIG18 can be any of the first devices described above. The training device 1800 can include a generating unit 1810 and a training unit 1820.

[0208] The generating unit 1810 is configured to generate a second data set based on the first data set, wherein the data in the second data set is low-dimensional representation data of the data in the first data set.

[0209] The training unit 1820 is configured to train a first model for wireless communication based on the second data set.

[0210] In some embodiments, the generating unit 1810 is configured to: train a second model based on the first data set; and process the first data set using the second model to generate the second data set.

[0211] In some embodiments, the second model comprises an encoder in a VAE model.

[0212] In some embodiments, the training unit 1820 is used to: use the second data set as the input of the first model to obtain the output result of the first model; and train the first model using the difference between the output result of the first model and the label data of the first model.

[0213] In some embodiments, the first model includes a codec model, and the label data of the first model is data in the first data set.

[0214] In some embodiments, the first model comprises a CSI feedback model.

[0215] FIG19 is a schematic structural diagram of an apparatus for using a model according to an embodiment of the present application. The apparatus 1900 for using a model shown in FIG19 can be any of the first devices described above. The apparatus 1900 can include a generating unit 1910 and a processing unit 1920.

[0216] The generating unit 1910 is configured to generate second data based on the first data, wherein the second data is low-dimensional representation data of the first data.

[0217] The processing unit 1920 is configured to obtain a processing result of the first model based on the second data and the first model for wireless communication.

[0218] In some embodiments, the generating unit 1910 is configured to: process the first data using a second model to generate the second data.

[0219] In some embodiments, the second model comprises an encoder in a VAE model.

[0220] In some embodiments, the first model comprises a CSI feedback model.

[0221] FIG20 is a schematic structural diagram of a terminal device provided in an embodiment of the present application. The terminal device 2000 shown in FIG20 can be any of the terminal devices described above. The terminal device 2000 can include a receiving unit 2010.

[0222] The receiving unit 2010 is used to receive a first model and a second model from a network device; wherein the second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than that of the first data, and the first model is used to process the second data.

[0223] In some embodiments, the receiving unit 2010 is also used to receive the AI ​​decoder in the first model from the network device; the terminal device 2000 also includes: a processing unit 2020, used to process the first data using the second model to generate the second data; a training unit 2030, used to train the first model using the second data.

[0224] In some embodiments, the first model includes an AI encoder and an AI decoder, and the training unit 2030 is used to fix the parameters of the AI ​​decoder and train the AI ​​encoder using the second data.

[0225] In some embodiments, the receiving unit 2010 is further used to: receive an updated first model from the network device, where the updated first model is obtained by training the AI ​​encoder using the second data.

[0226] In some embodiments, the terminal device 2000 further includes: a processing unit 2020, configured to process the first data using the second model to generate the second data; and the processing unit 2020, further configured to process the second data using the first model to obtain a processing result of the first model.

[0227] In some embodiments, the first model includes an AI encoder, and the processing result of the first model is encoded data. The terminal device 2000 also includes: a sending unit 2040, which is used to send the encoded data to the network device.

[0228] In some embodiments, the second model comprises an encoder in a VAE model.

[0229] In some embodiments, the first model comprises a CSI feedback model.

[0230] Figure 21 is a schematic structural diagram of a network device provided in an embodiment of the present application. The network device 2100 shown in Figure 21 can be any of the network devices described above. The network device 2100 can include a sending unit 2110.

[0231] The sending unit 2110 is used to send a first model and a second model to the terminal device; wherein the second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than the first data, and the first model is used to process the second data.

[0232] In some embodiments, the network device 2100 also includes: a processing unit 2120, used to process the first data using the second model to generate the second data; an updating unit 2130, used to update the first model using the second data to obtain an updated first model; the sending unit 2110 is also used to: send the updated first model to the terminal device.

[0233] In some embodiments, the first model includes an AI encoder, and the network device 2100 further includes: a receiving unit 2140, for receiving encoded data from the terminal device, wherein the encoded data is obtained by the AI ​​encoder processing the second data; and a processing unit 2120, for processing the encoded data using an AI decoder to generate the first data.

[0234] In some embodiments, the second model comprises an encoder in a VAE model.

[0235] In some embodiments, the first model comprises a CSI feedback model.

[0236] Figure 22 is a schematic diagram of a device according to an embodiment of the present application. The dashed lines in Figure 22 indicate that the unit or module is optional. Device 2200 may be used to implement the method described in the above method embodiment. Device 2200 may be a chip, a first device, a terminal device, or a network device.

[0237] The device 2200 may include one or more processors 2210. The processor 2210 may support the device 2200 to implement the method described in the above method embodiment. The processor 2210 may be a general-purpose processor or a special-purpose processor. For example, the processor may be a central processing unit (CPU). Alternatively, the processor may be another general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, etc. The general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0238] The apparatus 2200 may further include one or more memories 2220. The memories 2220 store programs that can be executed by the processor 2210, causing the processor 2210 to perform the methods described in the above method embodiments. The memories 2220 may be independent of the processor 2210 or integrated into the processor 2210.

[0239] The apparatus 2200 may further include a transceiver 2230. The processor 2210 may communicate with other devices or chips via the transceiver 2230. For example, the processor 2210 may transmit and receive data with other devices or chips via the transceiver 2230.

[0240] The present application also provides a computer-readable storage medium for storing a program. The computer-readable storage medium can be applied to a terminal or network device provided in the present application, and the program enables a computer to execute the method performed by the terminal or network device in each embodiment of the present application.

[0241] The present application also provides a computer program product. The computer program product includes a program. The computer program product can be applied to a terminal or network device provided in the present application, and the program causes a computer to execute the method performed by the terminal or network device in each embodiment of the present application.

[0242] The embodiments of the present application also provide a computer program. The computer program can be applied to the terminal or network device provided in the embodiments of the present application, and the computer program enables a computer to execute the method performed by the terminal or network device in each embodiment of the present application.

[0243] It should be understood that the terms "system" and "network" in this application can be used interchangeably. In addition, the terms used in this application are only used to explain the specific embodiments of this application and are not intended to limit this application. The terms "first", "second", "third", and "fourth" in the specification and claims of this application and the accompanying drawings are used to distinguish different objects rather than to describe a specific order. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions.

[0244] In the embodiments of this application, the term "indication" may refer to a direct indication, an indirect indication, or an indication of an association. For example, "A indicates B" may refer to a direct indication of B, e.g., B can obtain information through A; it may refer to an indirect indication of B, e.g., A indicates C, e.g., B can obtain information through C; or it may refer to an association between A and B.

[0245] In the embodiment of the present application, "B corresponding to A" means that B is associated with A and B can be determined based on A. However, it should be understood that determining B based on A does not mean determining B based solely on A, but B can also be determined based on A and / or other information.

[0246] In the embodiments of the present application, the term "corresponding" may indicate a direct or indirect correspondence between the two, or an association relationship between the two, or a relationship between indication and indication, configuration and configuration, etc.

[0247] In the embodiments of the present application, "pre-definition" or "pre-configuration" may be implemented by pre-storing corresponding codes, tables, or other methods that can be used to indicate relevant information in a device (e.g., a terminal device and a network device). The present application does not limit the specific implementation method. For example, pre-definition may refer to information defined in a protocol.

[0248] In the embodiments of the present application, the “protocol” may refer to a standard protocol in the communications field, for example, it may include an LTE protocol, an NR protocol, and related protocols used in future communication systems, and the present application does not limit this.

[0249] In the embodiments of this application, the term "and / or" is simply a description of the association relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this document generally indicates that the related objects are in an "or" relationship.

[0250] In various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean 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 the present application.

[0251] In the several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0252] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0253] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0254] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. 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 computer-readable storage medium. For example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that can be read by a computer or a data storage device such as a server or data center that includes one or more available media integrated therein. The available medium may be a magnetic medium (eg, a floppy disk, a hard disk, a magnetic tape), an optical medium (eg, a digital versatile disc (DVD)), or a semiconductor medium (eg, a solid state disk (SSD)).

[0255] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A training method, characterized in that: include: The first device generates a second data set according to the first data set, wherein the data in the second data set is low-dimensional representation data of the data in the first data set; The first device trains a first model for wireless communication based on the second data set.

2. The method according to claim 1, characterized in that The first device generates a second data set according to the first data set, including: The first device trains a second model according to the first data set; The first device processes the first data set using the second model to generate the second data set.

3. The method according to claim 2, characterized in that The second model includes an encoder in a variational autoencoder (VAE) model.

4. The method according to any one of claims 1 to 3, characterized in that The first device trains a first model for wireless communication according to the second data set, including: The first device uses the second data set as input of the first model to obtain an output result of the first model; The first device trains the first model by using the difference between the output result of the first model and the label data of the first model.

5. The method according to any one of claims 1 to 4, characterized in that The first model includes a coding model, and the label data of the first model is the data in the first data set.

6. The method according to any one of claims 1 to 5, characterized in that The first model includes a channel state information CSI feedback model.

7. A method of using a model, characterized in that include: The first device generates second data according to the first data, wherein the second data is low-dimensional representation data of the first data; The first device obtains a processing result of the first model according to the second data and the first model for wireless communication.

8. The method according to claim 7, characterized in that The first device generates second data according to the first data, including: The first device processes the first data using a second model to generate the second data.

9. The method according to claim 8, characterized in that The second model includes an encoder in a variational autoencoder (VAE) model.

10. The method according to any one of claims 7 to 9, characterized in that: The first model includes a channel state information CSI feedback model.

11. A wireless communication method, characterized in that: include: The terminal device receives the first model and the second model from the network device; The second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than that of the first data, and the first model is used to process the second data.

12. The method according to claim 11, characterized in that The method further comprises: The terminal device processes the first data using the second model to generate the second data; The terminal device trains the first model using the second data.

13. The method according to claim 12, characterized in that The first model includes an AI encoder and an AI decoder, and the terminal device performs online training on the first model using the second data, including: The terminal device fixes the parameters of the AI ​​decoder and uses the second data to train the AI ​​encoder.

14. The method according to claim 11, characterized in that The method further comprises: The terminal device receives the updated first model from the network device, where the updated first model is obtained by training the first model using the second data.

15. The method according to any one of claims 11 to 14, characterized in that The method further comprises: The terminal device processes the first data using the second model to generate the second data; The terminal device processes the second data using the first model to obtain a processing result of the first model.

16. The method according to claim 15, characterized in that The first model includes an AI encoder, and a processing result of the first model is encoded data. The method further includes: The terminal device sends the encoded data to the network device.

17. The method according to any one of claims 11 to 16, characterized in that The second model includes an encoder in a variational autoencoder (VAE) model.

18. The method according to any one of claims 11 to 17, characterized in that The first model includes a channel state information CSI feedback model.

19. A wireless communication method, characterized in that: include: The network device sends the first model and the second model to the terminal device; The second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than that of the first data, and the first model is used to process the second data.

20. The method according to claim 19, characterized in that The method further comprises: The network device processes the first data using the second model to generate the second data; The network device updates the first model using the second data to obtain an updated first model; The network device sends the updated first model to the terminal device.

21. The method according to claim 19 or 20, characterized in that The first model includes an AI encoder, and the method further includes: The network device receives encoded data from the terminal device, where the encoded data is obtained by processing the second data by the AI ​​encoder; The network device processes the encoded data using an AI decoder to generate the first data.

22. The method according to any one of claims 19 to 21, characterized in that The second model includes an encoder in a variational autoencoder (VAE) model.

23. The method according to any one of claims 19 to 22, characterized in that The first model includes a channel state information CSI feedback model.

24. A training device, characterized in that: include: A generating unit, generating a second data set according to the first data set, wherein the data in the second data set is low-dimensional representation data of the data in the first data set; A training unit is used to train a first model for wireless communication according to the second data set.

25. The device according to claim 24, characterized in that The generating unit is used for: Training a second model based on the first data set; The first data set is processed using the second model to generate the second data set.

26. The device according to claim 25, characterized in that The second model includes an encoder in a variational autoencoder (VAE) model.

27. The device according to any one of claims 24 to 26, characterized in that The training unit is used for: Using the second data set as input of the first model to obtain an output result of the first model; The first model is trained by utilizing the difference between the output result of the first model and the label data of the first model.

28. The device according to any one of claims 24 to 27, characterized in that The first model includes a coding model, and the label data of the first model is the data in the first data set.

29. The device according to any one of claims 24 to 28, characterized in that The first model includes a channel state information CSI feedback model.

30. A device for using a model, characterized in that: include: A generating unit, configured to generate second data according to the first data, wherein the second data is low-dimensional representation data of the first data; A processing unit is used to obtain a processing result of the first model based on the second data and the first model for wireless communication.

31. The device according to claim 30, characterized in that The generating unit is used for: The first data is processed using a second model to generate the second data.

32. The device according to claim 31, characterized in that The second model includes an encoder in a variational autoencoder (VAE) model.

33. The device according to any one of claims 30 to 32, characterized in that The first model includes a channel state information CSI feedback model.

34. A terminal device, characterized in that: include: A receiving unit, configured to receive a first model and a second model from a network device; The second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than that of the first data, and the first model is used to process the second data.

35. The terminal device according to claim 34, characterized in that: The terminal device further includes: a processing unit, configured to process the first data using the second model to generate the second data; A training unit is used to train the first model using the second data.

36. The terminal device according to claim 35, characterized in that: The first model includes an AI encoder and an AI decoder, and the training unit is used to: The parameters of the AI ​​decoder are fixed, and the AI ​​encoder is trained using the second data.

37. The terminal device according to claim 34, characterized in that: The receiving unit is also used for: An updated first model is received from the network device, where the updated first model is obtained by training the AI ​​encoder using the second data.

38. The terminal device according to any one of claims 34 to 37, characterized in that: The terminal device further includes: a processing unit, configured to process the first data using the second model to generate the second data; The processing unit is further used to process the second data using the first model to obtain a processing result of the first model.

39. The terminal device according to claim 38, characterized in that: The first model includes an AI encoder, a processing result of the first model is encoded data, and the terminal device further includes: A sending unit is used to send the encoded data to the network device.

40. The terminal device according to any one of claims 34 to 39, characterized in that: The second model includes an encoder in a variational autoencoder (VAE) model.

41. The terminal device according to any one of claims 34 to 40, characterized in that: The first model includes a channel state information CSI feedback model.

42. A network device, characterized in that: include: A sending unit, configured to send the first model and the second model to a terminal device; The second model is used to convert the first data of the terminal device into second data, the dimension of the second data is lower than that of the first data, and the first model is used to process the second data.

43. The network device according to claim 42, characterized in that: The network device also includes: a processing unit, configured to process the first data using the second model to generate the second data; an updating unit, configured to update the first model using the second data to obtain an updated first model; The sending unit is further used to: send the updated first model to the terminal device.

44. The network device according to claim 42 or 43, characterized in that: The first model includes an AI encoder, and the network device further includes: A receiving unit, configured to receive encoded data from the terminal device, wherein the encoded data is obtained by processing the second data by the AI ​​encoder; A processing unit is used to process the encoded data using an AI decoder to generate the first data.

45. The network device according to any one of claims 42 to 44, characterized in that: The second model includes an encoder in a variational autoencoder (VAE) model.

46. ​​The network device according to any one of claims 42 to 45, characterized in that: The first model includes a channel state information CSI feedback model.

47. A device, characterized in that The invention comprises a memory and a processor, wherein the memory is used to store a program, and the processor is used to call the program in the memory to execute the method according to any one of claims 1 to 23.

48. A device, characterized in that The invention comprises a processor, which is used to call a program from a memory to execute the method according to any one of claims 1 to 23.

49. A chip, characterized in that: It comprises a processor, which is used to call a program from a memory, so that a device equipped with the chip executes a method as claimed in any one of claims 1 to 23.

50. A computer-readable storage medium, characterized in that A program is stored thereon, the program causing a computer to execute the method according to any one of claims 1 to 23.

51. A computer program product, characterized in that The method comprises a program which causes a computer to execute the method according to any one of claims 1 to 23.

52. A computer program, characterized in that The computer program enables a computer to execute the method according to any one of claims 1 to 23.

Images