Communication method and apparatus

Abstract

The present application relates to the field of communications. Provided are a communication method and apparatus. The communication method comprises: after sending, to a network device, first information for requesting a model, a first terminal receiving from the network device a model data packet of a pre-trained first model matching a model label of the model requested by the first terminal, wherein the model data packet comprises model parameters of the first model, and also comprises a model label of the first model. In this way, a first terminal does not need to train a first model, and the first terminal also does not need to frequently acquire, from a network device, training data for training the first model, thereby reducing network communication overheads and the computing overheads of the first terminal.

Description

Communication method and apparatus

[0001] The present application claims priority to the Chinese Patent Application No. 202410897957.4, filed on July 4, 2024, and entitled "Communication method and apparatus", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to the field of communication, in particular to a communication method and apparatus. BACKGROUND

[0003] Physical layer artificial intelligence (AI) in future networks refers to a technology that applies AI techniques to the physical layer modules of communication devices. The physical layer is the lowest layer of a communication device, and is responsible for data transmission and reception, including encoding, modulation, demodulation, decoding, channel estimation, and other processes. The introduction of physical layer AI aims to improve the performance and efficiency of physical layer modules through the data processing and learning capabilities of AI technology. Physical layer AI can be applied to various functional units in the physical layer, such as encoding, modulation, demodulation, and channel estimation units, to optimize these functional units through AI technology and improve the performance and efficiency of the physical layer modules.

[0004] For example, in channel estimation, physical layer AI can use AI algorithms to more accurately predict and estimate channel states, thereby improving the transmission quality and efficiency of the communication system. In the application scenarios of physical layer AI, a neural network (NN) model is usually trained with a large amount of data in an offline environment to ensure that the NN model can efficiently and accurately operate in the inference stage.

[0005] To ensure that the NN model can accurately reflect the current environmental state, the physical layer AI needs to update the model parameters of the NN model as the environment changes, and then perform accurate channel prediction and estimation based on the updated NN model to provide optimal communication performance based on the estimated channel state. However, the update of the model parameters of the NN model requires additional computational resources and time, resulting in large network communication overhead and device computational overhead. SUMMARY

[0006] The present application provides a communication method and apparatus for reducing network communication overhead and device computational overhead in the application scenarios of physical layer AI.

[0007] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0008] In a first aspect, a communication method is provided. The method is applied to a first terminal. The execution subject of the method can be the first terminal, a component or device (e.g., a processor, a chip, or a chip system) applied to the first terminal, or a logic module or software capable of realizing all or part of the functions of the first terminal. The communication method comprises: after sending first information for requesting a model, receiving a model data packet of a pre-trained first model matching a model tag requested by the first terminal. The model data packet comprises model parameters of the first model, and can also comprise the model tag of the first model.

[0009] In the first aspect, the network device provides the first terminal with a model data packet of the first model corresponding to the model requested by the first terminal, without the first terminal training the first model or the first terminal frequently obtaining training data for training the first model from the network device, thereby reducing network communication overhead and the computing overhead of the first terminal.

[0010] In a possible design, the sending of the first information comprises: sending the first information to the network device; and the receiving of the model data packet of the first model comprises: receiving, by the network device, the model data packet of the first model from a server. The server stores a model tag of a second model trained by a second terminal and model parameters of the second model. The first terminal and the second terminal are the same terminal, or the first terminal and the second terminal are different terminals, and the second model comprises the first model.

[0011] In this design, first, through centralized storage and management of the server, the model tag of the second model and the model parameters of the second model are efficiently and securely shared with multiple network devices, improving data utilization and access efficiency. Second, the high-performance processing capability of the server ensures rapid response to requests for the model tag of the second model and the model parameters of the second model, supports high-concurrency access, and meets the access requirements of large-scale network devices. In addition, the data backup and recovery mechanism of the server guarantees the security and reliability of the model tag of the second model and the model parameters of the second model, and flexible permission control ensures the security and confidentiality of the data. Finally, the scalability and maintainability of the server provide good expansion space and maintenance convenience for the system, further improving the stability and availability of the entire network system.

[0012] In a possible design, the model tag of the second model comprises or is used to indicate at least one of the following: a physical layer module to which the second model is applicable, a collection time of training data of the second model, a collection location of the training data of the second model, a type of a terminal training the second model, a type of a service corresponding to the training of the second model, a type of the second model, a size of the second model, or a performance score of the second model.

[0013] In the design, a model label of a plurality of second models is designed, and based on the model label, it can be determined whether the second model can be reused.

[0014] In a possible design, the method can further include: receiving first test data; running the first model according to the first test data and the model parameters of the first model to obtain a performance indicator of the first model; when the performance indicator indicates that the first model meets a model updating condition, sending second information used to request an updated model; and receiving an updated model data packet of the first model corresponding to the second information. Optionally, the method can further include: sending third information used to request the first test data, and the third information can be sent when a preset condition is met; wherein the preset condition includes receiving fourth information indicating that the third information is to be sent, or the preset condition includes reaching a preset period.

[0015] In the design, by introducing the performance indicator of the first model, when the performance indicator of the first model meets the model updating condition, the updated model data packet of the first model is requested in time, so that the model of the terminal can be accurately and timely updated. Moreover, the third information used to request the first test data is sent, and the sending of the request is controlled based on the preset condition (such as receiving the fourth information or reaching the preset period), so that the flexibility of the communication system can be improved, resource utilization can be optimized, real-time performance can be enhanced, and predictability and controllability can be improved.

[0016] In a possible design, the first information is sent in the following manner: when accessing the network, the first information is sent; or before initiating a preset service, the first information is sent.

[0017] In the design, two time points for sending the first information are given, and the sending of the first information can be flexibly implemented.

[0018] In a second aspect, a communication method is provided, which is applied to a first network device. The execution subject of the method can be the first network device, a component or apparatus (such as a processor, a chip, or a chip system) applied to the first network device, or a logic module or software capable of realizing all or part of the functions of the first network device. The communication method includes: receiving first information used to request a model from a first terminal, obtaining a model data packet of a first model corresponding to the first information; sending the model data packet of the first model, the first model being a model in pre-trained models that matches a model label requested by the first terminal, the model data packet including model parameters of the first model; or the model data packet including the model parameters of the first model and a model label of the first model; and receiving the model data packet of the first model.

[0019] In the second aspect, the network device provides the first terminal with the model data packet of the first model corresponding to the model requested by the first terminal, without the first terminal training the first model and the first terminal frequently obtaining training data for training the first model from the network device, thereby reducing network communication overhead and computation overhead of the first terminal.

[0020] In a possible design, the model data packet of the first model corresponding to the first information is obtained by: sending fourth information to a server, the fourth information being determined according to the first information, the fourth information being used to indicate a model label of the first model, the server storing a model label of a second model trained by a second terminal and model parameters of the second model, the first terminal and the second terminal being the same terminal, or the first terminal and the second terminal being different terminals, the second model including the first model; and receiving the model data packet of the first model from the server.

[0021] In this design, first, through centralized storage and management of the server, the model label of the second model and the model parameters of the second model are efficiently and securely shared to multiple network devices, improving data utilization and access efficiency. Second, the high-performance processing capability of the server ensures rapid response to requests for the model label of the second model and the model parameters of the second model, supports high-concurrency access, and meets access requirements of large-scale network devices. In addition, the data backup and recovery mechanism of the server guarantees the security and reliability of the model label of the second model and the model parameters of the second model, and flexible permission control ensures the security and confidentiality of data. Finally, the scalability and maintainability of the server provide good expansion space and maintenance convenience for the system, further improving the stability and availability of the entire network system.

[0022] In a possible design, the model label of the second model includes or is used to indicate at least one of the following: a physical layer unit to which the second model is applicable, a collection time of training data of the second model, a collection location of the training data of the second model, a type of a terminal that trains the second model, a type of a service corresponding to training of the second model, a type of the second model, a size of the second model, or a performance score of the second model.

[0023] In this design, various model labels of the second model are designed, and whether the second model can be reused can be determined based on the model label.

[0024] In a possible design, the method can further include: sending first test data; receiving second information, wherein the performance indicator of the first model indicates that the first model meets a model update condition, the performance indicator of the first model being obtained by running the first model based on the first test data and the model parameters of the first model; obtaining an update model data packet of the first model corresponding to the second information; and sending the update model data packet of the first model.

[0025] Optionally, the first test data can be sent after receiving the third information for requesting the first test data. Alternatively, the first test data can also be sent according to a preset period, thereby enhancing the flexibility and adaptability of the communication system.

[0026] In this design, by introducing the performance indicator of the first model, the update model data packet of the first model is requested in time in the case that the performance indicator of the first model meets the model update condition, so that the model of the terminal can be accurately and timely updated. Moreover, the third information for requesting the first test data is sent, and the sending of the request is controlled based on a preset condition (such as receiving the fourth information or reaching a preset period), so that the flexibility of the communication system can be improved, the resource utilization can be optimized, the real-time performance can be enhanced, and the predictability and controllability can be improved.

[0027] In a third aspect, a communication method is provided, which is applied to a server. The execution subject of the method can be the server, a component or device (such as a processor, a chip, or a chip system, etc.) applied to the server, or a logic module or software capable of realizing all or part of the functions of the server. The communication method comprises: receiving, by the server, fourth information indicating a model tag of a first model requested by a first terminal; storing, by the server, a model tag of a second model trained by a second terminal and model parameters of the second model, the first terminal and the second terminal being the same terminal, or the first terminal and the second terminal being different terminals, the second model including the first model; sending a model data packet corresponding to the model tag of the first model, the model data packet including the model parameters of the first model; or the model data packet including the model parameters of the first model and the model tag of the first model.

[0028] In the third aspect, first, through the centralized storage and management of the server, the model tag of the second model and the model parameters of the second model are efficiently and securely shared to multiple network devices, thereby improving the data utilization rate and access efficiency. Second, the high-performance processing capability of the server ensures the rapid response to the request for the model tag of the second model and the model parameters of the second model, supports high-concurrency access, and meets the access requirements of large-scale network devices. In addition, the data backup and recovery mechanism of the server guarantees the security and reliability of the model tag of the second model and the model parameters of the second model, and the flexible permission control ensures the security and confidentiality of the data. Finally, the scalability and maintainability of the server provide good extension space and maintenance convenience for the system, further improving the stability and availability of the entire network system.

[0029] In a possible design, the model label of the second model includes or is used to indicate at least one of the following: a physical layer unit to which the second model is applicable, a collection time of training data of the second model, a collection location of the training data of the second model, a type of a terminal used to train the second model, a type of a service corresponding to the training of the second model, a type of the second model, a size of the second model, or a performance score of the second model.

[0030] In a fourth aspect, a communication apparatus is provided for implementing the method in any one of the first aspect to the third aspect. For example, the communication apparatus can be the first terminal in the first aspect, or an apparatus (for example, a chip or chip system) included in the first terminal; or the communication apparatus can be the network device in the second aspect, or an apparatus (for example, a chip or chip system) included in the network device; or the communication apparatus can be the server in the third aspect, or an apparatus (for example, a chip or chip system) included in the server. When the apparatus is a chip system, the apparatus can be composed of a chip or include a chip and other discrete devices.

[0031] The communication apparatus includes modules, units, or means corresponding to the method, which can be implemented by hardware, software, or by a combination of hardware and software. The hardware or software includes one or more modules or units corresponding to the functions.

[0032] In some possible designs, the communication apparatus can include a processing module and a transceiver module. The processing module can be used to implement the processing functions in any one of the aspects above and any possible implementation manner thereof. The transceiver module, which can also be referred to as a transceiving unit, is used to implement the functions of sending and / or receiving in any one of the aspects above and any possible implementation manner thereof. The transceiver module can be composed of a transceiver circuit, a transceiver, a transceiver, or a communication interface.

[0033] In some possible designs, the transceiver module includes a sending module and / or a receiving module, which are used to implement the functions of sending or receiving in any one of the aspects above and any possible implementation manner thereof.

[0034] In a fifth aspect, a communication apparatus is provided, which comprises: a processor and a communication interface; the communication interface is configured to communicate with a module outside the communication apparatus; the processor is configured to execute a computer program or instructions, so that the communication apparatus performs the method in any one of the aspects. For example, the communication apparatus can be the first terminal in the first aspect, or an apparatus included in the first terminal, such as a chip or chip system; or the communication apparatus can be the network device in the second aspect, or an apparatus included in the network device, such as a chip or chip system; or the communication apparatus can be the server in the third aspect, or an apparatus included in the server, such as a chip or chip system. When the apparatus is a chip system, the apparatus can be composed of a chip, or can include a chip and other discrete devices.

[0035] In a sixth aspect, a communication apparatus is provided, which comprises: at least one processor; the processor is configured to execute a computer program or instructions stored in a memory, so that the communication apparatus performs the method in any one of the aspects. The memory can be coupled with the processor, or the memory can exist independently of the processor, for example, the memory and the processor are two independent modules. The memory can be located outside the communication apparatus, or can be located inside the communication apparatus.

[0036] The communication apparatus is configured to implement the method in any one of the first aspect to the third aspect. For example, the communication apparatus can be the first terminal in the first aspect, or an apparatus included in the first terminal, such as a chip or chip system; or the communication apparatus can be the network device in the second aspect, or an apparatus included in the network device, such as a chip or chip system; or the communication apparatus can be the server in the third aspect, or an apparatus included in the server, such as a chip or chip system. When the apparatus is a chip system, the apparatus can be composed of a chip, or can include a chip and other discrete devices.

[0037] In a seventh aspect, a computer readable storage medium is provided, which stores computer programs or instructions, when the computer programs or instructions are run on a communication apparatus, the communication apparatus can perform the method in any one of the aspects.

[0038] In an eighth aspect, a computer program product is provided, which includes instructions, when the instructions are run on a communication apparatus, the communication apparatus can perform the method in any one of the aspects.

[0039] In a ninth aspect, a communication apparatus is provided, which is configured to perform the method in any one of the aspects.

[0040] It can be understood that, when the communication apparatus provided in any one of the fourth aspect to the sixth aspect is a chip, the sending action / function of the communication apparatus can be understood as outputting information, and the receiving action / function of the communication apparatus can be understood as inputting information.

[0041] The technical effects brought by any one of the fourth aspect to the ninth aspect can be referred to the technical effects brought by different design manners in the first aspect to the third aspect, which will not be repeated here.

[0042] In a tenth aspect, a communication system is provided, which includes the first terminal, the network device and the server according to the above aspects. BRIEF DESCRIPTION OF DRAWINGS

[0043] FIG. 1 is a physical layer AI deployment form diagram according to an embodiment of the present application;

[0044] FIG. 2 is a structure diagram of a communication system according to an embodiment of the present application;

[0045] FIG. 3 is a flow diagram of a communication method according to an embodiment of the present application;

[0046] FIG. 4 is a flow diagram of another communication method according to an embodiment of the present application;

[0047] FIG. 5 is a flow diagram of another communication method according to an embodiment of the present application;

[0048] FIG. 6 is a flow diagram of another communication method according to an embodiment of the present application;

[0049] FIG. 7 is a structure diagram of a communication apparatus according to an embodiment of the present application;

[0050] FIG. 8 is a structure diagram of another communication apparatus according to an embodiment of the present application;

[0051] FIG. 9 is a structure diagram of another communication apparatus according to an embodiment of the present application;

[0052] FIG. 10 is a structure diagram of another communication apparatus according to an embodiment of the present application. DETAILED DESCRIPTION

[0053] The network architecture and service scenarios described in the embodiments of the present application are used to more clearly illustrate the technical solutions of the embodiments of the present application, and do not constitute a limitation on the technical solutions provided by the embodiments of the present application. Those skilled in the art can know that, with the evolution of network architecture and the appearance of new service scenarios, the technical solutions provided by the embodiments of the present application are also applicable to similar technical problems.

[0054] Before introducing the embodiments of the present application, some terms related to the embodiments of the present application are explained.

[0055] AI is a cutting-edge technology that covers a wide range of fields, simulating, extending, and surpassing various aspects of human intelligence. Through techniques such as deep learning, machine learning, natural language processing, and computer vision, AI systems can process and analyze massive amounts of data, recognize patterns, make predictions and decisions, and even possess some degree of creativity and self-learning ability. These capabilities enable AI to play an important role in many fields, such as medicine, transportation, finance, education, and entertainment, not only improving production efficiency but also driving technological innovation, improving people's quality of life, and providing new possibilities for solving global challenges.

[0056] NN is a computational model that simulates the structure and function of biological neural networks, composed of a large number of artificial neurons (nodes) connected to each other. These neurons receive input signals, process them internally (such as weighted summation, activation function processing, etc.), generate output signals, and pass them on to the next layer of neurons. Neural networks learn and adapt to different tasks such as image recognition, speech recognition, and natural language processing by adjusting the connection weights and bias terms between neurons. It can automatically extract features from input data and optimize network parameters through training to achieve efficient classification, regression, or generation tasks. Neural networks have a wide range of applications in artificial intelligence and are an important part of modern machine learning and deep learning technologies.

[0057] Variational Autoencoder (VAE) is a neural network designed specifically for data compression and reconstruction. Unlike traditional autoencoders, VAE uses a probabilistic approach to model and generate data in the compressed latent space. The VAE architecture mainly consists of two parts: the encoder is responsible for mapping input data to a representation in the latent space, while the decoder uses these latent representations to reconstruct the original data. These two components are usually implemented by multi-layer neural networks. VAE optimizes its parameters by minimizing a specific loss function that evaluates both the accuracy of reconstructed data and the matching degree of latent space distribution with prior distribution. The main advantage of VAE is its ease of implementation and training, as well as the effectiveness of the learned compressed data representation. In addition, VAE allows for uncertainty estimation and generates different outputs with probabilistic properties, providing unique support for tasks such as signal classification and joint source-channel coding.

[0058] Variational Autoencoder (VAE) is mainly applied in physical layer AI for its powerful data generation and latent space modeling capabilities. VAE converts input data into latent representations through an encoder and reconstructs data from the latent space through a decoder. This capability is crucial in physical layer AI as it allows systems to extract key information from complex communication signals and represent and transmit these data in a more efficient manner. In physical layer AI, VAE can be used to simulate and generate communication data to evaluate network performance and optimize algorithms. By learning latent representations from actual communication data, VAE can generate synthetic data similar to the distribution of real data, which is very useful for training and testing physical layer AI algorithms. In addition, the latent space modeling capability of VAE also makes it suitable for handling uncertainties in the physical layer. Since communication signals may be affected by noise, interference and channel changes during transmission, understanding and modeling these uncertainties is crucial for improving the performance of physical layer AI systems. VAE models the data distribution in the latent space through a probabilistic approach and can generate data samples with probabilistic properties, thus helping physical layer AI systems better adapt to these complex environments.

[0059] The classification of physical layer AI, as shown in Figure 1, can be divided into the following three categories according to the deployment form:

[0060] The first type, as shown in (a) of Figure 1, is joint deployment of neural network (Tx-Rx Joint NN): In this deployment form, the transmitter (Tx) and receiver (Rx) participate in the training of the neural network model simultaneously. This approach requires another independent physical layer link to assist in training to synchronize and optimize the performance between Tx and Rx. Characteristics: Training overhead is extremely large, as it involves the coordinated training and optimization of both endpoints.

[0061] The second type, as shown in (b) of Figure 1, is receiver-only neural network (Rx-only NN): In this deployment form, the receiver (Rx) is used to train the neural network model. The training overhead of this deployment form is relatively small, and since it only focuses on the processing of the receiver, it usually has good standard compatibility. Characteristics: Training overhead is moderate, and standard compatibility is good, as it only focuses on the optimization of the receiver.

[0062] The third type, as shown in (c) of Figure 1, is transmitter-only neural network (Tx-only NN): In this deployment form, the transmitter (Tx) is used to train the neural network model. Although the training overhead is also moderate, this deployment form is usually only applicable to specific base station modes or specific scenarios, such as channel prediction. Characteristics: Training overhead is moderate, but the application range is limited, as it only focuses on the optimization of the transmitter and may only be applicable to certain specific base station configurations or functions.

[0063] As introduced in the background, the physical layer AI needs to update the model parameters of the NN model as the environment changes, which means that the update of the model parameters of the NN model needs additional computing resources and time, resulting in large network communication overhead and device computing overhead. This is usually caused by the following two reasons. On the one hand, when updating the model parameters of the NN model, the device training the model may need to frequently obtain new training data, which will result in large network communication overhead of transmitting new training data. For example, in the first type of deployment form of the physical layer AI scenario, the sender and the receiver need to transmit new training data when updating the model parameters of the NN model. On the other hand, when updating the model parameters of the NN model, the device training the model needs to train the NN model based on the new training data to update the NN model, which will result in large computing overhead of the device training the model. For example, in any type of physical layer AI scenario, the NN model needs to be retrained when updating the model parameters of the NN model.

[0064] To solve the above technical problems, the embodiment of the present application provides a communication method, which defines the model in a standardized manner from the perspective of whether the physical layer AI model can be standardized and managed. Specifically, the model is marked with a model label, so that when a model usage requirement is generated, it can be determined whether the trained model can be directly reused based on the requirement and the model label, without the need for retraining, thereby reducing the network communication overhead and the computing overhead of the device.

[0065] The method provided by the embodiment of the present application will be described below in conjunction with the accompanying drawings.

[0066] The communication method provided by the embodiment of the present application can be applied to various communication systems, such as a long term evolution (LTE) system, a 5th generation (5G) mobile communication system, a wireless fidelity (WiFi) system, a future communication system, or a system integrating multiple communication systems, etc., and the embodiment of the present application is not limited thereto. The 5G can also be referred to as new radio (NR).

[0067] The communication method provided by the embodiments of the present application can be applied to various communication scenarios, for example, can be applied to one or more of the following communication scenarios: enhanced mobile broadband (eMBB), ultra reliable low latency communication (URLLC), machine type communication (MTC), massive machine type communication (mMTC), device to device (D2D), vehicle to everything (V2X), vehicle to vehicle (V2V), and internet of things (IoT), etc.

[0068] The communication method provided by the embodiments of the present application will be described below taking the communication system shown in FIG. 2 as an example.

[0069] FIG. 2 is a schematic diagram of a communication system provided by an embodiment of the present application, as shown in FIG. 2, the communication system can include:

[0070] a first terminal 210, a network device 220, and a server 230.

[0071] The first terminal 210 is configured to request a model from the network device 220 by means of first information.

[0072] The network device 220 is configured to receive the first information and, in response to the request of the first terminal 210, acquire a model data packet of a first model corresponding to the first information, and then send the model data packet of the first model to the first terminal 210, the first model being a model in pre-trained models whose model label matches the model label requested by the first terminal 210, the model data packet of the first model including model parameters of the first model; or the model data packet including the model parameters of the first model and the model label of the first model.

[0073] In an example, as shown in Table 1, the model data packet includes two types of data, model parameters and model labels. The model parameters are numerical values or variables that describe how the model predicts the output from its input. These parameters are automatically determined by data in the model learning or training process, aiming to minimize the prediction error or loss function of the model. Different model types (such as linear regression, decision tree, neural network, etc.) have different types of parameters. For what the parameters of various model types are, please refer to the relevant technology, which will not be described here. After Table 1, the model label will be described.

[0074] Table 1

[0075] In the model label, the physical layer unit to which the model is applicable includes at least one of an encoding unit, a modulation unit, a demodulation unit, a decoding unit, a channel estimation unit, etc. The physical layer units to which different models are applicable can be the same or different, and are not limited. In a possible implementation, the physical layer unit to which the model is applicable can be corresponded to a model number, for example, number 1 corresponds to an encoding unit, and number 2 corresponds to a modulation unit. The model number is a unique identifier corresponding to the physical layer unit. It ensures that a specific physical layer unit can be accurately and unambiguously indicated within a network or across networks. This number can be based on a certain naming rule or coding system to indicate the physical layer unit to which the model is applicable, or it is simply an incremental serial number. The specific form of the model number is not limited. Recording the physical layer unit to which the model is applicable is very important for determining whether the model can be reused to meet the demand. For example, if the demand is a model applicable to an encoding unit, it can be determined whether there is a second model applicable to the encoding unit based on the second model stored by the server.

[0076] The collection time of the training data of the model depends on the availability of the data and the business demand. For offline training, the data can be collected at any time, but there is usually a fixed time period. The collection time of the training data of the model refers to the exact time point of the model data collection. Exemplarily, the collection time of the training data of the model can be a timestamp, which can be accurate to seconds or milliseconds. Recording the collection time of the training data of the model is very important for determining whether the model can be reused to meet the demand. For example, if the demand is a model trained by training data collected in period A, it can be determined whether there is a second model applicable to period A based on the collection time of the training data of the second model stored by the server.

[0077] The collection location of the training data of the model depends on the location where the data is generated. For example, the collection location of the training data can be a location coordinate with a certain precision, such as city-level precision, county / district-level precision, cell-level precision, global positioning system (GPS) level precision, etc. Recording the collection location of the training data of the model is very important for determining whether the model can be reused to meet the demand. For example, if the demand is a model trained by training data collected in region B, it can be determined whether there is a second model applicable to region B based on the collection location of the training data of the second model stored by the server.

[0078] The type of the terminal training the model can be various computing devices, such as servers of various specifications, etc. These computing devices have strong computing and storage capabilities and can support the training process of the AI model. For example, in the field of autonomous driving, training the AI model of an autonomous vehicle usually requires the use of a server cluster, that is, the type of the terminal training the model is a server cluster. Recording the type of the terminal training the model is very important for determining whether the model can be reused to meet the demand. For example, the demand is a model trained by a C-specification server with strong computing power, and whether there is a second model suitable for the C-specification server can be determined based on the type of the terminal stored by the server for training the second model.

[0079] The business corresponding to the training model can be the business to which the model obtained by training is expected to be applied. For example, the type of the business corresponding to the training model can be enhanced mobile broadband (eMBB), ultra-reliable low-latency communications (URLLC), etc. Recording the business corresponding to the training model is very important for determining whether the model can be reused to meet the demand. For example, the demand is a model for eMBB, and whether there is a second model trained for eMBB can be determined based on the business corresponding to the second model stored by the server for training.

[0080] The type of the model is various, for example, it can be a convolutional neural network (CNN), a recurrent neural network (RNN), a Transformer, etc. Recording the type of the model is very important for determining whether the model can be reused to meet the demand. For example, the demand is a CNN model, and whether there is a second model of the type of CNN can be determined based on the type of the second model stored by the server for training.

[0081] The size of the model is usually measured by the number of its parameters, and the more the number of parameters, the larger the size of the model. For example, large models such as GPT-4 have hundreds of billions of parameters. Recording the size of the model is very important for determining whether the model can be reused to meet the demand. For example, the demand is a model with tens of billions of parameters, and whether there is a second model with tens of billions of parameters can be determined based on the size of the second model stored by the server.

[0082] The performance score of the model is usually based on a series of evaluation criteria, such as accuracy, precision, recall, etc. These evaluation criteria can help comprehensively evaluate the performance of the model. The evaluation criteria used by the embodiments of the present application when scoring the performance of the model are not limited. For example, in an image recognition task, the accuracy of the model can be used to evaluate the performance of the model. If the model can correctly identify 90% of the test images, then the accuracy of the model can be said to be 90%. In addition, other evaluation criteria can also be used to further evaluate the performance of the model, such as the precision and recall of identifying images of different categories, etc. In another example, the performance of the physical layer can also be used as the performance score of the model. For example, the block error rate (BLER) of the physical layer after running the model can be used as the performance score of the model.

[0083] Recording the performance score of the model is very important for determining whether the model can meet the demand for reuse. For example, if the demand is a model with a performance score of D or above, then the server can determine whether there is a second model with a performance score of D based on the performance score of the second model stored by the server.

[0084] It can be understood that the model label described above is an example, and more or fewer model labels can also be used in specific implementation, which is not limited.

[0085] Optionally, the network device 220 can request and receive the model data packet of the first model from the server 230. The server 230 stores the model label of the second model trained by the second terminal and the model parameters of the second model, and the server 230 can determine the model data packet of the first model from the stored data.

[0086] It can be understood that the first terminal 210 and the second terminal can be the same terminal, or the first terminal 210 and the second terminal can also be different terminals, and the second model includes the first model.

[0087] It should be noted that FIG. 2 is only an example framework diagram, and the number of nodes and the state of the terminal included in FIG. 2 are not limited. In addition to the functional nodes shown in FIG. 2, other nodes such as core network devices, gateway devices, etc. can also be included, which are not limited. The network device and the core network device communicate with each other through wired or wireless means, such as through a next generation (NG) interface.

[0088] The network device is mainly used to implement at least one of resource scheduling, radio resource management, and radio resource control of the terminal. Specifically, the network device can include any of a base station, a wireless access point, a transmission receive point (TRP), a transmission point (TP), and some other access node. In a system using different wireless access technologies, the name of the device with the function of the base station can be different, for example, in an LTE system, it is called an evolved NodeB (eNB or eNodeB), in a third generation (3G) communication system, it is called a Node B (Node B), and the like. For convenience of description, in the embodiments of the present application, the device providing a wireless communication function for the terminal is collectively referred to as a network device. In the embodiments of the present application, the device for implementing the function of the network device can be a network device; it can also be a device capable of supporting the network device to implement the function, such as a chip system, which can be installed in the network device or used in matching with the network device.

[0089] The terminal (terminal equipment) can be a user equipment (UE), a mobile station (MS), or a mobile terminal (MT), etc. Specifically, the terminal can be a subscriber unit, a cellular phone, a smart phone, a wireless data card, a personal digital assistant (PDA) computer, a tablet computer, a wireless modem, a handset, a laptop computer, a machine type communication (MTC) terminal, etc. In the embodiments of the present application, the device for implementing the function of the terminal device can be a terminal, and can also be a device capable of supporting the terminal to implement the function, such as a chip system, which can be installed in the terminal or used in matching with the terminal.

[0090] A server (Server) is a high-performance computer used to provide services in a network environment, such as storing, processing, transmitting data or application programs, etc. It usually has higher processing power, larger storage capacity and more stable running environment to support simultaneous access and requests of multiple clients (Client). The server is one of the core devices of network computing, used to process and respond to requests from other devices in the network (such as network devices, terminals, etc.).

[0091] In combination with the above communication system, an embodiment of the present application provides a communication method. FIG. 3 shows a flowchart of the communication method provided by an embodiment of the present application. As shown in FIG. 3, the method can include the following steps:

[0092] S310, the first terminal sends first information to the network device, and correspondingly, the network device receives the first information from the first terminal.

[0093] The first information is used for requesting a model. Optionally, the timing of sending the first information can be when accessing the network; the timing of sending the first information can also be before initiating a preset service. The preset service can be a call service, a video service, etc., and is not limited.

[0094] S320, the network device acquires a model data packet of a first model corresponding to the first information.

[0095] The network device can locally check whether the model data packet of the first model corresponding to the first information is stored, and if found, perform S330. If the model data packet of the first model corresponding to the first information is not found locally, the network device can request the model data packet of the first model from a device storing the model data packet of the first model. In an embodiment, the network device can acquire the model data packet of the first model from a server. The specific description will be given in the embodiment shown in FIG. 4 below, which is not described in detail here.

[0096] S330, the network device sends the model data packet of the first model to the first terminal, and correspondingly, the first terminal receives the model data packet of the first model from the network device.

[0097] The first model is a model whose model label matches the model label requested by the first terminal in the pre-trained model, and the model data packet includes the model parameters of the first model. Optionally, the model data packet can also include the model label of the first model. The model label of the first model can be used to distinguish different models.

[0098] Optionally, if the first terminal has not accessed the network device at this time, the method can further include:

[0099] S340, the first terminal sends an access request to the network device, and correspondingly, the network device receives the access request from the first terminal.

[0100] When the first terminal needs to request a model, the access request can be sent to determine whether the first terminal can access the network device.

[0101] S350, the network device sends first indication information to the first terminal, and correspondingly, the first terminal receives the first indication information from the network device.

[0102] The network device can send the first indication information to the first terminal after determining that the first terminal meets the condition of network access. The first indication information indicates that the first terminal can access the network device. The condition of network access can be flexibly set and is not limited.

[0103] In the embodiment, the network device provides the first terminal with the model data packet of the first model corresponding to the model requested by the first terminal, so that the first terminal does not need to train the first model, and the first terminal does not need to frequently obtain the training data for training the first model from the network device, thereby reducing the network communication overhead and the computing overhead of the first terminal.

[0104] In a possible design, the network device can obtain the model data packet of the first model from the server, as shown in FIG. 4. In this case, S320 can include the following steps.

[0105] S3201, the network device sends fourth information to the server. Correspondingly, the server receives the fourth information from the network device.

[0106] The fourth information is determined according to the first information, and the fourth information is used to indicate the model label of the first model. In an implementation, the network device maintains the information of the first terminal. After receiving the first information, the network device determines that the terminal requesting the model is the first terminal. At this time, the network device can determine the label of the first model requested by the first terminal based on the information of the first terminal stored by itself. In another implementation, the first information can include the model label of the first model, that is, the first terminal directly sends the model label requested by itself to the network device.

[0107] The server stores the model label of the second model trained by the second terminal and the model parameter of the second model. The first terminal and the second terminal are the same terminal, or the first terminal and the second terminal are different terminals, and the second model includes the first model. The second terminal can record the corresponding model label and model parameter of the second model when training the second model, and send the recorded model label of the second model and the model parameter of the second model to the server for storage. The specific process will be introduced in the embodiment shown in FIG. 6 below, which will not be described here in detail.

[0108] The model label of the second model includes or is used to indicate at least one of the following: a physical layer unit to which the second model is applicable, a collection time of training data of the second model, a collection location of the training data of the second model, a type of a terminal used to train the second model, a type of a service corresponding to the training of the second model, a type of the second model, a size of the second model, or a performance score of the second model. It can be understood that the second model stored by the server can include multiple models or one model, and the description of the model label can refer to the description of the model label in the communication system provided in the embodiments of the present application, and will not be repeated here.

[0109] In S3202, the server sends a model data packet corresponding to the model label of the first model to the network device, and correspondingly, the network device receives the model data packet corresponding to the model label of the first model from the server.

[0110] The model data packet corresponding to the model label of the first model includes the model parameters of the first model, and optionally, the model data packet can also include the model label of the first model.

[0111] In the embodiments of the present application, first, through the centralized storage and management of the server, the model label of the second model and the model parameters of the second model are efficiently and securely shared to multiple network devices, improving the data utilization and access efficiency. Second, the high-performance processing capability of the server ensures the rapid response to the request for the model label of the second model and the model parameters of the second model, supports high-concurrency access, and meets the access demand of large-scale network devices. In addition, the data backup and recovery mechanism of the server guarantees the security and reliability of the model label of the second model and the model parameters of the second model, and the flexible permission control ensures the security and confidentiality of the data. Finally, the scalability and maintainability of the server provide good expansion space and maintenance convenience for the system, further improving the stability and availability of the entire network system.

[0112] In a possible design, as shown in FIG. 5, the method can further include:

[0113] In S510, the network device sends first test data to the first terminal, and correspondingly, the first terminal receives the first test data from the network device.

[0114] The network device can send the first test data according to a preset period (referred to as period A). Alternatively, the network device can also send the first test data after receiving third information for requesting the first test data from the first terminal. The first terminal can send the third information when a preset condition is met, and the preset condition includes receiving fourth information from the network device indicating to send the third information, or the preset condition includes reaching a preset period (referred to as period B). It should be understood that period A and period B can be the same or different, and are not limited.

[0115] In S520, the first terminal runs the first model according to the first test data and the model parameters of the first model to obtain a performance indicator of the first model.

[0116] The performance indicator can include a performance score. Based on the description of the performance score in the embodiments of the present application, the performance score of the model is usually based on a series of evaluation criteria such as accuracy, precision, recall rate, etc. These evaluation criteria can help to comprehensively evaluate the performance of the model, and then obtain the performance indicator of the first model. The evaluation criteria used by the embodiments of the present application when evaluating the performance score of the model are not limited. For example, the performance indicator of the physical layer can be used as the performance score of the model, for example, the BLER of the physical layer after running the model can be used as the performance score of the model. For example, assuming that the model update condition is that the BLER is greater than 30%, and the performance indicator of the first model obtained is 31%, then the first model meets the model update condition.

[0117] In the case where the performance indicator indicates that the first model meets the model update condition, the following is performed:

[0118] In S530, the first terminal sends second information for requesting to update the model to the network device, and correspondingly, the network device receives the second information from the first terminal.

[0119] The second information is used to request to update the model, in other words, the second information is used to request a new model.

[0120] In S540, the network device obtains an updated model data packet of the first model corresponding to the second information.

[0121] Wherein, similar to step S320 in the embodiment shown in FIG. 3, the network device can locally look up whether the update model data packet of the first model corresponding to the second information is stored, and if found, S550 is performed. If the update model data packet of the first model corresponding to the second information is not found locally, the network device can request the update model data packet of the first model from the device storing the update model data packet of the first model. In an implementation, the network device can obtain the update model data packet of the first model from the server, and the specific obtaining principle can refer to the description of the embodiment shown in FIG. 4, and will not be repeated here.

[0122] S550, the network device sends the update model data packet of the first model to the first terminal, and correspondingly, the first terminal receives the update model data packet of the first model from the network device.

[0123] Wherein, after obtaining the update model data packet of the first model, the network device can send it to the first terminal, and the first terminal can deploy the corresponding model based on the update model data packet of the first model.

[0124] In the embodiments of the present application, the following disadvantages are considered: after the terminal introduces the physical layer AI model to predict the pilot signal, although the communication process can be optimized, it may also lead to the loss of real pilot signals. When these real pilot signals are lost, the terminal loses a benchmark for comparing with the AI model prediction result. Therefore, when the communication quality decreases, the terminal cannot directly judge whether it is a problem of the channel itself or a problem of the AI model through the measurement of the pilot signal, so as to accurately and timely update the model. The embodiments of the present application introduce the performance index of the first model, and in the case that the performance index of the first model meets the model update condition, the update model data packet of the first model is requested in time, so as to ensure that the model of the terminal can be accurately and timely updated.

[0125] In an embodiment, as shown in FIG. 6, the method further comprises:

[0126] S610, the network device sends the training data to the second terminal, and correspondingly, the second terminal receives the training data from the network device.

[0127] The training data is used to train the second model. For example, the training data can be transmitted in the form of training frames. That is, the network device transmits at least one training frame to the second terminal each time. The network device can send the training data to the second terminal on the downlink air interface. When sending the training data, existing pilots such as channel state information-reference signals (CSI-RSs) and demodulation reference signals (DMRSs) can be used, without limitation.

[0128] In S620, the second terminal trains the second model according to the training data to obtain model parameters of the second model.

[0129] After receiving the training data, the second terminal can train the second model based on the training data and obtain the model parameters of the second model.

[0130] In S630, the second terminal associates the model parameters of the second model with the model label of the corresponding second model to obtain a model data packet of the second model.

[0131] During the training process, after the second terminal completes the training of the model parameters of the second model, the model label is directly associated with the model parameters to ensure the integrity and traceability of the data. This step generates a model data packet of the second model, which is a comprehensive data packet containing not only the model parameters, i.e., the core knowledge learned by the model, but also the model label to identify the specific version, purpose, or training condition of the model.

[0132] It should be understood that in the case of the first model, the model data packet is described as containing the model parameters and optionally containing the model label. Compared with the model data packet of the first model introduced earlier, although both contain the model parameters, the model data packet of the second model is more complete in composition and also includes the model label. In other words, the model data packet of the second model is a full-quantity data packet that comprehensively contains the model label and the model parameters, facilitating identification when the model is reused.

[0133] In S640, the second terminal sends the model data packet of the second model to the network device, and correspondingly, the network device receives the model data packet of the second model from the second terminal.

[0134] After the second terminal obtains the model data packet of the second model, it can send the model data packet to the network device for subsequent model reuse.

[0135] In S650, the network device sends the model data packet of the second model to the server, and correspondingly, the server receives the model data packet of the second model from the network device.

[0136] Wherein, the network device can send the model data packet of the second model to the server for storage after successfully receiving the complete model data packet of the second model. This process ensures that the model data packet of the second model can be safely saved on the server for subsequent model reuse.

[0137] Before performing step S640, as an optional step, the second terminal can confirm whether it has the permission or condition to upload the model data packet of the second model through the following process:

[0138] S660, the second terminal sends an upload request to the network device. Correspondingly, the network device receives the upload request from the second terminal.

[0139] Next, the network device will evaluate the upload request of the second terminal based on a series of preset rules or conditions. If these conditions are met, i.e., the network device determines that the second terminal can perform the upload operation, for example, it determines whether the second terminal is on the preset list that can perform the upload operation. After determining that the second terminal can perform the upload operation, the next step is:

[0140] S670, the network device sends second indication information to the second terminal. Correspondingly, the second terminal receives the second indication information from the network device.

[0141] Wherein, this second indication information explicitly tells the second terminal that it is now allowed to upload the model data packet of the second model.

[0142] Through this process, it is ensured that the second terminal will only perform the subsequent upload of the model data packet of the second model with the explicit permission of the network device, thereby ensuring the security and compliance of data transmission.

[0143] In the embodiments of the present application, the model parameters of the second model and the corresponding model tags are combined into a model data packet and stored in the server, which can be used for subsequent model reuse. This promotes the effective sharing of high-quality models among networks, significantly reducing unnecessary communication overhead and computational resource consumption caused by independent repeated training of the same or similar models by terminals. By sharing trained models, each terminal can directly utilize these efficient and accurate models without starting from scratch, thereby achieving efficient utilization of resources and significant cost reduction.

[0144] The above describes the scheme provided by the embodiments of the present application from the perspective of the logic of each step. It can be understood that each node, for example, the network device, includes a hardware structure and / or software module corresponding to each function in order to implement the above functions. Those skilled in the art should easily realize that, in combination with the algorithm steps of each example described in the embodiments disclosed herein, the method of the embodiments of the present application can be implemented in the form of hardware, software, or a combination of hardware and computer software. Whether a certain function is implemented in hardware or computer software driven hardware depends on the specific application of the technical solution and the design constraint conditions. The skilled person can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0145] The embodiments of the present application can divide the functional modules of the network device according to the above method examples, for example, each functional module can be divided according to each function, or two or more functions can be integrated in one processing module. The above integrated module can be implemented in the form of hardware or software functional module. It should be noted that the division of the modules in the embodiments of the present application is illustrative, and is only a logical function division. There can be another division method in actual implementation.

[0146] In a specific implementation, each network element shown in the present application, such as the first terminal or the second terminal, can adopt the constituent structure shown in Figure 7 or include the components shown in Figure 7. Figure 7 is a structural schematic diagram of a communication apparatus provided by the embodiments of the present application. When the communication apparatus has the function of the terminal described in the embodiments of the present application, the communication apparatus can be a terminal or a chip or system on chip in the terminal. When the communication apparatus has the function of the network device described in the embodiments of the present application, the communication apparatus can be a network device or a chip or system on chip in the network device. When the communication apparatus has the function of the server described in the embodiments of the present application, the communication apparatus can be a server or a chip or system on chip in the server.

[0147] Exemplarily, FIG. 7 shows a structural schematic diagram of a possible communication apparatus. It can be understood that the communication apparatus 700 includes necessary forms of means, such as modules, units, elements, circuits, or interfaces, and the like, which are configured together to perform the present solution appropriately. The communication apparatus 700 can be a terminal, a network device, or a server described in the above method embodiment, or can be a component (for example, a chip) of these devices, to implement the method described in the above method embodiment. The communication apparatus 700 includes one or more processors 701. The processor 701 can be a general purpose processor or a special purpose processor, and the like. For example, it can be a baseband processor or a central processing unit. The baseband processor can be used to process communication protocols and communication data, and the central processing unit can be used to control the communication apparatus, execute software programs, and process data of the software programs.

[0148] Optionally, in a design, the processor 701 can include a program 703 (which can also be referred to as code or instructions at times) that can be run on the processor 701, so that the communication apparatus 700 performs the method described in the above embodiment. In yet another possible design, the communication apparatus 700 includes a circuit (not shown in FIG. 7) for implementing the signal processing function in the above embodiment.

[0149] Optionally, the communication apparatus 700 can include one or more memories 702 that have a program 704 (which can also be referred to as code or instructions at times) stored thereon, which can be run on the processor 701, so that the communication apparatus 700 performs the method described in the above embodiment.

[0150] Optionally, the processor 701 and / or the memory 702 can include an AI module 707, 708 for implementing AI-related functions. The AI module can be implemented in a software, hardware, or software-hardware combined manner. For example, the AI module can include a RIC module. For example, the AI module can be a near-real-time RIC or a non-real-time RIC.

[0151] Optionally, the processor 701 and / or the memory 702 can also store data. The processor and the memory can be arranged separately or integrated together.

[0152] Optionally, the communication apparatus 700 can also include a transceiver 705 and / or an antenna 706. The processor 701 can also be referred to as a processing unit, which controls the communication apparatus. The transceiver 705 can also be referred to as a transceiving unit, a transceiver, a transceiving circuit, or a transceiver, and the like, which is used to implement the transceiving function of the communication apparatus through the antenna 706.

[0153] FIG. 8 shows a structural diagram of a communication apparatus 80 applied to the first terminal. The modules in the apparatus shown in FIG. 8 have functions of implementing the corresponding steps in the above method embodiments and can achieve their corresponding technical effects. The beneficial effects of the steps performed by the modules can be referred to the descriptions of the corresponding steps in the above method embodiments, which will not be repeated. The functions can be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The communication apparatus can be the first terminal or a chip or system on chip in the first terminal. For example, the communication apparatus includes:

[0154] The transceiver module 801 is configured to send the first information, the first information being used to request a model, and receive a model data packet of a first model, the first model being a model in pre-trained models and having a model label matching a model label requested by the first terminal, the model data packet including model parameters of the first model; or the model data packet including the model parameters of the first model and the model label of the first model.

[0155] In an embodiment, the transceiver module 801 is specifically configured to send the first information to a network device, and receive the model data packet of the first model from a server through the network device, the server storing a model label of a second model trained by a second terminal and model parameters of the second model, the first terminal and the second terminal being the same terminal, or the first terminal and the second terminal being different terminals, and the second model including the first model.

[0156] In an embodiment, the model label of the second model includes or is used to indicate at least one of the following: a physical layer module to which the second model is applicable, a collection time of training data of the second model, a collection location of the training data of the second model, a type of a terminal training the second model, a type of a service corresponding to training of the second model, a type of the second model, a size of the second model, or a performance score of the second model.

[0157] In an embodiment, the transceiver module 801 is further configured to receive the first test data.

[0158] The processing module 802 is configured to run the first model according to the first test data and the model parameters of the first model, to obtain a performance indicator of the first model.

[0159] The transceiver module 801 is configured to send second information, the performance indicator indicating that the first model meets a model update condition, the second information being used to request an updated model, and receive an updated model data packet of the first model corresponding to the second information.

[0160] In an embodiment, the transceiver module 801 is further configured to send third information, the third information being used to request the first test data.

[0161] In an embodiment, the transceiver module 801 is specifically configured to send the third information when a preset condition is met; wherein the preset condition comprises receiving fourth information indicating to send the third information, or the preset condition comprises reaching a preset period.

[0162] In an embodiment, the transceiver module 801 is specifically configured to send the first information when accessing the network, or send the first information before initiating the preset service.

[0163] FIG. 9 shows a structural diagram of a communication apparatus 90 applied to a network device. The modules in the apparatus shown in FIG. 9 have the functions of implementing the corresponding steps in the above method embodiments and can achieve their corresponding technical effects. For the beneficial effects of the steps performed by the modules, reference can be made to the descriptions of the corresponding steps in the above method embodiments, which will not be repeated here. The functions can be implemented by hardware, or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The communication apparatus can be a network device or a chip or system on chip in a network device. For example, the communication apparatus includes:

[0164] The transceiver module 901 is configured to receive first information from a first terminal, the first information being used to request a model, and obtain a model data packet of a first model corresponding to the first information, and send the model data packet of the first model, the first model being a model in pre-trained models whose model label matches a model label requested by the first terminal, the model data packet including model parameters of the first model; or the model data packet including the model parameters of the first model and a model label of the first model; and receive the model data packet of the first model.

[0165] In an embodiment, the transceiver module 901 is specifically configured to send fourth information to a server, the fourth information being determined according to the first information, the fourth information being used to indicate a model label of the first model, the server storing a model label of a second model trained by a second terminal and model parameters of the second model, the first terminal and the second terminal being the same terminal, or the first terminal and the second terminal being different terminals, the second model including the first model, and receive the model data packet of the first model from the server.

[0166] In an embodiment, the model label of the second model includes or is used to indicate at least one of the following:

[0167] a physical layer unit to which the second model is applicable, a collection time of training data of the second model, a collection location of the training data of the second model, a type of a terminal training the second model, a type of a service corresponding to training the second model, a type of the second model, a size of the second model, or a performance score of the second model.

[0168] In an embodiment, the transceiver module 901 is further configured to send the first test data, receive second information, wherein the performance indicator of the first model indicates that the first model meets the model updating condition, the performance indicator of the first model is obtained by running the first model based on the first test data and model parameters of the first model; obtain an updated model data packet of the first model corresponding to the second information; and send the updated model data packet of the first model.

[0169] In an embodiment, the transceiver module 901 is further configured to receive third information, wherein the third information is used to request the first test data.

[0170] In an embodiment, the transceiver module 901 is specifically configured to send the first test data according to a preset period.

[0171] FIG. 10 shows a structural diagram of a communication apparatus 10 applied to a server. The modules in the apparatus shown in FIG. 10 have the functions of implementing the corresponding steps in the above method embodiments and can achieve their corresponding technical effects. For the beneficial effects of the steps performed by the modules, reference can be made to the descriptions of the corresponding steps in the above method embodiments, which will not be repeated here. The functions can be implemented by hardware or by hardware executing corresponding software. The hardware or software includes one or more modules corresponding to the above functions. The communication apparatus can be a server or a chip or a system on chip in a server. For example, the communication apparatus includes:

[0172] The transceiver module 101 is configured to receive fourth information, wherein the fourth information is used to indicate a model label of a first model requested by a first terminal; the server stores a model label of a second model trained by a second terminal and model parameters of the second model, the first terminal and the second terminal are the same terminal, or the first terminal and the second terminal are different terminals, and the second model includes the first model; send a model data packet corresponding to the model label of the first model, wherein the model data packet includes the model parameters of the first model; or the model data packet includes the model parameters of the first model and the model label of the first model.

[0173] Embodiments of the present application also provide a communication system, which is a communication system corresponding to a high-speed private network information transmission scenario of a neighboring cell. The communication system can include a first terminal, a network device, and a server. The first terminal can have the functions of the communication apparatus 80 described above, the network device can have the functions of the communication apparatus 90 described above, and the server can have the functions of the communication apparatus 100 described above.

[0174] The embodiments of the present application further provide a computer readable storage medium. All or part of the flow of the method embodiments can be instructed by a computer program to relevant hardware to complete, and the program can be stored in the computer readable storage medium. When the program is executed, the program can include the flow of the method embodiments. The computer readable storage medium can be the terminal device of any of the foregoing embodiments, such as an internal storage unit of a data sending end and / or a data receiving end, for example, a hard disk or a memory of the terminal device. The computer readable storage medium can also be an external storage device of the terminal device, for example, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, and the like. Further, the computer readable storage medium can include both the internal storage unit and the external storage device of the terminal device. The computer readable storage medium is used to store the computer program and other programs and data required by the terminal device. The computer readable storage medium can also be used to temporarily store data that has been output or will be output.

[0175] The embodiments of the present application further provide a computer instruction. All or part of the flow of the method embodiments can be instructed by the computer instruction to relevant hardware (such as a computer, a processor, a network device, and a terminal, etc.) to complete. The program can be stored in the computer readable storage medium.

[0176] The embodiments of the present application further provide a chip system. The chip system can be composed of a chip, or can include a chip and other discrete devices, without limitation. The chip system includes a processor and a transceiver. All or part of the flow of the method embodiments can be completed by the chip system, for example, the chip system can be used to implement the functions performed by the first terminal or the network device in the method embodiments.

[0177] In a possible design, the chip system further includes a memory, and the memory is used to save program instructions and / or data. When the chip system is running, the processor executes the program instructions stored in the memory, so that the chip system performs the functions performed by the first terminal or the network device in the method embodiments.

[0178] In the embodiments of the present application, the processor can be a general processor, a digital signal processor, an application specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, and can implement or execute the disclosed methods, steps and logic block diagrams in the embodiments of the present application. The general processor can be a microprocessor or any conventional processor. The steps of the method disclosed in combination with the embodiments of the present application can be directly embodied as hardware processor execution or executed by a combination of hardware and software modules in the processor.

[0179] In the embodiments of the present application, the memory can be a non-volatile memory such as a hard disk drive (HDD) or a solid-state drive (SSD), and can also be a volatile memory such as a random-access memory (RAM). The memory can be any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but is not limited to this. The memory in the embodiments of the present application can also be a circuit or any other device capable of realizing a storage function, used for storing instructions and / or data.

[0180] It should be noted that the terms "first" and "second" and the like in the specification, claims and drawings of the present application are used to distinguish different objects, and are not used to describe a specific order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but can optionally include steps or units not listed or can optionally include other steps or units inherent to the process, method, product or device.

[0181] It should be understood that in the embodiments of the present application, "at least one" refers to one or more, "multiple" refers to two or more, "at least two" refers to two or three and three or more, and "and / or" is used to describe the association relationship of the associated objects, which means that there can be three relationships, for example, "A and / or B" can mean: only A, only B, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects before and after it. "At least one of the following" or similar expressions means any combination of these items, including any combination of single item or multiple items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple. It should be understood that in the embodiments of the present application, "B corresponding to A" means that B is associated with A. For example, B can be determined according to A. It should also be understood that determining B according to A does not mean that B is determined only according to A, but B can also be determined according to A and / or other information. In addition, "connection" appearing in the embodiments of the present application means direct connection or indirect connection and various connection manners to achieve communication between devices, which is not limited by the embodiments of the present application.

[0182] Unless otherwise specified, "transmit" and "transmission" appearing in the embodiments of the present application mean bidirectional transmission, including sending and / or receiving actions. Specifically, "transmit" in the embodiments of the present application includes data sending, data receiving, or data sending and data receiving. Or, data transmission here includes uplink and / or downlink data transmission. Data can include channels and / or signals, uplink data transmission is uplink channel and / or uplink signal transmission, and downlink data transmission is downlink channel and / or downlink signal transmission. "Network" and "system" appearing in the embodiments of the present application express the same concept, and the communication system is the communication network.

[0183] Through the description of the above embodiments, those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional modules is taken as an example for illustration, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above.

[0184] In several embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented by other manners. For example, the apparatus embodiments described above are merely illustrative, for example, the division of the modules or units is merely a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another apparatus, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some interfaces, apparatuses or units, and can be electrical, mechanical or other forms.

[0185] The units described as separate components can or can not be physically separate, and the components shown as units can be one physical unit or multiple physical units, that is, can be located in one place, or can be distributed to multiple different places. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0186] In addition, each functional unit in each embodiment of the present application can be integrated in one processing unit, or each unit can be physically present separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit. When the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of the present application essentially or say the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing an apparatus, such as a single-chip microcomputer, a chip, or a processor, to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a ROM, a RAM, a magnetic disk or an optical disk, and various storage media that can store program codes.

[0187] The above is merely a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto, and any change or replacement within the technical scope disclosed in the present application should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A communication method characterized by comprising: Applied to a first terminal, comprising: sending first information, the first information is used for requesting a model; receiving a model data packet of a first model, the first model is a model in a pre-trained model, a model tag of which matches a model tag requested by the first terminal, the model data packet comprises model parameters of the first model; or, the model data packet comprises model parameters of the first model and a model tag of the first model.

2. The communication method according to claim 1, characterized by, The sending of the first information comprises: sending the first information to a network device; The receiving of the model data packet of the first model comprises: receiving the model data packet of the first model from a server through the network device, the server storing a model tag of a second model trained by a second terminal and model parameters of the second model, the first terminal and the second terminal being the same terminal, or the first terminal and the second terminal being different terminals, the second model comprising the first model.

3. The communication method according to claim 2, wherein, The model tag of the second model comprises or is used to indicate at least one of the following: a physical layer module applicable to the second model, a collection time of training data of the second model, a collection location of the training data of the second model, a type of a terminal training the second model, a type of a service corresponding to the training of the second model, a type of the second model, a size of the second model, or a performance score of the second model.

4. The communication method according to any one of claims 1 to 3, characterized by, The method further comprises: receiving first test data; running the first model according to the first test data and the model parameters of the first model to obtain a performance indicator of the first model; sending second information, wherein the performance indicator indicates that the first model meets a model update condition, and the second information is used for requesting an updated model; receiving an updated model data packet of the first model corresponding to the second information.

5. The communication method according to claim 4, wherein, The method further comprises: sending third information, the third information being used for requesting the first test data.

6. The communication method according to claim 5, wherein, The sending of the third information comprises: sending the third information in a case where a preset condition is met; wherein the preset condition comprises receiving fourth information indicating the sending of the third information, or the preset condition comprises reaching a preset period.

7. The communication method according to any one of claims 1 to 6, characterized by, The sending of the first information comprises: sending the first information when accessing a network; or sending the first information before initiating a preset service.

8. A communication method characterized by comprising: Comprising: receiving first information from a first terminal, the first information being used for requesting a model; obtaining a model data packet of a first model corresponding to the first information; sending the model data packet of the first model, the first model being a model in a pre-trained model, a model tag of which matches a model tag requested by the first terminal, the model data packet comprising model parameters of the first model; or, the model data packet comprising model parameters of the first model and a model tag of the first model; receiving a model data packet of a first model.

9. The communication method according to claim 8, wherein, Obtaining a model data packet of a first model corresponding to the first information comprises: sending fourth information to a server, the fourth information being determined according to the first information, the fourth information being used to indicate a model label of the first model, the server storing a model label of a second model trained by a second terminal and model parameters of the second model, the first terminal and the second terminal being the same terminal, or the first terminal and the second terminal being different terminals, the second model including the first model; receiving a model data packet of the first model from the server.

10. The communication method according to claim 9, wherein, The model label of the second model includes or is used to indicate at least one of: a physical layer unit to which the second model is applicable, a collection time of training data of the second model, a collection location of the training data of the second model, a type of a terminal training the second model, a type of a service corresponding to training the second model, a type of the second model, a size of the second model, or a performance score of the second model.

11. The communication method according to any one of claims 8-10, characterized by, The method further includes: sending first test data; receiving second information, wherein a performance indicator of the first model indicates that the first model meets a model update condition, the performance indicator of the first model being obtained by running the first model based on the first test data and model parameters of the first model; obtaining an update model data packet of the first model corresponding to the second information; sending the update model data packet of the first model.

12. The communication method according to claim 11, wherein, The method further includes: receiving third information, the third information being used to request the first test data.

13. The communication method according to claim 11, wherein, The sending of the first test data includes: sending the first test data according to a preset period.

14. A communication method, comprising: It includes: A server receives fourth information, the fourth information being used to indicate a model label of a first model requested by a first terminal; The server stores a model label of a second model trained by a second terminal and model parameters of the second model, the first terminal and the second terminal being the same terminal, or the first terminal and the second terminal being different terminals, the second model including the first model; sending a model data packet corresponding to the model label of the first model, the model data packet including model parameters of the first model; or the model data packet including the model parameters of the first model and the model label of the first model.

15. The communication method according to claim 14, wherein, The model label of the second model includes or is used to indicate at least one of: a physical layer unit to which the second model is applicable, a collection time of training data of the second model, a collection location of the training data of the second model, a type of a terminal training the second model, a type of a service corresponding to training the second model, a type of the second model, a size of the second model, or a performance score of the second model.

16. A communications device, characterized by It includes a module for executing the method of any one of claims 1-7; or a module for executing the method of any one of claims 8-13; or a module for executing the method of claim 14 or 15.

17. A communications device, characterized by The communication device comprises a processor and a transceiver for supporting the communication device to perform the method according to any one of claims 1-15.

18. A communications device, characterized by The communication device comprises a processor for running a computer program or instructions to cause the communication device to perform the method according to any one of claims 1-15.

19. The communication apparatus according to claim 18, wherein The communication device further comprises a memory for storing the computer program or instructions.

20. A chip, characterized by The communication device comprises a processor configured to perform the method according to any one of claims 1-15.

21. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions which, when executed, perform the method according to any one of claims 1-15.

22. A computer program product, characterised in that, The computer program product comprises one or more computer programs which, when executed on a computer, cause the computer to perform the method according to any one of claims 1-15.

23. A communication system, characterized by The communication system comprises: a first terminal for performing the method according to any one of claims 1-7, a network device for performing the method according to any one of claims 8-13; and a server for performing the method according to claim 14 or 15.

Images