WO2025209453A1 - Communication method and apparatus, terminal, network side device, medium, and product - Google Patents

Abstract

The present application relates to the technical field of communications, and discloses a communication method and apparatus, a terminal, a network side device, a medium, and a product. The communication method in embodiments of the present application comprises: a first communication device acquires a first data set, the first data set comprising first input data; and the first communication device determines first information on the basis of the first data set and a first artificial intelligence (AI) model, wherein the first information comprises at least one of the following: first output data, the first output data being output data obtained by inputting the first input data to the first AI model, and the first output data being used for determining or assisting in determining at least one of a test result and a monitoring result of the first AI model; a first test result, the first test result being used for determining the test result of the first AI model; and first monitoring information, the first monitoring information being used for determining the monitoring result of the first AI model.

Description

Communication methods, devices, terminals, network-side equipment, media, and products

CROSS-REFERENCE TO RELATED APPLICATIONS

This application claims priority to Chinese Patent Application No. 202410403747.5 filed in China on April 3, 2024, the entire contents of which are incorporated herein by reference.

Technical Field

The present application belongs to the field of communication technology, and specifically relates to a communication method, apparatus, terminal, network-side equipment, medium and product.

Background Art

Artificial Intelligence (AI) is currently gaining widespread application across various fields, and its integration with the communications sector is increasingly deepening. When introducing AI functional models into communications equipment or systems, model training and application are required. However, the effectiveness of deploying AI functional models in these communications equipment or systems is currently difficult to determine, potentially leading to poor application results for these AI functional models in the communications sector.

Summary of the Invention

The embodiments of the present application provide a communication method, apparatus, terminal, network-side equipment, medium, and product, which can solve the problem of poor application effect of communication AI functional models in the communication field.

In a first aspect, a communication method is provided, the method comprising: a first communication device acquiring a first data set, the first data set comprising first input data;

The first communication device determines first information based on the first data set and a first artificial intelligence (AI) model;

The first information includes at least one of the following:

first output data, where the first output data is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

In a second aspect, a communication method is provided, the method comprising:

The second communication device receives the first information sent by the first communication device;

The second communication device determines at least one of a test result and a monitoring result of the first artificial intelligence AI model based on the first information;

Alternatively, the second communication device sends the first data set to the first communication device for the first communication device to determine the first information;

The first information includes at least one of the following:

First output data, where the first output data is output data obtained based on the first input data of the first data set and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

According to a third aspect, a communication method is provided, the method comprising:

The monitoring device receives fourth output data sent by the fourth communication device, where the fourth output data is decompressed data of the first output data;

The monitoring device determines at least one of a test result and a monitoring result of the first artificial intelligence (AI) model based on the fourth output data;

Alternatively, the monitoring device receives at least one of a test result and a monitoring result of the first artificial intelligence AI model sent by the first communication device or the fourth communication device;

The monitoring device determines at least one of the following based on at least one of the test result and the monitoring result of the first artificial intelligence (AI) model:

Target-first AI model;

Status information of the first AI model;

The target first AI model is at least one of the first AI models.

Among them, the fourth communication device is used to receive the first output data sent by the first communication device and decompress the first output data, where the first output data is output data obtained based on the first input number of the first data set and the first AI model.

In a fourth aspect, a communication device is provided, including:

A first acquisition module is configured to acquire a first data set, where the first data set includes first input data;

A first determination module, configured to determine first information based on the first data set and a first artificial intelligence (AI) model;

The first information includes at least one of the following:

first output data, where the first output data is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

According to a fifth aspect, a communication device is provided, the device comprising:

A fourth receiving module, configured to receive first information sent by the first communication device;

A second determination module is configured to determine at least one of a test result and a monitoring result of the first artificial intelligence AI model based on the first information;

Alternatively, the device comprises:

a second sending module, configured to send a first data set to a first communication device, so that the first communication device determines first information;

The first information includes at least one of the following:

First output data, where the first output data is output data obtained based on the first input data of the first data set and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

In a sixth aspect, a communication device is provided, the device comprising:

a fifth receiving module, configured to receive fourth output data sent by a fourth communication device, where the fourth output data is decompressed data of the first output data;

a third determining module, configured to determine at least one of a test result and a monitoring result of the first artificial intelligence (AI) model based on the fourth output data;

Alternatively, a sixth receiving module is configured to receive at least one of a test result and a monitoring result of the first artificial intelligence AI model sent by the first communication device or the fourth communication device;

A fourth determination module is configured to determine at least one of the following based on at least one of the test result and the monitoring result of the first artificial intelligence (AI) model:

Target-first AI model;

Status information of the first AI model;

The target first AI model is at least one of the first AI models.

Among them, the fourth communication device is used to receive the first output data sent by the first communication device and decompress the first output data, where the first output data is output data obtained based on the first input number of the first data set and the first AI model.

In the seventh aspect, a terminal is provided, which includes a processor and a memory, wherein the memory stores a program or instruction that can be run on the processor, and when the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented, or when the program or instruction is executed by the processor, the steps of the method described in the second aspect are implemented.

In an eighth aspect, a terminal is provided, comprising a processor and a communication interface, wherein the communication interface is configured to obtain a first data set, the first data set comprising first input data;

The processor is configured to determine first information based on the first data set and a first artificial intelligence (AI) model;

Alternatively, the communication interface is used to receive first information sent by a first communication device;

The processor is configured to determine at least one of a test result and a monitoring result of a first artificial intelligence (AI) model based on the first information;

Alternatively, the communication interface is used to send a first data set to the first communication device for the first communication device to determine the first information;

The first information includes at least one of the following:

first output data, where the first output data is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

In the ninth aspect, a network side device is provided, which includes a processor and a memory, wherein the memory stores a program or instruction that can be run on the processor, and when the program or instruction is executed by the processor, the steps of the method described in the first aspect are implemented, or when the program or instruction is executed by the processor, the steps of the method described in the second aspect are implemented, or when the program or instruction is executed by the processor, the steps of the method described in the third aspect are implemented.

In a tenth aspect, a network-side device is provided, comprising a processor and a communication interface, wherein the communication interface is used to obtain a first data set, wherein the first data set comprises first input data;

The processor is configured to determine first information based on the first data set and a first artificial intelligence (AI) model;

Alternatively, the communication interface is used to receive first information sent by a first communication device;

The processor is configured to determine at least one of a test result and a monitoring result of a first artificial intelligence (AI) model based on the first information;

Alternatively, the communication interface is used to send a first data set to the first communication device for the first communication device to determine the first information;

The first information includes at least one of the following:

first output data, where the first output data is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model;

Alternatively, the communication interface is used to receive fourth output data sent by a fourth communication device, where the fourth output data is decompressed data of the first output data;

The processor is configured to determine at least one of a test result and a monitoring result of the first artificial intelligence (AI) model based on the fourth output data;

Alternatively, the communication interface is used to receive at least one of a test result and a monitoring result of the first artificial intelligence AI model sent by the first communication device or the fourth communication device;

The processor is configured to determine at least one of the following based on at least one of a test result and a monitoring result of the first artificial intelligence (AI) model:

Target-first AI model;

Status information of the first AI model;

wherein the target first AI model is at least one of the first AI models;

Among them, the fourth communication device is used to receive the first output data sent by the first communication device and decompress the first output data, where the first output data is output data obtained based on the first input number of the first data set and the first AI model.

In the eleventh aspect, a readable storage medium is provided, on which a program or instruction is stored. When the program or instruction is executed by a processor, the steps of the method described in the first aspect are implemented, or the steps of the method described in the second aspect are implemented, or the steps of the method described in the third aspect are implemented.

In a twelfth aspect, a wireless communication system is provided, comprising: a terminal and a network-side device, wherein the terminal is configured to perform the steps of the method according to the first aspect, and the network-side device is configured to perform the steps of the method according to the second aspect; or, the terminal is configured to perform the steps of the method according to the second aspect, and the network-side device is configured to perform the steps of the method according to the first aspect;

Or, including: terminals, network side equipment and monitoring equipment;

The terminal can be used to execute the steps of the method described in the first aspect, the network side device is used to execute the steps of the method described in the second aspect, and the monitoring device is used to execute the steps of the method described in the third aspect.

In the thirteenth aspect, a chip is provided, which includes a processor and a communication interface, the communication interface and the processor are coupled, and the processor is used to run programs or instructions to implement the method as described in the first aspect, or the method as described in the second aspect, or the method as described in the third aspect.

In the fourteenth aspect, a computer program/program product is provided, which is stored in a storage medium, and is executed by at least one processor to implement the steps of the communication method described in the first aspect, or the computer program/program product is executed by at least one processor to implement the steps of the communication method described in the second aspect, or the computer program/program product is executed by at least one processor to implement the steps of the communication method described in the third aspect.

In an embodiment of the present application, a first communication device obtains a first data set, which includes first input data; the first communication device determines first information based on the first data set and a first artificial intelligence (AI) model; wherein the first information includes at least one of the following: first output data, which is output data obtained based on the first input data input and the first AI model, and the first output data is used to determine or assist in determining at least one of the test results and monitoring results of the first AI model; a first test result, which is used to determine the test result of the first AI model; first monitoring information, which is used to determine the monitoring result of the first AI model, that is, by obtaining at least one of the test results and monitoring results of the first AI model through the first data set and the first AI model, the deployment effect of the AI functional model in the communication device or system can be evaluated, thereby helping to improve the application effect of the communication AI functional model.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG1 is a block diagram of a wireless communication system to which embodiments of the present application may be applied;

FIG2 is a schematic diagram of a CSI compression solution applicable to an embodiment of the present application;

FIG3 is a schematic diagram of a CSI reporting solution applicable to an embodiment of the present application;

FIG4 is a schematic diagram of another CSI reporting solution applicable to an embodiment of the present application;

FIG5 is a flow chart of a communication method provided in an embodiment of the present application;

FIG6 is a schematic diagram of an AI model applicable to embodiments of the present application;

FIG7 is a flow chart of another communication method provided in an embodiment of the present application;

FIG8 is a schematic diagram of a monitoring method provided in an embodiment of the present application;

FIG9 is a schematic diagram of another monitoring method provided in an embodiment of the present application;

FIG10 is a flowchart of another communication method provided in an embodiment of the present application;

FIG11 is a schematic diagram of a communication device provided in an embodiment of the present application;

FIG12 is a schematic diagram of another communication device provided in an embodiment of the present application;

FIG13 is a schematic diagram of another communication device provided in an embodiment of the present application;

FIG14 is a schematic diagram of a communication device provided in an embodiment of the present application;

FIG15 is a schematic diagram of a terminal provided in an embodiment of the present application;

FIG16 is a schematic diagram of a network-side device provided in an embodiment of the present application;

Figure 17 is a schematic diagram of another network-side device provided in an embodiment of the present application.

DETAILED DESCRIPTION

The following will be combined with the accompanying drawings in the embodiments of this application to clearly describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field are within the scope of protection of this application.

The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the terms used in this way are interchangeable where appropriate, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same type, and do not limit the number of objects, for example, the first object can be one or more. In addition, "or" in this application represents at least one of the connected objects. For example, "A or B" covers three options, namely, Option 1: including A but not including B; Option 2: including B but not including A; Option 3: including both A and B. The character "/" generally indicates that the objects associated before and after are in an "or" relationship.

The term "indication" in this application can be either a direct indication (or explicit indication) or an indirect indication (or implicit indication). A direct indication can be understood as the sender explicitly informing the receiver of specific information, the operation to be performed, or the requested result, etc. in the instruction sent; an indirect indication can be understood as the receiver determining the corresponding information based on the instruction sent by the sender, or making a judgment and determining the operation to be performed or the requested result, etc. based on the judgment result.

It is worth noting that the technology described in the embodiments of the present application is not limited to the Long Term Evolution (LTE)/LTE-Advanced (LTE-A) system, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency Division Multiple Access (SC-FDMA) or other systems. The terms "system" and "network" in the embodiments of the present application are often used interchangeably, and the described technology can be used for the systems and radio technologies mentioned above, as well as for other systems and radio technologies. The following description describes a New Radio (NR) system for example purposes, and NR terminology is used in most of the following description, but these technologies can also be applied to systems other than NR systems, such as ^6th Generation (6G) communication systems.

FIG1 shows a block diagram of a wireless communication system applicable to embodiments of the present application. The wireless communication system includes a terminal 11 and a network-side device 12 . Among them, the terminal 11 can be a mobile phone, a tablet computer (Tablet Personal Computer), a laptop computer (Laptop Computer), a notebook computer, a personal digital assistant (PDA), a handheld computer, a netbook, an ultra-mobile personal computer (UMPC), a mobile Internet device (MID), an augmented reality (AR), a virtual reality (VR) device, a robot, a wearable device (Wearable Device), a flight vehicle, a vehicle user equipment (VUE), a ship-borne equipment, a pedestrian user equipment (PUE), a smart home (home appliances with wireless communication functions, such as refrigerators, televisions, washing machines or furniture, etc.), a game console, a personal computer (PC), an ATM or a self-service machine and other terminal-side devices. Wearable devices include: smart watches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart bracelets, smart rings, smart necklaces, smart anklets, smart anklets, etc.), smart wristbands, smart clothing, etc. Among them, the vehicle-mounted device can also be called a vehicle-mounted terminal, a vehicle-mounted controller, a vehicle-mounted module, a vehicle-mounted component, a vehicle-mounted chip or a vehicle-mounted unit, etc. It should be noted that the specific type of the terminal 11 is not limited in the embodiment of the present application. The network side device 12 may include an access network device or a core network device, wherein the access network device may also be called a radio access network (Radio Access Network, RAN) device, a radio access network function or a radio access network unit. The access network device may include a base station, a wireless local area network (Wireless Local Area Network, WLAN) access point (Access Point, AP) or a wireless fidelity (Wireless Fidelity, WiFi) node, etc. Among them, the base station can be called Node B (NB), Evolved Node B (eNB), the next generation Node B (gNB), New Radio Node B (NR Node B), access point, Relay Base Station (RBS), Serving Base Station (SBS), Base Transceiver Station (BTS), radio base station, radio transceiver, base The Basic Service Set (BSS), Extended Service Set (ESS), home Node B (HNB), home evolved Node B, transmission reception point (TRP) or other appropriate terms in the relevant field, as long as the same technical effect is achieved, the base station is not limited to specific technical vocabulary. It should be noted that in the embodiments of the present application, only the base station in the NR system is introduced as an example, and the specific type of the base station is not limited.

The core network equipment may include but is not limited to at least one of the following: core network nodes, core network functions, Mobility Management Entity (MME), Access and Mobility Management Function (AMF), Session Management Function (SMF), User Plane Function (UPF), Policy Control Function (PCF), Policy and Charging Rules Function (PCRF), Edge Application Server Discovery Function (EASDF), Unified Data Management (UDM), Unified Data Repository (UDR), and so on. The NR system comprises the following components: a central repository function (UDR), a home subscriber server (HSS), a centralized network configuration (CNC), a network repository function (NRF), a network exposure function (NEF), a local NEF (L-NEF), a binding support function (BSF), an application function (AF), a location management function (LMF), a gateway mobile location center (GMLC), and a network data analytics function (NWDAF). It should be noted that in the embodiments of the present application, only the core network device in the NR system is used as an example for introduction, and the specific type of the core network device is not limited.

For ease of understanding, some of the contents involved in the embodiments of this application are described below:

Artificial intelligence is currently being widely used in various fields. AI models can be implemented in a variety of ways, such as neural networks, decision trees, support vector machines, and Bayesian classifiers. Some of the embodiments in this application use neural networks as an example, but the specific type of AI model is not limited.

A neural network is composed of neurons, typically with a ₁ , a ₂ , ..., a _K as input, w as the weight (multiplicative coefficient), b as the bias (additive coefficient), and σ(.) as the activation function. Common activation functions include sigmoid, tanh, linear rectification function, or rectified linear unit (ReLU).

Neural network parameters are optimized using optimization algorithms. An optimization algorithm is a type of algorithm that minimizes or maximizes an objective function (sometimes called a loss function). The objective function is often a mathematical combination of model parameters and data. For example, given data X and its corresponding label Y, a neural network model f(.) can be constructed. With this model, the predicted output f(x) can be obtained based on the input x, and the difference between the predicted value and the true value (label) (f(x) - Y) can be calculated. This is the loss function. The goal of neural network training is to find the appropriate W (a vector of weights w) that minimizes the value of this loss function. The smaller the loss value, the closer the model is to the true state.

Currently, most common optimization algorithms are based on the error back propagation (BP) algorithm. The basic idea of the BP algorithm is that the learning process consists of two processes: forward signal propagation and back propagation of errors. During forward propagation, input samples are passed from the input layer, processed layer by layer by each hidden layer, and then passed to the output layer. If the actual output of the output layer does not match the expected output, the error back propagation phase begins. Error back propagation is the process of propagating the output error back through the hidden layer to the input layer layer by layer in some form, and distributing the error to all units in each layer, thereby obtaining an error signal for each unit in each layer. This error signal serves as the basis for correcting the weights of each unit. This process of adjusting the weights of each layer through forward signal propagation and back propagation of errors is repeated in a cycle. The process of continuous weight adjustment is the learning and training process of the network. This process continues until the error in the network output is reduced to an acceptable level, or until a pre-set number of learning times is reached. Common optimization algorithms include Gradient Descent, Stochastic Gradient Descent (SGD), mini-batch gradient descent, Momentum, Nesterov (name of the inventor, specifically stochastic gradient descent with momentum), Adaptive Gradient Descent (Adagrad), Adadelta, root mean square prop (RMSprop), and Adaptive Moment Estimation (Adam). During error backpropagation, these optimization algorithms calculate the derivative/partial derivative of the current neuron based on the error/loss obtained from the loss function, add the learning rate, previous gradients/derivatives, and partial derivatives, and finally obtain the gradient, which is then passed to the previous layer.

In the embodiments of the present application, the AI unit/AI model may also be referred to as an AI unit, an AI model, a machine learning (ML) model, an ML unit, an AI structure, an AI function, an AI characteristic, a neural network, a neural network function, a neural network function, etc., or the AI unit/AI model may also refer to a processing unit capable of implementing specific algorithms, formulas, processing flows, capabilities, etc. related to AI, or the AI unit/AI model may be a processing method, algorithm, function, module or unit for a specific data set, or the AI unit/AI model may be a processing method, algorithm, function, module or unit running on AI/ML related hardware such as a graphics processing unit (GPU), a neural network processor (NPU), a tensor processing unit (TPU), an application-specific integrated circuit (ASIC), etc., and the present application does not make specific limitations on this. Optionally, the specific data set includes the input and/or output of the AI unit/AI model.

Optionally, the identifier (identification information) of the AI unit/AI model may be an AI model identifier, an AI structure identifier, an AI algorithm identifier, or an identifier of a specific data set associated with the AI unit/AI model, or an identifier of a specific scenario, environment, channel feature, or device related to the AI/ML, or an identifier of a function, feature, capability, or module related to the AI/ML. This embodiment of the present application does not specifically limit this.

The present application involves the application of Channel State Information (CSI) compression scenarios. For ease of understanding, some relevant content related to CSI is introduced below.

Information theory shows that accurate channel state information is crucial to channel capacity. Especially for multi-antenna systems, the transmitter can optimize signal transmission based on CSI to better match the channel state. For example, the channel quality indicator (CQI) can be used to select an appropriate modulation and coding scheme (MCS) for link adaptation; the precoding matrix indicator (PMI) can be used to implement eigenbeamforming to maximize the strength of the received signal, or to suppress interference (such as inter-cell interference and multi-user interference). Therefore, since the introduction of multi-input multi-output (MIMO) technology, CSI acquisition has been a research hotspot.

Typically, access network equipment, taking a base station as an example, sends a Channel State Information Reference Signal (CSI-RS) on certain time-frequency resources in a certain time slot. The terminal performs channel estimation based on the CSI-RS, calculates the channel information on this slot, and feeds back the codebook information to the base station through the PMI. The base station combines the channel information based on the codebook information fed back by the terminal, and uses this to perform data precoding and multi-user scheduling before the next CSI report.

To further reduce CSI feedback overhead, the terminal can change the PMI reported for each subband to reporting it according to the delay. Since the channels in the delay domain are more concentrated, a PMI with less delay can approximately represent the PMI of all subbands, that is, the delay domain information is compressed before reporting. Similarly, to reduce overhead, the base station can pre-code the CSI-RS in advance and send the encoded CSI-RS to the terminal. The terminal sees the channel corresponding to the encoded CSI-RS. The terminal only needs to select several ports with higher strength from the ports indicated by the network side (for example, 32 ports per channel) and report the coefficients corresponding to these ports.

Furthermore, in order to better compress channel information, neural network or machine learning methods can be used.

Specifically, the terminal compresses and encodes the channel information, and the base station decodes the compressed content to recover the channel information. The base station's decoding network and the terminal's encoding network need to be jointly trained to achieve a reasonable match. A neural network is formed by the terminal's encoder and the base station's decoder, forming a joint neural network. Joint training is performed by the network. After training is complete, the base station sends the encoder network to the terminal. During inference (the model's application phase), the terminal estimates the CSI-RS and calculates the channel information. This calculated channel information or the original estimated channel information is passed through the encoding network to obtain an encoding result. The encoding result is then sent to the base station. The base station receives the encoded result and inputs it into the decoding network to recover the channel information.

The CSI compression use case is a typical two-end model use case, meaning the complete CSI compression model needs to be deployed on different communication nodes. Currently, most considerations involve deploying the encoder on the user equipment (UE) side and the decoder on the network (NW) side. The (sub-)models deployed on multiple nodes need to be paired with each other to function properly. Considering the aforementioned characteristics of the two-end model, the protocol defines several basic AI/ML CSI compression model training collaboration types:

1) Joint training at single entity (also called type 1)

This training framework refers to training a complete encoder and decoder model on a communication node (UE or NW or a third-party server node, etc.), and then deploying the corresponding model module to the target node through methods such as model transfer (for example, transferring the encoder part to the UE and the decoder part to the NW).

2) Joint training at multiple entities (also known as type 2)

This training framework involves multiple nodes participating in the training process, with each node independently calculating the forward and backpropagation information required for local model training and updating its own model parameters. Because the training process requires forward and backpropagation of the entire model (including the encoder and decoder), the corresponding forward and backpropagation information must be transferred between participating nodes. After training is complete, the model no longer needs to be transferred between nodes.

3) Separate training on multiple nodes (also called type 3)

This training framework involves first training a reference model on a specific node. Information about the reference model is then sent to the target node. The target node then uses this information to train its own model, ensuring that the node (or sub-)models are compatible. For example, the network (NW) first trains a complete encoder-decoder model and determines that the resulting decoder is the one that will be used. Information about the corresponding encoder (typically, the encoder's input and output data) is then sent to the user end (UE). The UE then uses this information to train its own encoder. This training framework can be further divided into two scenarios: UE-first training and NW-first training. UE-first training involves first training a complete model on the UE, then sending the information needed for the NW to train the matching model (typically, the input and output data of the model to be trained on the NW side). In contrast, NW-first training involves first training a complete model on the NW, then sending the information needed for the UE to train the matching model (typically, the input and output data of the model to be trained on the UE side).

An exemplary AI-based CSI/PMI compression process is shown in FIG2 , in which the CSI or target CSI (target CSI) or codebook W _N*B (where N is the number of CSI ports and B is the number of subbands) expected by the UE is compressed through AI, such as into an AI-based PMI value (AI based PMI value), and then reported to the network-side device. The network-side device performs decompression to obtain W′ _N*B .

Optionally, in the embodiments of the present application, the use of time-domain CSI correlation is introduced on the basis of space-frequency domain CSI compression, that is, the CSI on multiple slots can be combined for compression, thereby further reducing the overhead of CSI reporting or improving CSI reporting accuracy. For example, as shown in Figure 3, the CSI on four slots is jointly compressed and reported, and the CSI on each slot can be considered as a space-frequency domain CSI report, where the internal information stream corresponds to the output information of the encoder intermediate node.

Based on the reporting method of CSI on multiple slots, time-frequency-spatial domain CSI compression can be further divided into two types: packaged reporting and progressive reporting. Packaged reporting is to report CSI on multiple slots at one time (as shown in Figure 4), while progressive reporting is to report CSI on each slot in sequence in an autoregressive manner (as shown in Figure 3).

The following, in conjunction with the accompanying drawings, describes in detail the communication methods, devices, terminals, network-side equipment, media, and products provided in the embodiments of the present application through some embodiments and their application scenarios.

5 , which is a flow chart of a communication method provided in an embodiment of the present application, and is used for a first communication device. As shown in FIG5 , the method includes the following steps:

Step 501: A first communication device obtains a first data set, where the first data set includes first input data.

In the embodiments of the present application, the first communication device may be a terminal or a network-side device. In some embodiments, the first communication device may also be any one of a data compression device, a data encoding device, a data compression-decompression device, and a data encoding-decoding device. The data compression device, data encoding device, data compression-decompression device, and data encoding-decoding device may be a terminal or a network-side device.

In the embodiment of the present application, the first communication device may be a terminal or a core network device. In some embodiments, the first communication device may also be a location management unit, such as a LMF.

In an embodiment of the present application, the above-mentioned first data set can be understood as a data set used for testing or monitoring the first AI model, and the above-mentioned first data set can also be called a test case.

It is understood that the first AI model is typically obtained based on relevant training data or a training dataset. In this embodiment, the training data or training dataset of the first AI model is referred to as the second dataset. The second dataset is used to train the first AI model. Optionally, in one embodiment, the second dataset includes input data and data labels corresponding to the input data for training the first AI model.

In the embodiments of the present application, the first dataset used for testing and/or monitoring the first AI model is different from the second dataset. For example, the data sources may be different. Typically, the second dataset is collected historical data, while the first dataset in the embodiments of the present application may be standardized data or data from a second communication device.

In the embodiments of the present application, the second communication device can be understood as a peer device of the first communication device, and can be a network-side device, a terminal, a core network device such as a location management unit (NWDAF), or a monitoring device that monitors the performance of the communication system. The aforementioned monitoring device can also be referred to as a monitoring model or a monitoring entity.

In an embodiment of the present application, the data format of the first data set is the same as or different from the data format of the second data set. The second data set is usually with a data label (or called label data, label), and the above-mentioned data label is usually used to characterize the true value of the output corresponding to the input data. In an embodiment of the present application, the above-mentioned first data set may be with data labels or without data labels, for example, it may be at least one of a standardized data set with labeled data, a standardized data set with unlabeled data, a data set with labeled data sent by the second communication device, a data set with unlabeled data sent by the second communication device, and a data set with unlabeled data measured by the first communication device. For another example, the first data set may not correspond to the true value of the output, but the expected second output information obtained by the second communication device according to the selected model, such as the reference model.

For the CSI scenario, the first data set (test case) only includes the input information in the second data set (such as two types of information: codebook information or raw channel information), and does not include output information (such as compressed codebook information or raw channel information).

The second data set includes at least one of the following:

L codebook information;

L raw channel information;

L1 compressed codebook information or raw channel information.

Wherein, the above L and L1 are positive integers.

In the embodiment of the present application, the first data set includes at least first input data for inputting into the first AI model. Optionally, it may include or exclude label data corresponding to the first input data.

Optionally, the first data set further includes second output data, and the second output data includes at least one of the following:

First expected information;

First label data;

The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model;

The first label data is label data corresponding to the first input data.

In the embodiment of the present application, the first data set includes second output data corresponding to the first input data. It can be understood that the second output data can be understood as a reference value of the estimated actual output (first output data) of the first AI model. The closer the first output data is to the second output data, the better the performance of the first AI model or the smaller the error. The above-mentioned second output data can be conventional labeled data (for example, manually marked labels, the real value corresponding to the first input information), which has a similar meaning to the labels of the second data set.

In another optional embodiment, the second output data may also be the expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and the reference model. In the embodiment of the present application, the first AI model has a corresponding reference model, and the various devices have the same understanding of the reference model. When the first input data and the reference model are known, the second communication device or protocol may correspond to the output information (i.e., the first expected information) that can be obtained by inputting the reference model into the expected first input data. Similar to the first label data, the first expected information may also be used as a reference value for the first output data. The test of the AI model expects the first output data to be close to the second output data.

In an embodiment of the present application, optionally, the first data set (test case) is a special data set, characterized in that the output corresponding to the input of the data set is known (first label data), or the input of the data set is known to correspond to the output of the reference model (first expected information).

In some optional embodiments, if the first data set is a standardized first data set, the standard may specify multiple first data sets with different characteristics or different functions. Optionally, the first communication device needs to determine the specific first data set to be used, or the second communication device sends identification information or functional information of the first data set to indicate the first data set that the first communication device needs to use.

Optionally, the first data set further includes first expected information;

In a case where the first input data includes the channel state information or codebook information, the first expected information is the channel state information or codebook information compressed according to a reference model;

Or, in a case where the first input data includes original channel information, the first expected information includes original channel information compressed according to a reference model;

Or, in a case where the first input data includes the channel state information or codebook information, the first expected information is decompressed channel state information or codebook information obtained according to a reference model;

Alternatively, in a case where the first input data includes original channel information, the first expected information includes original channel information decompressed according to a reference model.

Optionally, the first data set further includes first label data corresponding to the first input data;

In a case where the first input data includes the channel state information or codebook information, the first label data is compressed channel state information or codebook information;

Or, in a case where the first input data includes original channel information, the first label data is compressed original channel information;

Or, when the first input data includes the channel state information or codebook information, the first tag data is decompressed channel state information or codebook information;

Or, in a case where the first input data includes original channel information, the first tag data is decompressed original channel information;

or, in a case where the first input data includes at least one of a delay profile (DP), a power delay profile (PDP), and a channel impulse response (CIR), the first tag data includes at least one of a location tag of the terminal, a distance tag between the terminal and other devices, a round-trip time (RTT), a reference signal time difference (RSTD) tag, and a time difference tag of a transmitting and receiving signal;

Alternatively, when the first input data includes beam information corresponding to set B, the first label data includes beam labels corresponding to set A.

For example, in the above embodiment, the first tag data includes a location tag of the terminal, which is the actual location of the terminal at the time of association with the first input data;

For example, in the above embodiment, the first tag data includes a location tag of the terminal which is an estimated location of the terminal at the time of association with the first input data;

For example, in the above embodiment, the beam label corresponding to set A is the set of the strongest top k beams at the time of association of the first input data;

For example, the beam tag corresponding to set A in the above embodiment is the beam information after the association time of the corresponding first input data.

As described in the preceding embodiments, the present application does not limit the application scenario of the first AI model. However, it is understood that, in the presence of a label, for a specific first AI model, a preset mapping relationship exists between the first input data and the second output data. This embodiment of the present application provides an exemplary illustration of the above-mentioned preset mapping relationship.

It can be understood that, when the first data set does not include the second output data label, the second communication device side also has knowledge of the first expected information corresponding to the first input data.

Optionally, the first data set is a data set determined by combining the standard standardized input (such as a codebook or channel information) (first input data) with a label (first expected information) obtained by a reference model.

Optionally, the first data set further includes first label data corresponding to the first input data; the first label data is associated with at least one of the following information:

The label data type of the second data set, the statistical information corresponding to the first input data, and the feature information corresponding to the first data set;

Among them, the second data set is the training data of the first AI model.

Optionally, the feature information corresponding to the first data set includes at least one of the following:

scene information associated with the first data set;

a second output information type associated with the first data set;

Optionally, the statistical information corresponding to the first input data includes at least one of the following:

The SNR or Signal to Interference plus Noise Ratio (SINR) of the input data;

The statistical mean corresponding to the input data.

In the embodiment of the present application, optionally, the data labels of the first data set and the second data set may correspond to each other, for example, the data label of the first data set and the data label of the second data set are both compressed channel state information or codebook information.

In an embodiment of the present application, optionally, the statistical information corresponding to the first input data and/or the feature information corresponding to the first data set can also be used as the first label data. For example, the signal to interference plus noise ratio (SNR) of the first data set can be used as the first label data of the first data set.

Optionally, the first data set includes at least one of the following:

a standard data set, a data set sent by the second communication device, and first input data measured by the first communication device.

It can be understood that the data set sent by the second communication device may be one or more data, or may be indication information of the data set.

In an embodiment of the present application, the above-mentioned standard data set and the data set sent by the second communication device can be understood as data prepared in advance, and the devices have the same understanding of the data set; the first input data measured by the above-mentioned first communication device can be understood as data obtained in real time, and the devices do not have the same understanding of the data.

Taking CSI-related data as an example, the first data set (test case) in the embodiment of the present application may include at least one of the following examples:

Example 1, standardized dataset (test case)

The data set includes at least K first communication device-sided model input information, and the data set includes at least one of the following:

1) K codebook information

2) K raw channel information (or encoder input).

Optionally, the standardized dataset may not include label information (second output data) or may include label information. The label information may be first label data or output information of the codebook information or raw channel information after compression and/or quantization by the reference model, i.e., "expected compressed codebook information or expected compressed raw channel information" (first expected information).

Optional, standardized datasets also include:

K1 first communication device side model output (first communication device-sided Model output) information (second output data), such as K1 expected compressed codebook information or raw channel information.

The above K1≤K, that is, the first input data in the standardized data set may partially correspond to the label, partially not correspond to the label, or may not correspond to the label at all, or may all correspond to the label.

Optionally, the first input data of the first data set includes at least one of the following: N first input data associated with the second output data, and M first input data not associated with the second output data, where N and M are both natural numbers.

The above examples use CSI-related information as an example. It is understood that the embodiments of the present application are not limited to CSI-related scenarios. For example, in a positioning embodiment, the second output data may also be at least one of position, distance, round-trip time (RTT), and reference signal time difference (RSTD).

Optionally, the first input data includes at least one of the following:

Channel state information;

Codebook information;

Original channel information;

Delay profile DP;

Power delay profile PDP;

Channel impulse response CIR;

Delay information corresponding to multipath;

Delay power information corresponding to multipath;

Beam information.

In the embodiment of the present application, the application scenario of the first AI model is not limited. The corresponding first AI model can be determined according to the specific application scenario, that is, the first input data corresponding to the first AI model can be determined, for example, CSI-related channel state information, codebook information, or original channel information, etc.; positioning-related delay profile (DP), power delay profile (PDP), channel impulse response (CIR), delay information corresponding to multipath, delay power information corresponding to multipath, etc.; it can also be beam-related information, such as beam set B, etc.

In an AI-based positioning embodiment, the input can be DP, PDP, CIR, or delay power on multiple paths, and the output can be a label corresponding to the input, such as the position of the UE corresponding to the input, the distance to other devices, or the reference signal time difference (RSTD), the Rx-Tx timing difference, etc.

For example, the UE first obtains the raw channel/PDP/multipath information of the positioning reference unit (PRU) and the position information of the PRU. The UE inputs the raw channel PDP/multipath information of the PRU into its own AI model. If the position output by the AI model has an error less than a certain threshold with the position of the PRU, the monitoring result is considered valid.

In an AI-based beam implementation, the input can be the beams corresponding to set B, and the output can be the beams corresponding to set A. Set B is the indices and/or Reference Signal Received Power (RSRP) of Y beams, and the output is the indices and/or RSRP of the Z best beams.

Example 2: Data set (test case) sent by the peer device (second communication device) to the first communication device

Similar to Example 1, the dataset includes at least K first communication device-side model input information (first communication device-sided Model input), and the dataset includes at least one of the following:

1) K codebook information;

2) K raw channel information (or called encoder input).

Optionally, the above dataset includes:

K1 first communication device side model output (first communication device-sided Model output) information (second output data), such as K1 expected compressed codebook information or raw channel information.

Similarly, the data set sent by the peer device (the second communication device) to the first communication device may or may not include label information (the second output data). The label information may be the first label data, or may be the output information of the codebook information or the raw channel information after compression and/or quantization by the reference model, i.e., the "expected compressed codebook information or the expected compressed raw channel information" (the first expected information).

Optionally, the K1 expected compressed codebook information or raw channel information (first expected data) is determined by the second communication device based on the codebook information or raw channel information in the reference model and test case.

Example 3: The first input data measured by the first communication device

The first input data measured by the first communication device can be understood as the data measured by the first communication device being directly used for testing and/or monitoring the first AI model during the operation or inference phase of the first AI model (the first AI model has completed training, or has completed training and testing). It is understood that the data measured by the first communication device only includes the first input data and does not include the corresponding second output data (label).

In the embodiment of the present application, the second communication device may further send at least one of the following to the first communication device: a reference model, (all or part of) parameters of the reference model, and a second data set.

In another embodiment, the reference model, (all or part of) the parameters of the reference model, and the second data set may also be sent to the first communication device by a third communication device. The third communication device may be, for example, a functional device on the core network side.

In yet another embodiment, the first AI model is determined by the first communication device itself.

Optionally, the method further includes:

The first communication device receives third information sent by the second communication device or the third communication device, where the second communication device includes at least one of a fourth communication device and a monitoring device;

The third information includes at least one of the following:

Second dataset;

Reference model;

Target model parameters associated with the reference model.

In an embodiment of the present application, the first communication device can determine the first AI model based on the above-mentioned third information.

In an embodiment of the present application, the fourth communication device may be any one of a network side device (eg, an access network device), a decoding device, and a decompression device.

Optionally, the first data set is associated with the third information.

In an embodiment of the present application, different second information can determine different first AI models. Correspondingly, the first data sets used for testing and/or monitoring the first AI model are also different, and the first data set and the second information can be considered to be associated. Typically, if the expected first output data is determined based on the reference model, the second output data of the first data set is associated with the reference model, or if the first data set and the second data set are different data belonging to the same third data set, the first data set is associated with the second data set.

In an embodiment of the present application, when the reference model and/or the parameters of the reference model are at least partially sent by the second communication device to the first communication device, the first data set can be used to determine whether the reference model and/or the parameters of the reference model are properly installed or used by the first communication device.

It is understood that the second communication device can determine whether the first AI model of the first communication device is used as expected based on the data in the first data set, such as whether the first output data is similar to the expected first output data (second output data). Alternatively, the first communication device can determine whether the first AI model of the first communication device is used as expected based on the data in the first data set.

Step 502: The first communication device determines first information based on the first data set and the first artificial intelligence (AI) model.

The first information includes at least one of the following:

first output data, where the first output data is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

In the embodiment of the present application, the first AI model can be understood as an AI model that has been trained and deployed on the first communication device. The first communication device can use the first data set to test or monitor the performance of the first AI model before or during the communication application of the first AI model.

The above test results can be understood as the results obtained by conducting a forward test before the first AI model is applied in communication;

The above monitoring results can be understood as the results obtained by backward testing in the communication application of the first AI model. The performance of the first AI model can be determined through the actual operation results of the communication system or equipment. For example, in the CSI scenario, the compression performance of the first AI model is determined by judging the throughput of the communication system. In the positioning scenario, the positioning performance of the first AI model is determined by position continuity, etc.

Optionally, the first communication device determines, based on the first data set and the first artificial intelligence (AI) model, first information including at least one of the following:

The first communication device inputs the first input data into the first AI model to obtain the first output data;

The first communication device determines at least one of the first test result and the first monitoring information according to the first data set and the first output data.

In the embodiment of the present application, the first output data can be understood as the output corresponding to the first input data after passing through the first AI model. The first test result and the first monitoring information are further determined based on the first output data.

Optionally, when the first input data includes the channel state information or codebook information, the first output data includes compressed channel state information or codebook information;

Or, in a case where the first input data includes original channel information, the first output data includes compressed original channel information;

Or, in a case where the first input data includes the channel state information or codebook information, the first output data includes decompressed channel state information or codebook information;

Or, in a case where the first input data includes original channel information, the first output data includes decompressed original channel information;

Or, when the first input data includes at least one of a delay profile DP, a power delay profile PDP, and a channel impulse response CIR, the first output data includes at least one of the location information of the terminal, the distance information between the terminal and other devices, the round-trip delay RTT, the reference signal time difference RSTD, and the time difference between transmitting and receiving signals;

Alternatively, when the first input data includes beam information corresponding to beam set B, the first output data includes beam information corresponding to beam set A.

As described in the aforementioned embodiments, the embodiments of this application do not limit the application scenarios of the first AI model. However, it is understood that, for a specific first AI model, a preset mapping relationship exists between the first input data and the first output data. The embodiments of this application provide an exemplary description of the preset mapping relationship.

For example, when the first communication device serves as a channel compression device, the corresponding first AI model is used for channel compression, and the first input data may be at least one of channel state information, codebook information, and original channel information, and the corresponding output data may be at least one of compressed channel state information, compressed codebook information, and compressed original channel information; when the first communication device serves as a channel compression-decompression device, the corresponding first AI model is used for channel compression-decompression, and the first input data may be at least one of channel state information, codebook information, and original channel information, and the corresponding output data may be at least one of decompressed channel state information, decompressed codebook information, and decompressed original channel information.

In an embodiment of the present application, the above-mentioned beam set B is a subset of beam set A; or, set B is a historical time relative to set A.

In an embodiment of the present application, the test results and/or monitoring results of the above-mentioned first AI model can be determined on the first communication device side, or on the second communication device side. The second communication device side obtains the above-mentioned test results and/or monitoring results based on the first information sent by the first communication device.

It can be understood that the above-mentioned first test result can be an intermediate parameter of the test result of the first AI model, or it can be the test result of the first AI model itself; the above-mentioned first monitoring information can be an intermediate parameter of the monitoring result of the first AI model (such as the error information of the first AI model), or it can be the monitoring result of the first AI model itself, that is, the above-mentioned first communication device can report the intermediate parameters of the above-mentioned test results and/or monitoring results to the second communication device, and the second communication device can judge the test results and/or monitoring results (such as the quality of the test results, the status information of the first AI model, whether the first AI model needs to be switched, etc.), or it can directly report the test results and/or monitoring results to the second communication device.

Optionally, the first AI model includes at least one of a first model and a second model, the first model is a reference model, and the second model is a model determined based on the first model and different from the first model.

Exemplarily, as shown in part (a) of Figure 6 , a first AI model includes a reference model and an optimization model (self-optimization). Different first AI models may include the same reference model or different reference models, and the reference model and optimization model in different first AI models may differ at least in part. Generally, a first AI model may include no reference model or one reference model; a first AI model may include no optimization model or one or more optimization models, as shown in part (b) of Figure 6 .

The optimization model and the reference model are different AI models. The optimization model is generated based on the reference model.

Optionally, the reference model includes at least one of the following:

Reference model agreed upon in the agreement;

a reference model determined in conjunction with target model parameters, the target model parameters including model parameters sent by the second communication device;

A reference model determined based on the reference model agreed upon in the protocol and the parameters of the target model;

A reference model trained based on a second data set;

A reference model based on the protocol agreement and a reference model determined by the second data set;

The reference model sent by the second communication device.

Optionally, the reference model may be one or more of a fully connected model, a convolutional model, or a transformer model, or a subsequently evolved model.

Optionally, the reference model agreed upon in the protocol includes at least one of the following information:

The number of fully connected layers;

The depth of the fully connected layers;

Fully connected connection mode;

Convolution kernel parameters, such as kernel size, feature map filling method, and output channels;

The number of convolution kernels;

The type of convolution kernel;

How the convolution kernels are connected;

Activation function;

Parameters of the multi-head attention module, such as the number of multi-head attention heads, the dimension of each attention head, the calculation method of the attention score, and whether there is an output projection matrix;

Parameters of the feedforward module, such as dimension expansion multiples and activation functions;

The number of multi-head attention modules;

How to connect the multi-head attention modules;

Optionally, the target model parameters include at least one of the following:

Fully connected weights;

Mapping method of attention module;

Loss function;

Quantitative methods;

The payload of the first output, etc.

Hyperparameters such as learning rate, batch size, optimization algorithm, etc.

In an embodiment of the present application, the first AI model includes at least one of the following:

1) A reference model, wherein the reference model is any one of the following:

A model generated by at least one of protocol standardization (model structure and/or model parameters and/or dataset (for training the model)),

Alternatively, a reference model determined based on a standardized model structure and model parameters received by the first communication device;

Alternatively, the model (including at least part of the model structure and/or part of the model parameters) received by the first communication device;

Alternatively, the first communication device generates or determines a reference model based on a third data set (optionally standardized, or sent by the opposite device to the first communication device);

Alternatively, the first communication device generates or determines a reference model based on a third data set (optionally standardized, or sent by the opposite device to the first communication device) and a standardized model structure;

2) An optimization model based on a reference model (this can also be understood as a model that does not perfectly implement the reference model, but can achieve functions similar to the reference model, or the mapping relationship between the input and output of the optimization model is the same as that of the reference model or is the same within a certain error range, or it can be understood that there is a certain consistency between the reference model and the optimization model);

It is understandable that the first communication device may maintain X (X is a positive integer) optimization models based on the reference model, and the X optimization models based on the reference model may be respectively applicable to different situations, for example, according to scenarios, conditions, or data distribution. Alternatively, performance similar to that of the reference model may be obtained through X optimization models based on the reference model, wherein the scale of each optimization model is much smaller than the reference model, that is, the optimization model may be a simplification of the reference model (for example, the number of neural network layers is reduced from 20 to 10).

Exemplarily, as shown in part (b) of FIG6 , the first communication device side includes multiple first AI models, which are at least two of the following: a reference model, an optimization model 1, an optimization model i, etc.;

The first communication device reports output associated with different first AI models, and the reported information includes association information associated with the first AI model.

For example, the associated information or output includes model ID, dataset ID, and reference model identification information;

The multiple first AI models may be associated with different at least one of the following:

A different first data set, a data set for training a first AI model,

Different reference models,

Different hardware capabilities,

Different quantification methods,

Different first output data (output) reporting bits (bit).

Optionally, the first communication device receives fourth information indicating (or requesting) information related to the first data set and/or the first AI model used by the first communication device;

Such as indicating the identification information of the first AI model (indicating which first AI model is used to obtain the output),

For example, identification information indicating a reference model of the first AI model,

For example, instructing to obtain information related to the first data set of the first AI model,

For example, indicating how many first AI models are used to obtain output, that is, indicating the number of first AI models.

Exemplarily, when the standardized data set does not include the second output data, optionally, the first communication device reports the “codebook information or raw channel information compressed by the first AI model” (first output data) after passing through the first AI model, and the second communication device determines the “expected compressed codebook information or raw channel information” (first expected information) based on its own model and test case. The second communication device compares the “codebook information or raw channel information compressed by the first AI model” with the “expected compressed codebook information or raw channel information” to determine whether the first communication device passes the test case test;

When the standardized data set includes the above-mentioned second output data, optionally, the first communication device obtains different "first AI model compressed codebook information or raw channel information" through different first AI models, such as, compressed codebook information or raw channel information obtained through a reference model, such as, a first optimization model obtained through a first optimization method and a reference model, and compressed codebook information or raw channel information obtained through the first optimization model; the first communication device determines whether the reference model and/or optimization model in the first AI model passes the test of the test case by comparing the expected compressed codebook information or raw channel information in the data set with the different "first AI model compressed codebook information or raw channel information" obtained by the first communication device through different first AI models.

In an embodiment of the present application, a first communication device obtains a first data set, which includes first input data; the first communication device determines first information based on the first data set and a first artificial intelligence (AI) model; wherein the first information includes at least one of the following: first output data, which is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of the test results and monitoring results of the first AI model; a first test result, which is used to determine the test result of the first AI model; and first monitoring information, which is used to determine the monitoring result of the first AI model, that is, by obtaining at least one of the test results and monitoring results of the first AI model through the first data set and the first AI model, the deployment effect of the AI functional model in the communication device or system can be evaluated, thereby helping to improve the application effect of the communication AI functional model.

It can be understood that in the embodiment of the present application, at least one of the test results and monitoring results of the first AI model is obtained through the first data set and the first AI model, so that the deployment effect of the AI functional model in the communication equipment or system can be evaluated. In the case of performance loss of the AI functional model, the performance loss node of the specific AI functional model can be located according to the test results and monitoring results of the AI model, for example, whether it is the performance loss caused by the reference model, or the performance loss caused by other models or reasons.

Optionally, the first test result includes at least one of the following:

first difference information between the first output data and the second output data;

statistical information of first difference information between the first output data and the second output data;

first ratio information between the first output data and the second output data;

statistical information of first ratio information between the first output data and the second output data;

first difference information between the first difference information and a reference model error;

Statistical information of first difference information between the first difference information and the reference model error;

an indication that the first AI model meets the test requirements;

an indication that the first AI model does not meet the test requirements;

Relevant information about the model that meets the test requirements in the first AI model;

The number of first AI models that meet testing requirements.

In an embodiment of the present application, the first data set includes second output data, that is, includes a reference object for comparison with the first output data. The first communication device can directly determine the test result of the first AI model, for example, an indication that the first AI model meets the test requirements; an indication that the first AI model does not meet the test requirements; relevant information of the model in the first AI model that meets the test requirements; the number of first AI models that meet the test requirements; it can also be to determine the difference comparison result between the first output data and the second output data, for example, the first difference information between the first output data and the second output data; statistical information of the first difference information between the first output data and the second output data; first ratio information between the first output data and the second output data, and statistical information of the first ratio information. The above comparison results can be sent to the second communication device, and the second communication device executes the judgment of the test result of the first AI model.

The difference information in the embodiment of the present application can be understood as the difference information between two objects, and can also be described as error information.

In an embodiment of the present application, the error of the above-mentioned reference model can also be used to determine the test result of the current first AI model or the intermediate value of the test result. The reference model error can refer to the difference between the reference model output and the label. The error of the reference model can be preset or sent by the second communication device to the first communication device.

In the embodiment of the present application, meeting the test requirements may mean that a preset model is executed or successfully deployed, or that the executed or deployed preset model meets the preset performance requirements.

In an embodiment of the present application, the relevant information of the model may be a model identifier (or identification information, or ID), a model type (reference model or optimized model), a reference model identifier corresponding to the first AI model, etc.

Taking the CSI scenario as an example, illustratively, the first information output by the first communication device may include at least one of the following:

1) Codebook information or raw channel information compressed by the first AI model (first output data);

Depending on the model type (reference model and/or optimization model) included in the first AI model, the first output data may include at least one of the following:

The compressed codebook information or raw channel information of the first AI model associated with the reference model;

The codebook information or raw channel information compressed by the first AI model associated with the optimization model.

Optionally, the compressed codebook information or raw channel information is associated with identification information of the model (e.g., model ID);

Optionally, the compressed codebook information or raw channel information is associated with the type of the model (such as whether it is a reference model, such as a method of optimizing the model);

In an optional embodiment, the output includes "first AI model compressed codebook information or raw channel information" corresponding to multiple different optimization models, which is used by the peer device to select an appropriate optimization model;

2) Difference information between the codebook information or raw channel information compressed by the first AI model (first output data) and the expected compressed codebook information or raw channel information (second output data).

Optionally, the difference information is associated with identification information of the model;

Optionally, the difference information is associated with the type of the model (such as whether it is a reference model, such as a method of optimizing the model);

3) Statistical information about the difference between the codebook information or raw channel information compressed by the first AI model and the expected compressed codebook information or raw channel information.

Optionally, the difference information is associated with identification information of the model;

Optionally, the difference information is associated with the type of the model (such as whether it is a reference model, such as a method of optimizing the model);

4) The ratio of the codebook information or raw channel information compressed by the first AI model to the expected compressed codebook information or raw channel information.

Optionally, the above output can be divided into two types:

In the case where the first data set does not include the label information, sending 1) of the output (output) to the second communication device;

If the first data set includes the label information, at least one of 2)-4) in the output (output) is sent to the second communication device. Optionally, the first communication device determines the first information based on the first data set and the first artificial intelligence (AI) model, including at least one of the following:

If the first condition is met, determining that the first AI model meets the test requirements;

If the first condition is not met, determining that the first AI model does not meet the test requirement;

The first condition includes at least one of the following:

The payload of the first output data is the same as the payload of the second output data;

First difference information between the first output data and the second output data is less than or equal to a first threshold;

Statistical information of the first difference information between the first output data and the second output data is less than or equal to a second threshold.

In an embodiment of the present application, the first communication device can directly determine the test result of the first AI model by comparing the first output data with the second output data. The first threshold and the second threshold can be preset or sent to the first communication device by the second communication device or the third communication device.

Optionally, the first output data includes at least one of decompressed codebook information and compressed original channel information, and when the first information includes a first test result, the first communications device determines, based on the first data set and the first artificial intelligence (AI) model, the first information, including at least one of the following:

determining the first test result according to at least one of second difference information between the first output data and the first input data and a statistical value of the second difference information;

The first test result is determined according to at least one of second ratio information of the first output data and the first input data and a statistical value of the second ratio information.

In an embodiment of the present application, in a CSI compression scenario, the pre-compression input information (first input data) may be compared with the final decompressed output data to determine the compression-decompression performance. For example, if at least one of the second difference information and the statistical value of the second difference information is less than or equal to a preset threshold, the compression-decompression performance of the first AI model is considered to be qualified. Alternatively, if at least one of the second ratio information of the first output data to the first input data and the statistical value of the second ratio information satisfies a preset threshold, the compression-decompression performance of the first AI model is considered to be qualified.

Optionally, the method further includes: the first communication device receiving first auxiliary information sent by a second communication device or a third communication device, where the second communication device includes at least one of a fourth communication device and a monitoring device;

The first auxiliary information includes at least one of the following:

first threshold;

second threshold;

First statistics;

The first threshold value includes at least one of the following:

difference threshold information between the first output data and the second output data;

threshold information of a ratio between the first output data and the second output data;

The second threshold value includes at least one of the following:

threshold information of statistical information of differences between the first output data and the second output data;

threshold information of statistical information of the ratio of the first output data to the second output data;

The first statistical information is statistical information of a second data set, and the second data set is training data of the first AI model;

The first communication device determines first information based on the first data set and the first artificial intelligence (AI) model, including:

The first communication device determines the first information based on the first data set, the first AI model and the first auxiliary information.

In an embodiment of the present application, the second communication device or the third communication device may provide auxiliary information to the first communication device for determining the first information, for example, for determining the first test result and the first monitoring information.

Optionally, the first monitoring information includes at least one of the following:

first difference information between the first output data and the second output data;

statistical information of first difference information between the first output data and the second output data;

first ratio information between the first output data and the second output data;

statistical information of first ratio information between the first output data and the second output data;

first difference information between the first difference information and a reference model error;

Statistical information of first difference information between the first difference information and the reference model error;

Monitoring results of the first AI model, such as status information of the first AI model, or the quality of the first AI model, or the degree of quality;

The instruction to switch the first AI model;

Information about well-monitored models in the first AI model;

The number of first AI models that meet monitoring indicators;

The first communication device reports the first monitoring result, such as information in the monitoring result, such as first difference information, statistical information of the first difference information, first ratio information, and statistical information of the first ratio information;

The first communication device obtains the first monitoring result, such as monitoring result information, such as the first AI model monitoring result, such as status information of the first AI model, or the quality of the first AI model, or the degree of quality;

Optionally, the first communication device determines the test result of the first AI model of the terminal based on the first monitoring result.

In an embodiment of the present application, statistical information of the training data set may be provided to the first communication device as reference information for determining first information, for example, for determining a first test result and/or first monitoring information.

Optionally, the first information includes first monitoring information, and the method further includes:

The first communication device receives second information sent by the second communication device;

The second information includes at least one of the following:

The decompressed codebook information acquired by the second communication device;

a throughput determined by the second communication device;

location information of the terminal determined by the second communication device;

Beam information of beam set A;

The first communication device determines first information based on the first data set and the first artificial intelligence (AI) model, including:

The first communication device determines the first monitoring information according to the first data set, the first output data and the second information.

In the embodiment of the present application, when the first monitoring information is determined by the first communication device, the second information sent by the second communication device may be received as reference information for determining the first monitoring information.

Optionally, the monitoring result of the first AI model may be determined by comparing the first output data with the second information. For example, the throughput in the first output data may be compared with the throughput determined by the second communication device to determine whether the current first AI model meets the communication performance requirements.

Optionally, if at least one of the test result and the monitoring result of the first AI model fails, perform at least one of the following operations:

Switch to the second AI model;

Switch to non-AI function;

Fall back to the reference model;

Fall back to non-AI functionality;

Update reference models;

Update reference model parameters;

receiving a reference model repeatedly sent by the second communication device or the third communication device;

receiving a reference model parameter repeatedly sent by the second communication device or the third communication device;

receiving an updated reference model sent by the second communication device or the third communication device;

receiving updated reference model parameters sent by the second communication device or the third communication device.

In an embodiment of the present application, the first communication device performs operations such as switching, rollback, and redeployment when at least one of the test results and monitoring results of the first AI model fails. The above-mentioned failure judgment can be determined by the first communication device or indicated to the first communication device after the second communication device determines it.

If at least one of the test results and monitoring results of the first AI model fails, it can be understood that the first AI model deployment was unsuccessful, and the model can be redeployed by re-receiving the same reference model and/or reference model parameters (the reference model and/or reference model parameters can also generate an optimized model). It can also be understood that the performance of the first AI model cannot meet current business needs, and an updated reference model and/or reference model parameters can be received for subsequent testing and/or application.

For example, the embodiment of the present application may include the following steps:

The first communication device determines the output and/or performs first monitoring based on the first data set (test case) and the first AI model; the performed first monitoring is used to determine whether the first AI model passes the test case or operates normally;

The output includes the result of the first monitoring, i.e., the result of the first monitoring is directly determined by combining the first data set (test case) and the first AI model;

Alternatively, the first monitoring includes at least one of the following:

The first monitoring determines the monitoring result based on the output (the compressed codebook of the first AI model) and the expected compressed codebook corresponding to the first data set (the test case);

If the monitoring result indicates “0” or “fail” or “invalid”, the first AI model fails the test case or does not work properly;

Switching to another AI model, or switching to a non-AI algorithm, function, or feature, or falling back to a non-AI algorithm, function, or feature;

If the monitoring result indicates "0" or "fail" or "invalid", updating or repeatedly sending the reference model or the parameters of the reference model;

If the monitoring result indicates "0" or "fail" or "invalid", the reference data is updated or sent repeatedly.

The first monitoring result includes at least one of the following:

Validity indication information;

Error information;

Error statistics.

According to the first monitoring result, the first communications device reports at least one of the following to a monitoring entity:

Output;

The output that corresponds one-to-one to the codebook information of the first dataset (test case);

The first data set (test case);

Identification information of the first dataset (test case);

The first monitoring includes at least one of the following:

The first monitoring determines the monitoring result based on the codebook after the output is decompressed and the codebook corresponding to the first data set (test case);

If the monitoring result indicates “0” or “fail” or “invalid”, the first AI model fails the test case or does not work properly;

Switching to another AI model, or switching to a non-AI algorithm, function, or feature, or falling back to a non-AI algorithm, function, or feature;

If the monitoring result indicates "0" or "fail" or "invalid", updating or repeatedly sending the reference model or the parameters of the reference model;

If the monitoring result indicates "0" or "fail" or "invalid", the reference data is updated or sent repeatedly.

The first monitoring result includes at least one of the following:

Validity indication information;

Error information, such as

Error information of the first AI model (difference between the codebook of the test case (first input data) and the codebook restored after output decompression);

Difference information between the codebook recovered after decompression of the reference model and the codebook recovered after decompression of the first AI model;

Difference information between the error information of the reference model and the error information of the first AI model;

Error statistical information, which is statistical information of the error information corresponding to the three difference information mentioned above.

Ratio information, such as a ratio of the acquired throughput of the first AI model to a throughput threshold, or a ratio of the throughput of the first AI model to the randomly selected throughput.

According to 2.2, the first monitoring result, the first communications device or the fourth communications device reports at least one of the following to the monitoring entity:

Output the recovered codebook after decompression;

The decompressed and recovered codebook corresponding to each test case;

The codebook after label decompression of the reference model;

Test cases;

Identification information of the test case.

The first communication device reports the first information, where the first information includes at least one of the following:

Report output information determined based on the input of multiple samples (test cases, or data) in the first data set, such as

The output information: output information after the first AI model (such as PMI information, i.e., codebook information compressed by the first AI model), or

The output information is the difference or ratio between the output information after passing through the first AI model and the label information in the first data set;

The output information is statistical information of the accuracy of the multiple output information compared to the label in the first data;

The output information is associated with the first data set;

Optionally, the configuration or reporting request information sent by the second communication device carries information related to the first data set;

Optionally, the information reported by the first communication device carries information related to the first data set.

The first information is associated with a sample in a first data set;

Optionally, the first information corresponds one-to-one to the samples in the first data set;

Optionally, the first information corresponds one-to-one to samples in a specified first data set (e.g., samples in the first data set specified by a network-side device, or samples in the first data set that satisfy certain rules); optionally, the second communication device sends indication information including samples to the first data set.

The first information is associated with a first AI model, such as a reference model or an optimization model.

The first information is associated with the functionality (functionality)/feature (feature) of the model and the identifier (ID) of the model.

Optionally, the method further includes:

The first communication device sends the first information to a second communication device, wherein the second communication device includes at least one of a fourth communication device and a monitoring device.

In an embodiment of the present application, the first communication device can send the first information to the second communication device, which can be to send the test results and monitoring results of the first AI model to the second communication device, or the test results and monitoring results of the first AI model can be determined by the second communication device by sending the first information to the second communication device.

The communication device needs to be tested when introducing new functions. When the AI model requires cooperation between both ends, the devices on both sides or at least one side need to have a certain understanding of the output of the other side. In this embodiment of the application, the first information is determined in the first communication device through the first data set, and interacted with the second communication device to determine whether the AI deployment of the first communication device meets expectations.

Optionally, the first information further includes at least one of the following:

identification information of the first data set;

Identification information of the second data set, wherein the second data set is training data for the first AI model;

identification information of the first AI model;

Type information of the first AI model;

Functional information of the first AI model.

In an embodiment of the present application, when sending at least one of the first output data, the first test result and the first monitoring information, the identifier of the relevant data set, the identifier, type or function information of the first AI model may also be sent accordingly.

Optionally, the first communication device sending the first information to the second communication device includes at least one of the following:

The first communication device sends the first output data to the second communication device;

The first communication device sends first output data corresponding to the plurality of first input data to the second communication device;

The first communication device sends, to the second communication device, first output data corresponding to first input data not associated with second output data;

The first communication device sends the first input data to the second communication device;

The first communication device sends multiple first information corresponding to a first input data to the second communication device, where different first information in the multiple first information is associated with different first AI models;

The second output data includes at least one of the following:

first expected information, where the first expected information is expected output information corresponding to the first input data, determined by the second communication device or protocol based on the first input data and a reference model;

First label data, where the first label data is label data corresponding to the first input data.

In the embodiment of the present application, sending the first output data may be sending the first output data corresponding to multiple first input data together, which can be understood as the packaged sending shown in FIG4 .

As described in the above embodiment, the first input data of the first data set includes at least one of the following items: N first input data associated with the second output data, and M first input data not associated with the second output data, where N and M are both natural numbers, that is, the first input data in the first data set can be labeled or unlabeled, and the first output data corresponding to the labeled first input data can directly obtain the test results and monitoring results of the first AI model on the first communication device side. For example, the target first AI model can be screened out from the first AI model based on the N first input data associated with the second output data and the corresponding second output data, and the M first input data not associated with the second output data are input into the target first AI model to obtain the corresponding first output data. The corresponding first output data (without the corresponding second output data) obtained from the target first AI model can be provided to the second communication device, so that the second communication device can judge the test results and monitoring results of the first AI model.

In an embodiment of the present application, the first output data sent is also the sum of the first output data sets after the same first input data is input into different first AI models.

It can be understood that the second communication device cannot obtain the first input data measured by the first communication device. Optionally, the first communication device provides the first input data measured by the first communication device to the second communication device.

In an embodiment of the present application, first input data can also be provided to the second communication device so that the second communication device determines the test results and monitoring results of the first AI model based on the first input data and the first information.

Optionally, the first AI model is associated with at least one of the following parameters:

The first data set, the second data set, the reference model, the hardware capability, the quantization method, the number of bits reported for the first information, the identification information of the model, the type information of the model, the function information of the model, the level information of the model, and the complexity information of the model;

Among them, the second data set is the training data set of the first AI model.

In an embodiment of the present application, a corresponding first AI model may be determined based on at least one of the first data set, the second data set, the reference model, the hardware capabilities, the quantization method, the number of bits reported in the first information, the identification information of the model, the type information of the model, the function information of the model, the level information of the model, and the complexity information of the model. The quantization method includes fixed-length quantization, variable-length quantization, vector quantization, and the like.

Referring to FIG. 7 , FIG. 7 is a flowchart of another communication method provided in an embodiment of the present application, which is used for a second communication device. As shown in part (a) of FIG. 7 , the method includes the following steps:

Step 701a: The second communication device receives the first information sent by the first communication device.

Optionally, the second communication device includes at least one of a fourth communication device and a monitoring device.

Step 702a: The second communication device determines at least one of a test result and a monitoring result of the first artificial intelligence (AI) model based on the first information.

The first information includes at least one of the following:

First output data, where the first output data is output data obtained based on the first input data of the first data set and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

In another optional embodiment, as shown in part (b) of FIG7 , the method includes the following steps:

Step 701b: The second communication device sends a first data set to the first communication device, for the first communication device to determine the first information;

The first information includes at least one of the following:

First output data, where the first output data is output data obtained based on the first input data of the first data set and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

Optionally, the first AI model includes at least one of a first model and a second model, the first model is a reference model, and the second model is a model determined based on the first model and different from the first model.

Optionally, the reference model includes at least one of the following:

Reference model agreed upon in the agreement;

a reference model determined in conjunction with target model parameters, the target model parameters including model parameters sent by the second communication device;

A reference model determined based on the reference model agreed upon in the protocol and the parameters of the target model;

A reference model trained based on a second data set;

A reference model based on the protocol agreement and a reference model determined by the second data set;

The reference model sent by the second communication device.

Optionally, the first data set further includes second output data, and the second output data includes at least one of the following:

First expected information,

First label data;

The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model;

The first label data is label data corresponding to the first input data.

Optionally, the first test result includes at least one of the following:

first difference information between the first output data and the second output data;

statistical information of first difference information between the first output data and the second output data;

first ratio information between the first output data and the second output data;

statistical information of first ratio information between the first output data and the second output data;

first difference information between the first difference information and a reference model error;

Statistical information of first difference information between the first difference information and the reference model error;

an indication that the first AI model meets the test requirements;

an indication that the first AI model does not meet the test requirements;

Relevant information about the model that meets the test requirements in the first AI model;

The number of first AI models that meet testing requirements.

Optionally, the method further includes at least one of the following:

The second communication device sends first auxiliary information to the first communication device;

The second communication device sends second information to the first communication device;

The second communication device sends third information to the first communication device;

The first auxiliary information includes at least one of the following:

first threshold;

second threshold;

First statistics;

The first threshold includes at least one of the following:

difference threshold information between the first output data and the second output data;

threshold information of a ratio between the first output data and the second output data;

The second threshold includes at least one of the following:

threshold information of statistical information of differences between the first output data and the second output data;

threshold information of statistical information of the ratio of the first output data to the second output data;

The first statistical information is statistical information of a second data set, and the second data set is training data of the first AI model;

The second information includes at least one of the following:

The decompressed codebook information acquired by the second communication device;

a throughput determined by the second communication device;

location information of the terminal determined by the second communication device;

Beam information of beam set A;

The third information includes at least one of the following:

Second dataset;

Reference model;

Target model parameters associated with the reference model.

Optionally, the first data set is associated with the third information.

Optionally, the first information further includes at least one of the following:

identification information of the first data set;

Identification information of the second data set, wherein the second data set is training data for the first AI model;

identification information of the first AI model;

Type information of the first AI model;

Functional information of the first AI model.

Exemplarily, as shown in FIG8 , the monitoring device may determine the first monitoring result by comparing the first output data (the compressed codebook of the first AI model) with the second output data (the expected compressed codebook).

Optionally, the second communication device receiving the first information sent by the first communication device includes:

The second communication device receives the first output data sent by the first communication device;

The second communication device receives first output data corresponding to the plurality of first input data sent by the first communication device;

The second communication device receives first output data corresponding to first input data not associated with second output data sent by the first communication device;

The second communication device receives the first input data sent by the first communication device;

The second communication device receives multiple first information corresponding to a first input data sent by the first communication device, wherein different first information in the multiple first information is associated with different first AI models;

The second output data includes at least one of the following:

first expected information, where the first expected information is expected output information corresponding to the first input data, determined by the second communication device or protocol based on the first input data and a reference model;

First label data, where the first label data is label data corresponding to the first input data.

Optionally, the second communication device determines, based on the first information, at least one of a test result and a monitoring result of the first artificial intelligence (AI) model in the first communication device, including at least one of the following:

If the second condition is met, determining that the first AI model meets the test requirements;

If the second condition is not met, determining that the first AI model does not meet the test requirement;

The second condition includes at least one of the following:

The payload of the first output data is the same as the payload of the first expected information;

The third difference information between the first output data and the first expected information is less than or equal to a third threshold;

Statistical information of the third difference information between the first output data and the first expected information is less than or equal to a fourth threshold;

The second difference information between the third difference information and the reference model error is less than or equal to a fifth threshold;

Statistical information of the third difference information and the second difference information of the reference model error is less than or equal to a sixth threshold;

The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model.

In an embodiment of the present application, at least one of the test results and monitoring results of the first artificial intelligence (AI) model is determined by the second communication device. The second communication device may determine the expected output information (first expected information) corresponding to the first input data based on the first input data and the reference model, or the expected output information (first expected information) corresponding to the first input data determined by the protocol based on the first input data and the reference model. The second communication device may obtain the above-mentioned first expected information and use it to determine the test results and monitoring results of the first artificial intelligence (AI) model.

Similar to how the first communication device determines the test results and monitoring results of the first artificial intelligence (AI) model, the second communication device can also further compare the test results and monitoring results with the reference model error to determine the test results and monitoring results of the first artificial intelligence (AI) model. The meaning of the reference model error has been explained in the previous embodiment and will not be repeated in this embodiment to avoid repetition.

Optionally, when the first input data includes the channel state information or codebook information, the first expected information is the channel state information or codebook information compressed according to a reference model;

Or, in a case where the first input data includes original channel information, the first expected information includes original channel information compressed according to a reference model;

Or, in a case where the first input data includes the channel state information or codebook information, the first expected information is decompressed channel state information or codebook information obtained according to a reference model;

Alternatively, in a case where the first input data includes original channel information, the first expected information includes original channel information decompressed according to a reference model.

Optionally, the second communication is set to a fourth communication device, and the first input data includes at least one of channel state information, codebook information, and original channel information, and the first output data includes at least one of compressed codebook information and compressed original channel information. When the second communication device determines the test result and monitoring result of the first artificial intelligence AI model in the first communication device based on the first information, the method includes:

The second communication device decompresses the first output data to obtain third output data;

If the third condition is met, determining that the first AI model meets the test requirement, or, if the third condition is not met, determining that the first AI model does not meet the test requirement;

The third condition includes at least one of the following:

The precision loss of the third output data relative to the first input data is less than or equal to a seventh threshold;

The accuracy loss statistics of the third output data relative to the first input data is less than or equal to an eighth threshold;

The precision of the third output data is greater than or equal to the precision of the first input data;

The precision loss of the third output data relative to the decompressed first tag data is less than or equal to a ninth threshold;

The accuracy loss statistics of the third output data relative to the decompressed first label data is less than or equal to a tenth threshold;

The precision of the third output data is greater than or equal to the precision of the decompressed first tag data;

Fourth difference information between the third output data and the decompressed first tag data is less than or equal to an eleventh threshold;

Statistical information of fourth difference information between the third output data and the decompressed first tag data is less than or equal to a twelfth threshold.

In an embodiment of the present application, for a CSI compression scenario, the fourth communication device may be a decompression device for the first output data. When the fourth communication device decompresses the first output data to obtain third output data, the fourth communication device compares the third output data with the first input data, or compares the third output data with the decompressed data of the first label data, to determine at least one of a test result and a monitoring result of the first artificial intelligence (AI) model. The decompression process of the first label data may also be performed by the fourth communication device or by the first communication device.

Exemplarily, as shown in FIG9 , the first input data (Test case-codebook) or the first label data after decompression (the decompressed codebook of the reference model label) is compared with the third output data (the restored codebook after decompression) to determine the monitoring result.

Optionally, the method further includes:

The second communication device sends feedback information to the first communication device;

The feedback information includes at least one of the following:

an indication that the first AI model meets the test requirements;

an indication that the first AI model does not meet the test requirements;

Relevant information about the model that meets the test requirements in the first AI model;

Training to obtain relevant information of a second data set of the first AI model;

The number of first AI models that meet testing requirements.

In an embodiment of the present application, the second communication device can also provide relevant results to the first communication device when determining at least one of the test results and monitoring results of the first artificial intelligence AI model. When the second communication device is a fourth communication device, the above-mentioned relevant results can also be sent by the fourth communication device to the monitoring device.

For example, the second communication device indicates the reference model or optimization model in the first AI model used by the first communication device, or indicates the specific optimization model used. It can be understood that when the first communication device reports the output of the optimization model, the opposite device can determine the model that the first communication device should use based on the output.

In one embodiment, the second communication device indication information may indicate whether to use a reference model;

In an optional embodiment, if the output obtained by the terminal (first communication device) through the first AI model fails to pass the test of the first data set (test case), the second communication device instructs the first communication device to fallback to the reference model.

In yet another embodiment, the second communication device indicates identification information of the model (eg, model ID).

In an optional embodiment, if the second communication device obtains the first information obtained by a different first AI model, the second peer device instructs the first communication device on the model to be subsequently applied.

In another embodiment, the second peer device indicates category information of the model.

Optionally, the first AI model is associated with at least one of the following parameters:

The first data set, the second data set, the reference model, the hardware capability, the quantization method, the number of bits reported for the first information, the identification information of the model, the type information of the model, the function information of the model, the level information of the model, and the complexity information of the model;

Among them, the second data set is the training data set of the first AI model.

It should be noted that this embodiment is an implementation method of the second communication device corresponding to the embodiment shown in Figures 5-6. Its specific implementation method can refer to the relevant descriptions in the embodiment shown in Figures 5-6, and both can achieve the same or similar beneficial effects. To avoid repetitive description, this embodiment will not be repeated.

Referring to FIG. 10 , FIG. 10 is a flow chart of another communication method provided in an embodiment of the present application, which is used for monitoring a device. As shown in part (a) of FIG. 10 , the method includes the following steps:

Step 1001a: The monitoring device receives fourth output data sent by the fourth communication device, where the fourth output data is decompressed data of the first output data;

Step 1002a: The monitoring device determines at least one of a test result and a monitoring result of the first artificial intelligence (AI) model based on the fourth output data.

Among them, the fourth communication device is used to receive the first output data sent by the first communication device and decompress the first output data, where the first output data is output data obtained based on the first input number of the first data set and the first AI model.

In another embodiment, as shown in part (b) of FIG10 , the method includes the following steps:

Step 1001b: The monitoring device receives at least one of the test result and the monitoring result of the first artificial intelligence AI model sent by the first communication device or the fourth communication device;

Step 1002b: The monitoring device determines at least one of the following based on at least one of the test result and the monitoring result of the first artificial intelligence (AI) model:

Target-first AI model;

Status information of the first AI model;

wherein the target first AI model is at least one of the first AI models;

Among them, the fourth communication device is used to receive the first output data sent by the first communication device and decompress the first output data, where the first output data is output data obtained based on the first input number of the first data set and the first AI model.

In an embodiment of the present application, the fourth communication device receives the first output data sent by the first communication device and decompresses the first output data. The fourth communication device sends the decompressed data to the monitoring device, and the monitoring device performs a judgment on at least one of the test results and the monitoring results of the first artificial intelligence (AI) model; or, the fourth communication device sends at least one of the test results and the monitoring results of the first artificial intelligence (AI) model to the monitoring device, and the monitoring device further determines the status information of the target first AI model and/or the first AI model. For example, the fourth communication device can send the difference information between the fourth output data and the first input data to the monitoring device, and the monitoring device specifically judges the status information of the first AI model, or selects the target first AI model from multiple first AI models.

Optionally, when the first input data includes at least one of channel state information, codebook information, and original channel information, and the first output data includes at least one of compressed codebook information and compressed original channel information, the monitoring device determines the test result and monitoring result of the first artificial intelligence AI model in the first communication device based on the fourth output data, including:

If the fourth condition is met, determining that the first AI model meets the test requirements, or, if the fourth condition is not met, determining that the first AI model does not meet the test requirements;

The fourth condition includes at least one of the following:

The precision loss of the fourth output data relative to the first input data is less than or equal to a thirteenth threshold;

The accuracy loss statistics of the fourth output data relative to the first input data is less than or equal to a fourteenth threshold;

The precision of the fourth output data is greater than or equal to the precision of the first input data;

The precision loss of the fourth output data relative to the decompressed first tag data is less than or equal to a fifteenth threshold;

The accuracy loss statistics of the fourth output data relative to the decompressed first label data is less than or equal to a sixteenth threshold;

The precision of the fourth output data is greater than or equal to the precision of the decompressed first tag data;

A third difference information between the fourth output data and the decompressed first tag data is less than or equal to a seventeenth threshold;

Statistical information of third difference information between the fourth output data and the decompressed first label data is less than or equal to an eighteenth threshold.

In the embodiment of the present application, the judgment logic of the test results and monitoring results of the first artificial intelligence AI model is similar to the judgment logic of the second communication device, and will not be repeated here.

In an embodiment of the present application, the test results and monitoring results of the first artificial intelligence (AI) model are determined based on the first output data of the first data set in the first AI model. The above-mentioned determination process can be performed in the first communication device and the second communication device (including the fourth communication device and/or the monitoring setting). Taking the CSI compression scenario, in which the terminal deploys the first AI model to perform compression as an example, the first data set (test case) is used to determine whether the first AI model of the terminal meets the first requirement (in an optional embodiment, meeting the first requirement can be understood as passing the first data set (test case) test), wherein the first requirement includes at least one of the following:

a) the output of the first AI model or the compressed codebook information or raw channel information reported by the first communication device is the same as the expected compressed codebook information or raw channel information payload;

b) the error between the output of the first AI model or the compressed codebook information or raw channel information reported by the first communication device and the expected compressed codebook information or raw channel information is less than a first preset threshold;

c) the variance between the output of the first AI model or the compressed codebook information or raw channel information reported by the first communication device and expected statistical information on the error between the compressed codebook information or raw channel information is less than a second preset threshold;

d) the codebook information or raw channel information recovered by the fourth communication device has similar or higher accuracy than the codebook information or raw channel information in the first data set (test case);

e) a loss in accuracy between the codebook information or raw channel information recovered by the fourth communication device and the codebook information or raw channel information in the first data set (test case) is less than a third preset threshold;

f) The statistical accuracy loss between the codebook information or raw channel information recovered by the fourth communication device and the codebook information or raw channel information in the first data set (test case) is less than a fourth preset threshold.

In an embodiment of the present application, when the process of determining the test results and monitoring results of the first artificial intelligence AI model is executed in different communication devices (first communication device, fourth communication device or monitoring setting), the values of the above-mentioned first preset threshold, second preset threshold, third preset threshold, and fourth preset threshold may be different.

In an embodiment of the present application, the first data set (test case) can assist the first communication device in determining a suitable AI model, thereby helping to optimize the output results of non-test cases in subsequent applications, or, the first data set (test case) can assist the first communication device in determining the monitoring results of the first communication device, or, the first data set (test case) can assist the first communication device in determining the requirement of the first communication device, for example, the lower limit of accuracy.

In an embodiment of the present application, the first data set (test case) is a special data set, and the different first communication devices need to confirm the performance of the reference model and/or optimization model based on the test case.

In an embodiment of the present application, during testing or actual use (or model performance monitoring), the second communication device tests or verifies the first AI model deployed by the first communication device in order to determine that the first communication device has implemented the first AI model in accordance with a prescribed or occurring reference model. Alternatively, during actual use, or in order to guide the first communication device to output results similar to test cases, some test cases are provided before obtaining the AI-based output.

It should be noted that this embodiment is an implementation method of the fourth communication device corresponding to the embodiment shown in Figures 5-7. Its specific implementation method can refer to the relevant descriptions in the embodiments shown in Figures 5-7, and both can achieve the same or similar beneficial effects. To avoid repetitive descriptions, this embodiment will not be repeated.

The communication device provided in the embodiment of the present application can be executed by a communication device. In the embodiment of the present application, the communication device 1100 provided in the embodiment of the present application is illustrated in FIG11 by taking the communication device executing the communication device as an example.

A first acquisition module 1101 is configured to acquire a first data set, where the first data set includes first input data;

A first determining module 1102 is configured to determine first information based on the first data set and a first artificial intelligence (AI) model;

The first information includes at least one of the following:

first output data, where the first output data is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

Optionally, the first data set further includes second output data, and the second output data includes at least one of the following:

First expected information;

First label data;

The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model;

The first label data is label data corresponding to the first input data.

Optionally, the first determining module includes at least one of the following:

A first determination submodule, configured to determine that the first AI model meets the test requirement when a first condition is met;

A second determining submodule, configured to determine that the first AI model does not meet the test requirement when the first condition is not met;

The first condition includes at least one of the following:

The payload of the first output data is the same as the payload of the second output data;

First difference information between the first output data and the second output data is less than or equal to a first threshold;

Statistical information of the first difference information between the first output data and the second output data is less than or equal to a second threshold.

Optionally, the device further includes:

a first receiving module, configured to receive first auxiliary information sent by a second communication device or a third communication device, where the second communication device includes at least one of a fourth communication device and a monitoring device;

The first auxiliary information includes at least one of the following:

first threshold;

second threshold;

First statistics;

The first threshold value includes at least one of the following:

difference threshold information between the first output data and the second output data;

threshold information of a ratio between the first output data and the second output data;

The second threshold value includes at least one of the following:

threshold information of statistical information of differences between the first output data and the second output data;

threshold information of statistical information of the ratio of the first output data to the second output data;

The first statistical information is statistical information of a second data set, and the second data set is training data of the first AI model;

The first determining module includes:

The third determination submodule is used to determine the first information based on the first data set, the first AI model and the first auxiliary information.

Optionally, the first information includes first monitoring information, and the apparatus further includes:

A second receiving module, configured to receive second information sent by a second communication device;

The second information includes at least one of the following:

decompressed codebook information acquired by the second communication device;

a throughput determined by the second communication device;

location information of the terminal determined by the second communication device;

Beam information of beam set A;

The first determining module includes:

The fourth determining submodule is configured to determine the first monitoring information according to the first data set, the first output data and the second information.

Optionally, the device further includes:

a third receiving module, configured to receive third information sent by a second communication device or a third communication device, where the second communication device includes at least one of a fourth communication device and a monitoring device;

The third information includes at least one of the following:

Second dataset;

Reference model;

Target model parameters associated with the reference model.

Optionally, when the first output data includes at least one of decompressed codebook information and compressed original channel information, and the first information includes a first test result, the first determining module includes at least one of the following:

a fifth determining submodule, configured to determine the first test result according to at least one of second difference information between the first output data and the first input data and a statistical value of the second difference information;

The sixth determining submodule is configured to determine the first test result based on at least one of second ratio information of the first output data to the first input data and a statistical value of the second ratio information.

Optionally, the device further includes:

an execution module, configured to, when at least one of the test result and the monitoring result of the first AI model fails, perform at least one of the following operations:

Switch to the second AI model;

Switch to non-AI function;

Fall back to the reference model;

Fall back to non-AI functionality;

Update reference models;

Update reference model parameters;

receiving a reference pattern repeatedly sent by a second communication device or a third communication device;

receiving a reference model parameter repeatedly sent by the second communication device or the third communication device;

receiving an updated reference model sent by the second communication device or the third communication device;

receiving updated reference model parameters sent by the second communication device or the third communication device.

Optionally, the device further includes:

The first sending module is configured to send the first information to a second communication device, wherein the second communication device includes at least one of a fourth communication device and a monitoring device.

Optionally, the first sending module includes:

A first sending submodule, configured to send the first output data to a second communication device;

A second sending submodule, configured to send first output data corresponding to the plurality of first input data to a second communication device;

a third sending submodule, configured to send, to the second communication device, first output data corresponding to the first input data not associated with the second output data;

a fourth sending submodule, configured to send the first input data to a second communication device;

A fifth sending submodule, configured to send a plurality of first information corresponding to a first input data to the second communication device, wherein different first information in the plurality of first information are associated with different first AI models;

The second output data includes at least one of the following:

first expected information, where the first expected information is expected output information corresponding to the first input data, determined by the second communication device or protocol based on the first input data and a reference model;

It should be noted that the communication device provided in the embodiment of the present application is a device capable of executing the above-mentioned communication method. Therefore, all implementation methods in the above-mentioned communication method embodiment are applicable to the communication device and can achieve the same or similar beneficial effects. To avoid repetition, this embodiment will not be described in detail.

The communication device provided in the embodiment of the present application can be executed by a communication device. In the embodiment of the present application, the communication device executing the communication device is taken as an example, as shown in Figure 12, which illustrates the communication devices 1200a and 1200b provided in the embodiment of the present application.

The apparatus 1200a includes:

The fourth receiving module 1201a is configured to receive first information sent by the first communication device;

A second determining module 1202a is configured to determine at least one of a test result and a monitoring result of a first artificial intelligence (AI) model based on the first information;

Alternatively, the apparatus 1200b includes:

A second sending module 1201b is configured to send a first data set to a first communication device, so that the first communication device determines first information;

The first information includes at least one of the following:

First output data, where the first output data is output data obtained based on the first input data of the first data set and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

Optionally, the first data set further includes second output data, and the second output data includes at least one of the following:

First expected information;

First label data;

The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model;

The first label data is label data corresponding to the first input data.

Optionally, the device further includes:

A third sending module, configured to send first auxiliary information to the first communication device;

a fourth sending module, configured to send second information to the first communication device;

a fifth sending module, configured to send third information to the first communication device;

The first auxiliary information includes at least one of the following:

first threshold;

second threshold;

First statistics;

The first threshold includes at least one of the following:

difference threshold information between the first output data and the second output data;

threshold information of a ratio between the first output data and the second output data;

The second threshold includes at least one of the following:

threshold information of statistical information of differences between the first output data and the second output data;

threshold information of statistical information of the ratio of the first output data to the second output data;

The first statistical information is statistical information of a second data set, and the second data set is training data of the first AI model;

The second information includes at least one of the following:

decompressed codebook information acquired by the second communication device;

a throughput determined by the second communication device;

location information of the terminal determined by the second communication device;

Beam information of beam set A;

The third information includes at least one of the following:

Second dataset;

Reference model;

Target model parameters associated with the reference model.

Optionally, the fourth receiving module includes:

a first receiving submodule, configured to receive the first output data sent by the first communication device;

a second receiving submodule, configured to receive first output data corresponding to a plurality of first input data sent by the first communication device;

a third receiving submodule, configured to receive first output data corresponding to first input data not associated with second output data and sent by the first communication device;

a fourth receiving submodule, configured to receive the first input data sent by the first communication device;

A fifth receiving submodule, configured to receive multiple first information corresponding to a first input data sent by the first communication device, wherein different first information in the multiple first information are associated with different first AI models;

The second output data includes at least one of the following:

first expected information, where the first expected information is expected output information corresponding to the first input data, determined by a second communication device or protocol based on the first input data and a reference model;

First label data, where the first label data is label data corresponding to the first input data.

Optionally, the second determining module includes at least one of the following:

a seventh determination submodule, configured to determine that the first AI model meets the test requirement when the second condition is met;

an eighth determining submodule, configured to determine that the first AI model does not meet the test requirement if the second condition is not met;

The second condition includes at least one of the following:

The payload of the first output data is the same as the payload of the first expected information;

The third difference information between the first output data and the first expected information is less than or equal to a third threshold;

Statistical information of the third difference information between the first output data and the first expected information is less than or equal to a fourth threshold;

The second difference information between the third difference information and the reference model error is less than or equal to a fifth threshold;

Statistical information of the third difference information and the second difference information of the reference model error is less than or equal to a sixth threshold;

The first expected information is expected output information corresponding to the first input data, determined by the second communication device or protocol based on the first input data and a reference model.

Optionally, the communication apparatus is a fourth communication device, and when the first input data includes at least one of channel state information, codebook information, and original channel information, and the first output data includes at least one of compressed codebook information and compressed original channel information, the second determination module includes:

a first decompression module, configured to decompress the first output data to obtain third output data;

a ninth determining submodule, configured to determine that the first AI model meets the test requirement if a third condition is met, or to determine that the first AI model does not meet the test requirement if the third condition is not met;

The third condition includes at least one of the following:

The precision loss of the third output data relative to the first input data is less than or equal to a seventh threshold;

The accuracy loss statistics of the third output data relative to the first input data is less than or equal to an eighth threshold;

The precision of the third output data is greater than or equal to the precision of the first input data;

The precision loss of the third output data relative to the decompressed first label data is less than or equal to a ninth threshold;

The accuracy loss statistics of the third output data relative to the decompressed first label data is less than or equal to a tenth threshold;

The precision of the third output data is greater than or equal to the precision of the decompressed first tag data;

The fourth difference information between the third output data and the decompressed first label data is less than or equal to an eleventh threshold;

Statistical information of fourth difference information between the third output data and the decompressed first label data is less than or equal to a twelfth threshold.

Optionally, the device further includes:

a sixth sending module, configured to send feedback information to the first communication device;

The feedback information includes at least one of the following:

an indication that the first AI model meets the test requirements;

an indication that the first AI model does not meet the test requirements;

Relevant information about the model that meets the test requirements in the first AI model;

Training to obtain relevant information of a second data set of the first AI model;

The number of first AI models that meet testing requirements.

It should be noted that the communication device provided in the embodiment of the present application is a device capable of executing the above-mentioned communication method. Therefore, all implementation methods in the above-mentioned communication method embodiment are applicable to the communication device and can achieve the same or similar beneficial effects. To avoid repetition, this embodiment will not be described in detail.

The communication device provided in the embodiment of the present application can be executed by a communication device. In the embodiment of the present application, the communication device executing the communication device is taken as an example, as shown in Figure 13, which illustrates the communication devices 1300a and 1300b provided in the embodiment of the present application.

The apparatus 1300a includes:

a fifth receiving module 1301a, configured to receive fourth output data sent by a fourth communication device, where the fourth output data is decompressed data of the first output data;

A third determining module 1302a is configured to determine at least one of a test result and a monitoring result of the first artificial intelligence (AI) model based on the fourth output data;

Alternatively, the apparatus 1300b includes:

a sixth receiving module 1301b, configured to receive at least one of a test result and a monitoring result of the first artificial intelligence AI model sent by the first communication device or the fourth communication device;

The fourth determining module 1302b is configured to determine at least one of the following based on at least one of the test result and the monitoring result of the first artificial intelligence (AI) model:

Target-first AI model;

Status information of the first AI model;

wherein the target first AI model is at least one of the first AI models;

Among them, the fourth communication device is used to receive the first output data sent by the first communication device and decompress the first output data, where the first output data is output data obtained based on the first input number of the first data set and the first AI model.

Optionally, when the first input data includes at least one of channel state information, codebook information, and original channel information, and the first output data includes at least one of compressed codebook information and compressed original channel information, the third determination module is configured to include at least one of the following:

A fifth determination module, configured to determine whether the first AI model meets the test requirement when the fourth condition is met;

a sixth determining module, configured to determine that the first AI model does not meet the test requirement if the fourth condition is not met;

The fourth condition includes at least one of the following:

The precision loss of the fourth output data relative to the first input data is less than or equal to a thirteenth threshold;

The accuracy loss statistics of the fourth output data relative to the first input data is less than or equal to a fourteenth threshold;

The precision of the fourth output data is greater than or equal to the precision of the first input data;

The precision loss of the fourth output data relative to the decompressed first label data is less than or equal to a fifteenth threshold;

The accuracy loss statistics of the fourth output data relative to the decompressed first label data is less than or equal to a sixteenth threshold;

The precision of the fourth output data is greater than or equal to the precision of the decompressed first tag data;

The third difference information between the fourth output data and the decompressed first label data is less than or equal to a seventeenth threshold;

Statistical information of the third difference information between the fourth output data and the decompressed first label data is less than or equal to an eighteenth threshold.

It should be noted that the communication device provided in the embodiment of the present application is a device capable of executing the above-mentioned communication method. Therefore, all implementation methods in the above-mentioned communication method embodiment are applicable to the communication device and can achieve the same or similar beneficial effects. To avoid repetition, this embodiment will not be described in detail.

The communication device 1100, communication device 1200a, communication device 1200b, communication device 1300a or communication device 1300b in the embodiments of the present application can be an electronic device, such as an electronic device with an operating system, or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal, or it can be other devices other than a terminal. For example, the terminal can include but is not limited to the types of terminal 11 listed above, and other devices can be servers, network attached storage (NAS), etc., which are not specifically limited in the embodiments of the present application.

The communication device 1100, communication device 1200a, communication device 1200b, communication device 1300a or communication device 1300b provided in the embodiments of the present application can implement the various processes implemented by the method embodiments of Figures 2 to 10 and achieve the same technical effects. To avoid repetition, they will not be repeated here.

As shown in Figure 14, an embodiment of the present application further provides a communication device 1400, including a processor 1401 and a memory 1402, wherein the memory 1402 stores a program or instruction that can be run on the processor 1401. For example, when the communication device 1400 is a terminal, the program or instruction, when executed by the processor 1401, implements the various steps of the communication method embodiment shown in Figure 5, Figure 7, or Figure 10, and can achieve the same technical effect. When the communication device 1400 is a network-side device, the program or instruction, when executed by the processor 1401, implements the various steps of the communication method embodiment shown in Figure 5, Figure 7, or Figure 10, and can achieve the same technical effect. To avoid repetition, they are not described here.

The present application also provides a terminal including a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is configured to execute a program or instruction to implement the steps of the method embodiment shown in Figure 5 or Figure 7. This terminal embodiment corresponds to the above-mentioned terminal-side method embodiment, and each implementation process and implementation method of the above-mentioned method embodiment can be applied to this terminal embodiment and can achieve the same technical effect. Specifically, Figure 15 is a schematic diagram of the hardware structure of a terminal implementing an embodiment of the present application.

The terminal 1500 includes but is not limited to: a radio frequency unit 1501, a network module 1502, an audio output unit 1503, an input unit 1504, a sensor 1505, a display unit 1506, a user input unit 1507, an interface unit 1508, a memory 1509 and at least some of the components of the processor 1510.

Those skilled in the art will appreciate that the terminal 1500 may also include a power supply (such as a battery) to power various components. The power supply may be logically connected to the processor 1510 via a power management system, thereby enabling the power management system to manage charging, discharging, and power consumption. The terminal structure shown in FIG15 does not limit the terminal. The terminal may include more or fewer components than shown, or may combine certain components, or have different component arrangements, which will not be described in detail here.

It should be understood that in the embodiment of the present application, the input unit 1504 may include a graphics processing unit (GPU) 15041 and a microphone 15042, and the graphics processor 15041 processes the image data of the static picture or video obtained by the image capture device (such as a camera) in the video capture mode or the image capture mode. The display unit 1506 may include a display panel 15061, and the display panel 15061 may be configured in the form of a liquid crystal display, an organic light emitting diode, etc. The user input unit 1507 includes a touch panel 15071 and at least one of the other input devices 15072. The touch panel 15071 is also called a touch screen. The touch panel 15071 may include two parts: a touch detection device and a touch controller. Other input devices 15072 may include but are not limited to a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, and a joystick, which will not be repeated here.

In the embodiment of the present application, after receiving downlink data from a network-side device, the radio frequency unit 1501 may transmit the data to the processor 1510 for processing. Furthermore, the radio frequency unit 1501 may send uplink data to the network-side device. Typically, the radio frequency unit 1501 includes, but is not limited to, an antenna, an amplifier, a transceiver, a coupler, a low-noise amplifier, a duplexer, and the like.

Memory 1509 can be used to store software programs or instructions and various data. Memory 1509 may primarily include a first storage area for storing programs or instructions and a second storage area for storing data. The first storage area may store an operating system, applications or instructions required for at least one function (such as a sound playback function, an image playback function, etc.). Furthermore, memory 1509 may include volatile memory or non-volatile memory. The non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be random access memory (RAM), static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct RAM (DRRAM). The memory 1509 in the embodiments of the present application includes, but is not limited to, these and any other suitable types of memory.

Processor 1510 may include one or more processing units. Optionally, processor 1510 integrates an application processor and a modem processor. The application processor primarily handles operations related to the operating system, user interface, and application programs, while the modem processor primarily processes wireless communication signals, such as a baseband processor. It is understood that the modem processor may not be integrated into processor 1510.

The radio frequency unit 1501 is configured to obtain a first data set, where the first data set includes first input data;

Processor 1510, configured to determine first information based on the first data set and a first artificial intelligence (AI) model;

Alternatively, the radio frequency unit 1501 is configured to receive first information sent by a first communication device;

The processor 1510 is configured to determine at least one of a test result and a monitoring result of a first artificial intelligence AI model based on the first information;

Alternatively, the radio frequency unit 1501 is configured to send a first data set to the first communication device, for the first communication device to determine the first information;

The first information includes at least one of the following:

first output data, where the first output data is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model;

a first test result, where the first test result is used to determine a test result of the first AI model;

First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model.

It can be understood that the implementation process of each implementation method mentioned in this embodiment can refer to the relevant description of the method embodiment Figure 5 or Figure 7, and achieve the same or corresponding technical effects. To avoid repetition, it will not be repeated here.

The present application also provides a network-side device, including a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is configured to execute a program or instruction to implement the steps of the method embodiment shown in Figure 5 or Figure 7. This network-side device embodiment corresponds to the aforementioned network-side device method embodiment, and each implementation process and implementation method of the aforementioned method embodiment is applicable to this network-side device embodiment and can achieve the same technical effects.

Specifically, embodiments of the present application also provide a network-side device. As shown in Figure 16, network-side device 1600 includes an antenna 161, a radio frequency device 162, a baseband device 163, a processor 164, and a memory 165. Antenna 161 is connected to radio frequency device 162. In the uplink direction, radio frequency device 162 receives information via antenna 161 and sends the received information to baseband device 163 for processing. In the downlink direction, baseband device 163 processes the information to be transmitted and sends it to radio frequency device 162. Radio frequency device 162 processes the received information and then sends it through antenna 161.

The method executed by the network-side device in the above embodiment may be implemented in the baseband device 163 , which includes a baseband processor.

The baseband device 163 may include, for example, at least one baseband board, on which multiple chips are arranged, as shown in Figure 16, one of the chips is, for example, a baseband processor, which is connected to the memory 165 through a bus interface to call the program in the memory 165 and execute the network device operations shown in the above method embodiment.

The network side device may also include a network interface 166, which is, for example, a Common Public Radio Interface (CPRI).

Specifically, the network side device 1600 of the embodiment of the present application also includes: instructions or programs stored in the memory 165 and executable on the processor 164. The processor 164 calls the instructions or programs in the memory 165 to execute the methods executed by the modules shown in FIG. 11 , FIG. 12 , or FIG. 13 , and achieves the same technical effect. To avoid repetition, it will not be elaborated here.

Specifically, an embodiment of the present application further provides a network-side device. As shown in FIG17 , the network-side device 1700 includes a processor 1701, a network interface 1702, and a memory 1703. The network interface 1702 is, for example, a common public radio interface (CPRI).

Specifically, the network side device 1700 of the embodiment of the present application also includes: instructions or programs stored in the memory 1703 and can be run on the processor 1701. The processor 1701 calls the instructions or programs in the memory 1703 to execute the methods executed by each module shown in Figure 11 or Figure 12 or Figure 13, and achieves the same technical effect. To avoid repetition, it will not be repeated here.

An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, the various processes of the communication method embodiment shown in Figure 5, Figure 7, or Figure 10 above are implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

The processor is the processor in the terminal described in the above embodiment. The readable storage medium includes a computer-readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk. In some examples, the readable storage medium may be a non-transitory readable storage medium.

An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the communication method embodiments shown in Figures 5, 7, or 10 above, and can achieve the same technical effects. To avoid repetition, they will not be repeated here.

It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

An embodiment of the present application further provides a computer program/program product, which is stored in a storage medium and is executed by at least one processor to implement the various processes of the communication embodiment shown in Figures 5, 7, or 10 above, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.

An embodiment of the present application further provides a communication system, including a terminal and a network-side device, wherein the terminal may be configured to perform the steps of the method shown in FIG5 , and the network-side device may be configured to perform the steps of the method shown in FIG7 ; or, the terminal may be configured to perform the steps of the method shown in FIG7 , and the network-side device may be configured to perform the steps of the method shown in FIG5 ;

Or, including: terminals, network side equipment and monitoring equipment;

The terminal may be used to execute the steps of the method described in FIG. 5 , the network-side device may be used to execute the steps of the method described in FIG. 7 , and the monitoring device may be used to execute the steps of the method described in FIG. 10 .

It should be noted that, in this article, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved. For example, the described method may be performed in an order different from that described, and various steps may also be added, omitted or combined. In addition, the features described with reference to certain examples may be combined in other examples.

Through the description of the above embodiments, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of a computer software product plus a necessary general-purpose hardware platform, or of course, by hardware. The computer software product is stored in a storage medium (such as ROM, RAM, magnetic disk, optical disk, etc.) and includes a number of instructions for enabling a terminal or network-side device to execute the methods described in each embodiment of the present application.

The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of this application, ordinary technicians in this field can also make many forms of implementation methods without departing from the purpose of this application and the scope of protection of the claims. These implementation methods are all within the protection of this application.

Claims (63)

A communication method, comprising: A first communication device acquires a first data set, where the first data set includes first input data; The first communication device determines first information based on the first data set and a first artificial intelligence (AI) model; The first information includes at least one of the following: first output data, where the first output data is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model; a first test result, where the first test result is used to determine a test result of the first AI model; First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model. The method according to claim 1, wherein the first communication device determines, based on the first data set and the first artificial intelligence (AI) model, first information including at least one of the following: The first communication device inputs the first input data into the first AI model to obtain the first output data; The first communication device determines at least one of the first test result and the first monitoring information according to the first data set and the first output data. The method according to claim 1 or 2, wherein the first data set includes at least one of the following: A standard data set, a data set sent by the second communication device, and first input data measured by the first communication device. The method according to any one of claims 1 to 3, wherein the first AI model includes at least one of a first model and a second model, the first model is a reference model, and the second model is a model determined based on the first model and different from the first model. The method according to claim 4, wherein the reference model includes at least one of the following: Reference model agreed upon in the agreement; a reference model determined in conjunction with target model parameters, the target model parameters including model parameters sent by the second communication device; A reference model determined based on the reference model agreed upon in the protocol and the parameters of the target model; A reference model trained based on a second data set; A reference model based on the protocol agreement and a reference model determined by the second data set; The reference model sent by the second communication device. The method according to any one of claims 1 to 5, wherein the first data set further includes second output data, and the second output data includes at least one of the following: First expected information; First label data; The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model; The first label data is label data corresponding to the first input data. The method according to claim 6, wherein the first input data of the first data set includes at least one of the following: N first input data associated with the second output data, and M first input data not associated with the second output data, where N and M are both natural numbers. The method according to any one of claims 6-7, wherein the first test result includes at least one of the following: first difference information between the first output data and the second output data; statistical information of first difference information between the first output data and the second output data; first ratio information between the first output data and the second output data; statistical information of first ratio information between the first output data and the second output data; first difference information between the first difference information and a reference model error; Statistical information of first difference information between the first difference information and the reference model error; an indication that the first AI model meets the test requirements; an indication that the first AI model does not meet the test requirements; Relevant information about the model that meets the test requirements in the first AI model; The number of first AI models that meet testing requirements. The method according to claim 8, wherein the first communication device determines, based on the first data set and the first artificial intelligence (AI) model, the first information, including at least one of the following: If the first condition is met, determining that the first AI model meets the test requirements; If the first condition is not met, determining that the first AI model does not meet the test requirement; The first condition includes at least one of the following: The payload of the first output data is the same as the payload of the second output data; First difference information between the first output data and the second output data is less than or equal to a first threshold; Statistical information of the first difference information between the first output data and the second output data is less than or equal to a second threshold. The method according to any one of claims 6 to 9, further comprising: The first communication device receives first auxiliary information sent by a second communication device or a third communication device, where the second communication device includes at least one of a fourth communication device and a monitoring device; The first auxiliary information includes at least one of the following: first threshold; second threshold; First statistics; The first threshold value includes at least one of the following: difference threshold information between the first output data and the second output data; threshold information of a ratio between the first output data and the second output data; The second threshold value includes at least one of the following: threshold information of statistical information of differences between the first output data and the second output data; threshold information of statistical information of the ratio of the first output data to the second output data; The first statistical information is statistical information of a second data set, and the second data set is training data of the first AI model; The first communication device determines first information based on the first data set and the first artificial intelligence (AI) model, including: The first communication device determines the first information based on the first data set, the first AI model and the first auxiliary information. The method according to any one of claims 1 to 10, wherein the first information includes first monitoring information, and the method further comprises: The first communication device receives second information sent by the second communication device; The second information includes at least one of the following: decompressed codebook information acquired by the second communication device; a throughput determined by the second communication device; location information of the terminal determined by the second communication device; Beam information of beam set A; The first communication device determines first information based on the first data set and the first artificial intelligence (AI) model, including: The first communication device determines the first monitoring information according to the first data set, the first output data and the second information. The method according to any one of claims 1 to 11, further comprising: The first communication device receives third information sent by the second communication device or the third communication device, where the second communication device includes at least one of a fourth communication device and a monitoring device; The third information includes at least one of the following: Second dataset; Reference model; Target model parameters associated with the reference model. The method of claim 12, wherein the first data set is associated with the third information. The method according to any one of claims 1 to 13, wherein the first input data includes at least one of the following: Channel state information; Codebook information; Original channel information; Delay profile DP; Power delay profile PDP; Channel impulse response CIR; Delay information corresponding to multipath; Delay power information corresponding to multipath; Beam information. The method according to claim 14, wherein In a case where the first input data includes the channel state information or codebook information, the first output data includes compressed channel state information or codebook information; Or, in a case where the first input data includes original channel information, the first output data includes compressed original channel information; Or, in a case where the first input data includes the channel state information or codebook information, the first output data includes decompressed channel state information or codebook information; Or, in a case where the first input data includes original channel information, the first output data includes decompressed original channel information; Or, when the first input data includes at least one of a delay profile DP, a power delay profile PDP, and a channel impulse response CIR, the first output data includes at least one of the location information of the terminal, the distance information between the terminal and other devices, the round-trip delay RTT, the reference signal time difference RSTD, and the time difference between transmitting and receiving signals; Alternatively, when the first input data includes beam information corresponding to beam set B, the first output data includes beam information corresponding to beam set A. The method according to claim 15, wherein the first output data includes at least one of decompressed codebook information and compressed original channel information, and when the first information includes a first test result, the first communications device determines, based on the first data set and the first artificial intelligence (AI) model, the first information, including at least one of the following: determining the first test result according to at least one of second difference information between the first output data and the first input data and a statistical value of the second difference information; The first test result is determined according to at least one of second ratio information of the first output data and the first input data and a statistical value of the second ratio information. The method according to any one of claims 14 to 16, wherein the first data set further includes first expected information; In a case where the first input data includes the channel state information or codebook information, the first expected information is the channel state information or codebook information compressed according to a reference model; Or, in a case where the first input data includes original channel information, the first expected information includes original channel information compressed according to a reference model; Or, in a case where the first input data includes the channel state information or codebook information, the first expected information is decompressed channel state information or codebook information obtained according to a reference model; Alternatively, in a case where the first input data includes original channel information, the first expected information includes original channel information decompressed according to a reference model. The method according to any one of claims 14 to 17, wherein the first data set further includes first label data corresponding to the first input data; In a case where the first input data includes the channel state information or codebook information, the first label data is compressed channel state information or codebook information; Or, in a case where the first input data includes original channel information, the first label data is compressed original channel information; Or, when the first input data includes the channel state information or codebook information, the first tag data is decompressed channel state information or codebook information; Or, in a case where the first input data includes original channel information, the first tag data is decompressed original channel information; Or, when the first input data includes at least one of a delay profile DP, a power delay profile PDP, and a channel impulse response CIR, the first label data includes at least one of a location label of the terminal, a distance label between the terminal and other devices, a round-trip delay RTT, a reference signal time difference RSTD label, and a time difference label of a transmitting and receiving signal; Alternatively, when the first input data includes beam information corresponding to set B, the first label data includes beam labels corresponding to set A. The method according to any one of claims 1 to 18, wherein the first data set further includes first label data corresponding to the first input data; the first label data is associated with at least one of the following information: The label data type of the second data set, the statistical information corresponding to the first input data, and the feature information corresponding to the first data set; Among them, the second data set is the training data of the first AI model. The method according to any one of claims 1 to 19, further comprising: If at least one of the test result and the monitoring result of the first AI model fails, perform at least one of the following operations: Switch to the second AI model; Switch to non-AI function; Fall back to the reference model; Fall back to non-AI functionality; Update reference models; Update reference model parameters; receiving a reference pattern repeatedly sent by a second communication device or a third communication device; receiving a reference model parameter repeatedly sent by the second communication device or the third communication device; receiving an updated reference model sent by the second communication device or the third communication device; Receive updated reference model parameters sent by the second communication device or the third communication device. The method according to any one of claims 1 to 20, further comprising: The first communication device sends the first information to a second communication device, wherein the second communication device includes at least one of a fourth communication device and a monitoring device. The method according to claim 21, wherein the first information further includes at least one of the following: identification information of the first data set; Identification information of a second data set, wherein the second data set is training data for the first AI model; identification information of the first AI model; Type information of the first AI model; Functional information of the first AI model. The method according to any one of claims 21-22, wherein the first communication device sends the first information to the second communication device, including at least one of the following: The first communication device sends the first output data to the second communication device; The first communication device sends first output data corresponding to the plurality of first input data to the second communication device; The first communication device sends, to the second communication device, first output data corresponding to first input data not associated with second output data; The first communication device sends the first input data to the second communication device; The first communication device sends multiple first information corresponding to a first input data to the second communication device, where different first information in the multiple first information is associated with different first AI models; The second output data includes at least one of the following: first expected information, where the first expected information is expected output information corresponding to the first input data, determined by the second communication device or protocol based on the first input data and a reference model; First label data, where the first label data is label data corresponding to the first input data. The method according to any one of claims 1 to 23, wherein the first AI model is associated with at least one of the following parameters: The first data set, the second data set, the reference model, the hardware capability, the quantization method, the number of bits reported for the first information, the identification information of the model, the type information of the model, the function information of the model, the level information of the model, and the complexity information of the model; Among them, the second data set is the training data set of the first AI model. A communication method, comprising: The second communication device receives the first information sent by the first communication device; The second communication device determines at least one of a test result and a monitoring result of the first artificial intelligence AI model based on the first information; Alternatively, the second communication device sends the first data set to the first communication device for the first communication device to determine the first information; The first information includes at least one of the following: First output data, where the first output data is output data obtained based on the first input data of the first data set and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model; a first test result, where the first test result is used to determine a test result of the first AI model; First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model. The method according to claim 25, wherein the first AI model includes at least one of a first model and a second model, the first model is a reference model, and the second model is a model determined based on the first model and different from the first model. The method of claim 26, wherein the reference model comprises at least one of the following: Reference model agreed upon in the agreement; a reference model determined in conjunction with target model parameters, the target model parameters including model parameters sent by the second communication device; A reference model determined based on the reference model agreed upon in the protocol and the parameters of the target model; A reference model trained based on a second data set; A reference model based on the protocol agreement and a reference model determined by the second data set; The reference model sent by the second communication device. The method according to any one of claims 25 to 27, wherein the first data set further includes second output data, and the second output data includes at least one of the following: First expected information; First label data; The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model; The first label data is label data corresponding to the first input data. The method according to claim 28, wherein the first test result includes at least one of the following: first difference information between the first output data and the second output data; statistical information of first difference information between the first output data and the second output data; first ratio information between the first output data and the second output data; statistical information of first ratio information between the first output data and the second output data; first difference information between the first difference information and a reference model error; Statistical information of first difference information between the first difference information and the reference model error; an indication that the first AI model meets the test requirements; an indication that the first AI model does not meet the test requirements; Relevant information about the model that meets the test requirements in the first AI model; The number of first AI models that meet testing requirements. The method according to any one of claims 28-29, wherein the method further comprises at least one of the following: The second communication device sends first auxiliary information to the first communication device; The second communication device sends second information to the first communication device; The second communication device sends third information to the first communication device; The first auxiliary information includes at least one of the following: first threshold; second threshold; First statistics; The first threshold includes at least one of the following: difference threshold information between the first output data and the second output data; threshold information of a ratio between the first output data and the second output data; The second threshold includes at least one of the following: threshold information of statistical information of differences between the first output data and the second output data; threshold information of statistical information of the ratio of the first output data to the second output data; The first statistical information is statistical information of a second data set, and the second data set is training data of the first AI model; The second information includes at least one of the following: decompressed codebook information acquired by the second communication device; a throughput determined by the second communication device; location information of the terminal determined by the second communication device; Beam information of beam set A; The third information includes at least one of the following: Second dataset; Reference model; Target model parameters associated with the reference model. The method of claim 30, wherein the first data set is associated with the third information. The method according to any one of claims 25 to 31, wherein the first information further includes at least one of the following: identification information of the first data set; Identification information of a second data set, wherein the second data set is training data for the first AI model; identification information of the first AI model; Type information of the first AI model; Functional information of the first AI model. The method according to any one of claims 25 to 32, wherein the second communication device receives the first information sent by the first communication device, comprising: The second communication device receives the first output data sent by the first communication device; The second communication device receives first output data corresponding to the plurality of first input data sent by the first communication device; The second communication device receives first output data corresponding to first input data not associated with second output data sent by the first communication device; The second communication device receives the first input data sent by the first communication device; The second communication device receives multiple first information corresponding to a first input data sent by the first communication device, wherein different first information in the multiple first information is associated with different first AI models; The second output data includes at least one of the following: first expected information, where the first expected information is expected output information corresponding to the first input data, determined by the second communication device or protocol based on the first input data and a reference model; First label data, where the first label data is label data corresponding to the first input data. The method according to any one of claims 28 to 33, wherein the second communication device determines, based on the first information, at least one of a test result and a monitoring result of the first artificial intelligence (AI) model in the first communication device, including at least one of the following: If the second condition is met, determining that the first AI model meets the test requirements; If the second condition is not met, determining that the first AI model does not meet the test requirement; The second condition includes at least one of the following: The payload of the first output data is the same as the payload of the first expected information; The third difference information between the first output data and the first expected information is less than or equal to a third threshold; Statistical information of the third difference information between the first output data and the first expected information is less than or equal to a fourth threshold; The second difference information between the third difference information and the reference model error is less than or equal to a fifth threshold; Statistical information of the third difference information and the second difference information of the reference model error is less than or equal to a sixth threshold; The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model. The method according to claim 34, wherein, when the first input data includes channel state information or codebook information, the first expected information is channel state information or codebook information compressed according to a reference model; Or, in a case where the first input data includes original channel information, the first expected information includes original channel information compressed according to a reference model; Or, in a case where the first input data includes the channel state information or codebook information, the first expected information is decompressed channel state information or codebook information obtained according to a reference model; Alternatively, in a case where the first input data includes original channel information, the first expected information includes original channel information decompressed according to a reference model. The method according to claim 34 or 35, wherein the second communication device is a fourth communication device, and the first input data includes at least one of channel state information, codebook information, and original channel information, and the first output data includes at least one of compressed codebook information and compressed original channel information, the second communication device determining the test result and monitoring result of the first artificial intelligence (AI) model in the first communication device based on the first information, comprising: The second communication device decompresses the first output data to obtain third output data; If the third condition is met, determining that the first AI model meets the test requirement, or, if the third condition is not met, determining that the first AI model does not meet the test requirement; The third condition includes at least one of the following: The precision loss of the third output data relative to the first input data is less than or equal to a seventh threshold; The accuracy loss statistics of the third output data relative to the first input data is less than or equal to an eighth threshold; The precision of the third output data is greater than or equal to the precision of the first input data; The precision loss of the third output data relative to the decompressed first label data is less than or equal to a ninth threshold; The accuracy loss statistics of the third output data relative to the decompressed first label data is less than or equal to a tenth threshold; The precision of the third output data is greater than or equal to the precision of the decompressed first tag data; The fourth difference information between the third output data and the decompressed first label data is less than or equal to an eleventh threshold; Statistical information of fourth difference information between the third output data and the decompressed first label data is less than or equal to a twelfth threshold. The method according to any one of claims 25 to 36, further comprising: The second communication device sends feedback information to the first communication device; The feedback information includes at least one of the following: an indication that the first AI model meets the test requirements; an indication that the first AI model does not meet the test requirements; Relevant information about the model that meets the test requirements in the first AI model; Training to obtain relevant information of a second data set of the first AI model; The number of first AI models that meet testing requirements. The method according to any one of claims 25 to 37, wherein the first AI model is associated with at least one of the following parameters: The first data set, the second data set, the reference model, the hardware capability, the quantization method, the number of bits reported for the first information, the identification information of the model, the type information of the model, the function information of the model, the level information of the model, and the complexity information of the model; Among them, the second data set is the training data set of the first AI model. The method according to any one of claims 25 to 38, wherein the second communication device comprises at least one of a fourth communication device and a monitoring device. A communication method, comprising: The monitoring device receives fourth output data sent by the fourth communication device, where the fourth output data is decompressed data of the first output data; The monitoring device determines at least one of a test result and a monitoring result of the first artificial intelligence (AI) model based on the fourth output data; Alternatively, the monitoring device receives at least one of a test result and a monitoring result of the first artificial intelligence AI model sent by the first communication device or the fourth communication device; The monitoring device determines at least one of the following based on at least one of the test result and the monitoring result of the first artificial intelligence (AI) model: Target-first AI model; Status information of the first AI model; wherein the target first AI model is at least one of the first AI models; Among them, the fourth communication device is used to receive the first output data sent by the first communication device and decompress the first output data, where the first output data is output data obtained based on the first input number of the first data set and the first AI model. The method according to claim 40, wherein the first input data includes at least one of channel state information, codebook information, and original channel information, and the first output data includes at least one of compressed codebook information and compressed original channel information, and the monitoring device determines, based on the fourth output data, a test result and a monitoring result of the first artificial intelligence (AI) model in the first communication device, including at least one of the following: If the fourth condition is met, determining that the first AI model meets the test requirements; If the fourth condition is not met, determining that the first AI model does not meet the test requirements; The fourth condition includes at least one of the following: The precision loss of the fourth output data relative to the first input data is less than or equal to a thirteenth threshold; The accuracy loss statistics of the fourth output data relative to the first input data is less than or equal to a fourteenth threshold; The precision of the fourth output data is greater than or equal to the precision of the first input data; The precision loss of the fourth output data relative to the decompressed first label data is less than or equal to a fifteenth threshold; The accuracy loss statistics of the fourth output data relative to the decompressed first label data is less than or equal to a sixteenth threshold; The precision of the fourth output data is greater than or equal to the precision of the decompressed first tag data; The third difference information between the fourth output data and the decompressed first label data is less than or equal to a seventeenth threshold; Statistical information of the third difference information between the fourth output data and the decompressed first label data is less than or equal to an eighteenth threshold. A communication device, comprising: A first acquisition module is configured to acquire a first data set, where the first data set includes first input data; A first determination module, configured to determine first information based on the first data set and a first artificial intelligence (AI) model; The first information includes at least one of the following: first output data, where the first output data is output data obtained based on the first input data and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model; a first test result, where the first test result is used to determine a test result of the first AI model; First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model. The apparatus of claim 42, wherein the first data set further comprises second output data, the second output data comprising at least one of the following: First expected information; First label data; The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model; The first label data is label data corresponding to the first input data. The apparatus according to claim 43, wherein the first determining module comprises at least one of the following: A first determination submodule, configured to determine that the first AI model meets the test requirement when a first condition is met; A second determining submodule, configured to determine that the first AI model does not meet the test requirement when the first condition is not met; The first condition includes at least one of the following: The payload of the first output data is the same as the payload of the second output data; First difference information between the first output data and the second output data is less than or equal to a first threshold; Statistical information of the first difference information between the first output data and the second output data is less than or equal to a second threshold. The apparatus according to claim 43 or 44, further comprising: a first receiving module, configured to receive first auxiliary information sent by a second communication device or a third communication device, where the second communication device includes at least one of a fourth communication device and a monitoring device; The first auxiliary information includes at least one of the following: first threshold; second threshold; First statistics; The first threshold value includes at least one of the following: difference threshold information between the first output data and the second output data; threshold information of a ratio between the first output data and the second output data; The second threshold value includes at least one of the following: threshold information of statistical information of differences between the first output data and the second output data; threshold information of statistical information of the ratio of the first output data to the second output data; The first statistical information is statistical information of a second data set, and the second data set is training data of the first AI model; The first determining module includes: The third determination submodule is used to determine the first information based on the first data set, the first AI model and the first auxiliary information. The apparatus according to any one of claims 42 to 45, wherein the first information includes first monitoring information, and the apparatus further comprises: A second receiving module, configured to receive second information sent by a second communication device; The second information includes at least one of the following: decompressed codebook information acquired by the second communication device; a throughput determined by the second communication device; location information of the terminal determined by the second communication device; Beam information of beam set A; The first determining module includes: The fourth determining submodule is configured to determine the first monitoring information according to the first data set, the first output data and the second information. The apparatus according to any one of claims 42 to 46, further comprising: a third receiving module, configured to receive third information sent by a second communication device or a third communication device, where the second communication device includes at least one of a fourth communication device and a monitoring device; The third information includes at least one of the following: Second dataset; Reference model; Target model parameters associated with the reference model. The apparatus according to any one of claims 42 to 47, wherein the first output data includes at least one of decompressed codebook information and compressed original channel information, and when the first information includes a first test result, the first determining module includes at least one of the following: a fifth determining submodule, configured to determine the first test result according to at least one of second difference information between the first output data and the first input data and a statistical value of the second difference information; The sixth determination submodule is configured to determine the first test result based on at least one of second ratio information of the first output data to the first input data and a statistical value of the second ratio information. The apparatus according to any one of claims 42 to 48, further comprising: an execution module, configured to, when at least one of the test result and the monitoring result of the first AI model fails, perform at least one of the following operations: Switch to the second AI model; Switch to non-AI function; Fall back to the reference model; Fall back to non-AI functionality; Update reference models; Update reference model parameters; receiving a reference pattern repeatedly sent by a second communication device or a third communication device; receiving a reference model parameter repeatedly sent by the second communication device or the third communication device; receiving an updated reference model sent by the second communication device or the third communication device; receiving updated reference model parameters sent by the second communication device or the third communication device. The apparatus according to any one of claims 42 to 49, further comprising: The first sending module is configured to send the first information to a second communication device, wherein the second communication device includes at least one of a fourth communication device and a monitoring device. The apparatus according to claim 50, wherein the first sending module comprises: A first sending submodule, configured to send the first output data to a second communication device; A second sending submodule, configured to send first output data corresponding to the plurality of first input data to a second communication device; a third sending submodule, configured to send, to the second communication device, first output data corresponding to the first input data not associated with the second output data; a fourth sending submodule, configured to send the first input data to a second communication device; A fifth sending submodule, configured to send a plurality of first information corresponding to a first input data to the second communication device, wherein different first information in the plurality of first information are associated with different first AI models; The second output data includes at least one of the following: first expected information, where the first expected information is expected output information corresponding to the first input data, determined by the second communication device or protocol based on the first input data and a reference model; First label data, where the first label data is label data corresponding to the first input data. A communication device, comprising: A fourth receiving module, configured to receive first information sent by the first communication device; A second determination module is configured to determine at least one of a test result and a monitoring result of the first artificial intelligence AI model based on the first information; Alternatively, the device comprises: a second sending module, configured to send a first data set to a first communication device, so that the first communication device determines first information; The first information includes at least one of the following: First output data, where the first output data is output data obtained based on the first input data of the first data set and the first AI model, and the first output data is used to determine or assist in determining at least one of a test result and a monitoring result of the first AI model; a first test result, where the first test result is used to determine a test result of the first AI model; First monitoring information, where the first monitoring information is used to determine a monitoring result of the first AI model. The apparatus of claim 52, wherein the first data set further comprises second output data, the second output data comprising at least one of the following: First expected information; First label data; The first expected information is expected output information corresponding to the first input data determined by the second communication device or protocol based on the first input data and a reference model; The first label data is label data corresponding to the first input data; The device further comprises: A third sending module, configured to send first auxiliary information to the first communication device; a fourth sending module, configured to send second information to the first communication device; a fifth sending module, configured to send third information to the first communication device; The first auxiliary information includes at least one of the following: first threshold; second threshold; First statistics; The first threshold includes at least one of the following: difference threshold information between the first output data and the second output data; threshold information of a ratio between the first output data and the second output data; The second threshold includes at least one of the following: threshold information of statistical information of differences between the first output data and the second output data; threshold information of statistical information of the ratio of the first output data to the second output data; The first statistical information is statistical information of a second data set, and the second data set is training data of the first AI model; The second information includes at least one of the following: decompressed codebook information acquired by the second communication device; a throughput determined by the second communication device; location information of the terminal determined by the second communication device; Beam information of beam set A; The third information includes at least one of the following: Second dataset; Reference model; Target model parameters associated with the reference model. The apparatus according to claim 52 or 53, wherein the fourth receiving module comprises: a first receiving submodule, configured to receive the first output data sent by the first communication device; a second receiving submodule, configured to receive first output data corresponding to a plurality of first input data sent by the first communication device; a third receiving submodule, configured to receive first output data corresponding to first input data not associated with second output data and sent by the first communication device; a fourth receiving submodule, configured to receive the first input data sent by the first communication device; A fifth receiving submodule, configured to receive multiple first information corresponding to a first input data sent by the first communication device, wherein different first information in the multiple first information are associated with different first AI models; The second output data includes at least one of the following: first expected information, where the first expected information is expected output information corresponding to the first input data, determined by a second communication device or protocol based on the first input data and a reference model; First label data, where the first label data is label data corresponding to the first input data. The apparatus according to any one of claims 52 to 54, wherein the second determining module comprises at least one of the following: a seventh determination submodule, configured to determine that the first AI model meets the test requirement when the second condition is met; an eighth determining submodule, configured to determine that the first AI model does not meet the test requirement if the second condition is not met; The second condition includes at least one of the following: The payload of the first output data is the same as the payload of the first expected information; The third difference information between the first output data and the first expected information is less than or equal to a third threshold; Statistical information of the third difference information between the first output data and the first expected information is less than or equal to a fourth threshold; The second difference information between the third difference information and the reference model error is less than or equal to a fifth threshold; Statistical information of the third difference information and the second difference information of the reference model error is less than or equal to a sixth threshold; The first expected information is expected output information corresponding to the first input data, determined by the second communication device or protocol based on the first input data and a reference model. The apparatus according to any one of claims 52 to 55, wherein the communication apparatus is a fourth communication device, and when the first input data includes at least one of channel state information, codebook information, and original channel information, and the first output data includes at least one of compressed codebook information and compressed original channel information, the second determining module includes: a first decompression module, configured to decompress the first output data to obtain third output data; a ninth determining submodule, configured to determine that the first AI model meets the test requirement if a third condition is met, or to determine that the first AI model does not meet the test requirement if the third condition is not met; The third condition includes at least one of the following: The precision loss of the third output data relative to the first input data is less than or equal to a seventh threshold; The accuracy loss statistics of the third output data relative to the first input data is less than or equal to an eighth threshold; The precision of the third output data is greater than or equal to the precision of the first input data; The precision loss of the third output data relative to the decompressed first label data is less than or equal to a ninth threshold; The accuracy loss statistics of the third output data relative to the decompressed first label data is less than or equal to a tenth threshold; The precision of the third output data is greater than or equal to the precision of the decompressed first tag data; The fourth difference information between the third output data and the decompressed first label data is less than or equal to an eleventh threshold; Statistical information of fourth difference information between the third output data and the decompressed first label data is less than or equal to a twelfth threshold. The apparatus according to any one of claims 52 to 56, further comprising: a sixth sending module, configured to send feedback information to the first communication device; The feedback information includes at least one of the following: an indication that the first AI model meets the test requirements; an indication that the first AI model does not meet the test requirements; Relevant information about the model that meets the test requirements in the first AI model; Training to obtain relevant information of a second data set of the first AI model; The number of first AI models that meet testing requirements. A communication device, comprising: a fifth receiving module, configured to receive fourth output data sent by a fourth communication device, wherein the fourth output data is decompressed data of the first output data; a third determining module, configured to determine at least one of a test result and a monitoring result of the first artificial intelligence (AI) model based on the fourth output data; Alternatively, a sixth receiving module is configured to receive at least one of a test result and a monitoring result of the first artificial intelligence AI model sent by the first communication device or the fourth communication device; A fourth determination module is configured to determine at least one of the following based on at least one of the test result and the monitoring result of the first artificial intelligence (AI) model: Target-first AI model; Status information of the first AI model; wherein the target first AI model is at least one of the first AI models; Among them, the fourth communication device is used to receive the first output data sent by the first communication device and decompress the first output data, where the first output data is output data obtained based on the first input number of the first data set and the first AI model. The apparatus according to claim 58, wherein the first input data includes at least one of channel state information, codebook information, and original channel information, and the first output data includes at least one of compressed codebook information and compressed original channel information, the third determining module is configured to include at least one of the following: A fifth determination module, configured to determine whether the first AI model meets the test requirement when the fourth condition is met; a sixth determining module, configured to determine that the first AI model does not meet the test requirement if the fourth condition is not met; The fourth condition includes at least one of the following: The precision loss of the fourth output data relative to the first input data is less than or equal to a thirteenth threshold; The accuracy loss statistics of the fourth output data relative to the first input data is less than or equal to a fourteenth threshold; The precision of the fourth output data is greater than or equal to the precision of the first input data; The precision loss of the fourth output data relative to the decompressed first label data is less than or equal to a fifteenth threshold; The accuracy loss statistics of the fourth output data relative to the decompressed first label data is less than or equal to a sixteenth threshold; The precision of the fourth output data is greater than or equal to the precision of the decompressed first tag data; The third difference information between the fourth output data and the decompressed first label data is less than or equal to a seventeenth threshold; Statistical information of the third difference information between the fourth output data and the decompressed first label data is less than or equal to an eighteenth threshold. A terminal comprises a processor and a memory, wherein the memory stores a program or instruction that can be run on the processor, and when the program or instruction is executed by the processor, the steps of the communication method according to any one of claims 1 to 24 are implemented, or when the program or instruction is executed by the processor, the steps of the communication method according to any one of claims 25 to 39 are implemented. A network-side device comprises a processor and a memory, wherein the memory stores a program or instruction that can be run on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the communication method according to any one of claims 1 to 24, or, when executed by the processor, implements the steps of the communication method according to any one of claims 25 to 39, or, when executed by the processor, implements the steps of the communication method according to any one of claims 40 to 41. A readable storage medium storing a program or instruction, wherein the program or instruction, when executed by a processor, implements the communication method according to any one of claims 1 to 24, or implements the steps of the communication method according to any one of claims 25 to 39, or implements the steps of the communication method according to any one of claims 40 to 41. A computer program product comprising computer instructions, wherein when the computer instructions are executed by a processor, the steps of the method according to any one of claims 1 to 24 are implemented, or when the computer instructions are executed by a processor, the steps of the method according to any one of claims 25 to 39 are implemented, or when the computer instructions are executed by a processor, the steps of the method according to any one of claims 40 to 41 are implemented.