Method and device for artificial intelligence-based network automation in wireless communication system

Abstract

The present specification proposes a method and device for artificial intelligence-based network automation in a wireless communication system. According to one embodiment of the present specification, network automation-related operations may be performed without user intervention by using a global large language model and a local large language model.

Description

Method and apparatus for artificial intelligence-based network automation in wireless communication systems

[0001] This specification relates to wireless communication, and more specifically, to a method and apparatus for artificial intelligence-based network automation in a wireless communication system.

[0002] With the advancement of artificial intelligence technology, the automation of wireless network operations based on artificial intelligence is being considered. In 5G NR, the Network Data Analytics Function (NWDAF) was introduced as one of the 5G Core (5GC) functions. To optimize and enhance the performance of each network function, NWDAF can collect data from various sources within the 5G network and provide analysis of the collected data. This includes other network functions such as Access and Mobility Management Function (AMF), Session Management Function (SMF), Policy Control Function (PCF), and Unified Data Management (UDM), as well as network exposure functions, application functions, and other service functions that provide information on user mobility, session management, quality of service, and network performance. NWDAF may include the Analytics Logical Function (AnLF), which provides inferred information based on AI / ML models, and the Model Training Logical Function (MTLF), which learns and trains AI / ML models.

[0003] For complete network automation, each network function needs to perform automated operations based on analysis reports received from the NWDAF. The NWDAF can provide AI / ML-based analysis data to network functions using datasets such as predefined analysis data formats and analysis IDs. However, complete network automation requires the automation of each individual network function, and considering the complexity of network operations, it may not be easy to achieve the automation of each network function solely by utilizing the aforementioned datasets such as predefined analysis data formats and analysis IDs.

[0004] This specification proposes a method and apparatus for artificial intelligence-based network automation in a wireless communication system.

[0005] According to one embodiment, a method is proposed to be performed by a communication device in a wireless communication system. The method is characterized in that it performs learning on a first AI / ML (Artificial Intelligence / Machine Learning) model and generates a second AI / ML model for a specific network function based on the learned first AI / ML model, wherein the communication device generates the second AI / ML model by performing learning on the specific network function, and the communication device transmits information regarding the second AI / ML model to a specific communication device, wherein the information includes at least one of information for the communication device to acquire the second AI / ML model and information for identifying the second AI / ML model among one or more AI / ML models stored in the specific communication device.

[0006] Here, the learning may include at least some of the steps of acquiring data for a plurality of network functions, performing preprocessing on the acquired data, and performing learning on the first AI / ML model based on the preprocessed data.

[0007] Here, a second AI / ML model for the specific network function can be generated based on data related to the specific network function and Key Performance Indicators (KPIs) related to the specific network function.

[0008] Here, data related to the specific network function may be predefined or determined by the communication device.

[0009] Here, the communication device receives information related to the specific network function from the specific communication device, and can perform an update to the first AI / ML model based on the information related to the specific network function.

[0010] Here, the communication device can exchange information about the specific communication device and the specific network function.

[0011] A communication device proposed according to another embodiment comprises one or more memories for storing instructions; one or more transceivers; and one or more processors connecting the one or more memories and the one or more transceivers, wherein the one or more processors execute the instructions to perform learning on a first AI / ML (Artificial Intelligence / Machine Learning) model and generate a second AI / ML model for a specific network function based on the learned first AI / ML model, wherein the communication device generates the second AI / ML model by performing learning on the specific network function, and the communication device transmits information regarding the second AI / ML model to a specific communication device, wherein the information includes at least one of information for the communication device to acquire the second AI / ML model and information for identifying the second AI / ML model among one or more AI / ML models stored in the specific communication device.

[0012] Here, the learning may include at least some of the steps of acquiring data for a plurality of network functions, performing preprocessing on the acquired data, and performing learning on the first AI / ML model based on the preprocessed data.

[0013] Here, a second AI / ML model for the specific network function can be generated based on data related to the specific network function and Key Performance Indicators (KPIs) related to the specific network function.

[0014] Here, data related to the specific network function may be predefined or determined by the communication device.

[0015] Here, the communication device receives information related to the specific network function from the specific communication device, and can perform an update to the first AI / ML model based on the information related to the specific network function.

[0016] Here, the communication device can exchange information about the specific communication device and the specific network function.

[0017] According to another embodiment, a method is proposed to be performed by a communication device in a wireless communication system. The method is characterized by receiving first information regarding a first AI / ML model, wherein the first information includes at least one of information for the communication device to acquire the first AI / ML model and information for identifying the first AI / ML model among one or more AI / ML models stored in the communication device, and performing training of the first AI / ML model for a specific network function, wherein the training is performed based on at least one of data related to the specific network function and KPIs related to the specific network function, and performing an operation related to the specific network function based on the first AI / ML model for the specific network function.

[0018] Here, the first information can be transmitted by a specific communication device.

[0019] Here, the communication device can exchange information related to the specific network function with the specific communication device.

[0020] Here, based on the fact that the first AI / ML model is a model trained for the specific network function, the communication device may omit the training of the first AI / ML model.

[0021] Here, the communication device can exchange information with another communication device having an AI / ML model for other network functions.

[0022] Here, the first AI / ML model may be an untrained basic model.

[0023] Here, the communication device receives second information from the specific communication device, and based on the second information, the communication device can adjust the specific network function.

[0024] Here, the adjustment of the specific network function may include retraining the first AI / ML model.

[0025] According to the present specification, AI / ML-based network operation for network automation is possible without user intervention, and furthermore, communication efficiency is increased.

[0026] The effects obtainable through the specific examples of this specification are not limited to those listed above. For example, there may be various technical effects that a person with ordinary skill in the related art can understand or derive from this specification. Accordingly, the specific effects of this specification are not limited to those explicitly described herein, but may include various effects that can be understood or derived from the technical features of this specification.

[0027] The following drawings are prepared to illustrate a specific example of the present specification. The names of specific devices or specific signals / messages / fields described in the drawings are presented as examples, and therefore the technical features of the present specification are not limited to the specific names used in the following drawings.

[0028] FIG. 1 shows an AI device (100) according to one embodiment of the present invention.

[0029] FIG. 2 shows an AI server (200) according to one embodiment of the present invention.

[0030] FIG. 3 shows an AI system (1) according to one embodiment of the present invention.

[0031] FIG. 4 illustrates an example of the data collection and analysis reporting operation of NWDAF according to one embodiment of the present specification.

[0032] Figure 5 illustrates an example of an AI / ML framework.

[0033] FIG. 6 illustrates an example of the configuration of a federated learning-based local AI / ML model according to one embodiment of the present specification.

[0034] FIG. 7 is a flowchart of an example of a method performed by a communication device according to one embodiment of the present specification.

[0035] FIG. 8 is a flowchart for another example of a method performed by a communication device according to one embodiment of the present specification.

[0036] Figure 9 is a diagram illustrating an example of a network failure monitoring method in a legacy mobile network operating system.

[0037] Figure 10 is a diagram illustrating an example of a network failure monitoring method in an NWDAF-based network operation system.

[0038] FIG. 11 is a flowchart for an example of a network failure recovery method performed by a communication device.

[0039] Figure 12 is a flowchart for another example of a network failure recovery method performed by a communication device.

[0040] The present invention is susceptible to various modifications and may have various embodiments; specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the invention to specific embodiments, and it should be understood that the invention includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the invention. Similar reference numerals have been used for similar components in the description of each drawing.

[0041] In this specification, terms such as "first," "second," "A," "B," etc., may be used to describe various components, but said components should not be limited by said terms. These terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be named the second component, and similarly, the second component may be named the first component. Additionally, the term "and / or" includes a combination of a plurality of related described items or any one of a plurality of related described items.

[0042] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

[0043] The terms used herein are merely for describing specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless the context clearly indicates otherwise. In this specification, terms such as “comprising” or “having” are intended to indicate the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0044] Unless otherwise defined, the terms used in this specification, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which the present invention pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this specification.

[0045] In this specification, singular expressions include plural expressions unless the context clearly specifies them as singular. Additionally, plural expressions include singular expressions unless the context clearly specifies them as plural. Throughout the specification, when a part is described as including a certain component, this means that, unless specifically stated otherwise, it does not exclude other components but may include additional components.

[0046] Additionally, the terms 'module' or 'part' as used in the specification refer to software or hardware components, and the 'module' or 'part' performs certain roles. However, the meaning of 'module' or 'part' is not limited to software or hardware. The 'module' or 'part' may be configured to reside in an addressable storage medium or configured to run on one or more processors. Thus, as an example, the 'module' or 'part' may include components such as software components, object-oriented software components, class components, and task components, and at least one of processes, functions, attributes, procedures, subroutines, segments of program code, drivers, firmware, microcode, circuits, data, databases, data structures, tables, arrays, or variables. The components and the functions provided within the 'module' or 'part' may be combined into a smaller number of components and 'modules' or 'parts', or further separated into additional components and 'modules' or 'parts'.

[0047] According to one embodiment of the present disclosure, a ‘module’ or ‘part’ may be implemented as a processor and memory. The term ‘processor’ should be broadly interpreted to include a general-purpose processor, a central processing unit (CPU), a microprocessor, a digital signal processor (DSP), a controller, a microcontroller, a state machine, etc. In some contexts, the term ‘processor’ may refer to an application-specific integrated circuit (ASIC), a programmable logic device (PLD), a field programmable gate array (FPGA), etc. The term ‘processor’ may also refer to a combination of processing devices, such as, for example, a combination of a DSP and a microprocessor, a combination of multiple microprocessors, a combination of one or more microprocessors combined with a DSP core, or any other combination of such configurations. Additionally, the term ‘memory’ should be broadly interpreted to include any electronic component capable of storing electronic information. 'Memory' may refer to various types of processor-readable media, such as Random Access Memory (RAM), Read-Only Memory (ROM), Non-Volatile Random Access Memory (NVRAM), Programmable Read-Only Memory (PROM), Erasable-Programmable Read-Only Memory (EPROM), Electrically Erasable PROM (EEPROM), Flash Memory, Magnetic or Optical Data Storage Devices, Registers, etc. If a processor can read information from memory and / or write information to memory, the memory is said to be in an electronic communication state with the processor. Memory integrated into a processor is in an electronic communication state with the processor.

[0048] In the present disclosure, the 'system' may include at least one of a server device and a cloud device, but is not limited thereto. For example, the system may be composed of one or more server devices. As another example, the system may be composed of one or more cloud devices. As yet another example, the system may be configured and operated with both a server device and a cloud device.

[0049] In the present disclosure, 'display' may refer to any display device associated with a computing device, for example, any display device capable of displaying any information / data controlled by or provided by the computing device.

[0050] Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the attached drawings.

[0051] Artificial Intelligence (AI) refers to the field of researching artificial intelligence or methodologies capable of creating it, while Machine Learning (ML) refers to the field of researching methodologies to define and solve various problems addressed within the field of artificial intelligence. Machine learning is also defined as an algorithm that improves performance on a task through continuous experience.

[0052] An Artificial Neural Network (ANN) is a model used in machine learning that can refer to any model capable of problem-solving, composed of artificial neurons (nodes) that form a network through the connection of synapses. An artificial neural network can be defined by connection patterns between neurons in different layers, a learning process that updates model parameters, and an activation function that generates output values.

[0053] An artificial neural network may include an input layer, an output layer, and optionally one or more hidden layers. Each layer may include one or more neurons, and the artificial neural network may include synapses connecting the neurons. In an artificial neural network, each neuron may output a function value of an activation function for input signals, weights, and biases input through the synapses.

[0054] Model parameters refer to parameters determined through learning, including synaptic connection weights and neuron biases. Hyperparameters, on the other hand, refer to parameters that must be set prior to training in a machine learning algorithm, including the learning rate, number of iterations, mini-batch size, and initialization function.

[0055] The objective of training an artificial neural network can be viewed as determining model parameters that minimize the loss function. The loss function can be used as an indicator to determine optimal model parameters during the training process of an artificial neural network.

[0056] Machine learning can be classified into supervised learning, unsupervised learning, and reinforcement learning depending on the learning method.

[0057] Supervised learning refers to a method of training an artificial neural network with labels provided for the training data; a label can refer to the correct answer (or result) that the neural network must infer when the training data is input. Unsupervised learning refers to a method of training an artificial neural network without labels provided for the training data. Reinforcement learning refers to a learning method in which an agent defined within an environment is trained to select an action or sequence of actions that maximizes the cumulative reward in each state.

[0058] Machine learning implemented using a Deep Neural Network (DNN) that includes multiple hidden layers among artificial neural networks is also called Deep Learning, and Deep Learning is a part of Machine Learning. Hereinafter, Machine Learning is used in a sense that includes Deep Learning.

[0059] FIG. 1 shows an AI device (100) according to one embodiment of the present invention.

[0060] The AI ​​device (100) can be implemented as a stationary device or a mobile device, such as a TV, projector, mobile phone, smartphone, desktop computer, laptop, digital broadcasting terminal, PDA (personal digital assistants), PMP (portable multimedia player), navigation, tablet PC, wearable device, set-top box (STB), DMB receiver, radio, washing machine, refrigerator, desktop computer, digital signage, robot, vehicle, etc.

[0061] Referring to FIG. 1, the terminal (100) may include a communication unit (110), an input unit (120), a learning processor (130), a sensing unit (140), an output unit (150), a memory (170), and a processor (180), etc.

[0062] The communication unit (110) can transmit and receive data with external devices, such as other AI devices (100a to 100e) or an AI server (200), using wired or wireless communication technology. For example, the communication unit (110) can transmit and receive sensor information, user input, learning models, control signals, etc., with external devices.

[0063] At this time, the communication technologies used by the communication unit (110) include GSM (Global System for Mobile communication), CDMA (Code Division Multi Access), LTE (Long Term Evolution), 5G, WLAN (Wireless LAN), Wi-Fi (Wireless-Fidelity), Bluetooth, RFID (Radio Frequency Identification), Infrared Data Association (IrDA), ZigBee, NFC (Near Field Communication), etc.

[0064] The input unit (120) can acquire various types of data.

[0065] At this time, the input unit (120) may include a camera for inputting a video signal, a microphone for receiving an audio signal, a user input unit for receiving information from a user, etc. Here, the camera or microphone may be treated as a sensor, and the signal obtained from the camera or microphone may be referred to as sensing data or sensor information.

[0066] The input unit (120) can obtain training data for model training and input data to be used when obtaining an output using a training model. The input unit (120) may also obtain unprocessed input data, in which case the processor (180) or the learning processor (130) can extract input feature points as a preprocessing step for the input data.

[0067] The learning processor (130) can train a model composed of an artificial neural network using training data. Here, the trained artificial neural network may be referred to as a learning model. The learning model can be used to infer a result value for new input data other than the training data, and the inferred value can be used as a basis for judgment to perform an action.

[0068] At this time, the learning processor (130) can perform AI processing together with the learning processor (240) of the AI ​​server (200).

[0069] At this time, the learning processor (130) may include memory integrated into or implemented in the AI ​​device (100). Alternatively, the learning processor (130) may be implemented using memory (170), external memory directly coupled to the AI ​​device (100), or memory maintained in an external device.

[0070] The sensing unit (140) can obtain at least one of internal information of the AI ​​device (100), surrounding environment information of the AI ​​device (100), and user information using various sensors.

[0071] At this time, the sensors included in the sensing unit (140) include a proximity sensor, an illuminance sensor, an accelerometer, a magnetic sensor, a gyroscope, an inertial sensor, an RGB sensor, an IR sensor, a fingerprint recognition sensor, an ultrasonic sensor, a light sensor, a microphone, a lidar, a radar, etc.

[0072] The output unit (150) can generate output related to sight, hearing, or touch.

[0073] At this time, the output unit (150) may include a display unit that outputs visual information, a speaker that outputs auditory information, a haptic module that outputs tactile information, etc.

[0074] The memory (170) can store data that supports various functions of the AI ​​device (100). For example, the memory (170) can store input data, training data, training models, training history, etc. obtained from the input unit (120).

[0075] The processor (180) can determine at least one executable action of the AI ​​device (100) based on information determined or generated using a data analysis algorithm or a machine learning algorithm. The processor (180) can perform the determined action by controlling the components of the AI ​​device (100).

[0076] To this end, the processor (180) can request, search, receive, or utilize data from the learning processor (130) or memory (170), and can control the components of the AI ​​device (100) to execute a predicted operation or a preferred operation among the at least one executable operation.

[0077] At this time, if the processor (180) requires the connection of an external device to perform a determined operation, it can generate a control signal to control the external device and transmit the generated control signal to the external device.

[0078] The processor (180) can obtain intent information regarding user input and determine the user's requirements based on the obtained intent information.

[0079] At this time, the processor (180) can obtain intent information corresponding to the user input by using at least one of a Speech To Text (STT) engine for converting voice input into a string or a Natural Language Processing (NLP) engine for obtaining intent information of natural language.

[0080] At this time, at least one of the STT engine or NLP engine may be composed of an artificial neural network in which at least a portion is learned according to a machine learning algorithm. Also, at least one of the STT engine or NLP engine may be learned by a learning processor (130), learned by a learning processor (240) of an AI server (200), or learned through distributed processing thereof.

[0081] The processor (180) may collect history information, including the operation details of the AI ​​device (100) or user feedback regarding the operation, and store it in memory (170) or a learning processor (130), or transmit it to an external device such as an AI server (200). The collected history information may be used to update a learning model.

[0082] The processor (180) can control at least some of the components of the AI ​​device (100) to run an application stored in memory (170). Furthermore, the processor (180) can operate two or more of the components included in the AI ​​device (100) in combination with each other to run the application.

[0083] FIG. 2 shows an AI server (200) according to one embodiment of the present invention.

[0084] Referring to FIG. 2, the AI ​​server (200) may refer to a device that trains an artificial neural network using a machine learning algorithm or uses a trained artificial neural network. Here, the AI ​​server (200) may be composed of multiple servers to perform distributed processing and may be defined as a 5G network. At this time, the AI ​​server (200) may be included as part of the configuration of the AI ​​device (100) and may perform at least part of the AI ​​processing together.

[0085] The AI ​​server (200) may include a communication unit (210), memory (230), a learning processor (240), and a processor (260), etc.

[0086] The communication unit (210) can transmit and receive data with external devices such as AI devices (100).

[0087] The memory (230) may include a model storage unit (231). The model storage unit (231) may store a model (or artificial neural network, 231a) that is being learned or has been learned through a learning processor (240).

[0088] The learning processor (240) can train the artificial neural network (231a) using training data. The training model may be used while mounted on the AI ​​server (200) of the artificial neural network, or it may be used while mounted on an external device such as an AI device (100).

[0089] The learning model may be implemented in hardware, software, or a combination of hardware and software. If part or all of the learning model is implemented in software, one or more instructions constituting the learning model may be stored in memory (230).

[0090] The processor (260) can use a learning model to infer a result value for new input data and generate a response or control command based on the inferred result value.

[0091] FIG. 3 shows an AI system (1) according to one embodiment of the present invention.

[0092] Referring to FIG. 3, the AI ​​system (1) is connected to a cloud network (10) at least one of an AI server (200), a robot (100a), an autonomous vehicle (100b), an XR device (100c), a smartphone (100d), or a home appliance (100e). Here, the robot (100a), the autonomous vehicle (100b), the XR device (100c), the smartphone (100d), or the home appliance (100e) to which AI technology is applied may be referred to as AI devices (100a to 100e).

[0093] A cloud network (10) may mean a network that constitutes part of a cloud computing infrastructure or exists within a cloud computing infrastructure. Here, the cloud network (10) may be configured using a 3G network, a 4G or LTE (Long Term Evolution) network or a 5G network, etc.

[0094] That is, each device (100a to 100e, 200) constituting the AI ​​system (1) can be connected to each other through a cloud network (10). In particular, each device (100a to 100e, 200) may communicate with each other through a base station, but may also communicate directly with each other without going through a base station.

[0095] The AI ​​server (200) may include a server that performs AI processing and a server that performs operations on big data.

[0096] The AI ​​server (200) is connected via a cloud network (10) to at least one of the AI ​​devices constituting the AI ​​system (1), such as a robot (100a), an autonomous vehicle (100b), an XR device (100c), a smartphone (100d), or a home appliance (100e), and can assist in at least some of the AI ​​processing of the connected AI devices (100a to 100e).

[0097] At this time, the AI ​​server (200) can train an artificial neural network according to a machine learning algorithm on behalf of the AI ​​devices (100a to 100e), and can directly store the training model or transmit it to the AI ​​devices (100a to 100e).

[0098] At this time, the AI ​​server (200) receives input data from the AI ​​devices (100a to 100e), infers a result value for the received input data using a learning model, and generates a response or control command based on the inferred result value and transmits it to the AI ​​devices (100a to 100e).

[0099] Alternatively, the AI ​​device (100a to 100e) may use a direct learning model to infer a result value for input data and generate a response or control command based on the inferred result value.

[0100] Hereinafter, the method / configuration proposed in this specification is described in detail.

[0101] In the present disclosure, a "machine learning model" may include any model used to infer an answer to a given input. According to one embodiment, a machine learning model may include an artificial neural network model comprising an input layer, a plurality of hidden layers, and an output layer. Here, each layer may include a plurality of nodes. In the present disclosure, a plurality of machine learning models are described as separate machine learning models, but are not limited thereto, and some or all of the plurality of machine learning models may be implemented as a single machine learning model. Additionally, a single machine learning model may include a plurality of machine learning models. In the present disclosure, the terms machine learning model and artificial neural network model may be used interchangeably to refer to the same or similar models. Additionally, in the present disclosure, a "language model" may refer to a machine learning model or an artificial neural network model configured to calculate probabilities for at least a part of a sequence of one or more words or a sentence, or to generate a sequence of words or a part of a sentence.

[0102] A large language model (LLM) is a language model composed of artificial neural networks with numerous parameters. LLMs can be trained on unlabeled text using self-supervised or semi-self-supervised learning.

[0103] LLM is one of the elements that enable artificial intelligence chatbot technology. The way LLM works can be categorized into tokenization, transformer models, prompts, etc.

[0104] Tokenization, as part of natural language processing, refers to the process of converting general human language into sequences that low-level machine systems (LLMS) can understand. This involves assigning numerical values ​​to sections and encoding them for rapid analysis. The purpose of tokenization may be to generate context vectors, such as learning guides or formulas, for artificial intelligence to predict sentence structures. The more language is studied and how sentences are constructed, the more accurate predictions regarding the next word in a specific type of sentence can become. This can lead to the development of models that replicate the various communication styles people use online.

[0105] A Transformer model is a type of neural network that examines sequential data to identify related patterns regarding which words are likely to follow each other, and it can be composed of layers that perform different analyses to determine which words are compatible. Such a model relies on algorithms to understand human-written words without learning language, and is trained to write standard text about coffee by, for example, providing "hipster coffee blog."

[0106] Prompts are information provided by developers to large-scale language models for analysis and tokenization; essentially, they serve as training data that aids LLM across various use cases. The more accurate the prompts received, the better the LLM can predict the next word and construct accurate sentences. Therefore, selecting appropriate prompts is crucial for the proper training of deep learning AI.

[0107] Network automation is explained below.

[0108] With the advancement of artificial intelligence technology, the automation of wireless network operations based on artificial intelligence is being considered. In 5G NR, the Network Data Analytics Function (NWDAF) was introduced as one of the 5G Core (5GC) functions. To optimize and enhance the performance of each Network Function (NF), NWDAF can collect data from various sources within the 5G network, such as other network functions like Access and Mobility Management Function (AMF), Session Management Function (SMF), Policy Control Function (PCF), and Unified Data Management (UDM), as well as network exposure functions, application functions, and other service functions that provide information on user mobility, session management, quality of service, and network performance. It can also provide analysis of the collected data. NWDAF may include the Analytics Logical Function (AnLF), which provides inferred information based on AI / ML models, and the Model Training Logical Function (MTLF), which learns and trains AI / ML models.

[0109] FIG. 4 illustrates an example of the data collection and analysis reporting operation of NWDAF according to one embodiment of the present specification.

[0110] Referring to FIG. 4, NWDAF may include functions such as DCCF (Data Collection Coordination Function), ADRF (Analytical Data Repository Function), MFAF (Messaging Framework Adaptor Function), ANLF, and MTLF. NWDAF can collect input data from 5GC, OAM (Operation, Administration and Maintenance), and service domains (or application functions (AF)), and transmit an analytical report based on the collected data to 5GC, OAM, and service domains. For example, a consumer network function can receive an analytical report from NWDAF without needing to request analytical information from all producers. As another example, NWDAF can obtain information regarding the resource status and processing capabilities of network functions from OAM and transmit an analytical report based on this. As yet another example, NWDAF can obtain information regarding service experience, traffic patterns, and application identifiers from service domains and transmit an analytical report based on this.

[0111] For complete network automation, each network function needs to perform automated operations based on analysis reports received from the NWDAF. The NWDAF can provide AI / ML-based analysis data to network functions using datasets such as predefined analysis data formats and analysis IDs. However, complete network automation requires the automation of each individual network function, and considering the complexity of network operations, it may not be easy to achieve the automation of each network function solely by utilizing the aforementioned datasets such as predefined analysis data formats and analysis IDs.

[0112] An AI / ML framework is described below. The AI / ML framework described below may be applied to the AI / ML models disclosed in this specification. FIG. 5 illustrates an example of an AI / ML framework.

[0113] Referring to FIG. 5, the AI / ML framework (500) may be composed of a data collection block (510), a model training block (520), a model management block (530), a model inference block (540), and a model storage block (550). FIG. 5 is merely an example of an AI / ML framework, and various entities / functions / blocks not disclosed in FIG. 5 may be added to the AI / ML framework, and at least some of the blocks disclosed in FIG. 5 may be omitted.

[0114] The data collection block (510) can be performed in the LCM for various purposes such as model training, model inference, model monitoring, model selection, and model updating. The data collection block (510) of FIG. 5 is a block that conceptually represents data sources and entities holding data for training, inference, and monitoring. Although the data collection block (510) of FIG. 5 is represented as a single block, data collection for training, inference, and monitoring may have various characteristics and requirements. Additionally, the timescale of training and monitoring (e.g., real-time or offline) may require individual consideration.

[0115] Regarding training, training data may be initially generated at the network and the UE (or terminal). The initial data may be collected (or transmitted) by one or more data collection entities. Data collection entities may be owned by various entities, such as internal or external UEs / chipset / network vendors, network operators, and positioning service providers.

[0116] With respect to inference, inference data for the UE-side model and / or the UE portion of both-sided models may be transmitted or provided directly from the UE. Inference data for the network-side model and / or the network portion of both-sided models may be transmitted or provided directly from the network, or may be transmitted from the UE.

[0117] Regarding monitoring, monitoring data for UE-side monitoring may be transmitted or provided directly from the UE. Monitoring data for network-side monitoring may be transmitted or provided directly from the network, or it may be transmitted from the UE.

[0118] Data collection for real-time operations such as real-time model monitoring, switching, and selection can cause significant signaling overhead. Conversely, infrequent data collection to reduce signaling overhead can result in latency for real-time model monitoring, switching, and selection.

[0119] The model training block (520) may include both initial training and model updates. Generally, model training can be divided into model training conducted alongside model development and subsequent training for the developed model. The model training block (520) of FIG. 5 is represented as a single block for simplification.

[0120] Depending on the location of the dataset and / or the region where the model (or untrained model) is located, training may be performed internally within the network or by external entities such as UEs, chipset / network vendors, network operators, and positioning service providers. Since AI / ML model development is generally an iterative process of data collection, model design, training, and performance validation, careful implementation considerations regarding power consumption, hardware domain, latency, and concurrency with other layer functions are required for AI / ML model development.

[0121] When large-scale field data is collected from a data collection entity, the vendor responsible for model development must have access to said data. Typically, model development is an offline engineering process performed by an engineering team that requires access to large datasets collected in the field. In other words, decisions regarding model structure, device-specific optimizations, and the number of models to develop (e.g., generalizable versus specific models) may depend on the large-scale field data. If the vendor owning the data collection entity is different from the vendor responsible for model development, the vendor responsible for model development must have access to the dataset. This can be achieved through explicit dataset sharing or by providing access to the collected dataset. Dataset sharing / access may be relevant to two-sided models where both the gNB vendor and the UE / chipset vendor must participate in the model development and training processes.

[0122] After the model is developed and trained, the model can be stored in a model repository or a model storage block (550) and delivered to a target device. The model can be compiled into an executable file for inference. Here, various methods may exist depending on the location where the model is trained, the model storage / delivery format, the location where the model is hosted before delivery, etc.

[0123] The model inference block (540) is a function that provides AI / ML model inference outputs, such as predictions or decisions. The model inference block (540) may also provide model performance feedback to the model training block (520). The model inference block (540) may be responsible for data preparation, such as data preprocessing, cleaning, formatting, and transformation, based on the inference data delivered by the data collection block (510).

[0124] Model management may include functionality / model monitoring, selection, activation, deactivation, switching, fallback, etc. FIG. 5 illustrates a single model management block (530), but not all aspects of model management may be implemented in a single location. Some aspects of model monitoring, activation / deactivation, selection, switching, and fallback may be performed on the network side, and other aspects may be performed on the UE side. With regard to model selection, activation, deactivation, switching, and fallback for UE-side models and both-side models, mechanisms related to decisions by the network initiated by the network, mechanisms related to decisions by the network initiated by the UE and requested by the network, mechanisms related to decisions by the UE that are event-triggered by the network and where the UE's decision is reported to the network, mechanisms related to decisions by the UE that are UE-autonomous and where the UE's decision is reported to the network, and mechanisms related to decisions by the UE that are UE-autonomous and where the UE's decision is not reported to the network may be considered.

[0125] Meanwhile, technology regarding AI agents has been advancing recently. Generally, an AI agent (or artificial intelligence agent, intelligent agent) refers to an intelligent software system that operates autonomously without user intervention, perceives and learns from its environment to achieve a given goal or solve a problem. In this specification, an AI agent may also be interpreted as an AI / ML model for the operation of the AI ​​agent.

[0126] As an example of network automation, the introduction of AI / ML models or AI agents (or AI / ML models to drive AI agents) specific to each network function can be considered. Specifically, a method can be considered in which an optimized AI agent, based on the role and purpose of each network function, generates necessary analytical information and controls other functions based on it. Here, training and inference of local AI / ML models corresponding to each network function can be performed based on the unique capabilities of each network function. Furthermore, by training local AI / ML models (or local fine-tuned AI / ML models) optimized for each network function based on a global AI / ML model (or global foundational AI / ML model) that trains the entire network, the inference results based on these models can be utilized for decision-making in each network function. Here, for example, a local AI / ML model may refer to an AI / ML model accessible within an internal network environment even without a wireless connection. Additionally, for example, at least one of the global AI / ML model and the local AI / ML model may be an LLM, an AI agent-based AI / ML model, or an AI / ML model for driving an AI agent.

[0127] Local AI / ML models that train and infer in each network function may have relatively limited resources. To minimize local resources, a method of creating and updating a global AI / ML model that performs federated learning on the local AI / ML models can be considered. Here, for example, network automation can be considered through a structure in which a global AI / ML model training the entire network is defined as a foundation model, optimized fine-tuning models are created based on the mission and data of each network function, and each network function (or the local AI / ML model corresponding to each network function) is controlled based on these models.

[0128] To generate global AI / ML models, scalable and privacy-assured real-time AI / ML models capable of understanding wireless communication systems or networks and automatically performing network optimization can be used. To this end, an AI / ML model capable of real-time inference can be generated by collecting data, preprocessing the collected data, and performing training.

[0129] Here, the global AI / ML model can be generated by collecting various data from control plane network functions and data regarding traffic flow, QoS information, network topology and resource allocation data, anomalies, and security threat logs from user plane network functions, and then performing preprocessing and learning on the data.

[0130] Furthermore, regarding local AI / ML models, it may be necessary to create and manage fine-tuned models based on relatively limited resources. To this end, methods for generating distilled models based on global AI / ML models or fine-tuning foundational models based on local data can be considered, and distributed cooperative learning models such as federated learning may be used for this purpose.

[0131] A method may be considered in which a local AI / ML model is generated for each network function based on the core functions and Key Performance Indicators (KPIs) of each network function, based on a global AI / ML model that understands the entire network, and an AI agent is implemented to perform real-time control of said model. The said AI agent can make decisions for the optimization of network functions based on the local AI / ML model and can have various control authority over the network functions.

[0132] FIG. 6 illustrates an example of the configuration of a federated learning-based local AI / ML model according to one embodiment of the present specification.

[0133] Referring to FIG. 6, an AI / ML model corresponding to each of the network functions can be generated by performing learning using the functions and KPIs of each network function, based on a global AI / ML model that understands the transactions and KPIs of the entire wireless communication network, such as 5G or next-generation wireless communication networks. Through this, local AI / ML models corresponding to each of the PCF (Policy Control Function), UPF (User Plane Function), SMF (Session Management Function), AMF (Access and Mobility Management Function), HSS (Home Subscriber Server), and various network functions not shown in FIG. 6 can be implemented. Here, the HSS in FIG. 6 can be separated into UDM (Unified Data Management) and AUSF (Authentication Server Function). In other words, learning is performed on the global AI / ML model (or global LLM) for each network function, and an AI agent based on the AI / ML model (or local LLM) generated through said learning can be implemented. The AI ​​agent can perform the role of the same network function as the subject of the learning. Specifically, the AI ​​agent can control various auxiliary functions and perform operations related to network functions based on analysis results, which are inference results based on local AI / ML models. In addition, the AI ​​agent can communicate with AI agents for other network functions to perform cooperative operations, such as exchanging necessary data or analysis information and exchanging control requests.Furthermore, the inference results of the AI ​​agent (e.g., accuracy of inference, etc.) and the results of operations related to network functions (e.g., performance of network functions, etc.) can be used for performance evaluation, and whether to update or retrain the local AI / ML model can be determined based on the performance evaluation results. Meanwhile, the global AI / ML model and the local AI / ML model can be received from or exported externally.

[0134] FIG. 7 is a flowchart illustrating an example of a method performed by a communication device according to one embodiment of the present specification. Herein, the communication device may include an AI / ML model for a wireless communication network. That is, the communication device may store and execute software or an application that implements an AI / ML model for a wireless communication network. Alternatively, the communication device may be a communication device that includes hardware that stores and executes software or an application that implements an AI / ML model for a wireless communication network.

[0135] Referring to FIG. 7, the communication device performs training on a first AI / ML model (S710). Here, the step S710 may include at least some of the steps of acquiring data for a plurality of network functions, performing preprocessing on the acquired data, and performing training on the first AI / ML model based on the preprocessed data. For example, the first AI / ML model may correspond to a global LLM or global AI / ML model disclosed herein.

[0136] The communication device generates a second AI / ML model for a specific network function based on the learned first AI / ML model (S720). Here, the second AI / ML model for the specific network function may be generated based on data related to the specific network function and / or KPIs related to the specific network function. Here, the data related to the specific network function may be predefined or determined by the communication device. Additionally, in step S720, the communication device may perform learning / training for the specific network function to generate the second AI / ML model. Also, for example, the second AI / ML model may correspond to a local LLM or local AI / ML model disclosed herein.

[0137] The communication device transmits information regarding the second AI / ML model to a specific communication device (S730). Here, the information may include at least one of information for the communication device to receive or acquire the second AI / ML model (e.g., an AI / ML model identifier, information for implementing the AI / ML model, information about the AI / ML model itself, etc.) and information for identifying the second AI / ML model among one or more AI / ML models stored in the specific communication device.

[0138] The specific communication device described above may perform operations related to the specific network function based on a second AI / ML model for the specific network function. Here, for example, if the specific network function is an AMF, the operations related to the specific network function may include performing authentication procedures with a terminal, mobility management, accessibility management, etc.

[0139] Although not illustrated in FIG. 7, the communication device may transmit the second AI / ML model as well as AI / ML models corresponding to each of the plurality of network functions to one or more specific communication devices. Subsequently, the communication device may perform learning, updating, etc., on the first AI / ML model by performing federated learning with the one or more specific communication devices. Specifically, the communication device may receive information related to the specific network function from the specific communication device and perform an update on the first AI / ML model based on the information related to the specific network function.

[0140] An example of FIG. 7 may be applied by combining at least some of the embodiments of FIG. 1 to 6 described above within a range where they are not arranged with each other. Accordingly, redundant descriptions are omitted. In addition, an example of FIG. 7 may be performed not only by operation between communication devices but also by operation of software programs, AI agents, applications, etc.

[0141] FIG. 8 is a flowchart for another example of a method performed by a communication device according to one embodiment of the present specification.

[0142] Referring to FIG. 8, the communication device receives first information regarding a first AI / ML model (S810). Here, the first information may include at least one of information for the communication device to receive or acquire the first AI / ML model from an external source (e.g., an AI / ML model identifier, information for implementing the AI / ML model, information about the AI / ML model itself, etc.) and information for identifying the first AI / ML model among one or more AI / ML models stored in the communication device. Additionally, the first information may be transmitted by a specific communication device. Here, the specific communication device may be the communication device of FIG. 7.

[0143] The communication device performs training of the first AI / ML model for a specific network function (S820). For example, the first AI / ML model for the specific network function may be generated / trained based on data related to the specific network function and / or KPIs related to the specific network function. As another example, the step S820 may include at least some of the steps of acquiring data for the specific network function, performing preprocessing on the acquired data, and performing training of the first AI / ML model based on the preprocessed data.

[0144] Here, the first AI / ML model may be an untrained basic model. Alternatively, it may be a trained model as in step S710 of FIG. 7 or a trained model as in step S720 of FIG. 7. Here, if the first AI / ML model is trained for a specific network function, step S820 may be omitted.

[0145] The communication device performs operations related to the specific network function based on a first AI / ML model for the specific network function (S830). Here, for example, if the specific network function is an AMF, the operations related to the specific network function may include performing an authentication procedure with a terminal, mobility management, accessibility management, etc.

[0146] Although not illustrated in FIG. 8, the communication device may exchange information with another communication device having an AI / ML model for a different network function. Alternatively, the communication device may exchange information related to the specific network function with a specific communication device that transmitted the first information.

[0147] An example of FIG. 8 may be applied by combining at least some of the embodiments of FIG. 1 to 6 described above within a range where they are not arranged with each other. Accordingly, redundant descriptions are omitted. In addition, an example of FIG. 8 may be performed not only by operation between communication devices but also by operation of software programs, AI agents, applications, etc.

[0148] Below, methods for analyzing the cause of failures and taking corrective measures are described in the event that network failures, such as performance degradation or quality degradation, occur.

[0149] FIG. 9 is a diagram illustrating an example of a network failure monitoring method in a legacy mobile network operating system. FIG. 10 is a diagram illustrating an example of a network failure monitoring method in an NWDAF-based network operating system.

[0150] Referring to Fig. 9, the legacy mobile network operating system collects data from various interfaces of the mobile network to generate respective Call Detail Reports (CDRs), collects the CDRs at a central analysis server to generate Cross CDRs (xCDRs) that organize customer and service-centric information, and develops a model to monitor failures and anomalies by applying various AI / ML algorithms based on the CDRs to perform network performance management and anomaly monitoring. At this time, regarding network performance degradation and anomalies detected by the operator, the operator can identify the cause of the performance degradation and anomalies and take direct action.

[0151] Referring to Fig. 10, in network operation based on NWDAF, a method may be used to provide real-time information (Network Performance) or predictive information regarding quality anomalies (Observed Service Experience) of specific users or applications, or performance degradation in specific regions, to a Network Function (NF) that has requested a subscription.

[0152] In legacy mobile network operating systems, significant computing power is consumed for the generation of CDRs, the collection of information from central servers, and the subsequent creation of xCDRs. This not only causes delays in monitoring but also requires operator capability and considerable time to identify performance and quality anomalies, analyze causes, and determine corrective measures for improvement; furthermore, depending on the operator's competence, disaster recovery may be delayed.

[0153] In the case of NWDAF, while phenomena such as a decline in the perceived quality of specific applications or users, and performance degradation in specific regions can be identified or detected, automated root cause analysis for various problems may not be possible.

[0154] On the other hand, according to the network structure of Fig. 6, the global LLM can identify network functions commonly traversed by the problematic service through real-time correlation analysis of occupied resources used or traversed by the service where the deterioration of KPI or KQI (Key Quality Indicator) has occurred, and can directly transmit the relevant information (i.e., network failures such as deterioration of KPI and / or KQI) to the network function (or the local LLM or local AI / ML model corresponding to the network function). The AI ​​agent of the network function (or the local LLM or local AI / ML model) that receives the relevant information can find the root cause by analyzing internal data of the network function with high correlation (e.g., system log (syslog), key indicators, internal success rate, etc.) based on the time and / or number of occurrences of the service problem related to the service problem, and then directly solve the problem based on the cause or determine a method of problem solving and exchange information related to the problem solving method with the global AI / ML model or another local AI / ML model.

[0155] FIG. 11 is a flowchart for an example of a network failure recovery method performed by a communication device. Here, the communication device may include an AI / ML model for a wireless communication network. That is, the communication device may store and execute software or an application that implements an AI / ML model for a wireless communication network. Alternatively, the communication device may be a communication device that includes hardware for storing and executing software or an application that implements an AI / ML model for a wireless communication network.

[0156] Referring to FIG. 11, the communication device detects a specific communication device related to a network failure (S1110). Here, the network failure may include degradation of communication performance, degradation of communication quality, delay, etc. Additionally, the specific communication device may be a device in which a local LLM or local AI / ML model related to a network function related to the network failure is implemented, that is, a device in which a local LLM or local AI / ML model related to a network function related to a service in which the network failure occurred is implemented. Additionally, as an example, the communication device may be a device in which a global LLM or global AI / ML model is implemented. Meanwhile, to perform the step S1110, the communication device may monitor the network failure periodically, semi-periodically, or non-periodically.

[0157] The communication device transmits network failure information to the specific communication device (S1120). Here, the network failure information may include an identifier and / or information indicating that the network failure has occurred.

[0158] The specific communication device can detect the cause of the network failure based on the network failure information. Subsequently, the specific communication device can perform adjustment / control of network functions implemented by the specific communication device to resolve / eliminate the cause of the network failure.

[0159] An example of FIG. 11 may be applied by combining at least some of the embodiments of FIG. 1 to 10 described above within a range where they are not arranged with each other. Additionally, an example of FIG. 11 may be performed not only by operation between communication devices but also by operation of software programs, AI agents, applications, etc.

[0160] Figure 12 is a flowchart for another example of a network failure recovery method performed by a communication device.

[0161] Referring to FIG. 12, the first communication device receives network failure information from the second communication device (S1210). Here, the network failure information may include an identifier and / or information indicating that the network failure has occurred. Additionally, as in the example of FIG. 11, the second communication device may detect the first communication device associated with the network failure. Here, the network failure may include degradation of communication performance, degradation of communication quality, delay, etc. Furthermore, the first communication device may be a device in which a local LLM or local AI / ML model related to the network function associated with the network failure is implemented, that is, a device in which a local LLM or local AI / ML model related to the network function associated with the service in which the network failure occurred is implemented. Additionally, as an example, the second communication device may be a device in which a global LLM or global AI / ML model is implemented. Meanwhile, the second communication device may monitor the network failure periodically, semi-periodically, or non-periodically.

[0162] The first communication device adjusts a network function executed by the first communication device based on the network failure information (S1220). Here, the step S1220 may include at least some of the following operations: adjusting the network function using an AI / ML model (or local LLM) implemented by the first communication device; determining whether to retrain / update the AI / ML model and performing retraining / updating; and reporting the result of adjusting the network function to the second communication device.

[0163] An example of FIG. 12 may be applied by combining at least some of the embodiments of FIG. 1 to 10 described above within a range where they are not arranged with each other. Additionally, an example of FIG. 12 may be performed not only by operation between communication devices but also by operation of software programs, AI agents, applications, etc.

[0164] In the exemplary system described above, methods that can be implemented according to the features of the present invention described above have been described based on flowcharts. For convenience, the methods have been described as a series of steps or blocks; however, the claimed features of the present invention are not limited to the order of the steps or blocks, and some steps may occur with other steps in a different order or simultaneously with those described above. Furthermore, those skilled in the art will understand that the steps shown in the flowcharts are not exclusive, and that other steps may be included, or that one or more steps of the flowcharts may be omitted without affecting the scope of the present invention.

[0165] Various embodiments of the present invention may be implemented by hardware, firmware, software, or a combination thereof. In the case of implementation by hardware, it may be implemented by one or more ASICs (Application Specific Integrated Circuits), DSPs (Digital Signal Processors), DSPDs (Digital Signal Processing Devices), PLDs (Programmable Logic Devices), FPGAs (Field Programmable Gate Arrays), general processors, controllers, microcontrollers, microprocessors, etc.

[0166] The scope of the present invention includes software or machine-executable instructions (e.g., operating systems, applications, firmware, programs, etc.) that enable operations according to the methods of various embodiments to be executed on a device or computer, and a non-transitory computer-readable medium on which such software or instructions, etc. are stored and which are executable on a device or computer. Examples of computer-readable media include hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, flash memory, etc. Examples of program instructions include machine code, such as that produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc. The hardware devices described above may be configured to operate as at least one software module to perform the operations of the present invention, and vice versa.

[0167] The methods according to the present invention may be implemented in the form of program instructions that can be executed through various computer means and recorded on a computer-readable medium. A computer-readable medium may include program instructions, data files, data structures, etc., either individually or in combination. The program instructions recorded on the computer-readable medium may be those specifically designed and configured for the present invention, or they may be those known and available to those skilled in the art of computer software. The operation of the method according to an embodiment of the present invention may be implemented as a computer-readable program or code on a computer-readable recording medium. A computer-readable recording medium includes any type of recording device in which information that can be read by a computer system is stored. Additionally, the computer-readable recording medium may be distributed across networked computer systems, allowing computer-readable programs or code to be stored and executed in a distributed manner.

[0168] Some aspects of the invention have been described in the context of a device, but may also be described according to a corresponding method, wherein a block or device corresponds to a method step or a feature of a method step. Similarly, aspects described in the context of a method may also be described according to a corresponding block or item or a feature of a corresponding device. Some or all of the method steps may be performed by (or using) a hardware device, such as, for example, a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, at least one of the most important method steps may be performed by such a device.

[0169] In the embodiments, a programmable logic device (e.g., a field-programmable gate array) may be used to perform some or all of the functions of the methods described herein. In the embodiments, the field-programmable gate array may operate with a microprocessor to perform one of the methods described herein. Generally, it is preferable that the methods be performed by some hardware device.

[0170] The exemplary methods of the present invention are described as a series of operations for clarity of description, but this is not intended to limit the order in which the steps are performed, and if necessary, each step may be performed simultaneously or in a different order. To implement the method according to the present invention, additional steps may be included in addition to the steps exemplified, steps excluding some steps and including the remaining steps, or steps excluding some steps and including additional steps.

[0171] The various embodiments of the present invention are not intended to list all possible combinations but are intended to explain representative aspects of the invention, and the matters described in the various embodiments may be applied independently or in combination of two or more.

[0172] Although the present invention has been described with reference to preferred embodiments, those skilled in the art will understand that various modifications and changes can be made to the invention without departing from the spirit and scope of the invention as described in the following claims.

[0173] The claims described in this specification may be combined in various ways. For example, the technical features of the method claims of this specification may be combined to be implemented as a device, and the technical features of the device claims of this specification may be combined to be implemented as a method. Furthermore, the technical features of the method claims and the technical features of the device claims of this specification may be combined to be implemented as a device, and the technical features of the method claims and the technical features of the device claims of this specification may be combined to be implemented as a method. Moreover, the embodiments described in this specification may be combined within a range that is not mutually incompatible.

Claims

1. A method performed by a communication device in a wireless communication system, Perform training on the first AI / ML (Artificial Intelligence / Machine Learning) model, and A second AI / ML model for a specific network function is generated based on the above-mentioned first AI / ML model, wherein The communication device generates the second AI / ML model by performing learning on the specific network function, and The above communication device transmits information regarding the second AI / ML model to a specific communication device, A method comprising at least one of the information for the communication device to acquire the second AI / ML model and information for identifying the second AI / ML model among one or more AI / ML models stored in the specific communication device.

2. In Paragraph 1, A method comprising at least some of the steps of acquiring data for a plurality of network functions, performing preprocessing on the acquired data, and performing training on the first AI / ML model based on the preprocessed data.

3. In Paragraph 1, A method in which a second AI / ML model for the specific network function is generated based on data related to the specific network function and a Key Performance Indicator (KPI) related to the specific network function.

4. In Paragraph 3, A method in which data related to the specific network function is predefined or determined by the communication device.

5. In Paragraph 1, A method in which the communication device receives information related to the specific network function from the specific communication device, and performs an update to the first AI / ML model based on the information related to the specific network function.

6. In Paragraph 1, A method in which the communication device exchanges information about the specific communication device and the specific network function.

7. The communication device is, One or more memories for storing instructions; One or more transceivers; and It includes one or more processors connecting the one or more memories and the one or more transceivers, wherein the one or more processors execute the instructions, Perform training on the first AI / ML (Artificial Intelligence / Machine Learning) model, and A second AI / ML model for a specific network function is generated based on the above-mentioned first AI / ML model, wherein The communication device generates the second AI / ML model by performing learning on the specific network function, and The above communication device transmits information regarding the second AI / ML model to a specific communication device, A communication device comprising at least one of the information for the communication device to acquire the second AI / ML model and information for identifying the second AI / ML model among one or more AI / ML models stored in the specific communication device.

8. In Paragraph 7, A communication device comprising at least some of the steps of acquiring data for a plurality of network functions, performing preprocessing on the acquired data, and performing training on the first AI / ML model based on the preprocessed data.

9. In Paragraph 7, A communication device, wherein a second AI / ML model for the specific network function is generated based on data related to the specific network function and a Key Performance Indicator (KPI) related to the specific network function.

10. In Paragraph 9, A communication device in which data related to the specific network function above is predefined or determined by the communication device.

11. In Paragraph 7, A communication device that receives information related to a specific network function from a specific communication device, and performs an update to a first AI / ML model based on the information related to the specific network function.

12. In Paragraph 7, The communication device above is a communication device that exchanges information about the specific communication device and the specific network function.

13. A method performed by a communication device in a wireless communication system, Receive first information regarding a first AI / ML model, The first information includes at least one of information for the communication device to acquire the first AI / ML model and information for identifying the first AI / ML model among one or more AI / ML models stored in the communication device. Training of the above-mentioned first AI / ML model for a specific network function, The above learning is performed based on at least one of the data related to the specific network function and the KPI related to the specific network function, and A method for performing an operation related to a specific network function based on a first AI / ML model for the specific network function.

14. In Paragraph 13, The above first information is transmitted by a specific communication device, a method.

15. In Paragraph 14, A method in which the communication device exchanges information related to the specific communication device and the specific network function.

16. In Paragraph 13, A method in which the communication device omits the learning of the first AI / ML model based on the fact that the first AI / ML model is a model learned for the specific network function.

17. In Paragraph 13, A method for a communication device to exchange information with another communication device having an AI / ML model for other network functions.

18. In Paragraph 13, The above first AI / ML model is a basic model that has not been trained, a method.

19. In Paragraph 14, The above communication device receives second information from the above specific communication device, and A method for the communication device to adjust the specific network function based on the second information above.

20. In Paragraph 19, A method comprising adjusting the specific network function described above, including retraining the first AI / ML model.

Images