Machine learning model processing method, wireless communication device, and system
Patent Information
- Application Number
- PCT/CN2023/093705
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2023-05-11
- Publication Date
- 2024-11-14
Smart Images

Figure CN2023093705_14112024_PF_FP_ABST
Abstract
Description
MACHINE LEARNING MODEL PROCESSING METHOD, WIRELESS COMMUNICATION DEVICE, AND SYSTEMTechnical Field
[0001] The present disclosure relates to the field of communication systems, and more particularly, to a machine learning model processing method and a wireless communication system.Background Art
[0002] Artificial intelligence (AI) and machine learning (ML) are two related fields of computer science that aim to create systems that can perform tasks that normally require human intelligence and learning. ML can be used to solve various problems in domains such as natural language processing, computer vision, robotics, and bioinformatics. Recently, AI / ML has been increasingly applied to telecommunication networks.Technical Problem
[0003] Current technical standards for various aspects of the system, such as model transmission, model deployment, and signal design, are not well-defined or coordinated. This leads to several challenges that stem from two main factors, as explained below. Hence, it is imperative to establish a clear and consistent system architecture that can be implemented in the network as a common platform in the near future.Technical Solution
[0004] An object of the present disclosure is to propose a wireless communication device, such as a user equipment (UE) or a base station, and a machine learning model processing method based on machine learning.
[0005] In a first aspect, an embodiment of the invention provides a machine learning system comprising at least one of:
[0006] an artificial intelligence Management and orchestrator (AI-MAO) ,
[0007] an agent of AI-MAO, or
[0008] a radio access network side artificial intelligence / machine learning model (RAN-side AI / ML model) ;
[0009] wherein the AI-MAO, the agent, and the RAN-side AI / ML model are located in one network entity or distributed to different network entities.
[0010] In a second aspect, an embodiment of the invention provides a machine learning model processing method for machine learning (ML) , executable in a wireless communication device, comprising: performing an AI / ML-model-related operation for an AI / ML model that is associated with a logical AI / ML model identifier (ID) , wherein the logical AI / ML model ID is derivable from a global AI / ML model ID of the AI / ML model, and / or AI / ML capability information, and / or RRC configuration.
[0011] In a third aspect, an embodiment of the invention provides a wireless communication device comprising a processor configured to call and run a computer program stored in a memory, to cause a device in which the processor is installed to execute the disclosed method.
[0012] The disclosed method may be implemented in a chip. The chip may include a processor, configured to call and run a computer program stored in a memory, to cause a device in which the chip is installed to execute the disclosed method.
[0013] The disclosed method may be programmed as computer-executable instructions stored in non-transitory computer-readable medium. The non-transitory computer-readable medium, when loaded to a computer, directs a processor of the computer to execute the disclosed method.
[0014] The non-transitory computer-readable medium may comprise at least one from a group consisting of: a hard disk, a CD-ROM, an optical storage device, a magnetic storage device, a Read Only Memory, a Programmable Read Only Memory, an Erasable Programmable Read Only Memory, EPROM, an Electrically Erasable Programmable Read Only Memory and a Flash memory.
[0015] The disclosed method may be programmed as a computer program product, which causes a computer to execute the disclosed method.
[0016] The disclosed method may be programmed as a computer program, which causes a computer to execute the disclosed method.Advantageous Effects
[0017] This disclosure proposes a novel AI / ML framework, model ID definition and protocol design with model ID under the AI / ML framework. This topic has not been reached by current research and standardization work. The disclosed method can enhance the wireless air interface by reducing signaling overhead, providing scalability to meet different models' requirements for one AI model type, and unifying the procedures for one-side or two-side models with a single AI model specific process.Description of Drawings
[0018] In order to more clearly illustrate the embodiments of the present disclosure or related art, the following figures will be described in the embodiments are briefly introduced. It is obvious that the drawings are merely some embodiments of the present disclosure. A person having ordinary skill in this field can obtain other figures according to these figures without paying the premise.
[0019] FIG. 1 illustrates a schematic view showing an AI / ML system in the disclosed AI / ML framework.
[0020] FIG. 2 illustrates a schematic view showing a wireless communication system comprising a user equipment (UE) , a base station, and a network entity.
[0021] FIG. 3 illustrates a schematic view showing a system for executing a machine learning model processing method using ML models.
[0022] FIG. 4 illustrates a schematic view showing an embodiment of the disclosed method.
[0023] FIG. 5 illustrates a schematic view showing a system for wireless communication according to an embodiment of the present disclosure.DETAILED DESCRIPTION OF EMBODIMENTS
[0024] Embodiments of the disclosure are described in detail with the technical matters, structural features, achieved objects, and effects with reference to the accompanying drawings as follows. Specifically, the terminologies in the embodiments of the present disclosure are merely for describing the purpose of the certain embodiment, but not to limit the disclosure.
[0025] Abbreviations used in the description are listed in the following:
[0026] Table 1
[0027] Embodiments of the disclosure are related to artificial intelligence (AI) and machine learning (ML) for new radio (NR) air interface and address problems of AI / ML model ID.
[0028] For simplicity, an AI / ML model, AI model, ML model, and model are interchangeably used in the description. In the description, device operations may comprise reference signal (RS) measurement, channel state information (CSI) reporting, beam prediction, positioning, and others. Model operations may comprise model selection, activation, deactivation, switching, fallback, model training, model monitoring, and / or fine-tuning at least for one-sided models and two-sided models.
[0029] A machine learning system is provided. The machine learning system comprises at least one of an artificial intelligence management and orchestrator (AI-MAO) , an agent of AI-MAO, a radio access network side artificial intelligence / machine learning model (RAN-side AI / ML model) . The AI-MAO, the agent, and the RAN-side AI / ML model are located in one network entity or distributed to different network entities. The RAN-side AI / ML model provides an AI / ML service to assist one or more device operations.
[0030] With reference to FIG. 1, a system architecture for AI / ML framework in a telecommunication network is provided. In the system architecture, a system 100 comprises an AI-MAO 34 and agents for the AI-MAO.
[0031] The AI-MAO may be deployed in a user equipment (UE) , a base station (BS) , or a network device that operates as an individual management platform separated from the UE and the base station.
[0032] The agents for the AI-MAO may be referred to as AI-MAO agents. The agents for the AI-MAO can be installed and executed in one or more network entities, such as UE (s) and / or base station (s) , in the RAN or CN. In FIG. 1, an agent 14 for the AI-MAO is installed and executed in UE 10, and an agent 24 for the AI-MAO is installed and executed in a base station 20. The agents can communicate with each other and with the AI-MAO 34 to share data and control signal used for or collected from model operations. For example, the model operation may comprise AI / ML model monitoring, activation / deactivation, update, switching, fallback. The agent of AI-MAO may be deployed as a UE-side agent or a BS-side agent. The RAN-side AI / ML model may be deployed as a UE-side AI / ML model or a BS-side AI / ML model. The UE-side agent 14 and the BS-side agent 24 are also RAN-side agents.
[0033] AI / ML models can be installed and executed in one or more network entities, such as UE (s) and / or base station (s) . In FIG. 1, a UE-side AI / ML model 15 is installed and executed in UE 10, and an BS-side AI / ML model 25 is installed and executed in a base station 20. The AI / ML models perform AI / ML-based operation to assist device operations and / or model operations. For example, the device operations may comprise RS measurement, CSI reporting, beam prediction, positioning, and others.
[0034] The system 100 enables operators to train custom AI / ML models and integrate third-party AI / ML models under their control. This system architecture allows network-based AI / ML services to support a wider range of partners and deliver services that meet market needs. The disclosed system and methods can improve service quality and increase market competitiveness.
[0035] The AL / ML system architecture has three main components: AI-MAO (AI-Management and orchestrator) , AI-MAO agent, and RAN-side AI. These components can be deployed in different ways, depending on the network entity (e.g., UE or gNB) that hosts them. For example, the AI-MAO, the AI-MAO agent, and the RAN-side AI can be co-located in one network entity, or they can be distributed across different entities. For example, the AI-MAO may be deployed in the BS side, the UE side, or a separate management platform that is not part of the UE or the base station. The functions of each component are described below:
[0036] 1. AI-MAO: A platform for AI / ML model lifecycle management (LCM) that covers model training, deployment, management, registration, activation, deactivation, storage and operational charging etc.
[0037] 2. AI-MAO Agent: An agent for AI-MAO is referred to as an AI-MAO agent. An AI-MAO agent is a component that is deployed on the UE or gNB side and interfaces with AI-MAO. An AI-MAO agent is responsible for synchronizing requirements between network entities and AI-MAO, including data collection, model deployment / monitoring / activation / deactivation, AI / ML model information lookup, etc.
[0038] 3. RAN-side AI capability: A feature that is deployed in network entities (e.g., UE, gNB, 5G core etc. ) . It is responsible for the application of the model in the network, including at least inference and monitoring.
[0039] With reference to FIG. 2, a telecommunication system including a UE 10a, a UE 10b, a base station (BS) 20a, and a network entity device 30 executes the disclosed method according to an embodiment of the present disclosure. FIG. 2 is shown for illustrative not limiting, and the system may comprise more UEs, BSs, and CN entities. Connections between devices and device components are shown as lines and arrows in the FIGs.
[0040] The base station 20a can operate as a gNB for 5G NR networks, an eNB for LTE networks, or a base station for future mobile network systems beyond 5G. A gNB is a 5G radio network node that connects to the core network via the NG interface. An eNB is a 4G radio network node that connects to the evolved packet core via the S1 interface. A base station for beyond 5G may be a smart virtual eNB (SVeNB) that can perform functions of EPS elements and reduce end-to-end delay.
[0041] The UE 10a may include a processor 11a, a memory 12a, and a transceiver 13a. The UE 10b may include a processor 11 b, a memory 12b, and a transceiver 13b. The base station 20a may include a processor 21a, a memory 22a, and a transceiver 23a. The network entity device 30 may include a processor 31, a memory 32, and a transceiver 33. Each of the processors 11a, 11b, 21a, and 31 may be configured to implement proposed functions, procedures and / or methods described in the description. Layers of radio interface protocol may be implemented in the processors 11a, 11b, 21a, and 31. Each of the memory 12a, 12b, 22a, and 32 operatively stores a variety of programs and information to operate a connected processor. Each of the transceivers 13a, 13b, 23a, and 33 is operatively coupled with a connected processor, and transmits and / or receives radio signals or wireline signals. The UE 10a may be in communication with the UE 10b through a sidelink. The base station 20a may be an eNB, a gNB, or one of other types of radio nodes, and may configure radio resources for the UE 10a and UE 10b.
[0042] Each of the processors 11a, 11b, 21a, and 31 may include an application-specific integrated circuit (ASICs) , other chipsets, logic circuits and / or data processing devices. Each of the memory 12a, 12b, 22a, and 32 may include read-only memory (ROM) , a random access memory (RAM) , a flash memory, a memory card, a storage medium and / or other storage devices. Each of the transceivers 13a, 13b, 23a, and 33 may include baseband circuitry and radio frequency (RF) circuitry to process radio frequency signals. When the embodiments are implemented in software, the techniques described herein may be implemented with modules, procedures, functions, entities, and so on, that perform the functions described herein. The modules may be stored in a memory and executed by the processors. The memory may be implemented within a processor or external to the processor, in which those may be communicatively coupled to the processor via various means are known in the art.
[0043] The network entity device 30 may be a node in a CN. CN may include LTE CN or 5G core (5GC) which includes user plane function (UPF) , session management function (SMF) , access and mobility management function (AMF) , unified data management (UDM) , policy control function (PCF) , control plane (CP) / user plane (UP) separation (CUPS) , authentication server (AUSF) , network slice selection function (NSSF) , and the network exposure function (NEF) .
[0044] An example of the UE in the description may include one of the UE 10a or UE 10b. An example of the base station in the description may include the base station 20a. The base station 20 may be a gNB or a base station of any other type of base station, such as an eNB or a base station for beyond 5G. Uplink (UL) transmission of a control signal or data may be a transmission operation from a UE to a base station. Downlink (DL) transmission of a control signal or data may be a transmission operation from a base station to a UE. A DL control signal may comprise downlink control information (DCI) or a radio resource control (RRC) signal, from a base station to a UE.
[0045] With reference to FIG. 3, the system 100 for the machine learning model processing method based on machine learning comprises units of data collection 101, model training unit 102, actor 103, and model inference 104. Please note that FIG. 3 does not necessarily limit the machine learning model processing method to the instant example. The machine learning model processing method is applicable to any design based on machine learning. The general steps comprise data collection and / or model training and / or model inference and / or (an) actor (s) .
[0046] The data collection unit 101 is a function that provides input data to the model training unit 102 and the model inference unit 104. AI / ML algorithm-specific data preparation (e.g., data pre-processing and cleaning, formatting, and transformation) is not carried out in the data collection unit 101.
[0047] Examples of input data may include measurements from UEs or different network entities, feedback from Actor 103, and output from an AI / ML model.
[0048] Training data is data needed as input for the AI / ML Model training unit 102.
[0049] Inference data is data needed as input for the AI / ML Model inference unit 104.
[0050] The model training unit 102 is a function that performs the ML model training, validation, and testing. The Model training unit 102 is also responsible for data preparation (e.g., data pre-processing and cleaning, formatting, and transformation) based on training data delivered by the data collection unit 101, if required.
[0051] Model deployment / update between units 102 and 104 involves deployment or update of an AI / ML model (e.g., a trained machine learning model 105a or 105b) to the model inference unit 104. The model training unit 102 uses data units as training data to train a machine learning model 105a and generates a trained machine learning model 105b from the machine learning model 105a.
[0052] The model inference unit 104 is a function that provides AI / ML model inference output (e.g., predictions or decisions) . The AI / ML model inference output is the output of the machine learning model 105b. The model inference unit 104 is also responsible for data preparation (e.g., data pre-processing and cleaning, formatting, and transformation) based on inference data delivered by the data collection unit 101, if required.
[0053] The output shown between unit 103 and unit 104 is the inference output of the AI / ML model produced by the model inference unit 104.
[0054] The actor 103 is a function that receives the output from the model inference unit 104 and triggers or performs corresponding actions. The actor 103 may trigger actions directed to other entities or to itself.
[0055] Feedback between unit 103 and unit 101 is information that may be needed to derive training or inference data or performance feedback.
[0056] FIG. 4 shows an embodiment of the disclosed method. At least one wireless communication device may be a network entity and executes a machine learning model processing method based on machine learning. In an embodiment, the at least one wireless communication device may comprise a user equipment (UE) . In another embodiment, the at least one wireless communication device may comprise a base station. In still another embodiment, the at least one wireless communication device may comprise a combination of UEs and base stations.
[0057] The at least one wireless communication device performs an AI / ML-model-related operation for an AI / ML model that is associated with a logical AI / ML model identifiers (ID) , wherein the logical AI / ML model ID of the AI / ML model is derivable from a global AI / ML model ID, and / or AI / ML capability information, and / or radio resource control (RRC) configuration (S001) .
[0058] The AI / ML-model-related operation comprises either or both of model operation and device operation. For example, the AI / ML-model-related operation comprises signaling or execution of one or more of:
[0059] ● a model deployment,
[0060] ● AI / ML model-related information,
[0061] ● AI / ML model activation,
[0062] ● AI / ML model deactivation,
[0063] ● AI / ML model monitoring,
[0064] ● reporting for AI / ML model monitoring,
[0065] ● AI / ML model update,
[0066] ● AI / ML model reselecting,
[0067] ● AI / ML model switching,
[0068] ● AI / ML model fallback, and
[0069] ● radio resource control (RRC) configuration.
[0070] In an embodiment, the logical AI / ML model ID comprises one or more of scenario, function, feature, or local ID of the AI / ML model. The global AI / ML model ID of the AI / ML model is mapped to the logical AI / ML model ID of the AI / ML model according to a mapping rule. The mapping rule may comprise direct mapping or hierarchical mapping. The direct mapping comprises:
[0071] ● a sequential mapping,
[0072] ● an inverse mapping, or
[0073] ● transparent mapping by which the logical AI / ML model ID is equal to the local ID.
[0074] In the hierarchical mapping of AI / ML models, the AI / ML models that share the same constraints, are arranged in a specific order based on IDs of each attribute involved in the constraints, the logical AI / ML model ID of the AI / ML model corresponds to a position of the AI / ML model in the arranged AI / ML models, which are arranged either in sequential or inverse order.
[0075] In an embodiment, when the logical AI / ML model ID of the AI / ML model is derived from the AI / ML capability information, ID used for configuring radio resource control (RRC) configuration is used as the logical AI / ML model ID of the AI / ML model which is associated with the AI / ML capability information. Alternatively, the logical AI / ML model ID of the AI / ML model is included in RRC configuration and is associated with the AI / ML capability information.
[0076] In an embodiment, when the logical AI / ML model ID of the AI / ML model is derived from the RRC configuration, an ID of the RRC configuration is used as the logical AI / ML model ID of the AI / ML model, and the ID of the RRC configuration comprises one or more of: measurement configuration ID, CSI measurement configuration ID, SRS resource set ID, SRS resource configuration ID, CSI-RS resource set ID, CSI-RS resource ID, and reporting configuration ID.
[0077] The AI / ML capability information comprises one or more of:
[0078] at least one type of AI / ML model related information;
[0079] at least one global AI / ML model ID; and
[0080] an AI / ML capability information index.
[0081] The AI / ML capability information is carried in an uplink transmission signal or a downlink transmission signal. The uplink transmission signal includes one or more of UE capability reporting, physical uplink control channel (PUCCH) , or physical uplink shared channel (PUSCH) . The downlink transmission signal includes one or more of RRC message, medium access control -control element (MAC-CE) , downlink control information (DCI) , physical downlink shared channel (PDSCH) , broadcast signaling, a system information block (SIB) , or a master information block (MIB) .
[0082] In an embodiment, the device operation comprises pilot signal configuration. For example, the pilot signal configuration comprises configuration of CSI-RS, SRS, TRS, PTRS, or DMRS. ID used for configuring pilot signal configuration, measurement, reporting, and event activation is used as the logical AI / ML model ID of the AI / ML model which is associated with AI / ML capability information.
[0083] Alternatively, the logical AI / ML model ID of the AI / ML model is included in pilot signal configuration, measurement, reporting, event activation, or other RRC configurations, and is associated with AI / ML capability information.
[0084] In an embodiment, the AI / ML-model-related operation comprises AI / ML model activation and / or AI / ML model deactivation. AI / ML model activate / deactivate information for the AI / ML model activation or AI / ML model deactivation comprises one or more of:
[0085] ● an AI / ML model ID,
[0086] ● activate / deactivate information,
[0087] ● activate timing,
[0088] ● deactivate timing, and
[0089] ● a timing window.
[0090] The AI / ML model activate / deactivate information may be transmitted in an uplink transmission or a downlink transmission. For example, the base station 20 may transmit the AI / ML model activate / deactivate information in a PDSCH, PDCCH, RRC message, MAC CE, SIB, MIB, or a unicast signal to the UE 10. In another example, the UE 10 may transmit the AI / ML model activate / deactivate information in a PUSCH, PUCCH, RRC message, or MAC CE to the base station 20. In another example, the UE 10 may transmit the AI / ML model activate / deactivate information in a PSSCH, PSCCH, RRC message, or MAC CE to the another UE.
[0091] In an embodiment, when the AI / ML model is used for configuring a reference signal related configuration, the reference signal related configuration comprises one or more of:
[0092] ● an AI / ML model ID,
[0093] ● reference signal information configuration, and
[0094] ● CSI reporting information configuration.
[0095] In an embodiment, the AI / ML model is used for CSI prediction.
[0096] Protocol designing under the system architecture:
[0097] The description proposes embodiments of the disclosure using the proposed system architecture in the following. Features are described in detail in the corresponding embodiments, comprising:
[0098] 1) Signaling between AI-MAO and its AI-MAO agent: Embodiment 1 includes the interface definition and the associated transmission elements or information for the disclosed system architecture.
[0099] 2) AI / ML model ID definition: Embodiment 2 defines a method for mapping between a Logical AI / ML model ID and a Global AI / ML model ID, as well as between the Logical AI / ML model ID and the RRC information configuration ID.
[0100] 3) RAN-side signaling with the Logical AI / ML model ID: Embodiments 3 to 5 provide signaling based on the Logical AI / ML model ID on the air interface.
[0101] Definitions:
[0102] Definition of terms is listed in the following:
[0103] 1) Global AI / ML model ID: A Global AI / ML model ID is a unique identifier for an AI / ML model that can be assigned at different levels of granularity, such as global, PLMN-specific, operator-specific, or AI / ML management platform-specific.
[0104] 2) Logical AI / ML model ID: A Logical AI / ML model ID is a unique identifier for an AI / ML model that is used in the network and that has a mapping relationship with the Global AI / ML model ID. The Logical ID can be assigned at different levels of granularity within the network, such as cell-specific, link-specific, TA-specific, CU-specific, DU-specific, UPF-specific, AMF-specific, RRC-specific, or network-slicing-specific. The term "cell-specific" means that each AI / ML model has a unique ID within a specific cell. Similarly, "link-specific" , "TA-specific" , "CU-Specific" , "DU-specific" , "UPF-specific" , "AMF-specific" , and "network-slicing-specific" mean that each AI / ML model has a unique ID within a specific context in the network.
[0105] 3) Hardware capability: The hardware capability of the CPU and / or GPU depends on various parameters, such as manufacturer, version, computational performance, number of threads, etc. The computational performance can be measured by metrics such as FLOPS, IPC, memory bandwidth, cache size, clock speed, etc.
[0106] 4) Software capability: The software capability of the system includes various components, such as manufacturer, version, operating system (OS) , application programming interface (API) , software development kit (SDK) , runtime environment, firmware, compiler, etc.
[0107] 5) Supported AI capability: The supported AI capability of the system refers to the types of AI / ML models that can be executed on the system, the computing resources and power consumption required for each model, and the file format used for storing or exchanging the models and related data. The types of AI / ML models include convolutional neural networks (CNNs) , recurrent neural networks (RNNs) , deep belief networks (DBNs) , support vector machines (SVMs) , random forest, k-nearest neighbors (KNN) , naive Bayes, etc.
[0108] 6) Real-time performance: The real-time performance of the system involves various indicators, such as CPU / GPU / memory / storage / network usage, temperature, fan speed, power consumption, voltage, error rate of hardware components, response time, thread utilization, etc.
[0109] 7) Network entity identifier: The network entity must be identified by at least one of the following network entity identifier:
[0110] ● UE identifier: IMEI (International Mobile Equipment Identity) or UUID; and
[0111] ● RAN identifier: cell ID, physical ID, PLMN, UUID, or another identifier.
[0112] 8) AI / ML model related information: The AI / ML model related information may include at least one of the following information:
[0113] ■ Global AI / ML model ID,
[0114] ■ Provider,
[0115] ■ Scenario,
[0116] ■ Feature,
[0117] ■ Function,
[0118] ■ Version,
[0119] ■ Accuracy,
[0120] ■ RRC descriptor,
[0121] ■ AI / ML model descriptor,
[0122] ■ Compute descriptor,
[0123] ■ Storage descriptor, and
[0124] ■ Power consumption.
[0125] Among the AI / ML model related information, some are detailed in the following.
[0126] ● Scenario: The scenario attribute for an AI / ML model specifies a scenario setting for the AI / ML model. For example, in a scenario, an AI / ML model is used for a certain feature in a FDD massive MIMO system. For example, the scenario attribute can be set to different values that indicate the type of environment for which the AI / ML model is optimized. Some examples of scenario values are indoor, outdoor, flying, water, and mobility speed. These values correspond to environments that have different characteristics.
[0127] ● Feature: The feature attribute may be configured to CSI compression, CSI prediction, beam management, positioning, handover, radio resource management, etc.
[0128] ● Function: The function attribute may be configured to CSI compression, CSI prediction, beam management, positioning, handover, radio resource management, etc. Additionally, the function attribute may be configured to represent usage of the AI / ML model. For example, the function attribute of the AI / ML model may be configured to monitoring, inference, and / or training.
[0129] ● RRC descriptor: The RRC descriptor defines the reference signal configuration and the CSI measurement parameters for the AI / ML model. These parameters include one or more of the following:
[0130] ■ reference signal timing configuration,
[0131] ■ reference signal frequency configuration,
[0132] ■ reference signal antenna port number, and
[0133] ■ CSI content configuration.
[0134] The reference signal timing configuration can include various parameters, such as the duration of each period, the interval between time steps, and the pattern of hopping between different time slots. Similarly, the reference signal frequency configuration can specify different aspects of the frequency domain, such as the allocation of frequency resources, the scheme of hopping between different frequency bands, and so on. The CSI content configuration can determine what kind of information is conveyed by the CSI, such as the channel matrix that describes the channel state, the eigen-vector that represents the dominant direction of the channel, and so on.
[0135] ● AI / ML model descriptor: The AI / ML model descriptor describes the model attributes of the AI / ML model. It includes information such as:
[0136] ■ AI / ML type,
[0137] ■ Accuracy,
[0138] ■ Input data requirement,
[0139] ■ Output data,
[0140] ■ Monitoring method,
[0141] ■ Input data distribution, and / or
[0142] ■ Output data distribution.
[0143] ● Compute descriptor: The compute descriptor specifies the computation resource and the type of CPU or GPU needed for the AI / ML model. It includes information such as:
[0144] ■ Number of CPU cores or GPU units,
[0145] ■ Clock speed,
[0146] ■ Memory bandwidth, and / or
[0147] ■ Cache size.
[0148] ● Storage descriptor: The storage descriptor specifies the storage resource and the type of memory needed for the AI / ML model. It includes information such as:
[0149] ■ Memory size,
[0150] ■ Memory type (e.g., RAM, ROM, flash, etc. ) ,
[0151] ■ Memory access speed, and / or
[0152] ■ Memory latency.
[0153] ● Power consumption: The power consumption attribute describes the amount of power consumed by the AI / ML model's computation. It includes information such as:
[0154] ■ Voltage,
[0155] ■ Current,
[0156] ■ Average power consumption,
[0157] ■ Peak power consumption, and / or
[0158] ■ Power efficiency.
[0159] 9) Data collection descriptor: The data collection descriptor specifies either one or both of the following: a training data descriptor or a monitoring descriptor.
[0160] ● Training data descriptor: The training data descriptor specifies the data collected for AI / ML model training. It includes one or more of the following: channel information (e.g., real-time channel matrix, long-term channel matrix, CSI, eigen-vector, etc. ) , system logging information (e.g., computation, storage, memory, etc. ) .
[0161] ● Monitoring descriptor: The monitoring descriptor describes the monitored data for AI / ML model, including one or more of the following: performance loss, performance gain.
[0162] Embodiment 1: Signaling between AI-MAO and agent of AI-MAO:
[0163] The following paragraph describes the system architecture and the example procedure of AI-MAO, a platform that enables the deployment and management of AI / ML models on various network entities. The system architecture consists of two main components: AI-MAO and the Agent of AI-MAO. AI-MAO is a centralized server that stores and manages the AI / ML models and metadata of the AI / ML models. The Agent of AI-MAO is a software module that runs on each network entity and communicates with the AI-MAO to perform one of more of requesting, deploying, activating, deactivating, and updating the AI / ML models. The example procedure is as follows:
[0164] 1) The Agent of AI-MAO sends a model deployment request message to AI-MAO, which includes at least one of the following information: network entity identifier, hardware capability, software capability, supported AI capability, real-time system performance, and AI / ML model-related information. This message indicates that the network entity requests to use an AI / ML model for a specific task or service. The information included in the message specifies attributes of the requested AI / ML model.
[0165] 2) Upon receiving the request from the Agent of AI-MAO, the AI-MAO determines which AI / ML models are available to be deployed or can be used based on the information in the model deployment request message. The AI-MAO selects the most suitable AI / ML model for the network entity according to its characteristics and requirements specified by the information in the model deployment request message. The AI-MAO provides signaling that indicates which AI / ML models are to be used based on information in the model deployment request message.
[0166] 3) AI-MAO sends information (e.g., AI / ML model-related information) to the Agent of AI-MAO, which includes at least AI / ML model-related information. This information contains the details of the selected AI / ML model as illustrated in the foregoing paragraphs.
[0167] 4) Based on the information received, the Agent of AI-MAO deploys and / or activates one or more AI / ML models indicated by the AI-MAO. The Agent of AI-MAO downloads the one or more AI / ML models determined by the AI-MAO from AI-MAO or a local cache and installs it on the network entity. The Agent of AI-MAO then initializes and runs the AI / ML model to provide the desired functionality or service.
[0168] 5) The Agent of AI-MAO sends at least one type of the following information to AI-MAO as feedback: real-time performance, AI / ML model-related information, data collection descriptor associated with the deployed one or more AI / ML models. This feedback helps AI-MAO to monitor and evaluate the performance and quality of the deployed AI / ML model. The data collection descriptor has been defined in the foregoing paragraphs. The data collection descriptor may specify what kind of data is collected by the network entity and how it is processed and stored.
[0169] 6) Based on the information in the feedback from the Agent of AI-MAO, the AI-MAO makes a decision and sends instructions reflecting the decision to the Agent of AI-MAO to deactivate or update the one or more AI / ML models. If the performance or quality of the deployed AI / ML model is unsatisfactory or outdated, AI-MAO can instruct the Agent of AI-MAO to stop using the deployed AI / ML model or to replace the deployed AI / ML model with a newer or better AI / ML model.
[0170] The interface between the AI-MAO and the agent of AI-MAO conveys information about various aspects of the system (i.e., the AI-MAO and the agent of AI-MAO) , such as hardware capability, software capability, supported AI capability, real-time performance, network entity identifier, AI / ML model-related information, and data collection descriptor.
[0171] Embodiment 2: Logical AI / ML model ID definition and mapping method:
[0172] Embodiment 2-1: Logical AI / ML model ID associated with the AI / ML Attribute:
[0173] The Logical AI / ML model ID is a unique identifier that represents a specific AI / ML model in a given context. The Logical AI / ML model ID includes one or more of the following components: scenario, function, feature, and local ID. The scenario describes the general use case or domain of the AI / ML model. The function describes the main task or objective of the AI / ML model. The feature describes the specific attribute or characteristic of the AI / ML model. The local ID is a numerical value that maps to the Global AI / ML model ID. The Global AI / ML model ID is a universal identifier that can be used across different platforms and systems. The following table shows an example of how to define the Logical AI / ML model ID for different AI / ML models.
[0174] Table 2: Example of the Logical AI / ML model ID definition
[0175] As shown below, there are different methods for mapping between Logical AI / ML model ID and Global AI / ML model, such as direct mapping and hierarchical mapping.
[0176] 1) Directly mapping method:
[0177] The mapping rule can be one of the following types: sequential mapping, inverse mapping, or transparent mapping.
[0178] Sequential mapping assigns local IDs to global AI / ML model IDs in ascending order, starting from 0. Inverse mapping does the same, but in descending order of global IDs. In the sequential mapping, all global AI / ML model IDs in the system are sorted in ascending order, and the corresponding local IDs for mapping are arranged in ascending order starting from 0 and assigned to the sorted global AI / ML model IDs. In the inverse mapping, all global AI / ML model IDs in the system are sorted in descending order, and the corresponding local IDs for mapping are arranged in ascending order starting from 0 and assigned to the sorted global AI / ML model IDs.
[0179] The transparent mapping sets the logical AI / ML model ID equal to the local ID.
[0180] For example, suppose there are 16 AI / ML models in a wireless system (corresponding to 2 scenarios, 2 features, 2 functions, and 2 RRC descriptors) , and 4 bits are used for local IDs. The table below shows how each mapping type works.
[0181] Table 3: Directly mapping method between Local ID and Global AI / ML model ID
[0182] 2) Hierarchical mapping
[0183] In hierarchical mapping, a local ID refers to a representation of AI / ML model information under certain constraints. For example, the constraints may be a combination of specific scenarios, features, and / or functions. All AI / ML models that share the same constraints are ordered by their unique IDs. Local ID corresponds to the position of each AI / ML model unique ID in the ordered sequence, either in ascending or descending order. The table in the following illustrates this mapping.
[0184] Table 4: Hierarchical mapping method between Local ID and AI / ML model unique ID
[0185] Embodiment 2-1: Logical AI / ML model ID associated with the RRC configuration:
[0186] Wireless communication systems use various ID information to identify RRC configurations, such as measurement configuration ID, CSI measurement configuration ID, SRS resource set ID, SRS resource configuration ID, CSI-RS resource set ID, CSI-RS resource ID, reporting configuration ID, and so on. To reduce the overhead of air interface signaling in wireless communication systems, several methods are proposed for associating Logical AI / ML model ID with RRC configurations. The following are some examples of these methods:
[0187] 1) Mapping to AI / ML capability information:
[0188] A wireless network entity (UE or gNB) supports AI / ML capabilities and transmits information of the AI / ML capabilities on the air interface. This AI / ML capability information can be conveyed in an uplink transmission signal or a downlink transmission signal. For example, the uplink transmission signal may include one or more of UE capability reporting, PUCCH, or PUSCH. For example, the downlink transmission signal may include one or more of RRC message, MAC-CE, DCI, or PDSCH, or broadcast signaling, or SIB / MIB. This information comprises one or more of the following:
[0189] ● At least one type of AI / ML model related information, as defined in the preceding paragraphs.
[0190] ● At least one global AI / ML model ID, as defined in the preceding paragraphs.
[0191] ● The AI / ML capability information index, which is used to identify an AI / ML capability.
[0192] The RRC configuration includes various parameters such as pilot signal configuration (CSI-RS, SRS, TRS, PTRS, DMRS, etc. ) , measurement configuration, reporting configuration, event activation configuration, etc. In a RRC configuration, the logical AI / ML model ID used for the RRC configuration is directly mapped to the AI / ML capability information supported by the network entity (UE or gNB) , as shown below.
[0193] ● Option A: The logical AI / ML model ID, which is associated with AI / ML capability information, is derived from the ID used for configuring RRC configuration. For example, the ID used for configuring RRC configuration may comprise ID used for configuring various reference signals (CSI-RS, SRS, TRS, PTRS, DMRS) and measurement / reporting / event activation procedures. For instance, the logical AI / ML model ID can be equal to the AI / ML capability information index, or can be derived from on a sorted sequence of AI / ML model related information or global AI / ML model ID carried by the AI / ML capability information, or computed from one of the above methods.
[0194] ● Option B: The configuration of CSI-RS, SRS, TRS, PTRS, DMRS, measurement, reporting, event activation, or other RRC configurations includes the logical AI / ML model ID, which corresponds to AI / ML capability information. For instance, the logical AI / ML model ID can be directly equal to the AI / ML capability information index, or derived from a sorted sequence of AI / ML model related information or Global AI / ML model ID that is carried by the AI / ML capability information, or calculated based on the results of one of the above methods.
[0195] In an embodiment the network entity uses the AI / ML model for CSI reporting. The CSI reporting may have the following scenarios:
[0196] ● Option A: The CSI information does not contain logical AI / ML model ID (or related information) .
[0197] ● Option B: The CSI information contains logical AI / ML model ID (or related information) , which could be one or more of the following: measurement configuration ID, reporting configuration ID, CSI-RS resource set / resource ID information, etc. That is, in an embodiment, the pilot signal configuration comprises configuration of CSI-RS, the logical AI / ML model ID of the AI / ML model is one or more of: measurement configuration ID, reporting configuration ID, CSI-RS resource set / resource ID information, and monitoring configuration ID.
[0198] ● Option C: The CSI information contains logical AI / ML model ID (or related information) , which is associated with AI / ML capability information. The logical AI / ML model ID could be equal to the AI / ML capability information index, or could be derived from a sorted sequence of AI / ML model related information or Global AI / ML model ID in the AI / ML capability information. Alternatively, the logical AI / ML model ID could be calculated using the AI / ML capability information index.
[0199] In an embodiment the network entity uses the AI / ML model for uplink, downlink, or sidelink transmission. When the network entity transmits data over channels such as PDSCH, PUSCH, or PSSCH, the logical AI / ML model ID used can be identified using the schemes detailed in the following options:
[0200] ● Option A: The logical AI / ML model ID used can be identified by CSI measurement configuration ID information, which includes one or more of the CSI measurement configuration ID, the reporting configuration ID, and the CSI-RS resource set / resource ID.
[0201] ● Option B: The logical AI / ML model ID that is associated with the AI / ML capability information. The logical AI / ML model ID can either be the same as the AI / ML capability information index, or can be derived from a sorted sequence of the AI / ML model related information or the global AI / ML model ID in the AI / ML capability information. Alternatively, the logical AI / ML model ID could be calculated using the AI / ML capability information index.
[0202] 2) Mapping to AI / ML RRC configuration / request / reconfiguration information
[0203] The RRC configuration includes pilot signal configuration (such as CSI-RS, SRS, TRS, PTRS, DMRS, etc. ) , measurement configuration, reporting configuration, event activation configuration, etc. The RRC configuration contains at least one of the following elements:
[0204] ● At least one type of AI / ML model related information, as defined in the previous paragraphs.
[0205] ● At least one global AI / ML model ID, as defined in the previous paragraphs.
[0206] In an embodiment the network entity uses the AI / ML model for CSI reporting. The CSI reporting may have the following scenarios:
[0207] ● Option A: The CSI information does not contain logical AI / ML model ID (or related information) .
[0208] ● Option B: The CSI information contains logical AI / ML model ID (or related information) , which may include one or more of the following: measurement configuration ID, reporting configuration ID, CSI-RS resource set / resource ID information, and / or monitoring configuration ID etc.
[0209] In an embodiment the network entity uses the AI / ML model for uplink, downlink, or sidelink transmission. During the process of transmitting data, such as decoding PDSCH, PUSCH, or PSSCH, or collecting data for AI / ML models (CSI measurement, inference, training, monitoring, etc. ) , the logical AI / ML model ID (or related information) can be one or more of the following:
[0210] ● The measurement configuration ID,
[0211] ● The reporting configuration ID,
[0212] ● The CSI-RS resource set / resource ID information,
[0213] ● The monitoring configuration ID.
[0214] Embodiment 3: Active or deactivate the AI / ML model:
[0215] An embodiment illustrates how the UE and network can activate or deactivate the AI / ML model for specific purposes, such as CSI compression, CSI prediction, beam management, or positioning. The following paragraphs show the AI / ML model Activate / Deactivate signaling and configuration information.
[0216] 1) AI / ML model activation / deactivation information: this information is used to activate or deactivate a specific AI / ML model and is carried by downlink or uplink transmission signaling. The signaling can be UE capability reporting, PUCCH, PUSCH, PDCCH, PDSCH, MAC-CE or RRC signaling. The AI / ML model activation / deactivation information may include one or more of the following elements: AI / ML model ID, activate / deactivate information, activate timing, deactivate timing, timing window.
[0217] ● AI / ML model ID: a unique identifier that can refer to one or more of the following:
[0218] - Logical AI / ML model ID,
[0219] - Global AI / ML model ID, and
[0220] - An identifier of AI / ML model-related information.
[0221] ● Activate / Deactivate information: a value that indicates whether an AI / ML model with the AI / ML model ID is active or inactive. The value can be represented by a Boolean (true or false) , a string (such as "on" or "off" ) , or a bitmap (asequence of bits) . The implementation details are described below:
[0222] ■ Option A-Direct Activate / Deactivate: Each bit may correspond to a specific feature, function, or model ID. If the bit is set to 1, it means that the corresponding item is active; otherwise, it is inactive. For example, if the bit for a scenario (e.g., a scenario-A) is 1, then scenario-Ais activated; otherwise, it is deactivated. The same applies to other items, such as feature, function, and model ID.
[0223] ■ Option B-Hierarchical Activate / Deactivate: Each bit corresponds to a group of features, functions, or model IDs that share a common property or characteristic. The items are bonded by a hierarchical relationship (s) as a group. If the bit is set to 1, it means that all the items in the group are activated to an active state; otherwise, if the bit is set to 0, it means that all the items in the group are deactivated to an inactive state. The hierarchical scheme applies to activation and deactivation. If any hierarchical relationship is present between multiple items, the activation by the hierarchical scheme can reduce the indication overhead. For example, if Scenario #1 contains Feature #1 and Feature #2, and Feature #1 contains Function #1 and Function #2, while Feature #2 contains Function #3 and Function #4, and so on. If Scenario #1 is activated / deactivated, all the features, functions, and model IDs contained in Scenario #1 will be activated / deactivated. If both Scenario #1 and Feature #1 are activated / deactivated, all the function and model IDs contained in Scenario #1 and Feature #1 will be activated / deactivated. The hierarchical scheme applies to activation and deactivation.
[0224] ● Activation time: The network entity (e.g., UE) will activate and start running the AI / ML model after the configured activation time.
[0225] ● Deactivation time: The network entity (e.g., UE) will deactivate and stop running the AI / ML model after the configured deactivation time.
[0226] ● Time window: A time window specifies the fixed duration of time for executing the AI / ML model.
[0227] 2 ) If the “AI / ML model activate / deactivate information” is transmitted on the uplink channel from a UE or to a gNB, the gNB will provide the UE with configuration comprising one or more of following information. The UE may perform the operations in response.
[0228] ● AI / ML model ID: a unique identifier that can refer to one or more of the following:
[0229] - Logical AI / ML model ID,
[0230] - Global AI / ML model ID, and
[0231] - An identifier of AI / ML model-related information.
[0232] The AI / ML model ID is carried in AI / ML model activate / deactivate information.
[0233] ● Reference signal information configuration: The parameters that specify the time-frequency allocation, duration, and activation of the reference signal for the AI / ML model ID. In addition, the reference signals referred may include CSI-RS, SRS, DMRS, PTRS, etc..
[0234] ● CSI reporting information configuration: The configuration for CSI measurement that is required for the AI / ML model ID includes measurement events, channel state information (CSI) reporting intervals, and other related information.
[0235] ● Activation time: The network entity (e.g., UE) will activate and start running the AI / ML model after the configured activation time.
[0236] ● Deactivation time: The network entity (e.g., UE) will deactivate and stop running the AI / ML model after the configured deactivation time.
[0237] ● Time window: A time window specifies the fixed duration of time for executing the AI / ML model.
[0238] 3) If the “AI / ML model active / deactivate information” is transmitted on the downlink channel from a gNB or to a UE, the UE may perform the operations in response. Some examples of the operations are provided in the following.
[0239] ● When receiving an activation signal, the UE initiates the procedures related to the AI / ML model (e.g., an AI / ML model for CSI measurement, CSI reporting, etc. ) , according to the RRC configuration that corresponds to the AI / ML model ID that was configured.
[0240] ● When receiving a deactivation signal, the UE terminates the procedures related to the AI / ML model (e.g., an AI / ML model for CSI measurement, CSI reporting, etc. ) , according to the RRC configuration that corresponds to the AI / ML model ID that was configured.
[0241] Embodiment 4: AI / ML model specific reference signal related procedures:
[0242] For the reference signal configuration, different types of reference signals are used for different purposes. For example, CSI-RS is used for downlink channel state information (CSI) acquisition and reporting. The UE needs to be configured with both the CSI-RS resource configuration and the corresponding CSI reporting configuration. For SRS, which is used for uplink CSI measurement, the UE needs to be configured with the SRS resource configuration. Other reference signals include DMRS, PTRS, SSB, etc. If an AI / ML model is used in the system, the reference signal related procedures would be enhanced as follows. Configuration for AI / ML model reference signal related procedures would include at least one of the following:
[0243] ● AI / ML model ID: The ID can be logical AI / ML model ID, global AI / ML model ID, and / or an identifier of AI / ML model related information.
[0244] ● Reference signal information configuration: The parameters that specify the time-frequency allocation, duration, and activation of the reference signal for the AI / ML model ID.
[0245] ● CSI reporting information configuration: The configuration for CSI measurement that is required for the AI / ML model ID includes measurement events, channel state information (CSI) reporting intervals, and other related information.
[0246] Embodiment 5: Reference signal reconfiguration requirement:
[0247] If the AI / ML model is used for CSI prediction or other purposes. In this case, the network can adjust its transmission behavior based on the feedback from the UE. For example, the UE can send a request for reference signal configuration on the uplink, and the gNB can respond to the request by sending confirmation or reconfiguration via RRC signaling to the UE. The detailed information is shown in the following:
[0248] ● Downlink reference signal configuration request (from UE to gNB) : This request includes one or more of the following information: reference signal indicator (e.g., CSI-RS resource set configuration ID, CSI-RS resource configuration ID, CSI-RS resource set ID, CSI-RS resource ID, CSI-RS antenna port ID, DMRS antenna port ID, etc. ) , timing information, frequency information, antenna port number, and AI / ML model ID.
[0249] ■ Timing information: This information includes one or more of the following: period and duration time.
[0250] ■ Frequency information: This information includes one or more of the following: frequency density and bandwidth.
[0251] ■ AI / ML model ID: This identifier can be one or more of the following: logical AI / ML model ID, global AI / ML model ID, or an identifier of AI / ML model related information. The AI / ML model ID is carried in the AI / ML model activate / deactivate information.
[0252] ● Downlink reference signal configuration (confirm) (from gNB to UE) : The gNB replies to the reference signal configuration request by sending the reference signal configuration to the UE. Downlink reference signal configuration comprises at least one of the following information: AI / ML model ID, confirmation flag, or RRC configuration.
[0253] ■ AI / ML model ID: The AI / ML model ID can be one or more of the following identifiers: logical AI / ML model ID, global AI / ML model ID, or an identifier of AI / ML model related information. The AI / ML model ID is carried in the AI / ML model activate / deactivate information.
[0254] ■ Confirmation flag: The confirmation flag which may a boolean value or a string indicates whether the request is valid or not.
[0255] ■ RRC configuration: The RRC configuration is a set of parameters that configure the radio resource control layer. If the UE receives the RRC configuration associated with the AI / ML model ID, the request is valid, and the UE receives the RRC configuration as a response that responds to the reference signal configuration request; otherwise, the request is invalid. The association between the RRC configuration and the AI / ML model ID can be based on the RRC configuration index being equal to the AI / ML model ID, or on a given timing for the RRC configuration. According to the given timing, the UE determines that the RRC configuration is associated with the AI / ML model ID when the UE receives the RRC configuration within the given timing after sending the reference signal configuration request.
[0256] In an embodiment, a reference signal configuration request is sent from the UE 10 to the base station 20. The reference signal configuration request comprises one or more of the following information: a reference signal indicator, timing information, frequency information, antenna port number, and an AI / ML model ID. The the reference signal indicator comprises one or more of: CSI-RS resource set configuration ID, CSI-RS resource configuration ID, CSI-RS resource set ID, CSI-RS resource ID, CSI-RS antenna port ID, and DMRS antenna port ID.
[0257] In an embodiment, the reference signal configuration / reconfiguration / confirmation is sent from the base station 20 to the UE 10. The reference signal configuration / reconfiguration / confirmation comprises one or more of the following information: an AI / ML model ID, confirmation information, and radio resource control (RRC) configuration.
[0258] FIG. 5 is a block diagram of an example system 700 for wireless communication according to an embodiment of the present disclosure. Embodiments described herein may be implemented into the system using any suitably configured hardware and / or software. FIG. 5 illustrates the system 700 including a radio frequency (RF) circuitry 710, a baseband circuitry 720, a processing unit 730, a memory / storage 740, a display 750, a camera 760, a sensor 770, and an input / output (I / O) interface 780, coupled with each other as illustrated.
[0259] The processing unit 730 may include circuitry, such as, but not limited to, one or more single-core or multi-core processors. The processors may include any combinations of general-purpose processors and dedicated processors, such as graphics processors and application processors. The processors may be coupled with the memory / storage and configured to execute instructions stored in the memory / storage to enable various applications and / or operating systems running on the system.
[0260] The radio control functions may include, but are not limited to, signal modulation, encoding, decoding, radio frequency shifting, etc. In some embodiments, the baseband circuitry may provide for communication compatible with one or more radio technologies. For example, in some embodiments, the baseband circuitry may support communication with 5G NR, LTE, an evolved universal terrestrial radio access network (EUTRAN) and / or other wireless metropolitan area networks (WMAN) , a wireless local area network (WLAN) , a wireless personal area network (WPAN) . Embodiments in which the baseband circuitry is configured to support radio communications of more than one wireless protocol may be referred to as multi-mode baseband circuitry. In various embodiments, the baseband circuitry 720 may include circuitry to operate with signals that are not strictly considered as being in a baseband frequency. For example, in some embodiments, baseband circuitry may include circuitry to operate with signals having an intermediate frequency, which is between a baseband frequency and a radio frequency.
[0261] In various embodiments, the system 700 may be a mobile computing device such as, but not limited to, a laptop computing device, a tablet computing device, a netbook, an ultrabook, a smartphone, etc. In various embodiments, the system may have more or less components, and / or different architectures. Where appropriate, the methods described herein may be implemented as a computer program. The computer program may be stored on a storage medium, such as a non-transitory storage medium.
[0262] The embodiment of the present disclosure is a combination of techniques / processes that can be adopted in 3GPP specification to create an end product.
[0263] If the software function unit is realized and used and sold as a product, it can be stored in a readable storage medium in a computer. Based on this understanding, the technical plan proposed by the present disclosure can be essentially or partially realized as the form of a software product. Or, one part of the technical plan beneficial to the conventional technology can be realized as the form of a software product. The software product in the computer is stored in a storage medium, including a plurality of commands for a computational device (such as a personal computer, a server, or a network device) to run all or some of the steps disclosed by the embodiments of the present disclosure. The storage medium includes a USB disk, a mobile hard disk, a read-only memory (ROM) , a random-access memory (RAM) , a floppy disk, or other kinds of media capable of storing program codes.
[0264] While the present disclosure has been described in connection with what is considered the most practical and preferred embodiments, it is understood that the present disclosure is not limited to the disclosed embodiments but is intended to cover various arrangements made without departing from the scope of the broadest interpretation of the appended claims.
Claims
1.A machine learning system comprising at least one of an artificial intelligence management and orchestrator (AI-MAO) , an agent of AI-MAO, a radio access network side artificial intelligence / machine learning model (RAN-side AI / ML model) ; wherein the AI-MAO, the agent, and the RAN-side AI / ML model are located in one network entity or distributed to different network entities.2.The machine learning system of claim 1, wherein the AI-MAO is deployed in a user equipment (UE) , a base station (BS) , or a network device that operates as an individual management platform separated from the UE and the base station.3.The machine learning system of claim 2, wherein the agent of AI-MAO is deployed as a UE-side agent or a BS-side agent.4.The machine learning system of claim 3, wherein the RAN-side AI / ML model is deployed as a UE-side AI / ML model or a BS-side AI / ML model.5.The machine learning system of claim 1, wherein the AI-MAO manages AI / ML models and metadata of the AI / ML models.6.The machine learning system of claim 5, wherein the RAN-side AI / ML model provides an AI / ML service to assist a device operation, and the agent of AI-MAO communicates with the AI-MAO to perform one of more of requesting, deploying, activating, deactivating, and updating the AI / ML models.7.The machine learning system of claim 6, wherein the agent of AI-MAO sends a model deployment request message to the AI-MAO;upon receiving the request from the agent of AI-MAO, the AI-MAO provides signaling that indicates which AI / ML models are to be used based on information in the model deployment request message;the AI-MAO sends AI / ML model-related information to the agent of AI-MAO; andthe agent of AI-MAO deploys and / or activates one or more AI / ML models indicated by the AI-MAO based on the AI / ML model-related information received.8.The machine learning system of claim 7, wherein the agent of AI-MAO sends at least one type of the following information to AI-MAO: real-time performance, AI / ML model-related information, data collection descriptor associated with the deployed one or more AI / ML models.9.The machine learning system of claim 8, wherein based on the information from the agent of AI-MAO, the AI-MAO sends instructions to the agent of AI-MAO to deactivate or update the one or more AI / ML models.10.A machine learning model processing method for artificial intelligence (AI) / machine learning (ML) , executable in a wireless communication device, comprising:performing an AI / ML-model-related operation for an AI / ML model that is associated with a logical AI / ML model identifier (ID) , wherein the logical AI / ML model ID is derivable from a global AI / ML model ID of the AI / ML model, and / or AI / ML capability information, and / or radio resource control (RRC) configuration.11.The machine learning model processing method of claim 10, wherein the AI / ML-model-related operation comprises signaling of at least one of a model deployment, AI / ML model-related information, AI / ML model activation, AI / ML model deactivation, AI / ML model monitoring, reporting for AI / ML model monitoring, AI / ML model update, AI / ML model reselecting, AI / ML model switching, AI / ML model fallback, or radio resource control (RRC) configuration.12.The machine learning model processing method of claim 10, wherein the logical AI / ML model ID comprises one or more of scenario, function, feature, or local ID of the AI / ML model.13.The machine learning model processing method of claim 10, wherein the global AI / ML model ID of the AI / ML model is mapped to the logical AI / ML model ID of the AI / ML model according to a mapping rule, and the mapping rule comprises direct mapping or hierarchical mapping.14.The machine learning model processing method of claim 13, wherein the direct mapping comprises:a sequential mapping,an inverse mapping, ortransparent mapping by which the logical AI / ML model ID is equal to the local ID.15.The machine learning model processing method of claim 13, wherein in the hierarchical mapping of AI / ML models, the AI / ML models that share the same constraints, are arranged in a specific order based on IDs of each attribute involved in the constraints, the logical AI / ML model ID of the AI / ML model corresponds to a position of the AI / ML model in the arranged AI / ML models, which are arranged either in a sequential or inverse order.16.The machine learning model processing method of claim 10, wherein when the logical AI / ML model ID of the AI / ML model is derived from the AI / ML capability information, ID used for configuring radio resource control (RRC) configuration is used as the logical AI / ML model ID of the AI / ML model which is associated with the AI / ML capability information; orthe logical AI / ML model ID of the AI / ML model is included in RRC configuration and is associated with the AI / ML capability information.17.The machine learning model processing method of claim 16, wherein the AI / ML capability information comprises one or more of:at least one type of AI / ML model related information;at least one global AI / ML model ID; andan AI / ML capability information index.18.The machine learning model processing method of claim 16, wherein the AI / ML capability information is carried in an uplink transmission signal or a downlink transmission signal;the uplink transmission signal includes one or more of UE capability reporting, physical uplink control channel (PUCCH) , or physical uplink shared channel (PUSCH) ; andthe downlink transmission signal includes one or more of RRC message, medium access control -control element (MAC-CE) , downlink control information (DCI) , physical downlink shared channel (PDSCH) , broadcast signaling, a system information block (SIB) , or a master information block (MIB) .19.The machine learning model processing method of claim 10, wherein when the logical AI / ML model ID of the AI / ML model is derived from the RRC configuration, an ID of the RRC configuration is used as the logical AI / ML model ID of the AI / ML model, and the ID of the RRC configuration comprises one or more of: measurement configuration ID, CSI measurement configuration ID, SRS resource set ID, SRS resource configuration ID, CSI-RS resource set ID, CSI-RS resource ID, and reporting configuration ID.20.The machine learning model processing method of claim 10, wherein the AI / ML-model-related operation comprises AI / ML model activation and / or AI / ML model deactivation; andAI / ML model activate / deactivate information for the AI / ML model activation or AI / ML model deactivation comprises one or more of:an AI / ML model ID,activate / deactivate information,activate timing,deactivate timing, anda timing window.21.The machine learning model processing method of claim 20, wherein the AI / ML model activate / deactivate information is transmitted in an uplink transmission or a downlink transmission.22.The machine learning model processing method of claim 10, wherein when the AI / ML model is used for configuring a reference signal related configuration, the reference signal related configuration comprises one or more of:an AI / ML model ID,reference signal information configuration, andCSI reporting information configuration.23.The machine learning model processing method of claim 10, wherein a reference signal configuration request is sent from a user equipment (UE) to a base station; the reference signal configuration request comprises one or more of the following information: a reference signal indicator, timing information, frequency information, antenna port number, and an AI / ML model ID.24.The machine learning model processing method of claim 23, wherein the reference signal indicator comprises one or more of CSI-RS resource set configuration ID, CSI-RS resource configuration ID, CSI-RS resource set ID, CSI-RS resource ID, CSI-RS antenna port ID, and DMRS antenna port ID.25.The machine learning model processing method of claim 23, wherein reference signal configuration / reconfiguration / confirmation is sent from the base station to the UE;the reference signal configuration / reconfiguration / confirmation comprises one or more of the following information:an AI / ML model ID,confirmation information, andradio resource control (RRC) configuration.26.A user equipment (UE) comprising:a processor configured to call and run a computer program stored in a memory, to cause a device in which the processor is installed to execute the method of any of claims 10 to 25.27.A chip, comprising:a processor, configured to call and run a computer program stored in a memory, to cause a device in which the chip is installed to execute the method of any of claims 10 to 25.28.A computer-readable storage medium, in which a computer program is stored, wherein the computer program causes a computer to execute the method of any of claims 10 to 25.29.A computer program product, comprising a computer program, wherein the computer program causes a computer to execute the method of any of claims 10 to 25.30.A computer program, wherein the computer program causes a computer to execute the method of any of claims 10 to 25.
Citation Information
Patent Citations
Dynamic accuracy-based deployment and monitoring of machine learning models in provider networks
CN111527499A
Center verification system of joint learning engine
CN114764475A
Machine learning model reasoning method and device
CN114912632A
Synchronization of artificial intelligence based microservices
US20220272794A1
Systems and methods to optimize training of ai / ML models and algorithms
WO2023017102A1
Cited By
Artificial intelligence (AI) edge service management platform
EP4841952A1
Secure multi-agent system for privacy-preserving distributed computation
US12316753B1
Data processing orchestrator utilizing semantic type inference and privacy preservation
US12437113B1
Reporting of performance monitoring
WO2025236722A1
Method and system for configuring adaptive model-based functionalities
WO2026209305A1
