Use of Artificial Intelligence / Machine Learning Model Identifiers

Abstract

Embodiments of the present disclosure relate to the use of artificial intelligence / machine learning (AI / ML) model identifiers (IDs). A terminal device sends a request to a core network device to register or identify an AI / ML model, the request including at least one AI / ML model capability of the terminal device. The terminal device receives a response to the AI / ML model registration or identification from the core network device, the response including a list of approved or identified AI / ML model IDs and at least one AI / ML model distribution priority. The solution for using AI / ML model identifiers (IDs) provided in the present disclosure provides a signaling procedure that must be defined between the terminal device and the network to determine the validity of a particular AI / ML model and allows the network to make additional inquiries to the terminal device regarding the capabilities of the AI / ML model.

Description

[Technical Field]

[0001] This application claims priority to Finnish application number 20235137, filed February 10, 2023, the contents of which are incorporated herein by reference in their entirety.

[0002] Various exemplary embodiments relate generally to the field of communications, and more particularly to devices, methods, apparatus, and computer-readable storage media relating to the use of artificial intelligence / machine learning (AI / ML) model identifiers (IDs). [Background technology]

[0003] As communications technology evolves, AI / ML models for the New Radio (NR) air interface are being researched. The 3rd Generation Partnership Project (3GPP®) Release 18 (Rel-18) Study Item (SI) explores the benefits of extending the air interface with the capability to support AI / ML-based algorithms for improved performance and / or reduced complexity and overhead.

[0004] Initial use cases targeted in Rel-18 SI include enhanced CSI feedback (e.g., reduced overhead, improved accuracy, prediction), beam management (e.g., reduced overhead and latency through beam prediction in time or spatial domains, improved beam selection accuracy), and improved location accuracy. Use cases for AI / ML approaches must be diverse enough to support diverse requirements at the next generation Node B (gNB) and User Equipment (UE) cooperation level, which defines the combination of ML models applied at least at the UE and / or gNB. Summary of the Invention

[0005] Generally, exemplary embodiments of the present disclosure provide a device, a method, an apparatus, and a computer-readable storage medium for using AI / ML model ID. Specifically, the solution can provide signaling procedures that need to be defined between a terminal device and a network.

[0006] In a first aspect, a terminal device is provided, which may include one or more transceivers and one or more processors communicatively coupled to the one or more transceivers, wherein the one or more processors are configured to cause the terminal device to: send a request for registration or identification of an AI / ML model to a core network device, the request including at least one AI / ML model capability of the terminal device; and receive a response to the registration or identification of the AI / ML model from the core network device, the response including a list of approved or identified AI / ML model IDs and at least one AI / ML model distribution priority.

[0007] In a second aspect, an access network device is provided, which may include one or more transceivers and one or more processors communicatively coupled to the one or more transceivers, wherein the one or more processors are configured to cause the access network device to receive, from a core network device, a message including a list of approved or identified AI / ML model IDs and contexts for a terminal device, and to store the approved or identified AI / ML model IDs and contexts.

[0008] In a third aspect, a first core network device is provided. The first core network device may include one or more transceivers and one or more processors communicatively coupled to the one or more transceivers, wherein the one or more processors are configured to cause the first core network device to: receive, from a terminal device, a request for registration or identification of an AI / ML model, where the request includes at least one AI / ML model capability of the terminal device; determine a list of AI / ML model IDs for the terminal device based on the at least one AI / ML model capability; send a validation request for the list of AI / ML model IDs to a second core network device; and receive, from the second core network device, a response including a list of approved or identified AI / ML model IDs and contexts.

[0009] In a fourth aspect, a second core network device is provided. The second core network device may include one or more transceivers and one or more processors communicatively coupled to the one or more transceivers, where the one or more processors are configured to cause the second core network device to: receive, from the first core network device, a request to verify a list of AI / ML model IDs associated with a terminal device, determine a list of approved or identified AI / ML model IDs and contexts for the terminal device by verifying the list of AI / ML model IDs based on a database of verified AI / ML models for the terminal device, and send, to the first core network device, a response including the list of approved or identified AI / ML model IDs and contexts.

[0010] In a fifth aspect, a method implemented in a terminal device is provided. The method may include sending a request for registration or identification of an AI / ML model to a core network device, the request including at least one AI / ML model capability of the terminal device, and receiving a response to the registration or identification of the AI / ML model from the core network device, the response including a list of approved or identified identifiers (IDs) of the AI / ML models and at least one AI / ML model distribution priority.

[0011] In a sixth aspect, a method implemented in an access network device is provided, and may include receiving, from a core network device, a message including a list of approved or identified AI / ML model IDs and contexts for a terminal device, and storing the approved or identified AI / ML model IDs and contexts.

[0012] In a seventh aspect, a method is provided implemented in a first core network device. The method may include receiving, from a terminal device, a request for registration or identification of an AI / ML model, the request including at least one AI / ML model capability of the terminal device, determining a list of AI / ML model IDs for the terminal device based on the at least one AI / ML model capability, sending a validation request for the list of AI / ML model IDs to a second core network device, and receiving, from the second core network device, a response including a list of approved or identified AI / ML model IDs and contexts.

[0013] In an eighth aspect, a method is provided that is implemented in a second core network device. The method may include receiving, from a first core network device, a request to verify a list of AI / ML model IDs associated with a terminal device, determining, at the second core network device, a list of approved or identified AI / ML model IDs and contexts for the terminal device by verifying the list of AI / ML model IDs based on a database of verified AI / ML models for the terminal device, and transmitting a response to the first core network device that includes the list of approved or identified AI / ML model IDs and contexts.

[0014] In a ninth aspect, an apparatus is provided for a terminal device, the apparatus may include: means for sending a request for registration or identification of an AI / ML model to a core network device, the request including at least one AI / ML model capability of the terminal device, and means for receiving, from the core network device, a response to the registration or identification of the AI / ML model, the response including an approved or identified list of AI / ML model identifiers and at least one AI / ML model distribution priority.

[0015] In a tenth aspect, an apparatus of an access network apparatus is provided, which may include: means for receiving, from a core network apparatus, a message including a list of approved or identified AI / ML model IDs and contexts for a terminal device; and means for storing the approved or identified AI / ML model IDs and contexts.

[0016] In an eleventh aspect, an apparatus is provided for a first core network device. The apparatus may include means for receiving a request for registration or identification of an AI / ML model from a terminal device, the request including at least one AI / ML model capability of the terminal device, means for determining a list of AI / ML model IDs for the terminal device based on the at least one AI / ML model capability, means for sending a validation request for the list of AI / ML model IDs to a second core network device, and means for receiving a response from the second core network device including a list of approved or identified AI / ML model IDs and contexts.

[0017] In a twelfth aspect, an apparatus is provided for a second core network device, which may include: means for receiving a validation request for a list of AI / ML model IDs associated with a terminal device from a first core network device; means for determining, at the second core network device, a list of approved or identified AI / ML model IDs and contexts for the terminal device by validating the list of AI / ML model IDs based on a database of validated AI / ML models for the terminal device; and means for transmitting a response to the first core network device, the response including the list of approved or identified AI / ML model IDs and contexts.

[0018] In a thirteenth aspect, there is provided a non-transitory computer readable medium comprising program instructions to cause an apparatus to perform at least a method according to the fifth or eighth aspects.

[0019] In a fourteenth aspect, a computer program is provided that includes instructions that, when executed by an apparatus, cause the apparatus to at least: send a request for registration or identification of an AI / ML model to a core network apparatus, the request including at least one AI / ML model capability of the terminal apparatus; and receive a response to the registration or identification of the AI / ML model from the core network apparatus, the response including an approved or identified list of AI / ML model IDs and at least one AI / ML model distribution priority.

[0020] In a fifteenth aspect, a computer program is provided that includes instructions that, when executed by an apparatus, cause the apparatus to at least receive a message from a core network apparatus that includes a list of approved or identified AI / ML model IDs and contexts of a terminal apparatus, and store the approved or identified AI / ML model IDs and contexts.

[0021] In a sixteenth aspect, a computer program is provided that includes instructions that, when executed by an apparatus, cause the apparatus to at least receive a request for registration or identification of an AI / ML model from a terminal device, the request including at least one AI / ML model capability of the terminal device; determine a list of AI / ML model IDs for the terminal device based on the at least one AI / ML model capability; send a validation request for the list of AI / ML model IDs to a second core network device; and receive a response from the second core network device that includes a list of approved or identified AI / ML model IDs and contexts.

[0022] In a seventeenth aspect, a computer program is provided that, when executed by an apparatus, causes the apparatus to at least receive, from a first core network apparatus, a request to verify a list of AI / ML model IDs associated with a terminal apparatus; determine a list of approved or identified AI / ML model IDs and contexts for the terminal apparatus by verifying the list of AI / ML model IDs based on a database of verified AI / ML models for the terminal apparatus; and send a response to the first core network apparatus including the list of approved or identified AI / ML model IDs and contexts.

[0023] In an eighteenth aspect, a terminal device is provided, the terminal device comprising: a transmitting circuit configured to perform transmitting a request for registration or identification of an AI / ML model to a core network device, the request including at least one AI / ML model capability of the terminal device; and a receiving circuit configured to perform receiving a response to the registration or identification of the AI / ML model from the core network device, the response including an approved or identified list of AI / ML model IDs and at least one AI / ML model distribution priority.

[0024] In a nineteenth aspect, an access network device is provided, the network device comprising: a receiving circuit configured to receive a message including a list of approved or identified AI / ML model IDs and a context of a terminal device from a core network device; and a storage circuit configured to store the approved or identified AI / ML model IDs and the context.

[0025] In a twentieth aspect, a first core network device is provided. The network device includes: a receiving circuit configured to cause receiving to receive a request for registration or identification of an AI / ML model from a terminal device, the request including at least one AI / ML model capability of the terminal device; a determining circuit configured to determine a list of AI / ML model IDs for the terminal device based on the at least one AI / ML model capability; a transmitting circuit configured to send a validation request for the list of AI / ML model IDs to a second core network device; and a receiving circuit configured to receive from the second core network device a response including a list of approved or identified AI / ML model IDs and contexts.

[0026] In a 21st aspect, a second core network device is provided, the network device comprising: a receiving circuit configured to receive a verification request for a list of AI / ML model IDs associated with a terminal device from the first core network device, a determining circuit configured to determine a list of approved or identified AI / ML model IDs and contexts for the terminal device by verifying the list of AI / ML model IDs based on a database of verified AI / ML models for the terminal device, and a transmitting circuit configured to transmit a response to the first core network device including the list of approved or identified AI / ML model IDs and contexts.

[0027] The Abstract is not intended to identify key features or essential features of the embodiments of the present disclosure, nor is it intended to be used to limit the scope of the present disclosure. Other features of the present disclosure will be readily apparent from the following description. [Brief explanation of the drawings]

[0028] Exemplary embodiments will now be described with reference to the accompanying drawings. [Figure 1A] FIG. 1A illustrates an example of a network environment in which exemplary embodiments of the present disclosure may be implemented. [Figure 1B] FIG. 1B illustrates a signaling architecture of an AI / ML function in which exemplary embodiments of the present disclosure may be implemented. [Figure 2] FIG. 2 illustrates an example of user equipment capability transfer related to an embodiment of the present disclosure. [Figure 3] FIG. 3 illustrates an example of an AI / ML model ID format for user equipment associated with an exemplary embodiment of the present disclosure. [Figure 4] FIG. 4 illustrates an example signaling process for AI / ML model identification and feature processing, according to some embodiments of the present disclosure. [Figure 5A] FIG. 5A illustrates another example signaling process for AI / ML model identification and feature processing, according to some embodiments of the present disclosure. [Figure 5B] FIG. 5B illustrates yet another example signaling process for AI / ML model identification and feature processing, according to some embodiments of the present disclosure. [Figure 6] FIG. 6 is a flowchart illustrating an example of a method implemented in a terminal device, according to some example embodiments of the present disclosure. [Figure 7] FIG. 7 is a flowchart illustrating an example of a method implemented in an access network device, according to some example embodiments of the present disclosure. [Figure 8] FIG. 8 illustrates an example flowchart of a method implemented in a first core network device according to some example embodiments of the present disclosure. [Figure 9] FIG. 9 illustrates an example flowchart of a method implemented in a second core network device according to some example embodiments of the present disclosure. [Figure 10] FIG. 10 illustrates an exemplary block diagram of an apparatus suitable for implementing embodiments of the present disclosure. [Figure 11] 11 shows a block diagram illustrating an example of a computer-readable medium according to some embodiments of the present disclosure. Throughout the drawings, the same or similar reference numbers represent the same or similar elements. DETAILED DESCRIPTION OF THE INVENTION

[0029] The principles of the present disclosure will be described with reference to some exemplary embodiments. It should be understood that these embodiments are provided merely to aid in the understanding of the present disclosure and are not intended to limit the scope of the present disclosure. The present disclosure can be implemented by various embodiments other than those described below.

[0030] In the following description and claims, unless otherwise defined, all technical and scientific terms used herein have the meaning commonly understood by one of ordinary skill in the art to which this invention pertains.

[0031] References in this disclosure to "one embodiment," "embodiment," "exemplary embodiment," and the like indicate that the described embodiment includes a particular feature, structure, or characteristic, but do not require that all embodiments include the particular feature, structure, or characteristic. Moreover, such references do not necessarily refer to the same embodiment. Furthermore, if a particular feature, structure, or characteristic is described in connection with an embodiment, it is believed to be within the knowledge of one skilled in the art to implement that feature, structure, or characteristic in connection with other embodiments, whether or not explicitly described.

[0032] Although terms such as "first" and "second" are used herein to describe various elements, it should be understood that these elements are not limited by these terms. These terms are used only to distinguish one element from another. For example, a first element could be referred to as a second element, and similarly, a second element could be referred to as a first element, without departing from the scope of the exemplary embodiments. As used herein, the term "and / or" includes all combinations of one or more of the listed terms.

[0033] The terms used herein are for the purpose of describing particular embodiments and are not intended to limit example embodiments. As used herein, the singular forms "a," "an," and "the" are intended to include the plural forms unless the context clearly dictates otherwise. Furthermore, it should be understood that the terms "comprises," "comprising," "has," "having," "includes," and / or "including" as used herein identify the presence of stated features, elements, and / or components, but do not exclude the presence or addition of one or more other features, elements, components, and / or combinations thereof. As used herein, the terms "at least one of: " and "at least one " and similar expressions, when a list of two or more elements is connected by "and" or "or," mean at least any element, or at least two or more elements, or at least all elements.

[0034] In this application, the term "circuit" means (a) hardware-only circuit implementations (e.g., implementations using only analog and / or digital circuitry); and (b) a combination of hardware circuitry and software (where applicable); (i) a combination of analog and / or digital hardware circuitry and software / firmware; and (ii) any portion of software (including digital signal processors), hardware processors with software and memory that work together to cause a device, such as a mobile phone or server, to perform various functions; and (c) A hardware circuit or processor that includes a microprocessor or part of a microprocessor that requires software (e.g., firmware), where the software may be absent if not necessary for operation; This refers to one or more, or all, of the following:

[0035] This definition of circuit applies to all uses of the term in this application, including the claims. Furthermore, as used herein, the term circuit also covers a simple hardware circuit or processor (or processors) or part of a hardware circuit or processor and its associated software and / or firmware implementation. The term circuit also covers, for example, baseband or processor integrated circuits for mobile devices, or similar integrated circuits found on servers, cellular network equipment, and other computing or network devices, when applied to certain claim elements.

[0036] As used herein, a "communication network" refers to a network conforming to an appropriate communication standard, such as Long Term Evolution (LTE), LTE-Advanced (LTE-A), Wideband Code Division Multiple Access (WCDMA), High-Speed ​​Packet Access (HSPA), or Narrowband Internet of Things (NB-IoT). Furthermore, communications between terminal devices and network devices within the communication network may be performed according to any appropriate generation of communication protocols, including, but not limited to, third-generation (3G), fourth-generation (4G), 4.5G, fifth-generation (5G) communication protocols, and / or beyond. Embodiments of the present disclosure may be applied to various communication systems. With the rapid development of communication technologies, it is natural that future communication technologies and systems capable of embodying the present disclosure will emerge. Therefore, the scope of the present disclosure is not limited to the above-mentioned systems.

[0037] The term "network device" as used herein refers to a node in a communication network through which a terminal device accesses the network and receives services from it. The network device may refer to, for example, a base station (BS) or an access point (AP). Depending on the applied terminology and technology, it may be referred to as a Node B (NodeB or NB), evolved Node B (eNodeB or eNB), NR NB (also referred to as gNB), remote radio unit (RRU), radio header (RH), remote radio head (RRH), relay, femto, pico, or other low-power node.

[0038] The term "terminal" refers to any terminal device capable of wireless communication. By way of example, a terminal device may also be referred to as communication equipment, user equipment (UE), subscriber station (SS), portable subscriber station, mobile station (MS), or access terminal (AT). Terminal devices include mobile phones, cellular phones, smartphones, VoIP phones, wireless local loop phones, tablets, wearable terminal devices, PDAs, portable computers, desktop computers, image capture terminal devices such as digital cameras, gaming terminal devices, music storage and playback devices, in-vehicle wireless terminal devices, wireless endpoints, mobile stations, laptop embedded devices (LEEs), laptop mounted devices (LMEs), USB dongles, smart devices, wireless customer premise equipment (CPEs), Internet of Things (IoT) devices, watches or other wearables, head-mounted displays (HMDs), vehicles, drones, medical devices and applications (e.g., remote surgery), industrial devices and applications (e.g., robots and / or other wireless devices operating in the context of industrial and / or automated processing chains), consumer electronics devices, devices operating on commercial networks, relay nodes, integrated access and backhaul (IAB) nodes, and / or industrial wireless networks. In the following description, the terms "terminal device," "communications equipment," "terminal," "user equipment," and "UE" may be used interchangeably.

[0039] As used herein, "resource," "transmission resource," "resource block," "physical resource block" (PRB), "uplink (UL) resource," or "downlink (DL) resource" refers to any resource for communication between a terminal device and a network device, such as a time domain resource, a frequency domain resource, a space domain resource, a code domain resource, a combination of multiple domains, or any other resource that enables communication. Hereinafter, a time domain resource (e.g., a subframe) is used as an example of a transmission resource to describe some exemplary embodiments of the present disclosure. It should be noted that the exemplary embodiments of the present disclosure are equally applicable to other resources in other domains.

[0040] The terms artificial intelligence and / or machine learning (AI / ML) refer to software-implemented methods based on mathematical algorithms or models that provide inference capabilities. Such models are typically mathematical algorithms that are trained with information to reproduce the decisions made by experts when provided with the same information. According to some embodiments, AI / ML functions can also provide data analysis. For example, an AI / ML training function associated with a model acquires data, passes the data through the AI / ML model, derives an associated loss, and adjusts the parameterization of the AI / ML model based on the calculated loss. Training methods include supervised learning, unsupervised learning, and reinforcement learning, and training can be performed offline or continuously. The inference function can be one of several known categories, such as regression-based, clustering or association-based, or reward-based behavior, and an appropriate training method is applied.

[0041] Examples of applications of AI and / or ML include, but are not limited to, use cases for speech recognition, image processing / computer vision, natural language processing, information retrieval, personalization and recommendations, robotics, data analytics including predictive and prescriptive analytics, and the design and / or planning and / or optimization and / or configuration and / or control and / or management of communication systems and / or networks.

[0042] Use cases include, but are not limited to: Use cases related to the physical layer of communication networks (modulation, coding, decoding, signal detection, channel estimation, prediction, compression, interference mitigation, etc.), Use cases related to the medium access control layer of communication networks (e.g., multiple access and resource allocation (e.g., power control, scheduling, spectrum management)); -Channel modeling, -Network optimization, - Cell capacity estimation in cellular networks, -routing, -Resource management, -Data traffic management, -security and anomaly detection, -Root cause analysis, -Transport protocol design and optimization, -User / network / application behavior analysis / prediction, -Transport layer congestion control, -User experience modeling and optimization, -User mobility and location management, -Network slicing, network virtualization, and software-defined networking, - Nonlinear impairment compensation in optical networks (e.g., visible light communication, fiber optic communication, and fiber-wireless converged networks), and -Transmission quality estimation and optical performance monitoring in optical networks.

[0043] An AI / ML entity refers to a network entity that contains one or more AI and / or ML capabilities. Examples of network entities include, but are not limited to: Radio access network entities such as base stations (e.g., eNodeBs in LTE and LTE-Advanced networks, gNodeBs used in 5G networks, and mobile phone base stations such as femtocells used in homes and business centers); -Relay stations, - a control station (e.g., a radio network controller, a base station controller, a network switching subsystem), - Access points in a local area network or ad-hoc network, Gateways and Radio Access Network Entities, - network management entities (e.g., operations, administration and monitoring (OAM) entities); -Network automation systems, - Distributed analytical entities (e.g., self-autonomous systems (D-SONs)), Network functions (e.g., Network Data Analysis Functions (NWDAF) as defined in current 3GPP standards), User Equipment (UE).

[0044] The goal of the Rel-18 SI is to lay the foundation for future air interface use cases leveraging AI / ML technologies. AI / ML use cases should be evaluated using developed methods and defined KPIs to assess their benefits and potential impact on specifications, including the PHY and protocol layers. One of the expected outcomes of the SI is that "AI / ML approaches in selected sub-use cases should be diverse enough to support diverse requirements at the gNB-UE collaboration level."

[0045] Note that additional use cases may be considered during the "AI / ML for Air Interface" Work Item (WI) phase. Since Rel-18, various use cases and applications of AI / ML in gNBs and UEs have been proposed. The goal is to explore the benefits of enhancing the air interface with features that enable improved performance or reduced complexity / overhead of AI / ML-based algorithms. Performance improvements here depend on the use case being considered and may include, for example, improved throughput, robustness, accuracy, or reliability. The goal is to explore enough use cases to identify a common AI / ML framework that includes functional requirements for the AI / ML architecture. It is also necessary to identify areas where AI / ML could improve the performance of air interface functions. It is necessary to evaluate the impact on specifications and develop a comprehensive understanding of the requirements necessary to apply AI / ML technologies to the air interface.

[0046] Discussions in RAN1 / RAN2 indicate that UEs that support ML models to extend specific features of the specification (e.g., beam management, CSI reporting) use an AI / ML model ID (also referred to as ML model ID) to identify the feature being used (e.g., CSI compression, CSI prediction, time-domain beam prediction, spatial-domain beam prediction, ML-based localization, etc.). However, it is unclear how the network knows or identifies which specific AI / ML model ID is valid. Validity can mean many things, such as the underlying ML model being valid, available for use by the UE, tested, verified, certified, authorized for use, and configurable for the UE. This is even more important because the network cannot understand the context of the ML model (which is typically a deep neural network, and the network cannot be expected to understand the architecture or implementation details, as these depend on the ML implementation choice, with hundreds of different options and variations, and the details are typically private and cannot be made public).

[0047] Once the network can reliably determine "valid" AI / ML model IDs, it is unclear how to obtain these corresponding capabilities from the UE so that the UE can be configured to take them into account. Furthermore, it is unclear how the network would transfer a representation of the ML model architecture (which might be encoded as a series of OCTETS using well-known tools such as the ONNX format) to the UE.

[0048] It may be necessary to define a signaling procedure between the UE and the network to determine the validity of specific ML models and for the network to query the UE about the capabilities associated with these ML models.

[0049] This application defines the following novel procedures: 1) storing ML models in newly defined network logical nodes, 2) verifying and activating AI / ML model IDs in UE and gNB, and 3) UE capability model query and configuration (which can also be achieved through model capability query and configuration).

[0050] Therefore, the present disclosure proposes AI / ML model identification and capability handling for UE(s). According to an embodiment of the present disclosure, a terminal device sends a request for registering or identifying an AI / ML model to a core network device. The request includes at least one AI / ML model capability of the terminal device. The terminal device receives a response to the AI / ML model registration or identification from the core network device. The response includes a list of approved or identified AI / ML model IDs.

[0051] It should be understood that the above steps operate partly in conjunction with one another and partly independently in the operational sequence described in the next section.

[0052] The AI / ML model ID usage solution of the present disclosure provides a signaling procedure that needs to be defined between a terminal device and a network to determine the validity of a given AI / ML model, and allows the network to further query the terminal device about capabilities related to the AI / ML model. The principles and some exemplary embodiments of the present disclosure are described in detail below with reference to the accompanying drawings.

[0053] For illustrative purposes, the principles and exemplary embodiments of the present disclosure relating to PHR operation are described below with reference to Figures 1A to 11. However, it should be noted that these embodiments are provided to enable those skilled in the art to understand the inventive concepts of the present disclosure and to implement the solutions proposed in the present disclosure, and are not intended to limit the scope of the present application in any way.

[0054] 1A illustrates an example of a network environment 100A in which exemplary embodiments of the present disclosure may be implemented. The network environment 100A, which is part of a communications network, includes a terminal device 110, an access network device 120, a first core network device 130, and a second core network device 140.

[0055] 1A, the terminal device 110 is also referred to as user equipment 110 or UE 110. The access network device 120 is also referred to as gNB 120. The first core network device 130 is also referred to as access and mobility management function (AMF) 130. The second core network device 140 is also referred to as user equipment machine learning function management function (UMLCMF) 140. In some embodiments, the AMF 130 and the UMLCMF 140 are logical entities. Therefore, the AMF 130 and the UMLCMF 140 may be combined, or the role of the UMLCMF 140 may be implemented in the AMF 130.

[0056] In some embodiments, the UMLCMF 140 may be required to store all AI / ML model IDs along with the corresponding AI / ML model context. The context may include, for example, metadata indicating the high-level details of the model (architecture, number of layers, etc.) and the applicable radio, configuration, and parameter conditions under which the model was trained.

[0057] In some embodiments, the UMLCMF 140 may include AI / ML model data, such as a model container that holds data corresponding to a given ML model.

[0058] 1B illustrates an AI / ML capability signaling architecture in which exemplary embodiments of the present disclosure may be implemented. A network environment 100B that is part of a communications network includes a UE 150, a radio access network (RAN) 160, an AMF 170, a UMLCMF 180, and a machine learning database (MLDB) 190.

[0059] 1B, the UE 150 corresponds to the terminal device 110. The RAN 160 corresponds to the access network device 120. The AMF 170 corresponds to the first core network device 130. The UMLCMF 180 corresponds to the second core network device 140.

[0060] In some embodiments, the MLDB 190 may be an ML external third-party database or an ML external operator database. AI / ML models may be stored in the MLDB 190. An operator can send the AI / ML models to the UMLCMF 180. The stored AI / ML model content and context may be visualized as a string ranging from several hundred kilobytes to several megabytes, depending on the manufacturer, version, and capabilities of the UE(s). The UMLCMF 180 uses an interface to link with the MLDB 190, allowing the operator to control the valid AI / ML model IDs used in a specified network. Nx1 refers to the service-based interface provided by the UMLCMF 180. Nx2 refers to the service-based interface provided by the MLDB 190.

[0061] FIG. 2 illustrates an example of user equipment capability transfer 200 associated with an exemplary embodiment of the present disclosure. FIG. 2 illustrates how a UE compiles and transfers its UE capability information upon receiving a UE Capability Enquiry from the network. The network 204 may initiate this procedure for a UE 202 in RRC_CONNECTED if UE radio access capability information or additional UE radio access capability information is required. The network 204 may acquire UE capabilities after activation of access stratum (AS) security. The network 204 cannot transfer UE capabilities acquired before activation of AS security to the CN.

[0062] Table 1 shows some of the terms used in this disclosure. [Table 1-1] [Table 1-2]

[0063] Table 2 shows the working hypotheses that consider “proprietary models” and “open format models” as two different model format categories in the discussion of RAN1. [Table 2]

[0064] From the perspective of the RAN1 discussion, RAN1 assumes the following: 1) proprietary models cannot be mutually recognized between vendors, and model design information is hidden from other vendors when shared, and 2) open format models can be mutually recognized between vendors, and model design information is not hidden from other vendors when shared.

[0065] Table 3 shows other working hypotheses that explain the two terms used herein. [Table 3]

[0066] At the RAN2 meeting, some agreement was reached. Table 4 shows the initial hypotheses made at the last RAN2#119-bis-e meeting. [Table 4]

[0067] Table 5 shows the initial assumptions made in the previous RAN2#120 meeting. [Table 5]

[0068] FIG. 3 illustrates an example of a user equipment AI / ML model ID format 300 associated with an exemplary embodiment of the present disclosure. As shown in FIG. 3, there are three fields. UF 310 indicates use case-specific information regarding the capabilities of the ML model (e.g., beam management, CSI compression, localization, mobility enhancement, power saving, etc.). The length of UF 310 may be a single hexadecimal digit.

[0069] Vendor ID 320 is another field. Vendor ID 320 is an identifier of the UE manufacturer. It is defined in the Private Enterprise Number values ​​issued by the Internet Assigned Numbers Authority (IANA) as the Private Enterprise Number Administrator and maintained at https: / / www.iana.org / assignments / enterprise-numbers / enterprise-numbers. The length of Vendor ID 320 is 8 hexadecimal digits. Version ID 330 means the current version ID set in UMLCMF180. The length of Version ID 330 is 2 hexadecimal digits.

[0070] 4 illustrates an example signaling process 400 for AI / ML model identification and capability processing in accordance with some embodiments of the present disclosure. As shown, terminal device 110 transmits 430 a request 410 for AI / ML model registration or identification to first core network device 130. Request 410 includes at least one AI / ML model capability of terminal device 110.

[0071] The first core network device 130 receives the request 410 (432). The first core network device 130 then determines a list of AI / ML model IDs for the terminal device 110 based on the at least one AI / ML model function (412). The first network device 130 sends a verification request 414 for the list of AI / ML model IDs to the second core network device 140 (434). The second core network device 140 receives the request 414 from the first core network device 130 (436).

[0072] The second core network device 140 determines a list of approved or identified AI / ML model IDs and contexts for the terminal device by verifying the list of AI / ML model IDs based on a database of verified AI / ML models for the terminal device (416). The second core network device 140 sends a response 418 including the list of approved or identified AI / ML model IDs and contexts to the first core network device 130 (440). The first core network device 130 receives the response 418 from the second core network device 140 (438).

[0073] The second core network device 140 transmits a response 420 to the terminal device 110 for the registration or identification of the AI / ML model (444). The response 420 includes a list of approved or identified AI / ML model IDs and at least one AI / ML model distribution priority. The terminal device 110 receives the response 420 (442).

[0074] The first core network device 130 sends (448) a message 422 including a list of approved or identified AI / ML model IDs and the context of the terminal device to the access network device 120. The access network device 120 receives (446) the message 422 from the first core network device 130. The access network device 120 stores (424) the approved or identified AI / ML model IDs and the context.

[0075] 4 can be implemented to provide for the use of an AI / ML model ID. In this way, a signaling procedure can be defined between the terminal device and the network to determine the validity of a particular AI / ML model, allowing the network to further query the terminal device about capabilities related to the AI / ML model.

[0076] FIG. 5A illustrates another example signaling process 500A for AI / ML model identification and capability processing in accordance with some embodiments of the present disclosure. The example signaling process 500A of FIG. 5A can be considered an example of the signaling process 400 of FIG. 4. Accordingly, the UE 502 of FIG. 5A is an example of the terminal device 110 of FIG. 4. The gNB 504 of FIG. 5A is an example of the access network device 120 of FIG. 4. The AMF 506 of FIG. 5A is an example of the first core network device 130 of FIG. 4. The UMLCMF 508 of FIG. 5A is an example of the second core network device 140 of FIG. 4. FIG. 5A describes two scenarios. Scenario 1 describes how AI / ML models are stored by a network entity. Scenario 2 describes registration / identification of UE-specific AI / ML models.

[0077] Scenario 1 is shown in dashed box 592. At 566B, MLDB 510 sends a store / delete AI / ML model request 512 to UMLCMF 508. At 566A, UMLCMF 508 receives request 512. At 514, UMLCMF 508 stores or deletes the AI / ML model according to request 512. At 568A, UMLCMF 508 sends a store / delete AI / ML model response 516 to MLDB 510.

[0078] In some embodiments, the UMLCMF 508 may receive requests for AI / ML models from an entity that maintains a database of validated ML models. The database may store UE-side models or the UE portion of two-sided models that are trained offline (offline model updates are also possible), and each trained model may be referenced by an AI / ML model ID along with the corresponding ML model context and ML model data. Here, the AI / ML model data (also referred to as ML-Model-Content) may optionally be sent to the UMLCMF 508. The UMLCMF 508 may store a list of the received AI / ML model IDs, ML-Model-Context, and ML-Model-Content. A response / acknowledgement confirming successful model reception is sent from the UMLCMF 508 to the MLDB 510.

[0079] In some embodiments, for a CSI compression use case, the network vendor and the UE vendor develop multiple two-sided ML models that take into account different deployment environments, parameters, and configurations. These models are stored in an operator database to ensure controllability of the models used on the air interface. Before using these models for CSI compression, the network entity UMLCMF retrieves the latest set of models (at least the UE-side portion of the two-sided models) from the operator-managed dataset using the procedure described above in dashed box 592.

[0080] Scenario 2 is shown in dashed box 594. The purpose of Scenario 2 is to allow the network to ensure that a given set of AI / ML model IDs is activated in the UE and also known to the gNB. At 518, the UE 502 may have established an RRC connection 518 with the gNB 504. At 520, the UE 502 may initiate an IMSI (International Mobile Subscriber Identity) attachment / registration procedure, which occurs as a result of the UE powering up or moving between different tracking areas within a PLMN.

[0081] At 570A, the UE 502 may send an AI / ML registration / identification request 522 to the AMF 506. At 570B, the AMF 506 receives the request 522. In some embodiments, during the IMSI attachment procedure, the UE 502 may send an ML model registration request (also referred to as an ML model identification request) to the AMF 506, where the ML model capabilities may be declared by the UE 502. The ML model capabilities may be generic functions known at a non-access stratum layer, for example, that allow the AMF to determine the types of ML capabilities already pre-programmed into the UE (such as the availability of hardware accelerators or GPUs, the amount of RAM for ML use, etc.).

[0082] In some embodiments, the request 522 may also include a list of AI / ML model IDs stored / supported by the UE. This list includes AI / ML model IDs that the device supports or uses by default (i.e., factory-programmed by the manufacturer). In some embodiments, the list of AI / ML model IDs may be a request to update the current list of AI / ML model IDs pre-identified / registered with the network. In some embodiments, the UE 502 may also notify the network of ML model delivery (content and context) priorities, as described at 544.

[0083] At 524, the AMF 506 determines a list of AI / ML model IDs for the terminal device 502 based on at least one AI / ML model capability. At 572A, the AMF 506 sends a request 526 for AI / ML model ID validation to the UMLCMF 508. At 572B, the UMLCMF 508 receives the request 526 from the AMF 506. At 528, the UMLCMF 508 verifies the AI / ML model ID.

[0084] In some embodiments, the AMF 506 may forward a list of AI / ML model IDs from the UE 502 to the UMLCMF 508, which may return a response list of allowed AI / ML model IDs and ML model context and content. The AMF 506 may interpret the ML capabilities and determine from the AI / ML model IDs that some of these AI / ML model IDs may be inappropriate, for example, because an operator has prohibited their use in a PLMN or zone within a PLMN, and filter the list for the UMLCMF.

[0085] In some embodiments, the UMLCMF 508 receives ML features and interprets them to perform filtering. The difference between ML content and context is that ML content is a list of strings containing the actual ML model parameters, while context contains metadata for interpreting the ML content (e.g., input / output formats, number of layers, etc.). This separation of ML content and context is understood as a standard way of specifying ML models.

[0086] At 574B, the UMLCMF 508 may send an AI / ML model ID validation response 530 to the AMF 506. At 574A, the AMF 506 may receive the response 530 from the UMLCMF 508. At 578B, the AMF 506 may send an initial context setup request 534 to the gNB 504. At 578A, the gNB 504 may receive the request 534. At 536, the gNB 504 may store the approved or identified AI / ML model ID and context.

[0087] At 580A, the gNB 504 sends an initial context setup response 538 to the AMF 506. At 580B, the AMF receives the response 538. In some embodiments, the gNB 504 is updated with the list of approved / identified AI / ML model IDs and the ML model content and content for each AI / ML model ID. In some embodiments, the gNB 504 may provide feedback to the AMF regarding the list of approved / identified ML model IDs by verifying the ML model context (e.g., by reviewing historical performance data for a set of visited cells).

[0088] In some embodiments, for each AI / ML model ID, there is an acknowledgement "ML model delivery priority" that is ACKed by the network, so that some ML model IDs are delivered directly to the UE via over-the-air (OTA) and some via the network.

[0089] At 539, the AMF 506 considers the feedback from the gNB 504. In some embodiments, the gNB 504 sends feedback to the AMF 506 regarding a list of approved / identified ML model IDs by reviewing the ML model context. In some embodiments, the gNB 504 reviews the model context and determines compatibility (e.g., reviews performance information in the ML model context). If the check with the model context is successful (e.g., above a 90% criteria threshold), the gNB 504 can continue forwarding to the UE 503. If not, the gNB 504 informs the AMF that the ML model ID cannot be used, and the AMF 506 may tag it as unqualified. In some embodiments, the operator may then decide to delete this model from the database.

[0090] At 576B, the UMLCMF 508 sends an ML registration / identification response 532 to the UE 502. At 576A, the UE 502 receives the response 532. In some embodiments, the UE 503 may receive an ML model registration response (also referred to as an ML model identification response) from the AMF 506 if an approved or updated list of AI / ML model IDs is known to the UE 502. Here, the UE 502 is expected to use the approved / identified / updated AI / ML model list.

[0091] In some embodiments, the response 532 may also include an updated list of AI / ML model IDs stored / supported by the UE, which may include replacing older versions of ML models with newer versions.

[0092] In some embodiments, the response 532 includes an ML model delivery priority for each approved / identified ML model ID, which may allow the network to approve or update the request from the UE 502, at 522.

[0093] In some embodiments, in a CSI compression use case, the UE 502 transmits ML model capabilities and supported AI / ML model IDs for a bidirectional ML model for model identification at 570A, where the ML model capabilities may indicate support for other types of models. Based on 524 through 530, the AMF 506 identifies other ML models that are more appropriate for the network and the UE 502 than the models supported by the UE 502. If the identified UE 502 is part of a bidirectional model and is within the ML capabilities supported by the UE 502, such models can be included in the updated list of AI / ML model IDs.

[0094] At 540, the UE 502 stores or updates a list of approved, identified, or updated AI / ML model IDs. In some embodiments, as an additional step, at 544, if model distribution is supported as part of the model registration / identification, the network initiates model distribution for the approved or updated list of AI / ML model IDs. The model distribution priority, indicated by the UE 502 at 570A and accepted or overridden by the AMF 506, determines how the ML model is distributed to the UE 502. In some embodiments, the UE 502 may request that the ML model distribution be performed by direct communication with a third-party server. In this case, the 3GPP network is transparent to the ML model distribution process (although a separate data / PDU session may need to be established between the UE and the third-party server for the ML model distribution).

[0095] In some embodiments, the UE 502 notifies the 3GPP® network that either the 3GPP® network intervenes and the ML model distribution (i.e., content and context) is transferred between the AMF 506 and the UE 502, or the AMF 506 requests the gNB 504 to perform the ML model distribution. In some embodiments, if a particular model in the approved or updated AI / ML model ID list is not available to the UE 502, the model distribution procedure is initiated by the UE 502.

[0096] Thus, at the end of Scenario 2, UE 502 and the network are fully synchronized with the AI / ML model IDs that UE 502 is authorized to use with the network.

[0097] FIG. 5B illustrates yet another example signaling process 500B for AI / ML model identification and capability processing in accordance with some embodiments of the present disclosure. The example signaling process 500B of FIG. 5B can be considered an example of the signaling process 400 of FIG. 4. Accordingly, the UE 502 of FIG. 5B is an example of the terminal device 110 of FIG. 4. The gNB 504 of FIG. 5B is an example of the access network device 120 of FIG. 4. The AMF 506 of FIG. 5B is an example of the first core network device 130 of FIG. 4. The UMLCMF 508 of FIG. 5B is an example of the second core network device 140 of FIG. 4. FIG. 5B illustrates Scenario 3. Scenario 3 is illustrated in dashed box 596. Scenario 3 illustrates UE model distribution, capability query, and ML model configuration.

[0098] Model registration / identification may be completed at 546. In some embodiments, prior to UE capability inquiry and reporting, the model used by UE 502 may undergo a model registration / identification procedure (i.e., scenarios 1 and / or 2).

[0099] At 548, the gNB 520 may obtain UE capabilities from the approved / identified model ID list. In some embodiments, obtaining may include steps 550, 552, and 554. At 582B, the gNB 504 may send a UE capability query 550 to the UE 502. At 582A, the UE 502 receives the query 550. At 552, the UE 502 generates AI / ML capabilities based on the query 550. At 584A, the UE 502 sends UE capability information 554 to the gNB 504.

[0100] In some embodiments, the gNB 504 may want to obtain the ML model capabilities of the UE 502, so the gNB 504 formats and sends a UE capability query 550 toward the UE 502. Based on the UE capability query 550, the UE 502 generates a UE capability report, which uses the approved / registered AI / ML model IDs stored as a result of the above procedure. The UE capabilities are then reported toward the gNB 504.

[0101] In some embodiments, UE capability reporting of ML capabilities is performed by reporting a Model-functionality ML (Model-functionality defines the associated radio control parameter functionality (typically defined in a specification) when supporting an ML function or ML use case).

[0102] In some examples, within a particular model capability, a UE may support multiple AI / ML model IDs, and these AI / ML model IDs may also be included in the capability report for the particular model capability. Here, different AI / ML model IDs may still refer to different ML-Model-Contexts and ML-Model-Contents, which are pre-identified by the network and continue to be used when configuring the UE with an ML configuration. In some embodiments, a model capability may also be identified by an ID (referred to as a model capability ID).

[0103] At 556, the gNB 504 may configure the UE 502 based on the UE capability information 554. In some embodiments, the configuration may include 558, 562, and 564. Based on the received UE capability information 554, the gNB 504 may configure the ML use case / function on the UE 502. In some embodiments, if multiple AI / ML model IDs are supported for a particular model function, the gNB sends an activate or select command to the UE to indicate the exact AI / ML model ID to be used for the ML use case / function.

[0104] In some embodiments, in a CSI compression use case, the UE 502 supports a Transformer-based NN architecture for some AI / ML model IDs and a non-Transformer-based NN architecture for other AI / ML model IDs for a particular model function. If the network wants to apply a Transformer-based NN architecture at the gNB 504 (the decoder / gNB portion in a two-sided model), an additional selection command may be sent from the gNB 504 to select an AI / ML model ID that uses a Transformer-based NN architecture.

[0105] In some embodiments, as an additional step, at 560, if model distribution is supported as part of the model configuration (where multiple model distributions may occur, as in 544), the network may initiate model distribution for the list of configured / selected / activated AI / ML model IDs.

[0106] At 589A and 589B, a PDU session 561 is initiated. The AMF 506 establishes the PDU session 561 to transfer the AI / ML model content to the UE 502. This is the case when the core network is involved in model distribution.

[0107] At 562, the UE 502 configures the ML capability in response to the configuration / selection / activation received from the gNB 504. At 590A, the UE 502 sends an RRC reconfiguration complete message 564 to the gNB 504. At 590B, the gNB 504 receives the message 564.

[0108] 6 illustrates an example flowchart 600 of a method implemented in a terminal device according to some example embodiments of the present disclosure, and is referenced in conjunction with FIG. 1A.

[0109] At 602, the terminal device 110 sends an AI / ML model registration or identification request to the core network device 130. The request includes at least one AI / ML model capability of the terminal device. At 604, the terminal device 110 receives a response to the AI / ML model registration or identification from the core network device 130. The response includes a list of approved or identified AI / ML model IDs and at least one AI / ML model distribution priority.

[0110] In some embodiments, terminal device 110 stores a list of approved or identified AI / ML model IDs. In some embodiments, terminal device 110 updates a previous list of AI / ML model IDs stored on the terminal device based on the list of approved or identified AI / ML model IDs.

[0111] In some embodiments, terminal device 110 receives a list of AI / ML model content for a list of approved or identified AI / ML model IDs from at least one of access network device 120, core network device 130, or a third-party database.

[0112] In some embodiments, terminal device 110 may receive a capability query from access network device 130 to obtain at least one capability for an approved or identified AI / ML model ID. Terminal device 110 may generate a capability report including the at least one capability for the approved or identified AI / ML model ID based on the capability query. Terminal device 110 may send the capability report to access network device 120.

[0113] In some embodiments, terminal device 110 receives a reconfiguration message from access network device 130. The reconfiguration message is generated based on at least one capability for at least one AI / ML model ID in terminal device 110 and includes at least one setting specific to the approved or identified AI / ML model ID. In some embodiments, terminal device 110 may configure at least one AI / ML model capability based on the received reconfiguration message. In some embodiments, terminal device 110 may send a reconfiguration complete message to access network device 120.

[0114] In some embodiments, terminal device 110 receives a list of AI / ML model content corresponding to a configured, selected, or activated list of AI / ML model IDs from at least core network device 130, a third-party device, or an operator. In some embodiments, the request to register or identify an AI / ML model further includes a list of AI / ML mode IDs stored or supported by terminal device 110.

[0115] 7 illustrates an example flowchart 700 of a method implemented in an access network device, in accordance with some example embodiments of the present disclosure. Reference is made in conjunction with FIG. 1A.

[0116] At 702, the access network device 120 receives a message from the core network device 130 that includes a list of approved or identified AI / ML model IDs and contexts for the terminal device 110. At 704, the access network device 130 stores the approved or identified AI / ML model IDs and contexts.

[0117] In some embodiments, the access network device 120 may evaluate or verify the approved or identified AI / ML model IDs based on the AI / ML model context. The access network device 120 may transmit to the terminal device 110 a list of AI / ML model contents corresponding to the list of approved or identified AI / ML model IDs.

[0118] In some embodiments, the access network device 120 determines the AI / ML capabilities of the terminal device 110 for the approved or identified AI / ML model ID. The access network device 120 sends a capability query for the terminal device 110. The capability query includes the AI / ML capabilities.

[0119] In some embodiments, access network device 120 may send a capability query to terminal device 110 to obtain at least one capability for the approved or identified AI / ML model ID. Access network device 120 may receive a capability report from terminal device 110 that includes at least one capability for the approved or identified AI / ML model ID.

[0120] In some embodiments, access network device 120 may send a reconfiguration message including at least one configuration specific to the approved or identified AI / ML model ID to terminal device 110. Access network device 120 may receive a reconfiguration complete message from terminal device 110.

[0121] In some embodiments, the message may further include a list of AI / ML model contents corresponding to a list of approved or identified AI / ML model IDs.

[0122] 8 shows a flowchart 800 illustrating an example of a method implemented in a first core network device, in accordance with some exemplary embodiments of the present disclosure, and is to be referenced in conjunction with FIG. 1A.

[0123] At 802, the first core network device 130 receives a request for registration or identification of an AI / ML model from the terminal device 110. The request includes at least one AI / ML model capability of the terminal device. At 804, the first core network device 130 determines a list of AI / ML model IDs for the terminal device 110 based on the at least one AI / ML model capability. At 806, the first core network device 130 sends a request to the second core network device 140 to verify the validity of the list of AI / ML model IDs. At 808, the first core network device 130 receives a response from the second core network device including a list of approved or identified AI / ML model IDs and contexts.

[0124] In some embodiments, the first core network device 130 sends a response to the registration or identification of the AI / ML model to the terminal device 110. The response includes a list of approved or identified AI / ML model IDs and at least one AI / ML model distribution priority.

[0125] In some embodiments, first core network device 130 may send a message to access network device 120 including a list of approved or identified AI / ML model IDs and context for terminal device 110. In some embodiments, first core network device 130 may send to terminal device 110 a list of AI / ML model content for the approved or identified list of AI / ML model IDs.

[0126] 9 illustrates an example flowchart 900 of a method implemented in a second core network device in accordance with some example embodiments of the present disclosure, and is referenced in conjunction with FIG. 1A.

[0127] At 902, the second core network device 140 receives from the first core network device 130 a request to verify a list of AI / ML model IDs associated with the terminal device 110. At step 904, the second core network device 140 determines a list of approved or identified AI / ML model IDs and contexts for the terminal device 110 by verifying the list of AI / ML model IDs based on a database of verified AI / ML models for the terminal device 110. At 906, the second core network device 140 sends a response to the first core network device 130 including the list of approved or identified AI / ML model IDs and contexts.

[0128] In some embodiments, the second core network device 140 may receive a request to store or delete an AI / ML model from the database. The request includes a model ID and model context for the AI / ML model. In some embodiments, the second core network device 140 may store or delete the AI / ML model based on the request. In some embodiments, the second core network device 140 may send a response to the request to the database. In some embodiments, the request may further include model content for the AI / ML model.

[0129] By implementing methods 600, 700, 800 and 900, the solution for using AI / ML model IDs provides a signaling procedure that needs to be defined between the terminal device and the network, which allows the validity of a particular AI / ML model to be determined, and further allows the network to query the terminal device about capabilities related to the AI / ML model.

[0130] In some exemplary embodiments, an apparatus capable of performing method 600 may comprise means for performing each step of method 600. The means may be implemented in any suitable form, for example, as a circuit or a software module.

[0131] In some exemplary embodiments, the device comprises means for sending a request for registration or identification of an AI / ML model to a core network device and means for receiving a response to the registration or identification of the AI / ML model from the core network device, the request including at least one AI / ML model capability of the terminal device, and the response including a list of approved or identified AI / ML model IDs and a delivery priority of the at least one AI / ML model.

[0132] In some example embodiments, the device may include means for storing a list of approved or identified AI / ML model IDs, and means for updating a list of previous AI / ML model IDs stored in the terminal device based on the list of approved or identified AI / ML model IDs.

[0133] In some example embodiments, the device may comprise means for receiving, from at least one of an access network device, a core network device, or a third party device, a list of AI / ML model content for a list of approved or identified AI / ML model IDs.

[0134] In some example embodiments, the apparatus may include means for receiving a capability query from an access network apparatus to obtain at least one capability for an approved or identified AI / ML model ID; means for generating a capability report including the at least one capability for the approved or identified AI / ML model ID based on the capability query; and means for transmitting the capability report to the access network apparatus.

[0135] In some example embodiments, the apparatus may comprise: means for receiving a reconfiguration message from an access network apparatus, the reconfiguration message being generated based on at least one capability of the terminal apparatus for at least one AI / ML model ID, the reconfiguration message including at least one configuration specific to the approved or identified AI / ML model ID; means for configuring the at least one AI / ML model capability based on the received reconfiguration message; and means for sending a reconfiguration complete message to the access network apparatus.

[0136] In some example embodiments, the device may comprise means for receiving, from at least one of a core network device, a third party device, or an operator, a list of AI / ML model content corresponding to a configured, selected, or activated list of AI / ML model IDs.

[0137] In some exemplary embodiments, the request for registration or identification of an AI / ML model further includes a list of AI / ML mode IDs stored or supported on the terminal device.

[0138] In some embodiments, the apparatus further comprises means for performing other steps in some embodiments of method 600. In some embodiments, the means comprises at least one processor and at least one memory containing computer program code, the at least one memory and the computer program code configured to cause the at least one processor to perform the operations of the apparatus.

[0139] In some exemplary embodiments, an apparatus capable of performing method 700 may comprise means for performing each step of method 700. The means may be implemented in any suitable form, for example, as a circuit or a software module.

[0140] In some example embodiments, the apparatus comprises means for receiving, from a core network apparatus, a message including a list of approved or identified AI / ML model IDs and contexts for the terminal apparatus, and means for storing the approved or identified AI / ML model IDs and contexts.

[0141] In some example embodiments, the apparatus may include means for evaluating or verifying approved or identified AI / ML model IDs based on the AI / ML model context, and means for transmitting, to the terminal device, a list of AI / ML model content corresponding to the approved or identified list of AI / ML model IDs.

[0142] In some demonstrative embodiments, the apparatus may include means for determining AI / ML capabilities of the terminal device for an approved or identified AI / ML model ID, and means for transmitting a capability query of the terminal device, where the capability query includes the AI / ML capabilities.

[0143] In some demonstrative embodiments, the apparatus may include means for sending a capability query to a terminal device to obtain at least one capability of an approved or identified AI / ML model ID, and means for receiving, from the terminal device, a capability report including the at least one capability of the approved or identified AI / ML model ID.

[0144] In some demonstrative embodiments, the apparatus may include means for sending a capability query to a terminal device to obtain at least one capability of an approved or identified AI / ML model ID, and means for receiving, from the terminal device, a capability report including the at least one capability of the approved or identified AI / ML model ID.

[0145] In some exemplary embodiments, the apparatus may include means for sending, to the terminal device, a reconfiguration message including at least one configuration specific to an approved or identified AI / ML model ID, and means for receiving, from the terminal device, a reconfiguration complete message. In some exemplary embodiments, the message may further include a list of AI / ML model content corresponding to the approved or identified list of AI / ML model IDs.

[0146] In some embodiments, the apparatus further comprises means for performing other steps in some embodiments of method 700. In some embodiments, the means comprises at least one processor and at least one memory containing computer program code, the at least one memory and the computer program code configured by the at least one processor to cause the apparatus to perform operations.

[0147] In some exemplary embodiments, an apparatus capable of performing method 800 may comprise means for performing each step of method 800. The means may be implemented in any suitable form, for example, as a circuit or a software module.

[0148] The apparatus may include means for receiving a request for registration or identification of an AI / ML model from a terminal device, the request including at least one AI / ML model capability of the terminal device; means for determining a list of AI / ML model IDs for the terminal device based on the at least one AI / ML model capability; means for sending a validation request for the list of AI / ML model IDs to a second core network device; and means for receiving a response from the second core network device including a list of approved or identified AI / ML model IDs and contexts.

[0149] In some exemplary embodiments, the apparatus comprises means for transmitting, to a terminal device, a response to the registration or identification of the AI / ML model, the response including the approved or identified AI / ML model IDs and at least one AI / ML model distribution priority. In some exemplary embodiments, the apparatus comprises means for transmitting, to an access network device, a message including a list of approved or identified AI / ML model IDs and contexts for the terminal device.

[0150] In some demonstrative embodiments, the apparatus may comprise means for transmitting, to the terminal device, a list of AI / ML model content corresponding to an approved or identified list of AI / ML model IDs.

[0151] In some embodiments, the apparatus further comprises means for performing other steps in some embodiments of method 800. In some embodiments, the means comprises at least one processor and at least one memory containing computer program code, the at least one memory and the computer program code configured to cause the at least one processor to perform the operations of the apparatus.

[0152] In some exemplary embodiments, an apparatus capable of performing method 900 may comprise means for performing each step of method 900. The means may be implemented in any suitable form, for example, as a circuit or a software module.

[0153] The device may include means for receiving a request to verify a list of AI / ML model IDs associated with the terminal device from a first core network device; means for determining, in a second core network device, a list of approved or identified AI / ML model IDs and contexts for the terminal device by verifying the list of AI / ML model IDs based on a database of verified AI / ML models for the terminal device; and means for sending a response to the first core network device including the list of approved or identified AI / ML model IDs and contexts.

[0154] In some exemplary embodiments, the apparatus comprises means for receiving a request to store or delete an AI / ML model from a database, the request including a model ID and a model context of the AI / ML model, means for storing or deleting the AI / ML model based on the request, and means for sending a response to the request to the database. In some exemplary embodiments, the request may further include model content of the AI / ML model.

[0155] In some embodiments, the apparatus further comprises means for performing other steps in some embodiments of method 900. In some embodiments, the means comprises at least one processor and at least one memory containing computer program code, the at least one memory and the computer program code configured by the at least one processor to cause the apparatus to perform operations.

[0156] 10, a simplified block diagram illustrating an example of an apparatus suitable for implementing embodiments of the present disclosure is shown. Apparatus 1000 may be provided to implement a communications apparatus such as, for example, terminal apparatus 110 shown in FIG. 1A. As shown, apparatus 1000 includes one or more processors 1010, one or more memories 1040 (connectable to processor 1010), and one or more communications modules 1040 (connectable to processor 1010).

[0157] The communication module 1040 is for two-way communication. The communication module 1040 includes at least one antenna to facilitate communication. The communication interface may represent any interface necessary for communication with other network elements, for example, the communication interface may be a wireless or wired interface to other network elements, or a software-based interface for communication.

[0158] The processor 1010 may be of any type suitable for a local technology network and may include, for example, one or more of, but not limited to, a general-purpose computer, a special-purpose computer, a microprocessor, a digital signal processor (DSP), and a processor based on a multi-core processor architecture. The device 1000 may comprise multiple processors, such as application-specific integrated circuit chips, time-slaved to a clock that synchronizes the main processor.

[0159] The memory 1020 may include one or more non-volatile memories and one or more volatile memories. Examples of non-volatile memory include, but are not limited to, read-only memory (ROM) 1024, electrically programmable read-only memory (EPROM), flash memory, hard disks, compact disks (CDs), digital video disks (DVDs), and other magnetic and / or optical storage devices. Examples of volatile memory include, but are not limited to, random access memory (RAM) 1022 and other volatile memory that is not retained during power loss.

[0160] The computer program 1030 includes computer-executable instructions that are executed by an associated processor 1010. The program 1030 may be stored in ROM 1024. The processor 1010 may load the program 1030 into RAM 1022 to perform any suitable operations and processes.

[0161] The embodiments of the present disclosure may be implemented by a program so that the device 1000 can perform all of the processes of the present disclosure described in Figures 4 to 9. The embodiments of the present disclosure may be implemented by hardware or a combination of software and hardware.

[0162] In some embodiments, the program 1030 may be tangibly stored on a computer-readable medium included in the device 1000 (such as memory 1020) or other storage accessible by the device 1000. The device 1000 may load the program 1030 from the computer-readable medium into RAM 1022 for execution. The computer-readable medium may include any type of tangible non-volatile storage device, such as ROM, EPROM, flash memory, hard disk, CD, DVD, and the like. Figure 11 shows an example of a computer-readable medium 1100 in the form of a CD or DVD. The computer-readable medium has the program 1030 stored thereon.

[0163] In general, various embodiments of the present disclosure may be implemented by hardware or special purpose circuits, software, logic, or any combination thereof. Some aspects may be implemented by hardware, while other aspects may be implemented by firmware or software executed by a controller, microprocessor, or other computing device. Although various aspects of the embodiments of the present disclosure have been illustrated and described using block diagrams, flowcharts, or other graphical representations, it should be understood that the blocks, devices, systems, techniques, or methods described herein can be implemented in, by way of non-limiting example, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller, other computing device, or combinations thereof.

[0164] The present disclosure also provides at least one computer program product tangibly embodied in a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions, such as those included in program modules, that execute on a device on a target real processor or target virtual processor to perform methods 600, 700, 800, or 900, as described above with respect to FIG. 6 or FIG. 9. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The functionality of the program modules may be combined or split between program modules as desired in various embodiments. The machine-executable instructions for a program module may be executed in a local or distributed device. In a distributed device, program modules may be located in both local and remote storage media.

[0165] Program code for carrying out the methods of the present disclosure can be written in any combination of one or more programming languages. Such program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, and when executed by the processor or controller, causes the functions / acts specified in the flowcharts and / or block diagrams to be performed. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine or server, or entirely on the remote machine or server.

[0166] In the context of the present disclosure, computer program code or associated data may be transmitted by any suitable carrier, such as a signal, a computer-readable medium, or the like, to enable a device, apparatus, or processor to perform the various processes and operations described above.

[0167] The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. Computer-readable media include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. More specific examples of computer-readable storage media include an electrical connection having one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. As used herein, the term "non-transitory" refers to a limitation on the medium itself (i.e., tangible, not a signal), rather than a limitation on the permanence of the data storage (e.g., RAM vs. ROM).

[0168] Furthermore, although operations are shown in a particular order, this does not imply that such operations must be performed in the particular order or sequential order shown, or that all of the operations shown must be performed, to achieve desirable results. Multitasking and parallel processing may be preferred in certain situations. Similarly, while the above description includes details of several specific embodiments, these should not be construed as limiting the scope of the disclosure, but rather as descriptions of functionality specific to particular embodiments. Certain features that are described in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment may also be implemented in multiple embodiments separately or in any suitable subcombination.

[0169] Although the present disclosure has been described using language specific to structural features and / or methodological acts, it is to be understood that the present disclosure, as defined by the appended claims, is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.

Claims

1. A terminal device, at least one processor; When executed by the at least one processor, the terminal device is configured to: sending, to a core network device, a request for registration or identification of an artificial intelligence / machine learning (AI / ML) model, the request including at least one AI / ML model capability of the terminal device; receiving a response to the registration or identification of the AI / ML model from the core network device, the response including a list of approved or identified AI / ML model identifiers (IDs) and at least one AI / ML model distribution priority; at least one memory storing instructions for causing the A terminal device comprising:

2. The terminal device further storing a list of said approved or identified AI / ML model IDs; or updating a historical list of AI / ML model IDs stored in the terminal device based on the list of approved or identified AI / ML model IDs; 2. The terminal device of claim 1, adapted to execute the following:

3. The terminal device further receiving, from at least one of an access network device, the core network device, or a third party device, a list of AI / ML model contents corresponding to the list of approved or identified AI / ML model IDs; 3. A terminal device according to claim 1, adapted to execute the following:

4. The terminal device further receiving a capability query from an access network device to obtain at least one capability of the approved or identified AI / ML model ID; generating a feature report based on the feature query, the feature report including at least one feature of the approved or identified AI / ML model ID; sending the capability report to the access network device; 4. A terminal device according to claim 1, adapted to execute the following:

5. The terminal device further receiving a reconfiguration message from an access network device, the reconfiguration message being generated based on at least one capability of the terminal device for at least one AI / ML model ID, the reconfiguration message including at least one configuration specific to the approved or identified AI / ML model ID; configuring at least one AI / ML model function based on the received reconfiguration message; sending a reconfiguration complete message to the access network device; 5. A terminal device according to claim 1, adapted to execute the following:

6. The terminal device further receiving, from at least one of the core network device, the third party device, or an operator, a list of AI / ML model content corresponding to a configured, selected, or activated list of AI / ML model IDs; 6. A terminal device according to any one of claims 1 to 5, adapted to execute the following:

7. The terminal device according to claim 1 , wherein the request for registration or identification of an AI / ML model further includes a list of AI / ML model IDs stored or supported in the terminal device.

8. An access network device, at least one processor; When executed by the at least one processor, the access network device is configured to: receiving a message from a core network device, the message including a list of authorized or identified AI / ML model identifiers (IDs) and contexts for the terminal device; storing the approved or identified AI / ML model ID and context; at least one memory storing instructions for executing the An access network device comprising:

9. The access network device further comprises: Evaluating or validating the approved or identified AI / ML model ID based on the AI / ML model context; transmitting to the terminal device a list of AI / ML model content corresponding to the approved or identified list of AI / ML model IDs; 9. The access network device of claim 8, adapted to execute:

10. The access network device further comprises: determining the AI / ML capabilities of the terminal device for the approved or identified AI / ML model ID; transmitting a capability query of the terminal device, the capability query including the AI / ML capabilities; 10. An access network device according to claim 8 or 9, adapted to execute:

11. The access network device further comprises: sending a capability query to the terminal device to obtain at least one capability of the approved or identified AI / ML model ID; receiving a capability report from the terminal device, the capability report including at least one capability of the approved or identified AI / ML model ID; 11. An access network device according to any one of claims 8 to 10, adapted to execute the following:

12. The access network device further comprises: sending to said terminal device a reconfiguration message including at least one configuration specific to said approved or identified AI / ML model ID; receiving a reconfiguration complete message from the terminal device; 12. An access network device according to any one of claims 8 to 11, adapted to execute:

13. 13. The access network device of claim 8, wherein the message further includes a list of AI / ML model contents corresponding to the approved or identified list of AI / ML model IDs.

14. A first core network device, at least one processor; When executed by the at least one processor, the method causes the first core network device to receiving, from a terminal device, a request for registration or identification of an artificial intelligence / machine learning (AI / ML) model, the request including at least one AI / ML model capability of the terminal device; determining a list of AI / ML model identifiers (IDs) for the terminal device based on the at least one AI / ML model capability; sending, to the terminal device, a response to the registration or identification of the AI / ML model, the response including a list of approved or identified AI / ML model IDs and at least one AI / ML model distribution priority; at least one memory storing instructions for executing the A first core network device comprising:

15. The first core network device further sending a request to a second core network device to verify the validity of the list of AI / ML model IDs; receiving a response from the second core network device, the response including a list of approved or identified AI / ML model IDs and contexts; The first core network device of claim 14, adapted to execute the following:

16. The first core network device further sending a message to an access network device containing said list of authorized or identified AI / ML model identifiers (IDs) and contexts for said terminal device; 16. The first core network device according to claim 14 or 15, adapted to execute the following:

17. The first core network device further sending to the terminal device a list of AI / ML model content corresponding to the approved or identified list of AI / ML model IDs; 17. The first core network device according to claim 14, wherein the first core network device is adapted to execute the following:

18. A second core network device, at least one processor; When executed by the at least one processor, the method causes the second core network device to receiving a request from a first core network device to verify a list of AI / ML model IDs associated with the terminal device; determining a list of approved or identified AI / ML model IDs and contexts for the terminal device by verifying the list of AI / ML model IDs based on a database of verified AI / ML models for the terminal device; sending a response to the first core network device including the list of approved or identified AI / ML model IDs and contexts; at least one memory storing instructions for executing the A second core network device comprising:

19. The second core network device further receiving a request to store or delete an AI / ML model from the database, the request including a model ID and a model context of the AI / ML model; storing or deleting the AI / ML model based on the request; sending a response to the request to the database; The second core network device of claim 18, adapted to execute the following:

20. The second core network device according to claim 18 or 19, wherein the request further includes model content of the AI / ML model.

21. sending, at a terminal device, to a core network device, a request for registration or identification of an artificial intelligence / machine learning (AI / ML) model, the request including at least one AI / ML model capability of the terminal device; receiving, at the terminal device, from the core network device, a response to the registration or identification of the AI / ML model, the response including a list of approved or identified identifiers (IDs) of AI / ML models and at least one AI / ML model distribution priority; A method comprising:

22. receiving, at the access network device, from the core network device, a message including a list of authorized or identified AI / ML model identifiers (IDs) and contexts for the terminal device; storing, in the access network device, the approved or identified AI / ML model ID and context; A method comprising:

23. receiving, at a first core network device, a request for registration or identification of an artificial intelligence / machine learning (AI / ML) model from a terminal device, the request including at least one AI / ML model capability of the terminal device; determining, at the first core network device, a list of AI / ML model identifiers (IDs) for the terminal device based on the at least one AI / ML model capability; sending, to the terminal device, a response to the registration or identification of the AI / ML model, the response including a list of approved or identified AI / ML model IDs and at least one AI / ML model distribution priority; A method comprising:

24. receiving, at a second core network device, from the first core network device, a request to verify a list of AI / ML model IDs associated with the terminal device; In the second core network device, determining a list of approved or identified AI / ML model IDs and contexts for the terminal device by verifying the list of AI / ML model IDs based on a database of verified AI / ML models for the terminal device; sending, at the second core network device, a response to the first core network device, the response including the list of approved or identified AI / ML model IDs and contexts; A method comprising:

25. A non-transitory computer readable medium comprising program instructions for causing an apparatus to perform at least the method of any of claims 21 to 24.

Images