Method and system for conditional model switching and activation in radio access networks

Abstract

The present invention disclosure provides a unified model identification framework and associated mapping mechanism for AI / ML model transfer in wireless communication networks including base station (e.g., gNB, TN, NTN) and mobile station (e.g., UE). The proposed framework ensures efficient model deployment, transfer, and lifecycle management across different network environments and involves pre-transferring multiple candidate models to a UE with predefined activation conditions.

Description

[0001] 202501727

[0002] - 1 -

[0003] TITLE

[0004] Method and system for conditional model switching and activation in radio access networks

[0005] TECHNNICAL FIELD

[0006] The present invention disclosure relates to AI / ML based model operation with wireless communication networks and, more particularly, to an optimized method for model switching or model activation to manage AI / ML model transfers in radio access networks by supporting a unified model identification framework and an associated mapping mechanism for efficient model transfer.

[0007] BACKGROUND

[0008] In 3GPP (Third Generation Partnership Project), one of the selected study items as the approved Release 18 package is AI / ML (artificial intelligence / machine learning) as described in the related document (RP-213599) addressed in 3GPP TSG (Technical Specification Group) RAN (Radio Access Network) meeting #94e. The official title of AI / ML study item is “Study on AI / ML for NR Air Interface”. The goal of this study item is to identify a common AI / ML framework and areas of obtaining gains using AI / ML based techniques with use cases. According to 3GPP, the main objective of this study item is to study AI / ML framework for air-interface with target use cases by considering performance, complexity, and potential specification impact. In particular, AI / ML model, terminology and description to identify common and specific characteristics for framework are included as one of key work scopes. Regarding AI / ML framework, various aspects are under consideration for investigation and one of key items is about lifecycle management of AI / ML model where multiple stages are included as mandatory for model training, model deployment, model inference, model monitoring, model updating etc.

[0009] Also in 3GPP, two-sided (AI / ML) model is defined as a paired AI / ML model(s) over which joint inference is performed, where joint inference comprises AI / ML Inference whose inference is performed jointly across the UE and the network. Also for onesided (AI / ML) model, UE-side (AI / ML) model is defined as an AI / ML model whose202501727

[0010] -2 -

[0011] inference is performed entirely at the UE and network-side (AI / ML) model is defined as an AI / ML model whose inference is performed entirely at the network. Currently, AI / ML specification work is at the stage of work item discussion for Release 19.

[0012] Earlier, in 3GPP TR 37.817 for Release 17, titled as Study on enhancement for Data Collection for NR and EN-DC, UE (user equipment) mobility was also considered as one of AI / ML use cases and one of scenarios for model training / inference is that both functions are located within RAN node. Followingly, in Release 18 the new work item of “Artificial Intelligence (AI)ZMachine Learning (ML) for NG-RAN” was initiated to specify data collection enhancements and signaling support within existing NG-RAN interfaces and architecture.

[0013] For the above active standardization works, RAN-based AI / ML model is considered very significant for both network and UE to meet any desired model operations (e.g., model training, inference, selection, switching, update, monitoring, etc.). Model information can be signaled to pair both network-side and UE-side models for various lifecycle management (LCM) operations. However, signaling overhead indicating model information can be very high especially when model based LCM is processed between base station (BS / gNB) and multiple UEs. Traditional AI / ML model deployment often lacks standardized compatibility across RAN, core networks (CN), and inter-operator environments, leading to inefficiencies in model execution, synchronization, and adaptation. There are several functional units defined in RAN such as RU (radio unit), DU (distributed unit), CU (central unit) as well as TRP (transmission reception point) as a network node. In conventional cellular networks, the association between a user equipment (UE) and a base station (BS) is based on geographic cell boundaries, limiting flexibility. In RAN, signaling is crucial for communication between the UE and the network as this signaling occurs across different layers of the protocol stack, primarily L1 (Layer 1), L2 (Layer 2), and RRC (radio resource control).

[0014] In LCM, model training is one of the most important parts for model deployment and currently there is no specification defined for signaling methods and network-UE behaviors so as to identify the required dataset when model updating / re-training as any activated model can be also impacted due to model / data drift. When ML202501727

[0015] - 3 -

[0016] condition changes, the enabled AI / ML model(s) can be impacted for model performance due to data / model drift. In this case, model re-training / updating can be executed. This invention proposes a novel conditional model switching and activation. The existing 3GPP framework does not support a conditional, preconfigured AI / ML model switching mechanism, which can dynamically adapt to radio conditions, computational resources, and inference quality. The present invention disclosure introduces a novel model switching / activation to address these issues efficiently.

[0017] WO2024110160A1 describes a request reception for a machine learning model, wherein the request comprises information on model adaptation constraint for training of the machine learning model or an inference of the machine learning model with response to the access node being able to adapt the machine learning model using model adaptation constraint.

[0018] WO2023211572A1 describes a method of implementing AI / ML for air interface optimization determining a collaboration level for AI / ML collaboration between a network and a device from among a plurality of predetermined collaboration levels.

[0019] W02023006205A1 describes a method of sending a request for a first machine learning model to a machine learning model provider and receiving a response on the request with an executable file or information to access a first part of the first machine learning model as a service, and metadata associated with the first part.

[0020] US2024265306A1 describes a method of collaborating with a network device with selected collaboration levels for machine learning operations with life cycle management type and model transfer format.

[0021] US2022400373A1 describes the method of determining neural network functions and configuring models for performing wireless communications management

[0022] procedures.202501727

[0023] - 4 -

[0024] WO2022161624A1 describes the method of receiving a request for retrieving or executing a ML model or a combination of ML models.

[0025] W02024200393A1 describes methods of using a list of AI / ML candidate models for model switching in wireless mobile communication system, where alternative model(s) is applied to model switching operation autonomously based on the preconfigured mapping relationship information between model IDs and other associated parameter values. TR38.843 introduced different options for model delivery / transfer to UE, training location, and model delivery / transfer format combinations for UE-side models and UE-part of two-sided models.

[0026] The present disclosure solves the cited problem by the proposed embodiments and describes as first aspect a method for AI / ML model switching and / or activation at network entity in a wireless communication system, comprising: preloading multiple AI / ML models onto a UE or pre-obtaining their associated model identifiers; transmitting, from a network entity to the UE, a candidate model indication message containing a list of candidate AI / ML models, associated model IDs, model activation conditions, and priority levels; monitoring, by the UE, the network and device conditions to determine whether an AI / ML model switch and / or activation is required; selectively transmitting, from the network entity to the UE, a target model activation message to activate a specific AI / ML model when the activation conditions are met; and allowing the UE to autonomously switch or activate an AI / ML model without receiving the target model activation message based on pre-configured conditions.

[0027] In some embodiments of the method according to the first aspect the method is characterized by, that the candidate model indication message is transmitted via a L1, L2, or RRC signaling.

[0028] In some embodiments of the method according to the first aspect the method is characterized by, that the candidate model indication message is included in a HandoverRequest message to facilitate AI / ML model preloading and activation in mobility scenarios.202501727

[0029] - 5 -

[0030] In some embodiments of the method according to the first aspect the method is characterized by, that the target model activation message comprises: a target model identifier selected from the candidate models; updated activation conditions specifying when the target model should be activated; and fine-tuned model parameters for optimizing the AI / ML model's execution.

[0031] In some embodiments of the method according to the first aspect the method is characterized by, that the target model activation message is transmitted via a L1 , L2, or RRC signaling.

[0032] In some embodiments of the method according to the first aspect the method is characterized by, that the candidate model indication message is transmitted periodically or event-triggered to ensure AI / ML model readiness at the UE.

[0033] In some embodiments of the method according to the first aspect the method is characterized by, that the two-step messaging approach optimizes radio resource usage by minimizing redundant AI / ML model transfers and reducing network overhead in AI / ML-driven wireless communication environments.

[0034] In some embodiments of the method according to the first aspect the method is characterized by, that further comprising a network entity configured to transmit a candidate model indication message and a target model activation message; and a user equipment (UE) configured to receive and process AI / ML model indication messages, monitor activation conditions, and autonomously switch AI / ML models as required.

[0035] The present disclosure solves the cited problem by the proposed embodiments and describes as second aspect a method for the unified model identification framework, comprising defining a model ID format consisting of fields: {type, architecture, version, parameter, mapping}; transmitting a model ID including a subset of said fields from a network entity to a UE; allowing the UE to infer missing model ID fields using pre-configured information or internal lookup tables; and enabling dynamic model retrieval based on the mapping field.202501727

[0036] -6 -

[0037] In some embodiments of the method according to the second aspect the method is characterized by, that the type field indicates model functionality, model task, or model category such as channel estimation model and / or QoS prediction model and / or beamforming prediction model and / or network traffic model.

[0038] In some embodiments of the method according to the second aspect the method is characterized by, that the model architecture field specifies the AI / ML model identification framework to define and execute the model via indication of how model is structured and implemented, ensuring compatibility between the network side and UE side and allowing the UE to correctly interpret and execute the received AI / ML model.

[0039] In some embodiments of the method according to the second aspect the method is characterized by, the version field specifies the version of the AI / ML model deployed in the network by ensuring consistency between the model stored at network entity and the model available at the UE with use of compatible model versions for model operation.

[0040] In some embodiments of the method according to the second aspect the method is characterized by, the parameter field specifies key configuration parameters of the AI / ML model that affect its behavior and decision-making process, defining how the model operates under different conditions and allowing dynamic fine-tuning of model behavior with parameter adaptation without requiring full model re-transfer.

[0041] In some embodiments of the method according to the second aspect the method is characterized by, the mapping field establishes a mapping index that associates a model ID with specific model operation conditions, UE types, or model applications by allowing flexible assignment of AI / ML models for model transfer, model switching, or model activation.

[0042] In some embodiments of the method according to the second aspect the method is characterized by, that model ID is allowed to be separately assigned with unique202501727

[0043] - 7 -

[0044] identifier while the field information in model ID format maintains mapping relation with each separate model ID.

[0045] In some embodiments of the method according to the second aspect the method is characterized by, that further comprising: a model repository storing AI / ML models accessible based on model mapping information; and a signaling mechanism utilizing L1, L2, or RRC message to transmit AI / ML model information such as a specific model ID format and the associated fields.

[0046] The present disclosure solves the cited problem by the proposed embodiments and describes as third aspect by an apparatus for AI / ML model switching and / or activation at network entity in a and / or for the unified model identification framework wireless communication system the apparatus comprising a wireless transceiver, a processor coupled with a memory in which computer program instructions are stored, said instructions being configured to implement steps according to the first and / or second aspect.

[0047] The present disclosure solves the cited problem by the proposed embodiments and describes as fourth aspect by a user equipment comprising an apparatus according to third aspect.

[0048] The present disclosure solves the cited problem by the proposed embodiments and describes as fifth aspect a gNB comprising an apparatus according to third aspect.

[0049] The present disclosure solves the cited problem by the proposed embodiments and describes as sixth aspect wireless communication system for model switching and / or activation at network entity in a and / or for the unified model identification framework, wherein the wireless communication systems comprises at least a user equipment according to the fourth aspect, at least a gNB according to the fifth aspect, whereby the user Equipment and the gNB each comprises a processor coupled with a memory in which computer program instructions are stored, said instructions

[0050] According to a seventh aspect, the present disclosure relates to a computer program product comprising instructions which, when executed by at least one processor,202501727

[0051] - 8 -

[0052] configure said at least one processor to carry out a method according to the first aspect said at least one processor to carry out a method for exchanging data according to any one of the embodiments of the present disclosure. The computer program product can use any programming language, and can be in the form of source code, object code, or in any intermediate form between source code and object code, such as in a partially compiled form, or in any other desirable form.

[0053] According to a eight aspect, the present disclosure relates to a computer-readable storage medium comprising instructions which, when executed by at least one processor, configure said at least one processor to carry out a method according to any one of the embodiments of the present disclosure.

[0054] SUMMARY OF THE INVENTION

[0055] The application disclosure provides a method for model switching and activation, wherein multiple candidate AI / ML models and associated model IDs are transferred to the UE in advance. The UE dynamically switches and / or activates to the most suitable model based on pre-defined conditions by minimizing additional signaling overhead, supporting a novel unified model identification framework for model ID assignment and management.

[0056] BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 is an exemplary table of two-step messaging.

[0058] Figure 2 is an exemplary signaling flow of network decision-based model switching and activation.

[0059] Figure 3 is an exemplary signaling flow of UE decision-based model switching and activation.

[0060] Figure 4 is an exemplary flow chart of configuring conditional model switching and activation at network side.

[0061] Figure 5 is an exemplary flow chart of the autonomous decision procedure for conditional model activation at the UE.

[0062] Figure 6 is an exemplary table of model ID framework.202501727

[0063] - 9 -

[0064] DETAILED DESCRIPTION

[0065] The detailed description set forth below, with reference to annexed drawings, is intended as a description of various configurations and is not intended to represent the only configurations in which the concepts described herein may be practiced. The detailed description includes specific details for the purpose of providing a thorough understanding of the various concepts. However, it will be apparent to those skilled in the art that these concepts may be practiced without these specific details. In particular, although terminology from 3GPP 5G NR may be used in this disclosure to exemplify embodiments herein, this should not be seen as limiting the scope of the invention.

[0066] Some of the embodiments contemplated herein will now be described more fully with reference to the accompanying drawings. Other embodiments, however, are contained within the scope of the subject matter disclosed herein, the disclosed subject matter should not be construed as limited to only the embodiments set forth herein; rather, these embodiments are provided by way of example to convey the scope of the subject matter to those skilled in the art.

[0067] Generally, all terms used herein are to be interpreted according to their ordinary meaning in the relevant technical field, unless a different meaning is clearly given and / or is implied from the context in which it is used. All references to a / an / the element, apparatus, component, means, step, etc. are to be interpreted openly as referring to at least one instance of the element, apparatus, component, means, step, etc., unless explicitly stated otherwise. The steps of any methods disclosed herein do not have to be performed in the exact order disclosed, unless a step is explicitly described as following or preceding another step and / or where it is implicit that a step must follow or precede another step. Any feature of any of the embodiments disclosed herein may be applied to any other embodiment, wherever appropriate. Likewise, any advantage of any of the embodiments may apply to any other embodiments, and vice versa. Other objectives, features and advantages of the enclosed embodiments will be apparent from the following description.202501727

[0068] - 10 -

[0069] In some embodiments, a more general term “network node” may be used and may correspond to any type of radio network node or any network node, which communicates with a UE (directly or via another node) and / or with another network node. Examples of network nodes are NodeB, MeNB, ENB, a network node belonging to MCG or SCG, base station (BS), multi-standard radio (MSR) radio node such as MSR BS, eNodeB, gNodeB, network controller, radio network controller (RNC), base station controller (BSC), relay, donor node controlling relay, base transceiver station (BTS), access point (AP), transmission points, transmission nodes, RRU, RRH, nodes in distributed antenna system (DAS), core network node (e.g. Mobile Switching Center (MSC), Mobility Management Entity (MME), etc), Operations & Maintenance (O&M), Operations Support System (OSS), Self Optimized Network (SON), positioning node (e.g. Evolved- Serving Mobile Location Centre (E-SMLC)), Minimization of Drive Tests (MDT), test equipment (physical node or software), etc.

[0070] In some embodiments, the non-limiting term user equipment (UE) or wireless device may be used and may refer to any type of wireless device communicating with a network node and / or with another UE in a cellular or mobile communication system. Examples of UE are target device, device to device (D2D) UE, machine type UE or UE capable of machine to machine (M2M) communication, PDA, PAD, Tablet, mobile terminals, smart phone, laptop embedded equipped (LEE), laptop mounted equipment (LME), USB dongles, UE category Ml, UE category M2, ProSe UE, V2V UE, V2X UE, etc.

[0071] Additionally, terminologies such as base station / gNodeB and UE should be considered non-limiting and do in particular not imply a certain hierarchical relation between the two; in general, “gNodeB” could be considered as device 1 and “UE” could be considered as device 2 and these two devices communicate with each other over some radio channel. And in the following the transmitter or receiver could be either gNodeB (gNB), or UE.202501727

[0072] - 11 -

[0073] As will be appreciated by one skilled in the art, aspects of the embodiments may be embodied as a system, apparatus, method, or program product. Accordingly, embodiments may take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, resident software, micro-code, etc.) or an embodiment combining software and hardware aspects.

[0074] For example, the disclosed embodiments may be implemented as a hardware circuit comprising custom very-large-scale integration (“VLSI”) circuits or gate arrays, off-the-shelf semiconductors such as logic chips, transistors, or other discrete components. The disclosed embodiments may also be implemented in programmable hardware devices such as field programmable gate arrays, programmable array logic, programmable logic devices, or the like. As another example, the disclosed embodiments may include one or more physical or logical blocks of executable code which may, for instance, be organized as an object, procedure, or function.

[0075] Furthermore, embodiments may take the form of a program product embodied in one or more computer readable storage devices storing machine readable code, computer readable code, and / or program code, referred hereafter as code. The storage devices may be tangible, non- transitory, and / or non-transmission. The storage devices may not embody signals. In a certain embodiment, the storage devices only employ signals for accessing code

[0076] Any combination of one or more computer readable medium may be utilized. The computer readable medium may be a computer readable storage medium. The computer readable storage medium may be a storage device storing the code. The storage device may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, holographic, micromechanical, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing.

[0077] More specific examples (a non-exhaustive list) of the storage device would include the following: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random-access memory (“RAM”), a read-only memory202501727

[0078] - 12 -

[0079] (“ROM”), an erasable programmable read-only memory (“EPROM” or Flash memory), a portable compact disc readonly memory (“CD-ROM”), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the context of this document, a computer readable storage medium may be any tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device.

[0080] Code for carrying out operations for embodiments may be any number of lines and may be written in any combination of one or more programming languages including an object- oriented programming language such as Python, Ruby, Java, Smalltalk, C++, or the like, and conventional procedural programming languages, such as the “C” programming language, or the like, and / or machine languages such as assembly languages. The code may execute entirely on the user’s computer, partly on the user’s computer, as a stand-alone software package, partly on the user’s computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user’s computer through any type of network, including a local area network (“LAN”), wireless LAN (“WLAN”), or a wide area network (“WAN”), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider (“ISP”)).

[0081] Furthermore, the described features, structures, or characteristics of the embodiments may be combined in any suitable manner. In the following description, numerous specific details are provided, such as examples of programming, software modules, user selections, network transactions, database queries, database structures, hardware modules, hardware circuits, hardware chips, etc., to provide a thorough understanding of embodiments. One skilled in the relevant art will recognize, however, that embodiments may be practiced without one or more of the specific details, or with other methods, components, materials, and so forth. In other instances, well-known structures, materials, or operations are not shown or described in detail to avoid obscuring aspects of an embodiment. Reference throughout this specification to “one embodiment,” “an embodiment,” or similar language means that a particular feature, structure, or characteristic described in connection with the202501727

[0082] - 13 -

[0083] embodiment is included in at least one embodiment. Thus, appearances of the phrases “in one embodiment,” “in an embodiment,” and similar language throughout this specification may, but do not necessarily, all refer to the same embodiment, but mean “one or more but not all embodiments” unless expressly specified otherwise. The terms “including,” “comprising,” “having,” and variations thereof mean “including but not limited to,” unless expressly specified otherwise. An enumerated listing of items does not imply that any or all of the items are mutually exclusive, unless expressly specified otherwise. The terms “a,” “an,” and “the” also refer to “one or more” unless expressly specified otherwise.

[0084] Aspects of the embodiments are described below with reference to schematic flowchart diagrams and / or schematic block diagrams of methods, apparatuses, systems, and program products according to embodiments. It will be understood that each block of the schematic flowchart diagrams and / or schematic block diagrams, and combinations of blocks in the schematic flowchart diagrams and / or schematic block diagrams, can be implemented by code. This code may be provided to a processor of a general-purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the flowchart diagrams and / or block diagrams

[0085] The code may also be stored in a storage device that can direct a computer, other programmable data processing apparatus, or other devices to function in a particular manner, such that the instructions stored in the storage device produce an article of manufacture including instructions which implement the function / act specified in the flowchart diagrams and / or block diagrams.

[0086] The code may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus, or other devices to produce a computer implemented process such that the code which execute on the computer or202501727

[0087] - 14 -

[0088] other programmable apparatus provide processes for implementing the functions / acts specified in the flowchart diagrams and / or block diagrams.

[0089] The flowchart diagrams and / or block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of apparatuses, systems, methods, and program products according to various embodiments. In this regard, each block in the flowchart diagrams and / or block diagrams may represent a module, segment, or portion of code, which includes one or more executable instructions of the code for implementing the specified logical function(s).

[0090] It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the Figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. Other steps and methods may be conceived that are equivalent in function, logic, or effect to one or more blocks, or portions thereof, of the illustrated Figures.

[0091] Although various arrow types and line types may be employed in the flowchart and / or block diagrams, they are understood not to limit the scope of the corresponding embodiments. Indeed, some arrows or other connectors may be used to indicate only the logical flow of the depicted embodiment. For instance, an arrow may indicate a waiting or monitoring period of unspecified duration between enumerated steps of the depicted embodiment. It will also be noted that each block of the block diagrams and / or flowchart diagrams, and combinations of blocks in the block diagrams and / or flowchart diagrams, can be implemented by special purpose hardware-based systems that perform the specified functions or acts, or combinations of special purpose hardware and code.

[0092] The description of elements in each figure may refer to elements of proceeding figures. Like numbers refer to like elements in all figures, including alternate embodiments of like elements.202501727

[0093] - 15 -

[0094] The detailed description set forth below, with reference to the figures, is intended as a description of various configurations and is not intended to represent the only configurations in which the concepts described herein may be practiced. The detailed description includes specific details for the purpose of providing a thorough understanding of the various concepts. However, it will be apparent to those skilled in the art that these concepts may be practiced without these specific details. For instance, although 3GPP terminology, from e.g., 5G NR, may be used in this disclosure to exemplify embodiments herein, this should not be seen as limiting the scope of the present disclosure.

[0095] The disclosure is related to wireless communication system, which may be for example a 5G NR wireless communication system. More specifically, it represents a RAN of the wireless communication system, which is used exchange data with UEs via radio signals. For example, the RAN may send data to the UEs (downlink, DL), for instance data received from a core network (CN). The RAN may also receive data from the UEs (uplink, UL), which data may be forwarded to the CN.

[0096] In the examples illustrated, the RAN comprises one base station, BS. Of course, the RAN may comprise more than one BS to increase the coverage of the wireless communication system. Each of these BSs may be referred to as NB, eNodeB (or eNB), gNodeB (or gNB, in the case of a 5G NR wireless communication system), an access point or the like, depending on the wireless communication standard(s) implemented.

[0097] The UEs are located in a coverage of the BS. The coverage of the BS corresponds for example to the area in which UEs can decode a PDCCH transmitted by the BS.

[0098] An example of a wireless device suitable for implementing any method, discussed in the present disclosure, performed at a UE corresponds to an apparatus that provides wireless connectivity with the RAN of the wireless communication system, and that can be used to exchange data with said RAN. Such a wireless device may be included in a UE. The UE may for instance be a cellular phone, a wireless modem, a202501727

[0099] - 16 -

[0100] wireless communication device, a handheld device, a laptop computer, or the like. The UE may also be an Internet of Things (loT) equipment, like a wireless camera, a smart sensor, a smart meter, smart glasses, a vehicle (manned or unmanned), a global positioning system device, etc., or any other equipment that may run applications that need to exchange data with remote recipients, via the wireless device.

[0101] The wireless device comprises one or more processors and one or more memories. The one or more processors may include for instance a central processing unit (CPU), a digital signal processor (DSP), a field-programmable gate array (FPGA), an application specific integrated circuit (ASIC), etc. The one or more memories may include any type of computer readable volatile and non-volatile memories (magnetic hard disk, solid-state disk, optical disk, electronic memory, etc.). The one or more memories may store a computer program product, in the form of a set of programcode instructions to be executed by the one or more processors to implement all or part of the steps of a method for exchanging data, performed at a UE’s side, according to any one of the embodiments disclosed herein.

[0102] The wireless device can comprise also a main radio, MR, unit. The MR unit corresponds to a main wireless communication unit of the wireless device, used for exchanging data with BSs of the RAN using radio signals. The MR unit may implement one or more wireless communication protocols, and may for instance be a 3G, 4G, 5G, NR, WiFi, WiMax, etc. transceiver or the like. In preferred embodiments, the MR unit corresponds to a 5G NR wireless communication unit.

[0103] AI / ML Model is a data driven algorithm that applies AI / ML techniques to generate set of outputs based on set of inputs.

[0104] AI / ML model delivery is a generic term referring to delivery of an AI / ML model from one entity to another entity in any manner. Note is An entity could mean network node / function (e.g., gNB, LMF, etc.), UE, proprietary server, etc.

[0105] AI / ML model Inference is a process of using trained AI / ML model to produce set of outputs based on set of inputs.202501727

[0106] - 17 -

[0107] AI / ML model testing is a subprocess of training, to evaluate the performance of final AI / ML model using dataset different from one used for model training and validation. Differently from AI / ML model validation, testing does not assume subsequent tuning of the model.

[0108] AI / ML model training is a process to train an AI / ML Model [by learning the input / output relationship] in data driven manner and obtain the trained AI / ML Model for inference.

[0109] AI / ML model transfer is a delivery of an AI / ML model over the air interface in manner that is not transparent to 3GPP signaling, either parameters of model structure known at the receiving end or new model with parameters. Delivery may contain full model or partial model.

[0110] AI / ML model validation is a subprocess of training, to evaluate the quality of an AI / ML model using dataset different from one used for model training, that helps selecting model parameters that generalize beyond the dataset used for model training.

[0111] Data collection is a process of collecting data by the network nodes, management entity, or UE for the purpose of AI / ML model training, data analytics and inference.

[0112] Federated learning I federated training is a machine learning technique that trains an AI / ML model across multiple decentralized edge nodes e.g., UEs, gNBs each performing local model training using local data samples. The technique requires multiple interactions of the model, but no exchange of local data samples.

[0113] Functionality identification is a process / method of identifying an AI / ML functionality for the common understanding between the NW and the UE. Note is Information regarding the AI / ML functionality may be shared during functionality identification. Where AI / ML functionality resides depends on the specific use cases and sub use cases.202501727

[0114] - 18 -

[0115] Model activation means enable an AI / ML model for specific AI / ML-enabled feature.

[0116] Model deactivation means disable an AI / ML model for specific AI / ML-enabled feature.

[0117] Model download means Model transfer from the network to UE.

[0118] Model identification is A process / method of identifying an AI / ML model for the common understanding between the NW and the UE. The process / method of model identification may or may not be applicable and regarding the AI / ML model may be shared during model identification.

[0119] Model monitoring is A procedure that monitors the inference performance of the AI / ML model.

[0120] Model parameter update is Process of updating the model parameters of model. Model selection is the process of selecting an AI / ML model for activation among multiple models for the same AI / ML enabled feature. Model selection may or may not be carried out simultaneously with model activation.

[0121] Model switching is deactivating currently active AI / ML model and activating different AI / ML model for specific AI / ML-enabled feature.

[0122] Model update is Process of updating the model parameters and / or model structure of model.

[0123] Model upload is Model transfer from UE to the network.

[0124] Network-side (AI / ML) model is an AI / ML Model whose inference is performed entirely at the network.

[0125] Offline field data is the data collected from field and used for offline training of the AI / ML model.202501727

[0126] - 19 -

[0127] Offline training is an AI / ML training process where the model is trained based on collected dataset, and where the trained model is later used or delivered for inference. Note is This definition only serves as guidance. There may be cases that may not exactly conform to this definition but could still be categorized as offline training by commonly accepted conventions.

[0128] Online field data is the data collected from field and used for online training of the AI / ML model.

[0129] Online training is an AI / ML training process where the model being used for inference) is (typically continuously) trained in (near) real-time with the arrival of new training samples. Note is the notion of (near) real-time vs. non real-time is context-dependent and is relative to the inference time-scale. This definition only serves as guidance.

[0130] There may be cases that may not exactly conform to this definition but could still be categorized as online training by commonly accepted conventions. Note is Fine-tuning / re-training may be done via online or offline training. This note could be removed when we define the term fine-tuning.

[0131] Reinforcement Learning (RL) is a process of training an AI / ML model from input (a.k.a. state) and feedback signal (a.k.a. reward) resulting from the model’s output (a.k.a. action) in an environment the model is interacting with.

[0132] Semi-supervised learning is a process of training model with mix of labelled data and unlabelled data.

[0133] Supervised learning is a process of training model from input and its corresponding labels.

[0134] Two-sided (AI / ML) model is a paired AI / ML Model(s) over which joint inference is performed, where joint inference comprises AI / ML Inference whose inference is performed jointly across the UE and the network, i.e, the first part of inference is202501727

[0135] - 20 -

[0136] firstly performed by UE and then the remaining part is performed by gNB, or vice versa.

[0137] UE-side (AI / ML) model is an AI / ML Model whose inference is performed entirely at the UE.

[0138] Unsupervised learning is a process of training model without labelled data.

[0139] Proprietary-format models is ML models of vendor-Zdevice-specific proprietary format, from 3GPP perspective. They are not mutually recognizable across vendors and hide model design information from other vendors when shared.

[0140] Open-format models is ML models of specified format that are mutually recognizable across vendors and allow interoperability, from 3GPP perspective. They are mutually recognizable between vendors and do not hide model design information from other vendors when shared.

[0141] The following explanation will provide the detailed description of the mechanism about pre-configuring and signaling the specific information about model online training by configuring a set of UE behaviors. AI / ML based techniques are currently applied to many different applications and 3GPP also started to work on its technical investigation to apply to multiple use cases based on the observed potential gains. AI / ML lifecycle can be split into several stages such as data collection / pre-processing, model training, model testing / validation, model deployment / update, model monitoring etc., where each stage is equally important to achieve target performance with any specific model(s). In applying AI / ML model for any use case or application, one of the challenging issues is to manage the lifecycle of AI / ML model. It is mainly because the data / model drift occurs during model deployment and inference and it results in performance degradation of AI / ML model. Fundamentally, the dataset statistical changes occur after model is deployed and model inference capability is also impacted with unseen data as input.202501727

[0142] - 21 -

[0143] In a similar aspect, the statistical property of dataset and the relationship between input and output for the trained model can be changed with drift occurrence. In this context, model training or re-training is one of key issues for model performance maintenance as model performance such as inferencing and / or training is dependent on different model execution environment with varying configuration parameters. To handle this issue, collaboration between UE and gNB is highly important to track model performance and re-configure model corresponding to different environments. AI / ML model needs model monitoring after deployment because model performance cannot be maintained continuously due to drift and update feedback is then provided to re-train / update the model or select alternative model.

[0144] When AI / ML model enabled wireless communication network is deployed, it is then important to consider how to handle AI / ML model in activation with re-configuration for wireless devices under operations such as model training, inference, updating, etc. For determining information about model identification (e.g., model ID), frequent model ID assignment / re-assignment processes might occur due to model drift related to model performance variation and / or model applicable condition change. In addition, if all identified models at UE side are to be assessed and monitored, the associated signaling overhead and computing power demand can increase significantly.

[0145] And signaling overhead can highly increase for model ID decision and assignment. When there are multiple models for model transfer / delivery, significant increase of signaling overhead and the limited UE capability supporting models to be transferred can be critical for the deployment of target models at UE side. In this method, a two-step messaging approach is proposed to facilitate efficient AI / ML model switching and / or activation in a wireless communication system. This mechanism ensures that multiple candidate models are preloaded onto a device (e.g., UE) or their associated model IDs are obtained in advance while deferring actual model switching and / or activation until triggered, thereby reducing latency and optimizing model selection. The proposed two-step messaging consists of candidate model indication message (that notifies the UE about available candidate models and their activation conditions)202501727

[0146] - 22 -

[0147] and target model activation message (that triggers the activation of a specific model when necessary).

[0148] The candidate model indication message is transmitted periodically or event-triggered to ensure AI / ML model readiness at the UE. However, target model activation message is allowed to be skipped when UE activates a specific model autonomously without receiving model activation signaling. In this case, candidate model indication message is only sent to UE for model switching and / or activation. Depending on different AI / ML model-based applications and collaboration levels of network-UE sides, two-step messaging is adjusted by sending candidate model indication message only or both candidate model indication message and target model activation message.

[0149] This approach enables adaptive AI / ML model selection while minimizing the impact on radio resource consumption and processing overhead at both the network side and UE side. Regarding candidate model indication message, it is transmitted from the gNB to the UE in advance to provide a list of candidate AI / ML models or their associated model IDs that may be utilized in the future under specific conditions. The candidate model indication message includes a set of model IDs (that have been preapproved for use by the UE, where each model ID corresponds to an AI / ML model available for specific conditions), model activation conditions (that defines the conditions under which a model switch and / or activation should occur, where the exemplary conditions include radio quality metric, UE mobility state, device computational resource constraints, etc.), and priority levels of candidate models for model selection.

[0150] When multiple models satisfy activation conditions, a priority level is used to determine the preferred model and priority is based on factors such as model properties, latency, energy efficiency, or accuracy of model inference. The candidate model indication message is delivered via layer 1 (L1), layer 2 (L2), or radio resource control (RRC) signaling. In addition, HandoverRequest message is utilized to send the candidate model indication message in mobility scenarios where the UE is expected to switch to a new cell by ensuring that the target gNB is aware of the UE’s202501727

[0151] - 23 -

[0152] AI / ML model capabilities before handover. If necessary, one or more AI / ML candidate models are allowed to be pre-transferred so that one or more models can be preloaded at UE in advance when model transfer is required. Therefore, when one or more AI / ML candidate models are pre-transferred, model transfer follows after the candidate model indication message is transmitted. However, model transfer such as pre-transferring models might not always be needed since obtaining candidate AI / ML model ID information can be enough for model switching and / or activation at UE. Regarding target model activation message, it is transmitted when the gNB determines that a specific AI / ML model should be activated.

[0153] This mechanism ensures that the network decides AI / ML model selection for UE while allowing rapid activation. The target model activation message includes target model ID(s) (by specifying the exact AI / ML model that should be activated on the UE and the selected model ID is chosen from the list provided in the candidate model indication message), the updated activation conditions (that contains refined conditions to justify the activation of the selected model, where exemplary conditions are handover execution with UE movement to a new cell requiring a model update, deteriorating radio link below a predefined threshold, or network traffic change with an increase of latency-sensitive applications requiring a different AI / ML model), and additional fine-tuned model parameters (if applicable, when the gNB may provide model-specific hyperparameters to fine-tune the AI / ML model’s operation as parameters may be related to model execution constraints based on processing capability). The target model activation message is delivered via L1 , L2, or RRC signaling. For example, MAC CE is used for scenarios where immediate model switching and / or activation is required, or RRCReconfiguration is used when additional AI / ML model parameters need to be configured along with model activation.

[0154] To support two-step messaging for model switching and / or activation, the unified model identification framework is required to ensure seamless AI / ML model exchange between different network vendors and UE device manufacturers and standardize model ID mapping allowing consistent identification across multi-vendor environments. Specifically, the unified model ID format consist of model ID fields =202501727

[0155] -24 -

[0156] {type, architecture, version, parameter, mapping}. Depending on different model ID transmission use cases or deployment scenarios, different combinations of fields in model ID format are used to represent model ID information. For each field of model ID format, they are also represented as the pre-configured identifiers as indexed information. RRC signaling protocol is used to deliver model ID format information and the associated configuration information. In addition, any further updates about model ID format information can be also sent to UE via L1 or L2 signaling. For example, model ID including fields of {type, version, mapping} is sent to UE by gNB. In this case, information of other fields (e.g., {architecture, parameter}) might be inferred using a pre-configured information at UE side or already available at UE side in advance (e.g., pre-stored using an internal lookup table). As another example, UE can retrieve the corresponding model structure and parameters based on the mapping field information via repository if necessary. On the other hand, model ID can be separately assigned with unique identifier while the field information in model ID fields can be allowed to maintain mapping relation with each separate model ID. UE can then infer missing model ID using pre-configured information or internal lookup tables based on the received combination of field information of a specific model ID format.

[0157] Figure 1 shows an exemplary table of two-step messaging. In this example, the table describes that the candidate model indication message is used to preload multiple AI / ML models at the UE in advance and the target model activation message is sent only when the network determines that a specific model needs to be activated. The candidate model indication message is transmitted proactively during RRC reconfiguration or handover preparation or via L1 or L2 signaling, ensuring that the UE has the necessary AI / ML models before they are needed. Meanwhile, the target model activation message is sent reactively, triggered by specific conditions. The candidate model indication message contains a list of candidate models, including their model IDs, activation conditions, and priority levels. This allows the UE to store potential model information in advance. On the other hand, the target model activation message specifies the exact model to be activated, along with updated activation conditions and fine-tuned model parameters to optimize performance. The candidate model indication message enables UE-side model caching, reducing202501727

[0158] - 25 -

[0159] activation latency and allowing for quicker AI / ML inference when needed. The target model activation message ensures network-driven AI / ML model adaptation.

[0160] Figure 2 shows an exemplary signaling flow of network decision-based model switching and / or activation. In this example, the candidate model indication (as Step- 1 message) is sent by network side (e.g., gNB) to UE so that model switching and / or activation is monitored based on the configured conditions. UE then reports the model monitoring measurement data to network side so that model switching and / or activation is allowed to be decided by network. The target model activation (as Step- 2 message) is sent by network side (e.g., gNB) to UE so that a specific model is activated or switched based on reception of the network side decision.

[0161] Figure 3 shows an exemplary signaling flow of UE decision-based model switching and / or activation. In this example, the target model activation (as Step-2 message) is not applied and the candidate model indication (as Step-1 message) is only utilized so that model switching and / or activation is decided by UE autonomously. Therefore, UE executes model monitoring, target model selection from candidate models, and model switching and / or activation on UE side without network side decision indication. After autonomous decision made by UE, model switching and / or activation status is reported back to network side.

[0162] Figure 4 shows an exemplary flow chart of configuring conditional model switching and / or activation at network side. In this example, the ML configuration information about conditional model switching and / or activation is generated on network side for model transfer and activation. For model transfer initiation, the network side selects a set of candidate AI / ML models and indication message (that contains model IDs, model activation conditions, model priority levels) is transmitted to UE. The UE receives and stores models while awaiting condition matching. When a condition is met (e.g., radio link quality drop, computational load change), UE autonomously activates a stored model among a set of candidate AI / ML models given by network side. The UE notifies the gNB about UE autonomous model activation and / or switching (e.g., via MAC-CE or RRC signaling).202501727

[0163] - 26 -

[0164] Figure 5 shows an exemplary flow chart of the autonomous decision procedure for conditional model activation at the UE. In this example, the autonomous decision procedure for conditional model switching and / or activation at the UE is shown. For model switching and / or activation, UE monitors real-time conditions (e.g., SINR, battery level, CPU load). UE checks the pre-configured model activation conditions received from the gNB. If conditions are met, the best-matching model is activated. If multiple models are qualified in parallel as candidate model for model switching and / or activation, prioritization level is referenced for a specific model selection when each model has their associated priority level indication. UE reports model switching and / or activation status update to network side. If no model switching and / or activation is needed, monitoring conditions continue.

[0165] Figure 6 shows an exemplary table of model ID framework. In this example, the unified model identification framework is described with fields. Specifically, the unified model ID format consist of model ID fields = {type, architecture, version, parameter, mapping}. Depending on different model ID transmission use cases or deployment scenarios, different combinations of fields in model ID format are used to represent model ID information. For each field of model ID format, they are also represented as the pre-configured identifiers as indexed information. For type field, it indicates model functionality, model task, or model category such as channel estimation model, QoS prediction model, beamforming prediction model, or network traffic model, etc. For architecture field, the model architecture field is a key attribute that specifies the AI / ML model identification framework used to define and execute the model and this information is crucial for ensuring compatibility between the network side and UE side, allowing the UE to correctly interpret and execute the received AI / ML model. This field is identified to indicate how model is structured and implemented by including specific model information such as LSTM, CNN, RNN, or transformer in association with ONNX, TensorFlow, or PyTorch, etc.

[0166] For version field, the model version field specifies the version of the AI / ML model deployed in the network and this helps ensure consistency between the model stored at network entity and the model available at the UE by ensuring that the UE and network entity use compatible model versions for model operation. Any mismatches202501727

[0167] - 27 -

[0168] between outdated and updated models can be prevented, reducing model performance errors. For parameter field, the model parameter field specifies key configuration parameters of the AI / ML model that affect its behavior and decisionmaking process. These parameters define how the model operates under different conditions and allows dynamic fine-tuning of model behavior with parameter adaptation without requiring full model re-transfer. Specifically, key configuration parameters of the AI / ML model is allowed to be the indexed values by including input feature set, inference time window, learning rate, etc. For mapping field, the model mapping field establishes a mapping index that associates a model ID with specific model operation conditions, UE types, or model applications so that it allows flexible assignment of AI / ML models depending on real-time requirements for model transfer, model switching, or model activation. For example, MapJD is assigned as Map_ID_1. Map_ID_1 is then mapped onto a specific model and it is derived from the configured relationship with any specific UE type, model application, and / or additional conditions.

[0169] Based on the mapping field information, the corresponding model structure and parameters is also retrieved if necessary. Dynamic model retrieval based on the mapping field is enabled. On the other hand, model ID can be separately assigned with unique identifier while the field information in model ID fields can be allowed to maintain mapping relation with each separate model ID. The proposed method above contributes to lower latency in AI / ML model switching and / or activation (real-time adaptation at UE level) and efficient bandwidth utilization by minimizing redundant model transfers, dynamically adapting to different conditions (e.g., network-level, device-level, model operation-level, and environmental-level).

Claims

202501727- 28 -CLAIMS1. A method for AI / ML model switching and / or activation at network entity in a wireless communication system, comprising:• preloading multiple AI / ML models onto a UE or pre-obtaining their associated model identifiers;• transmitting, from a network entity to the UE, a candidate model indication message containing a list of candidate AI / ML models, associated model IDs, model activation conditions, and priority levels;• monitoring, by the UE, the network and device conditions to determine whether an AI / ML model switch and / or activation is required;• selectively transmitting, from the network entity to the UE, a target model activation message to activate a specific AI / ML model when the activation conditions are met; and• allowing the UE to autonomously switch or activate an AI / ML model without receiving the target model activation message based on pre-configured conditions.

2. The method of claim 1 , wherein the candidate model indication message is transmitted via a L1, L2, or RRC signaling.

3. The method according to one of the previous claims, wherein the candidate model indication message is included in a HandoverRequest message to facilitate AI / ML model preloading and activation in mobility scenarios.

4. The method according to one of the previous claims, wherein the target model activation message comprises:• a target model identifier selected from the candidate models;• updated activation conditions specifying when the target model should be activated; and• fine-tuned model parameters for optimizing the AI / ML model's execution.202501727- 29 -5. The method according to one of the previous claims, wherein the target model activation message is transmitted via a L1, L2, or RRC signaling.

6. The method according to one of the previous claims, wherein the candidate model indication message is transmitted periodically or event-triggered to ensure AI / ML model readiness at the UE.

7. The method according to one of the previous claims, wherein the two-step messaging approach optimizes radio resource usage by minimizing redundant AI / ML model transfers and reducing network overhead in AI / ML-driven wireless communication environments.

8. The according to one of the previous claims, wherein further comprising:• a network entity configured to transmit a candidate model indication message and a target model activation message; and• a user equipment (UE) configured to receive and process AI / ML model indication messages, monitor activation conditions, and autonomously switch AI / ML models as required.

9. A method for the unified model identification framework, comprising:• defining a model ID format consisting of fields: {type, architecture, version, parameter, mapping};• transmitting a model ID including a subset of said fields from a network entity to a UE;• allowing the UE to infer missing model ID fields using pre-configured information or internal lookup tables; and• enabling dynamic model retrieval based on the mapping field.

10. The according to claim 9, wherein the type field indicates model functionality, model task, or model category such as channel estimation model and / or QoS prediction model and / or beamforming prediction model and / or network traffic model.202501727- 30 -11. The method according to one of the previous claims 9 to 10, wherein the model architecture field specifies the AI / ML model identification framework to define and execute the model via indication of how model is structured and implemented, ensuring compatibility between the network side and UE side and allowing the UE to correctly interpret and execute the received AI / ML model.

12. The method according to one of the previous claims 9 to 11 , wherein the version field specifies the version of the AI / ML model deployed in the network by ensuring consistency between the model stored at network entity and the model available at the UE with use of compatible model versions for model operation.

13. The method according to one of the previous claims 9 to 12, wherein the parameter field specifies key configuration parameters of the AI / ML model that affect its behavior and decision-making process, defining how the model operates under different conditions and allowing dynamic fine-tuning of model behavior with parameter adaptation without requiring full model re-transfer.

14. The method according to one of the previous claims 9 to 13, wherein the mapping field establishes a mapping index that associates a model ID with specific model operation conditions, UE types, or model applications by allowing flexible assignment of AI / ML models for model transfer, model switching, or model activation.

15. The method according to one of the previous claims 9 to 14, model ID is allowed to be separately assigned with unique identifier while the field information in model ID format maintains mapping relation with each separate model ID.

16. The method according to one of the previous claims 9 to 15, further comprising:• a model repository storing AI / ML models accessible based on model mapping information; and• a signaling mechanism utilizing L1 , L2, or RRC message to transmit AI / ML model information such as a specific model ID format and the associated fields.202501727- 31 -17. Apparatus for AI / ML model switching and / or activation at network entity in a and / or for the unified model identification framework wireless communication system the apparatus comprising a wireless transceiver, a processor coupled with a memory in which computer program instructions are stored, said instructions being configured to implement steps of the claims 1 to 16.

18. User Equipment comprising an apparatus according to claim 17.

19. gNB comprising an apparatus according to claim 17.

20. Wireless communication system for model switching and / or activation at network entity in a and / or for the unified model identification framework, wherein the wireless communication systems comprises at least a user equipment according to claim 18, at least a gNB according to claim 19, whereby the user Equipment and the gNB each comprises a processor coupled with a memory in which computer program instructions are stored, said instructions being configured to carry out the steps according of the claims 1 to 16.

Images