Method for matching signaling with AI / ML model

Abstract

The present application describes a method for data-driven AI / ML model signaling for pairable model search signaling based on AI / ML, wherein techniques are disclosed for pre-configuration and signaling with respect to model pairing searches in machine learning model operations in wireless mobile communication systems, including base stations (e.g., gNBs) and mobile stations (e.g., UEs). A minimum set of model information is provided for matching by searching for pairable models through a model pairing search process.

Description

Technical Field

[0001] This disclosure relates to AI / ML-based pairable model search signaling, wherein techniques for pre-configuration and signaling for model pair search in machine learning model operations are proposed. Background Technology

[0002] Within 3GPP (3rd Generation Partnership Project), one of the selected research projects as part of the approved Release 18 package is AI / ML (Artificial Intelligence / Machine Learning), as described in the relevant document (RP-213599) submitted at 3GPP TSG RAN (Technical Specification Group for Radio Access Networks) meeting #94e. The formal title of the AI / ML research project is "Study on AI / ML for NR Air Interface," and RAN WG1 and WG2 are currently actively developing specifications. The goal of this research project is to identify a general AI / ML framework and areas where AI / ML-based technologies and use cases can yield benefits.

[0003] According to 3GPP, the main objective of this research project is to study AI / ML frameworks for air interfaces by considering performance, complexity, and potential specification impacts using target use cases. Specifically, the AI / ML models, terminology, and descriptions used to establish common and specific characteristics of the framework will be a key area of ​​work. Regarding AI / ML frameworks, various aspects are being investigated, and a key project concerns the lifecycle management of AI / ML models, which mandatorily includes multiple phases such as model training, model deployment, model inference, model monitoring, and model updates.

[0004] Earlier, in 3GPP TR 37.817, version 17, entitled "Study on enhancement for Data Collection for NR and EN-DC," UE mobility was also considered as one of the AI / ML use cases, and one scenario for model training / inference was that both functions resided within the RAN node. Subsequently, in version 18, a new work project, "Artificial Intelligence (AI) / Machine Learning (ML) for NG-RAN," was initiated to specify data collection enhancements and signaling support within the existing NG-RAN interface and architecture.

[0005] In 3GPP AI / ML discussions, several working assumptions exist regarding the terminology. For example, a bilateral (AI / ML) model is a paired AI / ML model on which joint inference is performed, where joint inference includes AI / ML inference jointly performed across the UE and network, i.e., the first part of the inference is performed by the UE first and then the remainder by the gNB, or vice versa. Proprietary format models are vendor / device-specific proprietary format ML models from a 3GPP perspective, and they are not mutually identifiable across vendors and hide model design information from other vendors when shared. Open format models are specified format ML models that are mutually identifiable across vendors and allow interoperability from a 3GPP perspective, and they are mutually identifiable between vendors and do not hide model design information from other vendors when shared.

[0006] For the standardization of the aforementioned activities, there are currently no specifications defining the following signaling methods or network (e.g., gNB, BS) / mobile station (e.g., UE) behavior: AI / ML model operations that support model pairing between two nodes when multiple LCM (Lifecycle Management) operations (such as model training and inference and / or updates) are enabled on the device for one or more specific application scenarios with different sets of ML features / functionalities. Configuring bilateral models is very difficult when model identifiers (e.g., model IDs) are not pre-configured for model pairing, and significant configuration delays can occur in finding matching model pairs.

[0007] US 2021 344 469 A1 demonstrates that a machine learning model is implemented in the UE, and the UE estimates one or more features of the second frequency band based on the measurements it performs on a reference signal transmitted from a dedicated transmitter in the first frequency band.

[0008] US 2022 149 980 A1 demonstrates a method for dynamically selecting link adaptation strategies, wherein the method includes using channel quality information, additional information, and an ML model to select a link adaptation strategy from a predefined set of link adaptation strategies.

[0009] US 2022 150 727 A1 demonstrates a technique for sharing machine learning models between wireless nodes (such as UEs and BSs) and indicating the transmitting and receiving points (TRPs) to which the machine learning models are applicable.

[0010] US 2023 153 408 A1 demonstrates a method that includes the following steps: creating a machine learning model, using the machine learning model being trained, calculating and updating the parameters of the machine learning model based on the calculated loss. Attached Figure Description

[0011] The disclosed invention will now be discussed further based on the preferred embodiments presented in the accompanying drawings. However, the disclosed invention may be embodied in many different forms and should not be construed as limited to the preferred embodiments described. Rather, the preferred embodiments are provided for exhaustiveness and completeness, and to fully convey the scope of the invention to those skilled in the art. The following detailed description refers to the accompanying drawings, in which:

[0012] Figure 1 This is an exemplary block diagram of a blind pairing search method for ML models;

[0013] Figure 2 This is an exemplary block diagram of the minimum model information set table;

[0014] Figure 3 It is the signaling process for searching candidate matchable models for model matching;

[0015] Figure 4 This is a flowchart of the blind matching search process on the network side;

[0016] Figure 5 This is a flowchart of the blind pairing search process on the UE side;

[0017] Figure 6 It is a flowchart of the process of determining the combination of model pairs; and

[0018] Figure 7 This is a flowchart of the process for determining the type of candidate pairable models. Detailed Implementation

[0019] The specific embodiments described below with reference to the accompanying drawings are intended as descriptions of various configurations and are not intended to represent only configurations in which the concepts described herein can be practiced. The detailed description includes specific details and is intended to provide a thorough understanding of the various concepts. However, it will be apparent to those skilled in the art that these concepts can be practiced without these specific details. Specifically, although terms from 3GPP 5G NR may be used in this disclosure to exemplify embodiments herein, this should not be construed as limiting the scope of the invention.

[0020] Some embodiments of the ideas contemplated herein will now be described more fully with reference to the accompanying drawings. However, other embodiments are also included within the scope of the subject matter disclosed herein, and the disclosed subject matter should not be construed as being limited to 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.

[0021] Generally, all terms used herein should be interpreted according to their common meaning in the relevant art, unless the context in which the term is used explicitly gives and / or implies a different meaning. Unless otherwise expressly stated, all references to a / an / element, device, component, means, step, etc., are openly interpreted as referring to at least one instance of that element, device, component, means, step, etc. The steps of any method disclosed herein need not be performed in the exact order disclosed, unless a step is explicitly described as occurring after or before another step and / or where an implicit step must occur after or before another step. Where appropriate, any feature of any embodiment of the embodiments disclosed herein may be applied to any other embodiment. Similarly, any advantage of any embodiment of this embodiment may be applied to any other embodiment, and vice versa. Other objects, features, and advantages of the appended embodiments will be apparent from the following description.

[0022] In some embodiments, the more general term "network node" may be used, and this term may correspond to any type of radio network node or any network node that communicates with a UE (directly or via another node) and / or with another network node. Examples of network nodes are NodeB, MeNB, ENB, network nodes belonging to MCG or SCG, base station (BS), multi-standard radio (MSR) radio nodes (such as MSR BS), eNodeB, gNodeB, network controller, radio network controller (RNC), base station controller (BSC), relay, donor node control relay, base transceiver (BTS), access point (AP), transport point, transport node, RRU, RRH, nodes in distributed antenna system (DAS), core network nodes (e.g., mobile switching center (MSC), mobility management entity (MME), etc.), operations and maintenance (O&M), operations support system (OSS), self-optimizing network (SON), location node (e.g., evolved serving mobile location center (E-SMLC)), minimized drive test (MDT), test equipment (physical node or software), another UE, etc.

[0023] In some embodiments, the non-limiting terms User Equipment (UE) or Wireless Device may be used, and the term may refer to any type of wireless device that communicates with a network node and / or with another UE in a cellular or mobile communication system. Examples of UEs include target devices, device-to-device (D2D) UEs, machine-type UEs or UEs capable of machine-to-machine (M2M) communication, PDAs, PADs, tablet computers, mobile terminals, smartphones, laptop embedded devices (LEE), laptop mounted devices (LME), USB dongles, UE class M1, UE class M2, ProSe UE, V2V UE, V2X UE, etc.

[0024] Furthermore, terms such as base station / gNodeB and UE should be considered non-restrictive and, in particular, do not imply any hierarchical relationship between the two; generally speaking, "gNodeB" can be considered device 1 and "UE" can be considered device 2, and the two devices communicate with each other via a radio channel. And in the following text, the transmitter or receiver can be either a gNodeB (gNB) or a UE.

[0025] As those skilled in the art will understand, aspects of the embodiments may be embodied as systems, devices, methods, or computer program products. Therefore, embodiments may take the form of entirely hardware embodiments, entirely software embodiments (including firmware, resident software, microcode, etc.), or embodiments combining software and hardware aspects.

[0026] For example, the disclosed embodiments can be implemented as hardware circuitry, including custom-designed very large-scale integration (“VLSI”) circuitry or gate arrays, off-the-shelf semiconductors (such as logic chips, transistors, or other discrete components). The disclosed embodiments can also be implemented in programmable hardware devices such as field-programmable gate arrays, programmable array logic, programmable logic devices, etc. As another example, the disclosed embodiments may include one or more physical or logical blocks of executable code, which may, for example, be organized as objects, procedures, or functions.

[0027] Furthermore, embodiments may take the form of a computer program product embodied in one or more computer-readable storage devices, which store machine-readable code, computer-readable code, and / or program code, hereinafter referred to as code. The storage device may be tangible, non-transitory, and / or non-transferable. The storage device may not contain signals. In one embodiment, the storage device uses only signals to access the code.

[0028] Any combination of one or more computer-readable media may be used. A computer-readable medium may be a computer-readable storage medium. A computer-readable storage medium may be a storage device for storing code. A storage device may be, for example, but not limited to, electronic, magnetic, optical, electromagnetic, infrared, holographic, micromechanical, or semiconductor systems, devices, or apparatuses, or any suitable combination of the foregoing.

[0029] Further specific examples of storage devices (a non-exhaustive list) will include the following: electrical connections having one or more wires, portable computer floppy disks, hard disks, random access memory (“RAM”), read-only memory (“ROM”), erasable programmable read-only memory (“EPROM” or flash memory), portable optical disc read-only memory (“CD-ROM”), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. In the context of this document, a computer-readable storage medium can be any tangible medium that can contain or store programs for use by or in conjunction with an instruction execution system, device, or apparatus.

[0030] The code used to perform the operations of the embodiments can be any number of lines and can be written in any combination of one or more programming languages, including object-oriented programming languages ​​(such as Python, Ruby, Java, Smalltalk, C++, etc.), as well as conventional procedural programming languages ​​(such as the "C" programming language, etc.) and / or machine languages ​​(such as assembly language). The code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer via any type of network (including a local area network ("LAN"), a wireless LAN ("WLAN"), or a wide area network ("WAN"), or can be connected to an external computer (e.g., via the Internet through an Internet service provider ("ISP").

[0031] Furthermore, the features, structures, or characteristics described in the embodiments can be combined in any suitable manner. Numerous specific details, such as examples of programming, software modules, user selection, network transactions, database queries, database structures, hardware modules, hardware circuits, hardware chips, etc., are provided in the following description to provide a thorough understanding of the embodiments. However, those skilled in the art will recognize that the embodiments can be practiced without one or more of these specific details or using other methods, components, materials, etc. In other instances, well-known structures, materials, or operations are not shown or described in detail to avoid obscuring aspects of the embodiments. Throughout the specification, references to “an embodiment,” “embodiment,” or similar language mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment. Therefore, unless expressly specified otherwise, the phrases “in one embodiment,” “in an embodiment,” and similar language appearing throughout the specification may, but not necessarily all, refer to the same embodiment, but rather 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, the enumeration of items does not imply that any or all items in the item are mutually exclusive. Unless otherwise expressly specified, the terms “a”, “an” and “the” also mean “one or more”.

[0032] According to a first aspect of the present invention, a method for AI / ML model matching signaling between a first node and a second node in a wireless network to support an AI / ML model pairing process between model pairs consisting of a first AI / ML model of the first node and a second AI / ML model of the second node, comprising the step of receiving ML configuration information indicating a model pairing request matching the first AI / ML model. Further, the method includes the step of identifying a finite list of candidate pairable models. Further, the method includes the step of generating at least one minimum model information set for each candidate pairable model, each minimum model information set containing descriptive information of its corresponding candidate pairable model. Further, the method includes the step of receiving the at least one minimum model information set at the first node. Further, the method includes the step of using the at least one minimum model information set to determine the second AI / ML model as the candidate pairable model, the candidate pairable model forming an optimal combination with the first AI / ML model.

[0033] Advantageously, the determination of the second AI / ML model is based on performance measurements of the target performance metrics of the model pair.

[0034] Advantageously, the minimum model information set contains information about model functionality, and / or model inputs / outputs, and / or model descriptions, and / or model metadata.

[0035] Advantageously, the determination of the second AI / ML model employs a blind pairwise search.

[0036] Advantageously, the at least one minimal model information set is transmitted from the second node to the first node via L1, L2 or L3 signaling.

[0037] Advantageously, the minimum set of model information is indexed in a pre-configured format to support the generation of a finite list of candidate models.

[0038] Advantageously, the minimum model information set contains multiple tables.

[0039] Advantageously, the candidate pairable model is classified as a general reference model or a special reference model, and the method further includes the following steps:

[0040] • Receive at least one pre-configured range regarding the at least one minimum model information set.

[0041] If the at least one model information set is below the pre-configured range, then the general reference model is selected as the second AI / ML model.

[0042] • If the minimum model information set is higher than the pre-configured range, then the dedicated reference model is selected as the second AI / ML model.

[0043] Advantageously, the number of candidate pairable models for testing model performance can be adjusted based on implementation settings.

[0044] Advantageously, if no model pair provides sufficient target performance metrics, a search failure is issued to enable alternative AI / ML configurations or a fallback operation.

[0045] Advantageously, nodes can be UEs, networks, base stations, or combinations thereof, and communication between nodes can be performed via L1, L2, or L3 signaling, or based on sidelink communication.

[0046] According to a second aspect of the invention, an apparatus for AI / ML model matching signaling includes a processor coupled to a memory storing computer program instructions configured to implement the steps of the method described above.

[0047] According to a third aspect of the present invention, a user equipment includes the above-described equipment, characterized in that it performs the following steps: receiving the ML configuration information, identifying a finite list of candidate pairable models, determining the at least one minimum model information set, and transmitting the at least one minimum model information set.

[0048] According to a fourth aspect of the present invention, a base station (gNB) includes the above-described equipment, characterized in that it performs the following steps: transmitting the ML configuration information, receiving the at least one minimum model information set, and determining the second AI / ML model.

[0049] According to a fifth aspect of the present invention, a wireless communication system includes the aforementioned base station (gNB) and the aforementioned user equipment, characterized in that,

[0050] The base station (gNB) includes a processor coupled to a memory storing computer program instructions configured to implement the steps of the method described above.

[0051] The user equipment (UE) includes a processor coupled to a memory storing computer program instructions configured to implement the steps of the method described above.

[0052] The following description refers to schematic flowcharts and / or block diagrams of methods, apparatus, systems, and computer program products according to embodiments. It should be understood that each block of the schematic flowcharts and / or block diagrams, and combinations of blocks of the schematic flowcharts and / or block diagrams, can be implemented by code. This code can 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 instructions executable via the processor of the computer or other programmable data processing apparatus create means for implementing the functions / actions specified in the flowcharts and / or block diagrams.

[0053] The code may also be stored in a storage device that can instruct a computer, other programmable data processing device or other means to operate in a particular manner, such that the instructions stored in the storage device produce an article of art including instructions that implement the functions / actions specified in the flowchart and / or block diagram.

[0054] The code may also be loaded onto a computer, other programmable data processing device or other apparatus such that a series of operational steps to be performed on the computer, other programmable device or other apparatus produce a computer-implemented process, such that the code to be executed on the computer or other programmable device provides a process for implementing the functions / actions specified in the flowchart and / or block diagram.

[0055] The flowcharts and / or block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of devices, systems, methods, and program products according to various embodiments. In this regard, each box in the flowcharts and / or block diagrams may represent a module, segment, or portion of code, which includes one or more executable instructions of that code for implementing the specified logical function.

[0056] It should also be noted that in some alternative implementations, the functions mentioned in the boxes may not appear in the order shown in the figures. For example, depending on the functionality involved, two boxes shown consecutively may actually be executed substantially concurrently, or the boxes may sometimes be executed in reverse order. Other steps and methods that are functionally, logically, or effectively equivalent to one or more boxes or portions thereof shown in the figures can be envisioned.

[0057] While various arrow and line types may be used in flowcharts and / or block diagrams, they are not intended to limit the scope of the corresponding embodiments. In practice, some arrows or other connecting symbols may be used to indicate only the logical flow of the depicted embodiment. For example, arrows may indicate waiting or monitoring periods of unspecified duration between enumerated steps of the depicted embodiment. It should also be noted that each block in the block diagram and / or flowchart, and combinations of blocks in the block diagram and / or flowchart, may be implemented by a system based on dedicated hardware or a combination of dedicated hardware and code that performs the specified function or action.

[0058] The description of the elements in each figure can be referenced to the elements in the preceding figures. All identical numbers in the figures refer to the same elements, including alternative embodiments of the same elements.

[0059] The following explanation will provide a detailed description of the mechanism for data-driven AI / ML model signaling for ML model operations in a wireless mobile communication system, which includes base stations (e.g., gNBs) and mobile stations (e.g., UEs).

[0060] The AI / ML lifecycle can be broken down into several phases, such as data collection / preprocessing, model training, model testing / validation, model deployment / update, model monitoring, and model switching / selection. Each phase is equally important for achieving the target performance of any particular model. One of the challenging issues when applying AI / ML models to any use case or application is managing the AI / ML model lifecycle. This is primarily because data / model drift occurs during model deployment / inference, leading to performance degradation. Essentially, statistical changes in the dataset occur after the model is deployed, and the model's inference capabilities are affected when using unseen data as input.

[0061] Similarly, the statistical properties of the dataset and the relationship between the input and output of the trained model can change as drift occurs. Model adaptation is then required to support operations such as model switching, retraining, and rollback. Furthermore, model pairing is needed to allow collaboration between two or more models located on different nodes (such as gNB-UE or UE-UE). A crucial step in model pairing is using machine learning to find the optimal combination of paired models for different applications and / or functions. Configuring model pairs is particularly difficult when model identifiers (e.g., model IDs) are not pre-configured for model pairing. Significant delays can occur in the associated configurations when searching for candidate model pairs to find matching pairs.

[0062] In this approach, a finite list of candidate pairable models matching the UE model is determined using a minimal model information set (e.g., model functionality, model inputs / outputs), thereby identifying the optimal combination of model pairs through blind pairing search. For the blind pairing search method, minimal model information about the UE model is shared between two nodes, such as the network side and the UE side. A finite number of candidate pairable models are determined for searching for model pairing. Each pairable model (i.e., the determined candidate pairable models and the UE model) is measured for model performance based on at least one target performance metric (such as model accuracy). The minimal model information set refers to the information to be shared to support the search for a finite list of candidate models to be paired. The minimal model information set contains shareable model description information or metadata. For example, when the UE has a model without a model ID and shares the minimal model information set with the network side, the network side uses this shared information from the UE to search for candidate pairable models.

[0063] The selection criteria for blind pairing search are based on predefined metrics related to model performance measurement.

[0064] Key benefits include reduced latency for matching model pairs and improved search performance when explicit model identifiers are unavailable.

[0065] Figure 1 An exemplary block diagram of a blind pairing search method for ML models is shown. In this exemplary block diagram, the scenario involves multiple candidate pairable models on the network side. A blind search method is used to potentially match these models with the UE model to find the optimal combination of network-UE model pairs. One or more search steps are taken before obtaining the final candidate pairable models, and each search step can be based on a minimal model information set provided by the UE side. The minimal model information set relates to model attribute data, such as model description, metadata, model functionality, model type / format, model input / output, etc. The UE model itself can be based on a proprietary format or a hidden model design.

[0066] The number of search steps can be determined based on how much UE model information can be shared with the network side through a minimum model information set, in order to find all potential pairable models by narrowing down a finite list of candidate pairable models. After identifying candidate pairable models, model performance can be measured for each combination of model pairs based on a target performance metric (e.g., model accuracy) as a selection criterion. Pairable model search is based on predefined metrics related to model performance. While in this exemplary block diagram, multiple candidate pairable models identified at the network side may match a specific UE model, it is also applicable when multiple candidate pairable models are identified at the UE side that can match a specific network-side model. Furthermore, the same mechanism applies to model pairing searches between UEs based on sidelinks. The minimum model information set can be related to the UE-side model, network-side model, or a combination of UE / network-side models in terms of model information that helps find matching model pairs.

[0067] Figure 2 An exemplary block diagram of a minimum model information set table is shown. In this exemplary table, the index and associated format of the minimum model information set are pre-configured, and the table contains information to be shared with another node to aid in searching a limited list of candidate models to be paired. The minimum model information set contains model attribute data, such as model description, metadata, model functionality, model type / format, model input / output, etc. Multiple tables may exist, each with different content and / or number of indexes for the minimum information set, and these tables may be predefined based on different ML applications and model functionalities, etc. The pre-configured table of the minimum model information set can then be sent to another node via L1, L2, or L3 signaling.

[0068] Figure 3 The diagram illustrates the signaling flow for searching candidate pairable models for model matching. Based on ML configuration information provided by the network side, a minimum set of model information to be shared with the network side is determined to enable blind pairing search. The minimum model information set format associated with the content of each set can be pre-configured so that each indexed set of minimum model information can be generated based on ML application, applicable model, model functionality, etc. After receiving the minimum model information set from the UE, a blind pairing search is performed between the network and the UE to find the optimal combination of paired models among the candidate pairable models.

[0069] Figure 4A flowchart illustrating the blind pairing search process on the network side is shown. In this exemplary process, the network side pre-configures a format for an indexed minimum model information set to be shared with another node (e.g., a UE). The UE can then use this format to generate an available minimum model information set. The format of the indexed minimum model information set can vary, and multiple formats may exist for different ML applications, applicable models, model functions, etc. For the same purpose, there may even be multiple formats to be shared with the UE (e.g., formats of different sizes or versions). Based on the received minimum model information set, the network side performs a blind pairing search to find candidate pairable models using any reference model that can be determined using the received minimum model information set.

[0070] Figure 5 A flowchart illustrating the blind pairing search process on the UE side is shown. Upon receiving ML configuration information, the UE model is determined so that an associated minimal model information set can be generated based on an indexed, pre-configured format.

[0071] Figure 6 The flowchart illustrates the process of determining combinations of model pairs. After identifying candidate pairable models, model performance is measured for each combination of model pairs based on a target performance metric (e.g., model accuracy measure) as the selection criterion. Therefore, the pairable model search is based on predefined metrics related to model performance. Among the identified candidate pairable models, the number of models among them used for performance testing can be adjusted based on implementation settings. This model performance measurement step is also considered pre-inference. After measuring the model performance of the candidate pairable models, the best combination of matching model pairs can be selected. If no matching model pair is found, the search may fail, and alternative ML configurations or rollback operations may need to be enabled.

[0072] Figure 7A flowchart illustrates the process of determining the type of candidate pairable models. Two types of candidate pairable models can be defined, such as a general reference model and a dedicated reference model. A general reference model is used to pair with another model during or after blind pairing search, where the shared minimum model information from the other model is below a pre-configured range or threshold. The general reference model can be used directly among candidate pairable models or during the search for candidate pairable models. The choice of reference model (general or dedicated) may depend on how much minimum model information is available. A dedicated reference model is used to pair with another model during or after blind pairing search when the pre-configured range or threshold is not met. This shared minimum model information is used when the shared minimum model information from the other model is above a pre-configured range or threshold. Similar to the general reference model, a dedicated reference model can be used directly among candidate pairable models or during the search for candidate pairable models. The pre-configured range or threshold indicates the amount of minimum model information shared between the two models.

[0073] Regarding network / UE-side behavior, depending on the deployment scenario (including sidelink-based communication), both sides can be nodes performing blind pairing searches. In the diagrams discussed above, the format of the minimum model information set is pre-configured on the network side, where configuration information can be sent using RRC(L3) signaling, but other signaling types, such as L1 or L2, can also be considered. However, other node types (such as UEs or edge computing devices) can perform pre-configuration of the minimum model information set format as needed.

[0074] This application aims to provide a fundamental mechanism for AI / ML support in radio access network collaboration—particularly in the search for the optimal combination of UE models and associated pairable models in ML operations—and for the flow of data information.

[0075] Based on the proposed invention, the behavior of gNB-UE for AI / ML operations that leverage joint ML operations to support wireless communication can be significantly improved in potential scenarios.

[0076] Abbreviations:

[0077] AI (Artificial Intelligence)

[0078] BWP bandwidth portion

[0079] CBG code block group

[0080] CLI Cross-Link Interference

[0081] CP cyclic prefix

[0082] CQI Channel Quality Indicator

[0083] CPU CSI processing unit

[0084] CRB Public Resource Block

[0085] CRC Cyclic Redundancy Check

[0086] CRI CSI-RS Resource Indicator

[0087] CSI Channel State Information

[0088] CSI-RS Channel State Information Reference Signal

[0089] CSI-RSRP CSI Reference Signal Received Power

[0090] CSI-RSRQ CSI reference signal reception quality

[0091] CSI-SINR (CSI Signal-to-Noise and Interference Ratio)

[0092] CW coding

[0093] DCI Downlink Control Information

[0094] DL downlink

[0095] DM-RS demodulation reference signal

[0096] DRX discontinuous reception

[0097] EPRE is the energy of each resource element.

[0098] IAB-MT Integrated Access and Backhaul - Mobile Terminal

[0099] ID identifier

[0100] L1-RSRP Layer 1 Reference Signal Received Power

[0101] LI layer indicator

[0102] LCM Lifecycle Management

[0103] MCS modulation and coding scheme

[0104] ML Machine Learning

[0105] NW Network

[0106] PDCCH (Physical Downlink Control Channel)

[0107] PDSCH (Physical Downlink Shared Channel)

[0108] PSS Master Synchronization Signal

[0109] PUCCH (Physical Uplink Control Channel)

[0110] QCL Quasi-co-located

[0111] PMI Precoding Matrix Indicator

[0112] PRB (Physical Resource Block)

[0113] PRG precoded resource block group

[0114] PRS Positioning Reference Signal

[0115] PT-RS phase tracking reference signal

[0116] RAN (Radio Access Network)

[0117] RB resource block

[0118] RBG resource block group

[0119] RI rank indicator

[0120] RIV resource indicator value

[0121] RS reference signal

[0122] SCI sidelink control information

[0123] SLIV start and length indicators

[0124] SR scheduling request

[0125] SRS Detection Reference Signal

[0126] SS synchronization signal

[0127] SSS auxiliary synchronization signal

[0128] SS-RSRP SS reference signal received power

[0129] SS-RSRQ SS reference signal reception quality

[0130] SS-SINR (Signal-to-Noise and Interference Ratio)

[0131] TB transfer block

[0132] TCI Transport Configuration Indicator

[0133] TDM (Time Division Multiplexing)

[0134] UE User Equipment

[0135] UL uplink

[0136] WG Working Group

Claims

1. A method for AI / ML model matching signaling between a first node and a second node in a wireless network to support an AI / ML model pairing process between model pairs consisting of a first AI / ML model of the first node and a second AI / ML model of the second node, the method comprising the following steps: • Receive ML configuration information indicating a model pairing request that matches the first AI / ML model. • Identify a finite list of candidate pairable models. • For each candidate pairable model, generate at least one minimum model information set, where each minimum model information set contains descriptive information about its corresponding candidate pairable model. • Receive the at least one minimum model information set at the first node, and • The second AI / ML model is determined as the candidate pairable model using the at least one minimal model information set, which forms an optimal combination with the first AI / ML model.

2. The method according to claim 1, characterized in that, The determination of the second AI / ML model is based on the performance measurement of the target performance index of the model pair.

3. The method according to claim 1 or 2, characterized in that, The at least one minimum model information set contains information about model functionality, and / or model inputs / outputs, and / or model descriptions, and / or model metadata.

4. The method according to any one of the preceding claims, characterized in that, The second AI / ML model was determined using blind pairwise search.

5. The method according to any one of the preceding claims, characterized in that, The at least one minimum model information set is transmitted from the second node to the first node via L1, L2 or L3 signaling.

6. The method according to any one of the preceding claims, characterized in that, The at least one minimal model information set is indexed in a pre-configured format to support the generation of a finite list of candidate models.

7. The method according to claim 5, characterized in that, The minimum model information set contains multiple tables.

8. The method according to any one of the preceding claims, characterized in that, The candidate pairable models are classified as general reference models or special reference models, and the method further includes the following steps: • Receive at least one pre-configured range regarding the at least one minimum model information set. If the at least one model information set is below the pre-configured range, then the general reference model is selected as the second AI / ML model. • If the minimum model information set is higher than the pre-configured range, then the dedicated reference model is selected as the second AI / ML model.

9. The method according to claim 2, characterized in that, The number of candidate paired models that can be tested for model performance can be adjusted based on implementation settings.

10. The method according to claim 2, characterized in that, If no model pair provides sufficient target performance metrics, a search failure is issued to enable alternative AI / ML configurations or a fallback operation.

11. The method according to any one of the preceding claims, characterized in that, The nodes can be UEs, networks, base stations, or combinations thereof, and communication between nodes can be performed via L1, L2, or L3 signaling, or via sidelink-based communication.

12. An apparatus for AI / ML model matching signaling, comprising a processor coupled to a memory storing computer program instructions configured to implement the steps of claims 1 to 11.

13. A user equipment comprising the apparatus according to claim 12, characterized in that, The following steps are performed: receiving the ML configuration information, identifying a finite list of candidate pairable models, determining the at least one minimum model information set, and transmitting the at least one minimum model information set.

14. A base station (gNB), comprising the equipment according to claim 12, characterized in that, The following steps are performed: transmitting the ML configuration information, receiving the at least one minimum model information set, and determining the second AI / ML model.

15. A wireless communication system comprising a base station (gNB) according to claim 14 and a user equipment according to claim 13, characterized in that, The base station (gNB) includes a processor coupled to a memory storing computer program instructions configured to implement the steps of the method as described in claims 1 to 11. The user equipment (UE) includes a processor coupled to a memory storing computer program instructions configured to implement the steps of the method as claimed in claims 1 to 11.

Images