Method for offloading ML LCM operations in wireless networks

Abstract

A method of pre-configuring AI / ML operations when enabling multi-connectivity links in a wireless mobile communication system including base stations (e.g., gNBs) and mobile stations (e.g., UEs) is presented. If AI / ML models are applied to the radio access network, model performance such as inference and / or training depends on different model execution environments between the network side and the UE side. Therefore, by offloading model operations from the network side to a relay UE, potential performance impact due to out-of-coverage locations can be reduced with enhanced model performance.

Description

Technical Field

[0001] This disclosure relates to the pre-configuration of AI / ML operations, which proposes methods for reconfiguration and signaling specific information to enhance the performance of relay-based models. Background Technology

[0002] Within 3GPP (3rd Generation Partnership Project), one of the selected research projects up to the approved Release 18 package is AI / ML (Artificial Intelligence / Machine Learning), as described in the relevant document (RP-213599) submitted at 3GPP TSG (Technical Specification Group) RAN (Radio Access Networks) meeting #94e. The formal title of the AI / ML research project is "Study on AI / ML for NR Air Interface," and currently, RANWG1 (Working Group 1) and WG2 are actively developing the specifications. The goal of this research project is to identify a public AI / ML framework and areas where benefits can be derived from using AI / ML-based technologies through use cases.

[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, leveraging target use cases. Specifically, AI / ML models, terminology, and descriptions used to determine the common and specific characteristics of the framework will be a key area of ​​work. Regarding AI / ML frameworks, investigations are being considered from various aspects, and one of the key projects concerns the lifecycle management (LCM) of AI / ML models, which mandatorily includes multiple phases for model training, deployment, inference, monitoring, and updates.

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

[0005] Regarding the aforementioned positive standardization efforts, UE ML conditions used to support RAN-based AI / ML models can be considered crucial for both the gNB and UE to meet any desired model operations (e.g., model training, inference, selection, handover, updates, monitoring, etc.). Currently, there are no specifications defining signaling methods or gNB-UE behavior for distributing split LCM operations via sidelink trunk links while RAN-based AI / ML model operations continue. Therefore, it is necessary to investigate any specification impacts by considering model operations via sidelink trunk links. Furthermore, any mechanisms for additional signaling methods and / or gNB-UE behavior need to be addressed to support trunk-based model operations between the gNB and UE, minimizing any potential impact of UE ML conditions on model operations within the RAN while maintaining service continuity.

[0006] On the other hand, in 3GPP, the terminology in the work list contains a set of high-level descriptions regarding AI / ML model training, inference, validation, testing, UE-side models, network-side models, single-sided models, two-sided models, etc. UE-side models and network-side models respectively indicate that the AI / ML model operates on the UE side and the network side. In a similar context, single-sided models and two-sided models indicate that the AI / ML model is located on one side and both sides, respectively.

[0007] Not all signaling aspects supporting the aforementioned projects have been specified at present, as the definitions of the terminology are still under discussion and require further revision. Any potential standard implications of new or enhanced mechanisms for AI / ML models supporting the aforementioned work list projects are a key area to be investigated in AI / ML research projects.

[0008] US 2021 203 565 A1 discloses methods, systems, and apparatus for training and using machine learning models to classify network traffic into IoT traffic or non-IoT traffic, and for managing traffic based on said classification, including computer programs encoded on a computer storage medium. In some implementations, machine learning parameters of a local machine learning model trained by an edge device are received from each edge device in at least a subset of a set of edge devices. The machine learning parameters received from the edge devices are parameters of a local machine learning model trained by the edge devices based on local network traffic processed by the edge devices and used to classify network traffic into Internet of Things (IoT) traffic or non-IoT traffic. The machine learning parameters are used to generate a global machine learning model to classify network traffic processed by the edge devices into IoT traffic or non-IoT traffic.

[0009] US 2022 261 697 A1 discloses a system and method for federated machine learning. A central system receives satellite analytics from multiple satellite site systems and generates a central machine learning model based on the satellite analytics. Multiple federated machine learning rounds are performed. In each round, the central system transmits the central machine learning model to the multiple satellite site systems and then receives, accordingly, a set of satellite values ​​for a set of weights for the model from each satellite site system, where the satellite values ​​are generated by the respective satellite site system based on its local dataset. In each round, the central system then generates an updated version of the central machine learning model based on the satellite values ​​received from the satellite site systems.

[0010] US Patent 2023 037 893 A1 discloses a method for generating a real-time radio coverage map in a wireless network by a network device. The method includes: receiving real-time geospatial information from one or more geographical sources in the wireless network; determining handover information of at least one user equipment (UE) in the wireless network from multiple base stations based on the real-time geospatial information; and generating a real-time radio coverage map based on the real-time geospatial information and the handover information of at least one UE.

[0011] WO 2022 015 008 A1 discloses a method for determining a target cell for handover of a UE. The method includes: a mobility management platform monitoring multiple network characteristics associated with multiple UEs; and the mobility management platform determining a correlation between the multiple UEs based on the multiple network characteristics associated with the multiple UEs and location information of the multiple UEs. The method further includes: the mobility management platform receiving location information from the multiple UEs; and the mobility management platform determining a measurement report corresponding to the location information received from the UEs based on the correlation. The method also includes: the mobility management platform determining a target cell for handover of the UE based on the measurement report and the location information received from the UEs.

[0012] WO 2022 205 023 A1 discloses systems, methods, and apparatuses on wireless network architectures and air interfaces. In some embodiments, a sensing agent communicates with a user equipment (UE) or node using one of a variety of sensing modes via a non-sensor-based link or a sensor-based link, and / or an artificial intelligence (AI) agent communicates with the UE or node using one of a variety of AI modes via a non-AI-based link or an AI-based link. AI and sensing can operate independently or together. For example, an AI block can send a sensing service request to a sensing block to obtain sensing data from the sensing block, and the AI ​​block can generate a configuration based on the sensing data. Various other features, such as example interfaces, channels, and other aspects related to enabling AI and / or enabling sensing communication, are also disclosed. Attached Figure Description

[0013] The disclosed invention will be further discussed below 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 is taken with reference to the accompanying drawings, in which:

[0014] Figure 1 This is an example table showing the mapping between unloadable LCM operations and reference ML capabilities;

[0015] Figure 2 This is an exemplary flowchart of network-side behavior used for LCM operation offloading;

[0016] Figure 3 This is an exemplary flowchart of relay UE-side behavior for LCM operation offloading;

[0017] Figure 4 This is an exemplary flowchart of the target UE-side behavior for LCM operation offloading; and

[0018] Figure 5 This is an exemplary flowchart of the target UE-side behavior used for LCM operation offloading during in-coverage location. 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 the embodiments described herein may be exemplified using terminology from 3GPP 5G NR in this disclosure, this should not be construed as limiting the scope of the invention.

[0020] Some embodiments of the contemplated embodiments 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 ordinary meaning in the relevant art, unless a different meaning is expressly given and / or implied in the context of their use. Unless otherwise expressly stated, all references to a / an / said element, device, component, element, step, etc., should be interpreted openly as referring to at least one instance of said element, device, component, element, 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 the said embodiments may be applied to any other embodiment, and vice versa. Other objectives, features, and advantages of the appended embodiments will be apparent from the description below.

[0022] In some implementations, the more general term "network node" may be used, and the 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), repeater, donor node control repeater, 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 implementations, 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] Additionally, terms such as base station / gNodeB and UE should be considered non-limiting and, in particular, do not imply any hierarchical relationship between them; 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. Furthermore, in the following text, a transmitter or receiver can be a gNodeB (gNB) or a UE.

[0025] As those skilled in the art will understand, aspects of the implementation scheme can be embodied in a system, device, method, or computer program product. Therefore, the implementation scheme can take the form of a fully hardware implementation scheme, a fully software implementation scheme (including firmware, resident software, microcode, etc.), or a combination of 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, the implementation 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 implementation, 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] More specific examples of storage devices (a non-exhaustive list) will include the following: an electrical connection having one or more wires; a portable computer floppy disk; a hard disk; random access memory (“RAM”); read-only memory (“ROM”); erasable programmable read-only memory (“EPROM” or flash memory); a portable optical disc read-only 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 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 implementation scheme 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 execute 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 implementation scheme 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 offer a thorough understanding of the implementation scheme. However, those skilled in the art will recognize that the implementation scheme can be practiced without one or more of the specific details described herein, or by utilizing other methods, components, materials, etc.

[0032] 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,” “implementation,” or similar language mean that a particular feature, structure, or characteristic described in connection with an embodiment is included in at least one embodiment. Therefore, unless expressly specified otherwise, the phrases “in an 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,” “including.” Unless expressly specified otherwise, an enumerated list of items does not imply that any or all of the items are mutually exclusive. Unless expressly specified otherwise, the terms “an,” “a,” and “the” also mean “one or more.”

[0033] The following description of aspects of the embodiments is based on schematic flowcharts and / or block diagrams of methods, apparatus, systems, and computer program products according to the 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 establish components for implementing the functions / actions specified in the flowcharts and / or block diagrams.

[0034] 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.

[0035] Code can 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 executing on the computer or other programmable device provides a process for implementing the functions / actions specified in the flowchart and / or block diagram.

[0036] 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 the code for implementing the specified logical function.

[0037] It should also be noted that in some alternative implementations, the functions indicated in the boxes may not occur in the order shown in the figures. For example, in practice, depending on the functionality involved, the two boxes shown consecutively may be executed substantially simultaneously, or sometimes in reverse order. Other steps and methods that are functionally, logically, or effectically equivalent to one or more boxes or portions thereof shown in the figures can be envisioned.

[0038] While various arrow and line types may be used in flowcharts and / or block diagrams, they are understood not to limit the scope of the corresponding implementation. In practice, some arrows or other connecting symbols may be used to indicate only the logical flow of the depicted implementation. For example, arrows may indicate waiting or monitoring periods of unspecified duration between enumerated steps of the depicted implementation. It should also be noted that each box in a block diagram and / or flowchart, as well as combinations of boxes in block diagrams and / or flowcharts, 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.

[0039] The description of the elements in each figure can be referenced to the elements in the preceding figures. In all figures, the same numbers refer to the same elements, including alternative embodiments of the same elements.

[0040] The following explanation will provide a detailed description of the mechanism for pre-configuring AI / ML-based models before the handover occurs in a wireless mobile communication system that includes base stations (e.g., gNBs) and mobile stations (e.g., UEs).

[0041] 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, changes in dataset statistics occur after model deployment, and the model's inference capabilities are also affected when using unseen data as input.

[0042] 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. When deploying wireless communication networks that enable AI / ML models, it is important to consider how to handle AI / ML model adaptation during operations such as model training, inference, monitoring, and updates.

[0043] The applicability of ML for LCM (Lifecycle Management) operations can vary significantly depending on the use case and environmental attributes, depending on the specific network-UE ML collaboration in the deployment scenario (e.g., UE mobility scenarios). AI / ML-based technologies are currently applied to many different applications, and based on observed potential benefits, 3GPP has also begun its technology research for application to multiple use cases.

[0044] Similarly, the statistical properties of a dataset and the relationship between the inputs and outputs of a trained model can change as drift occurs. In this context, model performance, such as inference and / or training, depends on different model execution environments with varying configuration parameters.

[0045] To address this issue, it is crucial to track model performance and reconfigure the model during collaboration between the UE and gNB, and across different environments between the UE and different gNBs. When deploying wireless communication networks with AI / ML models enabled, it is important to consider how to handle active AI / ML models and reconfiguration for wireless devices during operations such as model training, inference, and updates. When the UE is out of coverage, the configured ML model may be affected by model execution failures.

[0046] A first aspect of the present invention discloses a method for offloading ML LCM operations in a wireless network. The wireless network includes at least a base station (gNB), a target UE located outside coverage, and a group of relay UEs including relay UEs not located outside coverage. The method includes the following steps:

[0047] Generate ML capability levels that define the requirements to support LCM operations.

[0048] Receive ML capability level,

[0049] For each relay UE, ML capability is measured based on ML capability level.

[0050] For each relay UE, the measured ML capability is compared with the ML capability level.

[0051] For each relay UE, an indication message is sent based on a corresponding comparison of ML capabilities to indicate whether the relay UE can perform LCM operations.

[0052] At least a first relay UE is selected from the relay UEs, and the at least first relay UE is capable of performing LCM operation.

[0053] Configure LCM operation to at least the first relay UE via system information or dedicated RRC message, and

[0054] The LCM operation is performed by at least the first relay UE.

[0055] Advantageously, the ML capability level is a pre-configured mapping between LCM operations and corresponding thresholds or ranges, and the ML capability level is transmitted via system information or dedicated RRC messages.

[0056] Advantageously, ML capability is measured by the base station or at least the first relay UE.

[0057] Advantageously, the measured ML capability is compared with the ML capability level by the base station or at least the first relay UE.

[0058] Advantageously, LCM offload operations can be performed across multiple relay UEs.

[0059] Advantageously, the pre-configured mapping relationships include:

[0060] A finite set of indexed, unloadable LCM operations, wherein the unloadable LCM operations are one of the following: data collection, model training, model inference, or model monitoring; and / or

[0061] Matching reference ML capability, which is one of the following: a threshold or range indicating any combination of parameters reflecting the ML applicability of the device and / or the ML applicability of the environment.

[0062] Advantageously, multiple target UEs are connected to at least a first relay UE to receive reconfigured ML configuration messages via dedicated RRC reconfiguration messages.

[0063] Advantageously, at least the first relay UE sends an updated ML configuration message to the target UE or multiple target UEs to reconfigure the LCM operation.

[0064] Advantageously, ML configuration messages are exchanged via side links.

[0065] Advantageously, side links are established via L1 / L2 / L3 signaling.

[0066] Advantageously, multiple different ML models can be applied to LCM operations based on different ML use cases, which are one of the following:

[0067] When the base station offloads the LCM operation, all relevant model information and ML configuration are initially set to be provided to at least the first relay UE and / or the target UE and / or multiple target UEs;

[0068] The initial ML model information and ML configuration are provided by the base station. Once the relay UE begins to offload LCM operations, including model management, a reconfiguration occurs between the relay UE and the target UE.

[0069] A second aspect of the present invention discloses an apparatus for offloading ML LCM operations in a wireless communication system, the apparatus comprising:

[0070] • Wireless transceivers; and

[0071] • A processor coupled to a memory storing computer program instructions configured to implement the steps of the methods described above, and the device being designed for use in a base station (gNB) or user equipment (UE).

[0072] A third aspect of the present invention discloses an apparatus for offloading ML LCM operations in a wireless communication system, the apparatus comprising:

[0073] • Wireless transceiver

[0074] • A processor coupled to a memory in which computer program instructions are stored, the instructions being configured to implement the steps of the methods described above, and the device being designed for use in a user equipment (UE).

[0075] A fourth aspect of the present invention discloses a base station (gNB) that includes the device described above.

[0076] A fifth aspect of the present invention discloses a user equipment comprising the equipment described above.

[0077] A sixth aspect of the present invention discloses a wireless communication system comprising at least one base station (gNB) and at least one user equipment (UE) as described above.

[0078] In this approach, the network side (e.g., gNB) determines whether to offload the configured LCM operation to the relay UE based on information from the relay UE, where the configured LCM operation to be performed by the NW can be offloaded to the relay UE. The network side configures a set of thresholds / ranges to compare with the relay ML capabilities to support the requested offloaded LCM operation, where configuration information about the thresholds can be sent via system information and / or dedicated RRC messages.

[0079] When pre-configured LCM information to be offloaded is sent to candidate relay UEs, the candidate relay UEs measure their relay ML capabilities and compare them to a threshold / range. Alternatively, when the relay ML capabilities of candidate relay UEs are sent to the network side, the network side can measure the reported relay ML capabilities and compare them to a threshold / range. Among the candidate relay UEs, if the relay ML capability is above a threshold or within a threshold range related to the associated LCM operation, one or more relay UEs can be selected to serve the offloaded LCM operation. When more than one relay UE is selected, distributed LCM operation offloading is performed across multiple relay UEs, as the network side can assign a separate offloaded LCM operation based on each relay ML capability.

[0080] For a set of thresholds / ranges to be compared with relay ML capabilities, pre-configured mapping information can be used. For example, a limited set of indexed offloadable LCM operations can be configured to form a matching reference ML capability, where the offloadable LCM operations indicate the execution of ML tasks such as data collection, model training / inference / monitoring, etc., and the reference ML capability includes thresholds or ranges that indicate any combination of parameters reflecting the applicable conditions of device ML (e.g., computing power, memory, etc.) and environmental ML (e.g., site, wireless link, etc.).

[0081] After all relay UEs are selected during the selection process, the target UE group to be connected to the relay UEs is reconfigured to have ML tasks assigned by the relay UEs (e.g., via RRC reconfiguration) to execute, while receiving initial ML configuration for the target UEs during in-coverage status. In the out-of-coverage location of the target UE, the relay UE can choose to update / reconfigure LCM operations and send the updated configuration to the target UE, as the relay UE is able to measure ML capabilities to distribute LCM operations to its associated target UEs. Relay ML and target ML (re)configuration information for distributed LCM operations is exchanged via sidelinks (e.g., L1 / L2 / L3 signaling). For distributed LCM operations, multiple different ML models can be applied depending on the different ML use cases and different LCM operations. When the network side offloads LCM operations to the relay UEs, all relevant model information and ML configurations are initially set up to be provided to the relay UEs and / or target UEs. When a relay UE takes over offloaded LCM operations, including model management, initial information about the model and ML configuration from the network side can be reconfigured between the relay UE and the target UE if necessary.

[0082] Figure 1 An exemplary table is shown showing the mapping information between offloadable LCM operations and reference ML capabilities. In this example, a limited set of indexed offloadable LCM operations is configured, forming matching reference ML capabilities. The offloadable LCM operations instruct the execution of ML tasks, such as data collection, model training / inference / monitoring, etc., and the reference ML capabilities include thresholds or ranges that indicate any combination of parameters reflecting the suitability of device ML (e.g., computing power, memory, etc.) and environmental ML (e.g., site, wireless link, etc.).

[0083] Figure 2 An exemplary flowchart for network-side behavior for LCM operation offloading is shown. In this example, the network side (e.g., gNB) determines an ML configuration that includes pre-configured mapping information regarding offloading LCM operations to a relay UE device. To identify the relay UE capable of performing the offloaded LCM operation, relay ML capability information needs to be measured at the network side or the relay UE side, depending on the implementation scenario.

[0084] Figure 3An exemplary flowchart illustrating the relay UE-side behavior for LCM operation offloading is shown. In this example, the relay UE measures its relay ML capabilities and compares them with pre-configured threshold information. An indication message regarding the measurement is sent to the network side, allowing the relay UE to confirm that the offloaded LCM operation can be performed. Upon receiving this indication message, the network side determines the relay ML reconfiguration, enabling the selection of all relay UE devices to perform the offloaded LCM operation, or the selection of a single relay UE or multiple relay UEs for distributed offloaded LCM operation.

[0085] Figure 4 An exemplary flowchart of target UE-side behavior for LCM operation offloading is shown. In this example, one or more target UEs may connect to a relay UE to perform LCM operations, and the selected target UE receives an instruction message from the relay UE regarding activating target UE ML execution.

[0086] Figure 5 An exemplary flowchart illustrates the target UE-side behavior for LCM operation offloading during in-coverage location. In this example, the target UE receives ML configuration directly from the network side while it is still in an in-coverage location. After the target UE reaches an out-of-coverage location, ML reconfiguration is performed with the support of a relay UE, enabling the target UE to activate its own ML operation when instructed.

[0087] abbreviation

[0088] BWP bandwidth portion

[0089] CBG code block group

[0090] CLI Cross-Link Interference

[0091] CP loop prefix

[0092] CQI Channel Quality Indicator

[0093] CPUCSI processing unit

[0094] CRB Public Resource Block

[0095] CRC Cyclic Redundancy Check

[0096] CRICSI-RS resource indicator

[0097] CSI Channel Status Information

[0098] CSI-RS Channel State Information Reference Signal

[0099] CSI-RSRP CSI Reference Signal Received Power

[0100] CSI-RSRQ CSI reference signal reception quality

[0101] CSI-SINR: CSI signal versus noise plus interference ratio

[0102] CW code

[0103] DCI downlink control information

[0104] DL downlink

[0105] DM-RS demodulation reference signal

[0106] DRX discontinuous reception

[0107] EPRE Energy per Resource Element

[0108] IAB-MT Integrated Access and Backhaul - Mobile Terminal

[0109] L1-RSRP Layer 1 Reference Signal Received Power

[0110] LI layer indicator

[0111] MCS modulation and coding scheme

[0112] PDCCH Physical Downlink Control Channel

[0113] PDSCH Physical Downlink Shared Channel

[0114] PSS master synchronization signal

[0115] PUCCH (Physical Uplink Control Channel)

[0116] QCL Quasi-Co-location

[0117] PMI Precoding Matrix Indicator

[0118] PRB Physical Resource Block

[0119] PRG precoded resource block group

[0120] PRS positioning reference signal

[0121] PT-RS phase tracking reference signal

[0122] RB resource blocks

[0123] RBG resource block group

[0124] RI rank indicator

[0125] RIV resource indicator value

[0126] RP resource pool

[0127] RS reference signal

[0128] SCI sidelink control information

[0129] SL CR sidelink channel occupancy rate

[0130] SL CBR sidelink channel busy ratio

[0131] SLIV start and length indicator values

[0132] SR scheduling request

[0133] SRS detection reference signal

[0134] SS Synchronization Signal

[0135] SSS auxiliary synchronization signal

[0136] SS-RSRP SS reference signal received power

[0137] SS-RSRQ SS reference signal reception quality

[0138] SS-SINRSS signal to noise plus interference ratio

[0139] TB transport block

[0140] TCI Transport Configuration Indicator

[0141] TDM Time Division Multiplexing

[0142] UE User Equipment

[0143] UL uplink

Claims

1. A method for offloading ML LCM operations in a wireless network, the wireless network comprising at least a base station (gNB), a target UE located outside coverage, and a group of relay UEs including relay UEs not located outside coverage, the method comprising the following steps: Generate ML capability levels that define the requirements to support the LCM operations. Receive the ML capability level, For each relay UE, ML capability is measured based on the ML capability level. For each relay UE, the measured ML capability is compared with the ML capability level. For each relay UE, an indication message is sent based on a corresponding comparison of the ML capabilities, indicating whether the relay UE is capable of performing the LCM operation. At least a first relay UE is selected from the relay UEs, and the at least first relay UE is capable of performing the LCM operation. Configure the LCM operation to the at least first relay UE via system information or a dedicated RRC message, and The LCM operation is performed by the at least first relay UE.

2. The method according to claim 1, characterized in that, The ML capability level is a pre-configured mapping between LCM operations and corresponding thresholds or ranges, and the ML capability level is transmitted via system information or dedicated RRC messages.

3. The method according to claim 1 or 2, characterized in that, The ML capability is measured by the base station or by the at least first relay UE.

4. The method according to any one of the preceding claims, characterized in that, The measured ML capability is compared with the ML capability level by the base station or the at least first relay UE.

5. The method according to any one of the preceding claims, characterized in that, The LCM offloading operation is performed across multiple relay UEs.

6. The method according to claim 2, characterized in that, The pre-configured mapping relationships include: A finite set of indexed, detachable LCM operations, wherein the detachable LCM operations are one of the following: data collection, model training, model inference, or model monitoring; and / or Matching reference ML capability, which is one of the following: a threshold or range indicating any combination of parameters reflecting the ML applicability of the device and / or the ML applicability of the environment.

7. The method according to any one of the preceding claims, characterized in that, Multiple target UEs connect to the at least first relay UE to receive reconfigured ML configuration messages via dedicated RRC reconfiguration messages.

8. The method according to any one of the preceding claims, characterized in that, The at least first relay UE sends an updated ML configuration message to the target UE or the plurality of target UEs to reconfigure the LCM operation.

9. The method according to claim 7 or 8, characterized in that, The ML configuration messages are exchanged via a side link.

10. The method of claim 9, wherein the side link is established via L1 / L2 / L3 signaling.

11. The method according to any one of the preceding claims, characterized in that, The LCM operation can apply multiple different ML models based on different ML use cases, wherein the ML use case is one of the following: When the base station offloads the LCM operation, all relevant model information and ML configuration are initially set to be provided to the at least first relay UE and / or the target UE and / or the plurality of target UEs; The initial ML model information and ML configuration are provided by the base station. Once the relay UE begins to offload the LCM operations, including model management, a reconfiguration occurs between the relay UE and the target UE.

12. An apparatus for offloading ML LCM operations in a wireless communication system, the apparatus comprising: • Wireless transceiver; as well as • A processor coupled to a memory storing computer program instructions configured to implement the steps of the method according to claims 1 to 11, and the device being designed for use in a base station (gNB) or user equipment (UE).

13. An apparatus for offloading ML LCM operations in a wireless communication system, the apparatus comprising: • Wireless transceiver • A processor coupled to a memory in which computer program instructions are stored, the instructions being configured to implement the steps of the method according to claims 1 to 11, and the device being designed for use in a user equipment (UE).

14. A base station (gNB) comprising the device of claim 12.

15. A user equipment comprising the device of claim 13.

16. A wireless communication system comprising at least one base station (gNB) as claimed in claim 14 and at least one user equipment (UE) as claimed in claim 15.

Images