Apparatuses, methods, and computer programs for a network node and for user equipment and for partition monitoring of a machine-learning model

Abstract

Provided are apparatuses, methods, and computer programs for a network node and for user equipment. The method (10) for the network node (300) of the mobile communication system (500) comprises transmitting (12), to a user equipment (400), UE; information relating to a plurality of dataset partitions, each dataset partition being associated with a respective partition identifier, the information enabling the UE (400) to identify and selectively retrieve a dataset partition associated with an inference accuracy issue of a machine-learning model. The method 10 also includes in response to receiving a request from the user equipment indicating a specific partition identifier, transmitting (14) the dataset partition associated with the specific partition identifier to the UE (400).

Description

[0001] 202500709 1

[0002] Apparatuses, Methods, and Computer Programs for a Network Node and for User Equipment and for Partition Monitoring of a Machine-Learning Model

[0003] Description

[0004] The present disclosure relates to apparatuses, methods, and computer programs for a network node and for user equipment, and more particularly, but not exclusively, to a concept for improving life cycle management of a machine-learning model at a UE by enabling partition monitoring of one or more partitions the machine learning model is based on.

[0005] Machine-learning models deployed in wireless communication systems, particularly those operating at user equipment (UE) or network nodes in accordance with 3GPP (3rd Generation Partnership Project) specifications for the NR (New Radio) air interface, are subject to the development and accumulation of bias throughout their operational lifetime. Bias manifests as systematic performance degradation that evolves across distinct phases of the model lifecycle, presenting significant challenges for maintaining reliable AI / ML-enabled (Artificial Intelligence / Machine-Learning) features in commercial deployments.

[0006] During the initial training phase, bias is first introduced into a machine-learning model through various mechanisms inherent to the development process. Training datasets may not comprehensively cover all real-world scenarios, configurations, or deployment sites that the model will encounter during operation. Ground truth labels used for supervised learning may contain systematic errors that become embedded in the model's learned representations. Furthermore, the choice of model architecture and hyperparameter optimization may inadvertently favor certain operating conditions over others. A particularly problematic form of initial bias arises from overfitting, wherein the model learns the training data too specifically, including noise and artifacts that do not generalize to operational environments. Models that are optimized for standardized test setups may demonstrate excellent performance during laboratory evaluation while exhibiting compromised robustness in real-world environments that do not align with the controlled training conditions.202500709 2

[0007] When a trained model transitions to the deployment phase in a live network, additional bias can emerge from mismatches between training and operational conditions.

[0008] Scenario mismatches occur when models are deployed in environments not adequately represented in the training data, such as a model trained on urban propagation data being deployed in rural or indoor factory settings. Configuration mismatches arise when the deployed network employs different antenna configurations, frequency bands, or network settings than those present during training. Site-specific differences introduce bias due to unique propagation characteristics of new deployment locations, while hardware variations across different UE implementations or chipsets can cause systematic prediction errors.

[0009] Throughout the operational phase, bias accumulates progressively due to phenomena commonly referred to as data drift and concept drift. Data drift, also known as covariate shift, occurs when the statistical distribution of input data changes over time relative to the training distribution. Environmental changes such as seasonal variations, construction of new buildings, and changes in foliage can alter radio propagation characteristics. Network evolution through the deployment of new cells or configuration updates modifies the operational context. Changes in user behavior patterns, including different mobility characteristics and usage scenarios, further contribute to input distribution shifts. Concept drift represents a more fundamental form of bias development wherein the underlying relationship between model inputs and desired outputs changes over time. Technology updates that introduce new network features not present during training, evolving interference patterns from new sources in the environment, and regulatory changes affecting power levels or frequency allocations all contribute to concept drift.

[0010] The temporal progression of bias accumulation follows a characteristic pattern in deployed machine-learning models. Initial performance levels achieved at deployment gradually degrade as data drift begins to affect prediction accuracy. This degradation accelerates as concept drift compounds the effects of data drift, eventually leading to performance levels that fall below acceptable thresholds. For specific use cases defined in 3GPP specifications, this bias evolution manifests in distinct ways. In channel state information prediction using UE-sided models, training data reflects specific channel202500709 3

[0011] conditions that become increasingly unrepresentative as seasonal changes affect propagation, new construction alters multipath characteristics, and network densification modifies interference patterns over months and years of operation. In beam management applications, models optimized for current beam configurations become biased when new beams are added, beam patterns are modified, or environmental changes such as the addition or removal of reflective surfaces alter the coverage area. For positioning applications, fingerprint databases that accurately reflect the environment at the time of training become progressively biased as facility layouts are modified, machinery is relocated, or new network infrastructure is deployed.

[0012] To address the development of bias over the model lifetime, lifecycle management mechanisms have been defined within the 3GPP framework for AI / ML-enabled features. Performance monitoring functions continuously evaluate model performance to detect when accumulated bias causes degradation below acceptable thresholds. Model switching techniques enable the selection among multiple scenario-specific, configuration-specific, or site-specific models to address bias arising from operational condition changes. Model update and fine-tuning processes enable flexible adaptation of model structures or parameters in response to detected drift. Fallback mechanisms ensure that when bias causes AI / ML-enabled features to underperform relative to legacy non-AI / ML-based operations, the system can revert to conventional methods to maintain reliable service. The functional framework for AI / ML lifecycle management encompasses data collection for ongoing operational monitoring, model training and adaptation for bias correction, model storage for maintaining multiple model versions, model inference for applying current models, and management functions for coordinating monitoring, switching, and update operations throughout the model lifetime.

[0013] Further details can be found in:

[0014] 1. US 2022400373 A1 , MACHINE LEARNING MODEL CONFIGURATION IN WIRELESS NETWORKS;

[0015] 2. US 2022108214 A1, MANAGEMENT METHOD OF MACHINE LEARNING MODEL FOR NETWORK DATA ANALYTICS FUNCTION DEVICE;

[0016] 3. US 2022337487 A1 , NETWORK ENTITY FOR DETERMINING A MODEL FOR DIGITALLY ANALYZING INPUT DATA;202500709 4

[0017] 4. WO 2023277780 A1 , ENABLING DOWNLOADABLE Al;

[0018] 5. WO 2022258149 A1, USER DEVICE, SERVER DEVICE, METHOD AND SYSTEM FOR PRIVACY PRESERVING MODEL TRAINING;

[0019] 6. WO 2022228666 A1 , SIGNALING OF TRAINING POLICIES; and

[0020] 7. WO 2022161624 A1, CANDIDATE MACHINE LEARNING MODEL IDENTIFICATION AND SELECTION

[0021] Various examples of the present disclosure are based on the finding that machinelearning models deployed in mobile communication systems may experience inference accuracy issues that require efficient mechanisms for updating or retraining. In particular, when a user equipment (UE) operates a machine-learning model that yields suboptimal inference results, there is a technical need to identify which portions of training data are associated with the accuracy degradation and to selectively retrieve only those relevant data partitions rather than downloading entire datasets. The present disclosure relates to a technique for addressing inference accuracy issues in machinelearning models by enabling selective retrieval of dataset partitions in a mobile communication system.

[0022] The proposed concept introduces a coordinated approach between a network node and user equipment for managing dataset partitions associated with machine-learning model training or refinement. By providing information relating to a plurality of dataset partitions, each associated with a respective partition identifier, the network node enables the user equipment to identify and selectively retrieve only those dataset partitions that are associated with an inference accuracy issue. This improves bandwidth efficiency and reduces latency in addressing model accuracy degradation, as the user equipment can pinpoint specific data partitions relevant to the accuracy issue rather than downloading complete datasets. The proposed concept results in more efficient use of network resources and faster model improvement cycles.

[0023] Some aspects of the present disclosure relate to a method for a network node of a mobile communication system. The method comprises transmitting, to a user equipment, UE, information relating to a plurality of dataset partitions. Each dataset partition is associated with a respective partition identifier. The information enables the user equipment to identify and selectively retrieve a dataset partition associated with an202500709 5

[0024] inference accuracy issue of a machine-learning model. In response to receiving a request from the user equipment indicating a specific partition identifier, the method further comprises transmitting the dataset partition associated with the specific partition identifier to the user equipment. By providing partition-specific information and responding to targeted retrieval requests, the network node facilitates efficient and selective data transfer that addresses only the relevant accuracy issues, thereby optimizing network resource utilization.

[0025] To enable the user equipment to autonomously determine when an inference accuracy issue has occurred, the network node may provide threshold information. For example, the method may further comprise transmitting, to the user equipment, an accuracy threshold for use by the user equipment in determining the inference accuracy issue. By transmitting the accuracy threshold, the network node establishes a consistent criterion against which the user equipment can evaluate its inference results, ensuring uniform detection of accuracy degradation across the system.

[0026] To facilitate straightforward retrieval of dataset partitions, the information relating to the plurality of dataset partitions may be structured to include direct access mechanisms. In some examples, the information relating to the plurality of dataset partitions may comprise a mapping between each partition identifier and a corresponding downloadable link. This allows the user equipment to directly download the required dataset partition without additional signaling, reducing retrieval latency and simplifying the data acquisition process.

[0027] For flexible delivery of partition information to user equipment under varying conditions, the network node may employ different transmission mechanisms. In various examples, the method may further comprise transmitting the information relating to the plurality of dataset partitions via at least one of a system information message broadcast in a cell served by the network node or a user equipment-specific message. The system information message approach enables efficient distribution to multiple user equipment simultaneously, while the user equipment-specific message allows for tailored information delivery to individual devices.202500709 6

[0028] To enhance the relevance of dataset partitions to specific operational contexts, the partitioning strategy may consider contextual parameters. For example, the method may further comprise partitioning the plurality of dataset partitions based on at least one of a time parameter or a location parameter. By partitioning based on time or location parameters, the dataset partitions can be organized to reflect temporal or spatial variations in the data, enabling the user equipment to retrieve partitions that are most relevant to its current operating conditions.

[0029] Some aspects of the present disclosure relate to a method for user equipment, UE, in a mobile communication system. The method comprises receiving, from a network node, information relating to a plurality of dataset partitions, each dataset partition being associated with a respective partition identifier. The method further comprises performing an inference operation using a machine-learning model, which is based on one or more of the plurality of dataset partitions. The method comprises determining that an inference result of the inference operation does not meet an accuracy criterion. The method further comprises identifying, from among the plurality of dataset partitions, at least one dataset partition associated with the inference result not meeting the accuracy criterion. The method comprises initiating retrieval of the identified at least one dataset partition based on the respective partition identifier. By enabling the user equipment to identify and selectively retrieve specific dataset partitions associated with accuracy issues, the proposed concept reduces unnecessary data transfer and accelerates model refinement.

[0030] To establish a concrete basis for accuracy evaluation, the user equipment may receive threshold information from the network. In some examples, the method may further comprise receiving, from the network node, an accuracy threshold. The determining that the inference result does not meet the accuracy criterion may comprise comparing the inference result against the accuracy threshold. This provides a quantitative measure for detecting accuracy degradation, enabling systematic and consistent identification of inference accuracy issues.

[0031] To streamline the data retrieval process, the partition information may include direct access mechanisms. In various examples, the information relating to the plurality of dataset partitions may comprise a mapping between each partition identifier and a202500709 7

[0032] corresponding downloadable link. Initiating retrieval may comprise downloading the identified at least one dataset partition from the corresponding downloadable link. By utilizing downloadable links, the user equipment can efficiently acquire the required data without additional request-response exchanges with the network node.

[0033] In scenarios where direct download is not available or where network-coordinated retrieval is preferred, the user equipment may request data from the network node. For example, initiating retrieval may comprise transmitting, to the network node, a request message including the partition identifier of the identified at least one dataset partition. This enables the network node to control and manage the data distribution process.

[0034] For compatibility with standard mobile communication signaling procedures, the partition information may be conveyed through established message types. In some examples, receiving the information relating to the plurality of dataset partitions may comprise receiving at least one of a system information message or a UE-specific / dedicated message, such as a Radio Resource Control (RRC) reconfiguration message. The use of system information messages enables broadcast delivery to multiple user equipment, while RRC reconfiguration messages allow for user equipment-specific configuration.

[0035] Some aspects of the present disclosure relate to a computer program having a program code for performing one of the methods described above, when the computer program is executed on a computer, a processor, or a programmable hardware component. By providing the methods in the form of a computer program, the proposed concept can be flexibly deployed and updated on various hardware platforms within the mobile communication system.

[0036] Some aspects of the present disclosure relate to an apparatus for a network node of a mobile communication system. The apparatus comprises one or more interfaces configured to communicate in the mobile communication system. The apparatus further comprises one or more processing devices configured to perform one of the network node methods described above. By implementing the proposed concept in an apparatus for a network node, the technique for managing dataset partitions and responding to user equipment requests can be integrated into the network infrastructure.202500709 8

[0037] Some aspects of the present disclosure relate to an apparatus for user equipment, UE, of a mobile communication system. The apparatus comprises one or more interfaces configured to communicate in the mobile communication system. The apparatus further comprises one or more processing devices configured to perform one of the user equipment methods described above. By implementing the proposed concept in an apparatus for user equipment, the technique for identifying and selectively retrieving dataset partitions associated with inference accuracy issues can be realized in mobile devices.

[0038] Some aspects of the present disclosure relate to a network node of a mobile communication system comprising the apparatus described above for a network node. The network node is thereby enabled to manage and distribute dataset partitions in support of machine-learning model refinement across the mobile communication system.

[0039] Some aspects of the present disclosure relate to user equipment, UE, of a mobile communication system comprising the apparatus described above for user equipment. The user equipment is thereby enabled to perform inference operations using a machine-learning model and to selectively retrieve dataset partitions when inference accuracy issues are detected.

[0040] Further aspects of the present disclosure relate to a mobile communication system comprising the network node and the UE as described above. This provides a complete system implementation that enables coordinated machine-learning model management across network infrastructure and user devices.

[0041] Some examples of apparatuses, computer programs and / or methods will be described in the following by way of example only, and with reference to the accompanying figures, in which

[0042] Fig. 1 shows an example of a flowchart of a method for a network node;

[0043] Fig. 2 depicts an example of a flowchart of a method for user equipment;202500709 9

[0044] Fig. 3 illustrates block diagrams of examples of apparatuses for a network node, user equipment, and a communication network;

[0045] Fig. 4 shows an evaluation procedure at user equipment in an example;

[0046] Fig. 5 depicts an example of partition evaluation at the UE; and

[0047] Fig. 6 illustrates a flowchart of another example of a method for user equipment and a flowchart of an example of a method for a network node.

[0048] Some examples are now described in more detail with reference to the enclosed figures. However, other possible examples are not limited to the features of these embodiments described in detail. Other examples may include modifications of the features, as well as equivalents and alternatives to the features. Furthermore, the terminology used herein to describe certain examples should not be restrictive of further possible examples.

[0049] Throughout the description of the figures, same or similar reference numerals refer to same or similar elements and / or features, which may be identical or implemented in a modified form while providing the same or a similar function. The thickness of lines, layers, and / or areas in the figures may also be exaggerated for the sake of clarification.

[0050] When two elements A and B are combined using an "or", this is to be understood as disclosing all possible combinations, i.e. , only A, only B, as well as A and B, unless expressly defined otherwise in the individual case. As an alternative wording for the same combinations, "at least one of A and B" or "A and / or B" may be used. This applies equivalently to combinations of more than two elements.

[0051] If a singular form, such as "a", "an", and "the" is used and the use of only a single element is not defined as mandatory either explicitly or implicitly, further examples may also use several elements to implement the same function. If a function is described below as implemented using multiple elements, further examples may implement the same function using a single element or a single processing entity. It is further202500709 10

[0052] understood that the terms "include", "including", "comprise", and / or "comprising", when used, describe the presence of the specified features, integers, steps, operations, processes, elements, components, and / or a group thereof, but do not exclude the presence or addition of one or more other features, integers, steps, operations, processes, elements, components, and / or a group thereof.

[0053] Fig. 1 shows an example of a flowchart of method 10 for a network node. Method 10 for the network node of a mobile communication system comprises transmitting 12, to a UE, information relating to a plurality of dataset partitions. Each dataset partition is associated with a respective partition identifier. The information enables the UE to identify and selectively retrieve a dataset partition associated with an inference accuracy issue of a machine-learning model. In response to receiving a request from the UE indicating a specific partition identifier, method 10 further comprises transmitting 14 the dataset partition associated with the specific partition identifier to the user equipment.

[0054] Fig. 2 depicts an example of a flowchart of method 20 for a UE. Method 20 for the UE in a mobile communication system comprises receiving 21, from a network node, information relating to a plurality of dataset partitions, each dataset partition being associated with a respective partition identifier. Method 20 also includes performing 22 an inference operation using a machine-learning model, which is based on one or more of the plurality of dataset partitions. Method 20 further comprises determining 23 that an inference result of the inference operation does not meet an accuracy criterion and identifying 24, from among the plurality of dataset partitions, at least one dataset partition associated with the inference result not meeting the accuracy criterion. Method 20 also includes initiating 25 retrieval of the identified at least one dataset partition based on the respective partition identifier.

[0055] Fig. 3 illustrates block diagrams of examples of apparatuses 30, 40 for a network node 300, user equipment 400, and a communication network 500. Fig. 3 illustrates an apparatus 30 for a network node 300 of a mobile communication system 500. The apparatus 30 comprises one or more interfaces 32 configured to communicate in the mobile communication system 500. The one or more interfaces 32 are coupled to one202500709 11

[0056] or more processing devices 34, which are configured to perform one of the methods 10 for the network node 300 as described herein.

[0057] Fig. 3 also shows an apparatus 40 for a UE 400 of the mobile communication system 500. The apparatus 40 comprises one or more interfaces 42 configured to communicate in the mobile communication system 500. The one or more interfaces 42 are coupled to one or more processing devices 44, which are configured to perform one of the methods 20 for the UE 400 as described herein.

[0058] As optional components (dashed lines), optional from the perspective of the apparatuses 30, 40, Fig. 3 also shows a network node 300 of the mobile communication system 500 comprising the apparatus 30 and a UE 400 of the mobile communication system 500 comprising the apparatus 40. Fig. 3 also illustrates an example of the mobile communication system 500 comprising the network node 300 and the UE 400.

[0059] As illustrated in Fig. 3, the respective one or more signal processing devices 34, 44 are coupled to the respective one or more interfaces 32, 42. The one or more interfaces 32, 42 may serve as an interface for communicating in the communication system 500. The one or more interfaces 32, 42 may correspond to one or more inputs and / or outputs for receiving and / or transmitting information, which may be in digital (bit) values or analog according to a specified code or protocol, within a module, between modules, or between modules of different entities. For example, an interface may comprise interface circuitry configured to receive and / or transmit information. In examples, an interface may comprise any means for obtaining, receiving, transmitting, or providing analog or digital signals or information, e.g., any connector, contact, pin, register, input port, output port, conductor, lane, etc., which allows providing or obtaining a signal or information.

[0060] The one or more interfaces 32, 42 may be configured to communicate (transmit, receive, or both) in a wireless and / or wired manner, and they may be configured to communicate, i.e., transmit and / or receive signals or information with further internal or external components. The one or more interfaces 32, 42 or the apparatuses 30, 40 may comprise further components to enable communication in a (mobile) communication system or network; such components may include transceiver (transmitter and / or202500709 12

[0061] receiver) components, such as one or more Low-Noise Amplifiers (LNAs), one or more Power-Amplifiers (PAs), one or more duplexers, one or more diplexers, one or more filters or filter circuitry, one or more converters, one or more mixers, accordingly adapted radio frequency components, one or more antennas, etc. For example, the respective one or more interfaces 32, 42 may enable radio communication with UEs and communication between base stations, which can be direct and / or indirect, wired and / or wireless, respectively.

[0062] The one or more (signal) processing devices 34, 44 may be implemented using one or more processing units, one or more circuitries, or any means for processing, such as a processor, a computer, or a programmable hardware component being operable with accordingly adapted software. In other words, the described function of the one or more processing devices 34, 44 may as well be implemented in software, which is then executed on one or more programmable hardware components. Such hardware components may comprise a general-purpose processor, a Digital Signal Processor (DSP), a micro-controller, etc..

[0063] In examples, network node 300 may also be referred to as a base station, server, network device, etc. It may belong to an access network or to a core network. A network entity / node may correspond to a remote radio head, a transmission point, an access point, a macro cell, a small cell, a micro cell, a pico cell, a femto cell, or a metro cell. The term small cell may refer to any cell smaller than a macro cell, e.g., a micro cell, a pico cell, a femto cell, or a metro cell. A network node / base station can be a wireless interface of a wired network, which enables transmission and reception of radio signals to a communication device. Such a radio signal may comply with radio signals as, for example, standardized by 3GPP or, generally, in line with one or more of the abovelisted systems. Thus, a network entity / node may be a base station and may correspond to a NodeB, an eNodeB, an ngNB, a gNB, a BTS (Base Transceiver Station), or an access point, all of which may be implemented in a terrestrial station, a satellite, plane, HAPS (High Altitude Platform), etc. In the case of a moving implementation in a satellite, an airplane, etc., the link towards a core network of the communication system may also be implemented in a wireless manner.202500709 13

[0064] The mobile communication system 500 may hence be cellular. The term cell refers to a coverage area of radio services provided by a transmission point, a remote unit, a remote head, a remote radio head, a communication device, a network entity, or a NodeB, an eNodeB, an ngNB, a gNB, a beam, or a satellite, respectively. The terms cell and base station may be used synonymously; a base station may generate multiple cells and it may be implemented in a terrestrial station, a high-altitude platform, a plane, a drone, a satellite, etc. A UE or wireless communication device can be registered or associated with at least one cell (e.g., the network entity / node); e.g., it can be associated with a cell such that data can be exchanged between the network and the mobile in the coverage area of the associated cell using a dedicated channel, connection, or link.

[0065] In general, the UE is capable of communicating wirelessly. In particular, however, the UE / communication device may be a mobile communication device, e.g., a communication device that is suitable for being carried around by a user. For example, the communication device may be a User Terminal (UT) or UE within the meaning of the respective communication standards being used for mobile communication. For example, the communication device may be a mobile phone, such as a smartphone, a network access device embedded in a vehicle, ship, or airplane, or another type of mobile communication device, such as a computer, a laptop computer, a tablet computer, and so on.

[0066] For example, the UE / communication device and the network entity / node may be configured to communicate in a cellular mobile communication system. Accordingly, the UE and the network node may be configured to communicate in a cellular mobile communication system, for example in a Sub-6GHz-based cellular mobile communication system (covering frequency bands between 400 MHz and, in the meantime, 7 GHz), in a mmWave-based cellular mobile communication system (covering frequency bands between 24 GHz and 71 GHz), or in the so-called mid-bands (covering frequency bands between 7 GHz and 24 GHz). For example, the UE and the network node may be configured to communicate in a mobile communication system / cellular mobile communication system.202500709 14

[0067] In general, the mobile communication system may, for example, correspond to one of the 3GPP-standardized mobile communication networks, where the term mobile communication system is used synonymously with mobile communication network. The mobile communication system may correspond to, for example, a 6th Generation system (6G), a 5th Generation system (5G), a New Radio (NR) system, Long-Term Evolution (LTE), an LTE-Advanced (LTE-A) system, High Speed Packet Access (HSPA), a Universal Mobile Telecommunication System (UMTS), ora UMTS Terrestrial Radio Access Network (UTRAN), an evolved-UTRAN (e-UTRAN), a Global System for Mobile communication (GSM) or Enhanced Data rates for GSM Evolution (EDGE) network, a GSM / EDGE Radio Access Network (GERAN), or mobile communication networks with different standards, for example, generally an Orthogonal Frequency Division Multiple Access (OFDMA) network, a Time Division Multiple Access (TDMA) network, a Code Division Multiple Access (CDMA) network, a Wideband-CDMA (WCDMA) network, a Frequency Division Multiple Access (FDMA) network, a Spatial Division Multiple Access (SDMA) network, etc..

[0068] As further illustrated in Fig. 3, the apparatus 30 for the network node 300 provides / transmits information related to the plurality of dataset partitions to the apparatus 40 in the UE 400. The apparatus 40 at the UE 400 can then perform the inference operation and identify a specific partition that has an accuracy issue (steps 22, 23, and 24 in Fig. 2). Subsequently, the specific partition can be requested by the apparatus 40, and in response to said request, apparatus 30 can provide the requested partition to apparatus 40.

[0069] Machine-learning and artificial intelligence models deployed at user equipment in examples may be in accordance with 3GPP specifications for the NR air interface. They may be categorized according to several distinguishing characteristics that determine their operational behavior, deployment requirements, and lifecycle management procedures. There are one-sided models and two-sided models, which are distinguished based on the location of inference operations. In a one-sided UE-sided model, the complete inference operation is performed locally at the user equipment, with the model receiving input data derived from measurements or observations made by the device and producing output predictions or decisions without requiring real-time collaboration with the network during inference.202500709 15

[0070] One-sided UE-sided models are applicable to use cases such as time-domain channel state information prediction, wherein the user equipment predicts future channel conditions based on historical measurements, beam management applications wherein the user equipment predicts optimal beam selections based on reference signal measurements of a subset of available beams, and positioning / handover prediction applications wherein the user equipment determines its location based on radio measurements using techniques such as fingerprinting. In contrast, two-sided models distribute the inference operation across both the user equipment and the network, with each entity performing a portion of the overall inference task. The primary application of two-sided models in 3GPP specifications is spatial-frequency domain channel state information compression, wherein the user equipment operates an encoder model to compress channel information into a reduced representation for transmission over the air interface, and the network operates a corresponding decoder model to reconstruct the full channel state information from the received compressed representation.

[0071] Machine-learning models at the user equipment may further be distinguished according to their identification and lifecycle management approach, example definitions can be found in 3GPP specifications. Under functionality-based lifecycle management, the network and user equipment maintain mutual understanding of AI / ML-enabled features or feature groups, referred to as functionalities, without necessarily identifying the specific underlying AI / ML model or models that implement each functionality. The user equipment reports its supported functionalities through capability signaling mechanisms, and the network controls the selection, activation, deactivation, switching, and fallback of functionalities through standardized signaling, while the specific model implementations remain proprietary and potentially unknown to the network. Under model-identification-based lifecycle management, each AI / ML model is assigned a unique model identifier that enables both the network and user equipment to reference and manage specific models explicitly.

[0072] The model identifier may be accompanied by additional conditions specifying the scenarios, sites, configurations, or datasets for which the model is applicable, enabling fine-grained control over model selection based on current operating conditions. Models may further be characterized according to their generalization properties, ranging from202500709 16

[0073] models designed to generalize across multiple scenarios, configurations, and sites to scenario-specific, configuration-specific, or site-specific models optimized for particular operating conditions. The selection among multiple available models, the switching between models in response to changing conditions, and the updating or fine-tuning of models to address performance degradation constitute key lifecycle management operations that may be performed by the user equipment autonomously, under network control, or through collaborative procedures between the user equipment and network depending on the deployment configuration and collaboration level defined for the AI / ML-enabled feature.

[0074] Partitioning in the context of machine-learning models, e.g. within 3GPP specifications, encompasses multiple dimensions that determine how data, model components, and training responsibilities are distributed across network entities and user equipment. Data partitioning represents a fundamental distinction in federated learning approaches, wherein horizontal federated learning employs partitioning by samples such that multiple participating entities possess datasets with the same feature space but containing different data samples, while vertical federated learning employs partitioning by features such that participating entities possess datasets with different feature spaces that relate to the same set of samples. Model partitioning also applies to two-sided AI / ML models, wherein the inference operation is split between the user equipment and the network, with each entity responsible for executing a designated portion of the overall model. In the representative use case of spatial-frequency domain channel state information compression, the model is partitioned such that the user equipment operates an encoder component that compresses channel information, while the network operates a corresponding decoder component that reconstructs the full channel state information from the compressed representation.

[0075] Dataset partitioning for model development follows established machine-learning practices, wherein collected data is divided into training datasets used to learn model parameters, validation datasets used to tune hyperparameters and prevent overfitting, and test datasets used to evaluate final model performance under conditions representative of operational deployment. The partitioning of lifecycle management responsibilities between the network and user equipment constitutes an additional dimension, wherein different collaboration levels determine which entity is responsible202500709 17

[0076] for functions such as data collection, model training, model selection, performance monitoring, and triggering of model updates or switching operations. The appropriate partitioning strategy for a given deployment depends on factors including the computational capabilities of participating entities, communication overhead constraints, privacy requirements governing data sharing, and the degree of coordination required between network and user equipment to achieve reliable AI / ML-enabled feature performance.

[0077] In examples, basically any partitioning strategy or concept is conceivable. The partitions may correspond to sub-models or sub-modules of the machine-learning model that is used at the UE. For example, there may be a functional split between the partitions such that there is one partition for CSI prediction, another for beam prediction and yet another for position prediction. While these functions are at a rather high level it is noted that in other examples the functional split may as well be at a lower level, e.g. there may be one partition for pathloss or reception power prediction and another partition for interference prediction, and yet another for network load prediction, all contributing to a higher level data rate predictor.

[0078] A baseline dataset, in the context of example machine-learning models deployed at UEs in wireless communication systems, comprises a defined set of parameters and associated reference values that enable the evaluation of model / partition performance under controlled and reproducible conditions. The baseline dataset serves as a standardized benchmark against which the inference outputs of a machine-learning model / partition can be compared to assess prediction accuracy, detect performance degradation, and determine whether the model continues to meet specified performance thresholds. The parameters constituting the baseline dataset may include input features representative of expected operational conditions, corresponding ground truth values or labels that represent correct model outputs, and metadata describing the scenarios, configurations, or environmental conditions under which the baseline data was collected or generated.

[0079] By maintaining a baseline dataset at the UE, the device can periodically evaluate its deployed machine-learning model partitions by providing baseline inputs to the inference function and comparing the resulting outputs against the known reference202500709 18

[0080] values contained in the dataset, thereby enabling autonomous performance assessment without requiring real-time ground truth measurements from the network.

[0081] The composition and characteristics of the baseline dataset are tailored to the specific use case for which the machine-learning model / partition is deployed. For channel state information prediction models, the baseline dataset may comprise historical channel measurements paired with corresponding actual channel realizations, enabling the user equipment to assess prediction accuracy by computing error metrics between model / partition outputs and reference values. For beam management models, the baseline dataset may include reference signal measurements for a subset of beams together with known optimal beam selections, allowing evaluation of beam prediction accuracy. For positioning models, the baseline dataset may contain radio measurements associated with known reference positions, enabling assessment of positioning error statistics. The baseline dataset may be provisioned to the UE during initial model deployment, updated periodically by the network to reflect changing operational conditions, or generated locally by the user equipment during periods when ground truth information is available. The use of a baseline dataset for model / partition evaluation provides a computationally efficient and overhead-reducing alternative to continuous ground-truth-based monitoring, enabling the user equipment to detect model performance degradation, trigger lifecycle management actions such as model switching or update requests, and ensure that AI / ML-enabled features maintain performance levels equal to or exceeding those of legacy non-AI / ML-based operations.

[0082] Fig. 4 shows an evaluation procedure at a UE in an example. Fig. 4 shows a UE 400, which evaluates an installed AI / ML model / partition based on a baseline dataset, which can also be partition-specific. The baseline dataset comprises a set of input parameters and the AI / ML model provides a prediction, e.g. channel state information. The baseline dataset also provides output values, e.g. for a given environment and network conditions (defined by the input parameters) the baseline dataset also provides the output, i.e. the channel state information that should result. In Fig. 4 this process is illustrated as prediction. The predicted value can then be compared to the desired output from the baseline dataset. As a result, an inaccuracy can be determined as shown on the right of Fig. 4. Without bias the upper graph would result and with the actual model / partition the lower graph results. The difference can be compared to the202500709 19

[0083] accuracy threshold obtained from the network node. The method 10 for the network node then further comprises transmitting, to the UE, an accuracy threshold (can be partition specific) for use by the UE in determining the inference accuracy issue. The method 20 for the UE then comprises receiving, from the network node, an accuracy threshold and wherein the determining that the inference result does not meet the accuracy criterion comprises comparing the inference result against the accuracy threshold.

[0084] If the threshold is exceeded as exemplified in the viewgraph of Fig. 4 with the value on the right, model / partition adaptation can be triggered. The bias may, for example, occur because of changing environmental data, disadvantageous retraining of the model, etc. From the viewgraph on the right, it can be seen at (“X”), due to model bias the accuracy drops compared to a model without bias and therefore there is a need for corrective measure.

[0085] In examples, an evaluation at the UE may comprise calculating a prediction accuracy of the machine-learning model / partition. For example, this may be carried out by comparing predicted values and measured values. If the outcome of the comparison is that the prediction accuracy is better than demanded by the accuracy threshold (threshold not exceeded), the machine-learning model may be kept unchanged. In case the accuracy requirement is not met, further evaluation may be carried out (if the prediction accuracy does not meet the accuracy requirement). The accuracy requirement may be implemented by a threshold comparison. The accuracy threshold may be defined as an upper or lower threshold and the requirement may be defined relative to the threshold, e.g., the determined accuracy is higher or lower than the threshold. In examples, equality to the threshold may be defined as meeting or not meeting the accuracy requirement.

[0086] The further evaluation may comprise testing the machine-learning model / partition to obtain a tested machine-learning model / partition based on the baseline dataset in case the accuracy of the machine-learning model / partition does not meet the accuracy requirement. This is shown in Fig. 4. The further evaluation comprises calculating a prediction accuracy of the tested machine-learning model / partition and comparing the prediction accuracy to the accuracy requirement.202500709 20

[0087] The requesting of the machine-learning model / partition adaptation is carried out if the prediction accuracy of the tested machine-learning model / partition fails to meet the accuracy requirement based on the accuracy threshold. On the network side, the request for machine-learning-model / partition adaptation is received from the UE and the partition can be identified using the assignment of the identifier as initially provided by the network node.

[0088] In examples different options for the model / partition adaptation are conceivable. For example, the request is a request for retraining of the machine learning model / partition. The method may then further comprise retraining the machine-learning model / partition at the UE and / or at a training node. Hence, the retraining may be carried out at the UE and the network node would then provide retraining data to the UE. At the UE side the requesting may comprise requesting a retraining of the machine-learning model / partition and the method may further comprise receiving retraining data from the network node and retraining the machine-learning model / partition based on the retraining data.

[0089] Another option is that the re-training of the model / partition is carried out at the network node or any other node in the network also referred to as re-training node. For example, there may be a dedicated maintenance server in the network that maintains, monitors and controls the machine-learning models / partitions and that may also be used for retraining. From the UE perspective the adaptation of the machine-learning model / partition may then be a download of a re-trained model. In method 20 for the UE apparatus 40 the requesting then comprises requesting a download of the machinelearning model or an adapted machine-learning model. Method 20 then further comprises downloading the machine-learning model or the adapted machine-learning model via the network node and installing the downloaded machine-learning model.

[0090] The request may be a request for downloading the machine-learning model / partition or an adapted machine-learning model / partition. The method 10 at the network node apparatus may then further comprise forwarding the machine-learning model or the adapted machine-learning model to the UE. As outlined above, in some examples the partition / model may just be re-downloaded, e.g. when the model at the UE was202500709 21

[0091] developed in a disadvantageous direction, e.g. by learning / training / adaptation algorithms at the UE.

[0092] In some examples, the network node method 10 may further comprise dynamically adjusting the accuracy threshold. The configuring then comprises dynamically signaling the accuracy threshold to the UE. Dynamic adaptation may result in dynamic model adaptation. The network may also use the accuracy threshold to coordinate partition / model updates. By adjusting the threshold the network may coordination requested partition updates.

[0093] At least in some examples, the machine-learning model is a federated machine-learning model. Federated learning represents a distributed machine-learning paradigm, also adopted within 3GPP specifications, that enables the collaborative training of AI / ML models across multiple decentralized entities without requiring the exchange or centralized aggregation of raw training data. In the context of 3GPP wireless communication systems, federated learning addresses fundamental challenges related to data privacy, communication efficiency, and the practical constraints of collecting training data from geographically distributed user equipment and network nodes. The 3GPP architecture supports federated learning through the Network Data Analytics Function (NWDAF), which can coordinate training operations across multiple decentralized NWDAF instances or other participating entities.

[0094] In a federated learning configuration, user equipment performs partial training operations based on locally available data, generating local model updates that are transmitted to a central aggregation entity, which combines the contributions from multiple participants to produce an updated global model without accessing the underlying local datasets. The 5G system provides assistance for application layer federated learning operations, including candidate federated learning member selection according to specific criteria such as user equipment performance, location, trajectory, and network resource availability. The Network Exposure Function (NEF) may assist AI / ML application servers in scheduling available user equipment to participate in federated learning operations, enabling dynamic coordination of distributed training tasks across the network.202500709 22

[0095] The 3GPP specifications distinguish between two primary variants of federated learning that differ in how data is partitioned across participating entities. Horizontal federated learning, defined in Release 18, applies to scenarios where multiple participating entities possess datasets that share the same feature space but contain different data samples. In horizontal federated learning, each participant trains a local model using its own data samples, and the resulting model updates are aggregated to produce a global model that benefits from the collective knowledge across all participants without exposing individual data samples. Vertical federated learning (VFL), which became a focus of Release 19 studies, addresses scenarios where participating entities possess datasets with different feature spaces that must be combined to jointly train a global model. In vertical federated learning, different entities contribute complementary features or measurements related to the same set of samples, enabling the training of models that leverage diverse data sources while preserving the privacy of each entity's proprietary features. The 3GPP architecture supports vertical federated learning through the NWDAF or Application Functions (AF), which may assume the role of VFL server or VFL clients. Security aspects of federated learning operations are addressed through authorization procedures for entities assuming VFL server roles and protection of information exchange during the federated learning process using Transport Layer Security (TLS). The 5G system further enables dynamic addition or removal of specific user equipment to or from AI / ML federated learning tasks, including scenarios where user equipment communicates via direct device connections, providing flexibility in the composition of federated learning participants based on evolving network conditions and application requirements.

[0096] In examples, life cycle management may require model update, model transfer, model parameters re-configuration between entities, which may have subsequent overhead. The UE receives the partitions of the baseline dataset from the gNB, which is associated with the partitioned dataset ID. Thereby, an overhead for obtaining a baseline dataset for model accuracy may be reduced. The baseline dataset can be modified as soon as model accuracy falls below the required accuracy.

[0097] The UE receives the partitions of the baseline dataset from the gNB, which is associated with the partitioned dataset ID (identifier). The UE receives the mapping between baseline dataset partition and partition ID / downloadable links through a system202500709 23

[0098] information message, e.g., a system information block (SIB), or through dedicated signaling, e.g., a UE-specific message such as an RRC Reconfiguration message. The UE receives the partitions of the baseline dataset from the gNB, which is associated with the partitioned dataset ID.

[0099] The gNB provides the partitioned baseline dataset based on network implementation, for example, time, location, etc. The UE receives the accuracy threshold from the gNB, which can be a single value or can be associated with the partitioned dataset. The UE receives this information via system information message / UE specific message (for example, RRC Reconfiguration message). The UE computes model inference based on dataset partitions and checks the inference against the model accuracy threshold. If the model accuracy threshold is crossed, the UE analyzes its model and checks which dataset partitions are resulting in error; the UE downloads the baseline dataset for that particular partition to check for model accuracy.

[0100] The UE requests the gNB with the partition ID for which the model has accuracy bias, or it can download the dataset partition from the downloadable link.

[0101] The UE receives the partitions of the baseline dataset from the gNB, which is associated with the partitioned dataset ID. For example, the UE receives the mapping between baseline dataset partition and partition ID / downloadable links through a system information message / UE specific message (for example RRC Reconfiguration message). The gNB provides the partitioned baseline dataset based on network implementation, for example, time, location, etc..

[0102] The UE receives the accuracy threshold from the gNB, which can be a single value or can be associated with the partitioned dataset. The UE may receive this information also via system information message or a UE specific message (for example RRC Reconfiguration message).

[0103] At the UE the data may be represented as shown in the following table:

[0104]

[0105] 202500709 24

[0106]

[0107] The UE can then compute the model inference based on dataset partitions and check the inference against the model accuracy threshold. If the model accuracy threshold is crossed, the UE analyses its model and checks which dataset partitions are resulting in error; the UE downloads the baseline dataset for that particular partition to check for model accuracy. The download can be requested as the links to the data are already available at the UE. The UE may request the gNB with the partition ID for which model has accuracy bias or it can download the dataset partition from the downloadable link.

[0108] Fig. 5 illustrates this scenario. Fig. 5 depicts an example of partition evaluation at the UE. Fig. 5 shows the baseline partitioned dataset on the left, where the respective partitions have identifiers and associated links. The evaluation result at the UE is shown in the middle of Fig. 5 where the partition with ID3 is assumed to not meet the accuracy requirement, which is indicated by the hachure. Consequently, the UE downloads the baseline data set which resulted in model bias and updates the respective partition. In this example the baseline data set is updated. In other examples a sub-model / module of the machine-learning model may be updated. The information relating to the plurality of dataset partitions may comprise a mapping between each partition identifier and a corresponding downloadable link, be it for the baseline dataset or for the model / partition adaptation. As outlined above, method 10 may comprise transmitting information relating to the plurality of dataset partitions via at least one of a system information message broadcast in a cell served by the network node or a user equipment-specific message. The partitioning of the plurality of dataset partitions may be based on at least one of a time parameter or a location parameter. Hence, different partitions may apply to different locations or different points in time (morning / evening / night).

[0109] Summarizing, the UE receives the partitions of the baseline dataset from the gNB which is associated with the partitioned dataset ID. The UE may also receive the mapping between baseline dataset partition and partition ID / downloadable links through system information message / UE specific message (for example RRC Reconfiguration202500709 25

[0110] message). The UE may receive the accuracy threshold from the gNB, which can be a single value or multiple values which are associated with the partitioned datasets. In other words, in some examples different accuracy thresholds may be used for different partitions. The UE may receive the accuracy threshold via broadcast, e.g. a system information message, or via dedicated signaling, e.g. a UE specific message (for example RRC Reconfiguration message). The UE may then request the gNB with the partition ID for which model has accuracy bias or it can download the dataset partition from the downloadable link.

[0111] Fig. 6 illustrates a flowchart of another example of a method for user equipment and a flowchart of an example of a method for a network node. Fig. 6 shows a gNB at the top which provides a mapping between the partitioned baseline data set together with partition ID and downloadable links to a UE in step 602. Fig. 6 shows an example of a UE process at the bottom. In step 604 the UE evaluates the model inference based on an accuracy threshold. If the accuracy requirement is met the UE continues in step 606 and keeps using the machine-learning model with the partition as they are. In case the accuracy requirement is not met, in step 608 the UE analyzes the problematic dataset and requests the gNB to download the partitioned dataset from the link. Hence, on the UE side the initiating of the retrieval comprises downloading the identified at least one dataset partition from the corresponding downloadable link. The initiating of the retrieval may also comprise transmitting, to the network node, a request message including the partition identifier of the identified at least one dataset partition.

[0112] Performance monitoring constitutes a critical component of the AI / ML lifecycle management framework, serving as the primary mechanism for detecting when accumulated bias causes model performance to degrade below acceptable thresholds. In examples, this is done by evaluating the accuracy requirement defined by the accuracy threshold. In the context of 3GPP specifications for AI / ML-enabled features in the NR air interface, performance monitoring may encompass the continuous evaluation of model outputs against expected performance criteria, the computation of relevant performance metrics, the reporting of monitoring results to management entities, and the triggering of corrective actions when performance degradation is detected. The implementation of such monitoring mechanisms may require the definition of202500709 26

[0113] appropriate key performance indicators (KPIs) that quantify model accuracy and reliability under operational conditions.

[0114] For channel state information (CSI) prediction, relevant metrics (accuracy criteria that may be compared to the accuracy threshold) may include the mean squared error between predicted and actual CSI values, the correlation coefficient between predicted and measured channel matrices, or the resulting throughput performance when predictions are used for link adaptation. For beam management applications, monitoring metrics may encompass beam prediction accuracy, the frequency of beam misalignment events, or the overhead reduction achieved relative to exhaustive beam sweeping.

[0115] For positioning use cases, metrics such as horizontal and vertical positioning error, the cumulative distribution function of positioning accuracy, or the percentage of estimates meeting specified accuracy requirements provide quantifiable measures of model performance.

[0116] The practical implementation of performance monitoring mechanisms in wireless communication systems may follow several architectural approaches. In a ground-truth-based monitoring approach (baseline dataset), the UE or network periodically obtains actual measurements or outcomes that can be compared against model predictions to compute error metrics directly. For example, in CSI prediction, the UE may periodically measure actual channel conditions and compare these measurements against previously generated predictions to calculate prediction error statistics. This approach provides accurate performance assessment but incurs overhead from the additional measurements required to obtain ground truth. In a reference-signal-based monitoring approach, dedicated reference signals or pilot transmissions enable the evaluation of model performance without requiring full ground-truth measurements. The network may configure specific monitoring occasions during which reduced-complexity measurements are performed to assess prediction quality. In an inference-consistency-based monitoring approach, the temporal or spatial consistency of model outputs is evaluated without explicit ground truth, wherein sudden changes in prediction patterns or statistical anomalies in output distributions may indicate performance degradation. A hybrid monitoring approach may combine elements of these methods, using lightweight202500709 27

[0117] consistency checks during normal operation while periodically invoking more comprehensive ground-truth-based evaluation during dedicated monitoring intervals.

[0118] The monitoring framework may further require mechanisms for aggregating performance metrics over appropriate time windows, comparing aggregated metrics against predefined thresholds, and communicating monitoring outcomes to trigger lifecycle management actions. When monitored performance falls below configured thresholds, the monitoring mechanism may initiate model switching to select an alternative model better suited to current conditions, trigger model update or fine-tuning procedures to adapt the current model, or activate fallback to legacy non-AI / ML-based operations to ensure continued reliable service. The signaling infrastructure supporting performance monitoring may utilize existing 3GPP reporting mechanisms enhanced with AI / ML-specific information elements, or may employ dedicated signaling procedures defined specifically for AI / ML lifecycle management. The balance between monitoring accuracy, signaling overhead, computational complexity, and power consumption at the UE represents a key design consideration in the implementation of practical performance monitoring mechanisms for AI / ML-enabled features in wireless communication systems.

[0119] The aspects and features described in relation to a particular one of the previous examples may also be combined with one or more of the further examples to replace an identical or similar feature of that further example or to additionally introduce the features into the further example.

[0120] Examples may further be or relate to a (computer) program including a program code to execute one or more of the above methods when the program is executed on a computer, processor or other programmable hardware component. Thus, steps, operations or processes of different ones of the methods described above may also be executed by programmed computers, processors or other programmable hardware components. Examples may also cover program storage devices, such as digital data storage media, which are machine-, processor- or computer-readable and encode and / or contain machine-executable, processor-executable or computer-executable programs and instructions. Program storage devices may include or be digital storage devices, magnetic storage media such as magnetic disks and magnetic tapes, hard disk202500709 28

[0121] drives, or optically readable digital data storage media, for example. Other examples may also include computers, processors, control units, (field) programmable logic arrays ((F)PLAs), (field) programmable gate arrays ((F)PGAs), graphics processor units (GPUs), application-specific integrated circuits (ASICs), integrated circuits (ICs) or system-on-a-chip (SoCs) systems programmed to execute the steps of the methods described above.

[0122] It is further understood that the disclosure of several steps, processes, operations or functions disclosed in the description or claims shall not be construed to imply that these operations are necessarily dependent on the order described, unless explicitly stated in the individual case or necessary for technical reasons. Therefore, the previous description does not limit the execution of several steps or functions to a certain order. Furthermore, in further examples, a single step, function, process or operation may include and / or be broken up into several sub-steps, -functions, -processes or -operations.

[0123] If some aspects have been described in relation to a device or system, these aspects should also be understood as a description of the corresponding method. For example, a block, device or functional aspect of the device or system may correspond to a feature, such as a method step, of the corresponding method. Accordingly, aspects described in relation to a method shall also be understood as a description of a corresponding block, a corresponding element, a property or a functional feature of a corresponding device or a corresponding system.

[0124] The following claims are hereby incorporated in the detailed description, wherein each claim may stand on its own as a separate example. It should also be noted that although in the claims a dependent claim refers to a particular combination with one or more other claims, other examples may also include a combination of the dependent claim with the subject matter of any other dependent or independent claim. Such combinations are hereby explicitly proposed, unless it is stated in the individual case that a particular combination is not intended. Furthermore, features of a claim should also be included for any other independent claim, even if that claim is not directly defined as dependent on that other independent claim.202500709 29

[0125] Reference numerals

[0126] 10 method for network node

[0127] 12 transmitting information relating to datasets

[0128] 14 in response to request, transmitting specific dataset 20 method for UE

[0129] 21 receiving information relating to datasets

[0130] 22 performing inference operation

[0131] 23 determining inference result

[0132] 24 identifying at least one partition

[0133] 25 initiating retrieval

[0134] 30 apparatus for network node

[0135] 32 one or more interfaces

[0136] 34 one or more processing devices

[0137] 40 apparatus for UE

[0138] 42 one or more interfaces

[0139] 44 one or more processing devices

[0140] 300 network node

[0141] 400 UE

[0142] 500 mobile communication system

[0143] 602 providing mapping

[0144] 604 evaluation

[0145] 606 continue model inference

[0146] 608 Analyze dataset and request download

Claims

202500709 30ClaimsWhat is claimed is:

1. A method (10) for a network node (300) of a mobile communication system (500), the method (10) comprising:transmitting (12), to a user equipment (400), UE; information relating to a plurality of dataset partitions, each dataset partition being associated with a respective partition identifier, the information enabling the UE (400) to identify and selectively retrieve a dataset partition associated with an inference accuracy issue of a machine-learning model; andin response to receiving a request from the user equipment indicating a specific partition identifier,transmitting (14) the dataset partition associated with the specific partition identifier to the UE (400).

2. The method (10) of claim 1, further comprising:transmitting, to the UE (400), an accuracy threshold for use by the user equipment in determining the inference accuracy issue.

3. The method (10) of one of the claims 1 or 2, wherein the information relating to the plurality of dataset partitions comprises a mapping between each partition identifier and a corresponding downloadable link.

4. The method (10) of one of the claims 1 to 3, further comprising transmitting the information relating to the plurality of dataset partitions via at least one of a system information message broadcast in a cell served by the network node (300) or a UE-specific message.

5. The method (10) of one of the claims 1 to 4, further comprising partitioning the plurality of dataset partitions based on at least one of a time parameter or a location parameter.202500709 316. A method (20) for user equipment (400), UE, in a mobile communication system (500), the method (20)comprising:receiving (21), from a network node (300), information relating to a plurality of dataset partitions, each dataset partition being associated with a respective partition identifier;performing (22) an inference operation using a machine-learning model, which is based on one or more of the plurality of dataset partitions;determining (23) that an inference result of the inference operation does not meet an accuracy criterion;identifying (24), from among the plurality of dataset partitions, at least one dataset partition associated with the inference result not meeting the accuracy criterion; andinitiating (25) retrieval of the identified at least one dataset partition based on the respective partition identifier.

7. The method (20) of claim 6, further comprising receiving, from the network node (300), an accuracy threshold and wherein the determining (23) that the inference result does not meet the accuracy criterion comprises comparing the inference result against the accuracy threshold.

8. The method (20) of one of the claims 6 or 7, wherein the information relating to the plurality of dataset partitions comprises a mapping between each partition identifier and a corresponding downloadable link; andwherein initiating (25) retrieval comprises downloading the identified at least one dataset partition from the corresponding downloadable link.

9. The method (20) of one of the claims 6 to 8, wherein initiating retrieval comprises:202500709 32transmitting, to the network node (300), a request message including the partition identifier of the identified at least one dataset partition.

10. The method (20) of one of the claims 6 to 9, wherein receiving the information relating to the plurality of dataset partitions comprises receiving at least one of: a system information message or a UE-specific message, in particular a Radio Resource Control, RRC, reconfiguration message.

11. A computer program having a program code for performing one of the methods (10; 20) of claims 1 to 10, when the computer program is executed on a computer, a processor, or a programmable hardware component.

12. An apparatus (30) for a network node (300) of a mobile communication system (500), the apparatus (30) comprisesone or more interfaces (32) configured to communicate in the mobile communication system (500); andone or more processing devices (34) configured to perform one of the methods (10) of claims 1 to 5.

13. An apparatus (40) for user equipment (400), UE, of a mobile communication system (500), the apparatus (40) comprisesone or more interfaces (42) configured to communicate in the mobile communication system (500); andone or more processing devices (44) configured to perform one of the methods (20) of claims 6 to 10.

14. A network node (300) of a mobile communication system (500) comprising the apparatus (30) of claim 12.202500709 3315. User equipment (400), UE, of a mobile communication system (500) comprising the apparatus (40) of claim 13.

16. A mobile communication system (500) comprising the network node (300) of claim 14 and the UE (400) of claim 15.

Images