Optimization on network (NW)-side ai / ML data management
A unified logging trigger configuration and structured data availability mechanisms address inefficiencies in AI/ML data management, ensuring consistent data collection and retention, optimizing power consumption and mobility handling, thus improving AI/ML model performance.
Patent Information
- Application Number
- PCT/US2026/026472
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-09
- Filing Date
- 2026-05-04
- Publication Date
- 2026-11-12
AI Technical Summary
Current wireless communication systems lack standardized mechanisms for efficient and effective management of AI/ML data, leading to inconsistent data collection, retention, and reporting across different UE implementations, which affects the performance and accuracy of AI/ML models.
Implement a unified logging trigger configuration structure, structured data availability indication, deterministic mobility handling, enhanced metadata handling, and optimized data transport mechanisms to ensure consistent and efficient AI/ML data management.
Ensures consistent data collection behavior across devices, optimizes UE power consumption, supports network-adaptive data retrieval, and maintains data integrity during mobility events, thereby enhancing the performance and accuracy of AI/ML models.
Smart Images

Figure US2026026472_12112026_PF_FP_ABST
Abstract
Description
OPTIMIZATION ON NETWORK (NW)-SIDE AI / ML DATA MANAGEMENTCROSS REFERENCE TO RELATED APPLICATION
[0001] This application claims priority to U.S. Provisional Patent Application No.63 / 802,615, titled “RAN2 DISCUSSION ON UE SIDE DATA COLLECTION AND RAN2 DISCUSSION ON AIML NW SIDE DATA COLLECTION FOR RAN2 130”, filed with the U.S. Patent and Trademark Office on May 9, 2025, which is hereby incorporated by reference in its entirety.TECHNICAL FIELD
[0002] The present disclosure relates to optimization on network-side (NW-side) Artificial Intelligence and / or Machine Learning (AI / ML) data management.BACKGROUND
[0003] The information disclosed in this background section is only for the enhancement of understanding of the general background of the disclosure and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.
[0004] In modem wireless communication systems or networks (e.g., 5G New Radio (NR)-based networks, Open Radio Access Network (O-RAN)-based networks, etc.), Artificial Intelligence (Al) and / or Machine Learning (ML) are increasingly utilized to optimize network performance and operations. A specific configuration may involve the deployment (e.g., training, updating, executing, etc.) of AI / ML models for optimizing network operations. For instance, theAI / ML models may be deployed within one or more network components, such as a base station (e.g., 5G gNodeB, etc.), a radio access network (RAN) node, a core network entity, or the like, to provide network-related operations or services (e.g., network optimization, network operation automation, etc.). For descriptive purposes, such a configuration may be referred to as the "NW-side AI / ML” herein.
[0005] The deployment and continuous operation of the NW-side AI / ML may involve the collection and processing of massive amounts of data characterized by a wide range of varieties and temporal dynamics. Effective and efficient management of this high-volume, heterogeneous data (“AI / ML data” herein) is important for enabling accurate and reliable AI / ML-driven operations, as the performance of the AI / ML models relies on the comprehensiveness, completeness, and timeliness of the AI / ML data utilized for training and inference.SUMMARY
[0006] Example embodiments of the present disclosure provide systems, methods, and the like, that effectively and efficiently optimize NW-side AI / ML data management.
[0007] According to example embodiments, a system may include a network entity that may be configured to transmit, to a user equipment (UE), a message comprising a logging trigger configuration. The logging trigger configuration may be associated with a logging session for collecting data associated with an AI / ML model, and may include at least one of: a trigger type field indicating a type of condition that triggers the logging session, an event identifier (ID) field indicating an event that triggers the logging session, a time to trigger field indicating a time duration for triggering the logging session, and a logging interval field indicating an interval forperforming the logging session. Further, the network entity may be configured to receive, from the UE, a message comprising data collected by the UE according to the logging trigger configuration.
[0008] According to example embodiments, a method may include transmitting, by a network entity and to a UE, a message comprising a logging trigger configuration. The logging trigger configuration may be associated with a logging session for collecting data associated with an AUML model, and may include at least one of: a trigger type field indicating a type of condition that triggers the logging session, an event ID field indicating an event that triggers the logging session, a time to trigger field indicating a time duration for triggering the logging session, and a logging interval field indicating an interval for performing the logging session. The method may further include receiving, by the network entity and from the UE, a message comprising data collected by the UE according to the logging trigger configuration.
[0009] According to example embodiments, a non-transitory computer-readable recording medium may have recorded thereon instructions executable by a network entity to cause the network entity to perform a method. The method may include transmitting, to a UE, a message comprising a logging trigger configuration. The logging trigger configuration may be associated with a logging session for collecting data associated with an AI / ML model, and may include at least one of a trigger type field indicating a type of condition that triggers the logging session, an event ID field indicating an event that triggers the logging session, a time to trigger field indicating a time duration for triggering the logging session, and a logging interval field indicating an interval for performing the logging session. The method may further include receiving, from the UE, a message comprising data collected by the UE according to the logging trigger configuration.
[0010] Additional aspects will be set forth in part in the description that follows and, in part, will be apparent from the description, or may be realized by practice of the presented embodiments of the disclosure.BRIEF DESCRIPTION OF THE DRAWINGS
[0011] Features, aspects, and advantages of embodiments of the disclosure will be described below with reference to the accompanying drawings, in which like reference numerals denote like elements, and wherein:
[0012] FIG. 1 illustrates an example system configuration, according to one or more example embodiments;
[0013] FIG. 2 to FIG. 10 each illustrates an example method, according to one or more example embodiments;
[0014] FIG. 11 illustrates an example device / apparatus that may implement one or more example embodiments; and
[0015] FIG. 12 illustrates an example environment in which systems, devices, and / or methods, according to one or more example embodiments, may be implemented.DETAILED DESCRIPTION
[0016] The following detailed description of example embodiments refers to the accompanying drawings. The foregoing disclosure provides illustration and description, but is not intended to be exhaustive or to limit the implementations to the precise forms disclosed. Modifications and variations are possible in light of the above disclosure or may be acquired from practice of the implementations. Further, one or more features or components of one embodimentmay be incorporated into or combined with another embodiment (or one or more features of another embodiment). Additionally, the flowchart and description of operations provided below relate to one of the various embodiments. It should be noted that it is possible to make other embodiments that do not exactly match the flowchart and its description. It is understood that in other embodiments one or more operations may be omitted, one or more operations may be added, one or more operations may be performed simultaneously (at least in part).
[0017] It will be apparent that systems and / or methods, described herein, may be implemented in different forms of hardware, firmware, or a combination of hardware and software. The actual specialized control hardware or software code used to implement these systems and / or methods is not limited to the described implementations. Thus, the operation and behavior of the systems and / or methods are described herein without reference to specific software code. It is understood that software and hardware may be designed to implement the systems and / or methods based on the description herein.
[0018] Even though particular combinations of features are disclosed in the claims and / or in the specification, these combinations are not intended to limit the disclosure of implementations. In fact, many of these features may be combined in ways not specifically recited in the claims and / or disclosed in the specification. Although each dependent claim listed below may directly depend on only one claim, the disclosure of implementations includes each dependent claim in combination with every other claim in the claim set.
[0019] No element, act, or instruction used herein should be construed as critical or essential unless explicitly described as such. Also, as used herein, the articles “a” and “an” are intended to include one or more items, and may be used interchangeably with “one or more.” Also, as used herein, the terms “has,” “have,” “having,” “include,” “including,” or the like are intendedto be open-ended terms. Further, the phrase “based on” is intended to mean “based, at least in part, on” unless explicitly stated otherwise. Furthermore, expressions such as “at least one of [A] and [B]”, “[A] and / or [B]”, or “at least one of [A] or [B]”, are to be understood as including only A, only B, or both A and B.
[0020] Expressions such as “at least one processor,” where configured to implement a plurality of operations, execute a plurality of instructions, etc., are to be understood as a single processor implementing the plurality of operations, etc., or each of plural processors implementing at least some (but not necessarily all) of the plurality of operations, etc.
[0021] Reference throughout this specification to “one embodiment,” “embodiment,” “non-limiting exemplary embodiment,” “example embodiment,” or similar language means that a particular feature, structure, or characteristic described in connection with the indicated embodiment is included in at least one embodiment of the present solution. Thus, the phrases “in one embodiment”, “in an embodiment,” “in one non-limiting exemplary embodiment,” and similar language throughout this specification may, but do not necessarily, all refer to the same embodiment.
[0022] Further, the described features, advantages, and characteristics of the present disclosure may be combined in any suitable manner in one or more example embodiments. One skilled in the relevant art will recognize, in light of the description herein, that the present disclosure can be practiced without one or more of the specific features or advantages of a particular embodiment. In other instances, additional features and advantages may be recognized in certain embodiments that may not be present in all embodiments of the present disclosure.
[0023] It shall be noted that, descriptions of example embodiments of the present disclosure may include terms and names defined in one or more standard organizations, such asthe Open Radio Access Network (O-RAN) Alliance, the 3rd Generation Partnership Project (3GPP) standard organization, the European Telecommunications Standards Institute (ETSI) standard organization, and the like. For instance, the terms “RRC message,” “MAC CE,” “IE,” “L3,” “LI,” “MDT,” “CSI,” “CSI-RS,” “SSB,” “SINR,” “RSRP,” “TRP,” “RSTD,” “PCI,” “SFN,” “PRS,” and the like, as well as the associated features, operations, interfaces, and messages involved therein, are to be interpreted as consistent with those specified in one or more technical specifications, unless being described otherwise.
[0024] In the present disclosure, specific tasks may be performed using Artificial Intelligence and / or Machine Leaning (AI / ML) models. An AI / ML model is a model generated using one or more Al technologies, one or more ML algorithm or both, and generates output data based on input data. This output data is used to perform tasks. Tasks performed using AI / ML models include those generally referred to as intellectual tasks, such as classification, prediction, natural language processing, etc. In the following, terms like “the Al model” may refer to “one or more Al models,” “one or more ML models,” “one or more Al models and ML models,” and the like.
[0025] Although Al and ML may be explained separately, ML is a technology included in Al. In ML, instead of being explicitly programmed for a specific task, systems can improve their performance over time by identifying patterns and making inferences from training data. Typically, the generation of ML models includes data collection, model training, and model inference. Data collection involves gathering and preprocessing data to be used for training and inference. Model training involves developing and validating models using the collected data. Model inference involves applying the trained models to new data to generate new output data and perform tasks.
[0026] Machine learning includes various types of learning methods such as supervised learning, unsupervised learning, reinforcement learning, semi-supervised learning, self-supervised learning, transductive learning, transfer learning, meta learning, and the like. These types of learning methods can be appropriately selected according to the embodiments. Unless otherwise specified, the application of types not mentioned in this description is not precluded. Additionally, the structure of ML models may vary depending on the embodiments and learning methods, and is not limited to the methods disclosed. Furthermore, ML includes deep learning, which uses models that include neural networks. Deep learning models may include, for example, deep neural networks (DNNs), convolutional neural networks (CNNs), etc.
[0027] It should be noted that the AI / ML models presented hereinafter are examples and are not limited to the illustrated AI / ML models. They can be modified or altered by using different Al or ML algorithms. The configuration of the neural network is not limited to the configuration disclosed in the present disclosure and can be modified.
[0028] In addition, reference throughout the specification to “data collection,” “data logging,” or similar terminology may refer to the process that involves one or more operations such as: monitoring, measuring, obtaining, sampling, buffering, or recording data associated with one or more AI / ML operations. Said terminologies may be used interchangeably herein, without departing from the scope of the present disclosure. Similarly, reference throughout the specification to “logs,” “AI / ML data,” “collected data,” “logged data,” “AI / ML logs,” or similar terminology may refer to data collected or logged by the UE for AI / ML purposes. Said terminologies may be used interchangeably herein, without departing from the scope of the present disclosure. Further, reference throughout the specification to “AI / ML data logging session,” “AI / ML logging session,” “data logging session,” or similar terminology may refer to a loggingsession for logging or collecting data associated with an AI / ML model. Said terminologies may be used interchangeably herein, without departing from the scope of the present disclosure.
[0029] As described above, in a network-side (NW-side) configuration, one or more AI / ML models may be deployed (e.g., deployed by one or more network entities such as a base station, a radio access network (RAN), a core network, etc.) to provision network-related operations or services (e.g., network resource optimization, network automation, etc.). In this regard, the efficacy and accuracy of these AI / ML models may rely on the quality, timeliness, and completeness of the data utilized for implementing the AI / ML models. For example, when training an AI / ML model designed to optimize beam management, if the collected training data fails to capture critical or timely radio conditions (e.g., due to insufficient logging durations or delayed reporting, etc.), the resulting AI / ML model (trained based on the collected training data) may be unable to accurately predict dynamic channel changes, leading to suboptimal beam management when the AI / ML model is executed for beam management operations.
[0030] As also described above, the deployment and continuous operation of the NW-side AI / ML configuration may involve the collection and processing of massive amounts of data (“AI / ML data” herein). This AI / ML data may be associated with one or more AI / ML models and / or may be characterized by a wide range of varieties and temporal dynamics. In this regard, a major portion of these high-volume, heterogeneous data may be collected or aggregated by one or more user equipment (UEs), such as customer devices (e.g., smartphones, tablets, wearables, etc.), specialized terminals (e.g., vehicles, industrial internet of things (loT) sensors, etc.), and the like. Further, the AI / ML data may include, for example, radio / network-related data (e.g., Channel State Information (CSI), Signal-to-Interference-plus-Noise Ratio (SINR), beam measurements, etc.),UE-related data (e.g., power consumption, supported features, operational states, etc ), user-related data (e.g., application usage statistics, traffic demand patterns, mobility routines, etc.), and the like.
[0031] In view of the above, the communication among the network component(s) and the UE is an important aspect in managing AI / ML data for NW-side AI / ML deployment. For instance, the signaling between the network component(s) and the UE may affect how efficiently and effectively the network component(s) may inform the UE regarding the specific requirements for collecting AI / ML data for NW-side AI / ML deployment. Similarly, the quality of this communication may determine the ability of the UE to accurately capture and efficiently report the targeted AI / ML data to the network component(s), ensuring that the collected datasets are sufficiently comprehensive for NW-side AI / ML deployment while minimizing the transmission of redundant information and reducing the signaling overhead on the radio interface.
[0032] Nevertheless, data management mechanisms in the related art, such as Minimization of Drive Tests (MDT) or standard periodic measurement reporting, have been used for general network optimization (coverage analysis, failure detection) and were not originally designed to meet the rigorous and dynamic requirements of AI / ML-related operations, such as AI / ML model training, AI / ML model inference, and the like. Rather, there are no standardized or specified mechanisms available in the related art that enable efficient and effective NW-side AI / ML data management. In the following, the shortcomings of the related art are described for the following aspects:(1) Logging Configuration and Triggering(2) Data- Availability and Constraint Reporting(3) Retention and State / Mobility Handling(4) Data Format and Metadata(5) Transport Bearer and SRB Use(6) Cross-Layer RS Referencing(7) Memory / Buffer Limits(8) Combined Trigger-Mask Field
[0033] Regarding aspect (1), while the related art suggests two modes for triggering data logging sessions: (a) periodic-based logging and (b) event-based logging (e.g., data logging session triggered by L3 radio-condition events derived from A1 / A2 or other L3 events), the configuration parameters associated with these trigger modes remain distributed across several Information Elements (IES) in the related art. Specifically, there is no single, unified container or IE that allows the network to configure a trigger condition (e.g., an A1 / A2 -based trigger condition, etc.) together with an independent logging interval. This fragmentation may create implementation ambiguity, as different UE vendors may interpret and implement the trigger configurations differently, leading to inconsistent data collection behavior across devices or UE implementations.
[0034] Furthermore, the logging interval parameter, which determines how frequently the UE performs data logging after a trigger condition is satisfied, remains unspecified in the related art (e.g., marked as "for further study" (FFS) items in multiple discussions within 3GPP, etc.). The logging interval is important for optimizing UE power consumption, especially in scenarios involving short-period reference signals (RS). Without a specified logging interval, the UE may default to logging at every RS transmission, which can unnecessarily drain UE battery when the network only requires data at longer intervals.
[0035] In addition, in the related art, each active logging session may be identified by a logging configuration identifier (e.g., logConfigID). However, the related art lacks a mechanism that allows the network to configure trigger-related parameters (e.g., trigger type, event identifier,time to trigger, logging interval, etc.) in a consolidated manner under a single configuration structure associated with the logging configuration identifier. As a result, the network may not efficiently instruct the UE to perform AI / ML data logging with specific trigger conditions and logging intervals while maintaining granular control over individual logging sessions.
[0036] Regarding aspect (2), while the related art may suggest causes for data availability signaling (e.g., buffer threshold reached, buffer full, low power, etc.), the mechanisms for reporting data availability in the related art may be limited to a single parameter (e.g., a single Boolean flag, an on / off indicator, etc.). Specifically, a single parameter (e.g., Boolean flag, etc.) may not adequately represent the different causes of data availability, and the network may not be able to prioritize urgent cases (e.g., low-power conditions where the UE may soon become unavailable, etc.) over routine events (e.g., buffer threshold reached conditions that may not require immediate action, etc.). This limitation may prevent the network from efficiently scheduling data retrieval based on the urgency or nature of the data availability indication.
[0037] Furthermore, the related art may capture diverging network operator requirements regarding how data availability triggers should be evaluated. For instance, some network operators may prefer network-controlled thresholds that provide deterministic behavior, while other network operators may prefer UE-autonomous heuristics that allow the UE to decide when to report data availability based on its own internal conditions. Nevertheless, the related art may lack a mechanism that allows the network to toggle between network-controlled and UE-autonomous availability-trigger logic, forcing vendors to choose one approach or the other during implementation. Without a configurable availability trigger mode, the network may not be able to adapt the data availability reporting behavior to different deployment scenarios or network operator preferences.
[0038] In addition, if a UE silently stops logging due to resource constraints (e.g., low memory, critical battery, high processor utilization, etc.), the network may observe unexplained data gaps in the collected AI / ML data, which may bias AI / ML model training or degrade model performance. Nevertheless, the related art may provide only an on / off availability flag that may not convey the reason why the data logging session paused or stopped. Without an explicit indication of the resource constraint cause, the network may not be able to distinguish between intentional pauses in data logging and unintentional interruptions caused by UE resource limitations, potentially leading to the ingestion of incomplete or biased datasets for AVML operations.
[0039] Regarding aspect (3), NW-side AI / ML data collection may run for extended periods (e g., hours or days), and the collected data or logs may span multiple network cell changes, Discontinuous Reception (DRX) cycles, and / or Radio Access Technology (RAT) transitions. Without a deterministic rule set for handling the collected data during mobility events (e.g., handover, etc.), different UE implementations may exhibit inconsistent behaviors. For instance, some UEs may discard logs whenever a handover command arrives, which may guarantee fresh context for the target cell but may result in the loss of any unsent data. On the other hand, some UEs may retain logs by default. However, if the target network entity (e.g., target gNB) is from a different vendor, those logs may never be fetched because the target network entity may lack the decoding knowledge to interpret the retained data. Furthermore, some UEs (e.g., UEs with dualmode implementations, etc.) may behave indeterminately, retaining data in some cases (e.g., intrafrequency handover) and flushing data in others (e.g., transition to RRC IDLE state, RRC_INACTIVE state, radio link failure (RLF), etc.), which may result in missing data in the time series collected for AI / ML operations.
[0040] Furthermore, while the related art may suggest a directive (e.g., a one-bit indicator set by the source network entity) to instruct the UE on whether to retain or discard logged data during mobility events, critical gaps may remain. Specifically, the signaling location of the directive may affect interoperability. For instance, placing the directive inside the handover command (e.g., RRCReconfiguration message from the target cell) may be problematic if the source and target network entities are from different vendors, as the proprietary container may be ignored by the target network entity. In this regard, pre-handover delivery of the directive from the source network entity may enhance interoperability, but such a mechanism is not specified or standardized in the related art. In addition, if retained logs are not fetched for extended intervals (e.g., the UE soon enters RRC IDLE state or executes inter-RAT handover), the retained logs may need to self-expire to satisfy privacy policies and free memory resources. Nevertheless, the related art may lack a specified lifetime management mechanism (e.g., a retention timer similar to those used in MDT) that defines when retained logs should be discarded. Without a deterministic expiry mechanism, vendors may differ on when logs are purged, leading to inconsistent data retention behavior across different UE implementations. A uniform rule set may guarantee that UEs do not hoard stale data indefinitely, thereby limiting privacy exposure and memory pressure. In view of the above, without a deterministic location and expiry timer, different vendors may implement different mechanisms on when collected data or logs are purged.
[0041] Furthermore, in scenarios where the network entity configures multiple concurrent AI / ML data logging sessions (e.g., beam-AI logging and CSI-AI logging, etc ), the network entity may not be able to separate the datasets unless each log entry declares its origin or associated logging session. Nevertheless, the log entry structures in the related art (e.g., logMeasEntry, aiLogMeasEntry, etc.) may lack information that enables the network entity to distinguish betweendata collected from different concurrent logging sessions. Without such information, the network entity may not be able to properly separate and route the collected data to the appropriate AVML model pipelines.
[0042] Regarding aspect (4), the network entity may aggregate logs or collected data from a large number of UEs (e.g., millions of UEs), and the data may arrive out of order due to various factors such as Hybrid Automatic Repeat Request (HARQ) retransmissions, DRX sleep cycles, multi-SIM split uplink transmissions, or other transmission-related events. In the related art, data logging mechanisms (e.g., MDT signaling, etc.) may rely on context information contained in the enclosing Radio Resource Control (RRC) message or Protocol Data Unit (PDU) to associate the logged data with its corresponding logging session or configuration. Nevertheless, if segmentation of the logged data occurs (e.g., due to large data volumes, transmission constraints, etc.), the context information may be lost, and the network entity may not be able to properly reassemble the interleaved segments of the logged data.
[0043] Furthermore, with the message-level context alone, the network entity may not have sufficient information to identify the logging session, determine the sequence of segments, identify whether a segment is the final segment of a transmission, and / or determine the priority of the logged data. Specifically, message-level context alone may not be sufficient to reassemble interleaved segments when data from multiple UEs or multiple logging sessions arrives out of order or in an interleaved manner.
[0044] In addition, the log entry structures in the related art may lack per-entry metadata that enables the network entity to associate each log entry with its corresponding network cell, identify gaps or discontinuities in the logged data, or determine the priority of individual entries within a segment. Without such per-entry metadata (e.g., cell identifier, gap indicator, entrypriority, etc.), the network entity may not be able to properly process, filter, or prioritize the logged data for NW-side AI / ML operations, potentially leading to incomplete or improperly ordered datasets that may degrade AI / ML model training or inference performance.
[0045] Regarding aspect (5), in the related art, AI / ML logs (or collected AI / ML data) and other data logging mechanisms (e.g., MDT signaling, etc.) may share the same signaling radio bearer (e.g., SRB-4) for transmitting logged / collected AI / ML data from the UE to the network entity. Nevertheless, when multiple types of logged data share the same signaling radio bearer with a single logical channel group (LCG), head-of-line blocking may occur, where one type of data (e.g., AI / ML data) may delay the transmission of another type of data (e.g., MDT data). This may result in inefficient utilization of the radio resources and may degrade the timeliness or completeness of the logged data received by the network entity for NW-side AI / ML purposes.
[0046] Furthermore, the related art may lack a mechanism that enables the network entity to distinguish between AI / ML data traffic and other types of logged data (e.g., MDT data) at the Medium Access Control (MAC) layer. Without such a mechanism, the MAC scheduler may not be able to allocate appropriate resources (e.g., fewer bytes per Transmission Time Interval (TTI), etc.) to AI / ML traffic, throttle bursts during congestion, or apply fair-queue or weighted deficit round-robin scheduling between AI / ML data and other types of logged data. As a result, the network entity may not be able to efficiently manage the transmission of AI / ML logs while maintaining the quality of service for other types of logged data.
[0047] In addition, the related art may lack a mechanism that enables the network entity to gain explicit knowledge of burst boundaries associated with AI / ML data transmissions. Specifically, when the UE transmits AI / ML data in multiple segments, the network entity may not be able to determine whether additional segments are pending transmission by the UE. Withoutsuch knowledge, the scheduler of the network entity may not be able to appropriately allocate resources for the ongoing burst transmission, potentially leading to inefficient resource utilization or premature termination of the AI / ML data transmission.
[0048] Regarding aspect (6), in the context of NW-side AI / ML data logging, the network entity may configure the UE to collect data (e.g., log measurements) associated with specific reference signal (RS) resources. In this regard, some network operators may prefer that the logging follows the same RS pool used for inference, in order to ensure feature alignment between the logged data and the data used for AI / ML model training. Nevertheless, other network operators may prefer custom RS sets (e.g., occasional Synchronization Signal / Physical Broadcast Channel blocks, etc.) that are not part of Channel State Information (CSI) inference, particularly for use cases such as mobility-prediction AI / ML operations.
[0049] Furthermore, embedding RS details (e.g., RS resource identifiers, RS resource set configurations, etc.) in the RRC message every time the network entity configures a logging session may duplicate the ASN.l definitions already specified in the related art, and may risk version skew between the logging configuration and the underlying RS configuration. Nevertheless, forcing all logging configurations to tie to a specific identifier may not work with custom RS sets that are not associated with any existing RS configuration.
[0050] In addition, the related art may lack a mechanism that allows the network entity to reference or inherit RS sets from an existing configuration (e.g., CSI-ReportConfig) for AI / ML data logging purposes. Without such a mechanism, the network entity may need to either duplicate the RS configuration in every logging configuration message (which may increase signaling overhead and risk version skew), or restrict the logging to only those RS resources associated with existing configurations (which may limit flexibility for custom RS sets). Accordingly, the networkentity may not be able to efficiently configure the UE to log measurements on the desired RS resources while maintaining flexibility for both inference-aligned logging and custom RS set logging.
[0051] Regarding aspect (7), network operators may configure a fixed buffer threshold (e.g., 50 kB) for AI / ML data logging. Nevertheless, a fixed buffer threshold may result in potential issues depending on the buffer capacity of different UEs. Specifically, for UEs with small buffer capacities, the configured buffer threshold may exceed the total buffer size of the UE, causing the UE to hit a buffer full condition constantly, ultimately resulting in signaling storms (e.g., frequent reporting of buffer status) and data truncation (e.g., loss of logged data due to buffer overflow). Conversely, for UEs with large buffer capacities, the configured buffer threshold may not be reached, causing the network to delay fetching the logged AEML data. In this scenario, old logs or collected AEML data may occupy memory and may be overwritten when new high-value events occur, potentially resulting in the loss of valuable AEML data.
[0052] Furthermore, the related art may lack a minimum baseline requirement for dedicated memory for AI / ML data. Without such a baseline (e.g., a minimum of 32 kB dedicated Access Stratum (AS)-layer memory for AI / ML data), the network may not be able to guarantee that any configured buffer threshold below the baseline is supportable by all UEs. This may lead to inconsistent data logging behavior across different UE implementations.
[0053] In addition, the related art may lack a mechanism that enables the UE to declare its actual buffer size capacity to the network. Without such a mechanism, the network entity (e g., Operations, Administration and Maintenance (0AM) systems, self-optimizing Self-Organizing Network (SON) functions, etc.) may not be able to dynamically adjust the buffer threshold based on the UE's actual buffer capacity. As a result, the network may configure buffer thresholds thatare either too high (causing the UE to never report buffer status) or too low (causing the UE to report buffer status too frequently), leading to suboptimal data logging and reporting behavior.
[0054] Furthermore, the related art may lack a mechanism that constrains the buffer threshold to a reasonable range relative to the UE's declared buffer capacity. Without such a constraint (e.g., constraining the buffer threshold to 50-90% of the UE's declared buffer capacity), suboptimal settings may occur, where low-capacity UEs may experience frequent buffer overflows and high-capacity UEs may experience delayed data retrieval. This may result in non-uniform data logging behavior across different UEs and may complicate cross-vendor configuration templates for AI / ML data logging.
[0055] Regarding aspect (8), some network deployments may require the UE to perform data logging based on multiple trigger conditions simultaneously. For instance, a network entity may configure the UE to perform data logging based on both periodic intervals and event-based triggers (e.g., X1 / X2 events derived from A1 / A2 radio conditions) at the same time, in order to capture comprehensive AI / ML data that reflects both time-based patterns and event-driven radio conditions. Conversely, other network deployments may require one trigger condition to suppress another trigger condition. For instance, a network entity may configure the UE to perform data logging based on periodic intervals only when no event-based trigger has occurred, or vice versa, in order to avoid redundant data logging or to prioritize certain types of data over others.
[0056] Nevertheless, the related art may lack a unified mechanism that enables the network entity to enable or disable multiple trigger conditions per logging session. Specifically, the related art may not provide a single configuration field or parameter that allows the network entity to specify which combination of trigger conditions (e.g., periodic, XI event, X2 event, low power condition, etc.) should be active for a given logging session. Without such a unified trigger mask,the network entity may need to configure each trigger condition separately, which may increase signaling overhead and may lead to ambiguity regarding how the UE should behave when multiple trigger conditions are configured.
[0057] Furthermore, without a unified trigger mask, the UE may not be able to determine whether multiple trigger conditions should be evaluated independently (e.g., log when any enabled condition is satisfied) or in combination (e.g., log only when all enabled conditions are satisfied). This ambiguity may lead to inconsistent data logging behavior across different UE implementations, potentially resulting in incomplete or redundant datasets for NW-side AI / ML operations.
[0058] Example embodiments of the present disclosure provide systems, methods, mechanisms, and the like, that efficiently and effectively address one or more of the problems described above with respect to aspects (1) to (8), thereby optimizing the management of data associated with NW-side AI / ML.
[0059] Specifically, with respect to aspect (1), example embodiments of the present disclosure introduce and exemplify a mechanism that optimizes logging configuration and triggering for NW-side AI / ML data management. For instance, example embodiments introduce a unified logging trigger configuration structure (e.g., "LoggingTriggerConfig", etc.) that consolidates trigger-related parameters under a single configuration container, thereby addressing the fragmentation of configuration parameters across multiple Information Elements (IES) in the related art.
[0060] According to example embodiments, the logging trigger configuration may include at least one of: a trigger type field indicating a type of condition that triggers the logging session (e.g., Layer 3 (L3) radio-condition events, etc.), an event identifier (ID) field indicating oridentifying an event that triggers the logging session (e.g., A1 / A2 derivative events such as XI, X2, etc.), a time to trigger field indicating a time duration for triggering the logging session, or a logging interval field indicating an interval for performing the logging session. In this regard, by consolidating these trigger-related parameters under a single configuration structure, a network entity (e.g., base station, core network, network function, etc.) may efficiently instruct a UE to perform AI / ML data logging with specific trigger conditions and logging intervals, thereby resolving implementation ambiguity across different UE vendors and ensuring consistent data collection behavior across devices.
[0061] Furthermore, example embodiments introduce the specification of the logging interval parameter, which determines how frequently the UE performs data logging after a trigger condition is satisfied. According to example embodiments, the logging interval field may be optional, and when the logging interval is absent, the UE may default to perform logging or data collection per reference signal (RS) period. In this regard, by specifying the logging interval parameter (or the default behavior when the logging interval is absent), example embodiments may optimize UE power consumption, especially in scenarios involving short-period reference signals, by allowing the network to configure longer logging intervals when data at every RS transmission is not required.
[0062] In addition, example embodiments introduce the association of the logging trigger configuration with a logging configuration identifier (e.g., logConfigID, etc.) that identifies each active logging session. In this regard, by associating the trigger-related parameters (e g., trigger type, event identifier, time to trigger, logging interval, etc.) with the logging configuration identifier under a single configuration structure, the network entity may maintain granular controlover individual logging sessions while efficiently instructing the UE to perform AI / ML data logging with specific trigger conditions and logging intervals.
[0063] According to example embodiments, the time to trigger field may indicate a time duration that has a same range as a time duration to trigger (TTT) associated with a radio resource management (RRM) measurement. In this regard, by aligning the time to trigger parameter with existing RRM parameters, example embodiments may ensure consistency with existing measurement triggering mechanisms and simplify UE implementation.
[0064] According to example embodiments, the logging trigger configuration may be included in a message (e.g., a radio resource control (RRC) message, etc.) transmitted by the network entity to the UE. Specifically, the logging trigger configuration may be included under an information element (IE) of the RRC message (e.g., "AIML-LoggingConfig", etc.), thereby providing a standardized and unified container for configuring AEML data logging triggers.
[0065] Advantageously, by implementing the mechanism described herein, the network entity may effectively and efficiently configure the UE for AI / ML data logging by providing a unified logging trigger configuration that consolidates trigger-related parameters under a single configuration structure. Accordingly, example embodiments may resolve implementation ambiguity across different UE vendors, optimize UE power consumption through configurable logging intervals, and maintain granular control over individual logging sessions, thereby addressing the shortcomings of aspect (1) as described above.
[0066] With respect to aspect (2), example embodiments of the present disclosure introduce and exemplify a mechanism that optimizes data availability and constraint reporting for NW-side AI / ML data management. Specifically, example embodiments introduce a structured data availability indication (e.g., "DataAvailabilitylndication", etc.) that enables the UE to reportthe cause of data availability and additional information associated with the data availability status, thereby addressing the limitation of single-parameter (e.g., Boolean flag, etc.) availability reporting in the related art.
[0067] According to example embodiments, the network entity may transmit, to a UE, a message that includes configuration associated with an AI / ML data logging session. The configuration may include a field (e.g., "availabilityTriggerMode", etc.) indicating a mode (e.g., network-controlled, UE-autonomous, etc.) for triggering transmission of data availability indication. This field may indicate which entity (e.g., network entity, UE, etc.) evaluates buffer and power thresholds. According to example embodiments, the field may indicate the network-controlled mode as the default mode. In this regard, by providing a configurable availability trigger mode, example embodiments may address the diverging network operator requirements, where some network operators may prefer network-controlled thresholds that provide deterministic behavior, while other network operators may prefer UE-autonomous heuristics that allow the UE to decide when to report data availability based on its own internal conditions.
[0068] According to example embodiments, the network entity may receive, from the UE, a message that includes a data availability indication. The message may include UE Assistance Information (UAI) that includes the data availability indication. The UAI may include a field (e.g., "resourceConstraintCause", etc.) indicating a resource constraint cause (e.g., low memory condition, low power condition, busy processor / CPU condition, etc.). In this regard, by providing an explicit indication of the resource constraint cause, the network entity may be able to distinguish between intentional pauses in data logging and unintentional interruptions caused by UE resource limitations, thereby avoiding the ingestion of incomplete or biased datasets for AI / ML operations.
[0069] Furthermore, according to example embodiments, the data availability indication may be implemented as a structure element in the message (e.g., UAI, etc.). The data availability indication may include a field (e.g., "cause", etc.) associated with at least one cause indicating why the UE is sending the data availability indication (e.g., buffer has reached a configured threshold, buffer is full, UE is in low power condition, etc.), and a field (e.g., "remainingBytes", etc.) indicating how many bytes of data remain in the UE's buffer. In this regard, by providing a structured data availability indication with multiple fields, the network entity may be able to prioritize urgent cases (e.g., low-power conditions where the UE may soon become unavailable) over routine events (e.g., buffer threshold reached conditions that may not require immediate action), thereby enabling efficient scheduling of data retrieval based on the urgency or nature of the data availability indication.
[0070] Advantageously, by implementing the mechanism described herein, the network entity may effectively and efficiently receive structured data availability indications that enable prioritization of data retrieval based on the cause of availability, configure the availability trigger mode to adapt to different deployment scenarios or network operator preferences, and receive explicit resource constraint cause indications that enable identification of the reason for data logging pauses or interruptions. Accordingly, example embodiments may enable efficient scheduling of data retrieval, support both network-controlled and UE-autonomous availabilitytrigger logic, and avoid the ingestion of incomplete or biased datasets for AI / ML operations, thereby addressing the shortcomings of aspect (2) as described above.
[0071] With respect to aspect (3), example embodiments of the present disclosure introduce and exemplify a mechanism that optimizes retention and mobility handling for NW-side AI / ML data management. Specifically, example embodiments introduce a deterministic rule setfor managing collected data during mobility events (e.g., handover, etc.) and a logging configuration identifier for separating datasets from concurrent logging sessions, thereby addressing the inconsistent data retention behaviors and dataset separation issues in the related art.
[0072] According to example embodiments, in response to determining that handover of the UE to a target cell is required, the network entity may provide, to the UE, a message (e.g., RRC Reconfiguration message, etc.) that includes information associated with data retention. The message may include a parameter (e.g., "retainLoggedData", etc.) that indicates whether the UE may retain logged data after handover (e.g., TRUE to retain, FALSE to discard, etc.). The message may also include a retention timer parameter (e.g., "retentionTimer", etc.) that indicates the time duration (e.g., between 24 to 168 hours, with a default of 48 hours, etc.) for which the UE may retain the logged data before discarding. In this regard, by providing an explicit directive on data retention and a configurable retention timer from the source network entity before handover execution, example embodiments may ensure deterministic data retention behavior across different UE implementations and vendors, and guarantee that UEs do not hoard stale data indefinitely, thereby limiting privacy exposure and memory pressure.
[0073] Furthermore, according to example embodiments, the network entity may receive, from the UE, a message that includes AI / ML data collected by the UE along with logging information that includes an identifier (e.g., "logConfiglD", etc.) indicating different types of logging entry structures associated with the collected AI / ML data. The identifier may include a 5-bit identifier. Based on the logging information, the network entity may determine the type of collected AI / ML data (e.g., logConfiglD = 00001 for beam-AI dataset, logConfiglD = 00002 for CSI-AI dataset, etc.). In this regard, by including a logging configuration identifier with each log entry, example embodiments may enable the network entity to properly separate and route thecollected data to the appropriate AI / ML model pipelines when multiple concurrent logging sessions are configured.
[0074] Advantageously, by implementing the mechanism described herein, example embodiments may ensure deterministic data retention behavior during mobility events, manage the lifetime of retained logs, and enable proper dataset separation for concurrent logging sessions, thereby addressing the shortcomings of aspect (3) as described above.
[0075] With respect to aspect (4), example embodiments of the present disclosure introduce and exemplify a mechanism that optimizes data format and metadata handling for NW-side AI / ML data management. Specifically, example embodiments introduce a log header and logged measurement entry metadata that enable the network entity to reassemble interleaved segments of AI / ML data, thereby addressing the inability to reassemble out-of-order data segments in the related art.
[0076] According to example embodiments, the network entity may receive, from the UE, a message that includes a log header (e.g., "AI / MLLogHeader", etc.) and / or logged measurement entry metadata (e.g., "LoggedMeasurementEntry", etc.) associated with AI / ML data collected by the UE. The log header may include information associated with at least one of: an identifier (e.g., "logConfigID", etc.) that indicates different types of logging entry structures and / or logging configuration associated with the collected AI / ML data, a parameter (e.g., "segmentNo", etc.) that indicates segment sequence number of the AI / ML data, a parameter (e.g., "lastSegment", etc.) that indicates whether the collected AI / ML data is the final segment, a parameter (e.g., "logPriority", etc.) that indicates the priority (e.g., high, normal, etc.) of the collected AI / ML data, or a parameter (e.g., "moreSegmentsPending", etc.) that indicates whether additional data segments of AI / ML data are pending transmission by the UE. The logged measurement entry metadata may includeinformation associated with at least one of: an identifier (e ., "cellTD", etc.) that indicates network cell associated with the collected AI / ML data, a parameter (e.g., "gaplnd", etc.) that indicates whether there was any measurement gap associated with the collected AI / ML data, or a parameter (e g., "entry Priority", etc.) that indicates priority level of the collected AI / ML data.
[0077] According to example embodiments, the network entity may reassemble interleaved segments of the AI / ML data based on the log header and / or logged measurement entry metadata. In this regard, by providing a fixed header and / or per-entry metadata with each segment of the logged data, example embodiments may enable the network entity to identify the logging session, determine the sequence of segments, identify whether a segment is the final segment, determine the priority of the logged data, and identify gaps or discontinuities in the logged data, even when data from multiple UEs or multiple logging sessions arrives out of order or in an interleaved manner.
[0078] Advantageously, by implementing the mechanism described herein, example embodiments may enable the network entity to properly reassemble interleaved segments of AI / ML data, identify and handle gaps in the logged data, and prioritize processing of high-priority data, thereby addressing the shortcomings of aspect (4) as described above.
[0079] With respect to aspect (5), example embodiments of the present disclosure introduce and exemplify a mechanism that optimizes transport bearer and signaling radio bearer (SRB) usage for NW-side AI / ML data management. Specifically, example embodiments introduce a separate logical channel group (LCG) assignment for logged AI / ML data and a mechanism for indicating pending AI / ML data segments, thereby addressing the head-of-line blocking issue and lack of burst boundary knowledge in the related art.
[0080] According to example embodiments, the network entity may transmit, to the UE, a message that includes a data transport configuration for transporting ALML data. The data transport configuration may include information associated with at least one of: an LCG assignment (e.g., "logicalChannelGroup", etc.) for transmitting AI / ML data (e.g., "aiModelData-LCG", etc.), wherein the LCG may be separated from the LCG used for transmitting MDT data, or a transport bearer (e.g., SRB-4, etc.) for transmitting AI / ML data. In this regard, by assigning a separate LCG to AI / ML traffic while sharing the same SRB (e.g., SRB-4) with MDT traffic, example embodiments may avoid head-of-line blocking between AI / ML logs and MDT data, while maintaining the existing security context of the SRB.
[0081] According to example embodiments, the network entity may receive, from the UE, a message that includes collected AI / ML data according to the data transport configuration. The message may include a parameter (e.g., "moreSegmentsPending", etc.) indicating whether additional data segments of the AI / ML data are pending transmission by the UE. Based on the parameter, the network entity may schedule transmission resources. The scheduling of transmission resources may include at least one of: throttling bursts during congestion based on pending AI / ML data, allocating transmission resources based on burst boundaries, or applying fair-queue and / or weighted deficit round-robin scheduling. In this regard, by providing explicit knowledge of burst boundaries through the pending segments indicator, example embodiments may enable the network entity scheduler to efficiently manage the transmission of AI / ML logs without starving other types of logged data.
[0082] Advantageously, by implementing the mechanism described herein, example embodiments may avoid head-of-line blocking between AI / ML logs and MDT data, enable efficient resource allocation based on burst boundary knowledge, and support fair schedulingbetween different types of logged data, thereby addressing the shortcomings of aspect (5) as described above.
[0083] With respect to aspect (6), example embodiments of the present disclosure introduce and exemplify a mechanism that optimizes cross-layer reference signal (RS) referencing for NW-side AI / ML data management. Specifically, example embodiments introduce a mechanism that allows the network entity to reference or inherit RS definitions from an existing configuration (e.g., CSI-ReportConfig, etc.) or specify explicit RS resources for the logging session, thereby addressing the lack of RS referencing mechanism in the related art.
[0084] According to example embodiments, the network entity may transmit, to the UE, a message that includes data reporting configuration associated with an AI / ML data logging session. The data reporting configuration may include information associated with at least one of an identifier (e.g., "linkedCSI-ReportConfigID", etc.) indicating configuration associated with CSI reporting from which the UE may inherit reference signal definitions, an identifier (e.g., "rsSetlD", etc.) indicating RS resources set within the configuration associated with CSI reporting, or an explicit RS ID indicating RS resources for the logging session. The message may also include instructions to instruct or guide the UE to determine whether to inherit RS definitions from the configuration associated with CSI reporting. In this regard, by providing a mechanism to reference existing configurations or specify explicit RS resources, example embodiments may avoid duplicating RS details in every logging configuration message (which may increase signaling overhead and risk version skew), while maintaining flexibility for both inference-aligned logging and custom RS set logging.
[0085] According to example embodiments, the network entity may receive, from the UE, a message that includes collected AI / ML data based on the data reporting configuration. In thisregard, by allowing the UE to inherit RS definitions from an existing configuration when the linked identifier is present, or to use explicit RS IDs when the linked identifier is absent, example embodiments may guarantee feature alignment between the logged data and the data used for AI / ML model training (when inference-aligned logging is desired), or support custom RS sets for use cases such as mobility-prediction AI / ML operations (when custom RS logging is desired).
[0086] Advantageously, by implementing the mechanism described herein, example embodiments may enable flexible RS referencing for AI / ML data logging, avoid duplication of RS configurations and version skew, and support both inference-aligned logging and custom RS set logging, thereby addressing the shortcomings of aspect (6) as described above.
[0087] With respect to aspect (7), example embodiments of the present disclosure introduce and exemplify a mechanism that optimizes memory and buffer limit management for NW-side AI / ML data management. Specifically, example embodiments introduce a UE capability reporting mechanism for buffer size capacity and a configurable buffer threshold that may be constrained relative to the UE's declared capacity, thereby addressing the issues of over-triggering on low-capacity UEs and under-triggering on high-capacity UEs in the related art.
[0088] According to example embodiments, the network entity may receive, from the UE, a message that includes capability information associated with an AI / ML data logging session. The capability information may include at least one of a parameter (e.g., "bufferSizeCap", etc.) indicating a buffer size or capacity of the UE for logging AI / ML data, or a parameter indicating that the UE supports minimum dedicated AS-layer memory for logging AI / ML data (e g., greater than or equal to 32 kilobytes, etc.). The message may include an RRC message (e.g., "UECapability Information- Al", etc.). In this regard, by enabling the UE to declare its actual buffer size capacity to the network, example embodiments may allow the network (e.g., 0AM systems,Self-Organizing Network (SON) optimization functions, etc.) to dynamically adjust the buffer threshold based on the UE's actual buffer capacity.
[0089] According to example embodiments, the network entity may transmit, to the UE, a message that includes buffer configuration associated with the AI / ML data logging session. The buffer configuration may include a parameter (e.g., "buffer Threshold", etc.) indicating a threshold in absolute bytes for the AI / ML data in the UE's buffer. The buffer threshold parameter may be constrained to a percentage range (e.g., 50 to 90 percent, etc.) of the UE's declared buffer size capacity parameter. In this regard, by constraining the buffer threshold to a reasonable range relative to the UE's declared buffer capacity, example embodiments may prevent suboptimal settings where low-capacity UEs may experience frequent buffer overflows and high-capacity UEs may experience delayed data retrieval, thereby maintaining uniform data logging behavior across different UEs and simplifying cross-vendor configuration templates.
[0090] Advantageously, by implementing the mechanism described herein, example embodiments may enable dynamic buffer threshold adjustment based on UE-declared buffer capacity, prevent over-triggering on low-capacity UEs and under-triggering on high-capacity UEs, and ensure consistent data logging behavior across different UE implementations, thereby addressing the shortcomings of aspect (7) as described above.
[0091] With respect to aspect (8), example embodiments of the present disclosure introduce and exemplify a mechanism that optimizes trigger configuration for NW-side AI / ML data management. Specifically, example embodiments introduce a unified trigger mask that enables the network entity to enable or disable multiple trigger conditions per logging session, thereby addressing the lack of a unified mechanism for configuring multiple trigger conditions in the related art.
[0092] According to example embodiments, the network entity may transmit, to the UE, a message that includes a trigger mask configuration associated with an AI / ML data logging session. The trigger mask configuration may include a bit string (e.g., "triggerMask BIT STRING SIZE X", etc.), wherein each bit indicates a respective condition for triggering the AI / ML data logging session. For example, the trigger mask configuration may include a 4-bit string with bit assignments such as: bit 0 corresponding to periodic logging, bit 1 corresponding to a first event (e.g., XI event, etc.), bit 2 corresponding to a second event (e.g., X2 event, etc.), and bit 3 corresponding to a low power condition. The message may include instructions to guide the UE to trigger the AI / ML data logging session when any of the enabled bit conditions is satisfied. In this regard, by providing a unified trigger mask, example embodiments may enable the network entity to configure multiple trigger conditions (e.g., periodic logging and event-based logging simultaneously) or configure one trigger condition to suppress another, using a single configuration field.
[0093] According to example embodiments, the network entity may receive, from the UE, a message that includes AI / ML data collected based on the trigger mask configuration. For example, if only bit 0 is enabled, the UE may be instructed to log AI / ML data only on periodic intervals. If bit 0 and bit 1 are enabled, the UE may be instructed to log AI / ML data when either the periodic interval is reached or the first event (e.g., XI event) occurs. In this regard, by allowing flexible combinations of trigger conditions through the trigger mask, example embodiments may support diverse deployment scenarios where different combinations of trigger conditions may be required.
[0094] Advantageously, by implementing the mechanism described herein, example embodiments may enable flexible configuration of multiple trigger conditions per logging session,reduce signaling overhead by consolidating trigger configurations into a single bit string field, and ensure consistent data logging behavior across different UE implementations, thereby addressing the shortcomings of aspect (8) as described above.
[0095] In view of the above, example embodiments of the present disclosure disclose or suggest various mechanisms that may effectively and efficiently addressing the shortcomings of aspects (1) to (8), thereby optimizing NW-side AI / ML data management. It is contemplated that features, advantages, and significances of example embodiments described hereinabove are merely a portion of the present disclosure, and are not intended to be exhaustive or to limit the scope of the present disclosure. Further descriptions of the features, components, configuration, operations, and implementations of the example embodiments of the present disclosure are provided in the following.Example System Architecture and Configurations
[0096] FIG. 1 illustrates a generic system configuration 100, according to one or more example embodiments. As illustrated in FIG. 1, the system configuration 100 includes a network entity 110 and a user equipment UE 120. The network entity 110 may be communicatively coupled to the UE 120 via one or more networks and / or interfaces, and may be configured to exchange information or messages for network-side (NW-side) AI / ML purposes.
[0097] As described above, in a network-side (NW-side) configuration, one or more network entities may implement or deploy (e.g., train, update, execute, etc.) one or more AI / ML models to provide network-related operations or services. In this regard, “NW-side AI / ML purposes,” “AI / ML-related operations,” and similar descriptions as used herein may include (but are not limited to) one or more of: the training of AI / ML model(s) at one or more network entities (e g., using aggregated or filtered data received the UE 120, etc.), the execution or inference ofAI / ML model(s) at the network entity(s) for network automations (e.g., automated fault recovery, automated load balancing, etc.) and / or network optimizations (e.g., beam management, radio resource allocation, etc.), the provisioning and management of AI / ML-driven services, the continuous lifecycle management of AI / ML model(s) (e.g., performance monitoring, model retraining / fine-tuning, model switching to adapt to dynamic radio environments, etc.), and any other suitable type of activities or utilization of AI / ML model(s) at the network entity(s).
[0098] Further, the “NW-side AI / ML data” described herein may include data associated with the network, the UE, and / or any other suitable data, that may be utilized by the network entity 110 for NW-side AI / ML purposes. For instance, as further described below with reference to FIG.2 to FIG. 6, the NW-side AI / ML data may include (but is not limited to) one or more of: radio measurement data (e.g., signal strength, interference trends, etc.), beam-related data (e.g., beam transitions, beam-level metrics, etc.), Channel State Information (CSI)-related data (e.g., CSI measurement trends, CSI feedback, etc.), mobility-related data (e.g., handover parameters, trajectory logs, mobility patterns, etc.), UE-intemal status data (e.g., power consumption, storage availability, motion trajectory, beam selection metric, etc.), network-related operational data (e.g., network traffic / load, resource block utilization, etc.), AI / ML -related metadata (e.g., dataset classification / categorization information, model identifiers, performance feedback, etc.), and the like.
[0099] The network entity 110 may include one or more physical or logical network nodes within a network (e.g., 4G LTE network, 5G network, Open Radio Access Network (O-RAN)-based network, etc.) that may be configured to perform management, control, and / or data processing for NW-side AI / ML data. For instance, the network entity 110 may include, but is not limited to: a base station (e.g., a 4G eNodeB, a 5G gNodeB, etc.), a RAN node (e.g., NextGeneration RAN (NG-RAN), Open RAN (O-RAN), etc.) or an associated component (e.g., a Radio Intelligent Controller (RIC), a Centralized Unit (CU), etc.), a core network function or entity (e.g., an Access and Mobility Management Function (AMF), a User Plane Function (UPF), a Network Data Analytics Function (NWDAF), etc.), an AI / ML management or orchestration node, a cloud-based server, or any other suitable network node that may be configured to manage or deploy AI / ML model(s) for NW-side AI / ML purposes.
[0100] The UE 120 may include any suitable device, terminal, apparatus, or system that may communicate with the network entity 110 via a wireless connection, a wired connection, or a combination thereof. The UE 120 may also communicate with the network entity 110 via any suitable interfaces, such as those defined in one or more technical documents of one or more standard organizations (e.g., 3GPP, O-RAN Alliance, etc.). By way of example, the UE 120 may include, but is not limited to: a mobile device (e.g., a mobile phone / smartphone, a tablet, a laptop, a wearable device smartwatches and smartglasses, a portable hotspot, a wireless sensor, etc.), a static device (e.g., an Internet of Things (loT) device like sensors and smart home / office devices, an internet router, etc.), a vehicle to everything (V2X) communication terminal, a drone or unmanned aerial vehicle (UAV), or any other suitable terminal or device that may be configured to provide a request and / or report data to the network entity 110 for NW-side AI / ML purposes.
[0101] As illustrated in FIG. 1, the network entity 110 and UE 120 may communicate and interoperate with each other to facilitate the management of AI / ML data or datasets involved in or associated with NW-side AI / ML purposes. Generally, the communication between the network entity 110 and UE 120 may involve the exchange of management configuration and / or signaling in the downlink and the transmission of AI / ML data, reports, and / or requests in the uplink.
[0102] The NW-side AI / ML data management configuration and / or signaling provided by the network entity 110 to the UE 120 may include various control-plane messages and / or lower-layer signals that may include configurations or parameters that instruct or guide the UE 120 to perform one or more operations to appropriately collect and manage the AI / ML data, such as (but not limited to): a logging trigger configuration (e.g., trigger type, event identifier, time to trigger, logging interval, etc.) for configuring trigger conditions of an AI / ML data logging session, an availability trigger mode (e.g., network-controlled, UE-autonomous, etc.) for configuring how data availability triggers are evaluated, a data retention configuration (e.g., retain logged data indicator, retention timer, etc.) for configuring how the UE handles logged data during mobility events, a data transport configuration (e.g., logical channel group assignment, transport bearer, etc.) for configuring how AI / ML data is transmitted from the UE to the network entity, a reference signal configuration (e.g., linked CSI-ReportConfig identifier, RS set identifier, explicit RS identifiers, etc.) for configuring which reference signal resources the UE uses for the logging session, a buffer configuration (e.g., buffer threshold, etc.) for configuring buffer-related thresholds for the AI / ML data logging session, a trigger mask configuration (e.g., trigger mask bit string, etc.) for configuring which combination of trigger conditions are enabled for the logging session, and the like. The network entity 110 may provide the NW-side AI / ML data management configuration and / or signaling to the UE 120 via various methods, such as (but not limited to): including the desired configurations or parameters into one or more Information Elements (lEs) of a Radio Resource Control (RRC) message or a Medium Access Control (MAC) Control Element (CE).
[0103] On the other hand, the NW-side AI / ML data, report, and / or request provided by the UE 120 to the network entity 110 may include the information required for the network entity 110 to deploy (e.g., train, manage, execute, etc.) one or more AI / ML models for NW-side AI / MLpurposes. In addition to or in alternative to the above-mentioned NW-side AI / ML data, the UE 120 may provide a measurement report (e.g., MDT report, RRC measurement report, etc.) that contains the collected or logged NW-side AI / ML data (which may be processed according to one or more configurations defined by the network entity), a request (e.g., a request for additional data collection / logging, a request for dataset adaptation based on internal conditions of the UE, etc.), or the like. According to example embodiments, the NW-side AI / ML data, report, and / or request may include information associated with at least one of collected AI / ML data (e.g., logged measurements, beam measurements, CSI measurements, etc.) associated with one or more logging sessions, a data availability indication (e.g., cause of data availability, remaining bytes in buffer, etc.) indicating availability of logged data at the UE, a resource constraint cause indication (e.g., low memory condition, low power condition, busy processor condition, etc.) indicating a reason for pausing or stopping a logging session, capability information (e.g., buffer size capacity, support for minimum dedicated AS-layer memory, etc.) indicating the UE's capabilities for AI / ML data logging, a log header (e.g., logging configuration identifier, segment sequence number, last segment indicator, log priority, more segments pending indicator, etc.) associated with segments of the collected AI / ML data, logged measurement entry metadata (e.g., cell identifier, gap indicator, entry priority, etc.) associated with individual log entries, a logging configuration identifier (e.g., logConfigID, etc.) indicating the type or origin of the collected AI / ML data, and the like.
[0104] According to example embodiments, the communication between the network entity 110 and UE 120 may involve a coordinated or hybrid interaction. Specifically, while the network entity 110 may provide the foundational configuration and oversight, the UE 120 may actively participate in the management of the NW-side AI / ML data by, for example, providing real-time (or near real-time) reports and data collection / logging triggering requests based on itslocalized observations of the radio environment and / or its internal operational status. Accordingly, this bidirectional information exchange may ensure that the NW-side AI / ML data is collected and managed in an efficient, responsive, and standardized manner.
[0105] It is contemplated that the configuration in FIG. 1, as well as the described features and mechanisms, are merely examples and the scope of the present disclosure should not be limited thereto. Specifically, in some example implementations, the network entity 110 may communicate with multiple UEs, the UE 120 may communicate with multiple network entities, the network entity 110 and UE 120 may communicate with each other via an intermediate node / terminal, and the like, without departing from the scope of the present disclosure. Descriptions of an example environment that may implement the network entity 110 and UE 120 are provided below with reference to FIG. 12.
[0106] Further, it is contemplated that the network entity 110 and UE 120 may communicate and interoperate with each other to implement various methods and operations, in addition to those described above. For instance, as further described below with reference to FIG.2 to FIG. 10, the network entity 110 and UE 120 may communicate and interoperate with each other to implement various methods and operations for addressing the shortcomings of the related art in each of the above-mentioned aspects (1) to (8).
[0107] Furthermore, it is contemplated that the network entity 110 and UE 120 may further include components that may be configured to implement the associated functionalities or operations. Descriptions of various example components that may be included in (or may be configured to implement) the network entity 110 and / or UE 120 are provided below with reference to FIG. 11.Example Methods and Operations
[0108] Several example methods and operations, according to one or more example embodiments, are described below with reference to FIG. 2 to FIG. 10. One or more features, parameters, messages, and operations associated with FIG. 2 to FIG. 10 may be similar to those described above with reference to FIG. 1, and redundant descriptions associated therewith may be omitted below for conciseness.
[0109] For descriptive purposes, the example methods and operations may be mainly described herein as being performed by one or more specific network entities, although it can be understood that, in actual implementations, another related network entity(s) and / or the UE may perform similar / related operations, without departing from the scope of the present disclosure. For instance, an operation of a network entity (e.g., network entity 110) receiving a data / message from a UE (e g., UE 120) may suggest or indicate an operation of the UE providing the data / message to the network entity, an operation of the network entity providing a data / message that includes data / information that enables the UE to implement an operation / function may suggest or indicate an operation of the UE receiving the data / message and utilize the data / information included therein to implement the operation / function, and the like.
[0110] According to example embodiments, one or more operations of the network entity (e g., base station, NG-RAN, core network, etc.) may be implemented in one or more apparatuses or hardware components. For instance, the network entity (or one or more associated operations) may be implemented in an apparatus / device that includes a processor and a memory storage (or any other suitable storage mediums), wherein the memory storage may include computerexecutable instructions which, when executed by the processor, cause the processor to perform one or more operations thereof. An example device / apparatus that may implement the network entity (or one or more associated operations) is further described below with reference to FIG. 11.
[0111] The example methods and operations in FIG. 2 to FIG. 10 may be implemented by the network entity to effectively and efficiently manage NW-side AI / ML data, thereby addressing the above-mentioned shortcomings of aspects (1) to (8). Generally, the example method and operations in FIG. 2 may be associated with aspect (1), the example method and operations in FIG.3 may be associated with aspect (2), the example methods and operations in FIG. 4 and FIG. 5 may be associated with aspect (3), the example method and operations in FIG. 6 may be associated with aspect (4), the example method and operations in FIG. 7 may be associated with aspect (5), the example method and operations in FIG. 8 may be associated with aspect (6), the example method and operations in FIG. 9 may be associated with aspect (7), and the example method and operations in FIG. 10 may be associated with aspect (8).
[0112] In this regard, although some of the methods and operations may be described separately from one another, it is contemplated that, in some example implementations, the methods and operations in two or more of FIG. 2 to FIG. 10 may be implemented by the same network entity. In this regard, it can be understood that the network entity may perform multiple methods and operations in two or more of FIG. 2 to FIG. 10 in any suitable manner (e.g., simultaneously, sequentially, etc.) and may achieve the associated technical advantages in combination. By way of a non-limiting example, although the methods and operations in FIG. 2 and FIG. 3 are associated with different aspects (and thus, are described independently from each other herein), it can be understood that the network entity may be configured to implement the methods / operations in FIG. 2 and FIG. 3 in any suitable manner (e g., simultaneously, sequentially, etc.) to thereby addressing the shortcomings of both aspect (1) and aspect (2) and achieve the technical advantages of addressing the shortcomings of both aspect (1) and aspect (2).
[0113] Further, it can be understood that one or more operations in one of the FIG. 2 to FIG. 10 may be similar to or be part of one or more operations in another one of the FIG. 2 to FIG.10. By way of a non-limiting example, the operations in FIG. 2 (e.g., operations associated with aspect (1), etc.) may involve one or more operations in FIG. 10 (e.g., operations associated with aspect (8), etc.), as the trigger type and event identifier configured in FIG. 2 may correspond to the trigger conditions enabled via the trigger mask in FIG. 10. Further, one or more operations in FIG. 3 (e.g., operations associated with aspect (2), etc.) may involve or be part of one or more operations in FIG. 9 (e.g., operations associated with aspect (7), etc.), as the buffer threshold configured in FIG. 9 may determine when the UE triggers the transmission of the data availability indication received in FIG. 3. Furthermore, one or more operations in FIG. 4 may involve or be part of one or more operations in FIG. 5 (e.g., operations associated with aspect (3), etc.), as the retained logged data configured in FIG. 4 may be the same data that is subsequently received with the logging configuration identifier in FIG. 5. In addition, one or more operations in FIG. 6 (e.g., operations associated with aspect (4), etc.) may involve or be part of one or more operations in FIG. 7 (e.g., operations associated with aspect (5), etc.), as the pending segments indicator in the log header received in FIG. 6 may be the same parameter used by the scheduler in FIG. 7 to manage burst boundaries and allocate transmission resources, and the like. Thus, it can be understood that the descriptions of an operation with reference to one of the FIG. 2 to FIG. 10 may be applicable to another operation of another one of the FIG. 2 to FIG. 10 without departing from the scope of the present disclosure, and redundant descriptions associated therewith may be omitted below for conciseness.
[0114] Further, the methods and operations in two or more of FIG. 2 to FIG. 10 may involve the collection and / or transmission of data associated with one or more AI / ML models. Inthis regard, it can be understood that the collected data may be utilized by the network entity for NW-side AI / ML purposes (e.g., training / fine-tuning one or more AI / ML models, inputting to one or more AI / ML models for inference purpose, providing AI / ML-related services, etc.), and the contents, characteristics, and intents of the collected data may be similar to the “AI / ML data,” “data collected for NW-side AI / ML purposes,” and the like, described above. Thus, it can also be understood that the characteristic and content of the data collected via the method and operations in one of FIG. 2 to FIG. 10 may be applicable to the data collected via the method and operations in another one of FIG. 2 to FIG. 10, and redundant descriptions associated therewith may be omitted for conciseness. Similarly, it can be understood that the characteristic and content of the messages or signals (e.g., RRC message, UAI, etc.) involved in the method and operations of one of FIG. 2 to FIG. 10 may be applicable to the messages or signals involved in the method and operations of another one of FIG. 2 to FIG. 10, and redundant descriptions associated therewith may be omitted for conciseness.> > Example Methods and Operations associated with Aspect (1)
[0115] FIG. 2 illustrates an example method 200, according to one or more example embodiments. The operations in method 200 may be performed by a network entity (e.g., network entity 110) to communicate with a UE (e.g., UE 120) to manage or configure logging trigger conditions for an AI / ML data logging session for NW-side AI / ML purposes.
[0116] As illustrated in FIG. 2, at operation S210, the network entity may be configured to transmit, to the UE, a message including a logging trigger configuration of an AI / ML data logging session. Specifically, the logging trigger configuration may include at least one of: (a) a trigger type field ( "triggerType", etc.) indicating a type of condition that triggers the logging session (e.g., Layer 3 (L3) network event, etc.), (b) an event identifier field (e.g., "eventID", etc.) indicating anevent that triggers the logging session (e.g., maps to A1 / A2 derivative events, etc.), (c) a time to trigger field (e.g., "timeToTrigger", etc.) indicating a time duration for triggering the logging session, or (d) a logging interval field (e.g., "logginginterval", etc.) indicating an interval for performing the logging session.
[0117] According to example embodiments, the trigger type field may indicate a type of condition that triggers the AI / ML data logging session. For instance, the trigger type field may indicate an L3 network event that triggers the AI / ML data logging session. In this regard, L3 network events (also referred to as "L3 radio-condition events") may include events that are evaluated based on L3-filtered measurements (e.g., measurements that have been processed through L3 filtering to smooth out short-term fluctuations, etc.) of one or more radio parameters. For instance, L3 network events may include, but are not limited to: routing and path selection events, connectivity and path failure events, network traffic processing events, security and traffic filtering events, events derived from A1 / A2 measurement events (e.g., events triggered when a serving cell measurement quantity crosses a configured threshold, etc.), events derived from A3 / A4 / A5 / A6 measurement events (e.g., events triggered based on inter-cell measurement comparisons, neighbor cell measurements crossing thresholds, etc.), events derived from B1 / B2 measurement events (e.g., events triggered based on inter-RAT measurement conditions, etc ), and the like. In this regard, by specifying the trigger type, the network entity may instruct the UE on the category of condition that may initiate the AI / ML data logging session.
[0118] According to example embodiments, the event identifier field may indicate a mapping between an identifier (ID) and at least one of: an Al derivative event or an A2 derivative event. In this regard, Al and A2 events may refer to measurement events defined in one or more technical documents of one or more standard organizations (e.g., 3GPP specifications, etc.) thatare triggered based on serving cell measurements. Specifically, an A 1 event may be triggered when a serving cell measurement quantity (e.g., Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Signal-to-Interference-plus-Noise Ratio (SINR), etc.) exceeds a configured threshold, while an A2 event may be triggered when a serving cell measurement quantity fall belows a configured threshold. The event identifier field may also map to specific derivative events (e.g., XI, X2, etc.) that are derived from or extend the A1 / A2 event framework for AI / ML data logging purposes. For instance, an XI derivative event may be triggered when a serving cell RSRP exceeds a configured upper threshold (derived from Al), while an X2 derivative event may be triggered when a serving cell RSRP falls below a configured lower threshold (derived from A2). In this regard, by specifying the event identifier, the network entity may instruct the UE on which specific event may trigger the AI / ML data logging session.
[0119] According to example embodiments, the time to trigger field may indicate a time duration for triggering the logging session. The time to trigger field may have a same scale (e.g., same value range and / or same step size granularity) as a time to trigger (TTT) parameter used for triggering a radio resource management (RRM) operation. In this regard, RRM operations may include, but are not limited to: measurement reporting (e.g., the UE reporting measurement results to the network entity when configured measurement events are triggered, etc.), cell reselection (e g., the UE selecting a more suitable cell based on measurement criteria, etc ), radio link monitoring (e.g., the UE monitoring the quality of the radio link to detect radio link failure conditions, etc.), and the like. The TTT parameter in RRM operations may define a time duration during which a measurement event condition may be required to be continuously satisfied before the event is considered triggered, thereby preventing premature triggering due to transient fluctuations in radio conditions. By aligning the time to trigger parameter for AI / ML data loggingwith the same scale as the TTT parameter used for RRM operations, example embodiments may ensure consistency with existing measurement triggering mechanisms and simplify UE implementation.
[0120] According to example embodiments, the logging interval field may indicate a time interval for performing the AI / ML data logging session. The logging interval field may be optional. In this regard, when the logging interval field is absent, the message may further include instructions to instruct the UE to perform the logging session per reference signal (RS) period. The RS period may refer to the periodicity at which reference signals (e.g., Channel State Information Reference Signal (CSI-RS), Synchronization Signal Block (SSB), etc.) are transmitted by the network entity. For instance, if the CSI-RS periodicity is configured as 20 ms, the UE may perform data logging every 20 ms when the logging interval field is absent. In this regard, by specifying the logging interval (or the default behavior when the logging interval is absent), example embodiments may optimize UE power consumption, especially in scenarios involving short-period reference signals, by allowing the network to configure longer logging intervals when data at every RS transmission is not required.
[0121] According to example embodiments, at least one of the event identifier field, the time to trigger field, or the logging interval field may include an integer value. For instance, the event identifier field may include an integer value in a range of 0 to 31, the time to trigger field may include an integer value in a range of 0 to 500 milliseconds (ms), and the logging interval field may include an integer value in a range of 1 to 1024 ms. An example logging trigger configuration, according to one or more example embodiments, may define the following fields in Table 1:Table 1: Example Parameters of Logging Trigger Configuration Field Type / Range NotestriggerType ENUM {13Condition(0)} Only L3 events in Rel-19 scope eventld INTEGER(0..31) Maps to A1 / A2 derivatives (e.g., XI,X2)timeToTrigger INTEGER(0..500) ms Same scale as RRM TTT logginginterval OPTIONAL 1NTEGER(1..IO24) ms If absent log per RS period
[0122] According to example embodiments, the message (that may be transmitted by the network entity to the UE at operation S210) may include a Radio Resource Control (RRC) message. In this regard, the RRC message may include an RRC Configuration / Reconfiguration message, such as: an RRC Connection Setup message, an RRC Connection Reconfiguration message, an RRC Connection Reestablishment message, and the like. In this regard, the logging trigger configuration (or the associated field(s)) may be included under an Information Element (IE) of the RRC message (e.g., "AIML-LoggingConfig", etc.), thereby providing a standardized and unified container for configuring AI / ML data logging triggers. Alternatively or additionally, the message (that may be transmitted by the network entity to the UE at operation S210) may include a Medium Access Control (MAC) Control Element (CE). In this regard, the logging trigger configuration may be implemented as a payload of the MAC CE that defines the intended configuration (e.g., X bits payload indicates trigger type X, event identifier X, time to trigger X, and / or logging interval X, etc.).
[0123] According to example embodiments, the message (that may be transmitted by the network entity to the UE at operation S210) may further include information or instructions that guide the UE on how to transmit the collected / logged data back to the network entity. Specifically, the message may include information or instructions that guide the UE to include AI / ML-relevant identifiers or information into one or more existing responding or reporting mechanisms. For instance, the message may further include information / instructions that guide the UE to extend acontainer of a Minimization of Drive Tests (MDT) measurement report to include the logged data or content into the associated field(s) of the logging trigger configuration (e.g., adding the new fields as specified in the message to the standard MDT measurement report), thereby enabling the UE to include the specific data needed for NW-side AI / ML operations when sending the MDT measurement report. As another example, the message may further include information / instructions that guide the UE to extend a response message (e.g., measurement report, etc.) (may be referred to as "UEInformationResponse" herein) to include a list of logging entry records with an associated tag / label (e.g., a tag / label that indicates that the purpose of the response message or the associated logging entry records are associated with the AI / ML model training, etc.). In this way, by implementing example embodiments, the existing responding or reporting mechanisms may be reused when possible while adding AI / ML-relevant identifiers as needed.
[0124] According to example embodiments, the message (that may be transmitted by the network entity to the UE at operation S210) may further include information or instructions that guide the UE on how to configure the logging interval. For instance, the message may further include information or instructions that guide the UE to implement logging per RS period as a default logging interval, thereby establishing a default UE behavior where the data logging automatically aligns with the RS periodicity in the absence of a specific logging interval configuration. In this way, example embodiments may support both fixed / default logging intervals and alignment with RS periodicity, thereby providing a balance between consistency and configurability, and ensuring that the logged data / content may be inherently synchronized or aligned with the most relevant downlink signals (e.g., CSI-RS, etc.) for consistent AI / ML operations (e.g., model training, etc.).
[0125] Referring still to FIG. 2, at operation S220, the network entity may be configured to receive, from the UE, a message that includes data collected by the UE according to the logging trigger configuration. In this regard, the UE may collect or log data based on the trigger conditions, event identifiers, time to trigger parameters, and / or logging intervals specified in the logging trigger configuration, and may transmit the collected data to the network entity for NW-side AI / ML purposes (e.g., AUML model training, AUML model inference, etc.). According to example embodiments, the message may include an MDT measurement report. In this regard, one or more containers of the MDT measurement report may be extended (e.g., by the UE according to the logging trigger configuration in the message provided by the network entity) to include one or more AI / ML-related identifiers or parameters (e.g., fields of the logging trigger configuration defined in the message, etc ). Additionally or alternatively, the message may include a response message (e.g., UEInformationResponse, etc.) that includes the collected data along with the associated logging trigger configuration parameters.
[0126] In view of the above, the method and operations described herein introduce and exemplify a mechanism that may be implemented by a network entity (e.g., Next Generation Radio Access Network (NG-RAN), etc.) to configure a unified logging trigger configuration that consolidates trigger-related parameters under a single configuration structure, thereby efficiently and effectively addressing the shortcomings of aspect (1) as described above.
[0127] Advantageously, by implementing the method and operations described herein, the network entity may effectively and efficiently configure the UE for AI / ML data logging by providing a unified logging trigger configuration that consolidates trigger-related parameters (e.g., trigger type, event identifier, time to trigger, logging interval, etc.) under a single configuration structure. Accordingly, example embodiments may resolve implementation ambiguity acrossdifferent UE vendors, optimize UE power consumption through configurable logging intervals, align trigger timing with existing RRM parameters, and maintain granular control over individual logging sessions, thereby addressing the shortcomings of aspect (1) as described above.Example Methods and Operations associated with Aspect (2)
[0128] FIG. 3 illustrates an example method 300, according to one or more example embodiments. The operations in method 300 may be performed by a network entity (e.g., network entity 110) to communicate with a UE (e.g., UE 120) to configure and receive data availability and constraint reporting for an AI / ML data logging session for NW-side AI / ML purposes.
[0129] As illustrated in FIG. 3, at operation S310, the network entity may be configured to transmit, to the UE, a message that includes configuration associated with an AI / ML data logging session (e.g., “AI / ML-LoggingConfig”, etc.). Specifically, the configuration may include a field (e.g., "availability Tri ggerMode", etc.) indicating a mode for triggering a transmission of an indication associated with availability of data during the logging session (e.g., “DataAvailabilitylndication”, etc.). In this regard, the mode may indicate which entity (e.g., the network entity, the UE, etc.) evaluates one or more thresholds (e.g., buffer thresholds, power thresholds, etc.) to determine when the UE should transmit the indication associated with the availability of the data.
[0130] According to example embodiments, the mode may include at least one of: (a) a network-controlled mode (e.g., "nwControlled", etc.) indicating that the network entity controls or sets the thresholds for when the UE should transmit the indication, or (b) a UE-autonomous mode (e.g., "ueAutonomous", etc.) indicating that the UE autonomously decides when to transmit the indication based on its own internal conditions or heuristics. In this regard, the network-controlled mode may provide deterministic behavior, where the network entity defines the specific thresholds(e g., buffer threshold in absolute bytes, power threshold, etc.) that the UE evaluates to determine when to transmit the indication. Conversely, the UE-autonomous mode may allow the UE to apply its own heuristics or internal logic to determine when to transmit the indication, which may be suitable for deployment scenarios where the UE has better knowledge of its own internal conditions (e.g., remaining battery capacity, available memory, processor utilization, etc.).
[0131] According to example embodiments, the availabilityTriggerMode field may be added to AI / ML-LoggingConfig that is defined in one or more technical specifications of 3GPP (e.g., TS 38.331). An example AI / ML-LoggingConfig that includes the availabilityTriggerMode field, according to one or more example embodiments, is presented in Table 2:Table 2: Example AI / ML-LoggingConfigField Type Default Semantics availabilityTriggerMode ENUM { nwControlled(O), nwControlled Decides which ueAutonomous(l) } entity evaluates buffer & power thresholds
[0132] As illustrated in Table 2, the network entity may configure the network-controlled mode (e.g., "nwControlled", etc.) as the default mode for triggering the transmission of the data availability indication. In this regard, by configuring the default mode, example embodiments may ensure deterministic behavior when no specific configuration is provided by the network entity, while still allowing the network entity to configure the UE-autonomous mode when desired.
[0133] According to example embodiments, the message (that may be transmitted by the network entity to the UE at operation S310) may include an RRC message. In this regard, the RRC message may include an RRC Configuration / Reconfiguration message, such as: an RRC Connection Setup message, an RRC Connection Reconfiguration message, an RRC ConnectionReestablishment message, and the like. In this regard, the configuration (e.g., the availability trigger mode field, etc.) may be included under an Information Element (IE) of the RRC message (e.g., "AIML-LoggingConfig", etc.). Alternatively or additionally, the message (that may be transmitted by the network entity to the UE at operation S310) may include a MAC Control Element CE. In this regard, the configuration may be implemented as a payload of the MAC CE that defines the intended configuration (e.g., X bits payload indicates availability trigger mode X, etc.).
[0134] Referring still to FIG. 3, at operation S320, the network entity may be configured to receive, from the UE, a message that includes a data availability indication. According to example embodiments, the message may include UE Assistance Information (UAI) (may be referred to as "UEAssistancelnformation" herein) that includes the data availability indication.
[0135] According to example embodiments, the data availability indication may be implemented as a structured element (e.g., "DataAvailability Indi cation", etc.) in the message (e.g., UAI, etc.). The structured element may be implemented as a sequence of fields (e.g., ASN.l SEQUENCE, etc.), including at least one of: (a) a field (e.g., "cause", etc.) indicating a cause of why the UE is sending the data availability indication (e.g., the buffer of the UE has reached a configured threshold, the buffer of the UE is full, the UE is in low power condition, etc.), or (b) a field (e.g., "remainingBytes", etc.) indicating a number of bytes of data remaining in the buffer of the UE.
[0136] According to example embodiments, the cause field may be implemented as an enumerated type (e.g., ENUMERATED, etc.) and may indicate at least one of: a buffer threshold reached condition (e.g., "buffer ThresholdReached(O)", indicating that the buffer has reached a configured threshold), a buffer full condition (e.g., "bufferFull(l)", indicating that the buffer isfull), or a low power condition (e.g., "lowPower(2)", indicating that the UE is in a low power state). The remaining bytes field may be optional and may indicate an integer value (e.g., INTEGER in a range of 0 to 262143, etc.) representing the number of bytes of logged data remaining in the UE's buffer. An example structured element of data availability indication, according to one or more example embodiments, may be presented in Table 3:Table 3: Example Structured Element of Data Availability Indication DataAvailabilitylndication ::= SEQUENCE {cause ENUMERATED {bufferThresholdReached (0),bufferFull (1),lowPower (2)},remamingBvtes INTEGER (0..262143) OPTIONALJ _
[0137] In this regard, by providing a structured data availability indication with a cause field and an optional remaining bytes field, the network entity may be able to prioritize urgent cases (e.g., low-power conditions where the UE may soon become unavailable) over routine events (e.g., buffer threshold reached conditions that may not require immediate action), thereby enabling efficient scheduling of data retrieval based on the urgency or nature of the data availability indication. For instance, when the network entity receives a data availability indication with a cause of "lowPower(2)", the network entity may prioritize fetching the logged AI / ML data from the UE before the UE becomes unavailable due to power depletion. Conversely, when the network entity receives a data availability indication with a cause of "bufferThresholdReached(O)", the network entity may schedule data retrieval at a lower priority, as the UE may still have sufficient buffer capacity and power to continue logging and transmission.
[0138] According to example embodiments, the message received at operation S320 may further include a field (e.g., "resourceConstraintCause", etc.) indicating a resource constraint cause of the UE. The resource constraint cause field may indicate at least one of: a low memory condition(e g., the UE's available memory has fallen below a threshold, also referred to as "memLow"), a low power condition (e.g., the UE's battery level has fallen below a threshold, also referred to as "powerLow"), or a busy processor condition (e.g., the UE's processor utilization has exceeded a threshold, also referred to as "cpuBusy"). According to example embodiments, the resource constraint cause field may be implemented as a bit string (e.g., BIT STRING of SIZE 3), wherein each bit of the bit string corresponds to a respective resource constraint cause. An example UAI message that includes the resourceConstraintCause field, according to one or more example embodiments, may be presented in Table 4:.Table 4: Example UAI Including ResourceConstraintCause Field Type / Size Bit Mapping resourceConstraintCause BIT STRING (SIZE 3) bO memLow, b 1 powerLow, b2 cpuBusy
[0139] In this regard, by implementing the resource constraint cause field as a bit string, multiple resource constraint causes may be indicated simultaneously (e.g., both low memory and low power conditions may be indicated at the same time by setting both bO and bl to " 1 ", etc.). By providing an explicit indication of the resource constraint cause, the network entity may be able to distinguish between intentional pauses in data logging and unintentional interruptions caused by UE resource limitations, and may identify the specific reason why the data logging session paused or stopped, thereby avoiding the ingestion of incomplete or biased datasets for AI / ML operations. For instance, if the network entity receives a resource constraint cause indicating "memLow" (bO = 1), the network entity may determine that the UE paused logging due to insufficient memory and may schedule immediate data retrieval to free the UE's buffer. Conversely, if the network entity receives a resource constraint cause indicating "cpuBusy" (b2 = 1), the network entity may determine that the UE paused logging due to high processor utilization and may defer data retrieval until the UE's processor load decreases.
[0140] According to example embodiments, the message (that may be received by the network entity from the UE at operation S320) may further include information or instructions that enable the network entity to manage the data retrieval process. Specifically, the network entity may utilize the data availability indication (e.g., cause field, remaining bytes field, etc.) and / or the resource constraint cause field to determine the appropriate timing, priority, and scheduling of data retrieval from the UE. For instance, the network entity may implement a priority-based data retrieval scheme where data availability indications with a cause of "lowPower" or resource constraint causes indicating "powerLow" are assigned a higher retrieval priority than data availability indications with a cause of "bufferThresholdReached".
[0141] In view of the above, the method and operations described herein introduce and exemplify a mechanism that may be implemented by a network entity to configure a data availability indication trigger mode and receive structured data availability indications and resource constraint cause reporting from the UE, thereby efficiently and effectively addressing the shortcomings of aspect (2) as described above.
[0142] Advantageously, by implementing the method and operations described herein, the network entity may effectively and efficiently configure the availability indication trigger mode to adapt to different deployment scenarios or network operator preferences (e.g., network-controlled thresholds for deterministic behavior, UE-autonomous heuristics for flexible behavior, etc.), receive structured data availability indications that enable prioritization of data retrieval based on the cause of availability (e.g., prioritizing urgent low-power cases over routine buffer-threshold events), and receive explicit resource constraint cause indications that enable identification of the specific reason for data logging pauses or interruptions (e.g., low memory, low power, busy processor, etc.). Accordingly, example embodiments may enable efficient scheduling of dataretrieval, support both network-controlled and UE-autonomous availability-trigger logic, and avoid the ingestion of incomplete or biased datasets for AI / ML operations, thereby addressing the shortcomings of aspect (2) as described above.Example Methods and Operations associated with Aspect (3)
[0143] FIG. 4 illustrates an example method 400, according to one or more example embodiments. The operations in method 400 may be performed by a network entity (e.g., network entity 110, acting as a source network entity such as a source gNB) to communicate with a UE (e.g., UE 120) to configure data retention behavior during mobility events (e.g., handover, etc.) for NW-side AI / ML purposes.
[0144] As illustrated in FIG. 4, at operation S410, the network entity may determine whether handover of the UE to a target cell is required. In this regard, it may be assumed that the network entity is a source base station that manages a source cell that is serving the UE, and the target cell is a neighboring cell of the UE that may be managed by another network entity (i.e., a target base station). The network entity may receive a measurement report (which may, but not necessarily, contain the AI / ML data) from the UE, which includes parameters that indicate the signal strength and quality of both the source cell (which is currently serving the UE) and one or more neighboring cells (which includes the target cell). Accordingly, the network entity may determine whether a handover is required by, for example, determining whether the signal strength / quality of a neighboring cell exceeds the signal strength / quality of the source cell by a predefined threshold. In this case, based on determining that the signal strength / quality of a neighboring cell exceeds the signal strength / quality of the source cell by the predefined threshold, the network entity may determine that a handover is required and select the neighboring cell as the target cell. According to example embodiments, the network entity may determine that handoveris required based on one or more factors, such as: a serving / source cell measurement quantity (e.g., RSRP, RSRQ, or SINR) has crossed a configured threshold, radio link quality degradation, load balancing considerations, and the like. It is contemplated that the network entity may perform additional or alternative operations when determining whether a handover is required, without departing from the scope of the present disclosure.
[0145] At operation S420, in response to determining that handover of the UE to a target cell is required, the network entity may provide, to the UE, a message that includes information associated with data retention. According to example embodiments, the message may include an RRC Reconfiguration message. In this regard, by providing the data retention information in an RRC Reconfiguration message from the source network entity before handover execution, example embodiments may enhance interoperability across different vendors, as the data retention directive may not be ignored even when the source and target network entities are from different vendors. Specifically, placing the data retention directive inside the handover command from the target cell may be problematic if the source and target network entities are from different vendors, as the proprietary container may be ignored by the target network entity. Pre-handover delivery of the directive (in RRC Reconfiguration message) from the source network entity may address this interoperability concern.
[0146] According to example embodiments, the message may include a parameter (e.g., "retainLoggedData", etc.) that indicates whether the UE may retain logged data after handover. The parameter may be implemented as a Boolean parameter. For instance, a first value (e.g., "TRUE") may indicate that the UE may retain logged data after handover, while a second value (e.g., "FALSE") may indicate that the UE may discard logged data after handover. In this regard, by providing an explicit directive on whether to retain or discard logged data during mobilityevents, example embodiments may address the indeterminate behavior in the related art, where some UEs may discard logs whenever a handover command arrives, other UEs may retain logs by default, and some UEs may behave inconsistently depending on the type of mobility event (e.g., retaining data during intra-frequency handover but flushing data during transition to RRC IDLE state, RRC INACTIVE state, or RLF).
[0147] According to example embodiments, the message may further include a parameter (e.g., "retentionTimer", etc.) that indicates the time duration for which the UE may retain the logged data before discarding. The retention timer parameter may be specified as an integer parameter (e.g., between 24 to 168 hours, with a default timer of 48 hours, etc.). In this regard, the retention timer may be similar to an upper bound timer used in MDT operations, which may define a maximum duration for retaining logged data. By providing a retention timer, example embodiments may address the lifetime management gap in the related art, ensuring that retained logs self-expire to satisfy privacy policies and free memory resources. For instance, if the UE enters RRC IDLE state or executes inter-RAT handover shortly after the handover event, the retained AI / ML-related logs may self-expire when the retention timer expires, rather than being retained indefinitely. Accordingly, example embodiments may guarantee that UEs do not hoard stale data indefinitely, thereby limiting privacy exposure and memory pressure.
[0148] According to example embodiments, the UE may discard the retained logs when at least one of the following conditions is satisfied: (a) the retention timer expires, (b) the UE executes a subsequent handover, (c) the UE transitions to a specific RRC state (e g., RRC IDLE state, etc.), or (d) the UE experiences a radio link failure (RLF). In this regard, by defining deterministic conditions under which the UE discards retained logs, example embodiments may ensure consistent data retention behavior across different UE implementations and vendors.
[0149] In view of the above, the method and operations described herein with reference to FIG. 4 introduce and exemplify a mechanism that may be implemented by a source network entity (e.g., source gNB) to instruct the UE on how to handle collected AI / ML data during mobility events, thereby ensuring deterministic data retention behavior across different UE implementations and vendors, and managing the lifetime of retained logs through a configurable retention timer.
[0150] FIG. 5 illustrates an example method 500, according to one or more example embodiments. The operations in method 500 may be performed by a network entity (e.g., network entity 110, acting as a target network entity such as a target gNB) to receive and process AI / ML data from a UE (e.g., UE 120) after a mobility event (e.g., handover, etc.) for NW-side AI / ML purposes. It is contemplated that, in some example implementations, the operations in FIG. 5 may be performed subsequent to the operations in FIG. 4. For instance, the operations in FIG. 4 may be performed by a source network entity (e.g., source gNB) prior to or during a handover event, and the operations in FIG. 5 may be performed by a target network entity (e.g., target gNB) after the handover event. Nevertheless, it can be understood that the operations in FIG. 4 and FIG. 5 may be performed independently, concurrently, or in any other suitable order, without departing from the scope of the present disclosure.
[0151] As illustrated in FIG. 5, at operation S510, the network entity (e.g., target network entity) may receive, from the UE, a message that includes AI / ML data collected by the UE along with associated logging information. According to example embodiments, the logging information may include an identifier (e.g., "logConfigID", etc.) that indicates different types of logging entry structures associated with the collected AI / ML data. The identifier may be implemented as a 5-bit integer (e.g., INTEGER (0..31)) or similar data type that may be prepended to each log entrystructure (e.g., "logMeasEntry", "aiLogMeasEntry", etc.). Tn this regard, by including a logging configuration identifier with each log entry structure, the target network entity may be able to identify the origin or associated logging session of each log entry, even when the target network entity receives data from multiple concurrent logging sessions that were configured by the source network entity.
[0152] At operation S520, the network entity (e.g., target network entity) may determine the type of collected AEML data based on the logging information. For instance, a first value of the logging configuration identifier (e.g., logConfigID = 00001) may indicate that the collected AI / ML data is associated with a beam-AI dataset, while a second value of the logging configuration identifier (e.g., logConfigID = 00002) may indicate that the collected AEML data is associated with a CSI-AI dataset. In this regard, by associating different logging configuration identifier values with different types of AI / ML datasets, example embodiments may enable the target network entity to properly separate and route the collected data to the appropriate AEML model pipelines when multiple concurrent AI / ML data logging sessions (e.g., beam-AI logging and CSI-AI logging) were configured by the source network entity.
[0153] An example log entry structure that has a 5-bit logConfigID prepended thereto, according to one or more example embodiments, may be presented in Table 5:Table 5: Example Log Entry StructureaiLogMeasEntry ::= SEQUENCE {logConfigID INTEGER (0..31),-- existing fieldsJ _
[0154] In view of the above, the method and operations described herein with reference to FIG. 5 introduce and exemplify a mechanism that may be implemented by a target network entity (e.g., target gNB) to receive and process AEML data from the UE after a mobility event, and todetermine the type of collected AI / ML data based on the logging configuration identifier, thereby enabling proper dataset separation and routing for AI / ML model training when multiple concurrent logging configurations are active.
[0155] Advantageously, by implementing the methods and operations described herein with reference to FIG. 4 and FIG. 5, the network entity may effectively and efficiently instruct the UE on how to handle collected AI / ML data during mobility events through an explicit data retention directive and a configurable retention timer (e.g., operations associated with FIG. 4), and may properly separate and route collected AI / ML data from multiple concurrent logging sessions based on the logging configuration identifier (e.g., operations associated with FIG. 5). Accordingly, example embodiments may ensure deterministic data retention behavior across different UE implementations and vendors, prevent indefinite data retention through configurable retention timers, enhance interoperability through pre-handover delivery of data retention directives, and enable proper dataset separation for concurrent logging sessions, thereby addressing the shortcomings of aspect (3) as described above.>> Example Methods and Operations associated with Aspect (4)
[0156] FIG. 6 illustrates an example method 600, according to one or more example embodiments. The operations in method 600 may be performed by a network entity (e.g., network entity 110) to receive and reassemble AI / ML data from a UE (e g., UE 120) for NW-side AI / ML purposes.
[0157] As illustrated in FIG. 6, at operation S610, the network entity may be configured to receive, from the UE, a message that includes a log header and / or logged measurement entry metadata associated with AI / ML data collected by the UE.
[0158] According to example embodiments, the log header (e.g., "AIMLLogHeader", etc.) may be a fixed header that is prepended to one or more segments of the AI / ML data transmitted by the UE. In this regard, the log header may include information associated with at least one of: (a) an identifier (e.g., "logConfigID", etc.) that indicates different types of logging entry structures and / or logging configuration associated with the collected AI / ML data, (b) a parameter (e.g., "segmentNo", etc.) that indicates a segment sequence number of the AI / ML data, (c) a parameter (e.g., "lastSegment", etc.) that indicates whether the collected AI / ML data is the final segment of the associated AI / ML data, (d) a parameter (e.g., "logPriority", etc.) that indicates the priority (e.g., high, normal, etc.) of the collected AI / ML data, or (e) a parameter (e.g., "moreSegmentsPending", etc.) that indicates whether additional data segments of AI / ML data are pending transmission by the UE.
[0159] According to example embodiments, the logging configuration identifier (e.g., "logConfigID") may be implemented as an integer (e.g., INTEGER (0..15), etc.) that identifies which logging session or configuration the data segment belongs to. In this regard, when the network entity aggregates AI / ML data from a large number of UEs (e.g., millions of UEs), and the data arrives out of order (e.g., in the form of non-continuous data segments, etc.) due to various factors (such as HARQ retransmissions, DRX sleep cycles, multi-SIM split uplink transmissions, or other transmission-related events), the logging configuration identifier may enable the network entity to identify and separate AI / ML data from different concurrent logging sessions (e.g., beam-AI logging session, CSLAI logging session, etc ).
[0160] According to example embodiments, the segment sequence number (e.g., "segmentNo") may be implemented as an integer (e.g., INTEGER (0..255), etc.) that indicates the position of the data segment within the overall data transmission. In this regard, when datasegments arrive out of order (e.g., segment 0, segment 2, segment 1 , segment 3), the network entity may use the segment sequence number to reorder the data segments in the correct sequence during reassembly.
[0161] According to example embodiments, the last segment indicator (e.g., "lastSegment") may be implemented as a Boolean parameter (e.g., “TRUE” indicating that the segment is the final segment, “FALSE” indicating that additional segments are expected, etc.). In this regard, the network entity may use the last segment indicator to determine when all data segments for a particular transmission (or a particular AI / ML data) have been received, and may proceed to process the complete dataset accordingly.
[0162] According to example embodiments, the log priority (e.g., "logPriority") may be implemented as an enumerated type (e.g., ENUM {normal, high}, etc.) that indicates the priority level of the AI / ML data. In this regard, the network entity may use the log priority to prioritize processing of high-priority data over normal-priority data. For instance, AI / ML data associated with time-sensitive use cases (e.g., beam prediction during high-mobility scenarios) may be assigned a high priority, while AI / ML data associated with less time-sensitive use cases may be assigned a normal priority.
[0163] According to example embodiments, the pending segments indicator (e.g., "moreSegmentsPending") may be implemented as a Boolean parameter (e.g., “TRUE” indicating that additional segments are pending transmission, “FALSE” indicating that no additional segments are pending, etc ). In this regard, the network entity may use the pending segments indicator to determine whether additional data is expected from the UE, and may manage resource allocation and scheduling accordingly.
[0164] An example log header structure, according to one or more example embodiments, may be presented in Table 6:Table 6: Example Log Header StructureAIMLLogHeader ::= SEQUENCE {logConfigID INTEGER (0 .15),segmentNo INTEGER (0..255),lastSegment BOOLEAN.logPriority ENUM {normal, high},moreSegmentsPending BOOLEANJ _
[0165] According to example embodiments, the logged measurement entry metadata (e.g., "LoggedMeasurementEntry", etc.) may include per-entry information that enables the network entity to associate each log entry with its corresponding context. Specifically, the logged measurement entry metadata may include information associated with at least one of: (a) an identifier (e.g., "celllD", etc.) that indicates a network cell associated with the collected AI / ML data, (b) a parameter (e.g., "gaplnd", etc.) that indicates whether there was any measurement gap associated with the collected AI / ML data, or (c) a parameter (e.g., "entryPriority", etc.) that indicates a priority level of the collected AI / ML data.
[0166] According to example embodiments, the cell identifier (e.g., "celllD") may indicate the network cell in which the UE performed the measurement. In this regard, the cell identifier may enable the network entity to associate logged AI / ML data with a specific network cell, which may be important for AI / ML model training that is cell-specific (e.g., cell-specific beam prediction models, etc.) or for mobility scenarios where the UE may traverse multiple network cells during a logging session.
[0167] According to example embodiments, the gap indicator (e.g., "gaplnd") may indicate whether there was a measurement gap (e.g., due to DRX sleep, measurement gaps configured by the network entity, inter-frequency measurements, or other interruptions) associated with thecollected AI / ML data. In this regard, the gap indicator may enable the network entity to identify discontinuities in the time series data, which may be critical for AI / ML training to avoid treating discontinuous data as continuous, potentially biasing the AI / ML model.
[0168] According to example embodiments, the entry priority (e.g., "entry Priority") may be an optional parameter that indicates the priority level of an individual log entry within a data segment. In this regard, the entry priority may enable the network entity to prioritize processing of specific entries within a data segment, which may be useful when some measurements are more valuable or critical for AI / ML training than others (e.g., measurements taken during rapid channel changes may be more valuable than measurements taken during stable channel conditions).
[0169] Referring still to FIG. 6, at operation S620, the network entity may be configured to reassemble interleaved segments of the AI / ML data based on the log header and / or logged measurement entry metadata.
[0170] According to example embodiments, the network entity may separate received data segments into different session buckets based on the logging configuration identifier (e.g., "logConfigID"). For instance, segments with logConfigID = 1 may be grouped into a beam-AI bucket, while segments with logConfigID = 2 may be grouped into a CSLAI bucket.
[0171] According to example embodiments, within each session bucket, the network entity may order the segments based on the segment sequence number (e.g., "segmentNo"). For instance, if segments arrive in the order of segmentNo = 0, 2, 1, 3, the network entity may reorder them to segmentNo = 0, 1, 2, 3.
[0172] According to example embodiments, the network entity may determine that all segments for a particular transmission have been received based on the last segment indicator (e.g.,"lastSegment" = TRUE). In this regard, the network entity may proceed to process the complete dataset once the final segment has been received.
[0173] According to example embodiments, the network entity may manage pending data based on the pending segments indicator (e.g., "moreSegmentsPending"). For instance, when the pending segments indicator is TRUE, the network entity may continue to allocate resources for receiving additional segments from the UE, and may defer processing until all pending segments have been received.
[0174] According to example embodiments, the network entity may identify and handle gaps or discontinuities in the logged data based on the gap indicator (e.g., "gaplnd") in the logged measurement entry metadata. In this regard, the network entity may mark portions of the time series data as discontinuous, thereby preventing the AI / ML model from treating discontinuous data as continuous during training.
[0175] In view of the above, the method and operations described herein introduce and exemplify a mechanism that may be implemented by a network entity to receive AVML data with a fixed log header and / or per-entry metadata, and to reassemble interleaved segments of the AVML data based on the log header and / or metadata, thereby efficiently and effectively addressing the shortcomings of aspect (4) as described above.
[0176] Advantageously, by implementing the method and operations described herein, the network entity may effectively and efficiently reassemble interleaved segments of AI / ML data from multiple UEs and multiple logging sessions, even when data arrives out of order due to HARQ retransmissions, DRX sleep cycles, multi-SIM split uplink transmissions, or other transmission-related events. Specifically, the log header may enable the network entity to identify the logging session (via logConfigID), determine the sequence of segments (via segmentNo),identify the final segment (via lastSegment), determine the priority of the data (via logPriority), and manage pending data (via moreSegmentsPending). Furthermore, the logged measurement entry metadata may enable the network entity to associate each log entry with its corresponding cell (via celllD), identify measurement gaps (via gaplnd), and determine the priority of individual entries (via entryPriority). Accordingly, example embodiments may enable proper reassembly of interleaved AI / ML data segments, identification and handling of measurement gaps, and prioritized processing of AI / ML data, thereby addressing the shortcomings of aspect (4) as described above.>> Example Methods and Operations associated with Aspect (5)
[0177] FIG. 7 illustrates an example method 700, according to one or more example embodiments. The operations in method 700 may be performed by a network entity (e.g., network entity 110) to communicate with a UE (e.g., UE 120) to configure transport bearer and signaling radio bearer (SRB) usage for AI / ML data transmission for NW-side AVML purposes.
[0178] As illustrated in FIG. 7, at operation S710, the network entity may be configured to transmit, to the UE, a message that includes a data transport configuration for transporting AI / ML data. Specifically, the data transport configuration may include information associated with at least one of: (a) a logical channel group (LCG) assignment (e.g., "logicalChannel Group", etc.) for transmitting AI / ML data, or (b) a transport bearer (e.g., SRB-4, etc.) for transmitting AI / ML data.
[0179] According to example embodiments, the LCG assignment may indicate a logical channel group (e g., "aiModelData-LCG", etc.) that is dedicated to AI / ML data traffic. In this regard, the LCG for AI / ML data may be separated from (or operate independently from) the LCG used for transmitting other types of logged data (e.g., MDT data). By assigning a dedicated LCG to AI / ML traffic, the MAC layer of the network entity may distinguish between AI / ML data trafficand other types of logged data, thereby enabling the MAC scheduler to allocate appropriate resources to each type of traffic independently. In the related art, AI / ML data and MDT data may share the same SRB (e.g., SRB-4) with a single LCG, which may result in head-of-line blocking, where one type of data (e.g., AI / ML data) may starve or delay the transmission of another type of data (e.g., MDT data), or vice versa. By assigning a dedicated LCG to AI / ML traffic while sharing the same SRB (e.g., SRB-4), example embodiments may avoid head-of-line blocking between AI / ML data and MDT data, while maintaining the existing security context of the SRB without redefining another SRB.
[0180] According to example embodiments, the transport bearer may indicate a signaling radio bearer (e.g., SRB-4) for transmitting AI / ML data. In this regard, SRB-4 may be a signaling radio bearer defined in one or more technical documents of one or more standard organizations (e.g., 3GPP, etc.) that provides a security context for transmitting signaling data between the UE and the network entity. By reusing SRB-4 for AI / ML data transmission, example embodiments may leverage the existing security context of SRB-4 without introducing additional security overhead or defining a new SRB.
[0181] According to example embodiments, the message (that may be transmitted by the network entity to the UE at operation S710) may include an RRC message. In this regard, the data transport configuration (e.g., LCG assignment, transport bearer, etc.) may be included in one or more IES of the RRC message. Alternatively or additionally, the message may include a MAC CE. In this regard, the data transport configuration may be implemented as a payload of the MAC CE that defines the intended configuration (e.g., X bits payload indicates LCG assignment X and / or transport bearer X, etc.).
[0182] Referring still to FIG. 7, at operation S720, the network entity may be configured to receive, from the UE, a message that includes collected AI / ML data according to the data transport configuration. According to example embodiments, the message may include a parameter (e.g., "moreSegmentsPending", etc.) indicating whether additional data segments of the AI / ML data are pending transmission by the UE. It is contemplated that this parameter may be similar to the parameter described above with reference to FIG. 6 (e.g., the "moreSegmentsPending" parameter in the log header), and thus it can be understood that the descriptions of this parameter with reference to FIG. 6 may be applicable herein, and redundant descriptions associated therewith may be omitted for conciseness. This parameter may be implemented as a Boolean parameter (e.g., “TRUE” indicating that additional segments are pending transmission, “FALSE” indicating that no additional segments are pending). In this regard, when the parameter indicates TRUE (e.g., moreSegmentsPending = 1), the UE may continue using the assigned SRB (e.g., SRB-4) and the assigned LCG (e.g., aiModelData-LCG) until all segments have been transmitted. By providing the pending segments indicator, the network entity may gain explicit knowledge of burst boundaries associated with AI / ML data transmissions, which may not be available in the related art.
[0183] At operation S730, the network entity may schedule transmission resources based on the data transport configuration and / or the pending segments indicator. The scheduling of transmission resources may include at least one of: (a) throttling bursts during congestion based on pending AI / ML data, (b) allocating transmission resources based on burst boundaries (e g., when the pending segments indicator transitions from TRUE to FALSE, the network entity may determine that the burst has completed and may reallocate resources accordingly), or (c) applying fair-queue and / or weighted deficit round-robin scheduling between AI / ML data and other types oflogged data (e.g., MDT data). In this regard, by utilizing the separate LCG assignment and the pending segments indicator, the MAC scheduler may allocate fewer bytes per Transmission Time Interval (TTI) to AI / ML traffic when congestion is detected, throttle bursts to prevent AI / ML data from starving MDT data, and apply fair scheduling algorithms to ensure equitable resource allocation between different types of logged data.
[0184] In view of the above, the method and operations described herein introduce and exemplify a mechanism that may be implemented by a network entity to configure a separate LCG for AI / ML data traffic while sharing the same SRB with other logged data (e.g., MDT data), and to utilize a pending segments indicator for managing burst boundaries and scheduling transmission resources, thereby efficiently and effectively addressing the shortcomings of aspect (5) as described above.
[0185] Advantageously, by implementing the method and operations described herein, the network entity may effectively and efficiently avoid head-of-line blocking between AI / ML data and other logged data by assigning a separate LCG to AI / ML traffic, maintain the existing security context of SRB-4 without redefining another SRB, gain explicit knowledge of burst boundaries through the pending segments indicator, and apply fair scheduling algorithms (e.g., fair-queue, weighted deficit round-robin, etc.) to ensure equitable resource allocation between AI / ML data and other types of logged data. Accordingly, example embodiments may enable efficient management of AI / ML data transmission while maintaining the quality of service for other types of logged data, thereby addressing the shortcomings of aspect (5) as described above.> > Example Methods and Operations associated with Aspect (6)
[0186] FIG. 8 illustrates an example method 800, according to one or more example embodiments. The operations in method 800 may be performed by a network entity (e.g., networkentity 110) to communicate with a UE (e.g., UE 120) to configure cross-layer reference signal (RS) referencing for an AI / ML data logging session for NW-side AI / ML purposes.
[0187] As illustrated in FIG. 8, at operation S810, the network entity may be configured to transmit, to the UE, a message that includes data reporting configuration associated with an AI / ML data logging session. Specifically, the data reporting configuration may include information associated with at least one of: (a) an identifier (e.g., "linkedCSI-ReportConfigID'', etc.) indicating configuration associated with CSI reporting (e.g., “CSI-ReportConfig”, etc.) from which the UE may inherit RS definitions, (b) an identifier (e.g., "rsSetID", etc.) indicating an RS resource set within the CSI-ReportConfig, or (c) an explicit RS ID indicating RS resources for the logging session.
[0188] According to example embodiments, the linked CSI-ReportConfig identifier may reference an existing CSI-ReportConfig. In this regard, CSI-ReportConfig may be a configuration structure defined in one or more technical documents of one or more standard organizations (e.g., 3GPP, etc.) that specifies how the UE should report CSI measurements to the network entity. The CSI-ReportConfig may include parameters such as which RS resources (e.g., CSLRS, SSB, etc.) the UE should measure, how the UE should report the measurements (e.g., reporting periodicity, reporting format, etc ), and associated RS resource set lists. The linked CSI-ReportConfig identifier may be implemented as an integer (e g., INTEGER (0..255), etc.) that references a specific CSI-ReportConfig. In this regard, when the linked CSI-ReportConfig identifier is present, the UE may inherit the RS definitions (e.g., RS resource identifiers, RS resource set configurations, etc.) from the referenced CSI-ReportConfig, thereby avoiding the need to duplicate RS details in the logging configuration message. By inheriting RS definitions from an existing CSI-ReportConfig, example embodiments may guarantee feature alignment between the logged dataand the data used for AI / ML model inference, as the logging session may use exactly the same RS pool used for inference.
[0189] According to example embodiments, the RS set identifier (e.g., "rsSetID") may indicate a stable identifier of an RS resource set within the CSI-ReportConfig. It is contemplated that this identifier may be exposed in the CSI-ReportConfig ASN.1 structure (e.g., via a liaison request to RAN-1, etc.) to provide a stable reference point for cross-layer RS referencing. In this regard, by referencing a specific RS resource set within the CSI-ReportConfig, the network entity may provide more granular control over which RS resources the UE uses for the logging session, without duplicating the entire CSI-ReportConfig structure.
[0190] According to example embodiments, when the linked CSI-ReportConfig identifier is absent, the UE may use explicit RS IDs provided in the data reporting configuration. In this regard, the explicit RS IDs may indicate specific RS resources (e.g., specific CSI-RS resources, specific SSB resources, etc.) that the UE should use for the AI / ML data logging session. By providing explicit RS IDs, the network entity may configure custom RS sets that are not associated with any existing CSI-ReportConfig, which may be suitable for use cases such as mobilityprediction AI / ML operations where the network entity may prefer RS resources (e.g., occasional SSB transmissions) that are not part of CSI inference.
[0191] According to example embodiments, the message may further include instructions to instruct or guide the UE to determine whether to inherit RS definitions from the configuration associated with CSI reporting. For instance, the UE may evaluate whether the linked CSI-ReportConfig identifier is present in the data reporting configuration. If the linked CSL ReportConfig identifier is present, the UE may inherit the RS definitions from the referenced CSL ReportConfig. If the linked CSI-ReportConfig identifier is absent, the UE may use the explicit RSIDs provided in the data reporting configuration. In this regard, by providing a mechanism that supports both inheritance-based and explicit RS referencing, example embodiments may address the diverging operator requirements, where some operators may prefer inference-aligned logging (using the same RS pool as CSI inference) and other operators may prefer custom RS set logging (using RS resources not part of CSI inference).
[0192] According to example embodiments, the message (that may be transmitted by the network entity to the UE at operation S810) may include an RRC message. In this regard, the data reporting configuration may be included under an IE of the RRC message (e.g., "AI / ML-LoggingConfig", etc.). Alternatively or additionally, the message may include a MAC CE. In this regard, the data reporting configuration may be implemented as a payload of the MAC CE that defines the intended configuration.
[0193] Referring still to FIG. 8, at operation S820, the network entity may be configured to receive, from the UE, a message that includes collected AI / ML data, based on the data reporting configuration. In this regard, the UE may collect or log measurements based on the RS resources determined according to the data reporting configuration (e.g., inherited from the referenced CSI-ReportConfig, or based on the explicit RS IDs), and may transmit the collected data to the network entity forNW-side AI / ML purposes.
[0194] In view of the above, the method and operations described herein introduce and exemplify a mechanism that may be implemented by a network entity to configure cross-layer RS referencing for AI / ML data logging by either referencing an existing CSI-ReportConfig configuration or specifying explicit RS resources, thereby efficiently and effectively addressing the shortcomings of aspect (6) as described above.
[0195] Advantageously, by implementing the method and operations described herein, the network entity may effectively and efficiently configure the UE to log measurements on the desired RS resources while maintaining flexibility for both inference-aligned logging and custom RS set logging. Specifically, by providing a linked CSI-ReportConfig identifier, example embodiments may avoid duplicating RS details in every logging configuration message (thereby reducing signaling overhead and avoiding version skew between the logging configuration and the underlying RS configuration), while guaranteeing feature alignment between the logged data and the data used for AI / ML model inference. Furthermore, by supporting explicit RS IDs when the linked CSI-ReportConfig identifier is absent, example embodiments may support custom RS sets for use cases that are not associated with any existing CSI-ReportConfig. Accordingly, example embodiments may enable flexible and efficient RS referencing for AI / ML data logging, thereby addressing the shortcomings of aspect (6) as described above.> > Example Methods and Operations associated with Aspect (7)
[0196] FIG. 9 illustrates an example method 900, according to one or more example embodiments. The operations in method 900 may be performed by a network entity (e.g., network entity 110) to communicate with a UE (e.g., UE 120) to configure memory and buffer limits for an AI / ML data logging session for NW-side Al / AIL purposes.
[0197] As illustrated in FIG. 9, at operation S910, the network entity may be configured to receive, from the UE, a message that includes AI / ML data logging capability information associated with the UE. Specifically, the AI / ML data logging capability information may include at least one of: (a) a parameter (e.g., "bufferSizeCap", etc.) indicating a buffer size or capacity of the UE for logging AI / ML data, or (b) a parameter indicating that the UE supports minimum dedicated AS-layer memory for logging AI / ML data (e.g., greater than or equal to 32 kB, etc.).
[0198] According to example embodiments, the buffer size capacity parameter (e.g., "bufferSizeCap") may indicate the maximum buffer size that the UE may dedicate to storing AI / ML data during a logging session. In this regard, different UEs may have different buffer capacities depending on their hardware configurations, device categories, or chipset capabilities. For instance, a low-end UE (e.g., an loT sensor, a wearable device, etc.) may have a smaller buffer capacity (e.g., 32 kB) compared to a high-end UE (e.g., a smartphone, a tablet, etc.) that may have a larger buffer capacity (e.g., 256 kB or more). By enabling the UE to declare its actual buffer size capacity to the network entity, example embodiments may allow the network entity to dynamically adjust the buffer threshold based on the UE's actual buffer capacity, rather than relying on a fixed buffer threshold that may not be suitable for all UEs.
[0199] According to example embodiments, the minimum dedicated AS-layer memory parameter may indicate that the UE supports a minimum baseline of dedicated memory for AVML data. In this regard, the minimum baseline (e.g., 32 kB) may guarantee that any configured buffer threshold at or below the baseline is supportable by the UE. By establishing a minimum baseline requirement, example embodiments may ensure consistent data logging behavior across different UE implementations, as all UEs that support AI / ML data logging may be required to dedicate at least the minimum baseline of memory for AI / ML data.
[0200] According to example embodiments, the message (that may be received by the network entity from the UE at operation S910) may include an RRC message. For instance, the AI / ML data logging capability information may be included in a UE capability information message (e.g., "UECapabilitylnformation-AI", etc.) or any other suitable message that conveys the UE's capabilities for AI / AIL data logging.
[0201] Referring still to FIG. 9, at operation S920, the network entity may be configured to transmit, to the UE, a message that includes buffer configuration associated with the AI / ML data logging session. Specifically, the buffer configuration may include a parameter (e.g., "buffer Threshold", etc.) indicating a threshold for the AI / ML data in the UE's buffer.
[0202] According to example embodiments, the buffer threshold parameter may be signaled in absolute bytes (e.g., rather than as a percentage or a relative value). In this regard, by signaling the buffer threshold in absolute bytes, example embodiments may provide a clear and unambiguous threshold value that the UE may evaluate to determine when the buffer has reached the configured threshold. For instance, the network entity may configure a buffer threshold of 32 kB, indicating that the UE should take action (e g., report buffer status, transmit logged data, etc.) when the amount of AI / L data in the UE's buffer reaches 32 kB.
[0203] According to example embodiments, the buffer threshold parameter may be constrained to a percentage range (e.g., 50 to 90 percent) of the UE's declared buffer size capacity parameter (e.g., "bufferSizeCap"). In this regard, the UE may clamp the buffer threshold to the constrained range. For instance, if the UE declares a buffer size capacity of 64 kB, the buffer threshold may be constrained to a range of 32 kB (50% of 64 kB) to 57.6 kB (90% of 64 kB). By constraining the buffer threshold to a reasonable range relative to the UE's declared buffer capacity, example embodiments may prevent suboptimal settings that may occur in the related art. Specifically, constraining the buffer threshold may prevent scenarios where: (a) low-capacity UEs experience frequent buffer overflows due to a buffer threshold that exceeds the UE's total buffer size, and (b) high-capacity UEs experience delayed data retrieval due to a buffer threshold that is never reached.
[0204] According to example embodiments, the network entity may dynamically adjust the buffer threshold based on the UE's declared buffer size capacity, one or more network conditions (e.g., network load, congestion level, etc.), and / or one or more UE’s conditions (e.g., power status, operational mode, etc.). In this regard, by dynamically adjusting the buffer threshold, example embodiments may maintain uniform data logging behavior across different UEs with varying buffer capacities, and may simplify cross-vendor configuration templates for AI / ML data logging.
[0205] According to example embodiments, the message (that may be transmitted by the network entity to the UE at operation S920) may include an RRC message. In this regard, the buffer configuration may be included under an IE of the RRC message (e.g., "AIML-LoggingConfig", etc.). Alternatively or additionally, the message may include a MAC CE. In this regard, the buffer configuration may be implemented as a payload of the MAC CE that defines the intended configuration.
[0206] In view of the above, the method and operations described herein introduce and exemplify a mechanism that may be implemented by a network entity to receive AI / ML data logging capability information associated with the UE (e.g., buffer size capacity, minimum dedicated AS-layer memory support, etc.) and configure buffer thresholds that are constrained to a reasonable range relative to the UE's declared buffer capacity, thereby efficiently and effectively addressing the shortcomings of aspect (7) as described above.
[0207] Advantageously, by implementing the method and operations described herein, the network entity may effectively and efficiently receive the UE's actual buffer size capacity and minimum dedicated AS-layer memory support, dynamically adjust the buffer threshold based on the UE's declared buffer capacity, and constrain the buffer threshold to a reasonable range (e.g.,50-90% of the UE's declared buffer capacity) to prevent suboptimal settings. Accordingly, example embodiments may prevent low-capacity UEs from experiencing frequent buffer overflows and high-capacity UEs from experiencing delayed data retrieval, maintain uniform data logging behavior across different UEs with varying buffer capacities, and simplify cross-vendor configuration templates for AI / ML data logging, thereby addressing the shortcomings of aspect (7) as described above.> > Example Methods and Operations associated with Aspect (8)
[0208] FIG. 10 illustrates an example method 1000, according to one or more example embodiments. The operations in method 1000 may be performed by a network entity (e.g., network entity 110) to communicate with a UE (e.g., UE 120) to configure a combined trigger mask for an AI / ML data logging session for NW-side AI / ML purposes.
[0209] As illustrated in FIG. 10, at operation S 1010, the network entity may be configured to transmit, to the UE, a message that includes a trigger mask configuration associated with an AI / ML data logging session. Specifically, the trigger mask configuration may include a bit string (e.g., "triggerMask BIT STRING SIZE X", etc.), wherein each bit of the bit string indicates a respective condition for triggering the AI / ML data logging session.
[0210] According to example embodiments, the trigger mask configuration may include a 4-bit string, wherein each bit corresponds to a respective trigger condition. For instance, the bit assignments may include: bit 0 corresponding to periodic logging, bit 1 corresponding to a first network event (e.g., XI event, etc.), bit 2 corresponding to a second network event (e.g., X2 event, etc.), and bit 3 corresponding to a low power condition at the UE. It is contemplated that the trigger conditions described herein (e.g., XI event, X2 event, etc.) may be similar to the A1 / A2 derivative events described above with reference to FIG. 2 (e.g., the event identifier field that maps to A1 / A2derivative events such as XI, X2, etc ), and thus it can be understood that the descriptions of the A1 / A2 derivative events with reference to FIG. 2 may be applicable herein, and redundant descriptions associated therewith may be omitted for conciseness.
[0211] According to example embodiments, the message may include instructions to guide the UE to trigger the AI / ML data logging session when any of the enabled bit conditions is satisfied. In this regard, the UE may evaluate each enabled bit condition independently, and may trigger the AI / ML data logging session when at least one of the enabled bit conditions is satisfied. For instance, if only bit 0 (periodic) is enabled, the UE may be instructed to log AI / ML data only on periodic intervals. If bit 0 (periodic) and bit 1 (XI event) are enabled, the UE may be instructed to log AI / ML data when either the periodic interval is reached or the XI event occurs. If all four bits are enabled, the UE may be instructed to log AI / ML data when any of the four conditions (periodic, XI event, X2 event, low power) is satisfied.
[0212] According to example embodiments, the trigger mask configuration may enable the network entity to configure multiple trigger conditions simultaneously for a single logging session, or to configure one trigger condition to suppress another trigger condition. For instance, in some deployment scenarios, the network entity may enable both periodic and event-based triggers simultaneously to capture comprehensive AI / ML data that reflects both time-based patterns and event-driven radio conditions. Conversely, in other deployment scenarios, the network entity may enable only event-based triggers (e.g., XI and X2 events) while disabling periodic logging, in order to reduce redundant data logging and focus on event-driven data collection. In this regard, by providing a unified trigger mask, example embodiments may enable the network entity to flexibly configure any combination of trigger conditions per logging session using a single configuration field, rather than configuring each trigger condition separately (which may increasesignaling overhead and may lead to ambiguity regarding how the UE should behave when multiple trigger conditions are configured).
[0213] According to example embodiments, the trigger mask configuration may be included under an IE of an RRC message (e.g., "LoggingTriggerConfig" under "AIML-LoggingConfig", etc.). It is contemplated that the trigger mask configuration described herein may be included as part of the logging trigger configuration described above with reference to FIG. 2, and thus it can be understood that the descriptions of the logging trigger configuration with reference to FIG. 2 may be applicable herein, and redundant descriptions associated therewith may be omitted for conciseness. Alternatively or additionally, the trigger mask configuration may be implemented as a payload of a MAC CE that defines the intended configuration.
[0214] Referring still to FIG. 10, at operation SI 020, the network entity may be configured to receive, from the UE, a message that includes AI / ML data collected based on the trigger mask configuration. In this regard, the UE may collect or log data based on the enabled trigger conditions specified in the trigger mask configuration, and may transmit the collected data to the network entity for NW-side AI / ML purposes. According to example embodiments, the message may include an MDT measurement report (e.g., LoggedMeasurementReport, etc.), a response message (e g., UEInformationResponse, etc.), or any other suitable message that conveys the collected AI / ML data to the network entity.
[0215] In view of the above, the method and operations described herein introduce and exemplify a mechanism that may be implemented by a network entity to configure a unified trigger mask that enables the network entity to enable or disable multiple trigger conditions per logging session using a single configuration field, thereby efficiently and effectively addressing the shortcomings of aspect (8) as described above.
[0216] Advantageously, by implementing the method and operations described herein, the network entity may effectively and efficiently configure multiple trigger conditions per logging session using a unified trigger mask, support flexible combinations of trigger conditions (e.g., periodic logging and event-based logging simultaneously, or one trigger condition suppressing another), reduce signaling overhead by consolidating trigger configurations into a single bit string field, and ensure consistent data logging behavior across different UE implementations by providing clear instructions on how the UE should evaluate multiple trigger conditions (e.g., log when any enabled condition is satisfied). Accordingly, example embodiments may enable flexible and efficient trigger configuration for AI / ML data logging, thereby addressing the shortcomings of aspect (8) as described above.>> Summary
[0217] In view of the above, FIG. 2 to FIG. 10 illustrate various example methods and operations that may be implemented to address the shortcomings of the above-described aspects (1 )-(8), thereby enabling effective and efficient optimization on NW-side A 1 / M I. data management.
[0218] It is contemplated that the methods, operations, advantages, and significances described above with reference to FIG. 2 to FIG. 10 are merely examples, and the scope of the present disclosure should not be limited thereto. Specifically, one or more operations in FIG. 2 to FIG. 10 may be performed differently, less or additional operations may be involved, the messages or commands involved therein may include less or additional parameters, additional technical advantages may be achieved, and the like, without departing from the scope of the present disclosure. Further, it is contemplated that one or more of the above-described parameters, messages, and the like, may be consistent with those specified in one or more standard specifications, with additional parameters, attributes, and mechanisms supplemented thereto(when required). Further, multiple methods and operations in FIG. 2 to FIG. 10 may be performed in combination with each other, without departing from the scope of the present disclosure.Examples of Device
[0219] One or more components of the example embodiments (e.g., network entity, UE, etc.), as well as the operations associated therewith, may be implemented in one or more devices or hardware components. For instance, one or more components / operations of the network entity may be implemented in one or more devices like a server(s), and the like.
[0220] In the following, descriptions of a device in which the example embodiments may be implemented are provided. It is contemplated that one or more features, operations, and methods described above may be performed by the device. For instance, the one or more operations or methods associated with the network entity may be performed by at least one processor of the device upon executing machine-readable instructions or computer-readable instructions stored in a memory or a storage component of the device.
[0221] FIG. 11 illustrates an embodiment of a device 1100. As shown in FIG. 11, the device 1100 may include a processor 1110, a memory 1120, a storage component 1130, an input component 1140, an output component 1150, a communication interface 1160, and a bus 1170.
[0222] The processor 1110, as used herein, means any type of computational circuit that may comprise hardware elements and software elements. The processor 1110 may be embodied as a multi-core processor, a single core processor, or a combination of one or more multi-core processors and / or one or more single core processors, a distributed processing system, or the like. The processor 1110 may be a Central Processing Unit (CPU), a graphics processing unit (GPU), an accelerated processing unit (APU), an application-specific integrated circuit (ASIC), or another type of processing component.
[0223] Memory 1120 includes a non-transitory computer readable medium. Memory 1120 includes a random-access memory (RAM), a read only memory (ROM), and / or another type of dynamic or static storage device (e.g., a flash memory, a magnetic memory, and / or an optical memory) that stores information and / or instructions for use by processor 1110. The memory 1120 comprises machine-readable instructions which are executable by the processor 1110. These machine-readable instructions when executed by the processor 1110 cause the processor 1110 to perform one or more method steps of an embodiment described above.
[0224] Storage component 1130 stores information and / or software related to the operation and use of the device 1100. For example, storage component 1130 may include a hard disk (e.g., a magnetic disk, an optical disk, a magneto-optic disk, and / or a solid-state disk), a compact disc (CD), a digital versatile disc (DVD), a floppy disk, a cartridge, a magnetic tape, and / or another type of non-transitory computer-readable medium, along with a corresponding drive.
[0225] Input component 1140 is configured to receive information, such as user input. For example, the input component 1140 may include, but not be limited to, a touch screen display, a keyboard, a keypad, a mouse, a button, a switch, and / or a microphone. Additionally, or alternatively, the input component 1140 may include a sensor for sensing information (e.g., a global positioning system (GPS), an accelerometer, a gyroscope, and / or an actuator).
[0226] Output component 1150 is configured to provide output information from the device 1100. For example, the output component 1150 may be, but not limited to, a display, a speaker, instructions to an external device, and / or one or more light-emitting diodes (LEDs).
[0227] Communication interface 1160 is an interface that provides a communication connection to other devices, such as external devices and internal devices. The connection by the communication interface 1160 can be a wired connection, a wireless connection, or a combinationof wired and wireless connections, and can be a direct connection or an indirect connection via a communication network that exists between the device 1100 and other devices. In other words, the standard of the communication interface 1160 is not limited.
[0228] The bus 1170 acts as an interconnect between the processor 1110, the memory 1120, the storage component 1130, the input component 1140, the output component 1150, and the communication interface 1160 of the device 1100. The bus 1170 may include a wired interconnection or a wireless interconnection.
[0229] The number and arrangement of components shown in FIG. 11 are provided as an example. In practice, device 1100 may include additional components, fewer components, different components, or differently arranged components than those shown in FIG. 11. Additionally, or alternatively, a set of components (e.g., one or more components) of device 1100 may perform one or more functions described as being performed by another set of components of device 1100. Further, one or more method steps described in any of the embodiments may be performed utilizing a plurality of devices 1100 in communication with one another.Example Implementation Environment
[0230] Example embodiments of the present disclosure may be implemented in any suitable type of environment. In the following, an example environment (in which the example embodiments may be implemented) is described.
[0231] FIG. 12 illustrates a diagram of an example environment 1200 in which systems and / or methods, described herein, may be implemented. The implementation environment 1200 includes a UE (User equipment) 1210, a service environment 1220, and a network 1230. The service environment 1220 includes one or more sub-environments 1221. To illustrate this, FIG. 12shows, for convenience, examples of a 1 st sub-environment 1221 -1 , a 2nd sub-environment 1221 - 2, and an N-th sub-environment 1221-N (where N is any natural number).
[0232] The UE 1210 is connected to the network 1230, and the network 1230 is connected to the service environment 1220. The connections may be wired, wireless, or a combination of both wired and wireless. The UE 1210 and the service environment 1220 are connected via the network 1230.
[0233] The UE 1210 is a device that communicates with the service environment 1220. The UE 1210 receives information from the service environment 1220 and / or sends information to the service environment 1220. Also, the UE 1210 may generate and / or store information to be transmitted, as necessary. Also, the UE 1210 may store and / or process information that is received, as necessary.
[0234] The example FIG. 12 refers to the “UE”. However, it should be understood by those skilled in the art that general terms such as “user device,” “terminal,” “terminal device,” “communication device,” and “communication terminal” can be used interchangeably with the term “UE.”
[0235] For example, the UE 1210 may include a computing device (e.g., a desktop computer, a laptop computer, a tablet computer, a handheld computer, a smart speaker, a server, etc.), a mobile phone (e.g., a smart phone, a radiotelephone, etc.), a wearable device (e.g., a pair of smart glasses or a smart watch), or a similar device.
[0236] The service environment 1220 is an environment that communicates with the UE 1210 to provide one or more services. The service environment 1220 receives information from the UE 1210 and / or sends information to the UE 1210. Also, the service environment 1220 may generate and / or store information to be transmitted, as necessary. Also, the service environment1220 may store and / or process information that is received, as necessary. For example, the service environment 1220 may provide computing resources as one of the services. It should be noted that the service is not limited to being provided to the UE; it may also be provided to devices other than the UE. For example, based on communication from the UE, the service may perform processes such as anomaly detection or traffic analysis and notify the results to a predetermined destination.
[0237] The example FIG. 12 refers to the “service environment”. The term "service environment" is used to refer to the broader context within which services operate. For example, cloud environments, platforms, computing systems, network systems, and cloud systems generally represent the environments in which services are conducted, and these are included within the "service environment." However, the "service environment" is not limited to these examples. Additionally, the specific types of environments within the "service environment" are not restricted. For instance, cloud environments and cloud systems can be categorized as private cloud, public cloud, hybrid cloud, or multi-cloud, all of which are included within the "service environment."
[0238] The one or more services provided by the service environment 1220 is not specifically limited and can be adjusted according to the embodiments. For example, the services may include a service that provides information to the UE 1210, a service that stores information from the UE 1210, or a service that performs processing based on information from the UE 1210 and returns the results of the processing.
[0239] In an embodiment, the Service Environments 1220 may also provide computing resources as the service. The computing resources can be hardware resources and / or software resources. For example, applications, processors, memory, and storage can be included in the provided computing resources. Each computing resource can communicate with other computingresources via wired connections, wireless connections, or a combination of wired and wireless connections.
[0240] The provided computing resources can be actual resources (also referred to as physical resources) and / or virtual resources. Furthermore, means of virtualization for virtual resources can be selected as appropriate. That is, in this disclosure, the use of adjectives such as "Virtual" or "Virtualized" to describe names does not imply that they are virtualized by a specific means of virtualization. For example, “virtual machine” refers to software that operates like an actual computer, realized through means of virtualization, and it is not intended to exclude those realized by specific means of virtualization such as Hypervisors or Containers. Conversely, when means of virtualization such as Hypervisors or containers are mentioned in this disclosure, it is merely cited as a general method of implementation. It should also be interpreted that embodiments implemented with other virtualization means are also disclosed. Also, the services may also be provided using resources virtualized by different means.
[0241] The service environment 1220 includes one or more devices, such as servers and network devices, which provide services or perform processes. The placement of these devices within the service environment 1220 can be determined as appropriate. Additionally, if the service environment 1220 includes one or more sub-environments 1221, the placement of devices can be determined based on predetermined policies for each sub-environment 1221. For example, devices related to the first service may be placed in the 1st sub -environment 1221-1, and devices related to the second service may be placed in the 2nd sub-environment 1221-2. In another example, devices expected to have a higher load than a predetermined threshold may be placed in the 1st sub-environment 1221-1, while devices expected to have a lower load than the predetermined threshold may be placed in the 2nd sub-environment 1221-2. In this way, specific devices can beplaced in specific sub-environments 1221. Conversely, each sub -environment 1221 can be specialized for a particular purpose.
[0242] In an embodiment, all processes executed in a single service may run within a single service environment, or in multiple service environments. Multiple processes executed in a single service could be provided by different service environments.
[0243] The network 1230 is a network that exchanges information between the UE 1210 and the service environment 1220. The network 1230 includes one or more wired and / or wireless networks.
[0244] For example, the network 1230 may include a cellular network (e.g., a fifth generation (5G) network, a long-term evolution (LTE) network, a third generation (3G) network, a code division multiple access (CDMA) network, etc.), a public land mobile network (PLMN), a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a telephone network (e.g., the Public Switched Telephone Network (PSTN)), a private network, an ad hoc network, an intranet, the Internet, a fiber optic-based network, or the like, a non-terrestrial network (NTN), and / or a combination of these or other types of networks.
[0245] The network 1230 can be a part of a network. For example, in a 5G network that includes a RAN, a transport network, and a core network, the network 1230 can be at least one of the RAN, the transport network, or the core network. For example, the service environment 1220 could be in the core network, in which case the network 1230 could correspond to a network that is a combination of a RAN and a transport network and is part of the 5G network.
[0246] In this regard, the UE 1210 in FIG. 12 may be similar to the UE 120 in FIG. 1, while the network entity 110 in FIG. 1 may be implemented in the network 1230. Further, theAI / ML-related operations described herein (e.g., AI / ML model training, AI / ML inference, etc.) may be implemented in the service environment 1220.
[0247] The number and arrangement of devices and networks shown in FIG. 12 are provided as an example. It should be understood that any changes that may be implemented by those skilled in the art, such as the addition or rearrangement of well-known devices or networks at the time of implementation, are included in this disclosure.Various Aspects of Embodiments
[0248] Example embodiments introduce new parameters, attributes, mechanisms, and features that supplement and enhance the disclosures of one or more standard specifications. As a non-limiting example, example embodiments supplement and enhance at least one technical documents associated with 3 GPP Technical Specification Group-Radio Access Network (TSG-RAN) Working Group 2 (WG2). Specifically, the following disclosures were introduced by the example embodiments of the present disclosure during the 3GPP TSG-RAN WG2 Meeting #130: 2.1 Logging Configuration & TriggeringRel-19 permits two trigger modes:(i) periodic logging and(ii) L3 radio-condition events derived from A1 / A2 other L3 events.The draft CR still have these parameters across several IES and lacks a container that also lets the network set an independent logginginterval to save UE power on short-period RS.This was agreed in RAN2#127bis and confirmed in RAN2#129However, the logging interval parameter remains an FFS item in multiple placesEach active logging session is already identified by logConf iglD; a boolean flag per ID fulfils granularity. A timer (similar to Logged-MDT’s 48 h rule) fulfils lifetime control with negligible extra signalling.Observation 1No single IE configures an A1 / A2 -based trigger together with an optional logging interval.Proposal 1Introduce LoggingTriggerConfig under AIML-LoggingConf ig in TS 38.331Field Type / Range NotestriggerType ENUM {13Condition(0)} Only L3 events in Rel-19 scopeeventld INTEGER(0..31) Maps to A1 / A2 derivatives (e.g., XI, X2) timeToTrigger INTEGER(0. 500) ms Same scale as RRM TTTlogginginterval OPTIONAL INTEGER) 1. 1024) ms If absent —> log per RS period2.2 Data-Availability & Constraint ReportingAgreed causes for availability signalling are bufferThresholdReached, bufferFull, and lowPower. A single Boolean flag cannot represent them.Observation 2-1The CR draft offers only one flag; the gNB cannot prioritise urgent low-power cases over routine buffer-threshold events.Proposal 2-1Add structured element in UEAssistancelnformation / UAI:DataAvailabilitylndication : : = SEQUENCE {cause ENUMERATED {bufferThresholdReached ( 0 ) ,bufferFull ( 1 ) ,lowPower (2 )} ,remainingBytes INTEGER ( 0. . 262143) OPTIONAL}Discussion 2-2RAN-2 bis captured diverging operator requirements:• Network-controlled thresholds — deterministic.• UE-autonomous heuristics.The current draft CR lacks a mechanism to toggle between these, forcing vendors to choose one or other.Observation 2-2No IE allows the network to select between nw-controlled and UE-autonomous availability-trigger logic.Proposal 2-2Add field availabilityTriggerMode to AIML-LoggingConf ig ill TS 38.331:Field Type Default Semantics ENUM { Decides which entity availabilityTriggerMode nwControlled(O), nwControlled evaluates buffer & ueAutonomous(l) } power thresholdsDiscussion 2-3If a UE silently stops logging due to low memory or critical battery, the network sees unexplained data holes that bias ML training. Chair notes (128 §6) requested an explicit “reason code”.Observation 2-3The spec provides only an on / off availability flag; it cannot convey why logging paused.Proposal 2-3Extend UEAssistancelnformation and UAI with:Field Type / Size Bit Mapping resourceConstraintCause BIT STRING (SIZE 3) bO memLow, bl powerLow, b2 cpuBusy2.3 Retention & State I Mobility HandlingNW-side training data collection may run for hours or days, so partially collected logs may span multiple cell changes, DRX cycles, or even RAT transitions. Without a deterministic rule set we can see three different behaviours:• Discard logs whenever a handover command arrives, guaranteeing fresh context for the target cell but losing any unsent data.• Keep logs by default, if the target gNB is from a different vendor, those logs may never be fetched because the new node lacks decoding knowledge.• Indetenninate devices — particularly dual-mode chipsets may retain data in some cases (intraFreq HO) and flush in others (Idle, Inactive, RLF), hence missing data in the time series.RAN-2 resolved the principle by agreeing on a one-bit directive set by the source gNB. Yet two critical gaps remain:1. Signalling location -Placing the bit inside the HO-Command (i.e. RRCReconfiguration from the target cell) is difficult : if the source and target are from different vendors, the proprietary container may be ignored. Pre-HO delivery from the source gNB will enhance interoperability.2. Lifetime management - If retained logs are not fetched for long intervals (e.g., UE soon enters Idle or executes inter-RAT HO), they must self-expire to satisfy privacy policies and free memory'. MDT use upper bound timer similar to MDT.A unifonn rule set guarantees that:UEs do not hoard stale data indefinitely, limiting privacy exposure and memory pressure.Observation 3Without a deterministic location and expiry timer, vendors differ on when logs are purged.Proposal 3• 3-1 Add Boolean retainLoggedData in source-cell RRCReconfiguration sent before HO execution. Include retentionTimer (24-168 h, default 48 h) inside AIML-LoggingConf ig. UE discards retained logs when the timer expires or after HO / RRC IDLE / RLF.Observation 3-2With concurrent configs (beam-AI + CSI-AI), gNB cannot separate datasets unless each entry declares its origin. Several companies flagged this in RRClogMeasEntry and aiLogMeasEntry lack a session tag.Proposal 3-2Prepend 5-bit logConfiglD to both entry structures:aiLogMeasEntry-v!9aO : : = SEQUENCE {logConfiglD INTEGER ( 0 . . 31 ) ,. . . — existing fields}2.4 Data Format & MetadataNetwork may aggregate logs from millions of UEs, and data may arrive out of order due to HARQ retransmissions, DRX sleep, or multi-SIM split UL. Current MDT messages rely on context in the enclosing RRC / PDU: if segmentation occurs, that context can be lost. For NW-side Al training data collection is important.Observation 4Message-level context alone cannot reassemble interleaved segments.Proposal 4Add fixed header to every UElnformationResp se segment:AIMLLogHeader : : = SEQUENCE {logConfigID INTEGER ( 0 . . 15 ) ,segmentNo INTEGER ( 0 . . 255 ) ,lastSegment BOOLEAN,logPriority ENUM { normal , high } ,moreSegmentsPending BOOLEAN}Add cellld, gaplnd, and optional entrypriority to each LoggedMeasurementEntry.2.5 Transport Bearer & SRB UseAIML logs and MDT must share SRB-4 without starving each other, assigning a new LCG to AIML traffic lets MAC allocate fewer bytes per TTI, throttle bursts during congestion, and still rely on SRB- 4’s security context. Coupled with a moreSegmentsPending flag, the scheduler gains explicit knowledge of burst boundaries, enabling fair-queue or weighted deficit round-robin without redefining another SRB.Observation 5Single LCG on SRB-4 can create head-of-line blocking.Proposal 5Map AI / ML log PDUs on SRB-4 to new logicalChannelGroup, ( aiModelData ) ; MDT remains on its existing LCG. Scheduler may throttle using moreSegmentsPending.When moreSegmentsPending=l UE continues using SRB-4 / aiModelData-LCG until all segments sent2.6 Cross-layer RS ReferencingRAN-l defines CSI-RcportConfig IDs and their associated RS-RcsourccSct lists. Some companies want logging to follow exactly the same RS pool used for inference (to guarantee feature alignment in model training). Others, custom RS sets (e.g., occasional SS / PBCH blocks) that are not part of CSI inference. Embedding RS details in RRC every time duplicates RAN-l ASN.l and risks version skew; but forcing all logging to tie to a CSI-ReportConfig ID will not work with those custom cases.Observation 6Spec provides no mechanism to reference or inherit RS sets.Proposal 6Add optional linkedCSI-ReportConfigld (INTEGER(0..255)) inside AIML-LoggingConfig. If present, UE inherits RS definitions from that CSI-ReportConfig; else it uses explicit RS IDs.Liaison RAN-l request: expose stable rsSetid in CSI-ReportConfig ASN.l.2.7 Memory I Buffer LimitsOperators configuring a fixed bufferThreshold (e.g., 50 kB) may experience opposite issues.• Threshold exceeds total buffer on small devices UE hits bufferFull constantly, causing signalling storms and data truncation.• Threshold never reached on large devices — > network delays fetching; old logs occupy memory and are overwritten when new7high-value events occur.An MDT-style minimum baseline (32 kB) guarantees that any threshold <32 kB is always supportable. Exposing the actual bufferSizeCap in UE capabilities lets 0AM or self-optimising SON adjust thresholds dynamically (e.g., 60 % cap). Constraining bufferThreshold to 50-90 % of bufferSizeCap prevents pathological settings, maintains uniform trigger cadence, and simplifies cross-vendor configuration templates.Observation 7Without these, low-cap UEs over-trigger and high-cap UEs under-trigger.Proposal 7• 7-1- Require > 32 kB dedicated AS-layer memory for Al logs; UE declares buffersizecap in UECapabilitylnf ormation-AI .• 7-2 Signal buf ferThreshold in absolute bytes; UE clamps to 50-90 % of declared capacity.2.8 Combined Trigger-Mask FieldSome deployments need periodic + event (X1 / X2) logging simultaneously; others want one to suppress the other.Observation 8No unified mask lets the network enable / disable multiple triggers per session.Proposal 8Inside LoggingTriggerConfig, add triggerMask BIT STRING (SIZE ( 4 ) ) — bits {periodic, XI, X2, lowPower}.UE logs when any enabled bit condition is satisfied.3. ConclusionIn this contribution, the observations and proposal are summarized as follows:Logging Configuration & TriggeringObservation 1No single IE configures an A1 / A2 -based trigger together with an optional logging interval. Proposal 1Introduce LoggingTriggerConfig under AIML-LoggingConf ig in TS 38.331Field Type / Range NotestriggerType ENUM {13Condition(0)} Only L3 events in Rel-19 scopeeventld INTEGER(0..31) Maps to A1 / A2 derivatives (e.g., XI, X2) timeToTrigger INTEGER(O..5OO) ms Same scale as RRM TTTlogginginterval OPTIONAL INTEGER! 1..1024) ms If absent log per RS periodData-Availabilitv & Constraint ReportingObservation 2-1The draft offers only one flag; the gNB cannot prioritise urgent low-power cases over routine bufferthreshold events.Proposal 2-1Add structured element in UEAssistancelnformation / UAI:DataAvailabilitylndication : : = SEQUENCE {cause ENUMERATED {buf f erThresholdReached ( 0 ) ,bufferFull ( 1 ) ,lowPower (2 )1,remainingBytes INTEGER ( 0. . 262143) OPTIONAL1Observation 2-2No IE allows the network to select between nw-controlled and UE-autonomous availability-trigger logic.Proposal 2-2Add field availabilityTriggerMode to AIML-LoggingConf ig in TS 38.331:Field Type DefaultENUM { Decides which entity availabilityTriggerMode nwControlled(O), nwControlled evaluates buffer & ueAutonomous(l) } power thresholdsObservation 2-3The spec provides only an on / off availability flag; it cannot convey why logging paused.Proposal 2-3Extend UEAssistancelnformation and UAI with:Field Type / Size Bit Mapping resourceConstraintCause BIT STRING (SIZE 3) b0 memLow, bl powerLow, b2 cpuBusyRetention & State / Mobility HandlingObservation 3Without a deterministic location and expiry timer, vendors differ on when logs are purged.Proposal 3• 3-1 Add Boolean retainLoggedData in source-cell RRCReconf iguration sent before HO execution. Include retentionTimer (24-168 h, default 48 h) inside AlML-LoggingConf ig. UE discards retained logs when the timer expires or after HO / RRC IDLE / RLF.Observation 3-alogMeasEntry and aiLogMeasEntry lack a session tag.Proposal 3-aPrepend 5-bit logConfigID to both entry structures:aiLogMeasEntry-v! 9a O : : = SEQUENCE {logConfigID INTEGER ( 0 . . 31 ) ,. . . — exi sting fields}Data Format & MetadataObservation 4Message-level context alone cannot reassemble interleaved segments.Proposal 4Add fixed header to every UEinformationResponse segment:AIMLLogHeader : : = SEQUENCE {logConfigID INTEGER ( 0 . . 15 ) ,segmentNo INTEGER ( 0 . . 255 ) ,lastSegment BOOLEAN,logPriority ENUM { normal , high } ,moreSegments Pending BOOLEAN}Add cellld, gaplnd, and optional entrypriority to each LoggedMeasurementEntry.Transport Bearer & SRB UseObservation 5Single LCG on SRB-4 can create hcad-of-linc blocking.Proposal 5Map AI / ML log PDUs on SRB-4 to new logicalChannelGroup, ( aiModelData ) ; MDT remains on its existing LCG. Scheduler may throttle using moreSegmentsPending.When moreSegmentsPending=l UE continues using SRB-4 / aiModelData-LCG until all segments sentCross-layer RS ReferencingObservation 6Spec provides no mechanism to reference or inherit RS sets.Proposal 6Add optional linkedCSI-ReportConfigld (INTEGER(0..255)) inside AIML-LoggingConfig. If present, UE inherits RS definitions from that CSI-ReportConfig; else it uses explicit RS IDs.Proposal 6-aLiaison RAN-1 request: expose stable rsSetid in CSI-ReportConfig ASN.l.Memory / Buffer LimitsObservation 7A method needed to Distinguish, low-cap UEs over-trigger and high-cap UEs under-trigger.Proposal 7• 7-1- Require > 32 kB dedicated AS-layer memory for Al logs; UE declares buffersizecap in UECapabilitylnf ormation-AI .• 7-2 Signal buf f erThreshold in absolute bytes; UE clamps to 50-90 % of declared capacity.Combined Trigger-Mask FieldObservation 8No unified mask lets the network enable / disable multiple triggers per session.Proposal 8Inside LoggingTriggerConfig, add triggerMask BIT STRING ( SIZE ( 4 ) ) — bits {periodic. XI, X2, lowPowcr}.UE logs when any enabled bit condition is satisfied.
[0249] In view of the above, example embodiments introduce clarified, specified, and standardized approaches for enhancing the management of NW-side AI / ML data in the 3GPP-based network architecture. Specifically, example embodiments clarify how specifically the 3GPP defined frameworks and architectures (e.g., MDT, RRC signaling, MAC CE signaling, etc.) can be utilized to optimize the management of the NW-side AI / ML data. Accordingly, the example embodiments may be implemented in the 3GPP -based networks in a clear and standardized manner.
[0250] It is contemplated that features, advantages, and significances of example embodiments described hereinabove are merely examples of the present disclosure, and are not intended to be exhaustive or to limit the scope of the present disclosure.
[0251] Specifically, the foregoing disclosure provides illustration and description, but is not intended to be exhaustive or to limit the implementations to the precise form disclosed. Modifications and variations are possible in light of the above disclosure or may be acquired from practice of the implementations.
[0252] Some embodiments may relate to a device, a system, a method, and / or a computer-readable medium at any possible technical detail level of integration. Further, one or more of the above components described above may be implemented as instructions stored on a computer-readable medium and executable by at least one processor (and / or may include at least one processor). The computer-readable medium may include a computer-readable non-transitory storage medium (or media) having computer-readable program instructions thereon for causing a processor to carry out operations.
[0253] The computer-readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer-readable storage medium may be, for example, but is not limited to, an electronic storage device, a magnetic storagedevice, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non-exhaustive list of more specific examples of the computer-readable storage medium includes the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), electrically erasable programmable read-only memory (EEPROM), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, and any suitable combination of the foregoing. A computer-readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire.
[0254] Computer-readable program instructions described herein can be downloaded to respective computing / processing devices from a computer-readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and / or a wireless network. The network may comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium within the respective computing / processing device.
[0255] Computer-readable program code / instructions for carrying out operations may beassembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state-setting data, configuration data for integrated circuitry, or either source code or object code written in any combination of one or more programming languages, including an object-oriented programming language such as Smalltalk, C++, or the like, and procedural programming languages, such as the "C" programming language or similar programming languages.
[0256] The computer-readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer-readable program instructions by utilizing state information of the computer-readable program instructions to personalize the electronic circuitry, in order to perform aspects or operations.
[0257] These computer-readable program instructions may be provided to a processor of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer-readable program instructions may also be stored in a computer-readable storage medium that candirect a computer, a programmable data processing apparatus, and / or other devices to function in a particular manner, such that the computer-readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function / act specified in the flowchart and / or block diagram block or blocks.
[0258] The computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer-implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions / acts specified in the flowchart and / or block diagram block or blocks.
[0259] The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer-readable media according to various embodiments. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). The method, computer system, and computer-readable medium may include additional blocks, fewer blocks, different blocks, or differently arranged blocks than those depicted in the Figures. In some alternative implementations, the functions noted in the blocks may occur out of the order noted in the Figures. For example, two blocks shown in succession may, in fact, be executed concurrently or substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flowchart illustration, and combinations of blocks in the block diagrams and / or flowchart illustration, can be implemented by special purpose hardware-based systems thatperform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions.
[0260] It will be apparent that systems and / or methods, described herein, may be implemented in different forms of hardware, firmware, or a combination of hardware and software. The actual specialized control hardware or software code used to implement these systems and / or methods is not limited to the implementations. Thus, the operation and behavior of the systems and / or methods were described herein without reference to specific software code — it is understood that software and hardware may be designed to implement the systems and / or methods based on the description herein.
[0261] In view of the above, various further respective aspects and features of embodiments of the present disclosure may be defined by the following items:Item [1]: A system comprising: a network entity configured to: transmit, to a user equipment (UE), a message comprising a logging trigger configuration, wherein the logging trigger configuration is associated with a logging session for collecting data associated with an Artificial Intelligence or Machine Learning (AI / ML) model, wherein the logging trigger configuration comprises at least one of: a trigger type field indicating a type of condition that triggers the logging session, an event identifier (ID) field indicating an event that triggers the logging session, a time to trigger field indicating a time duration for triggering the logging session, and a logging interval field indicating an interval for performing the logging session; and receive, from the UE, a message comprising data collected by the UE according to the logging trigger configuration.Item [2]: The system according to item [1], wherein the trigger type field indicates a layer 3 (L3) network event that triggers the logging session.Item [3]: The system according to one or more of items
[0001] -[2], wherein the event ID field indicates a mapping between an ID and at least one of: an Al derivative event or an A2 derivative event.Item [4]: The system according to one or more of items
[0001] -[3], wherein the time to trigger field indicates that the time duration for triggering the logging session has a same scale as a time duration for triggering a radio resource management (RRM) operation.Item [5]: The system according to one or more of items
[0001] -[4], wherein at least one of the event ID field, the time to trigger field, or the logging interval field comprises an integer.Item [6]: The system according to one or more of items
[0001] -[5], wherein the logging trigger configuration comprises at least one field associated with at least one of: the trigger type field, the event ID field, and the time to trigger field, and wherein the message further includes an instruction to instruct the LIE to collect the data per reference signal (RS) period.Item [7]: The system according to one or more of items [l]-[6], wherein the message comprises a radio resource control (RRC) message, and wherein the logging trigger configuration is included under an information element (IE) of the RRC message.Item [8]: A method comprising: transmitting, by a network entity and to a user equipment (UE), a message comprising a logging trigger configuration, wherein the logging trigger configuration is associated with a logging session for collecting data associated with an Artificial Intelligence or Machine Learning (AI / ML) model, wherein the logging trigger configuration comprises at least one of: a trigger type field indicating a type of condition that triggers the logging session, an event identifier (ID) field indicating an event that triggers the logging session, a time to trigger field indicating a time durationfor triggering the fogging session, and a fogging interval field indicating an interval for performing the logging session; and receiving, by the network entity and from the UE, a message comprising data collected by the UE according to the logging trigger configuration.Item [9]: The method according to item [8], wherein the trigger type field indicates a layer 3 (L3) network event that triggers the logging session.Item
[0010] : The method according to one or more of items [8]-[9], wherein the event ID field indicates a mapping between an ID and at least one of an Al derivative event or an A2 derivative event.Item
[0011] : The method according to one or more of items [8]-
[0010] , wherein the time to trigger field indicates that the time duration for triggering the logging session has a same scale as a time duration for triggering a radio resource management (RRM) operation.Item
[0012] : The method according to one or more of items [8]-[l 1], wherein at least one of the event ID field, the time to trigger field, or the logging interval field comprises an integer.Item
[0013] : The method according to one or more of items [8]-
[0012] , wherein the logging trigger configuration comprises at least one field associated with at least one of: the trigger type field, the event ID field, and the time to trigger field, and wherein the message further includes an instruction to instruct the UE to collect the data per reference signal (RS) period.Item
[0014] : The method according to one or more of items [8]-
[0013] , wherein the message comprises a radio resource control (RRC) message, and wherein the logging trigger configuration is included under an information element (IE) of the RRC message.Item
[0015] : A non -transitory computer-readable recording medium having recorded thereon instructions executable by a network entity to cause the network entity to perform a method comprising: transmitting, to a user equipment (UE), a message comprising a logging trigger configuration, wherein the logging trigger configuration is associated with a logging session for collecting data associated with an Artificial Intelligence or Machine Learning (AI / ML) model, wherein the logging trigger configuration comprises at least one of: a trigger type field indicating a type of condition that triggers the logging session, an event identifier (ID) field indicating an event that triggers the logging session, a time to trigger field indicating a time duration for triggering the logging session, and a logging interval field indicating an interval for performing the logging session; and receiving, from the UE, a message comprising data collected by the UE according to the logging trigger configuration.Item
[0016] : The non-transitory computer-readable recording medium according to item
[0015] , wherein the trigger type field indicates a layer 3 (L3) network event that triggers the logging session.Item
[0017] : The non-transitory computer-readable recording medium according to one or more of items
[0015] -
[0016] , wherein the event ID field indicates a mapping between an ID and at least one of: an Al derivative event or an A2 derivative event.Item
[0018] : The non-transitory computer-readable recording medium according to one or more of items
[0015] -
[0017] , wherein the time to trigger field indicates that the time duration for triggering the logging session has a same scale as a time duration for triggering a radio resource management (RRM) operation.Item
[0019] : The non-transitory computer-readable recording medium according to one or more of items
[0015] -[l 8], wherein at least one of the event ID field, the time to trigger field, or the logging interval field comprises an integer.Item
[0020] : The non-transitory computer-readable recording medium according to one or more of items
[0015] -
[0019] , wherein the logging trigger configuration comprises at least one field associated with at least one of: the trigger type field, the event ID field, and the time to trigger field, and wherein the message further includes an instruction to instruct the UE to collect the data per reference signal (RS) period.
[0262] It can be understood that numerous modifications and variations of the present disclosure are possible in light of the above teachings. It will be apparent that within the scope of the appended clauses, the present disclosures may be practiced otherwise than as specifically described herein.
Claims
What is claimed is:
1. A system comprising:a network entity configured to:transmit, to a user equipment (UE), a message comprising a logging trigger configuration, wherein the logging trigger configuration is associated with a logging session for collecting data associated with an Artificial Intelligence or Machine Learning (AI / ML) model, wherein the logging trigger configuration comprises at least one of: a trigger type field indicating a type of condition that triggers the logging session, an event identifier (ID) field indicating an event that triggers the logging session, a time to trigger field indicating a time duration for triggering the logging session, and a logging interval field indicating an interval for performing the logging session; andreceive, from the UE, a message comprising data collected by the UE according to the logging trigger configuration.
2. The system according to claim 1, wherein the trigger type field indicates a layer 3 (L3) network event that triggers the logging session.
3. The system according to claim 1, wherein the event ID field indicates a mapping between an ID and at least one of: an Al derivative event or an A2 derivative event.
4. The system according to claim 1, wherein the time to trigger field indicates that the time duration for triggering the logging session has a same scale as a time duration for triggering a radio resource management (RRM) operation.
5. The system according to claim 1, wherein at least one of the event ID field, the time to trigger field, or the logging interval comprises an integer.
6. The system according to claim 1, wherein the logging trigger configuration comprises at least one field associated with at least one of: the trigger type field, the event ID field, and the time to trigger field, and wherein the message further includes an instruction to instruct the UE to collect the data per reference signal (RS) period.
7. The system according to claim 1, wherein the message comprises a radio resource control (RRC) message, and wherein the logging trigger configuration is included under an information element (IE) of the RRC message.
8. A method comprising:transmitting, by a network entity and to a user equipment (UE), a message comprising a logging trigger configuration, wherein the logging trigger configuration is associated with a logging session for collecting data associated with an Artificial Intelligence or Machine Learning (AEML) model, wherein the logging trigger configuration comprises at least one of: a trigger type field indicating a type of condition that triggers the logging session, an event identifier (ID) field indicating an event that triggers the logging session, a time to trigger field indicating a time duration for triggering the logging session, and a logging interval field indicating an interval for performing the logging session; andreceiving, by the network entity and from the UE, a message comprising data collected by the UE according to the logging trigger configuration.
9. The method according to claim 8, wherein the trigger type field indicates a layer 3 (L3) network event that triggers the logging session.
10. The method according to claim 8, wherein the event ID field indicates a mapping between an ID and at least one of: an Al derivative event or an A2 derivative event.
11. The method according to claim 8, wherein the time to trigger field indicates that the time duration for triggering the logging session has a same scale as a time duration for triggering a radio resource management (RRM) operation.
12. The method according to claim 8, wherein at least one of the event ID field, the time to trigger field, or the logging interval comprises an integer.
13. The method according to claim 8, wherein the logging trigger configuration comprises at least one field associated with at least one of: the trigger type field, the event ID field, and the time to trigger field, and wherein the message further includes an instruction to instruct the UE to collect the data per reference signal (RS) period.
14. The method according to claim 8, wherein the message comprises a radio resource control (RRC) message, and wherein the logging trigger configuration is included under an information element (IE) of the RRC message.
15. A non-transitory computer-readable recording medium having recorded thereon instructions executable by a network entity to cause the network entity to perform a method comprising: transmitting, to a user equipment (UE), a message comprising a logging trigger configuration, wherein the logging trigger configuration is associated with a logging session for collecting data associated with an Artificial Intelligence or Machine Learning (AI / ML) model, wherein the logging trigger configuration comprises at least one of: a trigger type field indicating a type of condition that triggers the logging session, an event identifier (ID) field indicating an event that triggers the logging session, a time to trigger field indicating a time duration for triggering the logging session, and a logging interval field indicating an interval for performing the logging session; andreceiving, from the UE, a message comprising data collected by the UE according to the logging trigger configuration.
16. The non-transitory computer-readable recording medium according to claim 15, wherein the trigger type field indicates a layer 3 (L3) network event that triggers the logging session.
17. The non-transitory computer-readable recording medium according to claim 15, wherein the event ID field indicates a mapping between an ID and at least one of: an Al derivative event or an A2 derivative event.
18. The non-transitory computer-readable recording medium according to claim 15, wherein the time to trigger field indicates that the time duration for triggering the logging session has a same scale as a time duration for triggering a radio resource management (RRM) operation.
19. The non-transitory computer-readable recording medium according to claim 15, wherein at least one of the event ID field, the time to trigger field, or the logging interval comprises an integer.
20. The non-transitory computer-readable recording medium according to claim 15, wherein the logging trigger configuration comprises at least one field associated with at least one of: the trigger type field, the event ID field, and the time to trigger field, and wherein the message further includes an instruction to instruct the UE to collect the data per reference signal (RS) period.Ill
