Measurement gap for version performance control

Abstract

Example embodiments of the present disclosure relate to methods, devices, apparatuses, and computer-readable storage media for measurement gaps for version performance control. One method includes receiving a version control configuration from a second apparatus, the version control configuration indicating a measurement gap and a resource, the resource being associated with at least one of: an uplink data transmission repetition or a downlink data transmission repetition; obtaining performance data for a machine learning (ML) model associated with at least one version during the measurement gap, wherein the performance data is collected based on at least one of: the uplink data transmission repetition or the downlink data transmission repetition associated with a transceiver; and generating data for life cycle management of the ML model based on the collected performance data.

Description

Cross-references to related applications

[0001] This application claims priority and benefit to EP patent application No. 25173105.5, filed on April 29, 2025, the contents of which are incorporated herein by reference in their entirety. Technical Field

[0002] Various exemplary embodiments of this disclosure generally relate to the telecommunications field, and more specifically to methods, apparatuses, devices, and computer-readable storage media for measuring gaps in version performance control. Background Technology

[0003] Artificial intelligence (AI) based transceivers are widely deployed in various devices, and each transceiver can be pre-configured with multiple versions of machine learning (ML) models. Devices can select different ML models for different scenarios. Summary of the Invention

[0004] In a first aspect of this disclosure, a first apparatus is provided. The first apparatus includes: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the first apparatus to at least: receive a version control configuration from a second apparatus, the version control configuration indicating measurement gaps and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; acquire, during the measurement gap, performance data of an ML model associated with at least one version, wherein the performance data is collected based on at least one of: uplink data transmission repetition or downlink data transmission repetition associated with a transceiver; and generate data for lifecycle management of the ML model based on the collected performance data.

[0005] In a second aspect of this disclosure, a second apparatus is provided. The second apparatus includes: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the second apparatus to at least: send a version control configuration to a first apparatus, the version control configuration indicating a measurement gap and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; and receive uplink data transmission repetition associated with at least one version of an ML model from the first apparatus during the measurement gap; or send downlink data transmission repetition associated with at least one version of an ML model to the first apparatus during the measurement gap.

[0006] In a third aspect of this disclosure, a method is provided. The method includes: receiving, by a first means, a version control configuration from a second means, the version control configuration indicating a measurement gap and resources, the resources being associated with at least one of: uplink data transmission duplication or downlink data transmission duplication; acquiring, during the measurement gap, performance data of an ML model associated with at least one version, wherein the performance data is collected based on at least one of: uplink data transmission duplication or downlink data transmission duplication associated with a transceiver; and generating, based on the collected performance data, data for lifecycle management of the ML model.

[0007] In a fourth aspect of this disclosure, a method is provided. The method includes: sending a version control configuration from a second device to a first device, the version control configuration indicating a measurement gap and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; and receiving uplink data transmission repetition associated with at least one version of an ML model from the first device during the measurement gap; or sending downlink data transmission repetition associated with at least one version of an ML model to the first device during the measurement gap.

[0008] In a fifth aspect of this disclosure, a first apparatus is provided. The first apparatus includes: components for receiving a version control configuration from a second apparatus, the version control configuration indicating a measurement gap and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; components for acquiring performance data of a machine learning ML model associated with at least one version during the measurement gap, wherein the performance data is collected based on at least one of: uplink data transmission repetition or downlink data transmission repetition associated with a transceiver; and components for generating data for lifecycle management of the ML model based on the collected performance data.

[0009] In a sixth aspect of this disclosure, a second apparatus is provided. The second apparatus includes: components for sending a version control configuration to a first apparatus, the version control configuration indicating a measurement gap and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; and components for receiving uplink data transmission repetition associated with at least one version of an ML model from the first apparatus during the measurement gap; or components for sending downlink data transmission repetition associated with at least one version of an ML model to the first apparatus during the measurement gap.

[0010] In a seventh aspect of this disclosure, a computer-readable medium is provided. The computer-readable medium includes instructions stored thereon for causing a device to at least execute the method according to a third aspect.

[0011] In an eighth aspect of this disclosure, a computer-readable medium is provided. The computer-readable medium includes instructions stored thereon for causing a device to at least execute the method according to the fourth aspect.

[0012] It should be understood that the summary portion is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0013] Some exemplary embodiments will now be described with reference to the accompanying drawings, in which: Figure 1 An example communication environment in which example embodiments of this disclosure may be implemented is shown; Figure 2 An example of an AI transceiver is shown; Figure 3 Signaling diagrams according to some example embodiments of this disclosure are shown; Figure 4 Signaling diagrams according to some example embodiments of this disclosure are shown; Figure 5 Signaling diagrams according to some example embodiments of this disclosure are shown; Figure 6 Flowcharts are shown according to some example embodiments of the present disclosure; Figure 7 A flowchart is shown illustrating a method implemented at a first device according to some exemplary embodiments of the present disclosure; Figure 8 A flowchart is shown illustrating a method implemented at a second device according to some example embodiments of the present disclosure; Figure 9 A flowchart is shown illustrating a method implemented at an apparatus according to some example embodiments of the present disclosure; Figure 10 A simplified block diagram of a device suitable for implementing example embodiments of the present disclosure is shown; and Figure 11 A block diagram of an example computer-readable medium according to some example embodiments of the present disclosure is shown.

[0014] In all the accompanying drawings, the same or similar reference numerals denote the same or similar elements. Detailed Implementation

[0015] The principles of this disclosure will now be described with reference to some exemplary embodiments. It should be understood that these embodiments are described for illustrative purposes only and to assist those skilled in the art in understanding and implementing this disclosure, and do not imply any limitation on the scope of this disclosure. The embodiments described herein can be implemented in various ways other than those described below.

[0016] In the following description and claims, unless otherwise defined, all technical and scientific terms used herein may have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains.

[0017] References to "an embodiment," "an embodiment," "an example embodiment," etc., in this disclosure indicate that the described embodiment may include a particular feature, structure, or characteristic, but not every embodiment needs to include that particular feature, structure, or characteristic. Furthermore, such phrases do not necessarily refer to the same embodiment. Moreover, when a particular feature, structure, or characteristic is described in connection with an embodiment, it is claimed that, whether explicitly described or not, it is within the knowledge of those skilled in the art to influence such feature, structure, or characteristic in conjunction with other embodiments.

[0018] It should be understood that although the terms "first," "second," etc., may be used before nouns in this document to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another, and they do not restrict the order of the terms(s). For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. As used herein, the term "and / or" includes any and all combinations of one or more of the listed terms.

[0019] As used herein, “at least one of the following: ” and “at least one of ” and similar wording, wherein a list of two or more elements, combined with “and” or “or”, means at least one of the elements, or at least any two or more of the elements, or at least all of the elements.

[0020] As used herein, unless explicitly stated otherwise, the “responding to A” execution step does not indicate that the step is executed immediately after “A” occurs, but may include one or more intermediate steps.

[0021] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments. As used herein, the singular forms “a,” “an,” and “the” are also intended to include the plural forms unless the context clearly indicates otherwise. It will also be understood that the terms “comprising,” “including,” “having,” “possessing,” “containing,” and / or “covering” as used herein specify the presence of the stated features, elements, and / or components, etc., but do not exclude the presence or addition of one or more other features, elements, components, and / or combinations thereof.

[0022] As used in this application, the term "circuit" may refer to one or more or all of the following: (a) Hardware circuit implementation only (e.g., implemented with purely analog and / or digital circuits) and (b) A combination of hardware circuitry and software, such as (if applicable): (i) A combination of (multiple) analog and / or digital hardware circuits and software / firmware, and (ii) Any part of a hardware processor(s) having software (including (multiple) digital signal processors, software, and (multiple) memories, which work together to enable a device (such as a mobile phone or server) to perform various functions), and (c) The operation requires software (e.g., firmware) for the operation of (multiple) hardware circuits and / or (multiple) processors, such as (multiple) microprocessors or a portion thereof, but the software may be absent when the operation does not require the software.

[0023] This definition of "circuit" applies to all uses of the term in this application (including any claims). As another example, as used in this application, the term "circuit" also covers only hardware circuitry or processors (or processors), or a portion of hardware circuitry or processors and their accompanying software and / or firmware implementations. For example, if applicable to a particular claim element, the term "circuit" also covers baseband integrated circuits or processor integrated circuits for mobile devices or similar integrated circuits in servers, cellular network devices, or other computing or networking devices.

[0024] As used herein, the term "communication network" refers to a network that conforms to any suitable communication standard, such as New Radio (NR), Long Term Evolution (LTE), LTE-A Advanced (LTE-A), Wideband Code Division Multiple Access (WCDMA), High-Speed ​​Packet Access (HSPA), Narrowband Internet of Things (NB-IoT), etc. Furthermore, communication between terminal devices and network devices in a communication network can be performed according to any suitable generation of communication protocol, including but not limited to first-generation (1G), second-generation (2G), 2.5G, 2.75G, third-generation (3G), fourth-generation (4G), 4.5G, fifth-generation (5G), 5.5G, sixth-generation (6G) communication protocols and / or any other currently known or future-developed protocols. Embodiments of this disclosure can be applied to various communication systems. Given the rapid development of communications, there will certainly be future types of communication technologies and systems that embody the nature of this disclosure. The scope of this disclosure should not be considered limited to the aforementioned systems.

[0025] As used herein, the term "network device" refers to a node in a communications network through which terminal devices access the network and receive services. Depending on the terminology and technology applied, a network device can refer to a base station (BS) or access point (AP), such as a Node B (NodeB or NB), an evolved Node B (eNodeB or eNB), an NR NB (also known as a gNB), a Remote Radio Unit (RRU), a Radio Head (RH), a Remote Radio Head (RRH), a relay, an Integrated Access and Backhaul (IAB) node, a low-power node (such as a femtosecond or picosecond), a non-terrestrial network (NTN) or non-terrestrial network device (such as satellite network equipment, low Earth orbit (LEO) satellites, and geostationary Earth orbit (GEO) satellites), an aircraft network device, etc. In some example embodiments, the Radio Access Network (RAN) split architecture includes a centralized unit (CU) and a distributed unit (DU) at the IAB donor node. An IAB node includes: a mobile terminal (IAB-MT) portion, which is similar to a UE relative to its parent node; and a DU portion of the IAB node, which is similar to a base station relative to the next-hop IAB node.

[0026] The term "terminal device" refers to any terminal device capable of wireless communication. As an example and not a limitation, a terminal device may also be referred to as a communication device, user equipment (UE), subscriber station (SS), portable subscriber station, mobile station (MS), or access terminal (AT). Terminal devices may include, but are not limited to, mobile phones, cellular phones, smartphones, Voice over IP (VoIP) phones, wireless local loop phones, tablet computers, wearable terminal devices, personal digital assistants (PDAs), portable computers, desktop computers, image capture terminal devices such as digital cameras, gaming terminal devices, music storage and recycle bins, in-vehicle wireless terminal devices, wireless endpoints, mobile stations, laptop embedded devices (LEEs), laptop-mounted devices (LMEs), USB dongles, smart devices, wireless customer premises equipment (CPEs), Internet of Things (IoT) devices, watches or other wearable devices, head-mounted displays (HMDs), vehicles, drones, medical devices and applications (e.g., remote surgery), industrial devices and applications (e.g., robots and / or other wireless devices operating in the context of industrial and / or automated processing chains), consumer electronic devices, devices operating on commercial and / or industrial wireless networks, etc. The terminal device may also correspond to the mobile terminal (MT) portion of an IAB node (e.g., a relay node). In the following description, the terms "terminal device," "communication device," "terminal," "user equipment," and "UE" are used interchangeably.

[0027] As used herein, the terms “resource,” “transmission resource,” “resource block,” “physical resource block” (PRB), “uplink resource,” or “downlink resource” can refer to any resource used to perform communication (e.g., communication between a terminal device and a network device), such as resources in the time domain, resources in the frequency domain, resources in the spatial domain, resources in the code domain, or any other combination of time-domain, frequency-domain, spatial-domain, and / or code-domain resources for implementing communication. In the following, unless explicitly stated otherwise, resources in the frequency and time domains will be used as examples of transmission resources used to describe some exemplary embodiments of this disclosure. Note that the exemplary embodiments of this disclosure are equally applicable to other resources in other domains.

[0028] Figure 1 An example communication network 100 in which exemplary embodiments of the present disclosure can be implemented is shown. For example... Figure 1 As shown, the communication network 100 may include a first device 110, which may also be referred to as a terminal device or UE, for example.

[0029] The communication network 100 may also include a second device 120, which may be considered, for example, a network device or included in a network device. In some example embodiments, the network device may be discussed as a BS, gNB, or eNB.

[0030] The service area provided by the second device 120 is referred to as a cell. The second device 120 can provide services to one or more cells of the first device. For example, the first device 110 can communicate with the second device 120 within cell 102. In some scenarios, cell 102 can be considered as a cell serving the first device 110.

[0031] In the following description, for illustrative purposes, some exemplary embodiments are described in which the first device 110 operates as a terminal device and the second device 120 operates as a network device. However, in some exemplary embodiments, the operations described in connection with the terminal device can be implemented at the network device or other devices, and the operations described in connection with the network device can be implemented at the terminal device or other devices.

[0032] In some example embodiments, if the first device 110 is a terminal device and the second device 120 is a network device, the link from the second device 120 to the first device 110 is referred to as a downlink (DL), and the link from the first device 110 to the second device 120 is referred to as a UL. In the DL, the second device 120 is a transmitting (TX) device (or transmitter), and the first device 110 is a receiving (RX) device (or receiver). In the UL, the first device 110 is a TX device (or transmitter), and the second device 120 is an RX device (or receiver).

[0033] It should be understood that Figure 1 The number of devices and their connections shown are for illustrative purposes only and do not imply any limitation. Communication environment 100 may include any suitable number of devices configured to implement the exemplary embodiments of this disclosure. Although not shown, it should be understood that one or more additional devices may be located in cell 102, and one or more additional cells may be deployed in communication environment 100. Note that although shown as a network device, network device 120 may be another device besides a network device. Although illustrated as a terminal device, terminal device 110 may be another device besides a terminal device.

[0034] Communication in communication environment 100 may be implemented according to any suitable communication protocol(s), including but not limited to cellular communication protocols, wireless local area network communication protocols such as IEEE 802.11, and / or any other currently known or future-developed protocols. Furthermore, communication may utilize any suitable wireless communication technology, including but not limited to: Code Division Multiple Access (CDMA), Frequency Division Multiple Access (FDMA), Time Division Multiple Access (TDMA), Frequency Division Duplex (FDD), Time Division Duplex (TDD), Multiple Input Multiple Output (MIMO), Orthogonal Frequency Division Multiple Access (OFDM), Discrete Fourier Transform Extended OFDM (DFT-s-OFDM), and / or any other currently known or future-developed technologies.

[0035] In some of the proposed 6G solutions, AI / ML capabilities are natively supported. These AI / ML capabilities may include deep waveform learning, deep constellation learning, and more. On the other hand, the physical layer characteristics of 6G allow 6G designs to deviate from the OFDM-based designs of 5G. Native AI air interfaces in 6G networks are also discussed in some of the proposed solutions. Many companies are also working on standardizing native AI support in 6G.

[0036] Now for reference Figure 2 It shows some of the functional modules of the AI ​​transceiver 200.

[0037] like Figure 2 As shown, in 6G, some functions can be implemented or obtained by using AI / ML methods. For example, the data to be transmitted can first be encoded using encoding function 240, which can be implemented using a general method (i.e., a method without AI / ML). Based on the determined encoding method, the data can then be modulated via modulation function 210. Furthermore, pilot signals or sequences can be generated using pilot insertion function 215 (e.g., implicitly, explicitly, or optionally). The modulated signal can be further modulated using OFDM modulation function 245. The signal can then be processed by digital predistortion function 205.

[0038] The processed signal can be transmitted via channel 260 and received by the receiver. The receiver can process the signal using OFDM demodulation function 250. The signal can then be processed by digital post-distortion function 220. The receiver can also perform channel estimation using channel estimation function 225 based on the signal quality. If interference is detected, the receiver can apply interference cancellation to the signal using interference cancellation function 230 (e.g., implicitly or explicitly). The signal can then be processed by demapping function 235. Finally, the signal can be decoded using decoding function 255.

[0039] In the above process, multiple functions can be implemented using AI / ML methods. For example, modulation function 210, digital predistortion function 205, pilot insertion function 215, digital postdistortion function 220, channel estimation function 225, interference cancellation function 230, and demapping function 235 can be implemented using AI / ML methods.

[0040] On the other hand, the encoding function 240, the OFDM modulation function 245, the OFDM demodulation function 250, and the decoding function 255 can be implemented using a general method (i.e., without using AI / ML).

[0041] Among the schemes discussed, several methods can be used to generate models of lower complexity (e.g., model distillation, quantization, pruning, etc., via teacher and student models). These models can be implemented in either the receiver or the transmitter.

[0042] In some versions of cellular equipment implementing New Radio (NR) and 6G, several radio-related algorithms based on machine learning may exist. Examples of such algorithms could be channel estimation, modulator and demodulator, encoder and decoder, channel state information (CSI) estimation, Hybrid Automatic Repeat Request (HARQ) feedback estimation, etc.

[0043] The problem with these types of algorithms is that they can be quite complex, so a significant amount of energy consumption can be expected. The approach discussed is to design the performance of these algorithms to meet a given performance (e.g., throughput, error rate, etc.), regardless of the radio link conditions experienced at the device (uplink or downlink).

[0044] However, in cellular systems, the actual performance requirements are associated with a given Quality of Service (QoS) profile (or a set of profiles when a device has multiple active services). Therefore, the performance of these machine learning algorithms does not always need to be the algorithm that maximizes performance.

[0045] Consider two different scenarios: one where radio conditions are excellent (i.e., DL and UL are experiencing a high signal-to-noise ratio (SNIR)) and the other where conditions are the opposite (i.e., DL and UL are experiencing a low SNIR). In both scenarios, the machine learning algorithm can be adapted to the conditions regarding processing power. For example, in a device with good conditions, the algorithm's complexity can be relaxed, and therefore it performs worse because the SINR conditions are sufficient to support the configured QoS profile anyway. Conversely, a device with poor propagation conditions will require the machine learning algorithm to operate at maximum complexity to cope with the low SINR.

[0046] In summary, the radio is expected to be equipped with multiple versions of the AI ​​transceiver, each optimized for a subset of key performance indicators (KPIs). For example, some versions will be inherently triggered during power-saving modes, while others will improve link performance. Maintaining these versions within the radio TRX (UE or gNB) is a problem to be solved.

[0047] According to some example embodiments of this disclosure, a solution for version performance control targeting measurement gaps is provided. In this solution, a first device 110 receives a version control configuration from a second device 120, the version control configuration indicating measurement gaps and resources associated with at least one of uplink data transmission repetition or downlink data transmission repetition. The first device 110 acquires performance data of an ML model associated with at least one version during the measurement gap. The performance data is collected based on at least one of uplink data transmission repetition or downlink data transmission repetition associated with a transceiver. The first device 110 generates data for lifecycle management of the ML model based on the collected performance data. Lifecycle management as described herein may include, but is not limited to, validation, training, and fine-tuning of the ML model. Furthermore, processes related to the management of the ML model may also include updating the version of the ML model, switching the ML model, and modifying the implementation of the ML model.

[0048] Through the above solutions, embodiments of this disclosure provide a data collection framework that utilizes measurement gaps to support version performance control of AI TRX in a first device 110 (e.g., UE) or a second device 120 (e.g., gNB).

[0049] This solution ensures that the controlled version operates as expected and can update the controlled version's operations (e.g., by updating any combination of weights, biases, pruning and quantization adapters, associated reference performance, etc.).

[0050] This approach enables an effective way to implement version control, which improves the responsiveness of the version control process and provides feedback on key KPIs to effectively adjust the ML model.

[0051] The exemplary embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0052] Typically, the version control solutions provided by embodiments of this disclosure involve some solutions for AI TX and some other solutions for AI RX.

[0053] In the AI ​​TX scenario, the ML model can be deployed at the transmitter of the first device 110, and the Physical Uplink Shared Channel (PUSCH) is used to transmit uplink data to the second device 120.

[0054] In the AI ​​TX scenario, the first device 110 can use a first version of the ML model to transmit uplink data to the second device 120. For example, the Kth version of the ML model (or any predetermined version of the ML model).

[0055] In some embodiments, the radio TRX can be equipped with multiple versions of AI TRX (different versions using ML models), each version of which may have the following characteristics: Indicators of model complexity

[0056] Version computational cost

[0057] Based on the defined performance KPIs, the target performance (which we usually define as...) ).

[0058] The model weights and biases indexed by the layers to which they are applied - applicable to all versions and corresponding to base version V0.

[0059] Pruning masks indexed by the layers that apply them Weight and bias quantization functions (with layer index) Activate the output quantization function (with layer index) As an example, the following shows examples of different parameters for an ML model.

[0060]

[0061] As shown in the table, parameters can include the version of the ML model. For example, version 0 (i.e., V.0) can be the initial version of the ML model without any pruning or adjustments.

[0062] Parameters may also include the applicable SINR range for the ML model, indicating how to select the ML model based on signal quality. Parameters may also include predefined target performance for the ML model (i.e., This parameter indicates the expected SINR of the signal after the ML model is applied.

[0063] The table above also shows some parameters of the ML model, such as those derived from it. It can indicate how complex the model is. Normalized cost The parameters can indicate the computational cost of the ML model. Furthermore, the parameters can also include weights and biases for different versions of the model. The activation function for the ML model can also exist in the parameter set.

[0064] Some features used to tune the ML model can also exist in the parameter set. For example, different pruning functions can be implemented for different versions of the ML model, and different weights, biases, and activation output quantization functions can also be different for different versions of the ML model.

[0065] Multiple versions of an ML model can be derived from baseline training data, for example, using data points generated via simulation, and they may need to be periodically validated—to ensure that the versions work as expected and / or fine-tuned when actual performance drops below the target performance.

[0066] In some embodiments, the second device 120 may collect performance data based on the received data. The performance data may include multiple key performance indicators (KPIs) for evaluating the performance of the ML model (e.g., the KPIs shown in the table above).

[0067] In some embodiments, the second device 120 may determine and provide a version control configuration for the first device 110. The version control configuration may indicate measurement gaps for repetition of PUSCH or Physical Downlink Shared Channel (PDSCH) corresponding to different versions of the ML model. For example, measurement gaps may be used to perform uplink data transmission using an initial version of the ML model, without using uplink data transmission of the ML model as a baseline for performance evaluation.

[0068] In some embodiments, the second device 120 may trigger a measurement gap for the first device 110 by transmitting a Radio Resource Control (RRC) message to the first device 110. Alternatively or additionally, the second device 120 may also use downlink control information (DCI) to trigger the measurement gap.

[0069] During the measurement interval, the first device 110 can transmit repeats as needed, and the second device 120 can continue to collect performance data for those repeats.

[0070] When collecting performance data, the second device 120 can transmit feedback to the first device 110 to provide performance data to the first device 110.

[0071] Based on the received performance data, the first device 110 can generate corresponding data for the lifecycle management of the ML model and fine-tune the ML model to improve performance.

[0072] On the other hand, in the AI ​​RX scenario, the first device 110 can receive downlink data transmission from the second device 120 via PDSCH using the Kth version of the ML model. The first device 110 can collect performance data of the ML model.

[0073] Similarly, in this scenario, the first device 110 can also receive version control configuration and instructions to trigger measurement gaps from the second device 120. Based on this configuration, the first device 110 can receive PDSCH repetitions and collect corresponding performance data within the measurement gaps.

[0074] In some embodiments, the first device 110 may generate data for lifecycle management and adjust the ML model based on the data.

[0075] In the following text, we will combine Figures 3 to 6 Various embodiments of this disclosure are described.

[0076] Now for reference Figure 3 This illustrates signaling diagram 300 for communication according to some example embodiments of the present disclosure. For example... Figure 3 As shown, signaling diagram 300 relates to first device 110 and second device 120. For discussion purposes, reference is made to... Figure 1 To describe signaling diagram 300.

[0077] exist Figure 3 The scenario shown illustrates a data collection framework for version control when the first device 110 is equipped with AI TX (e.g., AI-enhanced digital predistortion, AI-enhanced precoder determination). In addition, an indicator indicating the activation of the measurement gap is transmitted along with the version control configuration.

[0078] like Figure 3 As shown, the first device 110 can use a first version of the ML model to transmit (305) uplink data transmission to the second device 120. For example, uplink data transmission can be transmitted to the second device 120 via PUSCH.

[0079] In some embodiments, the first version may include a predetermined version of the ML model. For example, the first device 110 may use an AI TX having a Kth version (e.g., denoted as version Vk) of the ML model to transmit data via PUSCH.

[0080] Based on the received data transmission, the second device 120 can collect (310) performance data of the ML model for the first version. The performance data may include a first set of KPIs associated with the first version of the ML model. For example, upon receiving a PUSCH transmission, the second device 120 can measure and collect performance data or performance KPIs for that transmission, which may be represented as KPI(Vk).

[0081] In some embodiments, the KPI may include the block error rate (BLER) or bit error rate (BER) for the transmission. In some other embodiments, the SINR may also be calculated and collected by the second device 120.

[0082] like Figure 3 As shown, the second device 120 can send (315) a version control configuration indicating the measurement gap to the first device 110. In some examples, the measurement gap can be used for performance control of the first version (i.e., Vk) of the ML model against the AI ​​TX version.

[0083] In some embodiments, the version control configuration may also indicate the relevant configuration for the measurement gap. For example, the version control configuration may indicate the measurement gap length (MGL), which may also be referred to as the measurement gap duration.

[0084] In some embodiments, the version control configuration may also include a measurement gap repetition period (MGRP) that indicates the frequency of the repetitive measurement gap.

[0085] In some other embodiments, the version control configuration may also include measurement gap patterns. Measurement gap patterns can be specific to AI / ML version performance control. For example, in NR, there may be 26 measurement gap patterns, where MGL varies from 1.5ms to 20ms, and MGRP varies from 20ms to 160ms. Therefore, the version control configuration may include measurement gap pattern identifiers (IDs) (from the existing 26 options).

[0086] In some further embodiments, the version control configuration may also include a set of values ​​for the measurement gap. For example, this set of values ​​may include a set of parameters for the measurement gap. As an example, the second device 120 may explicitly configure the MGL and MGRP values ​​for the measurement gap.

[0087] In some embodiments, the version control configuration may indicate a resource associated with at least one of the uplink data transmission duplicates. Alternatively or additionally, the version control configuration may also indicate the ML version to be used for the duplicate.

[0088] In some embodiments, uplink data transmission repetition may include PUSCH repetition.

[0089] In some embodiments, the resource may include at least one predetermined time slot. For example, the second device 120 may configure the first device 110 to repeat M time slots / symbols of the last transmitted PUSCH (transmitted by the first device 110 using AI TX version Vk). In other words, PUSCH repetition may include a portion of the entire transmission, and the version control configuration may indicate or define the time slots to be repeated or the time slots used for PUSCH transmission.

[0090] In some embodiments, at least one uplink data transmission repetition may include a repetition corresponding to a second version of the ML model. The second version of the ML model may include an initial version of the ML model. For example, the first device 110 may be configured to transmit a PUSCH repetition 0 of application AI TX V.0 (no pruning, no quantization).

[0091] In some embodiments, the version control configuration may also configure the first device 110 to transmit duplicates without using an ML model, which may also be referred to as baseline transmissions. For example, the duplicate may include a PUSCH duplicate 1 that applies a non-AI / ML TX.

[0092] In some embodiments, PUSCH repeats as described above can be configured within the same Physical Resource Block (PRB) or a set of different PRBs. If repeats require resource reconfiguration, resources can be specified for each repeat index.

[0093] As described above, the version to be controlled and the time slot / repetition to be repeated can be indicated by the second device 120 (via version control configuration). Alternatively or additionally, the first device 110 can also actively request repetition. For example, the first device 110 can transmit a request to the second device 120 indicating the number of uplink data transmission repetitions associated with the ML model. Furthermore, the second device 120 can also request control over which version of the ML model or which repetition to transmit.

[0094] In some embodiments, when returning to a scenario where the configuration of the second device 120 is repeated, the version control configuration can be received from the second device 120 via an RRC message. For example, the version control configuration can be carried in the information element (IE) of the RRC message.

[0095] In some embodiments, the second device 120 may transmit an indicator to the first device 110 indicating the activation of a measurement gap and version control configuration. For example, a new RRC trigger may be transmitted to instruct the first device 110 to apply a measurement gap, i.e., suspend standard PUSCH transmission. Once the first device 110 receives the RRC message, it can determine to trigger the AI ​​TX version control process.

[0096] Based on the activation indicator, the first device 110 can apply the measurement gap and repeat the PUSCH time slot as instructed.

[0097] Alternatively or additionally, the first device 110 can also trigger the measurement gap itself.

[0098] In some embodiments, if the first device 110 receives a version control configuration during a measurement gap, the first device 110 may transmit (320) uplink data transmission duplicates to the second device 120 using a second version of the ML model.

[0099] In some embodiments, a second version of the ML model differs from a first version of the ML model. For example, based on the defined time slots to be repeated, the resources for each repetition, and the AI ​​TX version to be controlled, the first device 110 may use an initial version of the ML model (represented as PUSCH repetition 0 using AI TX version V.0) to transmit repetitions.

[0100] It should be noted in this document that the first device 110 can also use various versions of the ML model (e.g., the (k-1)th version of the ML model) to transmit repetitions.

[0101] like Figure 3 As shown, still within the measurement gap, the first device 110 can use non-AI / ML (baseline) transmission to transmit (325) PUSCH repeat 1.

[0102] Through the above process, the first device 110 can transmit the repetitions indicated by the second device 120. Alternatively or additionally, if the second device 120 actively requests repetitions, it can also transmit uplink data transmission repetitions corresponding to the number requested by the first device 110 during measurement intervals. The number of repetitions in this document can be determined and requested by the first device 110. Furthermore, in this scenario, the version of the ML model can also be determined by the first device 110.

[0103] As described above, based on the transmitted repetitions, the second device 120 can receive uplink data transmission repetitions associated with at least one version of the ML model from the first device 110 during measurement intervals. In some examples, the at least one version may include at least one AI version associated with the ML model, and the transceiver may include a transceiver that can be implemented using an ML model corresponding to the at least one AI version.

[0104] Based on the received PUSCH repeats, the second device 120 can decode the repeats and collect KPIs (V.0) from PUSCH repeat 0 (using AI TX V.0) and collect the KPI baseline from PUSCH repeat 1 (using non-AI / ML TX).

[0105] Alternatively or additionally, if the first device 110 requests a specific number of repetitions, the second device 120 may decode the repetitions according to the request of the first device 110.

[0106] In some embodiments, the second device 120 may decode time slots (e.g., predetermined time slots for certain repetitions) and perform measurements on the decoded repetitions to obtain performance data.

[0107] In some embodiments, the second device 120 may collect (330) performance data by combining performance data collected based on at least one uplink data transmission repetition during the measurement gap and performance data acquired before the measurement gap (e.g., PCH data transmission using a first version of the ML model).

[0108] Based on the data collection process described above, the second device 120 can transmit (335) performance data to the first device 110. In some embodiments, performance data is collected based on uplink data transmission repetitions (e.g., uplink data transmission repetitions within or outside measurement intervals). In other words, the performance data may include at least a first set of KPIs associated with a first version of the ML model, a second set of KPIs associated with a second version of the ML model, and a third set of KPIs associated with data transmissions that do not use the ML model.

[0109] In some embodiments, performance data transmitted to the first device 110 may be transmitted in the form of KPI tuples. In some examples, performance data may be transmitted to the first device 110 within a feedback message.

[0110] In some embodiments, the first device 110 may acquire performance data of an ML model associated with at least one version of the ML model during measurement intervals. The performance data may be acquired, for example, by decoding KPI tuples.

[0111] like Figure 3 As shown, based on the collected performance data, the first device 110 can generate (340) training data, for example, for lifecycle management of the ML model (e.g., as label data). For example, the first device 110 can extract performance data from KPI tuples and use it to generate training data.

[0112] Based on the received performance data, if the first device 110 determines, based on the second set of KPIs and the third set of KPIs, that the performance of the second version of the ML model is worse than the performance of data transmission without using the ML model, then the first device 110 may update at least the first version of the ML model and the second version of the ML model based at least on the training data.

[0113] For example, the first device 110 may use the newly collected data to update at least model versions V.0 and Vk only if the KPI (V.0) is worse than the KPI transmitted from the baseline.

[0114] In some other examples, the first device 110 can also use the generated training data to validate or fine-tune the ML model.

[0115] The above process enables an effective version control process and improves the efficiency of updating AI models.

[0116] Now for reference Figure 4 This illustrates a signaling diagram 400 for communication according to some example embodiments of the present disclosure. For example... Figure 4 As shown, signaling diagram 400 relates to first device 110 and second device 120. For discussion purposes, reference is made to... Figure 1 To describe signaling diagram 400.

[0117] exist Figure 4 In the scenario shown, the RRC pre-configuration for AI TX is first performed by the second device 120, and then DCI quickly triggers the immediate initiation of the measurement gap and repetition process. Figure 4 In the process shown, some steps are similar to Figure 3 The steps are already covered, so some details will not be repeated here.

[0118] like Figure 4 As shown, the second device 120 can send an (405) RRC configuration message to the first device 110, which defines measurement gap parameters (including gap length, periodicity and precise timing offset) and PUSCH repetition details (number of time slots to be repeated, resource allocation and specific AI TX version to be used).

[0119] The first device 110 can use AI TX version Vk (or any predetermined version) to transmit (410) PUSCH data transmission.

[0120] Upon receiving PUSCH data transmission, the second device 120 may collect (415) performance KPIs (e.g., BLER, BER) for the initial transmission, referred to as KPI (Vk).

[0121] In some embodiments, the second device 120 may transmit (420) an indicator indicating the activation of the measurement gap to the first device 110. For example, the indicator may be transmitted via DCI.

[0122] As an example, in order to quickly initiate a measurement, the second device 120 can send a DCI trigger instructing the first device 110 to immediately activate the pre-configured measurement gap.

[0123] In some embodiments, the DCI carrying an indicator may include a measurement gap trigger indicator, which is a sign that explicitly notifies the first device 110 to immediately activate the measurement gap instead of waiting for its normal scheduling.

[0124] In some embodiments, the DCI may also include a measurement gap mode ID / parameter pointer, which may include a field pointing to a measurement gap configuration established via an RRC message. This may be a mode ID (from one of the 26 standard NR measurement gap modes defined in some of the discussed schemes) or an explicit subfield specifying MGL and MGRP. It ensures that the first device 110 can know exactly which measurement gap settings to apply (e.g., the duration of the gap and the exact timing offset within the frame).

[0125] In some other embodiments, the DCI may also include a timing offset for immediate activation, which may include a field (or part of the gap configuration) indicating the precise timing offset by which the measurement gap should begin from the current frame or time slot. This can help synchronize the gap with the current operation of the first device 110.

[0126] In some further embodiments, the DCI may also carry a PUSCH reconfiguration.

[0127] For example, the PUSCH repeat configuration may include a repeat count / index, which may include information about how many repeats should occur (e.g., indicating that two repeats will be sent) and a specific identifier for each repeat (e.g., “repeat 0” and “repeat 1”).

[0128] In some embodiments, the PUSCH repeat configuration may also include resource allocation details. For example, the DCI may include references to resource allocations (e.g., frequency domain resources or PRB sets) for each repeat. This can be particularly useful if the repeats are to be transmitted in the same or different resource blocks.

[0129] In some embodiments, DCI may also include AI TX version assignment, which may include a field indicating which AI TX version to use for each repetition (e.g., specifying that repetition 0 can use AI TX V.0 and repetition 1 can use the baseline non-AI / ML configuration).

[0130] In some embodiments, the DCI may also include an identifier indicating the association between at least one of the Hybrid Automatic Repeat Request (HARQ) process and the uplink data transmission repetition. For example, if the repetitioned transmission needs to be associated with the same HARQ process, the DCI may carry the HARQ process number.

[0131] In some embodiments, the DCI may also include a validity timer that indicates the duration for which the measurement gap is valid. The validity timer may indicate the duration for which the measurement gap is valid. For example, the DCI may include a field indicating how long a rapid trigger is valid, ensuring that if the first device 110 does not immediately initiate the gap (e.g., due to processing delays), the trigger may not be applied after a specific time.

[0132] like Figure 4 As shown, within the triggered measurement gap, the first device 110 can use AI TX version V.0 to transmit (425) PUSCH repeat 0. The first device 110 can also use non-AI / ML (baseline) transmission to transmit (430) PUSCH repeat 1.

[0133] Based on the received repetitions, the second device 120 can collect (435) KPIs (V.0) from repetition 0 and collect baseline KPIs from repetition 1.

[0134] Based on the collected performance data, the second device 120 can send (440) three sets of KPIs (KPI(Vk), KPI(V.0) and baseline KPI) back to the first device 110.

[0135] Upon receiving performance data, the first device 110 can use the received KPI tuples and repeated transmission data to generate (445) labeled training data and update its AI TX model (if necessary) to ensure that the controlled version is validated or updated based on its performance.

[0136] Through the above process, rapid triggering of measurement gaps can be achieved, thereby reducing the latency of collecting performance data and accelerating the iteration of ML models.

[0137] It should be noted that although a description was provided using AI TX, Figure 3 and Figure 4Some of the details and steps can also be applied to AI RX using PDSCH instead of PUSCH.

[0138] Now for reference Figure 5 This illustrates a signaling diagram 500 for communication according to some example embodiments of the present disclosure. For example... Figure 5 As shown, signaling diagram 500 relates to first device 110 and second device 120. For discussion purposes, reference is made to... Figure 1 To describe signaling diagram 500.

[0139] exist Figure 5 In the scenario shown, the DL AI RX version management process is implemented.

[0140] like Figure 5 As shown, the second device 120 can use a first version of the ML model to transmit (505) downlink data transmission to the first device 110. For example, the first device 110 can receive PDSCH transmissions using AI RX Vk, for which the first device 110 can have measured performance KPIs (Vk), such as KPI = BLER, BER, etc. Upon receiving a PDSCH transmission, the first device 110 may collect (510) performance data corresponding to the first version of the ML model.

[0141] In some embodiments, the second device 120 may transmit (515) version control configuration to the first device 110. The version control configuration may include configuration for measurement gaps and configuration for PDSCH repetition. For example, the measurement gap may be configured with MGL and MGRP from the measurement gap mode or with a new set of values, and PDSCH repetition may be specified using a specific version (e.g., V.0) of the ML model to be used.

[0142] In some embodiments, version control configuration may be transmitted via RRC messages, and during this period, the first device 110 may apply an RRC trigger (i.e., an indicator that triggers the activation of the measurement gap) for the measurement gap (i.e., interrupting standard PDSCH reception) which may also be transmitted along with the version control configuration.

[0143] Alternatively or additionally, an indicator that triggers activation of the measurement gap can also be transmitted to the first device 110 via DCI.

[0144] Although the repetition is configured by the second device 120 in the above process, the first device 110 may also actively request repetition for PDSCH.

[0145] In some embodiments, the first device 110 may transmit a request to the second device 120 indicating the number of downlink data transmission repetitions associated with the ML model.

[0146] In some embodiments, if the first device 110 receives version control configuration during a measurement interval, the first device 110 may receive downlink data transmission repetitions from the second device 120 using a second version of the ML model. In some embodiments, the second version of the ML model is different from the first version of the ML model.

[0147] In some embodiments, the second device 120 may transmit downlink data transmission repetitions associated with at least one version of the ML model to the first device 110 during measurement intervals. For example, the second device 120 may repeat the last PDSCH slot (previously received using AI RX Vk) by repeating PDSCH repetition 0, and the first device 110 may apply AIRX V.0 (no pruning, no quantization) and receive (520) PDSCH repetition 0.

[0148] As some other examples, the first device 110 can receive (525) PDSCH repeats 1 without using the ML model.

[0149] Alternatively or additionally, if the first device 110 actively requests a PDSCH repeat, the first device 110 may receive downlink data transmission repeats from the second device 120 during the measurement interval corresponding to that number (i.e., the number of requested repeats).

[0150] In some embodiments, the first device may decode a time slot using its available RX and acquire the corresponding performance data. For example, the first device 110 may collect (530) performance data based on downlink data transmission repetition. As an example, the first device 110 may collect the measured performance KPI (V.0) corresponding to AI RX V.0 and the measured performance KPI corresponding to the baseline of non-AI / ML reception.

[0151] like Figure 5 As shown, the first device 110 can use KPI tuples and generate (535) labels for the PDSCH repetition signal. The first device 110 can update model versions V.0 and Vk at least using newly collected data only if the KPI (V.0) is worse than the baseline KPI.

[0152] In this way, the measurement gap can be coordinated by the first device 110 and the second device 120, and the training data collection process can be synchronized. Therefore, the efficiency of data collection and alignment of the measurement gap can be improved.

[0153] Now for reference Figure 6 The diagram illustrates a flowchart 600 for communication according to some example embodiments of the present disclosure.

[0154] exist Figure 6 In the scenario shown, AI RX is implemented at the receiver. Figure 6 Some of the steps are similar to those shown in the previous diagram, and their details will not be repeated here. Figure 6 The process shown can be performed in Figure 1 The first device shown is implemented at location 110.

[0155] like Figure 6 As shown in box 605, an apparatus (e.g., first apparatus 110) can receive downlink data transmissions from a network device using a first version of the ML model. For example, the apparatus can receive a first PDSCH using AI RX version Vk (or a predetermined version of the ML model) and measure selected KPIs, such as BLER.

[0156] In some embodiments, the device can receive a version control configuration indicating measurement gaps from a network device. In some embodiments, the version control configuration is associated with an ML model.

[0157] In some embodiments, the version control configuration may include an MGL indicating the measurement duration of the measurement gap. The version control configuration may also indicate an MGRP, which can control the frequency at which the measurement gap is repeated.

[0158] In some embodiments, the version control configuration can also indicate the measurement gap mode, which can inform the device which mode can be selected for the upcoming measurement gap.

[0159] In some embodiments, the version control configuration can indicate a set of values ​​for the measurement gap. For example, it can specifically indicate the values ​​for MGL and MGRP.

[0160] In some embodiments, versioning configuration may indicate resources associated with repetitive downlink data transmission. Resources may include at least one predetermined time slot.

[0161] In some embodiments, downlink data transmission repetition may include PDSCH repetition.

[0162] In some embodiments, the device may receive version control configuration from a network device via RRC messages.

[0163] In some embodiments, an indicator indicating the activation of the measurement gap may be transmitted along with the version control configuration. For example, the indicator may be received along with the version control configuration via an RRC message.

[0164] Alternatively or additionally, an indicator indicating the activation of the measurement gap can be transmitted via DCI.

[0165] In some embodiments, the DCI may also include an identifier that indicates the association between the HARQ process and downlink data transmission repetition.

[0166] In some other embodiments, the DCI may also include a validity timer that indicates the duration for which the measurement gap is valid.

[0167] In some embodiments, the apparatus may, based on a version control configuration, receive repeated downlink data transmissions associated with at least one version of the ML model from a network device during measurement intervals.

[0168] In some embodiments, at least one version includes at least one AI version associated with the ML model. At least one AI version (or a repeating configuration) may be specified by a versioning configuration. At least one AI version may correspond to an ML model implemented at the transceiver of the device.

[0169] Alternatively or additionally, the device may also actively request repetition of downlink data transmission. For example, the device may transmit a request to a network device indicating the number of repetitions of downlink data transmission associated with the ML model.

[0170] At block 610, if the device receives a version control configuration during a measurement interval, the device can receive downlink data transmission duplicates from the network device using a second version of the ML model. In some embodiments, the second version of the ML model may include an initial version of the ML model, which may differ from the first version of the ML model.

[0171] For example, the device can receive the first PDSCH repetition using AI RX version V.0 (or any other version chosen by the device) and measure BLER without performing any combination with the first transmission. AI RX version V.0 can be referred to as a second version of the ML model, which is different from Vk of the ML model.

[0172] At box 615, the device may receive a second PDSCH repeat using a non-AI / ML version (or any other version selected by the device) and measure BLER without performing any combination with the first transmission / repeat.

[0173] Alternatively or additionally, if the device actively requests a PDSCH repeat, it may receive downlink data transmission repeats from the network device during the measurement interval corresponding to the number requested by the device.

[0174] Based on the repetition of downlink data transmission, the device can collect performance data.

[0175] In some embodiments, performance data may include at least a first set of KPIs associated with a first version of the ML model, a second set of KPIs associated with a second version of the ML model, and a third set of KPIs associated with the transfer of data that does not use the ML model.

[0176] In some embodiments, the KPIs described above may include BLER, BER, and SINR.

[0177] In some embodiments, at block 620, the apparatus may generate data for lifecycle management of the ML model based on performance data associated with downlink data transmission and downlink data transmission repetition.

[0178] For example, the device can collect a training dataset by using the corresponding KPIs and the number of repetitions, as well as timestamps to mark each transmission.

[0179] In some embodiments, the device may update at least a first version of the ML model and a second version of the ML model based at least on data used for lifecycle management.

[0180] This approach provides a flexible way to validate and update ML models. Therefore, it improves the overall efficiency and responsiveness of version control. By implementing the data collection process described above, embodiments of this disclosure also improve the efficiency of collecting performance data.

[0181] Figure 7 A flowchart of an example method 700 implemented at a first device according to some example embodiments of the present disclosure is shown. For the purposes of discussion, [the following will be discussed]. Figure 1 Method 700 is described by the angle of the first device 110 in the middle.

[0182] In block 710, the first device 110 receives a version control configuration from the second device, the version control configuration indicating measurement gaps and resources, the resources being associated with at least one of the following: uplink data transmission duplication or downlink data transmission duplication.

[0183] At box 720, the first device 110 acquires performance data of an ML model associated with at least one version during a measurement gap, wherein the performance data is collected based on at least one of the following associated with the transceiver: uplink data transmission duplication or downlink data transmission duplication.

[0184] At box 730, the first device 110 generates data for lifecycle management of the ML model based on the collected performance data.

[0185] In some example embodiments, method 700 further includes receiving version control configuration from a second device via an RRC message.

[0186] In some example embodiments, method 700 further includes receiving an indicator from a second device, together with version control configuration, the indicator indicating activation of the measurement gap.

[0187] In some example embodiments, an indicator indicating the activation of the measurement gap is transmitted via DCI.

[0188] In some example embodiments, the DCI also includes an identifier that indicates an association between the HARQ process and at least one of the following: uplink data transmission duplication or downlink data transmission duplication.

[0189] In some example embodiments, the DCI also includes a validity timer that indicates the duration for which the measurement gap is valid.

[0190] In some example embodiments, method 700 further includes: sending uplink data transmission to a second device using a first version of the ML model; and if a version control configuration is received, sending uplink data transmission to the second device repeatedly using a second version of the ML model during the measurement interval, wherein the second version of the ML model is different from the first version of the ML model.

[0191] In some example embodiments, method 700 further includes receiving performance data from a second device, wherein the performance data is collected based on uplink data transmission repetition.

[0192] In some example embodiments, method 700 further includes: sending a request to a second device indicating the number of uplink data transmission repetitions associated with the ML model; and sending uplink data transmission repetitions corresponding to the number to the second device during measurement intervals.

[0193] In some example embodiments, method 700 further includes: receiving downlink data transmission from a second device using a first version of the ML model; receiving downlink data transmission repetition from the second device using a second version of the ML model during measurement intervals if a version control configuration is received, wherein the second version of the ML model is different from the first version of the ML model; and collecting performance data based on the downlink data transmission repetition.

[0194] In some example embodiments, method 700 further includes: sending a request to a second device to indicate the number of downlink data transmission repetitions associated with the ML model; and receiving downlink data transmission repetitions corresponding to the number from the second device during measurement intervals.

[0195] In some example embodiments, the performance data includes at least a first set of key performance indicators (KPIs) associated with a first version of the ML model, a second set of KPIs associated with a second version of the ML model, and a third set of KPIs associated with the transfer of data that does not use the ML model.

[0196] In some example embodiments, method 700 further includes: generating data for lifecycle management as label data for the ML model based on performance data; and updating at least the first version of the ML model and the second version of the ML model based at least on the data for lifecycle management if it is determined based on a second set of KPIs and a third set of KPIs that the performance of the second version of the ML model is worse than the performance of data transmission without using the ML model.

[0197] In some example implementations, the KPIs include at least one of the following: BLER; BER; SINR.

[0198] In some example embodiments, a first version of the ML model includes a predetermined version of the ML model, and a second version of the ML model includes an initial version of the ML model.

[0199] In some example embodiments, the version control configuration indicates at least one of the following: MGL; MGRP; measurement gap mode; a set of values ​​for the measurement gap.

[0200] In some example embodiments, the resource includes at least one predetermined time slot.

[0201] In some example implementations, uplink data transmission repetition includes PUSCH repetition, and downlink data transmission repetition includes PDSCH repetition.

[0202] In some example embodiments, at least one version includes at least one AI version associated with the ML model, and the transceiver includes a transceiver implemented with an ML model corresponding to the at least one AI version.

[0203] In some example embodiments, the first device includes a terminal device, and the second device includes a network device.

[0204] Figure 8 A flowchart of an example method 800 implemented at a second device according to some example embodiments of the present disclosure is shown. For the purposes of discussion, [the following will be discussed]. Figure 1 Method 800 is described by the angle of the second device 120 in the middle.

[0205] In block 810, the second device 120 sends a version control configuration to the first device, the version control configuration indicating measurement gaps and resources, the resources being associated with at least one of the following: uplink data transmission duplication or downlink data transmission duplication.

[0206] At box 820, the second device 120 receives repeated uplink data transmissions associated with at least one version of the ML model from the first device during measurement intervals.

[0207] At box 830, the second device 120 repeatedly sends downlink data transmissions associated with at least one version of the ML model to the first device during the measurement interval.

[0208] In some example embodiments, method 800 further includes sending version control configuration to the first device via an RRC message.

[0209] In some example embodiments, method 800 further includes sending an indicator to a first device, together with version control configuration, the indicator indicating activation of the measurement gap.

[0210] In some example embodiments, an indicator indicating the activation of the measurement gap is sent via DCI.

[0211] In some example embodiments, the DCI also includes an identifier that indicates an association between the HARQ process and at least one of the following: uplink data transmission duplication or downlink data transmission duplication.

[0212] In some example embodiments, the DCI also includes a validity timer that indicates the duration for which the measurement gap is valid.

[0213] In some example embodiments, method 800 further includes: receiving uplink data transmission from a first device using a first version of the ML model; sending a version control configuration to the first device; and during measurement intervals, receiving uplink data transmission from the first device repeatedly using a second version of the ML model, wherein the second version of the ML model is different from the first version of the ML model.

[0214] In some example embodiments, method 800 further includes: sending performance data to a first device, wherein the performance data is repeatedly collected based on uplink data transmission.

[0215] In some example embodiments, method 800 further includes: receiving a request from a first device indicating a number of uplink data transmission repetitions associated with the ML model; and receiving uplink data transmission repetitions corresponding to the number from the first device during measurement intervals.

[0216] In some example embodiments, method 800 further includes: sending downlink data transmission to a first device using a first version; and during measurement intervals, sending downlink data transmission to the first device repeatedly using a second version of the ML model, wherein the second version of the ML model is different from the first version of the ML model.

[0217] In some example embodiments, method 800 further includes: receiving a request from a first device indicating a number of downlink data transmission repetitions associated with the ML model; and during measurement intervals, sending downlink data transmission repetitions corresponding to the number to the first device.

[0218] In some example embodiments, the performance data includes at least a first set of KPIs associated with a first version of the ML model, a second set of KPIs associated with a second version of the ML model, and a third set of KPIs associated with the transfer of data that does not use the ML model.

[0219] In some example implementations, the KPIs include at least one of the following: BLER; BER; SINR.

[0220] In some example embodiments, a first version of the ML model includes a predetermined version of the ML model, and a second version of the ML model includes an initial version of the ML model.

[0221] In some example embodiments, the version control configuration indicates at least one of the following: MGL; MGRP; measurement gap mode; a set of values ​​for the measurement gap.

[0222] In some example embodiments, the resource includes at least one predetermined time slot.

[0223] In some example implementations, uplink data transmission repetition includes PUSCH repetition, and downlink data transmission repetition includes PDSCH repetition.

[0224] In some example embodiments, at least one version includes at least one AI version associated with the ML model, and the transceiver includes a transceiver implemented with an ML model corresponding to the at least one AI version.

[0225] In some example embodiments, the first device includes a terminal device, and the second device includes a network device.

[0226] Figure 9 A flowchart of an example method 900 implemented at a device according to some example embodiments of the present disclosure is shown. For the purposes of discussion, [the following will be discussed]. Figure 1 The method 900 is described by the angle of the first device 110 in the middle.

[0227] At box 910, the device uses the first version of the ML model to receive downlink data transmission from the network device.

[0228] In box 920, the device receives a version control configuration from a network device that indicates measurement gaps and resources associated with downlink data transmission repetition, wherein the version control configuration is associated with an ML model.

[0229] At box 930, the device receives repeated downlink data transmissions associated with at least one version of the ML model from the network device during measurement intervals based on version control configuration.

[0230] At box 940, the device generates lifecycle management data for the ML model based on performance data associated with downlink data transmission and downlink data transmission repetition.

[0231] In some example embodiments, method 900 further includes receiving version control configuration from the network device via an RRC message.

[0232] In some example embodiments, method 900 further includes receiving an indicator from a network device, along with version control configuration, the indicator indicating activation of the measurement gap.

[0233] In some example embodiments, an indicator indicating the activation of the measurement gap is transmitted via DCI.

[0234] In some example embodiments, the DCI also includes an identifier that indicates the association between the HARQ process and downlink data transmission repetition.

[0235] In some example embodiments, the DCI also includes a validity timer that indicates the duration for which the measurement gap is valid.

[0236] In some example embodiments, method 900 further includes: if a version control configuration is received, during the measurement interval, receiving downlink data transmission repetitions from the network device using a second version of the ML model, wherein the second version of the ML model is different from the first version of the ML model; and collecting performance data based on the downlink data transmission repetitions.

[0237] In some example embodiments, method 900 further includes: transmitting a request to a network device indicating the number of downlink data transmission repetitions associated with the ML model; and receiving, during measurement intervals, downlink data transmission repetitions corresponding to the number from the network device.

[0238] In some example embodiments, the performance data includes at least a first set of KPIs associated with a first version of the ML model, a second set of KPIs associated with a second version of the ML model, and a third set of KPIs associated with data transmission that does not use the ML model.

[0239] In some example embodiments, method 900 further includes updating at least a first version of the ML model and a second version of the ML model, based at least on data used for lifecycle management.

[0240] In some example implementations, the KPIs include at least one of the following: BLER; BER; SINR.

[0241] In some example embodiments, a first version of the ML model includes a predetermined version of the ML model, and a second version of the ML model includes an initial version of the ML model.

[0242] In some example embodiments, the version control configuration indicates at least one of the following: MGL; MGRP; measurement gap mode; a set of values ​​for the measurement gap.

[0243] In some example embodiments, the resource includes at least one predetermined time slot.

[0244] In some example implementations, downlink data transmission repetition includes PDSCH repetition.

[0245] In some example embodiments, at least one version includes at least one AI version associated with the ML model, and the transceiver includes a transceiver implemented with an ML model corresponding to the at least one AI version.

[0246] In some example embodiments, the device includes a terminal device.

[0247] In some example embodiments, a first device capable of performing any method 700 (e.g., Figure 1 The first device 110 may include a component for performing the corresponding operation of method 700. This component may be implemented in any suitable form. For example, the component may be implemented in a circuit or software module. The first device may be implemented as or included in... Figure 1 In the first device 110.

[0248] In some example embodiments, the first device includes: components for receiving a version control configuration from a second device, the version control configuration indicating measurement gaps and resources, the resources being associated with at least one of: uplink data transmission duplication or downlink data transmission duplication; components for acquiring performance data of a machine learning ML model associated with at least one version during the measurement gap, wherein the performance data is collected based on at least one of: uplink data transmission duplication or downlink data transmission duplication associated with a transceiver; and components for generating data for lifecycle management of the ML model based on the collected performance data.

[0249] In some example embodiments, the first device further includes a component for receiving version control configuration from the second device via RRC messages.

[0250] In some example embodiments, the first device further includes a component for receiving an indicator from the second device, together with version control configuration, the indicator indicating activation of the measurement gap.

[0251] In some example embodiments, an indicator indicating the activation of the measurement gap is transmitted via downlink control information (DCI).

[0252] In some example embodiments, the DCI also includes an identifier that indicates an association between the HARQ process and at least one of the following: uplink data transmission duplication or downlink data transmission duplication.

[0253] In some example embodiments, the DCI also includes a validity timer that indicates the duration for which the measurement gap is valid.

[0254] In some example embodiments, the first device further includes: a component for sending uplink data transmission to the second device using a first version of the ML model; and a component for sending uplink data transmission repetition to the second device using a second version of the ML model during measurement intervals if a version control configuration is received, wherein the second version of the ML model is different from the first version of the ML model.

[0255] In some example embodiments, the first device further includes a component for receiving performance data from the second device, wherein the performance data is collected based on uplink data transmission repetition.

[0256] In some example embodiments, the first device further includes: a component for sending a request to the second device, the request indicating the number of uplink data transmission repetitions associated with the ML model; and a component for sending uplink data transmission repetitions corresponding to the number to the second device during measurement intervals.

[0257] In some example embodiments, the first device further includes: components for receiving downlink data transmission from the second device using a first version of the ML model; components for receiving repeated downlink data transmission from the second device using a second version of the ML model during measurement intervals if a version control configuration is received, wherein the second version of the ML model is different from the first version of the ML model; and components for collecting performance data based on repeated downlink data transmission.

[0258] In some example embodiments, the first device further includes: a component for sending a request to the second device, the request indicating the number of downlink data transmission repetitions associated with the ML model; and a component for receiving downlink data transmission repetitions corresponding to the number from the second device during measurement intervals.

[0259] In some example embodiments, the performance data includes at least a first set of key performance indicators (KPIs) associated with a first version of the ML model, a second set of KPIs associated with a second version of the ML model, and a third set of KPIs associated with the transfer of data that does not use the ML model.

[0260] In some example embodiments, the first apparatus further includes: a component for generating data for lifecycle management as labeled data for the ML model based on performance data; and a component for updating at least the first version of the ML model and the second version of the ML model based at least on the data for lifecycle management if it is determined based on a second set of KPIs and a third set of KPIs that the performance of the second version of the ML model is worse than the performance of data transmission without using the ML model.

[0261] In some example implementations, the KPIs include at least one of the following: BLER; BER; SINR.

[0262] In some example embodiments, a first version of the ML model includes a predetermined version of the ML model, and a second version of the ML model includes an initial version of the ML model.

[0263] In some example embodiments, the version control configuration indicates at least one of the following: MGL; MGRP; measurement gap mode; a set of values ​​for the measurement gap.

[0264] In some example embodiments, the resource includes at least one predetermined time slot.

[0265] In some example implementations, uplink data transmission repetition includes PUSCH repetition, and downlink data transmission repetition includes PDSCH repetition.

[0266] In some example embodiments, at least one version includes at least one AI version associated with the ML model, and the transceiver includes a transceiver implemented with an ML model corresponding to the at least one AI version.

[0267] In some example embodiments, the first device includes a terminal device, and the second device includes a network device.

[0268] In some example embodiments, a second device capable of performing any method 800 (e.g., Figure 1 The second device 120 may include a component for performing the corresponding operation of method 800. This component may be implemented in any suitable form. For example, the component may be implemented in a circuit or software module. The second device may be implemented as or included in... Figure 1 The second device 120 in the middle.

[0269] In some example embodiments, the second device includes: components for sending a version control configuration to the first device, the version control configuration indicating a measurement gap and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; and components for receiving uplink data transmission repetition associated with at least one version of the ML model from the first device during the measurement gap; or components for sending downlink data transmission repetition associated with at least one version of the ML model to the first device during the measurement gap.

[0270] In some example embodiments, the second device further includes a component for transmitting version control configuration to the first device via RRC messages.

[0271] In some example embodiments, the second device further includes a component for sending an indicator to the first device, together with version control configuration, the indicator indicating activation of the measurement gap.

[0272] In some example embodiments, an indicator indicating the activation of the measurement gap is transmitted via DCI.

[0273] In some example embodiments, the DCI also includes an identifier that indicates the association between the HARQ process and at least one of uplink data transmission duplication or downlink data transmission duplication.

[0274] In some example embodiments, the DCI also includes a validity timer that indicates the duration for which the measurement gap is valid.

[0275] In some example embodiments, the second device further includes: components for receiving uplink data transmission from the first device using a first version of the ML model; components for transmitting version control configuration to the first device; and components for repeatedly receiving uplink data transmission from the first device using a second version of the ML model during measurement intervals, wherein the second version of the ML model is different from the first version of the ML model.

[0276] In some example embodiments, the second device further includes a component for transmitting performance data to the first device, wherein the performance data is repeatedly collected based on uplink data transmission.

[0277] In some example embodiments, the second device further includes: a component for receiving a request from the first device indicating a number of uplink data transmission repetitions associated with the ML model; and a component for receiving, during measurement intervals, uplink data transmission repetitions corresponding to that number from the first device.

[0278] In some example embodiments, the second device further includes: a component for transmitting downlink data transmission to the first device using a first version; and a component for repeatedly transmitting downlink data transmission to the first device using a second version of the ML model during measurement intervals, wherein the second version of the ML model is different from the first version of the ML model.

[0279] In some example embodiments, the second device further includes: a component for receiving a request from the first device, the request indicating the number of downlink data transmission repetitions associated with the ML model; and a component for transmitting, during measurement intervals, downlink data transmission repetitions corresponding to the number to the first device.

[0280] In some example embodiments, the performance data includes at least a first set of KPIs associated with a first version of the ML model, a second set of KPIs associated with a second version of the ML model, and a third set of KPIs associated with the transfer of data that does not use the ML model.

[0281] In some example implementations, the KPIs include at least one of the following: BLER; BER; SINR.

[0282] In some example embodiments, a first version of the ML model includes a predetermined version of the ML model, and a second version of the ML model includes an initial version of the ML model.

[0283] In some example embodiments, the version control configuration indicates at least one of the following: MGL; MGRP; measurement gap mode; a set of values ​​for the measurement gap.

[0284] In some example embodiments, the resource includes at least one predetermined time slot.

[0285] In some example implementations, uplink data transmission repetition includes PUSCH repetition, and downlink data transmission repetition includes PDSCH repetition.

[0286] In some example embodiments, at least one version includes at least one AI version associated with the ML model, and the transceiver includes a transceiver implemented with an ML model corresponding to the at least one AI version.

[0287] In some example embodiments, the first device includes a terminal device, and the second device includes a network device.

[0288] In some example embodiments, the apparatus capable of performing any method 900 (e.g., Figure 1 The first device 110 may include a component for performing the corresponding operation of method 900. This component may be implemented in any suitable form. For example, the component may be implemented in a circuit or software module. The device may be implemented as or included in... Figure 1 In the first device 110.

[0289] In some example embodiments, the apparatus includes: components for receiving downlink data transmission from a network device using a first version of an ML model; components for receiving a version control configuration from the network device, the version control configuration indicating measurement gaps and resources associated with downlink data transmission repetitions, wherein the version control configuration is associated with an ML model; components for receiving downlink data transmission repetitions associated with at least one version of the ML model from the network device during measurement gaps based on the version control configuration; and components for generating data for lifecycle management of the ML model based on performance data associated with the downlink data transmission and the downlink data transmission repetitions.

[0290] In some example embodiments, the apparatus further includes a component for receiving version control configuration from a network device via RRC messages.

[0291] In some example embodiments, the apparatus further includes a component for receiving an indicator from a network device, together with version control configuration, the indicator indicating activation of a measurement gap.

[0292] In some example embodiments, an indicator indicating the activation of the measurement gap is transmitted via DCI.

[0293] In some example embodiments, the DCI also includes an identifier that indicates the association between the HARQ process and downlink data transmission repetition.

[0294] In some example embodiments, the DCI also includes a validity timer that indicates the duration for which the measurement gap is valid.

[0295] In some example embodiments, the apparatus further includes: components for receiving downlink data transmission duplicates from a network device using a second version of the ML model during measurement intervals if a version control configuration is received, wherein the second version of the ML model is different from the first version of the ML model; and components for collecting performance data based on downlink data transmission duplicates.

[0296] In some example embodiments, the apparatus further includes: a component for transmitting a request to a network device, the request indicating the number of downlink data transmission repetitions associated with the ML model; and a component for receiving, during measurement intervals, downlink data transmission repetitions corresponding to the number from the network device.

[0297] In some example embodiments, the performance data includes at least a first set of KPIs associated with a first version of the ML model, a second set of KPIs associated with a second version of the ML model, and a third set of KPIs associated with data transmission that does not use the ML model.

[0298] In some example embodiments, the apparatus further includes a component for updating at least a first version of the ML model and a second version of the ML model, based at least on data used for lifecycle management.

[0299] In some example implementations, the KPIs include at least one of the following: BLER; BER; SINR.

[0300] In some example embodiments, a first version of the ML model includes a predetermined version of the ML model, and a second version of the ML model includes an initial version of the ML model.

[0301] In some example embodiments, the version control configuration indicates at least one of the following: MGL; MGRP; measurement gap mode; a set of values ​​for the measurement gap.

[0302] In some example embodiments, the resource includes at least one predetermined time slot.

[0303] In some example implementations, downlink data transmission repetition includes PDSCH repetition.

[0304] In some example embodiments, at least one version includes at least one AI version associated with the ML model, and the transceiver includes a transceiver implemented with an ML model corresponding to the at least one AI version.

[0305] In some example embodiments, the device includes a terminal device.

[0306] Figure 10 This is a simplified block diagram of a device 1000 suitable for implementing exemplary embodiments of the present disclosure. Device 1000 can be used to implement a communication device, for example, as... Figure 1 The first device 110 or the second device 120 shown. As shown, the device 1000 includes one or more processors 1010, one or more memories 1020 coupled to the processors 1010, and one or more communication modules 1040 coupled to the processors 1010.

[0307] Communication module 1040 is used for bidirectional communication. Communication module 1040 has one or more communication interfaces to facilitate communication with one or more other modules or devices. The communication interface can represent any interface necessary for communication with other network elements. In some example embodiments, communication module 1040 may include at least one antenna.

[0308] As a non-limiting example, processor 1010 can be any type suitable for a local technology network and can include one or more of the following: general-purpose computer, special-purpose computer, microprocessor, digital signal processor (DSP), and processor based on a multi-core processor architecture. Device 1000 can have multiple processors, such as application-specific integrated circuit chips, which are time-dependent on the clock of a synchronous main processor.

[0309] Memory 1020 may include one or more non-volatile memories and one or more volatile memories. Examples of non-volatile memories include, but are not limited to, read-only memory (ROM) 1024, electrically programmable read-only memory (EPROM), flash memory, hard disk, optical disc (CD), digital video disc (DVD), optical disc, laser disc, and other magnetic and / or optical storage. Examples of volatile memories include, but are not limited to, random access memory (RAM) 1022 and other volatile memories that will not be maintained during the duration of a power outage.

[0310] Computer program 1030 includes computer-executable instructions that are executed by an associated processor 1010. The instructions of program 1030 may include instructions for performing operations / actions of some example embodiments of this disclosure. Program 1030 may be stored in memory, such as ROM 1024. Processor 1010 can perform any suitable actions and processes by loading program 1030 into RAM 1022.

[0311] Example embodiments of this disclosure can be implemented using program 1030, enabling device 1000 to execute as described in the reference. Figures 2 to 9 Any process discussed in this disclosure. Exemplary embodiments of this disclosure may also be implemented by hardware or by a combination of software and hardware.

[0312] In some example embodiments, program 1030 may be tangibly contained in a computer-readable medium, which may be included in device 1000 (such as in memory 1020) or other storage device accessible by device 1000. Device 1000 may load program 1030 from the computer-readable medium into RAM 1022 for execution. In some example embodiments, the computer-readable medium may include any type of non-transitory storage medium, such as ROM, EPROM, flash memory, hard disk, CD, DVD, etc. As used herein, the term "non-transitory" is a limitation on the medium itself (i.e., tangible, not tactile) rather than on the persistence of data storage (e.g., RAM versus ROM).

[0313] Figure 11 An example of a computer-readable medium 1100 is shown, which may be in the form of a CD, DVD, or other optical storage disc. A program 1030 is stored on the computer-readable medium 1100.

[0314] Generally, the various embodiments of this disclosure can be implemented in hardware or dedicated circuitry, software, logic, or any combination thereof. Some aspects may be implemented in hardware, while others may be implemented in firmware or software, which may be executed by a controller, microprocessor, or other computing device. Although various aspects of the embodiments of this disclosure are shown and described as block diagrams, flowcharts, or using some other graphical representation, it should be understood that, as non-limiting examples, the blocks, apparatuses, systems, techniques, or methods described herein may be implemented in hardware, software, firmware, dedicated circuitry or logic, general-purpose hardware or controllers or other computing devices, or some combination thereof.

[0315] Some exemplary embodiments of this disclosure also provide at least one computer program product tangibly stored on a computer-readable medium, such as a non-transitory computer-readable medium. The computer program product includes computer-executable instructions, such as those included in a program module, which execute in a device on a target physical or virtual processor to perform any of the methods described above. Typically, a program module includes routines, programs, libraries, objects, classes, components, data structures, etc., that perform a particular task or implement a particular abstract data type. In various embodiments, the functionality of a program module can be combined or split among program modules as needed. The machine-executable instructions for a program module can execute within a local or distributed device. In a distributed device, the program module can reside on both local and remote storage media.

[0316] Program code used to perform the methods of this disclosure may be written in any combination of one or more programming languages. The program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code enables the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a stand-alone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0317] In the context of this disclosure, computer program code or related data may be carried by any suitable carrier to enable a device, apparatus, or processor to perform the various processes and operations described above. Examples of carriers include signals, computer-readable media, etc.

[0318] Computer-readable media can be computer-readable signal media or computer-readable storage media. Computer-readable media can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any suitable combination thereof. More specific examples of computer-readable storage media will include electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable optical disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0319] Furthermore, although operations are described in a specific order, this should not be construed as requiring that such operations be performed in the specific order shown or sequentially, or that all shown operations be performed to achieve the desired result. In some cases, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure, but rather as descriptions of features that may be specified in a particular embodiment. Unless explicitly stated otherwise, certain features described in the context of a single embodiment may also be implemented in combination in a single embodiment. Conversely, unless explicitly stated otherwise, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0320] Although this disclosure has been described in language specific to structural features and / or methodological actions, it should be understood that the disclosure as defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are disclosed as exemplary forms for implementing the claims.

[0321] Example 1. A first means for communication, comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the first means to at least: receive a version control configuration from a second means, the version control configuration indicating measurement gaps and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; acquire, during the measurement gap, performance data of a machine learning ML model associated with at least one version, wherein the performance data is collected based on at least one of: uplink data transmission repetition or downlink data transmission repetition associated with a transceiver; and generate data for lifecycle management of the ML model based on the collected performance data.

[0322] Example 2. According to the first apparatus of Example 1, wherein the first apparatus is configured to receive version control configuration from the second apparatus via a Radio Resource Control (RRC) message.

[0323] Example 3. According to the first device of Example 2, wherein the first device is configured to receive an indicator from the second device, together with the version control configuration, the indicator indicating the activation of the measurement gap.

[0324] Example 4. According to the first device of Example 2, an indicator indicating the activation of the measurement gap is transmitted via downlink control information (DCI).

[0325] Example 5. According to the first apparatus of Example 4, wherein the DCI further includes an identifier indicating an association between the Hybrid Automatic Repeat Request (HARQ) procedure and at least one of the following: uplink data transmission duplication or downlink data transmission duplication.

[0326] Example 6. According to the first device of Example 4, the DCI further includes an validity timer that indicates the duration for which the measurement gap is valid.

[0327] Example 7. A first device according to any one of Examples 1-6, wherein the first device is configured to: send uplink data transmission to a second device using a first version of the ML model; and if a version control configuration is received, send uplink data transmission to the second device repeatedly using a second version of the ML model during measurement intervals, wherein the second version of the ML model is different from the first version of the ML model.

[0328] Example 8. According to the first device of Example 7, wherein the first device is configured to: receive performance data from the second device, wherein the performance data is collected based on uplink data transmission repetition.

[0329] Example 9. According to the first apparatus of Example 7, wherein the first apparatus is configured to: send a request to the second apparatus indicating the number of uplink data transmission repetitions associated with the ML model; and during measurement intervals, send uplink data transmission repetitions corresponding to the number to the second apparatus.

[0330] Example 10. A first apparatus according to any one of Examples 1-6, wherein the first apparatus is configured to: receive downlink data transmission from a second apparatus using a first version of an ML model; if a version control configuration is received, receive downlink data transmission from the second apparatus repeatedly using a second version of an ML model during measurement intervals, wherein the second version of the ML model is different from the first version of the ML model; and collect performance data based on the repeated downlink data transmission.

[0331] Example 11. According to the first apparatus of Example 10, wherein the first apparatus is configured to: send a request to the second apparatus indicating the number of downlink data transmission repetitions associated with the ML model; and during measurement intervals, receive downlink data transmission repetitions corresponding to the number from the second apparatus.

[0332] Example 12. A first apparatus according to any one of Examples 1-6, wherein the performance data includes at least a first set of key performance indicators (KPIs) associated with a first version of the ML model, a second set of KPIs associated with a second version of the ML model, and a third set of KPIs associated with the transfer of data without using the ML model; and wherein the first apparatus is configured to: generate data for lifecycle management as labeled data for the ML model based on the performance data; and if it is determined based on the second set of KPIs and the third set of KPIs that the performance of the second version of the ML model is worse than the performance of the transfer of data without using the ML model, at least update the first version of the ML model and the second version of the ML model based on the data for lifecycle management.

[0333] Example 13. A second means for communication, comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the second means to at least: send a version control configuration to a first means, the version control configuration indicating a measurement gap and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; and receive uplink data transmission repetition associated with at least one version of a machine learning ML model from the first means during the measurement gap; or send downlink data transmission repetition associated with at least one version of the ML model to the first means during the measurement gap.

[0334] Example 14. A method for communication, comprising: receiving, by a first device, a version control configuration from a second device, the version control configuration indicating a measurement gap and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; acquiring, during the measurement gap, performance data of a machine learning ML model associated with at least one version, wherein the performance data is collected based on at least one of: uplink data transmission repetition or downlink data transmission repetition associated with a transceiver; and generating, based on the collected performance data, data for lifecycle management of the ML model.

[0335] Example 15. A method for communication, comprising: sending a version control configuration from a second device to a first device, the version control configuration indicating a measurement gap and resources, the resources being associated with at least one of: uplink data transmission repetition or downlink data transmission repetition; and receiving, during the measurement gap, uplink data transmission repetition associated with at least one version of a machine learning ML model from the first device; or sending, during the measurement gap, downlink data transmission repetition associated with at least one version of the ML model to the first device.

Claims

1. A first device for communication, comprising: At least one processor; as well as At least one memory, the at least one memory storing instructions, the instructions, when executed by the at least one processor, cause the first device to at least: The second device receives a version control configuration that indicates measurement gaps and resources, the resources being associated with at least one of the following: uplink data transmission duplication or downlink data transmission duplication; During the measurement interval, performance data of a machine learning ML model associated with at least one version is acquired, wherein the performance data is collected based on at least one of the following associated with the transceiver: uplink data transmission duplication or downlink data transmission duplication; as well as Based on the collected performance data, data for the lifecycle management of the ML model is generated.

2. The first device according to claim 1, wherein the first device is configured to: The version control configuration is received from the second device via Radio Resource Control (RRC) messages.

3. The first device according to claim 2, wherein the first device is configured to: Along with the version control configuration, an indicator is received from the second device, the indicator indicating the activation of the measurement gap.

4. The first apparatus of claim 2, wherein an indicator indicating activation of the measurement gap is transmitted via downlink control information (DCI).

5. The first apparatus of claim 4, wherein the DCI further comprises an identifier indicating an association between the Hybrid Automatic Repeat Request (HARQ) process and at least one of the following: uplink data transmission duplication or downlink data transmission duplication.

6. The first apparatus of claim 4, wherein the DCI further comprises a validity timer indicating the duration for which the measurement gap is valid.

7. The first device according to any one of claims 1-6, wherein the first device is configured to: Sending uplink data transmission to the second device using the first version of the ML model; and If the version control configuration is received, during the measurement interval, the uplink data transmission repeat is sent to the second device using a second version of the ML model, wherein the second version of the ML model is different from the first version of the ML model.

8. The first device according to claim 7, wherein the first device is configured to: The performance data is received from the second device, wherein the performance data is collected based on the repetition of the uplink data transmission.

9. The first device according to claim 7, wherein the first device is configured such that: A request is sent to the second device, the request indicating the number of repetitions of the uplink data transmission associated with the ML model; and During the measurement interval, the uplink data transmissions corresponding to the stated number are repeatedly sent to the second device.

10. The first device according to any one of claims 1-6, wherein the first device is configured to: Receive downlink data transmission from the second device using the first version of the ML model; If the version control configuration is received, during the measurement interval, the downlink data transmission is repeatedly received from the second device using a second version of the ML model, wherein the second version of the ML model is different from the first version of the ML model; as well as The performance data is collected repeatedly based on the downlink data transmission.

Images