Systems and devices for model training in a network and associated methods thereof

Abstract

Systems (100), devices (102), and methods (300) for model training in a network are disclosed. The method (300) includes configuring data indicative of model training, communicating the data to a user device, initiating model training based on the data, and generating a plurality of parameters related to the model training.

Description

Technical Field

[0001] This disclosure generally relates to one or both of a system and a device for training a model in a network that is associated with, for example, a user equipment (UE) that can be used for communication. This disclosure further relates to a method that can be associated with the system and / or the device. Background Technology

[0002] Generally, energy efficiency and energy savings are helpful or desirable in communication networks. An example of a communication network would be a telecommunications network based on the 3rd Generation Partnership Project (3GPP) 5G (Fifth Generation) New Radio (NR) standard.

[0003] Typically, the conventional techniques used for model training (e.g., artificial intelligence machine learning (AI / ML) training, the entity executing the model (e.g., AI / ML model, such as a user equipment (UE) or network)) may not know the dataset identifiers associated with a particular model for correct inference.

[0004] This disclosure envisions that it would be helpful to address (or at least alleviate) one or more problems associated with conventional technologies used to promote energy efficiency and energy savings. Summary of the Invention

[0005] According to a first aspect of the present invention, a method for model training in a network is provided, comprising: configuring data to instruct model training; transmitting the data to a user device; initiating model training based on the data; and generating a plurality of parameters related to the model training.

[0006] Advantageously, the method described herein can optimize the network and provide an energy-efficient network by associating model datasets with model identifiers and thus avoiding dataset mismatches. This method also provides a fundamental mechanism for interoperability and data flow in AI / ML-enabled radio access network cooperation. AI / ML performance for wireless communication can be improved based on gNB-UE cooperative operation for AI / ML support.

[0007] In this embodiment, the data includes model identifiers and dataset identifiers.

[0008] In this embodiment, several parameters include a model identifier and an associated dataset identifier.

[0009] In an embodiment, the method further includes transmitting multiple parameters via physical layer (L1) signaling and / or data link layer (L2) signaling.

[0010] In an embodiment, the method further includes transmitting multiple parameters via user equipment (UE) specific messages and / or system information blocks (SIBs) based on model configuration.

[0011] In one embodiment, the method further includes validating multiple parameters used for model training and inference.

[0012] In this embodiment, generating multiple parameters includes periodic generation and / or generation based on pre-configured events.

[0013] In this embodiment, model training includes artificial intelligence machine learning (AI / ML) model training.

[0014] In an embodiment, transmitting multiple parameters includes transmitting via at least one of the following: system information messages and / or dedicated RRC messages.

[0015] In one embodiment, a computer program (not shown) is provided, which may include instructions that, when executed by a computer (not shown), cause the computer to perform the method of the first aspect.

[0016] In one embodiment, a computer-readable storage medium is provided that stores data representing software executable by a computer, the software including instructions that, when executed by the computer, are used to implement the method of the first aspect.

[0017] According to a second aspect of this disclosure, an apparatus for model training in a network is provided, comprising: a first module configured to receive at least one input signal associated with data instructing model training; a second module configured to perform processing and facilitate at least one of the methods of the first aspect to generate at least one output signal; and a third module configured to transmit at least one output signal, wherein the output signal corresponds to a control signal for model training in the network.

[0018] In an embodiment, the device may correspond to a user equipment (UE) that can communicate with a device corresponding to a base station. The base station may, for example, correspond to a next-generation node B (gNB) that can be configured to transmit one or more signals (e.g., input signals) to the UE.

[0019] In one embodiment, a system is provided that includes one or more devices and one or more means. The devices and means may be coupled, for example, via wired coupling and / or wireless coupling.

[0020] Advantageously, this system allows the UE (user device) or base station (gNB or network) executing a model (e.g., an AI / ML model) to know the dataset identifier associated with that particular model for correct inference. This ensures dataset consistency between both sides of the model (e.g., an AI / ML model) during model training and inference. Attached Figure Description

[0021] Embodiments of this disclosure are described below with reference to the accompanying drawings, in which:

[0022] Figure 1A A schematic diagram of an exemplary system for model training in a network according to an embodiment of the present disclosure is shown, the system including at least one device.

[0023] Figure 1B The embodiments of the present disclosure are shown. Figure 1A Example scenarios associated with the system.

[0024] Figure 2 Further detailed illustrations of embodiments according to this disclosure are shown. Figure 1A A schematic diagram of the equipment.

[0025] Figure 3 The embodiments of the present disclosure are shown. Figure 1A The system-related methods.

[0026] Figures 4A to 4D Examples of embodiments according to this disclosure are shown. Figure 3 A diagram illustrating example scenarios associated with the method. Detailed Implementation

[0027] The detailed description set forth below with reference to the accompanying drawings is intended as a description of various configurations and is not intended to represent only the configurations in which the concepts described herein may be practiced. The detailed description includes specific details for the purpose of providing a thorough understanding of the various concepts. However, it will be apparent to those skilled in the art that these concepts may be practiced without these specific details. Specifically, although terms from 3GPP 5G NR may be used in this disclosure to exemplify embodiments herein, this should not be construed as limiting the scope of the invention.

[0028] In addition, some of the embodiments contemplated herein will now be described more fully with reference to the accompanying drawings. However, other embodiments are also included within the scope of the subject matter disclosed herein, and the disclosed subject matter should not be construed as being limited to the embodiments set forth herein; rather, these embodiments are provided by way of example to convey the scope of the subject matter to those skilled in the art.

[0029] Generally, all terms used herein should be interpreted according to their ordinary meaning in the relevant art, unless a different meaning is expressly given and / or implied in the context of their use. Unless otherwise expressly stated, all references to elements, devices, components, means, steps, etc., should be openly interpreted as referring to at least one instance of that element, device, component, means, step, etc. The steps of any method disclosed herein need not be performed in the exact order disclosed, unless a step is explicitly described as following or preceding another step, and / or it implies that a step must follow or precede another step. Where appropriate, any feature of any embodiment of the embodiments disclosed herein may be applied to any other embodiment. Similarly, any advantage of any embodiment in the embodiments may be applied to any other embodiment, and vice versa. Other objects, features, and advantages of the appended embodiments will become apparent from the following description.

[0030] In some embodiments, the non-limiting terms User Equipment (UE) or Wireless Device or User Equipment may be used, and the term may refer to any type of wireless device communicating with a network node and / or with another UE in a cellular or mobile communication system. Examples of UEs are target devices, device-to-device (D2D) UEs, machine-type UEs or UEs capable of machine-to-machine (M2M) communication, PDAs, PADs, tablet computers, mobile terminals, smartphones, laptop embedded devices (LEE), laptop mounted devices (LME), USB dongles, UE class M1, UE class M2, ProSe UE, V2V UE, V2X UE, etc.

[0031] In some embodiments, the more general term "network node" may be used, and this term may correspond to any type of radio network node or any network node that communicates with user equipment (directly or via another node) and / or with another network node. Examples of network nodes are NodeB, MeNB, ENB, network nodes belonging to MCG or SCG, base stations (BS), multi-standard radio (MSR) radio nodes (such as MSR BS), eNodeB, gNodeB, network controller, radio network controller (RNC), base station controller (BSC), repeater, donor node controlling repeater, base transceiver station (BTS), access point (AP), transport point, transport node, RRU, RRH, nodes in distributed antenna system (DAS), core network nodes (e.g., mobile switching center (MSC), mobility management entity (MME), etc.), operations and maintenance (O&M), operations support system (OSS), self-optimizing network (SON), location nodes (e.g., evolved servicing mobile location center (E-SMLC)), minimized drive test (MDT), test equipment (physical node or software), etc.

[0032] Furthermore, terms such as base station / gNodeB and UE should be considered non-restrictive and, in particular, do not imply any hierarchical relationship between the two; generally, "gNodeB" can be considered device 1 and "UE" can be considered device 2, and the two devices communicate with each other via a radio channel. Also, in the following text, the sender or receiver can be either a gNodeB (gNB) or a UE.

[0033] Furthermore, the features, structures, or characteristics described in the embodiments can be combined in any suitable manner. Numerous specific details, such as examples of programming, software modules, user selection, network transactions, database queries, database structures, hardware modules, hardware circuits, hardware chips, etc., are provided in the following description to provide a thorough understanding of the embodiments. However, those skilled in the art will recognize that the embodiments can be practiced without one or more of these specific details or using other methods, components, materials, etc. In other instances, well-known structures, materials, or operations are not shown or described in detail to avoid obscuring aspects of the embodiments. Throughout the specification, references to “an embodiment,” “embodiment,” or similar language mean that a particular feature, structure, or characteristic described in connection with the embodiment is included in at least one embodiment. Therefore, unless expressly stated otherwise, the phrases “in one embodiment,” “in an embodiment,” and similar language appearing throughout the specification may, but not necessarily all, refer to the same embodiment, but rather mean “one or more, but not all, embodiments.” Unless expressly stated otherwise, the terms “comprising,” “including,” “having,” and variations thereof mean “including, but not limited to,” “including.” Unless expressly stated otherwise, the enumeration list of items does not imply that any or all of the items are mutually exclusive. Unless otherwise expressly specified, the terms “a,” “one,” and “the” also mean “one or more.”

[0034] The following explanation provides a detailed description of a mechanism for pre-configuring and signaling specific information about online model training by configuring a set of UE behaviors. Specifically, this disclosure envisions that model-based technologies (e.g., AI / ML models) are currently applicable to many different applications and can be applied to multiple use cases based on observed potential benefits. The model (e.g., AI / ML) lifecycle can be broken down into several phases, such as data collection / preprocessing, model training, model testing / validation, model deployment / update, model monitoring, etc., where each phase is equally important for achieving the target performance of any particular model. Managing the lifecycle of an AI / ML model can be a challenging problem when applying it to any use case or application. This is primarily because data / model drift occurs during model deployment / inference and can lead to performance degradation of the AI / ML model. Fundamentally, dataset statistics may change after the model (e.g., AI / ML model) is deployed, and model inference capabilities may also be affected by unseen data as input. In a similar respect, this disclosure envisions that the statistical properties of the dataset and the relationship between the input and output of the trained model may change as drift occurs.

[0035] This disclosure also envisions that model training or retraining can be a critical issue in maintaining model performance, as model performance, such as inference and / or training, can depend on different model execution environments with varying configuration parameters. To address such situations, this disclosure envisions that collaboration between the UE and gNB may be crucial for tracking model performance and reconfiguring models to correspond to different environments. Therefore, this disclosure envisions that models (e.g., AI / ML models) may require model monitoring after deployment, as model performance may not be continuously maintained due to drift. Update feedback can then be provided to retrain or update the model or select an alternative model. When deploying wireless communication networks that support models (e.g., AI / ML models), it may be important to consider how to handle active AI / ML models using reconfiguration for wireless devices during operations such as model training, inference, and updates. Since the UE's Radio Resource Control (RRC) state can vary depending on UE mobility or network / device condition states, any AI / ML models supported by the UE may be affected for machine learning (ML) operations. In such methods, a finite set of multi-model combinations associated with UE RRC states (e.g., active, inactive, idle) can be pre-configured, and a mapping table can be generated to indicate different combinations of multiple models to be activated via SI or dedicated RRC signaling or Layer 1 / Layer 2 (L1 / L2) signaling based on ML configuration with changing UE RRC states.

[0036] This disclosure further envisions that, for different model combinations, each combination of multi-model sets with different UE RRC states can be periodically updated or reconfigured depending on the state of the applicable model at the UE and / or the network state with ML conditions. The activation or deactivation of the applied multi-model combination can be dynamically indicated via L1 / L2 or RRC signaling. Additionally, a preset timer can be configured to determine the activation or deactivation of the applied multi-model combination based on the timer. Furthermore, different combinations of multi-model sets can be configured for each RRC state. For each combination of multi-model sets used for activation, different model behaviors can be configured for lifecycle management (LCM) operations as two-sided or one-sided models, or they can be associated in the mapping information (if necessary) by using indexes or identifiers. Therefore, for the mapping table of multi-model combinations and RRC states, additional information such as descriptions of model behaviors and / or LCM operations can be associated as ML auxiliary information. Regarding the use of multi-model combinations to activate any specific model, a UE-autonomous decision to activate the UE model can be performed based on the mapping relationship between the multi-model combinations and RRC states. In addition, when the UE provides ML measurements or condition updates with supported model states, the network side can send an indication message to activate the UE model based on the mapping relationship between multiple model combinations and RRC states.

[0037] This disclosure envisions that model identification can be categorized into the following types of AI / ML model identification for the UE side or UE portion of a two-sided model. An example of a model identification type includes type A, where a model can be identified for both the network (if applicable) and the UE (if applicable) without over-the-air signaling. In this example embodiment, a model identifier can be assigned to the model during model identification, and this model identifier can be referenced or used in over-the-air signaling after model identification.

[0038] A second example of a model identification type could be type B, where the model is identified via over-the-air signaling. A third example could be type B1, where model identification can be initiated by the UE, and the network assists in the remaining steps of model identification (if any), and a model identifier can be assigned to the model during model identification. Yet another example could be type B2, where model identification can be initiated by the network, and the UE responds (if applicable) to the remaining steps of model identification. A model identifier can then be assigned to the model during model identification, which may not imply that model identification is necessary.

[0039] This disclosure further envisions that the following network-UE cooperation levels can be considered as aspects for defining cooperation levels. A first cooperation level could be level x, where cooperation is absent. A second cooperation level could be level y, where signaling-based cooperation exists but model transfer is not. Such levels could include cases where there is no model delivery. A third cooperation level could be level z, where signaling-based cooperation exists and model transfer is present.

[0040] This disclosure further envisions that model identification with data collection-related configurations and / or indications (MI-Option 1) of Model Identification (MI) Type B can be further explored in the following aspects: the relationship between the model ID and the data collection-related configurations and / or indications, information transmitted from the network to the UE (if any), information transmitted from the UE to the network (if any), any associated procedures and uses, and applicable use cases for MI-Option 1.

[0041] According to embodiments of this disclosure, this disclosure generally envisions promoting, for example (e.g., associated with 3GPP-based standards / specifications, etc.) network and / or user equipment (UE) efficiency (e.g., energy / power efficiency).

[0042] This disclosure envisions the possibility of ensuring consistency of datasets used for training and inference of two-sided models (e.g., AI / ML models). Specifically, the entity (UE or network) performing the AI / ML model may know the dataset identifier associated with a particular model for correct inference. In example embodiments, the model identifier and dataset identifier may be included together with the dataset used for training and inference of the two-sided model (e.g., AI / ML model).

[0043] According to embodiments of this disclosure, the manner described above may be able to improve power and energy consumption efficiency.

[0044] The foregoing will be discussed in further detail below with reference to Figures 1 to 4.

[0045] refer to Figure 1A The present disclosure illustrates a system 100 for model training in a network according to embodiments thereof. According to embodiments thereof, system 100 may be adapted, for example, to save energy and promote energy / power efficiency in a network.

[0046] As shown in the figure, according to an embodiment of the present disclosure, system 100 may include one or more devices 102, at least one apparatus 104, and optionally a communication network 106.

[0047] Device 102 may be coupled to device 104. Specifically, according to embodiments of the present disclosure, device 102 may be coupled to device 104, for example, via communication network 106.

[0048] In one embodiment, device 102 may be coupled to communication network 106, and device 104 may be coupled to communication network 106. Coupling may be performed via wired coupling and wireless coupling, or both. According to embodiments of this disclosure, device 102 may generally be configured to communicate with device 104 via communication network 106.

[0049] According to embodiments of this disclosure, device 102 may, for example, be associated with / correspond to / include one or more user equipments (UEs) that may carry one or more computers. For example, according to embodiments of this disclosure, device 102 may correspond to a UE carrying at least one computer (e.g., according to embodiments of this disclosure, an electronic device / module with computing capabilities, such as an electronic mobile device that can be carried in a vehicle or an electronic module that can be installed in a vehicle), which may be configured to perform one or more processing tasks associated with adaptive / dynamic / progressive control. In a more specific example, according to embodiments of this disclosure, in one embodiment, device 102 may include one or more processors (not shown) that may be configured to perform one or more processing tasks associated with dynamic / adaptive / progressive control. In one embodiment, device 102 may, for example, be configured to receive one or more input signals and perform at least one processing task based on the input signals in a manner that generates one or more output signals. According to embodiments of this disclosure, input signals may, for example, be transmitted from device 104 and received by device 102. As a possible option, according to embodiments of this disclosure, output signals may, for example, be transmitted from device 102. According to embodiments of this disclosure, reference will be made later. Figure 2 Let’s discuss device 102 in more detail.

[0050] Device 104 may be associated with / correspond to at least one base station (e.g., at least one gNB). Furthermore, device 104 may be configured, for example, to carry / associate with / include one or more computers (e.g., electronic devices / modules with computing capabilities), which may be configured, for example, to perform one or more processing tasks associated with the base station. According to embodiments of this disclosure, device 104 may be configured to generate one or more input signals that can be transmitted to device 102. This will be discussed in further detail later in the context of example scenarios according to embodiments of this disclosure.

[0051] Communication network 106 may correspond, for example, to an Internet communication network, a cellular communication network, a wired communication network, a Global Navigation Satellite System (GNSS) communication network, a wireless communication network, or any combination thereof. Communication via communication network 106 (e.g., between devices 102 and / or between devices 102 and apparatus 104) may be conducted through one or both of wired and wireless communication.

[0052] As previously mentioned, device 102 may, for example, be configured to receive at least one input signal and perform at least one processing task associated with dynamic / adaptive / progressive control on the input signal in a manner that generates at least one output signal. Furthermore, according to embodiments of this disclosure, device 104 may, for example, be configured to generate (and transmit) an input signal to device 102. Embodiments of this disclosure will be described below with reference to... Figure 1B This will be discussed in the context of the example scenario.

[0053] Figure 1B An embodiment of the invention is shown with Figure 1A Example scenarios associated with the system. Specifically, Figure 1B An example embodiment for training an artificial intelligence machine learning (AI / ML) model is illustrated. Referring to the accompanying drawings, the gNB (or base station) and the corresponding UE (or user equipment) in the network may have the same machine learning model (e.g., an AI / ML model). The gNB (or base station) may transmit the relevant model to the UE (or user equipment) in the network, and each UE may then generate trained weights and transmit them back to the gNB (or base station). This disclosure envisions that AI / ML for the air interface in a general AI / ML framework may involve the collection of UE-side model training data. The framework may also include: identifying the corresponding content of UE data collection, and analyzing the UE data collection mechanism along with the effects and limitations of each method.

[0054] According to embodiments of this disclosure, as will be discussed in further detail in the context of an example scenario associated with system 100 according to embodiments of this disclosure, this disclosure envisions that considering some form of dynamic / adaptive / progressive configuration / determination strategy that may contribute to power / energy consumption efficiency may be helpful. According to embodiments of this disclosure, the dynamic / adaptive / progressive control configuration / determination strategy may, for example, be related to dynamic / adaptive / progressive control based on model training in a network.

[0055] The advantageous aspects of system 100 described above in this disclosure are similarly applicable to all aspects of device 102 described below in this disclosure. Similarly, all advantageous aspects of device 102 described below in this disclosure are similarly applicable to all aspects of system 100 described above in this disclosure.

[0056] The following will refer to Figure 2 The aforementioned device 102 will be discussed in further detail.

[0057] refer to Figure 2 According to embodiments of this disclosure, device 102 is shown in more detail in the context of example implementation 200.

[0058] In example implementation 200, device 102 may correspond to electronic module 200a. According to embodiments of this disclosure, in one example, electronic module 200a may correspond to, for example, a mobile device that can be carried by a user into a vehicle. In another example, according to embodiments of this disclosure, electronic module 200a may correspond to an electronic device that can be installed / assembled in a vehicle. In this regard, electronic module 200a can be considered as being carried by the vehicle (e.g., carried by a user into the vehicle or installed / assembled in the vehicle).

[0059] According to embodiments of this disclosure, it is conceivable that electronic module 200a may be able to perform one or more processing tasks associated with adaptive / dynamic / progressive control-related processing.

[0060] Electronic module 200a may include, for example, a housing 200b. Furthermore, electronic module 200a may, for example, carry any one or any combination of the first module 202, the second module 204, and the third module 206.

[0061] In one embodiment, electronic module 200a may carry first module 202, second module 204, and / or third module 206. In a specific example, according to an embodiment of this disclosure, electronic module 200a may carry first module 202, second module 204, and third module 206.

[0062] In this regard, it can be understood that, in one embodiment, the shape and size of the housing 200b may be designed to carry any one or any combination of the first module 202, the second module 204 and the third module 206.

[0063] The first module 202 may be coupled to one or both of the second module 204 and the third module 206. The second module 204 may be coupled to one or both of the first module 202 and the third module 206. The third module 206 may be coupled to one or both of the first module 202 and the second module 204. In one example, according to an embodiment of the present disclosure, the first module 202 may be coupled to the second module 204, and the second module 204 may be coupled to the third module 206. The coupling between the first module 202, the second module 204, and / or the third module 206 may be performed, for example, by one or both of wired coupling and wireless coupling. According to an embodiment of the present disclosure, each of the first module 202, the second module 204, and the third module 206 may correspond to one or both of a hardware-based module and a software-based module.

[0064] In one example, the first module 202 may correspond to a hardware-based receiver that can be configured to receive one or more input signals. According to embodiments of this disclosure, the input signals may be transmitted, for example, from device 104 (e.g., gNB).

[0065] According to embodiments of this disclosure, the second module 204 may correspond, for example, to a hardware-based processor that can be configured to perform one or more processing tasks (e.g., to generate one or more output signals), as will be referred to later. Figure 3 Further detailed discussion is needed.

[0066] The third module 206 may correspond to a hardware-based transmitter that can be configured to transmit one or more output signals from the electronic module 200a. According to embodiments of this disclosure, the output signals may, for example, include / correspond to one or more instruction / command / control signals associated with the aforementioned dynamic / adaptive / progressive control configuration / determination strategy in order to promote efficiency (e.g., power / energy efficiency and / or communication efficiency).

[0067] This disclosure envisions the possibility that the first module 202 and the second module 204 can be integrated software-hardware modules (e.g., electronic components carrying software programs / algorithms associated with receiving and processing functions / electronic modules programmed to perform receiving and processing functions). This disclosure further envisions the possibility that the first module 202 and the third module 206 can be integrated software-hardware modules (e.g., electronic components carrying software programs / algorithms associated with receiving and transmitting functions / electronic modules programmed to perform receiving and transmitting functions). This disclosure also further envisions the possibility that the first module 202 and the third module 206 can be integrated hardware modules capable of performing receiving and transmitting functions (e.g., hardware-based transceivers).

[0068] The aforementioned advantageous aspects of the device 102 of this disclosure are also similarly applicable to all aspects of the processing / communication methods described below in this disclosure. Similarly, all the aforementioned advantageous aspects of the processing / communication methods of this disclosure are also similarly applicable to all aspects of the device 102 described above in this disclosure. It should be understood that these statements also apply similarly to the system 100 discussed earlier in this disclosure.

[0069] refer to Figure 3 The embodiments of this disclosure illustrate a method (also referred to as a processing method) associated with system 100.

[0070] According to embodiments of this disclosure, method 300 may be, for example, adapted to / able to promote energy efficiency.

[0071] According to embodiments of the present disclosure, the processing method 300 may include any one or any combination of the input step 302, the processing step 304, and the output step 306.

[0072] In one embodiment, processing method 300 may include an input step 302. In another embodiment, processing method 300 may include an input step 302 and a processing step 304. In another embodiment, processing method 300 may include an input step 302, a processing step 304, and an output step 306. In yet another embodiment, processing method 300 may include a processing step 304 and one or both of input step 302 and output step 306. In still another embodiment, processing method 300 may include an input step 302, a processing step 304, and an output step 306. In yet another additional embodiment, processing method 300 may include a processing step 304. In yet another additional embodiment, processing method 300 may include any or any combination of input step 302, processing step 304, and output step 306 (i.e., input step 302, processing step 304, and / or output step 306).

[0073] Regarding input step 302, one or more input signals may be received. For example, according to embodiments of this disclosure, the input signals may be transmitted from device 104 and received by device 102.

[0074] Input step 302 may include receiving at least one input signal associated with data indicating model training. In an embodiment, the input signal may be generated by and transmitted from device 104 to device 102. Alternatively, the input signal may be generated and received by device 102 to proceed to processing step 304. For example, the input signal may be generated by a transmitting UE (or user equipment) and received by a receiving UE (or user equipment).

[0075] According to embodiments of this disclosure, with respect to processing step 304, at least a processing task associated with the received input signal can be performed in some way to generate one or more output signals.

[0076] Processing step 304 may include: configuring data to instruct model training; transmitting the data to a user device; initiating model training based on the data; and generating multiple parameters related to model training. The data may include a model identifier and a dataset identifier, and the multiple parameters may include the model identifier and an associated dataset identifier.

[0077] Processing step 304 may further include: transmitting multiple parameters via physical layer (L1) signaling and / or data link layer (L2) signaling; transmitting multiple parameters via user equipment (UE) specific messages and / or system information blocks (SIBs) based on model configuration; and verifying multiple parameters used for model training and inference. Generating multiple parameters may include periodic generation and / or generation based on pre-configured events, and model training may include artificial intelligence machine learning (AI / ML) model training.

[0078] In an embodiment, a method for dataset identification for model training and inference in a wireless communication system may exist, whereby a gNB (or base station or network) can provide a dataset and a model identifier associated with that dataset. The wireless communication system can configure the model identifier during model identification, and the dataset identifier can be configured to be use case specific.

[0079] In one embodiment, the wireless communication system can be configured with a model identifier that has a dataset identifier, thereby allowing the model identifier and dataset identifier to be configured together.

[0080] In one embodiment, after the UE (or user equipment) has collected the dataset and is performing training, the gNB (or base station) may send a use case indication with a dataset identifier.

[0081] In one embodiment, a method for dataset identification for model training and inference in a wireless communication system may exist, whereby a UE (or user device) receives a message via the wireless communication system for configuring a model identifier having a dataset identifier. Furthermore, the UE (or user device) may configure the model identifier and the dataset identifier, and collect the dataset and perform training during model identification. Then, the UE (or user device) may instruct the model identifier having the dataset identifier for data collection.

[0082] In an embodiment, after receiving a message for configuring a model identifier via a wireless communication system, and after receiving a use case identifier with a dataset set from a gNB (or base station) after model identification, the UE (or user device) may indicate a model identifier with a dataset identifier for data collection.

[0083] In an embodiment, the UE (or user equipment) may include a dataset identifier with a model identifier, wherein the dataset identifier and the model identifier may be configured by the network, and the UE (or user equipment) that generates the dataset may include the dataset identifier. The network (or gNB or base station) receiving the dataset may use the dataset identifier for model validation, and may also use the received dataset for training and inference. The UE (or user equipment) may generate datasets periodically or based on pre-configured events, and the UE (or user equipment) may also signal the dataset identifier and model identifier via L1 or L2 signaling. This may be signaled via UE-specific messages and SIBs based on how AIML model identification is configured. This may cause the entity performing model training (e.g., the UE or gNB) to associate the dataset with the model ID to avoid dataset mismatch.

[0084] Regarding output step 306, according to embodiments of the present disclosure, as an option, for example, an output signal may be transmitted. For example, the output signal may optionally be transmitted from device 102. In a more specific example, according to embodiments of the present disclosure, the output signal may optionally be transmitted from device 102 to one or both of at least one device 104 and another device 102.

[0085] This disclosure further envisions a computer program (not shown) that may include instructions that, when executed by a computer (not shown), cause the computer to perform the input step 302, processing step 304, and / or output step 306 as discussed in reference method 300. For example, according to an embodiment of the invention, the computer program may include instructions that, when executed by a computer, cause the computer to perform the input step 302 and / or processing step 304.

[0086] This disclosure further envisions a computer-readable storage medium (not shown) storing data representing software executable by a computer (not shown), the software including instructions that, when executed by a computer, perform input step 302, processing step 304, and / or output step 306 as discussed in reference method 300. For example, according to an embodiment of the invention, the computer-readable storage medium may store data representing software executable by a computer, the software including instructions that, when executed by a computer, cause the computer to perform input step 302 and / or processing step 304.

[0087] In light of the foregoing, it is understood that this disclosure generally envisions an energy-saving device 102 suitable for use in a network, which may include a first module 202, a second module 204, and / or a third module 206.

[0088] The first module 202 can be configured to receive one or more input signals. These input signals may, for example, be associated with data indicating model training.

[0089] The second module 204 may be configured to process the input signal and / or facilitate the processing of the input signal according to the method 300 discussed earlier, to generate one or more output signals.

[0090] The third module 206 can be configured to transmit one or more output signals. These output signals may, for example, correspond to one or more control signals used for model training in the network.

[0091] In one embodiment, device 102 may correspond to a user equipment (UE) that can communicate with device 104 corresponding to a base station. The base station may, for example, correspond to a next-generation node B (gNB) that can be configured to transmit one or more signals (e.g., input signals) to the UE.

[0092] Furthermore, in view of the foregoing, it should be understood that this disclosure generally contemplates a system 100 that may include one or more devices 102 and one or more apparatuses 104. Devices 102 and apparatuses 104 may be coupled, for example, via wired coupling and / or wireless coupling.

[0093] It should be understood that the embodiments described above can be combined in any way where appropriate (e.g., one or more embodiments as discussed in the “Detailed Description” section can be combined with one or more embodiments as described in the “Summary of the Invention” section).

[0094] Those skilled in the art will further understand that variations and combinations of the embodiments described above, rather than alternatives or substitutes, can be combined to form even further embodiments.

[0095] In one example, the possibility of transmitting an output signal from device 102 is discussed. It should be understood that transmitting an output signal from device 102 is not necessarily required. Specifically, according to embodiments of the invention, the possibility that an output signal does not necessarily need to be transmitted outside device 102 is envisioned. More specifically, according to embodiments of the invention, the output signal may correspond, for example, to internal commands / instructions (e.g., transmitted only within device 102) for adaptively controlling the operational configuration of device 102.

[0096] Figures 4A to 4D A schematic diagram illustrating an example scenario associated with method 300 according to an embodiment of the present disclosure is shown.

[0097] Figure 4AThe example context shown may have a model identifier and a dataset identifier configured together between the UE (or user equipment) and the gNB (or base station). Both the gNB (or base station) and the UE (or user equipment) may have a model on their side, and this model is provided on the gNB (or base station) side. The gNB (or base station) may send a message to the UE (or user equipment), and the network is configured with a model identifier containing the dataset identifier. After the UE (or user equipment) has received such a message, the UE (or user equipment) collects the dataset and performs training. Thereafter, the UE (or user equipment) may instruct the model identifier containing the dataset identifier for use in data collection.

[0098] Figure 4B The example context shown may have a use case-specific dataset identifier configured between the UE (or user equipment) and the gNB (or base station). The gNB (or base station) and the UE (or user equipment) may each have a model on their respective sides. The gNB (or base station) may send a message to the UE (or user equipment), and the network is configured with a model identifier containing the dataset identifier. After the UE (or user equipment) has received this message, it collects the dataset and performs training, and receives a use case indication with the dataset identifier after model identification. Thereafter, the UE (or user equipment) indicates a model ID with a dataset ID for data collection.

[0099] In such Figure 4C In the example context shown, the base station (or gNB or network) can configure model identifiers and dataset identifiers for the UE (or user equipment).

[0100] In such Figure 4D In the example context shown, the UE (or user device) can be configured to provide a dataset and a model identifier associated with that dataset.

[0101] Various embodiments of this disclosure for addressing at least one of the aforementioned disadvantages have been described in the foregoing manner. Such embodiments are intended to be covered by the appended claims and are not limited to the specific form or arrangement of the components so described, and in view of this disclosure, it will be apparent to those skilled in the art that many changes and / or modifications can be made, which are also intended to be covered by the appended claims.

Claims

1. A method (300) for model training in a network, the method comprising: Configure the data for training the instruction model; The data is transmitted to the user device; Based on the data, initiate model training; as well as Generate multiple parameters related to the training of the model.

2. The method (300) according to claim 1, wherein, The data includes model identifiers and dataset identifiers.

3. The method (300) according to claim 1, wherein, The parameters include the model identifier and the associated dataset identifier.

4. The method (300) according to claim 1, further comprising: The parameters are transmitted via physical layer (L1) signaling and / or data link layer (L2) signaling.

5. The method (300) according to claim 1, further comprising: Based on the model configuration, the multiple parameters are transmitted via User Equipment (UE) Specific Messages and / or System Information Blocks (SIBs).

6. The method (300) according to claim 1, further comprising: Validate the multiple parameters used for model training and inference.

7. The method (300) according to claim 1, wherein, Generating the multiple parameters includes: periodic generation and / or generation based on pre-configured events.

8. The method (300) according to claim 1, wherein, The model training includes artificial intelligence machine learning (AI / ML) model training.

9. A computer program comprising instructions that, when executed by a computer, cause the computer to perform the method (300) as described in any of the preceding claims.

10. A computer-readable storage medium storing data representing software executable by a computer, the software including instructions that, when executed by the computer, are used to perform the method (300) as claimed in claims 1 to 9.

11. An apparatus (102) for training a model in a network, the apparatus comprising: A first module (202) is configured to receive at least one input signal associated with data indicating model training. The second module (204) is configured to process and facilitate at least one of the methods (300) as claimed in claims 1 to 8 to generate at least one output signal; as well as The third module (206) is configured to transmit at least one output signal. The output signal corresponds to the control signal used for model training in the network.

12. The device (102) according to claim 11, in, The device (102) corresponds to a user equipment (UE) capable of communicating with a device (104) corresponding to a base station, and The base station corresponds to a next-generation node B (gNB) configured to transmit the at least one input signal to the UE.

13. A system (100), the system comprising: At least one device (102) according to any one of claims 11 and 12; as well as At least one device (104) according to claim 12, The device (102) and the apparatus (104) are coupled via at least one of wired coupling and wireless coupling.

Images