Apparatus and method for communication

Abstract

Embodiments of the present disclosure provide a solution for achieving consistency of model- or functionality-related information. The device determines information for achieving consistency of model- or functionality-related information between a first machine learning (ML) stage for implementing a model or functionality and a second ML stage of the model or the functionality; and performs at least one of the following based on the information: a calibration process with another device, the first ML stage, or the second ML stage.

Description

Technical Field

[0001] The exemplary embodiments disclosed herein relate generally to the field of communication technology, and more specifically to devices and methods for achieving consistency of model-related or functionality-related information. Background Technology

[0002] As communication networks and services grow in scale, complexity, and user numbers, operations within these networks can become increasingly complex. To improve communication performance, the use of machine learning (ML) / artificial intelligence (AI) techniques in wireless communication networks has been proposed. For example, terminal devices and network devices can use different ML models to assist in communication-related functionalities such as beam management (BM) and mobility management. Summary of the Invention

[0003] Typically, embodiments of this disclosure provide a solution for achieving consistency of model-dependent or functionality-related information.

[0004] In a first aspect, an apparatus is provided, the apparatus comprising: a processor configured to cause the apparatus to: determine information on the consistency of model-related or functional-related information between a first machine learning (ML) stage for implementing a model or functionality and a second ML stage of the model or functionality; and based on the information to perform at least one of: a calibration process with another device, the first ML stage, or the second ML stage.

[0005] In a second aspect, a communication method performed by a device is provided. The method includes: determining information on the consistency of model-related or functionality-related information between a first machine learning (ML) stage for implementing a model or functionality and a second ML stage of the model or functionality; and performing, based on the information, at least one of: a calibration process with another device, the first ML stage, or the second ML stage.

[0006] In a third aspect, a computer-readable medium is provided that stores instructions which, when executed on at least one processor, cause the at least one processor to perform the method according to the second aspect.

[0007] Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0008] The above and other objects, features and advantages of this disclosure will become more apparent from a more detailed description of some exemplary embodiments thereof in the accompanying drawings, wherein: Figure 1 An example communication environment in which an example implementation of the present disclosure can be carried out is illustrated; Figure 2A Examples of consistency across different ML stages are illustrated; Figure 2B Example sets of beams across different ML stages are illustrated; Figure 2C Example sets of beams across different ML stages are illustrated; Figure 2D An example mapping of beam sets is shown; Figure 3 Signaling flows of communications according to some embodiments of this disclosure are illustrated; Figure 4 A 2-D planar antenna structure is illustrated, where each column is a cross-polarization array; Figure 5 An example mapping between resources and weights is shown; Figure 6A Signaling flows of communications according to some embodiments of this disclosure are illustrated; Figure 6B Signaling flows of communications according to some embodiments of this disclosure are illustrated; Figure 7A Signaling flows of communications according to some embodiments of this disclosure are illustrated; Figure 7B Signaling flows of communications according to some embodiments of this disclosure are illustrated; Figure 8 Signaling flows of communications according to some embodiments of this disclosure are illustrated; Figure 9 Signaling flows of communications according to some embodiments of this disclosure are illustrated; Figure 10 Flowcharts illustrating methods implemented at a device according to some example embodiments of this disclosure are shown; and Figure 11 A simplified block diagram of an apparatus suitable for implementing an example embodiment of the present disclosure is shown.

[0009] Throughout the accompanying drawings, the same or similar reference numerals denote the same or similar elements. Detailed Implementation

[0010] 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 help those skilled in the art to understand and implement 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.

[0011] In the following description and claims, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains.

[0012] As used herein, the term "terminal device" refers to any device with wireless or wired communication capabilities. Examples of terminal devices include, but are not limited to: user equipment (UE); personal computers; desktop computers; mobile phones; cellular phones; smartphones; personal digital assistants (PDAs); portable computers; tablets; wearable devices; Internet of Things (IoT) devices; Ultra-reliable and Low-Latency Communication (URLLC) devices; Internet of Everything (IoE) devices; machine-type communication (MTC) devices; devices on vehicles for V2X communication, where X refers to pedestrians, vehicles, or infrastructure / networks; devices for Integrated Access and Backhaul (IAB); spacecraft or aerospace vehicles in non-terrestrial networks (NTNs), including satellites and high-altitude platforms (HAPs) covering Unmanned Aircraft Systems (UAS); and different types of reality (such as Augmented Reality (AR), Mixed Reality (MR)). Extended Reality (XR) devices, including those for Virtual Reality (VR) and Virtual Reality (VR); unmanned aerial vehicles (UAVs), often referred to as drones (aircraft without human pilots); devices on high-speed trains (HSTs); or image capture devices such as digital cameras and sensors; gaming devices; music storage and playback equipment; or internet devices enabling wireless or wired internet access and browsing. "Terminal devices" may also have "multicast / broadcast" capabilities to support public safety and mission-critical applications, V2X applications, transparent IPv4 / IPv6 multicast delivery, IPTV, smart TV, radio services, wireless software delivery, group communication, and IoT applications. "Terminal devices" may also incorporate one or more Subscriber Identity Modules (SIMs), a situation known as multi-SIM. The term "terminal device" is used interchangeably with UE, mobile station, subscriber station, mobile terminal, user terminal, or wireless device.

[0013] The term "network device" refers to a device that provides or hosts a cell or coverage area for terminal devices to communicate. Examples of network devices include, but are not limited to, NodeBs (or NBs), evolved NodeBs (eNodeBs or eNBs), next-generation NodeBs (gNBs), transmission reception points (TRPs), remote radio units (RRUs), radioheads (RHs), remote radio heads (RRHs), IAB nodes, low-power nodes (such as femtonodes and piconodes), reconfigurable intelligent surfaces (RISs), etc.

[0014] Terminal devices or network devices may have artificial intelligence (AI) or machine learning capabilities. Terminal devices or network devices typically include models that have been trained on specific functions based on a large amount of collected data and can be used to predict some information.

[0015] Terminal or network devices can operate within several frequency ranges, such as FR1 (e.g., 450MHz to 6000MHz), FR2 (e.g., 24.25GHz to 52.6GHz), bands greater than 100GHz, and terahertz (THz). Terminal or network devices can also operate on licensed / unlicensed / shared spectrum. In Multi-Radio Dual Connectivity (MR-DC) applications, terminal devices may connect to more than one network device. Terminal or network devices can operate in full-duplex, flexible-duplex, and cross-division duplex modes.

[0016] The embodiments of this disclosure can be executed in test equipment (e.g., signal generator, signal analyzer, spectrum analyzer, network analyzer, test terminal equipment, test network equipment, channel simulator). In some embodiments, the terminal equipment can be connected to a first network equipment and a second network equipment. One of the first network equipment and the second network equipment can be a master node, and the other can be a slave node. The first network equipment and the second network equipment can use different Radio Access Technologies (RATs). In some embodiments, the first network equipment can be a first RAT device, and the second network equipment can be a second RAT device. In some embodiments, the first RAT device is an eNB, and the second RAT device is a gNB. Information related to different RATs can be sent to the terminal equipment from at least one of the first network equipment or the second network equipment. In some embodiments, first information can be sent from the first network equipment to the terminal equipment, and second information can be sent from the second network equipment directly or via the first network equipment to the terminal equipment. In some embodiments, information configured by the second network equipment and related to the configuration of the terminal equipment can be sent from the second network equipment via the first network equipment. Information configured by the second network device and related to the reconfiguration of the terminal device can be sent directly from the second network device or via the first network device to the terminal device.

[0017] As used herein, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. The term “comprising” and its variations should be understood as open terms meaning “including, but not limited to.” The term “based on” should be understood as “at least partially based on.” The terms “one implementation” and “implementation” should be understood as “at least one implementation.” The term “another implementation” should be understood as “at least one other implementation.” The terms “first,” “second,” etc., may refer to different or the same objects. Other explicit and implicit definitions are given below.

[0018] In some examples, values, programs, or devices are described as “best,” “lowest,” “highest,” “smallest,” “maximum,” etc. It should be understood that such descriptions are intended to indicate that a choice can be made among many alternative functionalities used, and that such a choice is not necessarily better, smaller, higher, or otherwise preferred than other choices.

[0019] As used herein, the terms “resource,” “transmission resource,” “uplink resource,” or “downlink resource” can refer to any resource used to perform communication, such as a resource in the time domain, a resource in the frequency domain, a resource in the spatial domain, a resource in the code domain, or any other resource used to implement communication. In the following, unless explicitly stated otherwise, resources in both the frequency and time domains will be used as examples of transmission resources to describe some exemplary embodiments of this disclosure. It should be noted that the exemplary embodiments of this disclosure are equally applicable to other resources in other domains.

[0020] As used herein, the terms “UE expects,” “UE does not expect,” “terminal device expects,” and “terminal device does not expect” may imply a limitation on the configuration of network devices (also known as NW (network) configuration). The terms “not expecting UE” and “not expecting terminal device” may imply a specific terminal implementation (also known as a specific UE implementation). In some implementations, the terms “UE does not expect” and “does not expect UE” may be used interchangeably.

[0021] As discussed in this article, terminal devices and network devices can use different ML models to assist in communication-related functionalities, such as beam management (BM) and mobility management.

[0022] To date, the following BM scenario 1 and BM scenario 2 have been agreed upon for AI / ML-based beam management: BM Scenario 1: Based on the measurement results of beam set B, perform spatial downlink (or uplink) beam prediction for beam set A.

[0023] BM Scenario 2: Based on the historical measurement results of beam set B, perform time downlink (or uplink) beam prediction for beam set A.

[0024] Furthermore, for BM scenario 1 and BM scenario 2, the beams in the aforementioned sets A and B can be in the same frequency range (FR).

[0025] In BM Scenario 1 and BM Scenario 2, the following alternative schemes for beam prediction are supported: downlink transmit (TX) beam prediction, downlink receive (RX) beam prediction, and beam pair prediction (a beam pair consists of a downlink TX beam and a corresponding downlink RX beam).

[0026] For the sub-use cases of BM Scenario 1 and BM Scenario 2, the following alternative schemes for AI / ML output can be supported: TX and / or RX beam IDs of N predicted downlink TX and / or RX beams and / or predicted layer 1 (L1) reference signal receiving power (RSRP), for example, the N predicted beams can be the first N predicted beams; TX and / or RX beam IDs and other information of N predicted downlink TX and / or TX beams (e.g., probability that the beam is the best beam, associated confidence, beam application time / dwell time, predicted beam failure), where the N predicted beams can be the first N predicted beams; TX and / or RX beam angles of N predicted DL TX and / or RX beams and / or predicted L1-RSRP, where the N predicted beams can be the first N predicted beams.

[0027] For BM Scenarios 1 and BM Scenarios 2 that utilize UE-side AI / ML models, it is expected that the functionality and / or model necessity and potential BM-specific / additional conditions will be studied from at least the following aspects: information about model inference; set A / set B configuration; performance monitoring; data collection; and auxiliary information.

[0028] For BM Scenario 1 and BM Scenario 2 that utilize the UE-side AI / ML model, the following indications may be considered: indications from the network to the UE for associated set A, such as the association / mapping of beams within set A and beams within set B (if applicable); beam indications from the network for UE reception.

[0029] Regarding the data collection performed on the UE side for AI / ML model training, configurations will be provided (e.g., configurations related to set A and / or set B, information about the association / mapping of set A and set B), and the network may provide auxiliary information to the UE.

[0030] In the following text, "reference pattern information" / "reference pattern" will be used as an example of information used to achieve consistency with some specific example embodiments described in this disclosure. It should be noted that the example embodiments described with respect to "reference pattern information" / "reference pattern" also apply to information used to achieve consistency.

[0031] For better description, the following is a list of some of the terms used in this article: AI / ML model: refers to a data-driven algorithm that uses AI / ML technology to generate a set of outputs based on a set of inputs; AI / ML Model Transfer: This is a general term referring to the transfer of AI / ML models from one entity to another in any way. Note: Entity can refer to network nodes / functions (e.g., gNB, LMF, etc.), UE, dedicated servers, etc. Functionality: refers to the AI / ML enabled features / feature groups (FG) that are enabled by configuration, where the configuration is supported based on conditions indicated by the UE capabilities; Functional LCM: It operates based on at least one configuration of the AI / ML feature / FG or a specific configuration of the AI / ML feature / FG. Model ID-based LCM: Operation is based on an identified model, where the model can be associated with specific configurations / conditions and UE capabilities that enable AI / ML features / FG, as well as additional conditions (e.g., scene, site, and dataset) determined / identified between the UE side and the NW side. AI / ML model inference: refers to the process of using a trained AI / ML model to produce a set of outputs based on a set of inputs; AI / ML model testing refers to the training sub-process used to evaluate the performance of the final AI / ML model using a different dataset than that used for model training and validation. Unlike AI / ML model validation, testing does not assume subsequent model tuning. AI / ML model training: refers to the process of training an AI / ML model in a data-driven manner [by learning the input / output relationship] and obtaining a trained AI / ML model for inference. AI / ML model transmission: This refers to transmitting AI / ML models over the air interface in a manner opaque to 3GPP signaling. The transmitted content consists of parameters of a model structure known to the receiving end, or a new model containing parameters. The transmitted content can include a complete model or a partial model. AI / ML model validation: refers to the training sub-process used to evaluate the quality of AI / ML models using a different dataset than the dataset used for model training, in order to help select model parameters that demonstrate generalization ability outside the dataset used for model training. Data collection refers to the process by which network nodes, management entities, or UEs collect data for the purposes of AI / ML model training, data analysis, and inference. Federated learning / federated training is a machine learning technique that trains AI / ML models across multiple decentralized edge nodes (e.g., UE, gNB), with each node performing local model training using local data samples. This technique requires multiple interactions between models but does not exchange local data samples. Functionality identification: refers to the process / method of identifying AI / ML functionality for mutual understanding between the network and the UE. Note: Information regarding AI / ML functionality can be shared during the functionality identification process. The location where AI / ML functionality resides depends on the specific use case and sub-use case; Model activation: refers to enabling AI / ML models for specific AI / ML-enabled features; Model deactivation: refers to disabling AI / ML models for specific AI / ML-enabled features; Model download: refers to transferring the model from the network to the user interface (UE); Model identification: refers to the process / method of identifying AI / ML models for mutual understanding between the network and the UE. Note: The process / method of model identification may or may not be applicable; Information about AI / ML models can be shared during model identification; Model monitoring: refers to the process of monitoring the inference performance of AI / ML models; Model parameter update: refers to the process of updating the model parameters; Model selection: This refers to the process of choosing one AI / ML model from multiple models to activate the same AI / ML-enabled feature. Note: Model selection can be performed simultaneously with model activation, or they can be performed at different times. Model switching refers to deactivating the currently active AI / ML model and activating different AI / ML models for specific AI / ML-enabled features; Model update: refers to the process of updating the model parameters and / or model structure; Model upload: refers to the transmission of the model from the UE to the network; Network-side (AI / ML) models: These are AI / ML models where inference is performed entirely at the network level. Offline field data: refers to data collected from the field and used for offline training of AI / ML models; Offline training: refers to the AI / ML training process in which a model is trained based on a collected dataset, and the trained model is later used or delivered for inference. Note: This definition is for guidance only. There may be some cases that, while not perfectly conforming to this definition, can still be classified as offline training according to generally accepted conventions. Online field data: refers to data collected from the field and used for online training of AI / ML models; Online training: refers to the AI / ML training process where the model used for inference is trained (usually continuously) in (near) real-time as new training samples arrive. Note: The concepts of (near) real-time and non-real-time depend on the context and are relative to the inference timescale. Note: This definition is for guidance only. There may be cases that, while not perfectly fitting this definition, can still be classified as online training according to generally accepted conventions. Note: Fine-tuning / retraining can be accomplished via online or offline training. (This note may be removed after defining the term "fine-tuning"). Reinforcement learning (RL) refers to the process of training an AI / ML model in an environment that interacts with the model, based on inputs (also called states) and feedback signals (also called rewards) caused by the model's outputs (also called actions). Semi-supervised learning is the process of training a model using a mixture of labeled and unlabeled data. Supervised learning refers to the process of training a model based on the input and its corresponding labels. Two-sided (AI / ML) model: refers to a paired AI / ML model on which joint inference is performed, where joint inference includes AI / ML inference, which is jointly performed across the UE and the network. That is, the first part of the inference is first performed by the UE, and then the remaining part is performed by the gNB, and vice versa. UE-side (AI / ML) model: refers to an AI / ML model where inference is performed entirely on the UE; Unsupervised learning refers to the process of training a model without using labeled data. From a 3GPP perspective, proprietary format models are vendor / device-specific proprietary ML models. These models are not mutually recognizable between different vendors and hide model design information from other vendors when shared. Note: An example is a device-specific binary executable format; Open format models: From a 3GPP perspective, these are ML models with a specified format that can be mutually recognized and interoperable across different vendors. Such models are mutually recognizable across different vendors and do not hide model design information from other vendors when shared. Measurement results may include, but are not limited to, (L1 / L3)-Reference Signal Received Power (RSRP), (L1 / L3)-SINR, (L1 / L3)-Received Signal Strength Indicator (RSSI), or (L1 / L3)-Reference Signal Received Quality (RSRQ). "Conditions": Configuration supported by UE capability report indications related to model training, model inference, performance monitoring, verification process, and fallback associated with AI / ML model / functionality or a set of models / functionality; "Additional conditions": These include, for example, application conditions, scenarios, datasets, cell IDs, timestamps, and SNR. For AI / ML-enabled features / feature groups, additional conditions refer to any aspect assumed for model training but not part of the UE's capabilities for AI / ML-enabled features / feature groups. Additional conditions can be divided into two categories: NW-side additional conditions and UE-side additional conditions. (The last sentence is a repetition of the first and can be omitted.) "UE internal conditions": such as memory, battery, computing resources, overheating and other hardware limitations; NW / Network Equipment: This can be access network equipment or core network equipment, such as "Operation Administration and Maintenance (OAM)," "Server," and "Access and Mobility Management Function (AMF) / Location Management Function (LMF)."

[0032] In this disclosure, The terms "information used to achieve conformance", "calibration information", and "conformance information" are used interchangeably. The terms “model,” “functionality,” and “model / functionality” are used interchangeably. The terms “ID,” “index,” “indicator,” and “identifier” are used interchangeably. The terms “model,” “model group,” “a set of models,” and “model collection” are used interchangeably. The terms “functionality,” “functional group,” “a set of functionalities,” and “functional collection” are used interchangeably.

[0033] In this disclosure, a beam may correspond to a channel state information-reference signal (CSI-RS), a synchronization signal and physical broadcast channel (PBCH) block (SSB), a CSI-RS resource, or an SSB resource. Therefore, the beam identity (ID) may be a CSI-RS resource indicator (CRI), an SSB resource indicator (SSBRI), or an RS ID. It should also be understood that, in fact, a beam refers to a resource that enables space-oriented communication, and therefore may be identified by other suitable parameters in other embodiments. This disclosure is not limited in this respect.

[0034] As used herein, the term "ML phase" may be replaced by "ML period," "ML process," "LCM phase," "LCM period," or "LCM process," including but not limited to model delivery process / period / stage, model inference process / period / stage, model testing process / period / stage, model training process / period / stage, model monitoring process / period / stage, model transfer process / period / stage, model validation process / period / stage, data collection process / period / stage, model learning process / period / stage, etc.

[0035] As used in this article, a physical model ID can refer to a real AI / ML model or a real implementation; for the same purpose, a logical model ID can be associated with one or a group of physical models. Furthermore, global model IDs and local model IDs can also be used to identify models.

[0036] It should be noted that when using the term "a set of...", it can refer to one or more elements / items, and the term can be replaced with the terms "at least one", "a group", or "a list of...". For example, "a set of X" means "at least one X" or "one or more X".

[0037] As used herein, a model may be equivalent to at least one of the following: AI / ML model, ML model, AI model, data-driven, data processing model, algorithm, functionality, program, process, entity, function, feature, feature group, model identifier (ID), ID, functionality ID, configuration ID, scene ID, site ID, or dataset ID. Therefore, the above terms are used interchangeably.

[0038] In some implementations, the model may include a set of weights that can be learned during training, for example for a specific architecture or configuration, where the set of weights may also be referred to as a parameter set.

[0039] In some implementations, the model can be used to predict target cells, or to predict the measurement results of a set of beams for a future set of candidate cells based on historical measurements of a set of beams from at least one set of candidate cells (e.g., L1-RSRP, L1-SINR).

[0040] In some implementations, the input to the ML model (i.e., the AI ​​input) may refer to the input to the model and indicate the data input into the model, which may be equivalent to data.

[0041] In some implementations, the output of the ML model (i.e., the AI ​​output) can refer to the output of the model and indicate the result produced by the model, which is equivalent to the label / data.

[0042] The principles and specific implementations of this disclosure will now be described in detail with reference to the accompanying drawings.

[0043] Example Environment Figure 1 A schematic diagram of an example communication environment 100 in which an example embodiment of the present disclosure may be implemented is illustrated. In the communication environment 100, multiple communication devices (including device 110 and another device 120) can communicate with each other.

[0044] Furthermore, the communication environment 100 supports multiple input multiple output (MIMO), enabling another device 120 and device 110 to communicate with each other via different beams to achieve directional communication.

[0045] exist Figure 1 In some implementations, device 110 may be a terminal device or a network device (including access network device or core network device), and another device 120 may be a terminal device or a network device (including access network device or core network device).

[0046] Furthermore, in some embodiments, one of device 110 and the other device 120 may include a terminal device, which may be referred to as the first device, and the other of device 110 and the other device 120 may include a network device, which may be referred to as the second device. In this particular example embodiment, the link from the first device to the second device is referred to as an uplink, and the link from the second device to the first device is referred to as a downlink.

[0047] In the downlink, the second device is a transmitting (TX) device (or transmitter), and the first device is a receiving (RX) device (or receiver). Correspondingly, in the uplink, the second device is an RX device (or receiver), and the first device is a TX device (or transmitter).

[0048] exist Figure 1 In this configuration, device 110 and another device 120 can communicate with each other via one or more beams. For example... Figure 1 As illustrated, device 110 can communicate with another device 120 via beams 130-1 to 130-3. For discussion purposes, beams 130-1 to 130-3 are collectively or individually referred to as beam 130. Figure 1 As illustrated, another device 120 may communicate with device 110 via one or more of beams 140-1, 140-2, and 140-3. For the purposes of discussion, beams 140-1 to 140-3 are collectively or individually referred to as beam 140.

[0049] In some implementations, one or more models may be deployed at another device 120 and / or device 110. For example... Figure 1As illustrated, model 115 is deployed at device 110, and / or model 125 is deployed at another device 120. Furthermore, in Figure 1 In the example, model 115 and / or model 125 can assist in tasks such as BM, i.e., obtaining input data and deriving relevant outputs.

[0050] 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 example embodiments of this disclosure.

[0051] In some implementations, device 110 and another device 120 may communicate with each other via a channel, such as a wireless communication channel on an air interface (e.g., a Uu interface). The wireless communication channel may include a physical uplink control channel (PUCCH), a physical uplink shared channel (PUSCH), a physical random-access channel (PRACH), a physical downlink control channel (PDCCH), a physical downlink shared channel (PDSCH), and a physical broadcast channel (PBCH). Of course, any other suitable channel is also possible.

[0052] The communications in communication environment 100 may conform to any suitable standard, including but not limited to Global System for Mobile Communication (GSM), Long Term Evolution (LTE), LTE-Evolution, LTE-Advanced (LTE-A), New Radio (NR), Wideband Code Division Multiple Access (WCDMA), Code Division Multiple Access (CDMA), GSM EDGE Radio Access Network (GERAN), Machine Type Communication (MTC), etc. The embodiments of this disclosure may be implemented according to any generation of communication protocols currently known or to be developed in the future. Examples of communication protocols include, but are not limited to, first-generation (1G) communication protocols, second-generation (2G) communication protocols, 2.5G communication protocols, 2.75G communication protocols, third-generation (3G) communication protocols, fourth-generation (4G) communication protocols, 4.5G communication protocols, fifth-generation (5G) communication protocols, 5.5G, 5G-Advanced Networks, or sixth-generation (6G) networks.

[0053] Example process Generally speaking, for BM Scenario 1 and BM Scenario 2 utilizing UE-side AI / ML models, from a performance perspective, consistency / association between beams in set B and beams in set A across training and inference is beneficial. Therefore, for BM Scenario 1 and BM Scenario 2 utilizing UE-side AI / ML models, mechanisms are needed to ensure consistency (such as order / index consistency) between beams in set B and beams in set A across at least training and inference for all aspects related to the association / mapping of beams in set A and beams in set B.

[0054] Now for reference Figure 2A This exemplifies an example of consistency 200A across different ML stages. In some implementations, the consistency of set B beams and set A beams includes at least one or more of the following: Consistency of set size between set B and set A: Consistency of the number of beams and / or associated resources between set B and set A across training and inference; Sequence / index consistency: Consistency of resource order across training and inference for both set B beams and set A beams (e.g., resource index consistency). QCL Consistency: The consistency of the QCL relationship between beam A and beam B across training and inference. Beamform consistency: The difference in relative pointing direction and beamwidth between the physical beams of resources A and B across training and inference should be within a predefined tolerance range.

[0055] For better understanding, please refer to the following: Figure 2B and Figure 2C Examples of beam sets 200B and 200C across different ML phases are shown.

[0056] exist Figure 2B The example illustrates two phases of set A beams and set B beams based on different possible gNB codebooks (set B is not a subset of set A), where set A beams and set B beams are different during the different phases.

[0057] exist Figure 2C The example illustrates two cases of set A beams and set B beams based on different possible gNB codebooks (set B is a subset of set A), where set A beams and set B beams are different during different phases.

[0058] Furthermore, achieving consistency between beamforming sets A and B is difficult because beamforming algorithms can vary significantly across different network device / UE vendors. Even for network device / UE vendors, whether the same beamforming weights are consistently used depends on the specific network device / UE implementation. Additionally, the mapping between "physical" beams and "logical" resources also depends on the configuration / implementation.

[0059] For better understanding, please refer to the following: Figure 2D This illustrates an example mapping 200D of beam sets, where resource IDs 1 to 64 correspond to set A. Figure 2DIn embodiment (a), resource ID {18, 20, 22, 24, 42, 44, 46, 48} corresponds to set B. In contrast, in embodiment (b), resource ID {18, 20, 22, 24, 42, 44, 48} is associated with other beams, leading to inconsistency. In fact, resource ID {11, 27, 43, 59, 14, 30, 46, 62} in embodiment (b) is consistent with resource ID {18, 20, 22, 24, 42, 44, 46, 48} in embodiment (a).

[0060] according to Figure 2D The example shows that even for the same beam shape, the relationship can still be different depending solely on the resource ID.

[0061] For the reasons mentioned above, the beam IDs / resource IDs of beam sets B and A cannot guarantee the consistency of the required relationship between beam sets B and A.

[0062] To date, no detailed solution has been proposed to ensure consistency between ensemble B beams and ensemble A beams across training and inference.

[0063] The above problems can at least be avoided by using the examples of processing discussed below.

[0064] Based on several implementation schemes, a solution is proposed for achieving consistency of model-related or functionality-related information. Specifically, a reference (beam) pattern is introduced to achieve consistency of beams across different devices and across different ML stages (i.e., AI / ML lifecycle management (LCM) stages).

[0065] refer to Figure 3 This illustrates a signaling stream 300 for transmitting information about the number of predicted beams according to some embodiments of this disclosure. For discussion purposes, reference will be made to... Figure 1 (For example, by using device 110 and another device 120) Let’s discuss signaling flow 300.

[0066] It should be understood that the operations at device 110 and the other device 120 should be coordinated. In other words, the other device 120 and device 110 should have a common understanding regarding configuration, parameters, etc. This common understanding can be achieved through any suitable interaction between the other device 120 and device 110, or by both device 120 and device 110 applying the same rules / policies. Although some operations are described below from the perspective of device 110, it should be understood that the corresponding operations should be performed by the other device 120. Similarly, although some operations are described from the perspective of the other device 120, it should be understood that the corresponding operations should be performed by device 110. For the sake of brevity, some identical or similar content is omitted here.

[0067] In the following implementation, device 110 may be described as a terminal device or network device (including access network device or core network device), and another device 120 may be described as a terminal device or network device (including access network device or core network device).

[0068] The term "consistency" in this paper can refer to the consistency between the conditions / additional conditions / implementations / algorithms applied during model training and those applied during model inference, such as the consistency between the beams applied during model training and those applied during model inference. Furthermore, "consistency" can be extended to other LCM stages.

[0069] In operation, device 110 determines information on the consistency of model-related or functional-related information between a first ML stage for realizing a model or functionality and a second ML stage for realizing the model or functionality.

[0070] The details of the information used to achieve consistency will be discussed below.

[0071] In some implementations, the information used to achieve consistency may indicate at least one of the following: First information indicating at least one reference mode used or supported. Second information indicating at least one reference value used or supported. Third information indicating at least one hypothesis used or supported. A fourth piece of information indicating whether consistency is required. The fifth piece of information indicates the tolerance range associated with this information.

[0072] In some implementations, the information used to achieve conformance (also referred to as calibration information or conformance information) may include the following: Reference mode information (i.e., first information); additionally, for AI / ML models developed for beam management, the reference mode is the reference beam mode; Other calibration information, or consistency information, includes reference values ​​and / or consistency assumptions for parameters such as transmit power, receiver type, speed, rotation, etc., i.e., second and third information.

[0073] In some implementations, the reference pattern (first information) of the AI / ML model used for beam management may be represented by the following: antenna modeling information, beamforming information, the relationship between beamforming information of set B beams and the corresponding set A beams and resource IDs.

[0074] In some implementations, a model may correspond to a single piece of antenna modeling / beamforming information, or it may correspond to a relationship between beamforming information and at least one resource. In other words, the mapping between the model and the relationship between the antenna modeling / beamforming information and at least one resource is a one-to-one mapping. To clarify, although a one-to-one mapping is discussed, in some cases the mapping can be changed; for example, the mapping may be a many-to-one mapping, a one-to-many mapping, or a many-to-many mapping.

[0075] In summary, one or more reference patterns (first information) can be specific to one or more models or functionalities.

[0076] In some implementations, the first information may include at least one of the following: Antenna modeling information associated with at least one of device 110 or another device 120, Beamforming information associated with at least one of device 110 or another device 120, Orientation information between device 110 and another device 120, or The relationship between beamforming information and at least one resource, each of which is associated with a beam.

[0077] In some implementations, the first information may include NW-UE orientation information, for example, the NW panel and the UE panel are parallel to each other and point towards each other's wide side.

[0078] In some implementations, the first information may include the relationship or transformation between the local coordinate system and the global coordinate system, which can be used to define the radiation pattern, pointing direction, and azimuth / elevation information. Information such as NW-UE orientation.

[0079] In some implementations, the antenna modeling information may include at least one of the following: The number of antenna elements in the first spatial dimension is denoted as M; The number of antenna elements in the second spatial dimension is denoted as N; The number of panels in the first spatial dimension is denoted as M. g , The number of panels in the second spatial dimension is denoted as N. g , The spacing between antenna elements in the first spatial dimension is denoted as d. H , The spacing between antenna elements in the second spatial dimension is denoted as d. v , The panel spacing in the first spatial dimension is represented as d. g,V , The panel spacing in the second spatial dimension is represented as d. g,H , The polarization number, denoted by P, is Antenna array type, Antenna numbering rules Polarization angle, Polarization type, Antenna element radiation modes in the first spatial dimension Antenna element radiation modes in the second spatial dimension For the combination method of multidimensional antenna element modes, Maximum directional gain of antenna elements, For at least one complex weight of the antenna element at the elevation angle, or Antenna height.

[0080] For better understanding, some example implementations regarding antenna modeling information will be discussed below.

[0081] In some implementations, the antenna modeling information may include antenna modeling information for NW, for example, including (M, N, P, M g , N g ) and (d V , d H In addition to (d) g,V , d g,H ), where N is the number of columns (or the number of horizontal antenna elements), M is the number of antenna elements with the same polarization in each column, P is the number of polarizations, and M g N g These represent the number of vertical and horizontal panels, respectively. The antenna elements are arranged in the horizontal direction at angles d... H The intervals are evenly spaced, and in the vertical direction at d V The antenna panels are evenly spaced apart, and in addition, the antenna panels are spaced apart in both the horizontal and vertical directions at (d g,V ,d g,H They are evenly spaced.

[0082] Alternatively or in addition, in some implementations, the antenna modeling information may include one or more of the following information / parameters: Cross-polarization arrays, uniform linear arrays, etc. Antenna numbering, for example, assuming an antenna array viewed from the front (where the x-axis points to the wide side and the y-coordinate increases with the column number). Polarization angle, Polarization types: linear (crossing, vertical, horizontal), circular, elliptical, etc. Vertical radiation mode of antenna element (dB). Horizontal radiation mode of antenna element (dB) For the combination method of 3D antenna element modes (dB), Maximum directional gain of antenna elements, For the complex weights of antenna elements at the elevation angle, or Antenna height, etc.

[0083] Now for reference Figure 4 The illustration shows a 2-D planar antenna structure 400, where each column is a cross-polarized array. It should be noted that although the discussion is conducted by assuming a 2-dimensional (2D, 2-dimension) planar antenna array structure (i.e., antenna elements are placed in the vertical and horizontal directions), the assumed antenna array structure could also be a 1-dimensional (1D, 1-dimension) linear array or other antenna structures. This disclosure is not limited in this respect.

[0084] Alternatively or otherwise, in some implementations, the antenna modeling information may also include antenna modeling information for the UE, which is similar to the antenna modeling information for the NW. For the sake of brevity, identical or similar content will be omitted.

[0085] In some implementations, the antenna modeling information may also be different for set B beams and the corresponding set A beams.

[0086] In some implementations, multiple antenna model parameter sets may be associated with an AI / ML model / functionality.

[0087] As discussed above, the mapping between the model and antenna modeling information can be many-to-one, one-to-many, or many-to-many. Therefore, in some implementations, one or more sets of antenna modeling parameters may be associated with an AI / ML model / functionality. Alternatively, in some implementations, a set of antenna modeling parameters may be associated with one or more AI / ML models / functionalities.

[0088] In some implementations, the beamforming information includes at least one of the following: Beamforming type The number of transceiver units (TXRUs) The number of beams, or At least one beamforming weight.

[0089] In some implementations, the beamforming type is one of the following: analog beamforming, digital beamforming, or hybrid beamforming.

[0090] In some implementations, the number of TXRUs can be the total number of TXRUs, or it can be associated with at least one of the following: having a specific type of beam set, a specific beam type, a specific spatial dimension, or a specific spatial angle.

[0091] In some implementations, the number of beams can be the total number of beams, or it can be associated with at least one of the following: having a specific type of beam set, a specific beam type, a specific spatial dimension, or a specific spatial angle.

[0092] In some implementations, the at least one beamforming weight includes: at least one discrete Fourier transform (DFT) weight, a set of weights associated with a particular set of beams, and a correspondence between the beam and weight matrices.

[0093] For better understanding, some example implementations of beamforming information will be discussed below.

[0094] In some implementations, beamforming information for NW is discussed below.

[0095] In some implementations, beamforming information may include beamforming type, such as analog, digital, hybrid, etc.

[0096] In some implementations, beamforming information may include the number of TXRUs, and in addition, the number of antenna elements per TXRU, such as... The total number of TXRUs, the number of TXRUs used for beams in set A, the number of TXRUs used for beams in set B, the number of TXRUs used for first-type beams (e.g., wide beams or SSB beams), and the number of TXRUs used for second-type beams (e.g., narrow beams or CSI-RS beams). The number of TXRUs used for horizontal / vertical beams, and the number of TXRUs used for azimuth / elevation beams, respectively. The number of TXRUs used for beaming in the first / second dimension (e.g., horizontal / vertical). In some implementations, beamforming information may include the number of beams, and may also include one or more of the following: The total number of NW Tx beams, the number of beams in set A, the number of beams in set B, the number of first-type beams (e.g., wide beams or SSB beams), and the number of second-type beams (e.g., narrow beams or CSI-RS beams). The number of horizontal beams, the number of vertical beams, or Number of azimuth beams ( ), number of elevation beams ( ).

[0097] In some implementations, beamforming information may include beamforming weights for each beam (or TXRU weight mapping, codebook, etc.), for example, 2D DFT weights. Azimuth beam and The weight matrix W for each elevation angle beam can be expressed by the following formula:

[0098]

[0099] in Represents the Kronecker product. It is a weight matrix. and It is a beamforming vector. It is the number of azimuth beams. It is the number of elevation beams. They are basis vectors. It was applied and vector ,and It was applied and vector , It is the azimuth value, and It is the elevation angle value.

[0100] It should be understood that, in addition to the DFT method mentioned above, the weight matrix... It can also be defined using methods other than DFT.

[0101] In some implementations, beamforming weights may also include weights for beams in sets B and A, where set A may be a subset of all NW beams. Furthermore, if the weights of beams in sets B and A are known, the relationship between sets B and A can be determined.

[0102] In some implementations, beamforming weights may also include the order / index of beams in set B and set A, wherein each beam can be referenced by the following: Weight vector used to generate beams Regarding the position in the beamforming weight matrix W (e.g., the i-th column and j-th row, or respectively in the beamforming vector...) , The position in, for example, respectively in The i-th element and in The information of the j-th element in the data. Information, such as the azimuth and elevation angles of the beam, or The position / order within this set. For example, it could also be the azimuth of departure angle (AOD, possibly after scaling) or the zenith of departure angle (ZOD, possibly after scaling); for the Rx beam on the UE side, the azimuth of arrival angle (AOA) or the zenith of arrival angle (ZOA) can be used.

[0103] In some implementations, beamforming weights may also include further optimization terms for the weights, such as beam orthogonality, oversampling DFT, antenna gain pattern shaping, sidelobe / backlobe control weights, and gating control weights. These further optimization terms may also include beam shape information such as pointing direction and beamwidth.

[0104] Alternatively or otherwise, in some implementations, beamforming information may include beamforming information for the UE, which is similar to beamforming information for the NW. For the sake of brevity, identical or similar content will be omitted.

[0105] In some implementations, the beamforming information may also be different for set B beams and the corresponding set A beams.

[0106] In some implementations, multiple sets of beamforming information parameters can be associated with an AI / ML model / functionality.

[0107] As discussed above, the mapping between model and antenna modeling information can be many-to-one, one-to-many, or many-to-many. Therefore, in some embodiments, one or more sets of beamforming information parameters may be associated with an AI / ML model / functionality. Alternatively, in some embodiments, a set of beamforming information parameters may be associated with one or more AI / ML models / functionalities.

[0108] In some implementations, beamforming information may be a set of weights, and the relationship between beamforming information and the at least one resource may be a mapping between the set of weights and the at least one resource.

[0109] In some implementations, each element in the weight set may correspond to: a first value in a first spatial dimension and a second value in a second spatial dimension.

[0110] Alternatively, in some implementations, each element in the weight set may correspond to a specific azimuth value and a specific elevation angle.

[0111] In some implementations, the relationship between beamforming information and the at least one resource can be represented by the order in which the at least one resource is mapped to a subset of the weight set, the order including at least one of the following ascending or descending order: First, the first spatial dimension, then the second spatial dimension. First the second spatial dimension, then the first spatial dimension. First the azimuth dimension, then the elevation dimension. First the elevation angle dimension, then the azimuth angle dimension, or The order of multiple beam groups, at least one beam group is divided into the multiple beam groups.

[0112] Alternatively or in addition, in some embodiments, the relationship between beamforming information and the at least one resource may be represented by information indicating the subset of the weight set, which includes at least one of the following: An indication of the beginning position of the subset. The number of rows in the subset. The number of columns in the subset If the elements in this subset are not adjacent in weight, the first step length in the row is, or The second step size in the column when the elements in the subset are not adjacent weights.

[0113] In summary, the reference mode can be represented by the relationship between beamforming weights and resource IDs.

[0114] Now for reference Figure 5 The example mapping 500 between resources and weights is illustrated. Implementations (1) and (2) are resource IDs, while implementations (a), (b) and (c) are weights.

[0115] In some implementations, the relationship between beamforming information and the at least one resource can be represented by a resource ID number (e.g., first horizontal / azimuth, then vertical / elevation, or first vertical / elevation, then horizontal / azimuth), such as... Figure 5 As shown.

[0116] In some implementations, each resource ID may correspond to W (as... Figure 5 Specific implementations (a), (b)) or (as) Figure 5 The specific implementation of (c) in the elements.

[0117] Alternatively, or otherwise, beamgrouping can be applied first, and resource ID numbering as described above can be performed within the group, such as... Figure 5 The specific implementation (2) of the process. The sorting method for these groups can be: first horizontal / azimuth angle, then vertical / elevation angle, and vice versa.

[0118] In addition, numbering can be done across groups: for example, first number the same first relative index within each group across groups, and then number the same second relative index within each group across groups.

[0119] In some implementations, the relationship between beamforming information and the at least one resource can be represented by an indication of that relationship, which may indicate for each resource ID or for a set A / B. i,j or For example, the starting position (row / column, horizontal / vertical) and the number of elements per (row / column, horizontal / vertical). In addition, when beamforming weights are not adjacent weights, the step size information (such as...) is also provided. Figure 5 (The first step length and the second step length).

[0120] The following section will discuss how to determine / exchange the information used to achieve consistency.

[0121] In this disclosure, either device 110 or another device 120 may determine information for achieving consistency and send that information to the other. Therefore, in some embodiments, device 110 may receive 310-1 a first message including information for achieving consistency, and then determine the information for achieving consistency based on the first message.

[0122] Alternatively, in some implementations, device 110 may determine the information used to achieve consistency itself, and then send a second message including the information used to achieve consistency 310-2 to another device 120.

[0123] In some implementations, the first message may be received or the second message may be sent during at least one of the following periods: For the process of transferring models or functional datasets, For the model or functional model transfer process, For data collection processes related to models or functionalities, For the reasoning process of a model or a functional model, For the monitoring process of models or functional models, The training process for a model or a functional model. The registration or identification process for a model or functionality, or Calibration process.

[0124] In some implementations, this information may be included in at least one of the following: Model description, Features or feature groups Ability-related information, Supporting conditional information associated with the model or functionality, or Additional conditional information associated with the model or functionality.

[0125] In some implementations, information for achieving consistency of model-related or functionality-related information, particularly reference mode information (i.e., the first information), may be indicated during at least one of the following: dataset transfer / transfer, data collection for model training, model inference, model monitoring, model transfer / transfer, model / functionality registration / identification, UE capability reporting, or feature / feature group reporting. Alternatively or additionally, the information for achieving consistency of model-related or functionality-related information may be part of conditions / additional conditions associated with the AI / ML model / functionality. In this way, calibration information can be exchanged between different devices.

[0126] Because transmitting information representing the reference mode may require a large data size, it may be more appropriate to use RRC messages, MAC CE, and other suitable signaling.

[0127] In some implementations, if more than one reference mode (or other calibration information) is available / usable, a reference mode ID / index can be provided for each configured / reported / activated reference mode.

[0128] In some implementations, for the corresponding AI / ML model / functionality, the reference mode (or other calibration information) may be indicated in at least one of the following: Dataset transfer / exchange process: Reference pattern information is indicated along with the dataset, or the mapping between reference pattern IDs / indexes and dataset IDs / indexes is predefined, NW configured, or reported by the UE; alternatively, the reference pattern may be implied by the dataset, i.e., implicitly indicated. In some implementations, reference pattern information may be associated with a dataset, the reference pattern information may be part of the dataset, or the reference pattern information may be reflected in the dataset information.

[0129] Data collection process: Reference pattern information is indicated along with data collection trigger information or data collection configuration; alternatively, the reference pattern may be implied by the data collection trigger information or data collection configuration, i.e., implicitly indicated. In some embodiments, the reference pattern information may be associated with the data collection trigger information or data collection configuration, and the reference pattern information may be part of the data collection trigger information or data collection configuration, or the reference pattern information may be reflected by the data collection trigger information or data collection configuration.

[0130] Model transfer / transmission process: Reference mode information is indicated along with the model, for example, when an AI / ML model is transferred from the NW to the UE, or from a third party to the UE. Alternatively, the reference mode may be implied by the model information, i.e., implicitly indicated. In some implementations, reference mode information may be associated with model information; the reference mode information may be part of the model information, or the reference mode information may be reflected by the model information.

[0131] The model / functionality registration / identification process; for model ID-based AI / ML lifecycle management, reference schema information can be indicated in the model ID or model description for each model; for functionality-based AI / ML lifecycle management, reference schema information can be indicated for each AI / ML feature / feature group. Alternatively, the reference schema can be implied by the model ID or model description, i.e., implicitly indicated. In some implementations, reference schema information can be associated with the model ID or model description, and the reference schema information can be part of the model ID or model description, or the reference schema information can be reflected by the model ID or model description.

[0132] Alternatively or otherwise, calibration information (especially reference mode information) may be part of conditions / additional conditions associated with the AI / ML model / functionality, for example, as one of the auxiliary information. Furthermore, if more than one set of condition / additional condition parameters is available / usable, a condition / additional condition ID / index may be provided for each set of condition / additional condition parameters. Alternatively, the reference mode may be implied by the condition / additional conditions, i.e., implicitly indicated. In some implementations, reference mode information may be associated with conditions / additional conditions, and the reference mode information may be part of the condition / additional conditions, or the reference mode information may be reflected by the condition / additional conditions.

[0133] In some implementations, for the UE-side model, the UE can provide a reference pattern (or other calibration information) via a UE capability report or via a feature / feature group. Furthermore, the reference pattern (or other calibration information) can be used for online model training, model inference, and model monitoring.

[0134] In some implementations, the UE capability report or feature / feature group report can provide information on whether the UE requires consistency between model training and model inference to activate / apply AI / ML models or functionality.

[0135] In some implementations, the UE capability report or feature / feature group report can provide information on whether the UE requires consistency between offline (or non-real-time) model training and online (or real-time) model training (and / or fine-tuning), and / or consistency between model inference and model monitoring.

[0136] In some implementations, UE capability reports or feature / feature group reports may provide a support reference pattern, such as, These patterns can be used during model training or offline model training, or The number of reference patterns may be the same or different for different AI / ML models / functionalities.

[0137] As an example, for supported AI / ML models / functional spatial domain beam prediction, where set A has 64 beams and set B is 1 / 8 the size of set A, the supported reference modes could be: 1) having a beam-based... 2DDFT weights A set of beams A, and having as follows Figure 2D The specific implementation in (a) or (b) is a set of 8 beams B that are uniformly selected; 2) having a basis 2D DFT weights A set of beams A, and a set of beams B, etc.

[0138] In some implementations, UE capability reports or feature / feature group reports can provide tolerance ranges / thresholds for support of differences from the reference mode, within which model inference can still function. For example, for sets A and B in the reference mode, the permissible beams in set B can satisfy... ,in It is the tolerance range / threshold. These are the weights of the beams in set B. It is the weight of the reference beam.

[0139] For NW-side models, typically, at least for DL ​​Tx beam prediction, the NW may not need to provide a reference mode. However, to control UE-side beamforming used for measurement and reporting, the NW may also provide a reference mode, which can be done via QCL / Rx beam indication. In some implementations, the NW may request the UE to apply specific QCL assumptions / Rx beamforming weights to measurement and reporting, or simply require the UE to apply the same QCL assumptions / Rx beamforming.

[0140] The following section will discuss how to utilize information used to achieve consistency.

[0141] In some implementations, device 110 may perform a first ML phase 320 with another device 120 based on information used to achieve consistency. Alternatively or in addition, in some implementations, device 110 may perform a second ML phase 330 with another device 120 based on information used to achieve consistency.

[0142] In some implementations, where device 110 is a network device 110 and another device 120 is a terminal device 110, device 110 may send at least one of the following to the other device 120 based on information for achieving consistency: measurement configuration to be used by the other device 120 for data collection, or information about updated beamforming information.

[0143] In some implementations, where device 110 is a terminal device 110 and another device 120 is a network device 110, device 110 may perform measurements on at least one beam based on information used to achieve consistency, and may send the measurement results of at least a portion of the at least one beam to the other device 120.

[0144] For better understanding, please refer to the following: Figure 6A and Figure 6B The following illustrates signaling flows 600A and 600B of communications according to some embodiments of the present disclosure.

[0145] exist Figure 6A and Figure 6B In the example, calibration information (especially the reference mode) can be used in the following processes: data collection for model training, model inference, model monitoring and other ML stages, enabling calibration on the UE / NW side for inputs collected from different NW devices.

[0146] In some implementations, for the UE-side model, based on a reference mode, the NW can determine / adjust its resource configuration and transmit beamforming weights for each resource for use in UE-side model training, inference, and monitoring, respectively.

[0147] like Figure 6A As illustrated, in some implementations, the NW may determine / adjust its resource configuration (i.e., set B) or transmit beamforming weights for each resource for model inference (i.e., for the UE to use to predict the best beam in set A).

[0148] For model training, model monitoring, or other LCMs, NW can determine / adjust its resource configuration and / or transmit beamforming weights for each resource used in both set B and set A, as well as the relationship between set B and set A.

[0149] It should be noted that when different UEs have different reference modes, NW equipment has difficulty meeting the requirements of all UEs, which means that it is almost impossible to adjust beamforming weights, but resource configuration can still be adjusted.

[0150] For the NW-side model, the NW can provide reference mode information to the UE to collect data from the UE for model training, model inference, and model monitoring on the NW side.

[0151] like Figure 6B As illustrated, in some implementations, the NW may require the UE to measure set B beams with reference to the reference Rx beam and provide beam reports for NW-side model inference (i.e., for the NW to use to predict the best beam in set A based on the UE report set B measurement results).

[0152] In some implementations, for model training, model monitoring, or other LCMs, the NW may require the UE to also adjust its Rx beamforming weights for each resource used in both set B and set A.

[0153] In some implementations, for NW-side models or third-party models, reference patterns can be used to exchange data between different NW / UE devices (e.g., different gNBs, cells, etc.).

[0154] As discussed above, calibration information (especially reference mode information) can be used to maintain consistency, which can be provided via UE capabilities or feature / feature group reports, and can be used in processes such as data collection for model training, model inference, and model monitoring, enabling consistency during different stages of AI / ML model LCM (e.g., data collection from different devices).

[0155] In some implementations, the calibration process can be used to check whether consistency has been maintained, so that calibration can be performed based on calibration information (especially reference mode) in the event of a lack of consistency.

[0156] In some implementations, the calibration process may be used during the model inference period (or other ML phases, such as before data collection for model training, before or after model updates) or during the model monitoring period, for example, as a solution for handling model failures or performance degradation.

[0157] In some implementations, the calibration process can also be used in conjunction with other processes, such as before or after cell selection, handover, beam training, etc. Furthermore, the calibration process can be performed periodically, or triggered by signaling or based on predefined events, and can be initiated by the UE or NW.

[0158] Furthermore, it should be clarified that the calibration process can be used in conjunction with other ML processes, or can be part of other ML processes, or can be in other LCM phases. For example, the calibration process can be used in conjunction with or can be part of the following processes: dataset transfer / delivery, data collection for model training, model inference, model monitoring, model transfer / delivery, model / functionality registration / identification, UE capability reporting, or feature / feature group reporting. More details are discussed below.

[0159] In some implementations, device 110 may perform a calibration process with another device 120 based on information used to achieve consistency.

[0160] In this disclosure, either device 110 or another device 120 can initiate a calibration process. Therefore, in some embodiments, device 110 can send a request 340-2 to another device 120 to initiate a calibration process.

[0161] Alternatively, in some embodiments, device 110 may receive a request from another device 120 340-1 to initiate a calibration process.

[0162] In this disclosure, the sending of a request may be performed based on a triggering event or by initiating a calibration process, as described below.

[0163] In some implementations, device 110 may send a request in response to detecting at least one triggering event for initiating a calibration process, wherein the at least one triggering event may include at least one of the following: Community handover Network device 110 switchover, Transmitter / Receiver Point (TRP) switching, Performance degradation of the model or its functionality, or The channel quality between device 110 and another device 120 is degraded.

[0164] The details of the request will be discussed below.

[0165] In some implementations, the request may include at least one of the following: The first indication used to trigger the calibration process A second instruction for a resource used to request resource allocation or to request an update to its configuration. Model or functional identifier, The identifier of functionality associated with the model or functionality, or Information used to achieve consistency.

[0166] In some implementations, the request may be based on information used to achieve conformance (i.e., calibration information), particularly reference mode information. In some implementations, the request may include information used to achieve conformance, particularly reference mode information.

[0167] More specifically, the request may include calibration information (including reference mode information) used in the first and / or second phases, such as calibration information (including reference mode information) used in model training and / or model inference / monitoring.

[0168] In some implementations, the request may be included in any suitable message, including but not limited to: dedicated signaling, handover request, handover command, beam switching request, or beam switching command.

[0169] In some implementations, upon receiving the request, device 110 may send a response to the request to another device 120, wherein the response indicates confirmation information, which includes at least one of the following: an indication of whether consistency has been maintained, an identifier of the reference mode used by device 110, an indication of whether the calibration information (including reference mode information) used by device 110 in the first and / or second phase is the same as the reference calibration information already exchanged with the other device 120, and the difference between the calibration information (including reference mode information) used by device 110 and the reference calibration information already exchanged with the other device 120.

[0170] In some implementations, the response indicates confirmation information, which includes at least one of the following: an indication of whether consistency has been maintained, an identifier of the reference mode used by device 110, an indication of whether the calibration information (including reference mode information) used by device 110 in the first and / or second phase is the same as the reference calibration information, and the difference between the calibration information (including reference mode information) used by device 110 and the reference calibration information that has been exchanged with another device 120.

[0171] In some implementations, the response indicates confirmation information, which includes at least one of the following: an indication of whether consistency has been maintained, an identifier of the reference mode used by device 110, an indication of whether the calibration information (including reference mode information) used by device 110 in model training and / or model inference / monitoring is the same as the reference calibration information, and the difference between the calibration information (including reference mode information) used by device 110 in model training and / or model inference / monitoring and the reference calibration information that has been exchanged with another device 120.

[0172] In some implementations, the response indicates confirmation information, which includes at least one of the following: an indication of whether consistency has been maintained, an identifier of the reference mode used by device 110, an indication of whether the mode used by device 110 is the same as the reference mode, a difference between the mode used by device 110 and the reference mode, or adjusted beamforming information for communication with another device 120.

[0173] Additionally, in some implementations, when device 110 is a network device 110, after receiving the request, the device sends a response to the request to another device 120, the response indicating the updated resource configuration.

[0174] Additionally, in some implementations, when device 110 is a terminal device 110, after receiving the request, the device sends a response to the request to another device 120, the response including a second indication for requesting resource allocation or requesting an update of the configuration of resources.

[0175] refer to Figure 7A Some example implementations are discussed, and the diagram illustrates the signaling flow 700A of communications according to some implementations of this disclosure. Figure 7A In the example, the UE initiates a calibration process to request the changes needed to maintain consistency.

[0176] In some implementations, an event-based (or condition-based) approach can be used to enable UE-initiated calibration procedures. For example, the event or condition can be defined based on the following: The UE switches to a new TRP / gNB / cell; in this case, the new NW device may or may not be able to follow the reference mode for beamforming and configure set B and / or set A; For UE-side model monitoring, the monitoring results indicate that the AI / ML model performance is below the threshold, which means that the NW configuration used for model inference (e.g., set B configuration) may be inconsistent with the trained model.

[0177] In some implementations, the UE can initiate the calibration process by transmitting a calibration request.

[0178] In some implementations, a calibration request may be a simple field used to trigger calibration of the AI / ML model / functionality for the current application. Alternatively or otherwise, a calibration request may include model / functionality information.

[0179] In some implementations, the calibration request may be based on a reference mode; for example, the calibration request may also include reference mode information.

[0180] In some implementations, a calibration request may include information for: requesting resources / configuration, or requesting updates to the current configuration, such as requesting to add / remove / update resources in set B for measurement via resource ID or via weights / angles described in the reference pattern, requesting updates on the number of predicted beams to be reported, or requesting updates on the number of historical measurements used for future beam prediction.

[0181] In some implementations, calibration requests may be signaled via dedicated signaling or carried in other signaling (such as handover requests / commands, beam switching requests / commands, etc.).

[0182] In some implementations, the NW can transmit a response to a request by providing acknowledgment information, which may be a simple message indicating whether consistency has been maintained.

[0183] In some implementations, the confirmation information may also be based on a reference pattern associated with the AI / ML model / functionality of the current application, for example, which reference pattern is assumed at NW if multiple patterns are associated; whether the applied pattern is the same as the reference pattern, or whether it is within the tolerance range compared to the reference pattern; or the difference from the reference pattern.

[0184] In some implementations, the NW may adjust the configuration based on UE requests and / or reference modes, for example, adding / removing / updating resource configurations for measurement for model inference, monitoring, truth reporting and / or model training, etc.; adding / removing / updating reporting configurations for model inference, monitoring, truth reporting and / or model training, etc.

[0185] In some implementations, Tx beamforming weights can be adjusted based on UE requests and / or reference modes. For example, the relationship between set B beams and set A beams can be adjusted for use in model inference, monitoring, truth reporting, and / or model training, or the Tx beamforming weights can be adjusted.

[0186] In some implementations, a UE capability report can be generated regarding whether the UE can support initiating calibration.

[0187] In some implementations, the UE may be provided with configuration information about events or conditions that trigger calibration or predefined or NW-configured periodic calibration.

[0188] In some implementations, the UE may be provided with configuration information about the resources used to transmit calibration requests, such as dedicated UL resources, like dedicated PUSCH, dedicated PUCCH, etc.

[0189] exist Figure 7B In the example, NW initiates a calibration process to maintain consistency.

[0190] exist Figure 7B In this context, NW can initiate the calibration process by sending a calibration request.

[0191] In some implementations, a calibration request may be a simple field used to trigger calibration of the AI / ML model / functionality for the current application. Alternatively, the calibration request may include model / functionality information. In addition, the calibration request may be based on a reference mode, for example, the calibration request may also include reference mode information.

[0192] In some implementations, calibration requests may be signaled via dedicated signaling or carried in other signaling (such as handover requests / commands, beam switching requests / commands, etc.).

[0193] In some implementations, the UE can transmit a response to a request by providing confirmation information, which may be a simple message indicating whether consistency has been maintained.

[0194] In some implementations, the confirmation information may also be based on reference patterns associated with the AI / ML model / functionality of the current application, for example, If multiple modes are associated, which reference mode is assumed at the UE; Is the applied pattern the same as the reference pattern, or is it within the tolerance range compared to the reference pattern? Differences from the reference mode.

[0195] In some implementations, the UE may adjust the Rx beamforming weights based on NW requests and / or reference modes, for example, adjusting the Rx beamforming weights used for measurement and reporting of set B beams and set A beams for model inference, monitoring, truth reporting and / or model training, etc.

[0196] In some implementations, the UE may transmit a response based on an NW request and / or reference mode to further request resources / configuration, or to request updates to the current configuration, for example, Request to add / remove / update resource and reporting configurations for measurement purposes, such as model inference, monitoring, truth reporting, and / or model training. Request an update on the number of predicted beams to be reported. Request an update on the number of historical measurements used for future beam prediction.

[0197] In some implementations, a UE capability report can be generated regarding whether the UE can support NW-initiated calibration.

[0198] In some implementations, the UE may be provided with configuration information about resources used to transmit acknowledgments or other requests (such as dedicated UL resources, such as dedicated PUSCH, dedicated PUCCH, etc.).

[0199] In this disclosure, the NW can initiate early calibration for candidate cells to prepare for consistency, enabling calibration for measurement and reporting on the UE side to support mobility. Details are discussed below.

[0200] In some implementations, additional information may be exchanged or utilized to ensure consistency of model-related or functionally relevant information between different ML stages used to achieve another model or another function used in candidate cells. Such processes will be discussed.

[0201] In some implementations, the calibration process for candidate cells may be instructed by the source cell or the candidate cell.

[0202] In some implementations, device 110 is a network device 110 that provides a source cell, and device 110 may determine additional information for achieving consistency of model-related or functional-related information between different ML stages of another model or another function used in a candidate cell, and then send a request to another device 120 to initiate a calibration process for the candidate cell.

[0203] Upon receiving a request to initiate a calibration process for a candidate cell, another device 120 may respond to device 110 with a response to the request for the candidate cell. Device 110 may then send that response to the candidate cell.

[0204] In some implementations, device 110 is a terminal device 110, and device 110 may receive a request from a source cell or a candidate cell to initiate a calibration process for the candidate cell. Device 110 may then send a response to the request for the candidate cell to the source cell or the candidate cell based on additional information for achieving consistency of model-related or functional-related information between different ML stages used in another model or another functionality in the candidate cell.

[0205] Now for reference Figure 8 This illustrates a signaling flow 800 of communication according to some embodiments of this disclosure.

[0206] exist Figure 8 In the example, NW can initiate a calibration process for other cells by transmitting a calibration request, where other cells can refer to neighboring cells, cells with different cell IDs, different PCIs, etc.

[0207] In addition, other cells can be configured as candidate cells, or they can be target cells for handover / handover.

[0208] In some implementations, the source / serving cell and the candidate / target cell may exchange calibration information, including a reference mode.

[0209] With the introduction of another piece of information, it is necessary to further distinguish between the information for the source cell and the other information for the candidate cell.

[0210] In some implementations, the request to initiate a calibration process for a candidate cell and the response to the request for a candidate cell indicate an identifier associated with another piece of information. In some implementations, the identifier associated with the other information may be an index of that other information. Alternatively, the identifier associated with the other information may be implied by any suitable identifier that corresponds to that other information. For example, the information corresponds to a source cell, and the other information corresponds to a candidate cell. In this case, the identifier of the candidate cell may imply / indicate the other information for the candidate cell.

[0211] In addition, a mapping between cell ID and reference mode ID can be established, and the mapping can be notified to the UE. Candidate cell configuration can be used to provide a reference mode assumed in other cells.

[0212] In some implementations, the reference pattern ID can also be implicitly signaled by signaling the cell ID.

[0213] In some implementations, the UE can transmit a response to a request by providing acknowledgment information to the serving cell / other cells, which may be a simple message indicating whether consistency has been maintained.

[0214] In some implementations, the confirmation information may also be based on reference patterns from multiple cells associated with the AI / ML model / functionality of the current application, for example, Which cell has been assumed, for example, the cell ID? If multiple modes are associated, which reference mode is assumed at the UE? Is the applied pattern the same as the reference pattern, or is it within the tolerance range compared to the reference pattern, or... Differences from the reference mode.

[0215] In some implementations, the source cell and the candidate / target cell may exchange UE confirmation information.

[0216] In some implementations, the UE can adjust the Rx beamforming weights based on NW requests and / or reference modes, for example, Adjust the Rx beamforming weights for measurement and reporting of the set B beams and set A beams used for the source / serving cell for model inference, monitoring, truth reporting, and / or model training, etc.; or Adjust the Rx beamforming weights for measurements and reporting of set B and set A beams used in other cells for model inference, monitoring, truth reporting, and / or model training, etc. The model can be used to predict beams in other cells.

[0217] In some implementations, the UE may transmit responses to the serving cell or other cells based on NW requests and / or reference modes to: request resources / configuration, or request updates to the current configuration, for example, Request to add / remove / update resource and reporting configurations for measurement purposes, such as model inference, monitoring, truth reporting, and / or model training. Request an update on the number of predicted beams to be reported; or Request an update on the number of historical measurements used for future beam prediction.

[0218] In some implementations, a UE capability report can be generated regarding whether the UE can support NW-initiated calibration for a cell different from the serving cell.

[0219] In this disclosure, the UE can initiate early calibration for candidate cells to prepare for consistency, enabling calibration for measurement and reporting on the UE side to support mobility.

[0220] In some implementations, the calibration process for candidate cells can be instructed by the terminal device, as discussed below.

[0221] In some implementations, device 110 is a terminal device 110, and device 110 can send a request to a source cell or a candidate cell to initiate a calibration process for the candidate cell, and then device 110 can receive a response to the request for the candidate cell from the source cell or the candidate cell.

[0222] In some implementations, device 110 is a network device 110 that provides a source cell, and another device 120 is a terminal device 110. Device 110 may receive a request from the other device 120 to initiate a calibration process for a candidate cell, and may then send the request to the candidate cell to initiate a calibration process for the candidate cell.

[0223] As a candidate cell, a candidate cell can respond to a request for a candidate cell by responding to a request for a candidate cell. Therefore, after receiving a response to a request for a candidate cell, device 110 can send that response to another device 120.

[0224] Now for reference Figure 9This illustrates a signaling flow 900 of communication according to some embodiments of this disclosure.

[0225] exist Figure 9 In the example, the UE can initiate a calibration process for other cells by transmitting a calibration request, where other cells can refer to neighboring cells, cells with different cell IDs, different PCIs, etc.

[0226] In addition, other cells can be configured as candidate cells, or they can be target cells for handover / handover.

[0227] In some implementations, the source / serving cell and the candidate / target cell may exchange calibration information, including a reference mode.

[0228] With the introduction of another piece of information, it is necessary to further distinguish between the information for the source cell and the other information for the candidate cell.

[0229] In some implementations, the request to initiate a calibration process for a candidate cell and the response to the request for a candidate cell indicate an identifier associated with another piece of information. In some implementations, the identifier associated with the other information may be an index of that other information. Alternatively, the identifier associated with the other information may be implied by any suitable identifier that corresponds to that other information. For example, the information corresponds to a source cell, and the other information corresponds to a candidate cell. In this case, the identifier of the candidate cell may imply / indicate the other information for the candidate cell.

[0230] In some implementations, a mapping between cell IDs and reference pattern IDs can be established, and candidate cell configurations can be used to provide reference patterns assumed at other cells.

[0231] In some implementations, the reference pattern ID can also be implicitly signaled by signaling the cell ID.

[0232] In some implementations, the calibration request may include information for requesting resources / configuration for the serving cell and / or other cells, or for requesting updates to the current configuration; for example, the cell ID may be included in the request.

[0233] In some implementations, the NW can transmit a response to a request by providing acknowledgment information to the serving cell / other cells, which may be a simple message indicating whether consistency has been maintained.

[0234] In some implementations, the confirmation information may also be based on a reference pattern of multiple cells associated with the AI / ML model / functionality of the current application, for example, which cell is assumed, such as the cell ID being included in the response / confirmation information.

[0235] In some implementations, the source cell and the candidate / target cell may exchange UE confirmation information.

[0236] In some implementations, the source cell and candidate / target cell can be configured for the serving cell / other cells based on UE requests and / or reference patterns.

[0237] In some implementations, the source cell and candidate / target cell can be adjusted for Tx beamforming weights used for the serving cell / other cells based on UE requests and / or reference patterns.

[0238] In some implementations, the source cell and the candidate / target cell may exchange information about adjusted weights or adjusted configurations.

[0239] In some implementations, the NW may be provided with a UE capability report regarding whether the UE can support initiating calibration for other cells.

[0240] As discussed above, in Figure 8 and Figure 9 In this example, capability-related and configuration information can be exchanged between device 110 and another device 120. For better understanding, the example capability-related and configuration information is outlined below.

[0241] Additionally, in some implementations, capability-related information can be exchanged between device 110 and another device 120. In this way, the corresponding device can better understand the capability-related information of other devices.

[0242] In some implementations, where device 110 is a terminal device 110 and another device 120 is a network device 110 providing the source cell, device 110 may send capability-related information to the other device 120, the capability-related information indicating at least one of the following: Does device 110 support initiating a calibration process for the source cell or candidate cell, or... Does device 110 support another device 120 initiating a calibration process?

[0243] In some implementations, device 110 may receive relevant information that can be used during the calibration process. For example, device 110 may receive configuration information from another device 120 that indicates at least one of the following: At least one triggering event is used to initiate the calibration process. Used to initiate the periodicity of the calibration process, The resource requested by device 110, or The resource used by device 110 to send a response to a request.

[0244] The above process ensures the consistency of model-related or functionally relevant information across different ML stages.

[0245] Example Method Figure 9 A flowchart illustrating a communication method 900 implemented at a device according to some embodiments of this disclosure is shown. For discussion purposes, [the following will be discussed]. Figure 1 The angle description method of device 110 in 900.

[0246] At box 910, the device determines information on the consistency of model-related or functional-related information between a first machine learning (ML) stage used to implement a model or functionality and a second ML stage of the model or functionality.

[0247] At box 920, the device performs at least one of the following based on the information: a calibration process with another device, the first ML stage, or the second ML stage.

[0248] In some example implementations, the processor is further configured to cause the device to perform one of the following: determining the information based on a first message including information for achieving consistency, the first message being sent by another device; or, after determining the information for achieving consistency, sending a second message including the information to another device, wherein the information indicates at least one of the following: first information indicating at least one used or supported reference mode, second information indicating at least one used or supported reference value, third information indicating at least one used or supported assumption, fourth information indicating whether consistency is required, and fifth information indicating a tolerance range associated with the information.

[0249] In some example implementations, the first information includes at least one of the following: antenna modeling information associated with at least one of the device or another device, beamforming information associated with at least one of the device or another device, orientation information between the device and the other device, or the relationship between the beamforming information and at least one resource, each of the at least one resource being associated with a beam.

[0250] In some example implementations, the antenna modeling information includes at least one of the following: the number of antenna elements in a first spatial dimension, the number of antenna elements in a second spatial dimension, the number of panels in a first spatial dimension, the number of panels in a second spatial dimension, the spacing between antenna elements in a first spatial dimension, the spacing between antenna elements in a second spatial dimension, the spacing between panels in a first spatial dimension, the spacing between panels in a second spatial dimension, the number of polarizations, the antenna array type, the antenna numbering rule, the polarization slant angle, the polarization type, the radiation pattern of the antenna elements in a first spatial dimension, the radiation pattern of the antenna elements in a second spatial dimension, the combination method for the multidimensional antenna element patterns, the maximum directional gain of the antenna elements, at least one complex weight for the antenna elements at the elevation angle, or the antenna height.

[0251] In some example implementations, beamforming information includes at least one of the following: beamforming type, number of transceiver units (TXRUs), number of beams, or at least one beamforming weight.

[0252] In some example implementations, the beamforming type is one of the following: analog beamforming, digital beamforming, hybrid beamforming; the number of TXRUs is the total number of TXRUs or is associated with at least one of the following: a beam set of a specific type, a specific beam type, a specific spatial dimension, or a specific spatial angle; the number of beams is the total number of beams or is associated with at least one of the following: a beam set of a specific type, a specific beam type, a specific spatial dimension, or a specific spatial angle; the at least one beamforming weight includes: at least one discrete Fourier transform (DFT) weight, a set of weights associated with a specific beam set, and a correspondence between beams and weight matrices.

[0253] In some example implementations, beamforming information is a set of weights, and the relationship between beamforming information and the at least one resource is a mapping between the weight set and the at least one resource, wherein the relationship between beamforming information and the at least one resource is represented by at least one of the following: an order for mapping the at least one resource to a subset of the weight set, the order including ascending or descending order of at least one of the following: "first spatial dimension first, then second spatial dimension", "second spatial dimension first, then first spatial dimension", "azimuth dimension first, then elevation dimension", "elevation dimension first, then azimuth dimension", or "the order of multiple beam groups, at least one beam group being divided into the multiple beam groups"; or information indicating the subset of the weight set, the information including at least one of the following: "an indication of the start position of the subset", "the number of rows in the subset", "the number of columns in the subset", "the first step length in the row when the elements in the subset are not adjacent weights", or "the second step length in the column when the elements in the subset are not adjacent weights".

[0254] In some example implementations, each element in the weight set corresponds to: a first value in a first spatial dimension and a second value in a second spatial dimension, or a specific azimuth value and a specific elevation angle.

[0255] In some example implementations, a first message is received or a second message is sent during at least one of the following: a dataset transfer process for a model or function, a model transfer process for a model or function, a data collection process for a model or function, a model inference process for a model or function, a model monitoring process for a model or function, a model training process for a model or function, a registration or identification process for a model or function, or a calibration process.

[0256] In some example implementations, the information used to achieve consistency is included in at least one of the following: model description, features or feature groups, capability-related information, supporting conditional information associated with the model or functionality, or additional conditional information associated with the model or functionality.

[0257] In some example implementations, the device is a network device and the other device is a terminal device, which may send at least one of the following to the other device based on information for achieving consistency: a measurement configuration to be used by the other device for data collection, or information about updated beamforming information; and wherein the device is a terminal device and the other device is a network device, the processor is further configured to cause the device to: perform measurements on at least one beam based on the information for achieving consistency; and send the measurement results of at least a portion of the at least one beam to the other device.

[0258] In some example implementations, the device may perform one of the following: send a request to another device to initiate a calibration process, or receive a request from another device to initiate a calibration process.

[0259] In some example implementations, the device may send a request in response to detecting at least one triggering event for initiating a calibration process, the at least one triggering event including at least one of the following: cell handover, network device handover, transmit / receive point (TRP) handover, model or functional performance degradation, or channel quality degradation between the device and another device.

[0260] In some example implementations, the request includes at least one of the following: a first indication for triggering the calibration process, a second indication for requesting resource allocation or requesting an update of the configuration of a resource, an identifier of a model or functionality, an identifier of a functionality associated with a model or functionality, or information for achieving consistency.

[0261] In some example implementations, the request is one of the following: dedicated signaling, handover request, handover command, beam switching request, or beam switching command.

[0262] In some example implementations, upon receiving the request, the device may send a response to the request to another device, the response indicating confirmation information including at least one of the following: an indication of whether consistency has been maintained, an identifier of the reference mode used by the device, an indication of whether the mode used by the device is the same as the reference mode, a difference between the mode used by the device and the reference mode, or adjusted beamforming information for communication with the other device.

[0263] In some example implementations, if the device is a network device, upon receiving the request, the device may send a response to the request to another device indicating an updated resource configuration; or if the device is an end device, upon receiving the request, the device may send a response to the request to another device including a second indication for requesting resource allocation or requesting an updated configuration of resources.

[0264] In some example implementations, the device is a network device that provides a source cell and can: determine additional information for achieving consistency of model-related or functional-related information between different ML stages used in another model or another function used in the candidate cell; send a request to another device to initiate a calibration process for the candidate cell; receive a response to the request for the candidate cell from the other device; and send the response to the candidate cell.

[0265] In some example implementations, the device is a terminal device, and the processor is further configured to cause the device to: receive from a source cell or a candidate cell a request to initiate a calibration process for the candidate cell; and send a response to the request for the candidate cell to the source cell or the candidate cell based on additional information for achieving consistency of model-related or functional-related information between different ML stages of another model or another functionality used in the candidate cell.

[0266] In some example implementations, the device is a terminal device, and the processor is further configured to cause the device to: send a request to a source cell or a candidate cell to initiate a calibration process for the candidate cell; and receive a response to the request for the candidate cell from the source cell or the candidate cell.

[0267] In some example implementations, the device is a network device that provides the source cell, and the other device is a terminal device that can: receive from the other device a request to initiate a calibration process for the candidate cell; send to the candidate cell a request to initiate a calibration process for the candidate cell; receive from the candidate cell a response to the request for the candidate cell; and send the response to the other device.

[0268] In some example implementations, the request to initiate a calibration process for a candidate cell and the response to the request for a candidate cell indicate an identifier associated with another piece of information.

[0269] In some example implementations, the device is a terminal device, and the other device is a network device providing the source cell, and the device can send capability-related information to the other device, the capability-related information indicating at least one of the following: whether the device supports initiating a calibration process for the source cell or candidate cell, or whether the device supports another device initiating a calibration process.

[0270] In some example implementations, the device may receive configuration information from another device indicating at least one of the following: at least one triggering event for initiating a calibration process, periodicity for initiating a calibration process, resources used by the device to send a request, or resources used by the device to send a response to a request.

[0271] In some example implementations, the device is a terminal device or a network device, and the other device is a terminal device or a network device.

[0272] Figure 10 This is a simplified block diagram of a device 1000 suitable for implementing embodiments of the present disclosure. Device 1000 can be considered as follows: Figure 1 Another example implementation of any of the devices shown. Thus, device 1000 may be implemented at or as a part of device 110.

[0273] As shown in the figure, device 1000 includes a processor 1010, a memory 1020 coupled to the processor 1010, a suitable transceiver 1040 coupled to the processor 1010, and a communication interface coupled to the transceiver 1040. The memory 1020 stores at least a portion of a program 1030. Depending on the requirements, the transceiver 1040 can be used for bidirectional or unidirectional communication. The transceiver 1040 may include at least one of a transmitter 1042 and a receiver 1044. The transmitter 1042 and receiver 1044 may be functional modules or physical entities. The transceiver 1040 has at least one antenna to facilitate communication; however, in practice, the access node mentioned in this application may have several antennas. The communication interface can represent any interface necessary for communication with other network elements, such as the X2 / Xn interface for bidirectional communication between eNBs / gNBs, the S1 / NG interface for communication between the Mobility Management Entity (MME) / Access and Mobility Management Function (AMF) / SGW / UPF and eNBs / gNBs, the Un interface for communication between eNBs / gNBs and relay nodes (RNs), or the Uu interface for communication between eNBs / gNBs and terminal equipment.

[0274] Assume that program 1030 includes program instructions that, when executed by the associated processor 1010, enable device 1000 to operate according to embodiments of this disclosure, as referenced herein. Figures 1 to 9 The embodiments discussed herein may be implemented by computer software executable by the processor 1010 of device 1000, or by hardware, or by a combination of software and hardware. The processor 1010 may be configured to implement various embodiments of this disclosure. Furthermore, the combination of the processor 1010 and the memory 1020 may form a processing unit 1050 suitable for implementing various embodiments of this disclosure.

[0275] Memory 1020 can be of any type suitable for a local technology network and can be implemented using any suitable data storage technology, such as, as non-limiting examples, non-transitory computer-readable storage media, semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory, and removable memory. Although only one memory 1020 is shown in device 1000, several physically different memory modules may exist in device 1000. Processor 1010 can be of any type suitable for a local technology network and may include one or more of the following: as non-limiting examples, general-purpose computers, special-purpose computers, microprocessors, digital signal processors (DSPs), and processors based on multi-core processor architectures. Device 1000 may have multiple processors, such as application-specific integrated circuit chips, which are time-dependent on a clock that synchronizes the main processor.

[0276] According to embodiments of this disclosure, an apparatus including circuitry is provided. The circuitry is configured to: determine information regarding the consistency of model-related or functionality-related information between a first machine learning (ML) stage for implementing a model or functionality and a second ML stage of the model or functionality; and based on that information, perform at least one of the following: a calibration process with another device, the first ML stage, or the second ML stage. According to embodiments of this disclosure, the circuitry may be configured to perform any method implemented by the apparatus as discussed above.

[0277] As used herein, the term "circuit" can refer to hardware circuitry and / or a combination of hardware circuitry and software. For example, a circuit can be a combination of analog and / or digital hardware circuitry with software / firmware. As another example, a circuit can be any part of a hardware processor with software, including digital signal processors, software, and memory, which work together to enable a device (such as a terminal device or network device) to perform various functions. In yet another example, a circuit can be hardware circuitry and / or a processor (such as a microprocessor or a portion thereof) that requires software / firmware to operate, but which may be absent when operation is not required. As used herein, the term "circuit" also encompasses a specific implementation of hardware circuitry or a processor, or a portion thereof, and its accompanying software and / or firmware.

[0278] According to embodiments of this disclosure, an apparatus is provided. The apparatus includes: components for determining the consistency of model-related or functionality-related information between a first machine learning (ML) stage for implementing a model or functionality and a second ML stage of the model or functionality; and components for performing at least one of the following based on the information: a calibration process with another device, the first ML stage, or the second ML stage. In some embodiments, the first apparatus may include components for performing corresponding operations of method 900. In some example embodiments, the first apparatus may also include components for performing other operations of method 900 in some example embodiments. The components may be implemented in any suitable form. For example, the components may be implemented as circuitry or software modules.

[0279] In summary, the implementation scheme disclosed herein provides the following aspects.

[0280] In one aspect, an apparatus is proposed, comprising: a processor configured to cause the apparatus to: determine information on the consistency of model-related or functional-related information between a first machine learning (ML) stage for implementing a model or functionality and a second ML stage of the model or functionality; and based on the information to perform at least one of: a calibration process with another device, the first ML stage, or the second ML stage.

[0281] In some implementations, the processor is further configured to cause the device to perform one of the following: determining the information based on a first message including information for achieving consistency, the first message being sent by another device; or, after determining the information for achieving consistency, sending a second message including the information to another device, wherein the information indicates at least one of the following: first information indicating at least one used or supported reference mode, second information indicating at least one used or supported reference value, third information indicating at least one used or supported assumption, fourth information indicating whether consistency is required, and fifth information indicating a tolerance range associated with the information.

[0282] In some implementations, the first information includes at least one of the following: antenna modeling information associated with at least one of the device or another device, beamforming information associated with at least one of the device or another device, orientation information between the device and the other device, or the relationship between the beamforming information and at least one resource, each of the at least one resource being associated with a beam.

[0283] In some implementations, the antenna modeling information includes at least one of the following: the number of antenna elements in a first spatial dimension, the number of antenna elements in a second spatial dimension, the number of panels in a first spatial dimension, the number of panels in a second spatial dimension, the spacing between antenna elements in a first spatial dimension, the spacing between antenna elements in a second spatial dimension, the spacing between panels in a first spatial dimension, the spacing between panels in a second spatial dimension, the number of polarizations, the antenna array type, the antenna numbering rule, the polarization slant angle, the polarization type, the radiation pattern of the antenna elements in a first spatial dimension, the radiation pattern of the antenna elements in a second spatial dimension, the combination method for the multidimensional antenna element patterns, the maximum directional gain of the antenna elements, at least one complex weight for the antenna elements at the elevation angle, or the antenna height.

[0284] In some implementations, beamforming information includes at least one of the following: beamforming type, number of transceiver units (TXRUs), number of beams, or at least one beamforming weight.

[0285] In some implementations, the beamforming type is one of the following: analog beamforming, digital beamforming, hybrid beamforming; the number of TXRUs is the total number of TXRUs or is associated with at least one of the following: a beam set of a specific type, a specific beam type, a specific spatial dimension, or a specific spatial angle; the number of beams is the total number of beams or is associated with at least one of the following: a beam set of a specific type, a specific beam type, a specific spatial dimension, or a specific spatial angle; the at least one beamforming weight includes: at least one discrete Fourier transform (DFT) weight, a set of weights associated with a specific beam set, and a correspondence between beams and weight matrices.

[0286] In some implementations, the beamforming information is a set of weights, and the relationship between the beamforming information and the at least one resource is a mapping between the weight set and the at least one resource, wherein the relationship between the beamforming information and the at least one resource is represented by at least one of the following: an order for mapping the at least one resource to a subset of the weight set, the order including ascending or descending order of at least one of the following: "first spatial dimension first, then second spatial dimension", "second spatial dimension first, then first spatial dimension", "azimuth dimension first, then elevation dimension", "elevation dimension first, then azimuth dimension", or "the order of multiple beam groups, at least one beam group being divided into the multiple beam groups"; or information indicating the subset of the weight set, the information including at least one of the following: "an indication of the start position of the subset", "the number of rows in the subset", "the number of columns in the subset", "the first step length in the row when the elements in the subset are not adjacent weights", or "the second step length in the column when the elements in the subset are not adjacent weights".

[0287] In some implementations, each element in the weight set corresponds to: a first value in a first spatial dimension and a second value in a second spatial dimension, or a specific azimuth value and a specific elevation angle.

[0288] In some implementations, a first message is received or a second message is sent during at least one of the following: a dataset transfer process for a model or function, a model transfer process for a model or function, a data collection process for a model or function, a model inference process for a model or function, a model monitoring process for a model or function, a model training process for a model or function, a registration or identification process for a model or function, or a calibration process.

[0289] In some implementations, the information used to achieve consistency is included in at least one of the following: model description, features or feature groups, capability-related information, supporting conditional information associated with the model or functionality, or additional conditional information associated with the model or functionality.

[0290] In some implementations, the device is a network device and the other device is a terminal device, and the processor is further configured to cause the device to: send to the other device at least one of the following based on information for achieving consistency: a measurement configuration to be used by the other device for data collection, or information about updated beamforming information; and wherein the device is a terminal device and the other device is a network device, the processor is further configured to cause the device to: perform measurements on at least one beam based on information for achieving consistency; and send to the other device the measurement results of at least a portion of the at least one beam.

[0291] In some implementations, the processor is further configured to cause the device to perform one of the following: send a request to another device to initiate a calibration process, or receive a request from another device to initiate a calibration process.

[0292] In some implementations, the processor is further configured to cause the device to send a request in response to detecting at least one triggering event for initiating a calibration process, the at least one triggering event including at least one of the following: cell handover, network device handover, transmit / receive point (TRP) handover, model or functional performance degradation, or channel quality degradation between the device and another device.

[0293] In some implementations, the request includes at least one of the following: a first indication for triggering the calibration process, a second indication for requesting resource allocation or requesting an updated configuration of resources, an identifier of the model or functionality, an identifier of the functionality associated with the model or functionality, or information for achieving consistency.

[0294] In some implementations, the request is one of the following: dedicated signaling, handover request, handover command, beam switching request, or beam switching command.

[0295] In some implementations, the processor is further configured to cause the device, upon receiving the request, to send a response to the request to another device, the response indicating confirmation information including at least one of the following: an indication of whether consistency has been maintained, an identifier of the reference mode used by the device, an indication of whether the mode used by the device is the same as the reference mode, a difference between the mode used by the device and the reference mode, or adjusted beamforming information for communication with the other device.

[0296] In some implementations, the processor is further configured such that the device, if it is a network device, upon receiving the request, sends a response to the request to another device indicating an updated resource configuration, or if it is a terminal device, upon receiving the request, the device sends a response to the request to another device, the response including a second indication for requesting resource allocation or requesting an updated configuration of resources.

[0297] In some implementations, the device is a network device that provides a source cell, and the processor is further configured to cause the device to: determine additional information for consistency of model-related or functional-related information between different ML stages used to achieve another model or another function used in the candidate cell; send a request to another device to initiate a calibration process for the candidate cell; receive a response to the request for the candidate cell from the other device; and send the response to the candidate cell.

[0298] In some implementations, the device is a terminal device, and the processor is further configured to cause the device to: receive from a source cell or a candidate cell a request to initiate a calibration process for the candidate cell; and send a response to the request for the candidate cell to the source cell or the candidate cell based on additional information for achieving consistency of model-related or functional-related information between different ML stages of another model or another functionality used in the candidate cell.

[0299] In some implementations, the device is a terminal device, and the processor is further configured to cause the device to: send a request to a source cell or a candidate cell to initiate a calibration process for the candidate cell; and receive a response to the request for the candidate cell from the source cell or the candidate cell.

[0300] In some implementations, the device is a network device providing the source cell, the other device is a terminal device, and the processor is further configured to cause the device to: receive from the other device a request to initiate a calibration process for the candidate cell; send to the candidate cell a request to initiate a calibration process for the candidate cell; receive from the candidate cell a response to the request for the candidate cell; and send the response to the other device.

[0301] In some implementations, the request to initiate a calibration process for a candidate cell and the response indication to the request for a candidate cell are associated with an identifier and other information.

[0302] In some implementations, the device is a terminal device and the other device is a network device providing the source cell, wherein the processor is further configured to cause the device to: send capability-related information to the other device, the capability-related information indicating at least one of the following: whether the device supports initiating a calibration process for the source cell or candidate cell, or whether the device supports the other device initiating a calibration process.

[0303] In some implementations, the processor is further configured to cause the device to: receive configuration information from another device, the configuration information indicating at least one of the following: at least one triggering event for initiating a calibration process, periodicity for initiating a calibration process, resources used by the device to send a request, or resources used by the device to send a response to a request.

[0304] In some implementations, the device is a terminal device or a network device, and the other device is a terminal device or a network device.

[0305] In one aspect, an apparatus includes: at least one processor; and at least one memory coupled to the at least one processor and storing instructions that, when executed by the at least one processor, cause the apparatus to perform the methods implemented by the apparatus as discussed above.

[0306] In one aspect, a computer-readable medium storing instructions that, when executed on at least one processor, cause the at least one processor to perform the methods implemented by the device discussed above.

[0307] In one aspect, a computer program includes instructions that, when executed on at least one processor, cause the at least one processor to perform the methods implemented by the device as discussed above.

[0308] Generally, various embodiments of this disclosure can be implemented in hardware or special-purpose circuitry, software, logic, or any combination thereof. Some aspects can be implemented in hardware, while others can be implemented in firmware or software executable by a controller, microprocessor, or other computing device. Although various aspects of embodiments of this disclosure are illustrated and described using block diagrams, flowcharts, or other illustrations, it should be understood that, as non-limiting examples, the blocks, apparatuses, systems, techniques, or methods described herein can be implemented in hardware, software, firmware, special-purpose circuitry or logic, general-purpose hardware or controllers or other computing devices, or any combination thereof.

[0309] This disclosure also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer-executable instructions (such as those included in program modules) that execute on a target real or virtual processor in a device to perform the functions described above. Figures 1 to 10 The described process or method. Generally, a program module includes routines, programs, libraries, objects, classes, components, data structures, etc., that perform specific tasks or implement specific abstract data types. The functionality of a program module can be combined in various implementation schemes or split among program modules as needed. The machine-executable instructions used for a program module can be executed on a local or distributed device. In a distributed device, a program module can reside on both local and remote storage media.

[0310] Program code used to perform the methods of this disclosure may be written in any combination of one or more programming languages. This 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 causes 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.

[0311] The aforementioned program code may be embodied on a machine-readable medium, which may be any tangible medium containing or storing a program used by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media may include, but are not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. More specific examples of machine-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 compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0312] Furthermore, although the operations are described in a specific order, this should not be construed as requiring such operations to be performed in the specific order shown or in sequential order, or to perform all the illustrated operations to achieve the desired result. In some environments, multitasking and parallel processing can be advantageous. While several specific implementation details are included in the foregoing discussion, these details should not be construed as limiting the scope of this disclosure, but rather as descriptions of features that may be specific to particular embodiments. Certain features described in the context of individual embodiments may also be implemented in a single embodiment in combination. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0313] Although this disclosure has been described using 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 examples of implementing the claims.

Claims

1. An apparatus, the apparatus comprising: Processor, the processor being configured to cause the device to: Information determining the consistency of model-related or functionality-related information between the first machine-learning (ML) stage used to implement the model or functionality and the second ML stage of the model or functionality; and Based on the information, perform at least one of the following: The calibration process with another device, The first ML stage, or The second ML stage.

2. The device of claim 1, wherein the processor is further configured to cause the device to perform one of the following: The information is determined based on a first message including information for achieving the consistency, the first message being sent by the other device; or After determining the information used to achieve the consistency, a second message including the information is sent to the other device. The information stated therein indicates at least one of the following: First information indicating at least one reference mode used or supported. Second information indicating at least one reference value used or supported. Third information indicating at least one hypothesis used or supported. Indicates whether the fourth piece of information regarding consistency is required, or The fifth piece of information indicates the tolerance range associated with the aforementioned information.

3. The device according to claim 2, wherein the first information includes at least one of the following: Antenna modeling information associated with at least one of the devices. Beamforming information associated with at least one of the devices. Orientation information between the device and the other device, or The relationship between the beamforming information and at least one resource, each of which is associated with a beam.

4. The device of claim 3, wherein the beamforming information is a weight set, and the relationship between the beamforming information and the at least one resource is a mapping between the weight set and the at least one resource. And the relationship between the beamforming information and the at least one resource is represented by at least one of the following: The order used to map the at least one resource to a subset of the weight set includes, but is not limited to, ascending or descending order of at least one of the following: First, the first spatial dimension, then the second spatial dimension. First the second spatial dimension, then the first spatial dimension. First the azimuth dimension, then the elevation dimension. First, the elevation angle dimension, then the azimuth angle dimension, or... The order of multiple beam groups, wherein at least one beam group is divided into the multiple beam groups; or Information indicating the subset of the weight set, the information including at least one of the following: An indication of the starting position of the subset. The number of rows in the subset, The number of columns in the subset, If the elements in the subset are not adjacent in weight, the first step length in the row is, or The second step size in the column when the elements in the subset are not adjacent weights.

5. The device of claim 4, wherein each element in the weight set corresponds to: The first value in the first spatial dimension and the second value in the second spatial dimension, or Specific azimuth and specific elevation angles.

6. The device of claim 2, wherein the first message is received or the second message is sent during at least one of the following: Regarding the data transfer process for the aforementioned model or functionalities, Regarding the model or the functional model transfer process, For the data collection process of the aforementioned model or functionality, The reasoning process for the model or the functional model, The monitoring process for the model or the functional model, The training process for the model or the functional model, The registration or identification process for the model or the functionality, or The calibration process.

7. The device of claim 2, wherein the information for achieving the consistency is included in at least one of the following: Model description, Features or feature groups Ability-related information, Supporting conditional information associated with the model or the functionality, or Additional conditional information associated with the model or the functionality.

8. The device of claim 1, wherein the device is a network device, and the other device is a terminal device, and the processor is further configured to cause the device to: Based on the information used to achieve the consistency, at least one of the following shall be sent to the other device: The measurement configuration used for data collection by the other device, or Information regarding the updated beamforming information; Furthermore, wherein the device is a terminal device, and the other device is a network device, and the processor is further configured to cause the device to: Measurements are performed on at least one beam based on the information used to achieve the consistency; and The measurement results of at least a portion of the at least one beam are transmitted to the other device.

9. The device of claim 1, wherein the processor is further configured to cause the device to perform one of the following: Send a request to the other device to initiate the calibration process, or Receive the request from the other device to initiate the calibration process.

10. The device of claim 9, wherein the request comprises at least one of the following: The first indication used to trigger the calibration process. A second instruction for a resource used to request resource allocation or to request an update to its configuration. The model or the identifier of the functionality. The identifier of the functionality associated with the model or the functionality, or The information used to achieve the aforementioned consistency.

11. The device of claim 9, wherein the processor is further configured such that the device: Upon receiving the request, a response to the request is sent to the other device, the response indicating confirmation information, the confirmation information including at least one of the following: An indication of whether the consistency has been maintained. The identifier of the reference mode used by the device. An indication of whether the mode used by the device is the same as the reference mode. The difference between the mode used by the device and the reference mode, or Adjusted beamforming information used for communication with the other device.

12. The device of claim 9, wherein the processor is further configured such that the device: If the device is a network device, after receiving the request, it sends a response to the request to the other device, the response indicating the updated resource configuration, or If the device is a terminal device, after receiving the request, it sends a response to the request to the other device, the response including a second indication for requesting resource allocation or requesting an update of the configuration of resources.

13. The device of claim 9, wherein the device is a network device providing a source cell, and the processor is further configured to cause the device to: Further information to determine the consistency of model-related or functionally relevant information between different ML stages used to achieve another model or another function used in candidate cells; Send the request to the other device to initiate the calibration process for the candidate cell; Receive a response from the other device to the request for the candidate cell; as well as Send the response to the candidate cell.

14. The device of claim 9, wherein the device is a terminal device, and the processor is further configured to cause the device to: Receive the request from the source cell or candidate cell to initiate the calibration process for the candidate cell; Based on additional information regarding the consistency of model-related or functional-related information between different ML stages used to achieve consistency of another model or another function used in the candidate cell, a response to the request for the candidate cell is sent to the source cell or the candidate cell.

15. The device of claim 9, wherein the device is a terminal device, and the processor is further configured such that the device: Send a request to the source cell or candidate cell to initiate the calibration process for the candidate cell; and Receive a response to the request for the candidate cell from the source cell or the candidate cell.

16. The apparatus of claim 9, wherein the apparatus is a network device providing a source cell, and the other apparatus is a terminal device, and the processor is further configured to cause the apparatus to: Receive a request from the other device to initiate the calibration process for the candidate cell; Send the request to the candidate cell to initiate the calibration process for the candidate cell; Receive a response to the request for the candidate cell from the candidate cell; as well as Send the response to the other device.

17. The device according to any one of claims 13 to 16, wherein the request for initiating the calibration process for the candidate cell and the response indication for the request for the candidate cell are associated with the identifier of the other information.

18. The device of claim 9, wherein the device is a terminal device, and the other device is a network device providing the source cell. And wherein the processor is further configured to cause the device to: Send capability-related information to the other device, the capability-related information indicating at least one of the following: Does the device support initiating the calibration process for the source cell or candidate cell, or Does the device support the other device initiating the calibration process? 19. The device of claim 9, wherein the processor is further configured such that the device: Receive configuration information from the other device, the configuration information indicating at least one of the following: At least one triggering event is used to initiate the calibration process. The periodicity used to initiate the calibration process The resource used by the device to send the request, or Resources used by the device to send a response to the request.

20. A communication method implemented at a device, the communication method comprising: Information determining the consistency of model-related or functionality-related information between a first machine learning (ML) stage used to implement the model or functionality and a second ML stage of the model or functionality; and Based on the information, perform at least one of the following: The calibration process with another device, The first ML stage, or The second ML stage.

Images