Devices and methods for communication
By incorporating model-related procedures like identification, transfer, training, and validation, the management of AI/ML models in communication networks is improved, addressing inefficiencies and reducing overhead in UE-side models.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-09-27
- Publication Date
- 2026-04-02
AI Technical Summary
Existing communication networks face challenges in managing AI/ML models due to the lack of model-related procedures during lifecycle management, particularly in complex scenarios involving UE-side models, leading to inefficiencies and increased overhead.
Implement model-related procedures such as model identification, transfer, training, validation, and performance monitoring to align network and user equipment (UE) parts, using standardized signaling and collaboration methods for UE-side models.
Enhances model management efficiency by ensuring proper alignment and reduction of overhead, particularly in two-sided model use cases like CSI compression, through coordinated model-related procedures.
Smart Images

Figure CN2024122052_02042026_PF_FP_ABST
Abstract
Description
DEVICES AND METHODS FOR COMMUNICATION
[0001] FIELDS
[0002] Example embodiments of the present disclosure generally relate to the field of communication techniques and in particular, to devices and methods for model management.BACKGROUND
[0003] As communication networks and services increase in size, complexity, and number of users, operations in the communication networks may become increasingly more complicated. In order to improve the communication performance, ML / artificial intelligence (AI) technology is proposed to be used in the wireless communication network. Lifecycle management (LCM) is one of the core parts for AI / ML related specification. LCM may include the following stages / phases / procedure: data collection, model training, model registration, model deployment, model configuration, model inference, model selection, model activation, model deactivation, model switching, and fallback operation, model monitoring, model update, model transfer, UE capability reporting and so on. In previous study, both functionality based LCM and model based LCM were discussed. Functionality based LCM may be the basis of LCM.SUMMARY
[0004] In general, embodiments of the present disclosure provide a solution for model management.
[0005] In a first aspect, there is provided a first device. The first device comprises: a processor configured to cause the first device to: transmit, to a second device, capability-related information comprising a set of functionalities supported by the first device; perform at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model; and determine, from the set of functionalities, at least one applicable functionality after completion of performing the at least one model-related procedure.
[0006] In a second aspect, there is provided a second device. The second device comprises: a processor configured to cause the second device to: receive, from a first device, capability-related information comprising a set of functionalities supported by the first device; and determine, from the set of functionalities, at least one applicable functionality after the first device performing at least one model-related procedure, the at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model.
[0007] In a third aspect, there is provided a communication method performed by a first device. The method comprises: transmitting, to a second device, capability-related information comprising a set of functionalities supported by the first device; performing at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model; and determining, from the set of functionalities, at least one applicable functionality after completion of performing the at least one model-related procedure.
[0008] In a fourth aspect, there is provided a communication method performed by a second device. The method comprises: receiving, from a first device, capability-related information comprising a set of functionalities supported by the first device; and determining, from the set of functionalities, at least one applicable functionality after the first device performing at least one model-related procedure, the at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model.
[0009] In a fifth aspect, there is provided a computer readable medium having instructions stored thereon, the instructions, when executed on at least one processor, causing the at least one processor to carry out the method according to the third, or fourth aspect.
[0010] Other features of the present disclosure will become easily comprehensible through the following description.BRIEF DESCRIPTION OF THE DRAWINGS
[0011] Through the more detailed description of some example embodiments of the present disclosure in the accompanying drawings, the above and other objects, features and advantages of the present disclosure will become more apparent, wherein:
[0012] FIG. 1A illustrates an example communication environment in which example embodiments of the present disclosure can be implemented;
[0013] FIG. 1B illustrates an example signaling flow for functionality based LCM;
[0014] FIG. 1C illustrates example blocks of inference procedure for different use cases;
[0015] FIG. 2 illustrates a signaling flow for communication in accordance with some embodiments of the present disclosure;
[0016] FIGS. 3A to FIG. 6B illustrate other signaling flows for communication in accordance with some embodiments of the present disclosure;
[0017] FIG. 7 illustrates a flowchart of a method implemented at a first device according to some example embodiments of the present disclosure;
[0018] FIG. 8 illustrates a flowchart of a method implemented at a second device according to some example embodiments of the present disclosure;
[0019] FIG. 9 illustrates a simplified block diagram of an apparatus that is suitable for implementing example embodiments of the present disclosure.
[0020] Throughout the drawings, the same or similar reference numerals represent the same or similar element.DETAILED DESCRIPTION
[0021] Principle of the present disclosure will now be described with reference to some example embodiments. It is to be understood that these embodiments are described only for the purpose of illustration and help those skilled in the art to understand and implement the present disclosure, without suggesting any limitation as to the scope of the disclosure. Embodiments described herein can be implemented in various manners other than the ones described below.
[0022] In the following description and claims, unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skills in the art to which this disclosure belongs.
[0023] As used herein, the term ‘terminal device’ refers to any device having wireless or wired communication capabilities. Examples of the terminal device include, but not limited to, user equipment (UE) , personal computers, desktops, mobile phones, cellular phones, smart phones, personal digital assistants (PDAs) , portable computers, tablets, wearable devices, internet of things (IoT) devices, Ultra-reliable and Low Latency Communications (URLLC) devices, Internet of Everything (IoE) devices, machine type communication (MTC) devices, devices on vehicle for V2X communication where X means pedestrian, vehicle, or infrastructure / network, devices for Integrated Access and Backhaul (IAB) , Space borne vehicles or Air borne vehicles in Non-terrestrial networks (NTN) including Satellites and High Altitude Platforms (HAPs) encompassing Unmanned Aircraft Systems (UAS) , eXtended Reality (XR) devices including different types of realities such as Augmented Reality (AR) , Mixed Reality (MR) and Virtual Reality (VR) , the unmanned aerial vehicle (UAV) commonly known as a drone which is an aircraft without any human pilot, devices on high speed train (HST) , or image capture devices such as digital cameras, sensors, gaming devices, music storage and playback appliances, or Internet appliances enabling wireless or wired Internet access and browsing and the like. The ‘terminal device’ can further have ‘multicast / broadcast’ feature, to support public safety and mission critical, V2X applications, transparent IPv4 / IPv6 multicast delivery, IPTV, smart TV, radio services, software delivery over wireless, group communications and IoT applications. It may also incorporate one or multiple Subscriber Identity Module (SIM) as known as Multi-SIM. The term “terminal device” can be used interchangeably with a UE, a mobile station, a subscriber station, a mobile terminal, a user terminal or a wireless device.
[0024] The term “network device” refers to a device which is capable of providing or hosting a cell or coverage where terminal devices can communicate. Examples of a network device include, but not limited to, a Node B (NodeB or NB) , an evolved NodeB (eNodeB or eNB) , a next generation NodeB (gNB) , a transmission reception point (TRP) , a remote radio unit (RRU) , a radio head (RH) , a remote radio head (RRH) , an IAB node, a low power node such as a femto node, a pico node, a reconfigurable intelligent surface (RIS) , and the like. Further, the network device also may be an operation administration and maintenance (OAM) , a server, location management function (LMF) , access and mobility management function (AMF) , and other core network node.
[0025] The terminal device or the network device may have Artificial intelligence (AI) or Machine learning capability. It generally includes a model which has been trained from numerous collected data for a specific function, and can be used to predict some information.
[0026] The terminal device or the network device may work on several frequency ranges, e.g., FR1 (e.g., 450 MHz to 6000 MHz) , FR2 (e.g., 24.25GHz to 52.6GHz) , frequency band larger than 100A GHz as well as Tera Hertz (THz) . It can further work on licensed / unlicensed / shared spectrum. The terminal device may have more than one connection with the network devices under Multi-Radio Dual Connectivity (MR-DC) application scenario. The terminal device or the network device can work on full duplex, flexible duplex and cross division duplex modes.
[0027] The embodiments of the present disclosure may be performed in test equipment, e.g., signal generator, signal analyzer, spectrum analyzer, network analyzer, test terminal device, test network device, channel emulator. In some embodiments, the terminal device may be connected with a first network device and a second network device. One of the first network device and the second network device may be a master node and the other one may be a secondary node. The first network device and the second network device may use different radio access technologies (RATs) . In some embodiments, the first network device may be a first RAT device and the second network device may be a second RAT device. In some embodiments, the first RAT device is eNB and the second RAT device is gNB. Information related with different RATs may be transmitted to the terminal device from at least one of the first network device or the second network device. In some embodiments, first information may be transmitted to the terminal device from the first network device and second information may be transmitted to the terminal device from the second network device directly or via the first network device. In some embodiments, information related with configuration for the terminal device configured by the second network device may be transmitted from the second network device via the first network device. Information related with reconfiguration for the terminal device configured by the second network device may be transmitted to the terminal device from the second network device directly or via the first network device.
[0028] 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 ‘includes’ and its variants are to be read as open terms that mean ‘includes, but is not limited to. ’ The term ‘based on’ is to be read as ‘at least in part based on. ’ The term ‘one embodiment’ and ‘an embodiment’ are to be read as ‘at least one embodiment. ’ The term ‘another embodiment’ is to be read as ‘at least one other embodiment. ’ The terms ‘first, ’ ‘second, ’ and the like may refer to different or same objects. Other definitions, explicit and implicit, may be included below.
[0029] In some examples, values, procedures, or apparatus are referred to as ‘best, ’ ‘lowest, ’ ‘highest, ’ ‘minimum, ’ ‘maximum, ’ or the like. It will be appreciated that such descriptions are intended to indicate that a selection among many used functional alternatives can be made, and such selections need not be better, smaller, higher, or otherwise preferable to other selections.
[0030] As used herein, the term “resource, ” “transmission resource, ” “uplink resource, ” or “downlink resource” may refer to any resource for performing a communication, such as a resource in time domain, a resource in frequency domain, a resource in space domain, a resource in code domain, or any other resource enabling a communication, and the like. In the following, unless explicitly stated, a resource in both frequency domain and time domain will be used as an example of a transmission resource for describing some example embodiments of the present disclosure. It is noted that example embodiments of the present disclosure are equally applicable to other resources in other domains.
[0031] As discussed above, in previous study, both functionality based LCM and model based LCM were discussed. Functionality based LCM may be the basis of LCM.
[0032] So far, radio access network 1 (RAN1) had some progresses on model identification mechanisms and model transfer / delivery mechanisms. In recent meeting, RAN2 agreed on procedures for functionality based LCM for UE-side model for beam management use case, which did not include any model-related procedures.
[0033] For two-sided model (such as, channel state information, CSI, compression) , if similar LCM procedure is applied, model-related procedure is a must for aligning NW part and UE part of two-sided model. For one sided model use case (such as, UE-side model for beam management) , on top of the agreed procedures for functionality based LCM, model-related procedure may also be needed to reduce the overhead and complexity.
[0034] Below table introduces different options for model delivery / transfer to UE, training location, and model delivery / transfer format combinations for UE-side models and UE-part of two-sided models:
[0035] Table. Model delivery / transfer cases
[0036] When a model of a known structure at UE (e.g., the above Case z4) is transferred from the network, the new model being identified (e.g., via Type B2) has the same structure as a previously identified model at the network and UE.
[0037] To facilitate the discussion, RAN1 studies the model identification type A (where Model is identified to NW (if applicable) and UE (if applicable) without over-the-air signalling) with more details related to use cases. Further, to facilitate the discussion, RAN1 studies the following options as starting point for model identification (MI) type B (where Model is identified via over-the-air signalling) with more details related to all use cases: MI-Option 1: Model identification with data collection related configuration (s) and / or indication (s) ; MI-Option 2: Model identification with dataset transfer; MI-Option 3: Model identification in model transfer from NW to UE; and MI-Option 4. Model identification via standardization of reference models (for CSI compression) ; and MI-Option 5. Model identification via performance monitoring.
[0038] Regarding MI-Option 1 (i.e., model identification with data collection related configuration (s) and / or indication (s) ) of model identification type B, RAN1 further study the following aspects: Relationship between model ID and data collection related configuration (s) and / or indication (s) ; Information transmitted from NW to UE (if any) ; Information transmitted from UE to NW (if any) ; The associated procedure; and Usage / Applicable use case (s) of MI-Option 1.
[0039] To alleviate / resolve the issues related to inter-vendor training collaboration of AI / ML-based CSI compression using two-sided model, study the following options: Option 1: Fully standardized reference model (structure + parameters) ; Option 2: Standardized dataset ( “Dataset” refers to a set of data samples of CSI feedback and associated target CSI) ; Option 3: Standardized reference model structure + Parameter exchange between NW-side and UE-side; Option 4: Standardized data / dataset format +Dataset exchange between NW-side and UE-side (it is clarified that Option 4 refers to the Option 1 of CSI compression) ; and Option 5: Standardized model format + Reference model exchange between NW-side and UE-side.
[0040] From RAN1 perspective, for UE-side model (s) developed (e.g., trained, updated) at UE side, following procedure is an example (noted as AI-Example1) of MI-Option1 for further study (including the feasibility / necessity) :
[0041] - A: For data collection, NW signals the data collection related configuration (s) and it / their associated ID (s) ; Associated IDs for each sub use case in relation with NW-sided additional conditions.
[0042] - B: UE (s) collects the data corresponding to the associated ID (s) .
[0043] - C: AI / ML models are developed (e.g., trained, updated) at UE side based on the collected data corresponding to the associated ID (s) .
[0044] - D: UE reports information of its AI / ML models corresponding to associated IDs to the NW. Model ID is determined / assigned for each AI / ML model.
[0045] Below alternatives may be used for determining model ID (s) : Alt. 1: NW assigns Model ID; Alt. 2: UE assigns / reports Model ID; Alt. 3: Associated ID (s) is assumed as model ID (s) (as for Alt. 3, “Model ID is determined / assigned for each AI / ML model” in D is not needed) ; Alt. 4: Model ID is determined by pre-defined rule (s) in the specification.
[0046] For Option 3, further define the two sub-options: 3a: Parameters received at the UE or UE-side goes through offline engineering at the UE-side (e.g., UE-side over-the- top (OTT) server) or a third party server, e.g., potential re-training, re-development of a different model, and / or offline testing. 3b: Parameters received at the UE are directly used for inference at the UE without offline engineering, potentially with on-device operations.
[0047] For Option 5, further define the two sub-options: 5a: Model received at the UE or UE-side goes through offline engineering at the UE-side (e.g., UE-side OTT server) , e.g., potential re-training, re-development of a different model, and / or offline testing. 5b: Model received at the UE are directly used for inference at the UE without offline engineering, potentially with on-device operations.
[0048] For Option 4, it is clarified that: Dataset received at the UE or UE-side goes through offline engineering at the UE-side (e.g., UE-side OTT server) , e.g., model training or offline testing. The descriptions under each option are only for the purpose of simplified discussion and do not mean deprioritizing any other Flavors (such as an exchange originating from the UE-side and ending at the NW-side) from potential specification.
[0049] As for Option 3a / 5a, focus further discussion on the following assumptions:
[0050] - The model (5a) / parameter (3a) exchange originates from the NW-side and ends at the UE-side.
[0051] - Model (5a) / parameters (3a) exchanged from the NW-side to UE-side is either CSI generation or reconstruction part or both.
[0052] - Option 3a-1 / 5a-1: Model / Parameters exchanged from the NW-side to UE-side is CSI generation part.
[0053] - Option 3a-2 / 5a-2: Model / Parameters exchanged from the NW-side to UE-side is CSI reconstruction part.
[0054] - Option 3a-3 / 5a-3: Model / Parameters exchanged from the NW-side to UE-side are both CSI generation part and CSI reconstruction part.
[0055] - Some additional information, if necessary, may be shared from the NW-side to help UE-side offline engineering and provide performance guidance, such as, Performance target, Dataset or information related to collecting dataset.
[0056] - Study different methods of exchanging, e.g., over the air-interface, offline delivery, and so on.
[0057] As for Option 3b, focus further discussion on the following assumptions: The method of exchanging is over the air-interface via model transfer / delivery Case z4. The parameter exchange is from NW to UE. Parameters exchanged from the NW-side to UE-side is CSI generation part.
[0058] As for Option 5b, focus further discussion on the following assumptions: The method of exchanging is over the air-interface via model transfer / delivery Case z4, assuming that the model structure is aligned based on offline inter-vendor collaboration. The model exchange is from NW to UE. Model exchanged from the NW-side to UE-side is CSI generation part.
[0059] As for Option 4, focus further discussion on the following assumptions: The dataset exchange originates from the NW-side and ends at the UE-side. Option 4-1: Dataset exchanged from the NW-side to UE-side consists of (target CSI, CSI feedback) . Option 4-2: Dataset exchanged from the NW-side to UE-side consists of (CSI feedback, reconstructed target CSI) . Option 4-3: Dataset exchanged from the NW-side to UE-side consists of (target CSI, CSI feedback, reconstructed target CSI) . Some additional information, if necessary, may be shared from the NW-side to help UE-side offline engineering and provide performance guidance, such as, performance target; Study different methods of exchanging, e.g., over the air-interface, offline delivery and so on.
[0060] Regarding the associated ID, the UE assumes that NW-side additional conditions with the same associated ID are consistent at least within a cell.
[0061] From RAN1 perspective, for UE part of two-sided model, further study the following example of MI-Option2 (AI-Example2-1, including the feasibility / necessity) : Step A: A dataset is transferred from the NW / NW-side to UE / UE-side via standardized signaling; Step B: B: UE part of two-sided model (s) is (are) developed based on at least the above dataset; Step C: UE reports information of its UE part of two-sided model (s) corresponding to the above dataset to the NW.
[0062] From RAN1 perspective, for model delivery / transfer Case z4, further study the following alternatives (including the necessity / feasibility / benefits) :
[0063] - Alt. A: Step A-1: UE reports the supported known model structure (s) to network; Step A-2: NW transfers to UE the parameters for one or more of supported known model structure (s) reported in Step A-1.
[0064] - Alt. B: Step B-0: UE reports to NW its support of model transfer / delivery case z4 (Step B-0 may be before or after Step B-1, or not necessary) ; Step B-1: NW indicates to UE the candidate known model structure (s) ; Step B-2: UE reports to NW which model structure (s) out of the candidate known model structure (s) indicated in Step B-1 is supported; Step B-3: NW transfers to UE the parameters for one or more of supported known model structure (s) reported in Step B-2.
[0065] From RAN1 perspective, the “known model structure (s) ” of the model transfer / delivery Case z4 at least include known information on the following aspects:
[0066] - Model type / backbone (e.g., Transformer, convolution neural networks (CNN) , and so on);
[0067] - In case model type is a neural network, Number of layers, Layer types / structure (e.g., full connected, activation layer and so on) , Layer size (e.g., the number of parameters of a layer) , Connection between different layers;
[0068] - model input / output related information.
[0069] As used herein, an AI / ML model may be equivalent to at least one of the following: a model, an ML model, an AI model, a data-driven, a data processing model, an algorithm, a functionality, a procedure, a process, an entity, a function, a feature, a feature group, a model ID, a functionality ID, a configuration ID, a scenario ID, a site ID, or a dataset ID. As a result, the above terms may be used interchangeably.
[0070] In some embodiments, the AI / ML model may comprise a set of weights values that may be learned during training, for example for a specific architecture or configuration, where a set of weights values may also be called a parameter set.
[0071] In some embodiments, the AI / ML model may be used to predict a target cell, or measurements of a set of beams of a set of candidate cells in future based on at least historical measurements (e.g., layer 1 (L1) -reference signal receiving power (RSRP) , L1-signal to interference plus noise ratio (SINR) ) of a set of beams of a set of candidate cells.
[0072] In some embodiments, an input of the AI / ML model (i.e., AI input) may refer to the input of a model and indicate data inputted into the model, which may be equivalent to data, input data or input data item.
[0073] In some embodiments, an output of AI / ML model (i.e., AI output) may refers to the output of a model and indicate result (s) outputted by the model, which is equivalent to label, data, output data, or label data.
[0074] In some embodiments, for UE-part / UE-side models, in case of functionality-based LCM procedure, indication of activation / deactivation / switching / fallback may be implemented based on individual AI / ML functionality.
[0075] In some embodiments, for UE-part / UE-side models, in case of model-ID-based LCM procedure, indication of model selection / activation / deactivation / switching / fallback may be implemented based on individual model IDs.
[0076] In some embodiments, the UE may have one AI / ML model for the functionality, Alternatively, in some embodiments, the UE may have multiple AI / ML models for the functionality.
[0077] In some embodiments, for UE-side models and UE-part of two-sided models: 1) for AI / ML functionality identification legacy 3GPP framework of features may be reused, UE may indicate the supported functionalities / functionality for a given sub-use-case, and the UE capability reporting may be taken as starting point; 2) for AI / ML model identification, models may be identified by model ID at the Network and UE may indicate the supported AI / ML models; 3) in functionality-based LCM, network may indicate activation / deactivation / fallback / switching of AI / ML functionality via a signaling (e.g., a radio resource control (RRC) signalling, downlink control information (DCI) ) ; models may not be identified at the Network, and UE may perform model-level LCM; 4) in model-ID-based LCM, models are identified at the network, and network / UE may activate / deactivate / select / switch individual AI / ML models via model ID.
[0078] In some embodiments, for functionality identification, there may be either one or more than one functionality defined within an AI / ML-enabled feature.
[0079] In some embodiments, for AI / ML model identification and model-ID-based LCM of UE-side models and / or UE-part of two-sided models, the model-ID-based LCM may operate based on identified models, where a model may be associated with specific configurations / conditions associated with UE capability of an AI / ML-enabled feature / feature group and additional conditions (e.g., scenarios, sites, and datasets) as determined / identified between UE-side and network-side.
[0080] In some embodiments, for model identification of UE-side or UE-part of two- sided models, model identification types may be categorized as follows:
[0081] Type A: Model is identified to network (if applicable) and UE (if applicable) without over-the-air signaling, where the model may be assigned with a model ID during the model identification, which may be referred / used in over-the-air signaling after model identification;
[0082] Type B: Model is identified via over-the-air signaling,
[0083] ■ Type B1: model identification initiated by the UE, and network assists the remaining steps (if any) of the model identification and the model may be assigned with a model ID during the model identification;
[0084] ■ Type B2: model identification initiated by the network, and UE responds (if applicable) for the remaining steps (if any) of the model identification, and the model may be assigned with a model ID during the model identification.
[0085] Further, for better understanding, some terminologies related with ML model are described in below table.
[0086] Table descriptions of terminology
[0087] Further, for better descriptions, some terms used herein are listed as below:
[0088] Conditions: referred to as configurations supported indicated via UE capability reporting (e.g., component field) , related to model training, model inference, performance monitoring, validation procedure, fallback, of an AI / ML model / functionality or a group of AI / ML models / functionalities.
[0089] Additional conditions (e.g., scenarios, sites, and datasets) : including but not limited to, application conditions, scenarios, datasets, cell ID, timestamp and SNR, and so on; For an AI / ML-enabled feature / FG, additional conditions refer to any aspects that are assumed for the training of the model but are not a part of UE capability for the AI / ML-enabled feature / FG. It does not imply that additional conditions are necessarily specified. Additional conditions can be divided into two categories: NW-side additional conditions and UE-side additional conditions. Note: whether specification impact is needed is a separate discussion.
[0090] Applicable condition: signal to noise ratio (SNR) , line of sight (LOS) / Non-line of sight (NLOS) , channel condition, and so on.
[0091] UE internal conditions: including but not limited to, memory, battery, computation resource, overheating and other hardware limitations on functionality / model operations.
[0092] Physical AI / ML model (s) : referred to as an actual implementation of such a model.
[0093] A logical AI / ML model: referred to as a model that is identified and assigned a model ID.
[0094] Complexity / processing capability: TOPs, Floating point operations (FLOPs) , MACs, the number of parameters and / or size.
[0095] Abeam: downlink beam, uplink beam, transmit beam, receive beam, beam pair, reference signal (RS) resource, RS resource set, antenna port, antenna port group, antenna element (s) , antenna array (s) , beam group.
[0096] Functionality: in case of AI / ML functionality identification and functionality-based LCM of UE-side models and / or UE-part of two-sided models, functionality is referred to an AI / ML-enabled Feature / feature group enabled by configuration (s) , where configuration (s) is (are) supported based on conditions indicated by UE capability. Correspondingly, functionality-based LCM may operate based on, at least, one configuration of AI / ML-enabled Feature / feature group, or specific configurations of an AI / ML-enabled Feature / Feature group.
[0097] Amodel inference configuration: a configuration to report information related to the predicted results (e.g., predicted beam, predicted CSI, predicted position or label) , or the compressed results (e.g., compressed CSI, compressed PMI) , and / or a configuration of resources for collect data used for prediction or compression.
[0098] Amode performance monitoring configuration: a configuration to report information related to the performance metrics (e.g., accuracy, confidence, data distribution, squared generalized cosine similarity (SGCS) , ground truth, difference from ground truth, etc. ) , and / or a configuration of resources for collect data used for calculating the performance metrics.
[0099] Amodel training configuration, a configuration to collect data used as model input and model output, and / or a configuration to enable / support dataset transfer from NW to UE.
[0100] In the present disclosure,
[0101] Terms of “ML model” , “AI model” , “ML function” , “AI function” and “algorithm” may be used interchangeably.
[0102] Terms of “model” “functionality” and “model / functionality” may be used interchangeably.
[0103] Terms of “model” , “model set” and “model group” may be used interchangeably.
[0104] Terms of “functionality” , “functionality group” , “functionality set” may be used interchangeably.
[0105] Terms of “ID” , “index” , “indicator” and “identifier” may be used interchangeably.
[0106] Terms of “known / unknown” , “applicable / non-applicable” or “suitable / non-suitable” may be used interchangeably.
[0107] Terms of “precoding information and number of layers” , “precoding matrix indicator (PMI) ” , “precoding matrix indicator” , “transmission precoding matrix indication” , “precoding matrix indication” , “transmission configuration indication state (TCI state) ” , “UL TCI state” , “joint TCI state” , “transmission configuration indicator” , “quasi co-location (QCL) ” , “quasi-co-location” , “QCL parameter” , “QCL assumption” , “QCL relationship” and “spatial relation” may be used interchangeably. Terms of “RSRP” , “L1-RSRP” “L3-RSRP” , “filtered RSRP” may be used interchangeably. If “RSRP” is used as a beam quality metric, the methods is readily extended to other metrics like “SINR” , “RSRQ” , “RSSI” and so on.
[0108] Terms of “model development” , “model re-development” , “model training” , “model re-training” , “model fine-tuning” and “model update” may be used interchangeably.
[0109] Principles and implementations of the present disclosure will be described in detail below with reference to the figures.
[0110] Example Environment
[0111] FIG. 1A illustrates a schematic diagram of an example communication environment 100A in which example embodiments of the present disclosure can be implemented. In the communication environment 100A, a plurality of communication devices, including a first device 110 and a second device 120, can communicate with each other.
[0112] In the example of FIG. 1A, in some embodiments, the first device 110 may include a terminal device and the second device 120 may include a network device serving the terminal device. It should be understood that, in the other embodiments, the first device 110 / second device 120 may be any of: a terminal device, a network device, an OTT (server) , an operation administration and maintenance (OAM) (server) , an edge cloud (server) , a neutral site, transmission reception point (TRP) core network and so on. In present disclosure is not limited in this regard.
[0113] In a case that the first device 110 is a terminal device and the second device 120 is a network device, a link from the first device 110 to the second device 120 is referred to as uplink, while a link from the second device 120 to the first device 110 is referred to as a downlink.
[0114] In downlink, the second device 120 is a transmitting (TX) device (or a transmitter) and the first device 110 is a receiving (RX) device (or a receiver) , and the second device 120 may transmit downlink transmission to the first device 110. Correspondingly, in uplink, the second device 120 is an RX device (or a receiver) and the first device 110 is a TX device (or a transmitter) , and the first device 110 may transmit uplink transmission to the second device 120.
[0115] Further, in FIG. 1A, one or more ML models may be deployed at the first device 110 and / or the second device 120. Reference is now made to FIG. 1B, which illustrates an example signaling flow 100B for functionality based LCM.
[0116] In the example of FIG. 1B, supported functionalities refer to functionalities that UE can indicate by using UE capability information (via RRC / location positioning protocol (LPP) signalling) , applicable functionalities refers to functionalities that the UE is ready to apply for inference, and activated functionalities refers to functionalities already enabled for performing inference.
[0117] At Step 1: Network sends UECapabilityEnqiry message to initiate the procedure to a UE reporting its AI / ML supported functionalities. At Step 2: UE sends UECapablityInformation message to network, containing supported functionalities at the UE side. At “Step 3” : Following configurations are provided from NW to UE: 1) UE is allowed to do UE assistance information (UAI) reporting via other configuration; 2) Network may provide NW-side additional condition. FFS on the RRC signalling and whether it is mandatory or optional; 3) configuration (e.g., inference configuration) of supported functionalities.
[0118] UE decides the applicable functionalities based on NW-side additional conditions (if provided) , UE-side additional conditions (internally known by UE) and model availability in device.
[0119] At Step 4: UE reports applicable functionality in the following scenarios: 1) Upon being configured to provide applicable functionality and upon change of applicable functionality via UAI; 2) As response to NW-side additional condition requesting applicable functionality reporting in step 3.
[0120] At Step 5: 1) Network configures inference configuration to UE after applicable functionality reporting, if inference configuration based on supported functionality is not provided in Step 3 (i.e., inference configuration is provided in Step 5) ; 2) If inference configuration based on supported functionality is provided in Step 3, it is up to network implementation whether to provide an updated configuration or not.
[0121] The applicable functionality may be activated by receiving its inference configuration when it is provided in Step 5.
[0122] In some embodiments, the granularity of functionality may be a use case (e.g., AI / ML based beam management, CSI prediction, CSI compression, positioning, etc. ) , a sub-use case (e.g., spatial domain beam prediction, temporal domain beam prediction, etc. ) , a model or a specific input / output (e.g., spatial domain beam prediction with 4 beams in Set B and 16 beams in Set A, spatial domain beam prediction with 8 beams in Set B and 64 beams in Set A and so on) .
[0123] For example, assuming that the granularity of functionality is a sub-use case, capability may also comprise signalling of detailed information such as Set A and Set B, observation / prediction window and so on, as in the following table.
[0124] For the evaluation of the AI / ML based CSI compression sub use cases, a two-sided model may be considered as a starting point, including an AI / ML-based CSI generation part to generate the CSI feedback information and an AI / ML-based CSI reconstruction part which is used to reconstruct the CSI from the received CSI feedback information. At least for inference, the CSI generation part is located at the UE side, and the CSI reconstruction part is located at the gNB side. Some example use cases of the model (s) deployed at the first device 110 and the second device 120 will be discussed with reference to FIG. 1C.
[0125] Example (A) in FIG. 1C provides an example for the inference procedure for CSI compression. For generating the input of CSI generation model, it may need some further pre-processing on the measured channel; for the output of the CSI reconstruction model, some further post-processing may also be applied. Besides CSI feedback of quantization output, there may also be other CSI / PMI related information transmitted. There may be other examples of merging quantization / dequantization into the inference for CSI generation / reconstruction, CSI generation model / CSI reconstruction model, respectively.
[0126] Example (B) in FIG. 1C provides an example for the inference procedure for CSI prediction. For generating the input of CSI prediction model, it may need some further pre-processing on the measured channel; for the output of the CSI prediction model, some further post-processing may also be applied.
[0127] Example (C) in FIG. 1C provides an example for the inference procedure for beam management for BM-Case1 and BM-Case2. Measurements based on Set B of beams are used as model input. In addition, beam ID information may be also provided as input to the AI / ML model. Based on model output (e.g., probability of each beam in Set A to be the Top-1 beam, predicted L1-RSRPs) , Top-1 / N beam (s) among Set A of beams can be predicted and / or potentially with predicted L1-RSRPs (depending on the labeling) . In the evaluation, for BM-Case 1, the measurements of Set B (otherwise stated) are used as model input to predict Top-1 / N beams from Set A, and for BM-Case2, the measurements from historic time instance (s) are used as model input for temporal DL beam prediction of beams from Set A. In the evaluation, the cases that Set A and Set B are different (Set B is NOT a subset of Set A) , and Set B is a subset of Set A for both BM-Case1 and BM-Case2, and case that Set A and Set B are the same for BM-Case2 are considered. And the performance of DL Tx beam prediction and DL Tx-Rx beam pair prediction is evaluated.
[0128] For both BM-Case1 and BM-Case2, UE can report the prediction result to NW based on the output of a UE-side model, or NW can predict the Top-1 / N beam (s) based on the reported measurements of Set B for a NW-side model.
[0129] It is to be understood that the number of devices and their connections shown in FIG. 1A are only for the purpose of illustration without suggesting any limitation. The communication environment 100A may include any suitable number of devices configured to implementing example embodiments of the present disclosure.
[0130] In some embodiments, the first device 110 and the second device 120 may communicate with each other via a channel such as a wireless communication channel on an air interface (e.g., Uu interface) or a PC5 interface. The wireless communication channel may comprise a sidelink, 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 channels are also feasible.
[0131] The communications in the communication environment 100A may conform to any suitable standards including, but not limited to, Global System for Mobile Communications (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) and the like. The embodiments of the present disclosure may be performed according to any generation communication protocols either currently known or to be developed in the future. Examples of the communication protocols include, but not limited to, the first generation (1G) , the second generation (2G) , 2.5G, 2.75G, the third generation (3G) , the fourth generation (4G) , 4.5G, the fifth generation (5G) communication protocols, 5.5G, 5G-Advanced networks, or the sixth generation (6G) networks.
[0132] Example Process
[0133] In some case, UE needs to decide the applicable functionalities based on model availability. However, during the functionality based LCM procedure, there is no steps for model-related procedures, such as, model identification and / or model transfer / delivery and / or model training and / or model validation / performance monitoring.
[0134] In this event, UE may not have available model and therefore cannot decide the applicable functionalities before model training completed / before model transfer completed. In some cases, UE may not have available model due to the existing model being not validated or out-of-date, or cannot being used (immediately) due to other reasons. In some other cases, UE may not have available model due to the existing model failing to meet the performance requirement. In some further cases, UE may not have available model due to some internal conditions being met / not met (such as lacking power, memory, computational resources and so on) . This problem is more severe for two-sided model use cases such as CSI compression, because model identification and model transfer / delivery are considered more useful for those use cases.
[0135] According to some example embodiments of the present disclosures, the mode-related procedure may be considered.
[0136] Reference is made to FIG. 2, which illustrates a signaling flow 200 for communication in accordance with some embodiments of the present disclosure. For the purposes of discussion, the signaling flow 200 will be discussed with reference to FIG. 1A and FIG. 1B, for example, by using the first device 110 and the second device 120.
[0137] In the following descriptions, while operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are contained in the above discussions, these should not be construed as limitations on the scope of the present disclosure, but rather as descriptions of features that may be specific to particular embodiments. Certain features that are described in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment may also be implemented in multiple embodiments separately or in any suitable sub-combination.
[0138] It is to be understood that the operations at the first device 110 and the second device 120 should be coordinated. In other words, the second device 120 and the first device 110 should have common understanding about configurations, parameters and so on. Such common understanding may be implemented by any suitable interactions between the second device 120 and the first device 110 or both the second device 120 and the first device 110 applying the same rule / policy. In the following, although some operations are described from a perspective of the first device 110, it is to be understood that the corresponding operations should be performed by the second device 120. Similarly, although some operations are described from a perspective of the second device 120, it is to be understood that the corresponding operations should be performed by the first device 110. Merely for brevity, some of the same or similar contents are omitted here.
[0139] For the purpose of discussion, the first device 110 may be a terminal device and the second device 120 may be a network device.
[0140] In operation, the first device 110 transmits (210-1) capability-related information to the second device 120, and the second device 120 receives (210-2) the capability-related information accordingly, where the capability-related information comprises a set of functionalities supported by the first device 110.
[0141] In some embodiments, the capability-related information may further comprise an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality and / or an indication of the at least one model-related procedure needed for determining the at least one applicable functionality.
[0142] Alternatively, or in addition, in some embodiments, the capability-related information may further comprise information about at least one functionality, wherein a functionality of the at least one functionality is applicable on a condition of the completion of the model-related procedure.
[0143] Alternatively, or in addition, in some embodiments, the capability-related information may further comprise: an indication of whether a model training is needed for a functionality supported by the first device 110, an indication of whether a model identification is needed for a functionality supported by the first device 110, an indication of whether an associated identity is needed, an indication of whether a model structure alignment procedure is needed, or an indication of whether a dataset transfer is needed, an indication of whether a model transfer is needed for a functionality supported by the first device 110, an indication of whether a validation is needed for a functionality supported by the first device 110, or an indication of whether a performance monitoring is needed for a functionality supported by the first device 110.
[0144] It is noted that a functionality supported by the first device 110 may refer to at least one of the following: a functionality supported by the first device 110, to be supported by the first device 110, a functionality to be applicable at the first device 110, a functionality to be activated at the first device 110.
[0145] It is to be understood that the above discussed capability-related information are given for illustrative purpose only. It should be understood that any suitable capability-related information associated with the embodiments discussed herein may be communicated between the first device 110 and the second device 120. The present disclosure is not limited in this regard.
[0146] It should be understood that the capability-related information may include UE capability (ies) and / or a capability (ies) of the AI / ML model, such as supported input and output, processing time, computation complexity. Further, the capability-related information may be comprised in one or more messages. For one example, the capability-related information may be comprised on a single message, such as, as initial capability reporting. In another example, part of the capability-related information may be reported as initial capability reporting and the remaining capability-related information may be comprised in following message (s) on demand.
[0147] The first device 110 perform (220) at least one model-related procedure which comprises at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model and other model-related procedures. After completion of performing the at least one model-related procedure, the first device 110 determines (230-1) at least one applicable functionality from the set of functionalities, and the second device 120 also determines the at least one configuration for the at least one applicable functionality correspondingly, such as, an inference configuration, a performance monitoring configuration and / or other configurations.
[0148] According to the present disclosure, the at least one model-related procedure may be performed at any suitable phase of the LCM. Such examples will be discussed with reference to FIG. 3A and FIG. 4B.
[0149] Reference is now made to FIG. 3A. As illustrated, the first device 110 may receive (305) first configuration information comprising at least one network additional condition from the second device 120, and the at least one model-related procedure may be performed (310) after receiving the first configuration information.
[0150] In the example of FIG. 3A, the first configuration information may further comprise at least one configuration comprising at least one of the following:
[0151] - a configuration used by the first device 110 to trigger the at least one model-related procedure. The configuration may be an UL resource configuration, UL resource request configuration, trigger condition configuration, time-related configuration and so on,
[0152] - at least one model inference configuration associated with the set of functionalities supported by the first device 110, or
[0153] - at least one performance monitoring configuration associated with the set of functionalities supported by the first device 110.
[0154] In some embodiments, the at least one model inference configuration may be considered invalid or deactivated before the completion of model-related procedure.
[0155] More details about the process of FIG. 3A are discussed. At Step 1: Network sends UECapabilityEnqiry message to initiate the procedure to a UE reporting its AI / ML supported functionalities. At Step 2: UE sends UECapablityInformation message to network, containing supported functionalities at the UE side. Further, UE may report whether / which model-related procedure is needed for supported functionality, e.g., to determine applicable functionality.
[0156] Then, following configurations may be provided (305) from NW to UE: UE is allowed to do UAI reporting via OtherConfig; Network may provide NW-side additional condition; Network may provide configuration (e.g., inference configuration) of supported functionalities. In some embodiments, the model inference configuration may be considered not valid / activated before the completion of model-related procedure. For example, reference signal may be (assumed) not transmitted in the configured resources, UE does not perform measurements and report based on the configured resources, UE does not need to perform rate matching around those resources and so on.
[0157] Further, NW may provide associated ID. For more than one functionality, NW may provide more than one NW-side additional condition and / or more than one associated ID. It is not limited that NW-side additional condition is functionality specific. In other words, NW-side additional condition may be the same or different for different functionalities.
[0158] Using the Set A / Set B association as an example, if the granularity of functionality is a sub-use case, the Set A / Set B association may be the same or different for BM-Case 1 and BM-Case 2. In addition, if associated ID is configured, it may be configured independently for each inference configuration for example, for BM-Case1 and BM-Case2, respectively. In some other examples, the associated ID may be configured jointly for both inference configurations for BM-Case1 and BM-Case2.
[0159] When the NW-side additional condition is not provided, the default assumption may be that UE assumes that the consistency between model training and model inference is ensured by the NW.
[0160] In some other examples, when the NW-side additional condition is not provided, the default assumption should be that UE cannot assume that the consistency between model training and model inference is ensured by the NW.
[0161] In some embodiments, to solve the potential performance issue, performance validation or performance monitoring needs to be configured.
[0162] In some embodiments, network may provide performance monitoring related configurations, such as target performance, measurement and report resources and so on.
[0163] UE may perform (310) model-related procedure needed for UE to determine the applicable functionality.
[0164] At Step 4: UE may report the applicable functionality (ies) in the following scenarios:
[0165] - Upon the completion of model-related procedure;
[0166] - Upon being configured to provide applicable functionality and upon change of applicable functionality via UAI;
[0167] - As response to NW-side additional condition requesting applicable functionality reporting, and / or other network configuration (e.g., inference configuration) in step 3, via signalling like UAI or via RRCReconfigurationComplete.
[0168] In some embodiments, UE may report model-related information together with the applicable functionality reporting, such as, Model ID, associated ID, NW part ID of the model, UE part ID of the model, associated model inference configuration, associated performance monitoring configuration and so on.
[0169] At Step 5: Network may configure or update the inference configuration to UE after the applicable functionality reporting. The applicable functionality may be activated by receiving its inference configuration when it is provided in Step 5.
[0170] In some embodiments, at Step 5, NW may also provide / update model training and performance monitoring configurations.
[0171] In some embodiments, an initial activation state of the applicable functionality may be activated or deactivated. In some embodiments, an initial state of the model inference (also model training, performance monitoring) configuration in the first configuration information which may be activated or deactivated.
[0172] In some embodiments, if the initial state of the applicable functionality and / or the related inference configuration are deactivated, additional L1 / L2 signaling (including but not limited to DCI, MAC CE or other suitable L1 / L2 signaling) may be used for activation / deactivation of the applicable functionality and / or the related inference configuration.
[0173] In some embodiments, one or multiple applicable functionalities may be activated at the same time, which may be based on UE capability or UAI report.
[0174] A Step 6: model inference and / or performance monitoring may be performed accordingly.
[0175] In some embodiments, the functionality may be activated immediately or after X time units upon receiving inference configuration in step 5. In some embodiments, the functionality is activated immediately or after X time units upon receiving L1 / L2 signaling for activation in step 5.
[0176] X time units can be NW configured, based UE reported capability, or pre-defined.
[0177] In the example of FIG. 3A, it is more reasonable that the first device 110 initiates to perform the model-related procedure. Reference is now made to FIG. 3B. As illustrated in FIG. 3B, in some embodiments, during the model-related procedure 310, the first device 110 may transmit (360) a message used for informing the second device 120 starting of performing the at least one model-related procedure (for example, inform the second device 120 that the at least one model-related procedure is to be performed or the at least one model-related procedure has been started / performed) .
[0178] In some embodiments, the message used for informing the starting of performing the at least one model-related procedure comprises at least one of the following:
[0179] - an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or
[0180] - an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, or
[0181] - an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure,
[0182] - an indication of whether a model training is needed for a functionality supported by the first device 110,
[0183] - an indication of whether a model identification is needed for a functionality supported by the first device 110,
[0184] - an indication of whether an associated identity is needed,
[0185] - an indication of whether a model structure alignment procedure is needed,
[0186] - an indication of whether a dataset transfer is needed,
[0187] - an indication of whether a model transfer is needed for a functionality supported by the first device,
[0188] - an indication of whether a validation is needed for a functionality supported by the first device, or
[0189] - an indication of whether a performance monitoring is needed for a functionality supported by the first device.
[0190] The first device 110 may perform (365) the at least one model-related procedure and then may transmit (270) a message used for informing the second device 120 the completion of performing the at least one model-related procedure.
[0191] More details about the process of FIG. 3B will be discussed in the following. As illustrated in FIG. 3B, if the model is not currently available at UE, to determine applicable functionality reporting, additional steps are needed, which may be done by one or more of the following mechanisms, for example, before applicable functionality reporting, for example, UE may trigger model-related procedure.
[0192] In some embodiments, RRC reconfigurations (in action 305) may provide configurations for UE to trigger model-related procedure are needed, such that UE is enabled to initiate the related procedure via UAI reporting, dynamic UE capability reporting, or other dedicated signalling.
[0193] In some embodiments, model transfer / model training / model identification / validation (i.e., different model-related procedures) may require independent and different procedures, but in general, the following steps may be needed.
[0194] In operation, UE may report (360) at least one candidate applicable functionality: this functionality is applicable conditioning on the model-related procedure completed. Then, model transfer / model training / model identification / validation may be performed.
[0195] Next, the completion information may be provided (370) from UE to NW. In some embodiments, the applicable functionality determination may be made upon the completion of model-related procedure. In some examples, the applicable functionality (ies) may be reported at action 370.
[0196] At Step 5, the applicable functionality (ies) is reported. Alternatively, applicable functionality reporting (in step 5) may be not needed, if NW configuration may determine the one or more applicable functionalities, e.g., different NW configuration is related to different applicable functionalities.
[0197] Other examples will be discussed with reference to FIG. 4A and 4B. Reference is now made to FIG. 4A. As illustrated, the first device 110 may transmit (405) a first message to the second device 120, then the first device 110 may perform (410) the at least one model-related procedure after transmitting the first message.
[0198] In some embodiments, the first message may comprise at least one of the following:
[0199] - an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or
[0200] - an indication of the at least one model-related procedure needed for determining the at least one applicable functionality,
[0201] - an indication of there is no applicable functionality, or
[0202] - an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure.
[0203] In some embodiments, the first device 110 may transmit the first message to the second device 120 based on a determination that:
[0204] - there is no available model,
[0205] - at least one of the following comprised in first configuration information transmitted by the second device 120 is not supported by the first device 110: at least one additional condition, a model inference configuration, or a performance monitoring configuration, or
[0206] - an internal condition of the first device 110 causes no available model or no applicable functionality.
[0207] More details about the process of FIG. 4A are discussed. In the example of FIG. 4A, the model-related procedure may also be performed after the applicable functionality reporting (e.g., action 405) . In this case, the report content in action 405 may be ‘no applicable functionality’ due to model unavailability issue, or ‘at least one candidate applicable functionality’ which may be conditioning on the completion of model-related procedures.
[0208] In some embodiments, UE may report (405) ‘no applicable functionality’ in the following scenarios:
[0209] - No model available,
[0210] - No NW side additional condition matched as provide in step 3 (if provided) ,
[0211] - Not support of the inference configuration in step 3 (if provided) ,
[0212] - Performance is not satisfactory (if NW provides target performance in step 3) ,
[0213] - Others: Existing model is not validated, is out-of-date, and so on. Some internal condition met / not met, such as lack of power, memory, computational resources.
[0214] At Step 5: after model-related procedure, since model is available at UE side and also identified at NW side, Network may configure or update inference configuration to UE after applicable functionality reporting. The applicable functionality associated with the identified / transferred / trained model may be activated by receiving its inference configuration. At Step 6: model inference and / or performance monitoring may be performed accordingly.
[0215] In the example of FIG. 4A, it is more reasonable that the second device 120 initiates to perform the model-related procedure. Reference is now made to FIG. 4B. As illustrated in FIG. 4B, in some embodiments, the first device 110 receive (455) a message used for triggering the first device 110 to start performing the at least one model-related procedure from the second device 120.
[0216] In some embodiments, the message used for triggering the first device 110 to start performing the at least one model-related procedure comprises: configuration information used by the fist device to perform the at least one model-related procedure.
[0217] The first device 110 may perform (460) the at least one model-related procedure in response to receiving the message, and then transmit (465) a message for informing the second device 120 the completion of performing the at least one model-related procedure.
[0218] More details about the process of FIG. 4B are discussed. In the example of FIG. 4B, during the applicable functionality reporting (e.g., step 4 in FIG. 4B) , UE may report that there is no applicable functionality, due to no available model. Further, at Step 4, UE may report at least one candidate applicable functionality: this functionality is applicable conditioning on the model-related procedure completed. In some examples, plural candidate functionalities may be reported.
[0219] In some embodiments, NW may trigger model-related procedure. Although model transfer / model training / model identification may require independent and different procedures, in general, the following steps may be needed: triggering information is provided (455) from NW to UE; model transfer / model training / model identification is performed (460) ; the completion information may be transmitted (465) from UE to NW. In some examples, additional applicable functionality reporting may be carried with the completion information.
[0220] The applicable functionality determination may be made upon the completion of model-related procedure. Further, the additional applicable functionality report may be not needed because the previously reported candidate applicable functionality has become an applicable functionality.
[0221] As discussed above, the at least one model-related procedure may comprise at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model and other model-related procedures. In the following, example where the at least one model-related procedure comprises the model transfer will be discussed. In such examples, the applicable functionality determination may be made upon the completion of model transfer. Further, model structure alignment may be needed before model transfer, and UE may report supported functionality and whether model transfer is needed for the functionality.
[0222] Below example embodiments about model transfer are at least suitable for two-sided model, and further may be applied to UE-side model use cases, if model transfer is needed and supported with signalling for UE-side model use case.
[0223] Since two-sided model requires both NW part and UE part, and NW and UE may be from different vendors, inter-vendor collaboration may be needed. To do inter-vendor collaboration, below multiple options may be possible: Option 1: Fully standardized reference model (structure + parameters) ; Option 2: Standardized dataset; Option 3: Standardized reference model structure + Parameter exchange between NW-side and UE-side; Option 4: Standardized data / dataset format + Dataset exchange between NW-side and UE-side; Option 5: Standardized model format + Reference model exchange between NW-side and UE-side.
[0224] Although below discussions may be focused on Option 3, similar methods also may be applied for Option 5. In Option 5, the full model is transferred, instead of parameter exchange, therefore may be with higher overhead.
[0225] It is not limited whether the transferred model / parameter is CSI generation part and / or CSI reconstruction part. In some examples, NW and UE need to align whether CSI generation part, CSI reconstruction part, or CSI generation part and CSI reconstruction part are to be transmitted.
[0226] In addition, the model identification may be supported, as in MI-Option 3: Model identification in model transfer from NW to UE, which suggests additional steps to identify the model transferred.
[0227] Some parameters related to AI / ML model in below table may be useful for transferring a model, and they can be considered as additional parameters.
[0228] In some embodiments, during the model transfer, the first device 110 may receive a set parameters comprising at least one parameter of a reference model structure from the second device 120, and then may apply the set of parameters to obtain a valid mode.
[0229] In some embodiments, the set of parameters further comprises at least one of the following: a model identity, a network side model identity of a two-sided model, a user equipment (UE) side model identity of a two-sided model.
[0230] In some embodiments, a model structure alignment may be needed. Specifically, during the model-related procedure, the first device 110 may transmit information of a reference model structure to the second device 120 or receive the information of the reference model structure from the second device 120, wherein the information of the reference model structure indicates at least one of the following: an index of model structure, a model type, a number of layers of a model, a required minimum number of layers of a model, a supported maximum number of layers of a model, a layer structure type of a model, a number of parameters of a layer, a required minimum number of parameters of a layer, a supported maximum number of parameters of a layer, a connection type between different layers, input / output related information of a model, quantization information of a model, a set of training related parameters of a model, or a set of generalization related parameters of a model.
[0231] More examples will be discussed with reference to FIG. 5A and FIG. 5B. Referenced is made to FIG. 5A. At Step 2 of UE capability reporting, UE may report supported functionality and whether model transfer is needed for the functionality.
[0232] In some embodiments, during the model-related procedure 500, UE may request (510) model transfer from NW to UE, together with the following information: whether model structure alignment is needed; the candidate applicable functionality upon the completion of model transfer; NW may respond (not shown in the figure) to the request with configurations for model transfer, e.g., model structure alignment.
[0233] In some embodiments, the model structure alignment may be performed (522) . In some embodiments, UE may report its support model structure, which may be indexing information of predefined model structures, or details of one or many or the following information to determine a model structure:
[0234] - Model type / backbone, e.g., Transformer, CNN and so on;
[0235] - Number of layers, minimum required / maximum supported number of layers;
[0236] - Layer types / structure, e.g., full connected layer, activation layer, input layer, output layer, normalization layer, pooling layer, quantization layer, recurrent layer, residual layer, convolutional layer, attention layer, transposed convolution Layer, specialized layer and so on. In some embodiments, it may be indicated for each of layers, or indicated for group of layers. In some embodiments, layer index may be needed;
[0237] - Layer size (e.g., the number of parameters of a layer) , including minimum required / maximum supported number of parameters of a layer;
[0238] - Connection between different layers, e.g., feedforward connection, recurrent, skip connection, residual connection, full connected, locally connected, sparse connection, weighted connection or shared weight, convolutional connection, attention, gated connection and so on.
[0239] - Model input / output related information: inputs / output size, format, fixed point representation and so on. For CSI compression use case, input dimensions, support antenna configuration, output dimensions, number of compressed bits and so on. For beam management case, set A and set B information, observation window, prediction window and so on. For CSI prediction use case: antenna ports, precoders, observation window, prediction window and so on.
[0240] - Quantization: scalar, vector and so on.
[0241] Alternatively, NW may provide model structure information and UE may feedback the supported model structure. To save overhead, UE feedback may be some simple information such as supported or not, or model structure ID.
[0242] In some embodiments, for a known model structure, NW may send (524) parameters of the model structure to the UE. Optionally, in some embodiments, Model ID, NW part ID of the model, UE part ID of the model may be signalled to the UE.
[0243] In some embodiments, UE applies (526) the parameters to have a valid model. In addition, parameters received at the UE or UE-side goes through offline engineering at the UE-side (e.g., UE-side OTT server) , e.g., potential re-training, re-development of a different model, and / or offline testing.
[0244] In some embodiments, UE reports (528) information of its AI / ML models. In some embodiments, Model ID may be determined / assigned for each AI / ML model.
[0245] The applicable functionality determination may be made upon the completion of model transfer. The completion information may be transmitted (530) from UE to NW.
[0246] In some embodiments, the model transfer procedure may need to be completed in a reasonable time duration. Further, different model size may correspond to different time length, and the relationship between model size and required time length may be based on NW configuration, UE report (UE capability or UAI) or pre-defined relationship.
[0247] In some embodiments, model size may be related to model structure, e.g., one or many of number of layers, number of parameters of a layer, connection between different layers and so on. In some embodiments, the duration may be depending on the number of applicable functionalities, if multiple applicable functionalities.
[0248] Reference is now made to FIG. 5B. As illustrated, at Step 4, UE may report ‘no applicable functionality’ in the following scenarios:
[0249] - No model available,
[0250] - No NW side additional condition matched as provide in step 3 (if provided) ,
[0251] - Not support of the inference configuration in step 3 (if provided) ,
[0252] - Performance is not satisfactory (if NW provides target performance in step 3) ,
[0253] - Others: Existing model is not validated, is out-of-date and so on. Some internal condition met / not met, such as lack of power, memory, computational resources.
[0254] In some embodiments, during the model-related procedure 550, the NW may trigger (560) the model transfer. Further, NW may provide configuration for UE to report whether model structure alignment is needed. In some embodiments, the target applicable functionality may be determined upon the completion of model transfer.
[0255] In some embodiments, NW may also signal the model ID, and / or corresponding associated ID, dataset ID and so on to the UE. In some embodiments, during the model transfer procedure 550, the model structure alignment may be performed (570) .
[0256] In some embodiments, UE may be configured to reports its support model structure (as discussed with FIG. 5A) . Similarly, NW also may provide model structure information and UE may feedback the support model structure.
[0257] In some embodiments, for a known model structure, NW may send (574) parameters of the model structure to the UE. Optionally, in some embodiments, Model ID, NW part ID of the model, UE part ID of the model may be signalled to the UE.
[0258] In some embodiments, UE applies (576) the parameters to have a valid model. In addition, parameters received at the UE or UE-side goes through offline engineering at the UE-side (e.g., UE-side OTT server) , e.g., potential re-training, re-development of a different model, and / or offline testing.
[0259] In some embodiments, UE reports (578) information of its AI / ML models. In some embodiments, Model ID is determined / assigned for each AI / ML model.
[0260] Applicable functionality determination may be made upon the completion of model transfer. The completion information may be transmitted (580) from UE to NW.
[0261] In the following, example where the at least one model-related procedure comprises the model identification will be discussed. In such embodiments, applicable functionality determination may be made upon the completion of model identification, and UE may report supported functionality and whether model identification is needed for the functionality.
[0262] In some embodiments, below embodiments are suitable for both one sided model and two-sided model. To do model identification, multiple options can be possible:
[0263] -MI-Option 1: Model identification with data collection related configuration (s) and / or indication (s) ,
[0264] - MI-Option 2: Model identification with dataset transfer,
[0265] - MI-Option 3: Model identification in model transfer from NW to UE,
[0266] - MI-Option 4: Model identification via standardization of reference models.
[0267] - MI-Option 5: Model identification via performance monitoring.
[0268] Although below discussions may focus on MI-Option 1 and MI-Option 2, other MI-Options are also applicable. In some embodiments, model identification options above may involve model training based on collected or transferred data.
[0269] In some embodiments, during the UE capability reporting, UE may report supported functionality and whether model identification is needed for the functionality. For example, UE may report at least one of the following,
[0270] - whether associated ID is needed for the supported functionality, or for model identification of associated functionality, if associated ID is needed, whether UE already has associated ID,
[0271] - whether dataset transfer is needed for the supported functionality, or for model identification of associated functionality; If dataset ID is needed, whether UE already has dataset ID or the number (size) of required dataset. In addition, whether the associated ID is required for the corresponding dataset.
[0272] In some embodiments, during the model-related procedure, the first device 110 may receive at least one associated identity from the second device 120, where each associated identity may correspond to a configuration for data collection or model training. The first device 110 may perform data collection or model training based on at least one configuration corresponding to the at least one associated identity. The first device 110 may transmit, to the second device 120, a message comprising model-related information corresponding to the at least one associated identity upon completion of performing the data collection or model training.
[0273] In some embodiments, the model-related information may comprise at least one model identity, each model identity corresponding to a functionality.
[0274] More examples will be discussed with reference to FIG. 6A and FIG. 6B. In the example of FIG. 6A, during the mode-related procedure 550, UE may request (560) to initiate model identification, together with the information such as:
[0275] - Whether model training is needed for the supported functionality, or for model identification of associated functionality;
[0276] - Whether associated ID is needed for the supported functionality, or for model identification of associated functionality; If associated ID is needed, whether UE already has associated ID;
[0277] - Whether dataset transfer is needed for the supported functionality, or for model identification of associated functionality; If dataset transfer ID is needed, whether UE already has dataset ID or the number (size) of required dataset; In addition, whether the associated ID is required for the corresponding dataset;
[0278] - The candidate applicable functionality upon the completion of model identification.
[0279] In some embodiments, for data collection, NW signals (622) the data collection related configuration (s) and / or it / their associated ID (s) . In some embodiments, associated IDs for each sub use case may be in relation with NW-sided additional conditions. In some embodiments, UE may send the information about whether it needs the information related to NW-side additional condition in the followed steps. Alternatively, in some embodiments, NW transmits the dataset to UE. In some examples, NW also signals associated ID or dataset ID accordingly.
[0280] In some embodiments, UE (s) collects (624) the data corresponding to the associated ID (s) and AI / ML models (or UE part of two-sided models) are developed (e.g., trained, updated) at UE side based on the collected data corresponding to the associated ID(s) . Alternatively, in some embodiments, UE may perform model training based on received dataset, in some examples, associated ID or dataset ID may be used to categorize the data for training.
[0281] In some embodiments, UE may report (626) information of its AI / ML models corresponding to associated IDs to the NW. In some embodiments, model ID may be determined / assigned for each applicable functionality. The completion information may be transmitted (630) from UE to NW.
[0282] Reference is now made to FIG. 6B. In the example of FIG. 6B, at Step 4, the UE reports ‘no applicable functionality’ in the following scenarios:
[0283] - No model available,
[0284] - No NW side additional condition matched as provide in step 3 (if provided) ,
[0285] - Not support of the inference configuration in step 3 (if provided) ,
[0286] - Performance is not satisfactory (if NW provides target performance in step 3) ,
[0287] - Others: Existing model is not validated, is out-of-date and so on. Some internal condition met / not met, such as lack of power, memory, computational resources.
[0288] In some embodiments, during the model-related procedure 650, NW may trigger (660) model identification. The target applicable functionality upon the completion of model training / model identification. In some embodiments, the NW may also signal (672) the model ID and / or corresponding associated ID, dataset ID to the second device 120.
[0289] In some embodiments, for data collection, NW signals (672) the data collection related configuration (s) and / or it / their associated ID (s) during model training / model identification 670. In some embodiments, associated IDs for each sub use case are in relation with NW-sided additional conditions. In some embodiments, UE can send the information about whether it needs the Information related to NW-side additional condition in the followed steps. Alternatively, NW transmit the dataset to UE, in some examples, NW also signals associated ID or dataset ID accordingly.
[0290] In some embodiments, UE (s) collects (674) the data corresponding to the associated ID (s) and AI / ML models (or UE part of two-sided model (s) ) are developed (e.g., trained, updated) at UE side based on the collected data corresponding to the associated ID (s) .
[0291] In some embodiments, UE reports (676) information of its AI / ML models corresponding to associated IDs to the NW. Model ID is determined / assigned for each applicable functionality. The completion information may be transmitted (680) from UE to NW.
[0292] The above discussions are mainly about how to determine the at least one applicable functionality. In the following, example processes after the determination of the at least one applicable functionality will be discussed with by continuing referring to FIG. 2.
[0293] As illustrated in FIG. 2, in some embodiments, after the completion of the model-related procedure, the first device 110 may transmit (240-1) a message indicating the at least one applicable functionality and especially indicating model-related information to a second device 120, and the second device 120 may receive (240-2) the message accordingly.
[0294] In some embodiments, the second device 120 may transmit (250-1) second configuration information to the first device 110, and the first device 110 may receive (250-2) the second configuration information accordingly. In some embodiments, the second configuration information may be associated with the at least one model-related procedure.
[0295] In particular, in some embodiments, the second configuration information may indicate at least one of the following:
[0296] - a model inference configuration,
[0297] - a performance monitoring configuration,
[0298] - a configuration of the performance monitoring,
[0299] - an initial activation state of the at least one applicable functionality, or
[0300] - an initial activation state of a previously-received model-related configuration comprising at least one of the following: a model inference configuration, a performance monitoring configuration, a model training configuration.
[0301] In some embodiments, the at least one model-related procedure may be performed within a reasonable duration. Further, the time length of the duration may be configured by the second device 120, reported by the first device 110, defined as a default value (such as, pre-defined in the 3GPP specification) , or associated with AI / ML model.
[0302] Further, the control of time can be realized by setting timer (s) , defining time window (s) for different procedure (s) respectively.
[0303] In some embodiments, in a case that the at least one model-related procedure comprises the model transfer, the duration may be further associated with at least one of the following: a mode size, a model structure, a model type, or the number of applicable functionalities.
[0304] Example Method
[0305] FIG. 7 illustrates a flowchart of a communication method 700 implemented at a first device in accordance with some embodiments of the present disclosure. For the purpose of discussion, the method 700 will be described from the perspective of the first device 110 in FIG. 1A.
[0306] At block 710, the first device transmits, to a second device, capability-related information comprising a set of functionalities supported by the first device.
[0307] At block 720, the first device performs at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model.
[0308] At block 730, the first device determines, from the set of functionalities, at least one applicable functionality after completion of performing the at least one model-related procedure.
[0309] In some example embodiments, the capability-related information further comprises at least one of the following: information about at least one functionality, wherein a functionality of the at least one functionality is applicable on a condition of the completion of the model-related procedure, an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of whether a model training is needed for a functionality supported by the first device, an indication of whether a model identification is needed for a functionality supported by the first device, an indication of whether an associated identity is needed, an indication of whether a model structure alignment procedure is needed, an indication of whether a dataset transfer is needed, an indication of whether a model transfer is needed for a functionality supported by the first device 110, an indication of whether a validation is needed for a functionality supported by the first device 110, or an indication of whether a performance monitoring is needed for a functionality supported by the first device 110.
[0310] In some example embodiments, the first device may receive, from the second device, first configuration information comprising at least one network additional condition, wherein the at least one model-related procedure is performed after receiving the first configuration information.
[0311] In some example embodiments, the first configuration information further comprises at least one configuration comprising at least one of the following: a configuration used by the first device to trigger the at least one model-related procedure, at least one model inference configuration associated with the set of functionalities supported by the first device, or at least one performance monitoring configuration associated with the set of functionalities supported by the first device.
[0312] In some example embodiments, the at least one model inference configuration is considered invalid or deactivated before the completion of model-related procedure.
[0313] In some example embodiments, the first device may transmit, a message used for informing the second device starting of performing the at least one model-related procedure; perform the at least one model-related procedure; and transmit, a message used for informing the second device the completion of performing the at least one model-related procedure.
[0314] In some example embodiments, the message used for informing the starting of performing the at least one model-related procedure comprises at least one of the following: an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure, an indication of whether a model training is needed for a functionality supported by the first device, an indication of whether a model identification is needed for a functionality supported by the first device, an indication of whether an associated identity is needed, an indication of whether a model structure alignment procedure is needed, an indication of whether a dataset transfer is needed, an indication of whether a model transfer is needed for a functionality supported by the first device 110, an indication of whether a validation is needed for a functionality supported by the first device 110, or an indication of whether a performance monitoring is needed for a functionality supported by the first device 110.
[0315] In some example embodiments, the first device may transmit, to the second device, a first message comprising at least one of the following: an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of there is no applicable functionality, or an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure; and perform the at least one model-related procedure after transmitting the first message.
[0316] In some example embodiments, the first device may transmit the first message to the second device based on a determination that: there is no available model, at least one of the following comprised in first configuration information transmitted by the second device is not supported by the first device: at least one additional condition, a model inference configuration, or a performance monitoring configuration, or an internal condition of the first device causes no available model or no applicable functionality.
[0317] In some example embodiments, the first device may receive, from the second device, a message used for triggering the first device to start performing the at least one model-related procedure; perform the at least one model-related procedure in response to receiving the message; and transmit, a message for informing the second device the completion of performing the at least one model-related procedure.
[0318] In some example embodiments, the message used for triggering the first device to start performing the at least one model-related procedure comprises: configuration information used by the fist device to perform the at least one model-related procedure.
[0319] In some example embodiments, after the completion of the model-related procedure, the first device may transmit, to a second device, a message indicating the at least one applicable functionality and model-related information.
[0320] In some example embodiments, the first device may receive, from the second device, second configuration information indicating at least one of the following: a model inference configuration, a performance monitoring configuration, a configuration of the performance monitoring an initial activation state of the at least one applicable functionality, or an initial activation state of a previously-received model-related configuration comprising at least one of the following: a model inference configuration, a performance monitoring configuration, a model training configuration.
[0321] In some example embodiments, the second configuration information is associated with the at least one model-related procedure.
[0322] In some example embodiments, the at least one model-related procedure is performed within a duration, wherein time length of the duration is configured by the second device, reported by the first device or defined as a default value.
[0323] In some example embodiments, the at least one model-related procedure comprises the model transfer, and the duration is associated with at least one of the following: a mode size, a model structure, a model type, or the number of applicable functionalities.
[0324] In some example embodiments, the at least one model-related procedure comprises the model transfer and the first device is further caused to: receive, from the second device, a set parameters comprising at least one parameter of a reference model structure; and apply the set of parameters to obtain a valid mode.
[0325] In some example embodiments, the set of parameters further comprises at least one of the following: a model identity, a network side model identity of a two-sided model, a user equipment (UE) side model identity of a two-sided model.
[0326] In some example embodiments, the at least one model-related procedure comprises the model transfer and the first device is further caused to: transmit information of a reference model structure to the second device or receive the information of the reference model structure from the second device, wherein the information of the reference model structure indicates at least one of the following: an index of model structure, a model type, a number of layers of a model, a required minimum number of layers of a model, a supported maximum number of layers of a model, a layer structure type of a model, a number of parameters of a layer, a required minimum number of parameters of a layer, a supported maximum number of parameters of a layer, a connection type between different layers, input / output related information of a model, quantization information of a model, a set of training related parameters of a model, or a set of generalization related parameters of a model.
[0327] In some example embodiments, the at least one model-related procedure comprises the model identification and the first device is further caused to: receive, from the second device, at least one associated identity, each associated identity corresponding to a configuration for data collection or model training; perform, based on at least one configuration corresponding to the at least one associated identity, data collection or model training; and transmit, to the second device, a message comprising model-related information corresponding to the at least one associated identity upon completion of performing the data collection or model training.
[0328] In some example embodiments, the model-related information comprises at least one model identity, each model identity corresponding to a functionality.
[0329] In some example embodiments, the first device is a terminal device and the second device is a network device.
[0330] FIG. 8 illustrates a flowchart of a communication method 800 implemented at a second device in accordance with some embodiments of the present disclosure. For the purpose of discussion, the method 800 will be described from the perspective of the second device 120 in FIG. 1A.
[0331] At block 810, the second device receives, from a first device, capability-related information comprising a set of functionalities supported by the first device.
[0332] At block 820, the second device determines, from the set of functionalities, at least one applicable functionality after the first device performing at least one model-related procedure, the at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model.
[0333] In some example embodiments, the capability-related information further comprises at least one of the following: information about at least one functionality, wherein a functionality of the at least one functionality is applicable on a condition of the completion of the model-related procedure, an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of whether a model training is needed for a functionality supported by the first device, an indication of whether a model identification is needed for a functionality supported by the first device, an indication of whether an associated identity is needed, an indication of whether a model structure alignment procedure is needed, an indication of whether a dataset transfer is needed, an indication of whether a model transfer is needed for a functionality supported by the first device 110, an indication of whether a validation is needed for a functionality supported by the first device 110, or an indication of whether a performance monitoring is needed for a functionality supported by the first device 110.
[0334] In some example embodiments, the second device may transmit, to the first device, first configuration information comprising at least one network additional condition, wherein the at least one model-related procedure is performed after receiving the first configuration information.
[0335] In some example embodiments, the first configuration information further comprises at least one configuration comprising at least one of the following: a configuration used by the first device to trigger the at least one model-related procedure, at least one model inference configuration associated with the set of functionalities supported by the first device, or at least one performance monitoring configuration associated with the set of functionalities supported by the first device.
[0336] In some example embodiments, the at least one model inference configuration is considered invalid or deactivated before the completion of model-related procedure.
[0337] In some example embodiments, the second device is further caused to: receive, from the first device, a message used for informing the second device starting of performing the at least one model-related procedure; and receive, from the first device, a message used for informing the second device the completion of performing the at least one model-related procedure.
[0338] In some example embodiments, the message used for informing the starting of performing the at least one model-related procedure comprises at least one of the following: an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure, an indication of whether a model training is needed for a functionality supported by the first device, an indication of whether a model identification is needed for a functionality supported by the first device, an indication of whether an associated identity is needed, an indication of whether a model structure alignment procedure is needed, an indication of whether a dataset transfer is needed, an indication of whether a model transfer is needed for a functionality supported by the first device 110, an indication of whether a validation is needed for a functionality supported by the first device 110, or an indication of whether a performance monitoring is needed for a functionality supported by the first device 110.
[0339] In some example embodiments, the second device may receive, from the first device, a first message comprising at least one of the following: an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of there is no applicable functionality, or an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure.
[0340] In some example embodiments, the second device may transmit, to the first device, a message used for triggering the first device to start performing the at least one model-related procedure; and receive, from the first device, a message for informing the second device the completion of performing the at least one model-related procedure.
[0341] In some example embodiments, the message used for triggering the first device to start performing the at least one model-related procedure comprises: configuration information used by the fist device to perform the at least one model-related procedure.
[0342] In some example embodiments, the second device is further caused to: receive, from the first device, a message indicating the at least one applicable functionality and model-related information.
[0343] In some example embodiments, the second device may transmit, to the first device, second configuration information indicating at least one of the following: a model inference configuration, a performance monitoring configuration, a configuration of the performance monitoring an initial activation state of the at least one applicable functionality, or an initial activation state of a previously-received model-related configuration comprising at least one of the following: a model inference configuration, a performance monitoring configuration, a model training configuration.
[0344] In some example embodiments, the second configuration information is associated with the at least one model-related procedure.
[0345] In some example embodiments, the at least one model-related procedure is performed within a duration, wherein time length of the duration is configured by the second device, reported by the first device or defined as a default value.
[0346] In some example embodiments, the at least one model-related procedure comprises the model transfer, and the duration is associated with at least one of the following: a mode size, a model structure, a model type, or the number of applicable functionalities.
[0347] In some example embodiments, the at least one model-related procedure comprises the model transfer and the second device is further caused to: transmit, to the first device, a set parameters comprising at least one parameter of a reference model structure; and apply the set of parameters to obtain a valid mode.
[0348] In some example embodiments, the set of parameters further comprises at least one of the following: a model identity, a network side model identity of a two-sided model, a user equipment (UE) side model identity of a two-sided model.
[0349] In some example embodiments, the at least one model-related procedure comprises the model transfer and the second device is further caused to: transmit information of a reference model structure to the second device or receive the information of the reference model structure from the second device, wherein the information of the reference model structure indicates at least one of the following: an index of model structure, a model type, a number of layers of a model, a required minimum number of layers of a model, a supported maximum number of layers of a model, a layer structure type of a model, a number of parameters of a layer, a required minimum number of parameters of a layer, a supported maximum number of parameters of a layer, a connection type between different layers, input / output related information of a model, quantization information of a model, a set of training related parameters of a model, or a set of generalization related parameters of a model.
[0350] In some example embodiments, the at least one model-related procedure comprises the model identification and the second device is further caused to: transmit, to the first device, at least one associated identity, each associated identity corresponding to a configuration for data collection or model training; perform, based on at least one configuration corresponding to the at least one associated identity, data collection or model training; and receive , form the first device, a message comprising model-related information corresponding to the at least one associated identity upon completion of performing the data collection or model training.
[0351] In some example embodiments, the model-related information comprises at least one model identity, each model identity corresponding to a functionality.
[0352] In some example embodiments, the first device is a terminal device and the second device is a network device.
[0353] Example device and apparatus
[0354] FIG. 9 is a simplified block diagram of a device 900 that is suitable for implementing embodiments of the present disclosure. The device 900 can be considered as a further example implementation of any of the devices as shown in FIG. 1A. Accordingly, the device 900 can be implemented at or as at least a part of the terminal device 110 or the network device 120.
[0355] As shown, the device 900 includes a processor 910, a memory 920 coupled to the processor 910, a suitable transceiver 940 coupled to the processor 910, and a communication interface coupled to the transceiver 940. The memory 920 stores at least a part of a program 930. The transceiver 940 may be for bidirectional communications or a unidirectional communication based on requirements. The transceiver 940 may include at least one of a transmitter 942 and a receiver 944. The transmitter 942 and the receiver 944 may be functional modules or physical entities. The transceiver 940 has at least one antenna to facilitate communication, though in practice an Access Node mentioned in this application may have several ones. The communication interface may represent any interface that is necessary for communication with other network elements, such as X2 / Xn interface for bidirectional communications between eNBs / gNBs, S1 / NG interface for communication between a Mobility Management Entity (MME) / Access and Mobility Management Function (AMF) / SGW / UPF and the eNB / gNB, Un interface for communication between the eNB / gNB and a relay node (RN) , or Uu interface for communication between the eNB / gNB and a terminal device.
[0356] The program 930 is assumed to include program instructions that, when executed by the associated processor 910, enable the device 900 to operate in accordance with the embodiments of the present disclosure, as discussed herein with reference to FIGS. 2 to 8. The embodiments herein may be implemented by computer software executable by the processor 910 of the device 900, or by hardware, or by a combination of software and hardware. The processor 910 may be configured to implement various embodiments of the present disclosure. Furthermore, a combination of the processor 910 and memory 920 may form processing means 950 adapted to implement various embodiments of the present disclosure.
[0357] The memory 920 may be of any type suitable to the local technical network and may be implemented using any suitable data storage technology, such as a non-transitory computer readable storage medium, semiconductor-based memory devices, magnetic memory devices and systems, optical memory devices and systems, fixed memory and removable memory, as non-limiting examples. While only one memory 920 is shown in the device 900, there may be several physically distinct memory modules in the device 900. The processor 910 may be of any type suitable to the local technical network, and may include one or more of general purpose computers, special purpose computers, microprocessors, digital signal processors (DSPs) and processors based on multicore processor architecture, as non-limiting examples. The device 900 may have multiple processors, such as an application specific integrated circuit chip that is slaved in time to a clock which synchronizes the main processor.
[0358] According to embodiments of the present disclosure, a first device comprising a circuitry is provided. The circuitry is configured to: transmit, to a second device, capability-related information comprising a set of functionalities supported by the first device; perform at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model; and determine, from the set of functionalities, at least one applicable functionality after completion of performing the at least one model-related procedure. According to embodiments of the present disclosure, the circuitry may be configured to perform any method implemented by the first device as discussed above.
[0359] According to embodiments of the present disclosure, a second device comprising a circuitry is provided. The circuitry is configured to: receive, from a first device, capability-related information comprising a set of functionalities supported by the first device; and determine, from the set of functionalities, at least one applicable functionality after the first device performing at least one model-related procedure, the at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model. According to embodiments of the present disclosure, the circuitry may be configured to perform any method implemented by the second device as discussed above.
[0360] The term “circuitry” used herein may refer to hardware circuits and / or combinations of hardware circuits and software. For example, the circuitry may be a combination of analog and / or digital hardware circuits with software / firmware. As a further example, the circuitry may be any portions of hardware processors with software including digital signal processor (s) , software, and memory (ies) that work together to cause an apparatus, such as a terminal device or a network device, to perform various functions. In a still further example, the circuitry may be hardware circuits and or processors, such as a microprocessor or a portion of a microprocessor, that requires software / firmware for operation, but the software may not be present when it is not needed for operation. As used herein, the term circuitry also covers an implementation of merely a hardware circuit or processor (s) or a portion of a hardware circuit or processor (s) and its (or their) accompanying software and / or firmware.
[0361] According to embodiments of the present disclosure, a first apparatus is provided. The first apparatus comprises means for transmitting, to a second device, capability-related information comprising a set of functionalities supported by the first device; means for performing at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model; and means for determining, from the set of functionalities, at least one applicable functionality after completion of performing the at least one model-related procedure. In some embodiments, the first apparatus may comprise means for performing the respective operations of the method 700. In some example embodiments, the first apparatus may further comprise means for performing other operations in some example embodiments of the method 700. The means may be implemented in any suitable form. For example, the means may be implemented in a circuitry or software module.
[0362] According to embodiments of the present disclosure, a second apparatus is provided. The second apparatus comprises means for receiving, from a first device, capability-related information comprising a set of functionalities supported by the first device; and means for determining, from the set of functionalities, at least one applicable functionality after the first device performing at least one model-related procedure, the at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model. In some embodiments, the second apparatus may comprise means for performing the respective operations of the method 800. In some example embodiments, the second apparatus may further comprise means for performing other operations in some example embodiments of the method 800. The means may be implemented in any suitable form. For example, the means may be implemented in a circuitry or software module.
[0363] In summary, embodiments of the present disclosure provide the following aspects.
[0364] In an aspect, it is proposed a first device comprising: a processor configured to cause the first device to: transmit, to a second device, capability-related information comprising a set of functionalities supported by the first device; perform at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model; and determine, from the set of functionalities, at least one applicable functionality after completion of performing the at least one model-related procedure.
[0365] In some embodiments, the capability-related information further comprises at least one of the following: information about at least one functionality, wherein a functionality of the at least one functionality is applicable on a condition of the completion of the model-related procedure, an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of whether a model training is needed for a functionality supported by the first device, an indication of whether a model identification is needed for a functionality supported by the first device, an indication of whether an associated identity is needed, an indication of whether a model structure alignment procedure is needed, an indication of whether a dataset transfer is needed, an indication of whether a model transfer is needed for a functionality supported by the first device 110, an indication of whether a validation is needed for a functionality supported by the first device 110, or an indication of whether a performance monitoring is needed for a functionality supported by the first device 110.
[0366] In some embodiments, the first device is further caused to: receive, from the second device, first configuration information comprising at least one network additional condition, wherein the at least one model-related procedure is performed after receiving the first configuration information.
[0367] In some embodiments, the first configuration information further comprises at least one configuration comprising at least one of the following: a configuration used by the first device to trigger the at least one model-related procedure, at least one model inference configuration associated with the set of functionalities supported by the first device, or at least one performance monitoring configuration associated with the set of functionalities supported by the first device.
[0368] In some embodiments, the at least one model inference configuration is considered invalid or deactivated before the completion of model-related procedure.
[0369] In some embodiments, the first device is further caused to: transmit, a message used for informing the second device starting of performing the at least one model-related procedure; perform the at least one model-related procedure; and transmit, a message used for informing the second device the completion of performing the at least one model-related procedure.
[0370] In some embodiments, the message used for informing the starting of performing the at least one model-related procedure comprises at least one of the following: an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure, an indication of whether a model training is needed for a functionality supported by the first device, an indication of whether a model identification is needed for a functionality supported by the first device, an indication of whether an associated identity is needed, an indication of whether a model structure alignment procedure is needed, an indication of whether a dataset transfer is needed, an indication of whether a model transfer is needed for a functionality supported by the first device, an indication of whether a validation is needed for a functionality supported by the first device, or an indication of whether a performance monitoring is needed for a functionality supported by the first device.
[0371] In some embodiments, the first device is further caused to: transmit, to the second device, a first message comprising at least one of the following: an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of there is no applicable functionality, or an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure; and perform the at least one model-related procedure after transmitting the first message.
[0372] In some embodiments, the first device is further caused to: transmit the first message to the second device based on a determination that: there is no available model, at least one of the following comprised in first configuration information transmitted by the second device is not supported by the first device: at least one additional condition, a model inference configuration, or a performance monitoring configuration, or an internal condition of the first device causes no available model or no applicable functionality.
[0373] In some embodiments, the first device is further caused to: receive, from the second device, a message used for triggering the first device to start performing the at least one model-related procedure; perform the at least one model-related procedure in response to receiving the message; and transmit, a message for informing the second device the completion of performing the at least one model-related procedure.
[0374] In some embodiments, the message used for triggering the first device to start performing the at least one model-related procedure comprises: configuration information used by the fist device to perform the at least one model-related procedure.
[0375] In some embodiments, the first device is further caused to: after the completion of the model-related procedure, transmit, to a second device, a message indicating the at least one applicable functionality and model-related information.
[0376] In some embodiments, the first device is further caused to: receive, from the second device, second configuration information indicating at least one of the following: a model inference configuration, a performance monitoring configuration, a configuration of the performance monitoring an initial activation state of the at least one applicable functionality, or an initial activation state of a previously-received model-related configuration comprising at least one of the following: a model inference configuration, a performance monitoring configuration, a model training configuration.
[0377] In some embodiments, the second configuration information is associated with the at least one model-related procedure.
[0378] In some embodiments, the at least one model-related procedure is performed within a duration, wherein time length of the duration is configured by the second device, reported by the first device or defined as a default value.
[0379] In some embodiments, the at least one model-related procedure comprises the model transfer, and the duration is associated with at least one of the following: a mode size, a model structure, a model type, or the number of applicable functionalities.
[0380] In some embodiments, the at least one model-related procedure comprises the model transfer and the first device is further caused to: receive, from the second device, a set parameters comprising at least one parameter of a reference model structure; and apply the set of parameters to obtain a valid mode.
[0381] In some embodiments, the set of parameters further comprises at least one of the following: a model identity, a network side model identity of a two-sided model, a user equipment (UE) side model identity of a two-sided model.
[0382] In some embodiments, the at least one model-related procedure comprises the model transfer and the first device is further caused to: transmit information of a reference model structure to the second device or receive the information of the reference model structure from the second device, wherein the information of the reference model structure indicates at least one of the following: an index of model structure, a model type, a number of layers of a model, a required minimum number of layers of a model, a supported maximum number of layers of a model, a layer structure type of a model, a number of parameters of a layer, a required minimum number of parameters of a layer, a supported maximum number of parameters of a layer, a connection type between different layers, input / output related information of a model, quantization information of a model, a set of training related parameters of a model, or a set of generalization related parameters of a model.
[0383] In some embodiments, the at least one model-related procedure comprises the model identification and the first device is further caused to: receive, from the second device, at least one associated identity, each associated identity corresponding to a configuration for data collection or model training; perform, based on at least one configuration corresponding to the at least one associated identity, data collection or model training; and transmit, to the second device, a message comprising model-related information corresponding to the at least one associated identity upon completion of performing the data collection or model training.
[0384] In some embodiments, the model-related information comprises at least one model identity, each model identity corresponding to a functionality.
[0385] In some embodiments, the first device is a terminal device and the second device is a network device.
[0386] In an aspect, it is proposed a second device comprising: a processor configured to cause the second device to: receive, from a first device, capability-related information comprising a set of functionalities supported by the first device; and determine, from the set of functionalities, at least one applicable functionality after the first device performing at least one model-related procedure, the at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model.
[0387] In some embodiments, the capability-related information further comprises at least one of the following: information about at least one functionality, wherein a functionality of the at least one functionality is applicable on a condition of the completion of the model-related procedure, an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of whether a model training is needed for a functionality supported by the first device, an indication of whether a model identification is needed for a functionality supported by the first device, an indication of whether an associated identity is needed, an indication of whether a model structure alignment procedure is needed, an indication of whether a dataset transfer is needed, an indication of whether a model transfer is needed for a functionality supported by the first device, an indication of whether a validation is needed for a functionality supported by the first device, or an indication of whether a performance monitoring is needed for a functionality supported by the first device.
[0388] In some embodiments, the second device is further caused to: transmit, to the first device, first configuration information comprising at least one network additional condition, wherein the at least one model-related procedure is performed after receiving the first configuration information.
[0389] In some embodiments, the first configuration information further comprises at least one configuration comprising at least one of the following: a configuration used by the first device to trigger the at least one model-related procedure, at least one model inference configuration associated with the set of functionalities supported by the first device, or at least one performance monitoring configuration associated with the set of functionalities supported by the first device.
[0390] In some embodiments, the at least one model inference configuration is considered invalid or deactivated before the completion of model-related procedure.
[0391] In some embodiments, the second device is further caused to: receive, from the first device, a message used for informing the second device starting of performing the at least one model-related procedure; and receive, from the first device, a message used for informing the second device the completion of performing the at least one model-related procedure.
[0392] In some embodiments, the message used for informing the starting of performing the at least one model-related procedure comprises at least one of the following: an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure, an indication of whether a model training is needed for a functionality supported by the first device, an indication of whether a model identification is needed for a functionality supported by the first device, an indication of whether an associated identity is needed, an indication of whether a model structure alignment procedure is needed, an indication of whether a dataset transfer is needed, an indication of whether a model transfer is needed for a functionality supported by the first device, an indication of whether a validation is needed for a functionality supported by the first device, or an indication of whether a performance monitoring is needed for a functionality supported by the first device.
[0393] In some embodiments, the second device is further caused to: receive, from the first device, a first message comprising at least one of the following: an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, or an indication of the at least one model-related procedure needed for determining the at least one applicable functionality, an indication of there is no applicable functionality, or an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure.
[0394] In some embodiments, the second device is further caused to: transmit, to the first device, a message used for triggering the first device to start performing the at least one model-related procedure; and receive, from the first device, a message for informing the second device the completion of performing the at least one model-related procedure.
[0395] In some embodiments, the message used for triggering the first device to start performing the at least one model-related procedure comprises: configuration information used by the fist device to perform the at least one model-related procedure.
[0396] In some embodiments, the second device is further caused to: receive, from the first device, a message indicating the at least one applicable functionality and model-related information.
[0397] In some embodiments, the second device is further caused to: transmit, to the first device, second configuration information indicating at least one of the following: a model inference configuration, a performance monitoring configuration, a configuration of the performance monitoring an initial activation state of the at least one applicable functionality, or an initial activation state of a previously-received model-related configuration comprising at least one of the following: a model inference configuration, a performance monitoring configuration, a model training configuration.
[0398] In some embodiments, the second configuration information is associated with the at least one model-related procedure.
[0399] In some embodiments, the at least one model-related procedure is performed within a duration, wherein time length of the duration is configured by the second device, reported by the first device or defined as a default value.
[0400] In some embodiments, the at least one model-related procedure comprises the model transfer, and the duration is associated with at least one of the following: a mode size, a model structure, a model type, or the number of applicable functionalities.
[0401] In some embodiments, the at least one model-related procedure comprises the model transfer and the second device is further caused to: transmit, to the first device, a set parameters comprising at least one parameter of a reference model structure; and apply the set of parameters to obtain a valid mode.
[0402] In some embodiments, the set of parameters further comprises at least one of the following: a model identity, a network side model identity of a two-sided model, a user equipment (UE) side model identity of a two-sided model.
[0403] In some embodiments, the at least one model-related procedure comprises the model transfer and the second device is further caused to: transmit information of a reference model structure to the second device or receive the information of the reference model structure from the second device, wherein the information of the reference model structure indicates at least one of the following: an index of model structure, a model type, a number of layers of a model, a required minimum number of layers of a model, a supported maximum number of layers of a model, a layer structure type of a model, a number of parameters of a layer, a required minimum number of parameters of a layer, a supported maximum number of parameters of a layer, a connection type between different layers, input / output related information of a model, quantization information of a model, a set of training related parameters of a model, or a set of generalization related parameters of a model.
[0404] In some embodiments, the at least one model-related procedure comprises the model identification and the second device is further caused to: transmit, to the first device, at least one associated identity, each associated identity corresponding to a configuration for data collection or model training; perform, based on at least one configuration corresponding to the at least one associated identity, data collection or model training; and receive , form the first device, a message comprising model-related information corresponding to the at least one associated identity upon completion of performing the data collection or model training.
[0405] In some embodiments, the model-related information comprises at least one model identity, each model identity corresponding to a functionality.
[0406] In some embodiments, the first device is a terminal device and the second device is a network device.
[0407] In an aspect, a first device comprises: at least one processor; and at least one memory coupled to the at least one processor and storing instructions thereon, the instructions, when executed by the at least one processor, causing the device to perform the method implemented by the first device discussed above.
[0408] In an aspect, a second device comprises: at least one processor; and at least one memory coupled to the at least one processor and storing instructions thereon, the instructions, when executed by the at least one processor, causing the device to perform the method implemented by the second device discussed above.
[0409] In an aspect, a computer readable medium having instructions stored thereon, the instructions, when executed on at least one processor, causing the at least one processor to perform the method implemented by the first device discussed above.
[0410] In an aspect, a computer readable medium having instructions stored thereon, the instructions, when executed on at least one processor, causing the at least one processor to perform the method implemented by the second device discussed above.
[0411] In an aspect, a computer program comprising instructions, the instructions, when executed on at least one processor, causing the at least one processor to perform the method implemented by the first device discussed above.
[0412] In an aspect, a computer program comprising instructions, the instructions, when executed on at least one processor, causing the at least one processor to perform the method implemented by the second device discussed above.
[0413] Generally, various embodiments of the present disclosure may be implemented in hardware or special purpose circuits, software, logic or any combination thereof. Some aspects may be implemented in hardware, while other aspects may be implemented in firmware or software which may be executed by a controller, microprocessor or other computing device. While various aspects of embodiments of the present disclosure are illustrated and described as block diagrams, flowcharts, or using some other pictorial representation, it will be appreciated that the blocks, apparatus, systems, techniques or methods described herein may be implemented in, as non-limiting examples, hardware, software, firmware, special purpose circuits or logic, general purpose hardware or controller or other computing devices, or some combination thereof.
[0414] The present 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, being executed in a device on a target real or virtual processor, to carry out the process or method as described above with reference to FIGS. 1 to 9. Generally, program modules include routines, programs, libraries, objects, classes, components, data structures, or the like that perform particular tasks or implement particular abstract data types. The functionality of the program modules may be combined or split between program modules as desired in various embodiments. Machine-executable instructions for program modules may be executed within a local or distributed device. In a distributed device, program modules may be located in both local and remote storage media.
[0415] Program code for carrying out methods of the present disclosure may be written in any combination of one or more programming languages. These program codes may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the program codes, when executed by the processor or controller, cause the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may execute entirely on a machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
[0416] The above program code may be embodied on a machine readable medium, which may be any tangible medium that may contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine readable medium may be a machine readable signal medium or a machine readable storage medium. A machine readable medium may include but not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine readable storage medium would include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM) , a read-only memory (ROM) , an erasable programmable read-only memory (EPROM or Flash memory) , an optical fiber, a portable compact disc read-only memory (CD-ROM) , an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0417] Further, while operations are depicted in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing may be advantageous. Likewise, while several specific implementation details are contained in the above discussions, these should not be construed as limitations on the scope of the present disclosure, but rather as descriptions of features that may be specific to particular embodiments. Certain features that are described in the context of separate embodiments may also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment may also be implemented in multiple embodiments separately or in any suitable sub-combination.
[0418] Although the present disclosure has been described in language specific to structural features and / or methodological acts, it is to be understood that the present disclosure defined in the appended claims is not necessarily limited to the specific features or acts described above. Rather, the specific features and acts described above are disclosed as example forms of implementing the claims.
Claims
A first device comprising:a processor configured to cause the first device to:transmit, to a second device, capability-related information comprising a set of functionalities supported by the first device;perform at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model; anddetermine, from the set of functionalities, at least one applicable functionality after completion of performing the at least one model-related procedure.The first device of claim 1, wherein the capability-related information further comprises at least one of the following:information about at least one functionality, wherein a functionality of the at least one functionality is applicable on a condition of the completion of the model-related procedure,an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality,an indication of the at least one model-related procedure needed for determining the at least one applicable functionality,an indication of whether a model training is needed for a functionality supported by the first device,an indication of whether a model identification is needed for a functionality supported by the first device,an indication of whether an associated identity is needed,an indication of whether a model structure alignment procedure is needed,an indication of whether a dataset transfer is needed,an indication of whether a model transfer is needed for a functionality supported by the first device,an indication of whether a validation is needed for a functionality supported by the first device, oran indication of whether a performance monitoring is needed for a functionality supported by the first device.The first device of claim 1, wherein the first device is further caused to:receive, from the second device, first configuration information comprising at least one network additional condition, wherein the at least one model-related procedure is performed after receiving the first configuration information.The first device of claim 3, wherein the first configuration information further comprises at least one configuration comprising at least one of the following:a configuration used by the first device to trigger the at least one model-related procedure,at least one model inference configuration associated with the set of functionalities supported by the first device, orat least one performance monitoring configuration associated with the set of functionalities supported by the first device.The first device of claim 4, wherein the at least one model inference configuration is considered invalid or deactivated before the completion of model-related procedure.The first device of claim 1, wherein the first device is further caused to:transmit, a message used for informing the second device starting of performing the at least one model-related procedure;perform the at least one model-related procedure; andtransmit, a message used for informing the second device the completion of performing the at least one model-related procedure.The first device of claim 6, wherein the message used for informing the starting of performing the at least one model-related procedure comprises at least one of the following:an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, oran indication of the at least one model-related procedure needed for determining the at least one applicable functionality,an indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure,an indication of whether a model training is needed for a functionality supported by the first device,an indication of whether a model identification is needed for a functionality supported by the first device,an indication of whether an associated identity is needed,an indication of whether a model structure alignment procedure is needed,an indication of whether a dataset transfer is needed,an indication of whether a model transfer is needed for a functionality supported by the first device,an indication of whether a validation is needed for a functionality supported by the first device, oran indication of whether a performance monitoring is needed for a functionality supported by the first device.The first device of claim 1, wherein the first device is further caused to:transmit, to the second device, a first message comprising at least one of the following:an indication of whether the at least one model-related procedure is needed for determining the at least one applicable functionality, oran indication of the at least one model-related procedure needed for determining the at least one applicable functionality,an indication of there is no applicable functionality, oran indication of at least one candidate functionality, wherein a candidate functionality of the at least one candidate functionality is applicable on a condition of the completion of the model-related procedure; andperform the at least one model-related procedure after transmitting the first message.The first device of claim 1, wherein the first device is further caused to:transmit the first message to the second device based on a determination that:there is no available model,at least one of the following comprised in first configuration information transmitted by the second device is not supported by the first device: at least one additional condition, a model inference configuration, or a performance monitoring configuration, oran internal condition of the first device causes no available model or no applicable functionality.The first device of claim 9, wherein the first device is further caused to:receive, from the second device, a message used for triggering the first device to start performing the at least one model-related procedure;perform the at least one model-related procedure in response to receiving the message; andtransmit, a message for informing the second device the completion of performing the at least one model-related procedure.The first device of claim 10, wherein the message used for triggering the first device to start performing the at least one model-related procedure comprises: configuration information used by the fist device to perform the at least one model-related procedure.The first device of claim 1, wherein the first device is further caused to:after the completion of the model-related procedure, transmit, to a second device, a message indicating the at least one applicable functionality and model-related information.The first device of claim 1, wherein the first device is further caused to:receive, from the second device, second configuration information indicating at least one of the following:a model inference configuration,a performance monitoring configuration,a configuration of the performance monitoringan initial activation state of the at least one applicable functionality, oran initial activation state of a previously-received model-related configuration comprising at least one of the following: a model inference configuration, a performance monitoring configuration, a model training configuration.The first device of claim 13, wherein the second configuration information is associated with the at least one model-related procedure.The first device of claim 1, wherein the at least one model-related procedure is performed within a duration, wherein time length of the duration is configured by the second device, reported by the first device or defined as a default value.The first device of claim 15, wherein the at least one model-related procedure comprises the model transfer, and the duration is associated with at least one of the following:a mode size,a model structure,a model type, orthe number of applicable functionalities.The first device of any of claims 1 to 16, wherein the at least one model-related procedure comprises the model transfer and the first device is further caused to:receive, from the second device, a set parameters comprising at least one parameter of a reference model structure; andapply the set of parameters to obtain a valid mode.The first device of claim 17, wherein the set of parameters further comprises at least one of the following: a model identity, a network side model identity of a two-sided model, a user equipment (UE) side model identity of a two-sided model.The first device of any of claims 1 to 18, wherein the at least one model-related procedure comprises the model transfer and the first device is further caused to: transmit information of a reference model structure to the second device or receive the information of the reference model structure from the second device, wherein the information of the reference model structure indicates at least one of the following:an index of model structure,a model type,a number of layers of a model,a required minimum number of layers of a model,a supported maximum number of layers of a model,a layer structure type of a model,a number of parameters of a layer,a required minimum number of parameters of a layer,a supported maximum number of parameters of a layer,a connection type between different layers,input / output related information of a model,quantization information of a model,a set of training related parameters of a model, ora set of generalization related parameters of a model.The first device of claim 1, wherein the at least one model-related procedure comprises the model identification and the first device is further caused to:receive, from the second device, at least one associated identity, each associated identity corresponding to a configuration for data collection or model training;perform, based on at least one configuration corresponding to the at least one associated identity, data collection or model training; andtransmit, to the second device, a message comprising model-related information corresponding to the at least one associated identity upon completion of performing the data collection or model training.The first device of claim 20, wherein the model-related information comprises at least one model identity, each model identity corresponding to a functionality.The first device of any of claims 1 to 21, wherein the first device is a terminal device and the second device is a network device.A second device comprising:a processor configured to cause the second device to:receive, from a first device, capability-related information comprising a set of functionalities supported by the first device; anddetermine, from the set of functionalities, at least one applicable functionality after the first device performing at least one model-related procedure, the at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model.A communication method implemented at a first device, comprising:transmitting, to a second device, capability-related information comprising a set of functionalities supported by the first device;performing at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model; anddetermining, from the set of functionalities, at least one applicable functionality after completion of performing the at least one model-related procedure.A communication method implemented at a second device, comprising:receiving, from a first device, capability-related information comprising a set of functionalities supported by the first device; anddetermining, from the set of functionalities, at least one applicable functionality after the first device performing at least one model-related procedure, the at least one model-related procedure comprising at least one of the following: a model identification, a model transfer, a model training, a model validation, or a performance monitoring of a model.A computer readable medium having instructions stored thereon, the instructions, when executed on at least one processor, causing the at least one processor to perform the method according to any of claims 24-25.
Citation Information
Patent Citations
Method and device for identifying artificial intelligence / machine learning function and model
CN118509882A
Lifecycle management supporting ai / ML for air interface enhancement
WO2024148935A1
Transmission by the UE of the requirements upon which ML features combinations are supported by the ue
WO2024193940A1