[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

WO2024210500A1 - Procédé et appareil de sélection d'un ue pour un apprentissage fédéré - Google Patents

Procédé et appareil de sélection d'un ue pour un apprentissage fédéré Download PDF

Info

Publication number
WO2024210500A1
WO2024210500A1 PCT/KR2024/004312 KR2024004312W WO2024210500A1 WO 2024210500 A1 WO2024210500 A1 WO 2024210500A1 KR 2024004312 W KR2024004312 W KR 2024004312W WO 2024210500 A1 WO2024210500 A1 WO 2024210500A1
Authority
WO
WIPO (PCT)
Prior art keywords
nef
ues
list
filtering criteria
data volume
Prior art date
Application number
PCT/KR2024/004312
Other languages
English (en)
Inventor
David Gutierrez Estevez
Tingyu XIN
Original Assignee
Samsung Electronics Co., Ltd.
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Samsung Electronics Co., Ltd. filed Critical Samsung Electronics Co., Ltd.
Publication of WO2024210500A1 publication Critical patent/WO2024210500A1/fr

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W24/00—Supervisory, monitoring or testing arrangements
    • H04W24/02—Arrangements for optimising operational condition
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/098—Distributed learning, e.g. federated learning
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W8/00—Network data management
    • H04W8/18—Processing of user or subscriber data, e.g. subscribed services, user preferences or user profiles; Transfer of user or subscriber data

Definitions

  • Certain examples of the present disclosure provide approaches for selecting User Equipment (UE) for federated learning.
  • UE User Equipment
  • certain examples of the present disclosure provide methods, apparatus and systems for selecting UEs for Artificial Intelligence/Machine Learning federated learning in 3rd Generation Partnership Project (3GPP) networks such as 5th Generation (5G) and 6th Generation (6G) networks.
  • 3GPP 3rd Generation Partnership Project
  • 5G mobile communication technologies define broad frequency bands such that high transmission rates and new services are possible, and can be implemented not only in “Sub 6GHz” bands such as 3.5GHz, but also in “Above 6GHz” bands referred to as mmWave including 28GHz and 39GHz.
  • 6G mobile communication technologies referred to as Beyond 5G systems
  • terahertz bands for example, 95GHz to 3THz bands
  • IIoT Industrial Internet of Things
  • IAB Integrated Access and Backhaul
  • DAPS Dual Active Protocol Stack
  • 5G baseline architecture for example, service based architecture or service based interface
  • NFV Network Functions Virtualization
  • SDN Software-Defined Networking
  • MEC Mobile Edge Computing
  • multi-antenna transmission technologies such as Full Dimensional MIMO (FD-MIMO), array antennas and large-scale antennas, metamaterial-based lenses and antennas for improving coverage of terahertz band signals, high-dimensional space multiplexing technology using OAM (Orbital Angular Momentum), and RIS (Reconfigurable Intelligent Surface), but also full-duplex technology for increasing frequency efficiency of 6G mobile communication technologies and improving system networks, AI-based communication technology for implementing system optimization by utilizing satellites and AI (Artificial Intelligence) from the design stage and internalizing end-to-end AI support functions, and next-generation distributed computing technology for implementing services at levels of complexity exceeding the limit of UE operation capability by utilizing ultra-high-performance communication and computing resources.
  • FD-MIMO Full Dimensional MIMO
  • OAM Organic Angular Momentum
  • RIS Reconfigurable Intelligent Surface
  • Wireless or mobile (cellular) communications networks in which a mobile terminal (e.g., user equipment (UE), such as a mobile handset) communicates via a radio link with a network of base stations, or other wireless access points or nodes, have undergone rapid development through a number of generations.
  • a mobile terminal e.g., user equipment (UE), such as a mobile handset
  • 3GPP 3 rd Generation Partnership Project
  • 4G and 5G systems are now widely deployed, and development of Sixth Generation (6G) Systems is in progress.
  • 3GPP standards for 4G systems include an Evolved Packet Core (EPC) and an Enhanced-UTRAN (E-UTRAN: an Enhanced Universal Terrestrial Radio Access Network).
  • EPC Evolved Packet Core
  • E-UTRAN Enhanced-UTRAN
  • LTE Long Term Evolution
  • LTE is commonly used to refer to the whole system including both the EPC and the E-UTRAN, and LTE is used in this sense in the remainder of this document.
  • LTE should also be taken to include LTE enhancements such as LTE Advanced and LTE Pro, which offer enhanced data rates compared to LTE.
  • 5G New Radio 5G New Radio
  • 5G NR 5G New Radio
  • NR is designed to support the wide variety of services and use case scenarios envisaged for 5G networks, though builds upon established LTE technologies.
  • New frameworks and architectures are also being developed as part of 5G networks in order to increase the range of functionality and use cases available through 5G networks.
  • 3GPP has also started studying the benefits of introducing Artificial Intelligence (AI)/Machine Learning (ML) solutions to communications networks, for example, enhancement of management and orchestration, performance, resource allocation, in addition to reduction of complexity and overhead in the network
  • AI Artificial Intelligence
  • ML Machine Learning
  • AI/ML models and/or data might be transferred across the AI/ML applications (AFs), 5GC and UEs.
  • AFs AI/ML applications
  • 5GC 5GC
  • UE User Equipment
  • the AI/ML works could be divided into two main phases: model training and inference. During model training and inference, multiple rounds of interaction may be required.
  • the AI/ML operation types could be categorised into three types: model splitting, model sharing, and distributed/federated learning.
  • the requirements, frequency and volume of data transmission may differ for different AI/ML processing phrases and/or operation types.
  • the AI/ML operation/model is split into multiple parts according to the current task and environment.
  • the intention is to offload the computation-intensive, energy-intensive parts to network endpoints, whereas leave the privacy-sensitive and delay-sensitive parts at the end device.
  • the device executes the operation/model up to a specific part/layer and then sends the intermediate data to the network endpoint.
  • the network endpoint executes the remaining parts/layers and feeds the inference results back to the device.
  • Multi-functional mobile terminals might need to switch the AI/ML model in response to task and environment variations.
  • the condition of adaptive model selection is that the models to be selected are available for the mobile device.
  • online model distribution i.e. new model downloading
  • an AI/ML model can be distributed from a NW endpoint to the devices when they need it to adapt to the changed AI/ML tasks and environments.
  • the model performance at the UE needs to be monitored constantly.
  • the cloud server trains a global model by aggregating local models partially-trained by each end devices.
  • a UE performs the training based on the model downloaded from the AI server using the local training data. Then the UE reports the interim training results to the cloud server via 5G UL channels.
  • the server aggregates the interim training results from the UEs and updates the global model. The updated global model is then distributed back to the UEs and the UEs can perform the training for the next iteration.
  • This sub-clause describes the list of 5GC enablers to support the following AI/ML operations in the Application layer: Distributed/Federated Learning, Model sharing and Model operation splitting.
  • the AF that aims to provide an AI/ML operation e.g. for Distributed/Federated Learning may request assistance from the 5GC as described in clause 5.46.2 of TS 23.501 by subscribing to the NEF to be notified about the list of UEs that fulfil certain filtering criteria. Details of the procedures are described in clause 4.15.x of TS 23.502. This list of UEs may become the list of candidate UEs for this AI/ML operation depending on the AF internal policies. Alternatively, depending on operator policies, the AF may select the list of UEs for the AI/ML operation e.g. Distributed/Federated Learning without NEF involvement as described in Annex I of TS 23.502.
  • the AF that selected a list of UEs may request the network to provide the preferred time window for the AI/ML operation using the Planned Data Transfer with QoS (PDTQ) requirements described in clause 6.1.2.7 of TS 23.503.
  • PDTQ Planned Data Transfer with QoS
  • the AF discovers a suitable NEF and requests the NEF to provide QoS for the list of UEs, each UE identified by its UE IP address, that were selected as described in clause 4.15.13 of TS 23.502.
  • the AF may subscribe to QoS Monitoring for those AF requests for QoS that result in a successful resource allocation.
  • the AF provides one or more of the following that are derived from the performance requirements listed in clause 7.10 of TS 22.261: QoS parameters, QoS profiles, QoS requirements, corresponding 5QIs etc.
  • the AF may be notified about changes in the list of candidate UEs, as such the AF may request a new preferred time window for the AI/ML operation using the Planned Data Transfer with QoS or may request to provide QoS with an updated list of UEs.
  • the updated UEs some of them that don't require resources allocation any longer, and some new selected UEs may require resource allocation and QoS Monitoring.
  • the AF that aims to provide an AI/ML operation e.g. Model Sharing may request assistance from the 5GC as described in clause 4.15.3.2.3 in TS 23.502 by subscribing to the NEF to be notified on the traffic volume shared between the UE and the AI/ML application server, this may help the AI/ML application server to determine how large the model is, and then e.g. select large models less frequently than small models.
  • ⁇ AF requests to the 5G System in the context of 5GS assistance to AI/ML operations in the application layer shall be authorized by the 5GC using existing mechanisms (before Release 18).
  • ⁇ UPF can perform traffic detection for AI/ML traffic as defined in clause 5.8.2.4 of TS 23.501 and the network needs to support charging for AI/ML traffic.
  • 5G system may support UE member selection assistance functionality to assist the AF in selecting member/ UE(s)(s) that can be used in application operations such as AI/ML based applications (e.g., Federated Learning).
  • the UE member selection assistance functionality is hosted by NEF, and the features of the UE member selection assistance functionality hosted by NEF include (see clause 4.15.13 in TS 23.502 for details of UE member selection procedures):
  • Receiving a request from the AF that shall include a list of target member/ UE(s)(s), optionally (a) time window(s), and one or more filtering criteria as specified in Table 4.15.Y.2-1 of TS 23.502 (e.g., number of UEs, UE location, trajectory and direction, QoS requirements, Area of Interest, DNN, S-NSSAI, expected application operation time duration, preferred access/RAT type, Desired UE transmission latency performance).
  • filtering criteria e.g., number of UEs, UE location, trajectory and direction, QoS requirements, Area of Interest, DNN, S-NSSAI, expected application operation time duration, preferred access/RAT type, Desired UE transmission latency performance.
  • NEF interacts with 5GC NFs using existing services to collect the corresponding data from relevant 5GC NFs (e.g. PCF, NWDAF and AMF) to derive the list(s) of candidate UE(s) (i.e., UE(s) among the list of target member/ UE(s)(s) provided by the AF).
  • relevant 5GC NFs e.g. PCF, NWDAF and AMF
  • Providing the AF with the UE member selection assistance information including one or more lists of candidate UE(s), and optionally other additional information (e.g., one or more recommended time window(s) for performing the application operation, QoS of each target UE, UE(s) location, Access/RAT type, etc.).
  • the UE member selection assistance information provided by the NEF can be used by the AF to select member/ UE(s)s. (See clause 5.2.6.X in TS 23.502 for details of parameters).
  • AF in either trusted or untrusted domain can select the UE members e.g., participating in federating learning operation, by collecting the corresponding data using network exposure information as described in clause 4.15.Y of TS 23.502, e.g. UE location reporting from the AMF, user plane information from the UPF and data analytics from NWDAF.
  • network exposure information as described in clause 4.15.Y of TS 23.502, e.g. UE location reporting from the AMF, user plane information from the UPF and data analytics from NWDAF.
  • the AIML operation model training and inference, multiple rounds of interaction may be required.
  • High volume traffic will be transfer between the AIML server and the UE(s).
  • a number of UEs will participate the AIML operation, the traffic volume will be significant high.
  • the member/ UE(s) selection performed by the NEF should be carefully specified. In the current spec, the NEF assists with the member/ UE(s) selection using the filter/ filtering criteria indicated by the AF.
  • the filter/ filtering criteria may include number of UEs, UE location, trajectory and direction, QoS requirements, Area of Interest, DNN, S-NSSAI, expected application operation time duration, preferred access/RAT type, Desired UE transmission latency performance.
  • more filter/ filtering criteria should be also considered, e.g. the load of the candidates UEs and the theirs serving gNB(s) , the RRC/CM state of the UEs ect. to balance the load of the network and avoid the signalling storm due to paging etc.
  • the NEF should provide the UEs that fulfil the requirements of the AIML operation, therefore allowing a service quality of the AIML to be maintained.
  • the AIML application/AF will provide the filters/ filtering criteria for member selection/recommendation to the NEF. Based on the requirement in the filters/ filtering criteria from the AF (e.g. the analytics ID, event ID, application ID, DNN.
  • the NEF collects data from different 5GC NFs (e.g. NWDAF, SMF, UPF, UDM, AMF, PCF etc.), AFs or OAM etc. for the required data (e.g. end-to-end transmission time/ latency, end-to-end transmission time/ latency variation, UE's RM/CM/RRC state, UE's abnormal behaviours related requirement etc.).
  • 5GC NFs e.g. NWDAF, SMF, UPF, UDM, AMF, PCF etc.
  • AFs or OAM etc. for the required data (e.g. end-to-end transmission time/ latency, end-to-end transmission time/ latency variation, UE's RM/CM/RRC state, UE's abnormal behaviours related requirement etc.).
  • the NEF may choose the UEs that fulfil the filtering criteria from the list of candidate UEs provided by the AF.
  • the NEF is able to assist the AF to select the UEs that can operate the corresponding AIML stably and efficiently, and therefore, improve and maintain the high service quality of AIML-based services.
  • the member/ UE(s) selection related procedures, 5GC services, filters/ filtering criteria etc. could be deployed to any other service that requires member selection, not limited to member/ UE(s) selection of federated learning.
  • Certain examples of the present disclosure provide a method for selecting User Equipment (UE) for a service operation in a 3GPP wireless communications system, the method comprising: receiving at a Network Exposure Function (NEF) from an Application Function (AF) a Nnef_UEMemberSelectionAssistance_Subscribe request including a subscription correlation ID and one or more filtering criteria; correlating by the NEF the Nnef_UEMemberSelectionAssistance_Subscribe message to an existing subscription according to the subscription correlation ID; collecting by the NEF from one or more Network Functions (NFs) information for each UE in a list of target member UEs based on the filtering criteria; deriving by the NEF a list of one or more candidate UEs associated with the existing subscription using the collected information; and transmitting from the NEF to the AF a Nnef_UEMemberSelectionAssistance_Notify message including the list of candidate UEs.
  • NEF Network Exposure Function
  • AF Application Function
  • the service operation may comprise an Artificial Intelligence/Machine Learning (AI/ML) federated learning operation.
  • AI/ML Artificial Intelligence/Machine Learning
  • the method may further comprise: receiving at the NEF from the AF an indication to modify the filtering criteria (e.g. the presence of a subscription correlation ID, but the absence of a list of target member UEs, in an Nnef_UEMemberSelectionAssistance_Subscribe request).
  • an indication to modify the filtering criteria e.g. the presence of a subscription correlation ID, but the absence of a list of target member UEs, in an Nnef_UEMemberSelectionAssistance_Subscribe request.
  • modifying the filtering criteria may comprise one or more of: adding one or more new filtering criteria; deleting one or more existing filtering criteria; updating one or more existing filtering criteria; and modifying one or more parameters associated with one or more existing filtering criteria.
  • the method may further comprise: deriving by the NEF an updated list of one or more candidate UEs based on the modified filtering criteria.
  • the list of target member UEs may be received by the NEF from the AF.
  • the list of candidate UEs may be selected from the list of target member UEs associated with the existing subscription.
  • the list of target member UEs may have been received with a previous request associated with the existing subscription.
  • the Nnef_UEMemberSelectionAssistance_Subscribe request does not include a list of target member UEs.
  • the list of candidate UEs may include a list of one or more UEs that do not meet the one or more filtering criteria or a list of one or more UEs that do meet the one or more filtering criteria.
  • the Nnef_UEMemberSelectionAssistance_Subscribe message may further include an indication of a periodicity to update the member UEs.
  • the method may further comprise, in accordance with the periodicity, periodically deriving by the NEF an updated list of candidate UEs and transmitting from the NEF to the AF a Nnef_UEMemberSelectionAssistance_Notify message including the updated list of candidate UEs.
  • the one or more filtering criteria may include a UE end-to-end data volume transfer time, wherein the end-to-end data volume transfer time may include one or more of an end-to-end data volume transfer time for a specific data volume between a UE and the AF, a variance of an end-to-end data volume transfer time for a specific data volume between a UE and the AF, and an average of an end-to-end data volume transfer time for a specific data volume between a UE and the AF.
  • information associated with the filtering criteria may include one or more of an area of interest indicating a location area of candidate UEs, and a time window indicating a start and stop time for selecting the list of candidate UEs.
  • the Nnef_UEMemberSelectionAssistance_Notify message may include a cause of one or more UEs not meeting the one or more filtering criteria.
  • the deriving may include collecting and consolidating information from one or more core network Network Functions (NFs) associated with the one or more filtering criteria.
  • NFs core network Network Functions
  • Certain examples of the present disclosure provide a Network Exposure Function (NEF) of a 3GPP wireless communications system, the NEF configured to implement the method according to any example, aspect, embodiment and/or claim disclosed herein.
  • NEF Network Exposure Function
  • Certain examples of the present disclosure provide a method for selecting User Equipment (UE) for Artificial Intelligence/Machine Learning (AI/ML) federated learning in a 3GPP wireless communications system, the method comprising: receiving at a Network Exposure Function (NEF) from an Application Function (AF) a Nnef_UEMemberSelectionAssistance_Subscribe message including an application ID of the AF, a list of one or more target UEs, and one or more filtering criteria including an end-to-end data volume transfer time filtering criteria; deriving by the NEF a list of candidate UEs from the list or more of more target UEs using the one or more filtering criteria; and transmitting from the NEF to the AF a Nnef_UEMemberSelectionAssistance_Notify message including the list of candidate UEs.
  • NEF Network Exposure Function
  • AF Application Function
  • AF Application Function
  • filtering criteria including an end-to-end data volume transfer time filtering criteria
  • the list of candidate UEs may include one or more UEs from the target UEs that meet or do not meet the end-to-end data volume transfer time filtering criteria.
  • the end-to-end data volume transfer time filtering criteria may include one or more of an end-to-end data volume transfer time for a specific data volume between a UE and an AF, a variance of an end-to-end data volume transfer time for a specific data volume between a UE and the AF, and an average of an end-to-end data volume transfer time for a specific data volume between a UE and the AF.
  • the one or more filtering criteria may include one or more of an area of interest indicating a location area of candidate UEs, a variance of an end-to-end data volume transfer time, and a time window indicating a start and stop time for selecting the list of candidate UEs.
  • the deriving may include collecting and consolidating analytics associated with the one or more filtering criteria received from a Network Data Analytics Function (NWDAF).
  • NWDAAF Network Data Analytics Function
  • the collecting and consolidating information associated with the one or more filtering criteria received from a Network Data Analytics Function may include: transmitting from the NEF to the NWDAF a Nnwdaf_AnalyticsSubscription_Subscribe / Nnwdaf_AnalyticsInfo_Request message including an analytics ID indicating the end-to-end data volume transfer time, the application ID of the AF, and the list of one or more target UEs, and receiving at the NEF from the NWDAF a Nnwdaf_AnalyticsSubscription_Notify / Nnwdaf_AnalyticsInfo_Request response including analytics on the end-to-end data volume transfer time for the one or more target UEs.
  • NWDAF Network Data Analytics Function
  • the method may further comprise, deriving, by the NEF, a Single Network Slice Selection Assistance Information (S-NSSAI) and a Data Network Name (DNN) the AF has access to, and including the S-NSSAI and DNN in the Nnwdaf_AnalyticsSubscription_Subscribe / Nnwdaf_AnalyticsInfo_Request message.
  • S-NSSAI Single Network Slice Selection Assistance Information
  • DNN Data Network Name
  • the method may further comprise verifying, by the NEF, the authorization of the Nnef_UEMemberSelectionAssistance_Subscribe message, and identifying, by the NEF, information to be collected and executed based on the end-to-end data volume transfer time related filtering criteria provided by the AF.
  • NEF Network Exposure Function
  • Embodiments of the present disclosure provides methods and apparatus for selecting UEs for the federated learning (FL) operations based on filtering criteria.
  • FL federated learning
  • Figure 1a provides an example of 5GC assistance to UE member selection for end-to-end data volume transfer time related filtering criteria
  • Figure 1b provides an example of 5GC assistance to member selection based on the UE's travel history, direction and separation distance;
  • Figure 1c provides an example of 5GC assistance to UE member selection and update
  • Figure 2 provides an example general procedure of 5GC assistance to UE/Member selection and update
  • Figure 3 provides an example detailed specific procedure for the 5GC assistance to member selection and update based on the various filter/ filtering criteria
  • Figure 4 provides a block diagram of an exemplary network entity/function that may be used in certain examples of the present disclosure.
  • X for Y (where Y is some action, process, operation, function, activity or step and X is some means for carrying out that action, process, operation, function, activity or step) encompasses means X adapted, configured or arranged specifically, but not necessarily exclusively, to do Y.
  • 3GPP 4G e.g., LTE
  • 5G e.g., NR
  • the techniques disclosed herein are not limited to these examples or to 3GPP 4G (e.g., LTE) and/or 5G (e.g., NR), and may be applied in any suitable system or standard, for example one or more existing and/or future generation wireless communication systems or standards (e.g., B5G, 5G-Advanced, 6G etc.).
  • 3GPP 4G e.g., LTE
  • 5G e.g., NR
  • 5G Advanced and/or 6G 3GPP Release 17, 18, 19, 20, etc.
  • 3GPP Release 17, 18, 19, 20, etc. 3GPP Release 17, 18, 19, 20, etc.
  • the functionality of the various network entities and other features disclosed herein may be applied to corresponding or equivalent entities or features in other communication systems or standards.
  • Corresponding or equivalent entities or features may be regarded as entities or features that perform the same or similar role, function, operation or purpose within the network.
  • model and model functionality may be used interchangeably.
  • a particular network entity may be implemented as a network element on dedicated hardware, as a software instance running on a dedicated hardware, and/or as a virtualised function instantiated on an appropriate platform, e.g. on a cloud infrastructure.
  • One or more of the messages in the examples disclosed herein may be replaced with one or more alternative messages, signals or other type of information carriers that communicate equivalent or corresponding information.
  • One or more non-essential elements, entities and/or messages may be omitted in certain examples.
  • ⁇ Information carried by a particular message in one example may be carried by two or more separate messages in an alternative example.
  • ⁇ Information carried by two or more separate messages in one example may be carried by a single message in an alternative example.
  • the transmission of information between network entities is not limited to the specific form, type and/or order of messages described in relation to the examples disclosed herein.
  • an apparatus/device/network entity configured to perform one or more defined network functions and/or a method therefor.
  • Such an apparatus/device/network entity may comprise one or more elements, for example one or more of receivers, transmitters, transceivers, processors, controllers, modules, units, and the like, each element configured to perform one or more corresponding processes, operations and/or method steps for implementing the techniques described herein.
  • an operation/function of X may be performed by a module configured to perform X (or an X-module).
  • Certain examples of the present disclosure may be provided in the form of a system (e.g., a network) comprising one or more such apparatuses/devices/network entities, and/or a method therefor.
  • UE RRC, CM and RM Status as a Filter/ Filtering Criteria for Member Selection
  • the UE In order to transmit data between UE and the network, the UE should be enter the proper RRC, CM and RM status.
  • Two RM/ registration states are used in the UE and the AMF that reflect the registration status of the UE in the selected PLMN:
  • Two CM states are used to reflect the NAS signalling Connection of the UE with the AMF:
  • RRC has three distinct states:
  • the data transmission between the UE and network could be performed when the UE is in RM-REGISTERED, CM-CONNECTED and RRC_CONNECTED. Even though in some cases, e.g. IoT devices or small data transmission, the UE may be able to transmit uplink date in RRC_IDLE and RRC_INACTIVE.
  • the UE may be able to perform the AIML operation only in the required RM, CM and/or RRC state or reachability status, e.g. CM-CONNECTED and RRC_CONNECTED.
  • the network may initiate CN or RAN paging to move the UE to the required state, or the UE may initiate the registration procedure, and/or the connection establishment procedures to participate the AIML operation.
  • the above procedures will generate significant signalling between the UE and the network, especially for the service that may involve a large number of UEs, e.g. federated learning, this may bring signal storm to the network.
  • the UEs For some cases, there are probably no sufficient time to the UEs to the required state, e.g. urgent AIML services, member update etc.
  • the member update might be triggered by UE's interests, network requirements, member/ UE(s) variation due to some reasons during the federated learning. If the selected UEs are not in the required RM, CM and/or RRC state before the service starts, the UE may not be able to join the AIML operation quickly/on time as required, the performance of the service operation will be degraded.
  • the UE registration state, connectivity state and UE Reachability e.g. one or more of the following states could be considered as the filter/ filtering criteria for member/ UE(s) selection:
  • the RM state registration state (Registered or Deregistered),
  • CM state connectivity state (IDLE or CONNECTED)
  • the AF may include the required sates of the UEs to the NEF, e.g. via Nnef_UEMemberSelectionAssistance_Subscribe or Nnef_UEMemberSelectionAssistance_update, as the specific parameters depending on the UE member filtering criteria corresponds to the filter/ filtering criteria.
  • Table 1 An example of the UE Reachability/ UE status of the selected UEs in Table 1 could be RM-REGISTERED, CM-CONNECTED and RRC_CONNECTED or RRC_INACITVE.
  • Table 1 provides a description of UE Member filer/ filtering criteria
  • the UE Reachability/ UE status may include the RM, CM and RRC state.
  • UE Member filtering criteria Description of filtering criterion UE filtering information Detailed description clause UE Reachability/ UE status/ UE registration state, connectivity state and reachability Indicate the registration state (Registered or Deregistered), connectivity state (IDLE or CONNECTED), and/or UE reachability of the selected UEs. e.g. the current RM, CM and/or RRC state of the selected UEs Service operation: Namf_EventExposure service Nudm_EventExposure_Subscribe service Filter: a list of GPSI(s) or SUPI(s) Event ID: Reachability Filter Connectivity state changes (IDLE or CONNECTED), UE reachability status
  • the AF may derive the required UE Reachability/ UE status, e.g. based on the requirements of the service, policies etc. And then the AF may indicate the required UE Reachability/ UE status for member/ UE(s) selection.
  • the required UE Reachability/ UE status could be indicated by the AF by invoking Nnef_UEMemberSelectionAssistance as a part of filter/ filtering criteria information.
  • the NEF executes the corresponding service operation based on the UE member filtering criteria provided by the AF, e.g. UE Reachability/ UE status, events, analytics and/or notifications. NEF interacts with different 5GC network functions to collect the required information.
  • NEF The set of interactions between NEF and among 5GC NFs are dependent on the UE member filtering criteria provided by the AF. Based on the collected information from other 5GC NFs, NEF consolidates all the information collected from other 5GC NFs to derive the list(s) of candidate UE(s) which fulfill the UE member filtering criteria in the AF request, including the UE Reachability/ UE status. NEF sends a Nnef_UEMemberSelectionAssistance_Notify request to the AF including the list(s) of candidate UE(s) and possibly additional information.
  • the NEF may also subscribe to the AMF of the UE Reachability/ UE status. Once the UE status is changed, the AMF may send notification to the NEF and may along with the UE identifier and the new status. The NEF may update the member/ UE(s) list correspondingly and inform the new UE list to the AF.
  • multiple rounds of UL and DL data transmission might be needed, e.g. multiple round of iterations for model training between the UEs and the AIML application.
  • asynchronous mode operation where the application server can aggregate models and provide a global model update as soon as a new model update arrives.
  • hybrid modes are possible, e.g. application server would wait for selected group of UEs sharing a set of local models/model updates but not wait for other UEs sharing their models/model updates before the aggregation of models (or models updates) into a global model.
  • Jitter/ latency variation + latency actual UE latency
  • the AF may setup the different UE jitter/ latency variation for member/ UE(s) selection.
  • the AF transmits the corresponding requirements to NEF to assistant the member/ UE(s) selection.
  • the filer/ filtering criteria as shown in Table 2.
  • the filter/ filtering criteria of the member/ UE(s) selection may include the maximum allowed/ threshold of UE jitter/ latency variation for federated learning that will be used NEF to select UEs.
  • the latency variation indicated by the filtering criteria could be interpreted as the latency variation that the selected UEs should meet and/or not exceed.
  • the UEs with relative high variation of the latency/ End-to-end data volume transfer time, or the UEs that exceed or cannot match the require variation will/may not be recommended by the NEF to the AF for the corresponding AIML services.
  • Table 2 provides a description of UE Member filter/ filtering criteria.
  • the AF may also have the requirements of the latency/ latency variation/ jitter performance of a group of UEs.
  • the AF have requirements on the Ratio of UEs of some E2E data volume transfer time class, e.g. the Ratio of UEs of low latency class should be lower than a threshold, the Ratio of UEs of low latency class should be higher than a threshold.
  • the application server needs to wait for all local updates from the UEs before the aggregation takes place.
  • the slowest UE(s) will determine the performance of the federated learning.
  • the jitter or the latency variation of the member/ UE(s)s may also determine the overall performance of the federated learning. If the Jitter/ latency variation of the member/ UE(s)s, especially the slow UE are significant, the network may wait for longer time for the local updates, which will degrade federated learning performance.
  • the AF may have a relatively restructure requirements UE jitter/ latency variation to maintain the performance of the federated learning.
  • the AF may setup the filter of the ratios of the UE in the high latency group to a very low number or to 0. And send the filtering criteria to the NEF. Using the filtering criteria, the NEF may abandon the slow UEs from the candidate list, and only choose the UE group whose performance can satisfy the filtering criteria.
  • the application server can aggregate models and provide a global model update as soon as a new model update arrives or a set of local models/model updates are received.
  • the AF may have a relatively relaxing requirements UE jitter/ latency variation. In this case, if the jitter/ latency variation of a group of UEs are relatively small, the AF may be able to receive the local model from higher number of UEs within a short period. By aggregating those model from UEs, the AF can achieve better performance of model update or training, compared to using a small amount of uploaded models.
  • the AF may also have requirements of ratios of the UE in one or more than more transfer time classes.
  • the AF can set the range of UEs ratios in transfer time class, e.g. 10 - 20% of UEs in high latency class, 30 - 40% UEs in low latency class, and other UEs in other class.
  • the AF may also set the minimum or maximum / threshold Ratio of UEs in one or more than more transfer time classes.
  • the filtering criteria representing the variation of the latency/ End-to-end data volume transfer time of a UEs or a list of UEs could work in conjunction with the End-to-end data volume transfer time that has been specified in Table 4.15.13.2-1 of TS 23.502 V18.1.1.
  • latency variation + latency actual UE latency (latency could be replaced by End-to-end data volume transfer time in this disclosure).
  • the variation of the End-to-end data volume transfer time could be interpreted as the maximum variation of the transmission time (jitter) among the selected UEs, or the group of the UEs.
  • the UEs are indicated by the AF in the member selection request,
  • the NEF can select the UEs that fulfil both criteria.
  • the NEF may notify that the AF, optionally with the reason for that this UE should not be recommended.
  • the variation of End-to-end data volume transfer time and End-to-end data volume transfer time might be included in the same filter/ filtering criteria from the AF.
  • the NEF behaves the same as indicating the variation of End-to-end data volume transfer time and End-to-end data volume transfer time by two separated filtering criteria.
  • UE Member filtering criteria Description of filtering criterion UE filtering information Detailed description clause Flexibility of federated learning or Variation / Jitter of UE End-to-end data volume transfer time/ Variation of End-to-end data volume transfer time Indicate the end-to-end data volume transfer time that refers to the variation of the time for completing the transmission of a specific data volume between UE to AF E.g., the maximum allowed/ threshold of UE jitter/ latency variation for federated learning.
  • Performance/ latency performance of member/ UE(s) group/ a group of UEs Indicate the criteria of the latency performance of a group of UEs May include the criteria of jitter/ latency variation/ of a group of UEs, and/or the ratio of UEs in the defined transfer time/ latency class(es). E.g., the maximum and/or minimum allowed of UE ratios in the one or more latency classes; or the threshold of UE ratios in the each latency class, the requirements could be above or below the threshold.
  • Service operation Nsmf_EventExposure Nupf_EventExposure Filter: jitter or UE latency variation, ratios of UEs in different transfer time class related filter;
  • the NEF may collect data related to jitter/ UE latency variation from UPF (if the NEF is allowed to collected data from UPF) or SMF.
  • the SMF collected the required data from UPF via N4 session.
  • the NEF may collect the data related to jitter/ UE latency variation from NWDAF, e.g. using the output of End-to-end data volume transfer time analytics: E2E data volume transfer time UL/DL of a Data volume; using the Ratio of UEs per E2E data volume transfer time class to choose the UE group. Some of the outputs are shown in Table 3. If the ratio cannot satisfy the requirement, the NEF may remove some UEs from the UE group based on the UE ID(s), to satisfy the requirements Then the NEF may require the NWDAF to run the End-to-end data volume transfer time analytics again to update the results. If the new output of the analytics can satisfy the filter criteria, the NEF choose this UE group for federated learning. Table 3 provides E2E data volume transfer time statistics in TS 23.288.
  • Data volume UL Indicates uplink data volume used to derive E2E data volume transfer time UL
  • Data volume DL Indicates downlink data volume used to derive E2E data volume transfer time DL
  • UE ID(s) Identifies the UE(s) in the transfer time class with respect to the threshold of the corresponding transfer time class.
  • An AF may invoke Nnef_UEMemberSelectionAssistance_Subscribe service operation with end-to-end data volume transfer time related filtering criteria for receiving a list of UEs that match or not exceed such criteria.
  • the AF may also include in the request:
  • End-to-end data volume transfer time filtering criteria including the end-to-end data volume transfer time for a specific data volume between UE and AF and/or the variation of the transfer time.
  • an Area of Interest location area of the candidate UEs.
  • Optional time windows for selecting the candidate UEs start time and stop time.
  • the AF provides an end-to-end data volume transfer time related filtering criteria, including end-to-end data volume transfer time and/or the variation of the transfer time associated to location for the duration of the subscription, e.g., AoI, S-NSSAI, time window, Application ID.
  • end-to-end data volume transfer time and the may be provided to assist the UE member recommendation for a AIML service in a particular location.
  • Figure 1a provides an example of 5GC assistance to UE member selection for end-to-end data volume transfer time related filtering criteria. The description below relates to Figure 1a.
  • Nnef_UEMemberSelectionAssistance_subscribe request including the Application ID, DNN/S NSSAI, AoI, and the end-to-end data volume transfer time related filtering criteria including the end-to-end data volume transfer time and/or the variation of the transfer time.
  • NEF verifies the authorization of the AF Request and identifies which information needs to be collected and executed based on the with end-to-end data volume transfer time related filtering criteria provided by the AF.
  • the NEF derives the S-NSSAI, DNN and DNAI(s) to which this Application ID have access. NEF discovers querying UDM and NRF the SMFs that are deployed in the Area of Interest. The NEF may also restrict the discovery to those SMFs that serve some S-NSSAI and DNN combination (this S-NSSAI and DNN combination has been derived from the Application ID).
  • the NWDAF collects data from multiple resources for end-to-end data volume transfer time analytics as specified in clause 6.18.4 of TS 23.288.
  • the NWDAF provides the required analytics, e.g. end-to-end data volume transfer times and/or the variations of the transfer time for one or more volumes of data to the consumer NF by means of either Nnwdaf_AnalyticsInfo_Request response or Nnwdaf_AnalyticsSubscription_Notify, depending on the service used in step 4.
  • the NEF consolidates results and derives the list(s) of candidate UE(s) that fulfil the filtering criteria requested by the AF.
  • the NEF may use the end-to-end data volume transfer time of specific volumes of data and/or the variations to derive the list(s) of candidate UEs that can match or do not exceed the requirements from the AF.
  • NEF sends a Nnef_ UEMemberSelectionAssistance_Notify request to the AF including the list(s) of candidate UE(s) and additional information.
  • the specific UE(s) may have abnormal behaviours, e.g. being misused or hijacked. If the member/ UE(s)(s) of AIML operation behaves abnormally, e.g. Unexpected UE location, Unexpected long-live/large rate flows, the performance of the AIML operation will be degraded. For example, if the UE has Unexpected long-live/large rate flows, the UE may be not able to upload the model to the AF as required, e.g. the UE's performance may not satisfy the QoS or transmission time requirement. In this case, the AF may not be able to receive the model uploaded by this UE within the expected table or cannot receive the model from the UE at all. As a result, for synchronise FL, the AF cannot update or aggregated the model, as the AF cannot implement the model training/ update before receive the models from all the UEs.
  • the member/ UE(s)(s) of AIML operation behaves abnormally, e.g. Unexpected
  • the AF or NEF should also avoid to choose the UEs that might be hijacked into the member/ UE(s) group. If the UE is hijacked, the UE access services frequently. In this case, for asynchronous / hybrid FL, the AF may update the model frequently which is not needed at all. This will increase the load of both the network and the AF and may collapse the AIML operation. Therefore, to maintain the AIML service quality, the AF or NEF may setup the UE condition/behaviour as a filter/ filtering criteria for member selection. The UEs that cannot fulfil the requirements of AF for this AIML service, should be not be recommended by the NEF with deploying the Assistance for UE member selection functionality. Table 4 provides potential UE abnormal behaviours (Table 6.7.5.1-1: Relation between expected analytics type and Exception IDs in TS 23.288)
  • Table 4a provides potential UE abnormal behaviours (Table 6.7.5.1-2: Description of Expected UE Behaviour parameters per Exception ID)
  • UE abnormal behaviours related information as a filter/ filtering criteria for member selection will help the NEF or AF to avoid involving the UE that may behave abnormally to the member/ UE(s) groups, and therefore, to further avoid the performance of the AIML operation might be degraded due to UE abnormal behaviours.
  • the AF may include the UE Abnormal behavior/or and Expected UE behavioural parameters related network data analytics or as a filter/ filtering criteria for member selection in the request of member/ UE(s) selection.
  • the NEF may use the filter to choose the UE for the initial member selection.
  • the NEF may also subscribes to the NFs that could provide the UE Abnormal behavior and or Expected UE behaviors information. If the behavior of UE(s) in the candidate list or the chosen UE list cannot satisfy the requirement of AIML operation anymore, the NEF or AF may update the member/ UE(s) group with removing the UE(s) whose behavior cannot fulfill the filtering criteria.
  • the NEF may collect the abnormal behavior and/or Expected UE behavioral related parameters:
  • the NEF or AF may indicate the filter to NWDAF for the UE Abnormal/ expected behaviour for assisting AIML operation.
  • the analytics are subscribed by the NEF for the 5GC assistance member UE selection, the NEF subscribes to the NWDAF analytics based on the requirements/ filtering criteria from AF.
  • the AF may include the one or more Exception IDs matching the expected analytics type in Table 4 and also the corresponding Exception Level, Exception trend, Ratio, Amount etc. If the UE cannot fulfil the requirement of the UE Abnormal behaviours, the UE will be discarded from the candidate UEs or will be not recommended by the NEF to the AF.
  • the Exception Level is a Scalar value indicating the severity of the abnormal behaviour.
  • the Exception Level of an Exception ID is set to M, based on the output of the NWDAF, the UE's Exception Level of this Exception ID is higher than M, this UE will be discarded. Or in other example, the Exception trend of the UE corresponds to an exception ID is up, which is against the filtering criteria, this UE will be discarded.
  • the Description of Expected UE Behaviour parameters per Exception are in Table 4a.
  • the Exception Level of an Exception ID is set to N, e.g. the Exception ID could be Ping-ponging across neighbouring cells, the corresponding UE behaviour parameters could be Expected UE Moving Trajectory and/or Stationary Indication.
  • the NEF determines the UEs that fulfill the requirements in filtering criteria and other information will be recommended to the AF for AIML operation, and send the recommended UEs to the AF via Nnef_UEMemberSelectionAssistance_Notify service operation;
  • the NEF determines the UEs that cannot fulfill the requirements in filtering criteria and other information will NOT be recommended to the AF for AIML operation, and send the NOT recommended UEs to the AF via Nnef_UEMemberSelectionAssistance_ NotifyDiscard/Delete/Remove service.
  • the expected UE behaviour parameters, expected analytics type or list of Exception IDs with associated thresholds for the Exception Level, where the expected analytics type can be mobility related, communication related or both etc. might be sent to the NWDAF by NEF or AF to choose the candidate UEs for FL operation.
  • TS 23.502 The Expected UE Behaviour parameters stored as AMF-Associated Expected UE Behaviour parameters which is per UE level and SMF-Associated Expected UE Behaviour parameters which is per PDU session level in UDM.
  • AMF retrieves the AMF-Associated Expected UE Behaviour parameters from UDM which may related to both PDU session(s) and SMS transmission.
  • SMF retrieves the SMF-Associated Expected UE Behaviour parameters from UDM for the specific PDU session.
  • the NEF may also indicate the time window the data resources to indicate Expected UE Behaviour shall have an associated validity time.
  • the AMF, SMF, UDM send the Expected UE Behaviour parameters to NEF based on the request, optionally Expected UE Behaviour shall have an associated validity time, confidence level and the accuracy level indicates.
  • the NEF determines the UEs that fulfill the requirements in filtering criteria and other information will be recommended to the AF for AIML operation, and send the recommended UEs to the AF via Nnef_UEMemberSelectionAssistance_Notify service operation;
  • the NEF determines the UEs that cannot fulfill the requirements in filtering criteria and other information will NOT be recommended to the AF for AIML operation, and send the NOT recommended UEs to the AF via Nnef_UEMemberSelectionAssistance_ NotifyDiscard/Delete/Remove service.
  • Table 4b provide a description of Expected UE Behaviour parameters.
  • the NEF may discard the UE(s) that cannot fulfil the filtering criteria of UE Abnormal behaviour.
  • the AF may indicate the lowest acceptable level of level of UE Abnormal behaviour, if the level of the corresponding event of this UE given by the NWDAF is higher than the threshold, the NEF will discard this UE.
  • the NEF may also notify the AF of the discarding UE via Nnef_UEMemberSelectionAssistance_Notify service operation or Nnef_UEMemberSelectionAssistance_Discard/Delete/Remove service operation.
  • Table 5 provides a description of UE Member filer/ filtering criteria.
  • UE Member filtering criteria Description of filtering criterion UE filtering information Detailed description clause UE Abnormal behaviour Indicate the Exceptions UE Abnormal behaviour, e.g. if the Exceptions level of some Exceptions IDs are higher than the filtering criteria, this UE will be discarded; or if the Exceptions trend cannot satisfy the filtering criteria, the UE will be discarded etc..
  • the network load may need to be considered for the member/ UE(s) selection.
  • the AF, or the NEF or any other network function that may perform, instruct or be involved into member/ UE(s) selection procedure need to consider the NF load, the load of the gNB etc. of the candidate UEs. If the serving NF(s) and/or gNB(s) of the candidate UE are overloaded, the performance of model training or inference will be degraded.
  • NF load NF load
  • Slice load gNB load etc.
  • the AF might indicate the NEF of the filter/ filtering criteria for member selection, the NEF will choose the UEs for federated learning if the filtering criteria could be satisfied.
  • the NF load, Slice load, gNB load could be collected by the NEF from NWDAF, OAM, AF, UPF or other 5GC NFs.
  • the statics and prediction of the NF load, slice load, gNB status information could be provided by NWDAF.
  • the NWDAF is able to provide the statics and prediction of the NF load, slice load and gNB status information from the following analytics: NF load analytics, Network Performance Analytics, Slice load level related network data analytics etc.
  • Some of the above analytics could be provided for a single UE (e.g. SUPI), or a group of UEs or any UE among the candidate member/ UE(s)s indicated by the AF.
  • the NEF may invoke the Nnwdaf_AnalyticsInfo or Nnwdaf_AnalyticsSubscription service to the NWDAF, the Analytics ID could set to NF load information, Network Performance, Load level information etc. as shown in Table 6.
  • the AF may provide the filter/ filtering criteria of UE Member.
  • the AF may indicate the maximum allowed load of NF(s) or gNBs.
  • the the load/status/ information of the network NF or nodes of the corresponding UE(s) should not exceed the maximum/ need to satisfy the requirements.
  • the maximum allowed load might be applied within a time window, the time window might be indicated by the AF, or derived by other NF(s) (e.g. PCF) and then indicate to the NEF.
  • the NEF collects data from NWDAF, OAM, NRF, UPF or any other NFs. If one or more of the following could satisfy the requirements, the corresponding UEs could be potentially chosen: the real time data, or historical data, or statics, or predilection the historical.
  • the gNB load or status could also be collected by the NEF from the OAM.
  • the RAN status (up/down), load (i.e. Radio Resource Utilization) and performance per Cell Id for the traffic type of interest and in the Area of Interest could be collected from the OAM as defined in TS 28.552.
  • the NFs, OAMs, AFs that provide the above network load/status information may notify the NEF if the load/status changed/exceed the allowance.
  • Table 6 provides a description of UE Member filer/ filtering criteria.
  • UE Member filtering criteria Description of filtering criterion UE filtering information Detailed description clause Network load/ status Indicate the load/status of the network NF or node that satisfy the filtering criteria.
  • Service operation Nnwdaf_AnalyticsInfo or Nnwdaf_AnalyticsSubscription.
  • Nnrf_NFManagement_NFStatusSubscribe Filter: Analytics ID NF load information, Network Performance, Load level information etc.
  • the UE current location it indicates the certain area that the selected UE currently located in. and the service deployed to collect the UE current location by the NEF is Namf_EventExposure with including the list of GPSI(s) or SUPI(s) of the corresponding UEs as the filter of this AMF service.
  • the AMF can only provide the UE current location in relatively coarse granularities, e.g. cell or TA level granularity.
  • relatively coarse granularities e.g. cell or TA level granularity.
  • the UE currently location as a filter/ filtering criteria for member selection, the UE location higher than the Cell-ID level cannot be provided.
  • the higher granularity of the UE location might be need as a filter/ filtering criteria for member selection.
  • the AF may require the NEF to filter the UEs using higher UE location, e.g. the TRP or beam level UE location information.
  • the higher granularity UE current location information may assist the AF which provide the federated learning in a more efficient way.
  • the higher granularity UE current location might be also required by QoS requirements/ attribute.
  • the NEF can collect the UE location data from GMLC via invoking the GMLC service, e.g. Ngmlc_Location_ProvideLocation.
  • the NEF may send the Ngmlc_Location_ProvideLocation_Request to the GMLC.
  • the GMLC responds the request from NEF and send the UE location to NEF due to the requirement in the request, e.g. periodic event type, area event type, motion event type etc., via Ngmlc_Location_ProvideLocation_response or Ngmlc_Location_LocationUpdateNotify etc.
  • the NEF may also request the GMLC to notify NEF the updated UE location or event when the notify procedure is triggered.
  • the NEF may require the notify from GMLC by in invoking LocationUpdateNotify or EventNotify services.
  • the NEF may also require to cancel the subscription to GLMC via CancellLocation service for periodic or triggered location reporting.
  • the Event types of GLMC service include UE available, change of area, motion or periodic location, applicable to deferred location requests only etc. Different event types could be setup for the GLMC service.
  • the types of event could be setup as (change of) area. This event indicates that where the UE enters, leaves or remains within a pre-defined geographical area. At least one type of area event can be defined (i.e. entering, leaving or remaining within the area).
  • the LCS (LoCation Services) client or AF or the service consumer may define the target area as a geographical area or as a geopolitical name of an area.
  • the GMLC will send notify/ response to the consume with other information
  • the information may include one or more of the following: UE ID(s) (GPSI, SUPI, Internal Group Identifier or External Group Identifier), Geodetic location, Local Location including Coordinate ID, civic location, age of location, etc.
  • Table 7 The corresponding services used by NEF to collected UE location from GLMC is documented in Table 7.
  • Table provides a description of UE Member filtering criteria.
  • UE Member filtering criteria Description of filtering criterion UE filtering information Detailed description clause UE current location Indicate the certain area that the selected UE currently located in.
  • Service operation Namf_EventExposure Ngmlc_Location service, e.g. Ngmlc_Location_ProvideLocation, Ngmlc_Location_LocationUpdateNotify ect.
  • Service operation Filter a list of GPSI(s) or SUPI(s), or a GPSI/ SUPI. 4.15.Y.3
  • the Ngmlc_Location service include different service operations, as illustrated in Table 8.
  • Table 8 provides a List of GMLC Services (Table 8.4.1-1 in TS 23.273).
  • AF requests 5GS assistance to support the UE selection by considering the UE's historical location, UE's current locationand direction. AF includes the UE lists and the following criteria as part of the FL UE selection request:
  • ⁇ UE historical location The Target AoI where the UEs have been roving over the historical nomadic period before moving into the FL coverage area.
  • ⁇ UE current location The current AoI which is the coverage area of the FL training server where the selected UEs located in to participate in the FL operation
  • ⁇ Direction Select the UE with the different direction in the FL coverage Area.
  • the AF may provide sub-areas, and provide a maximum number of UEs that should take part in FL from each sub-area.
  • NEF translates the GPSIs to SUPIs and maps the filtering criteria into the corresponding UE filtering information.
  • AMF will provide a list of UEs that are within the current AoI to the NEF using the Namf_EventExposure_Notify service operation.
  • the NEF obtains the list of possible targets UEs from AMF within the current AoI.
  • GMLC will provide UE ID(s) as to the NEF using the Ngmlc_Location_ProvideLocation (or Ngmlc_Location_EventNotify service operation, Ngmlc_Location_LocationUpdateNotify).
  • the NEF obtains the list of possible targets UEs from AMF within the current AoI
  • the NEF can determine the FL candidate UEs which are now within the FL coverage area but were roving within the target AOI over the historical nomadic period and UE with the different direction as requested by AF.
  • the NEF notifies AF for such UE candidate list.
  • step 10 identifies any UE which is moving out of the FL coverage area, the NEF may further notify AF of the given UE which is moving out the coverage.
  • step 10 If step 10 identifies any UE which is moving out of the FL coverage area, the NEF may further notify AF of the given UE which is moving out the coverage.
  • the NEF may collect data from different sources/ providers, e.g. the AMF, SMF, NWDAF etc., as required by the filters indicated by the AF for member/ UE(s) selection.
  • the NEF may also subscribes to the data providers with setting up filter /criteria/events for reporting/ notification. Once the filter /criteria/event is triggered, the data/service provider may update the latest the result to the NEF. For example,
  • the consumer of the member selection service e.g. AF, triggers the member/ UE(s) update procedures.
  • AF the consumer of the member selection service in this embodiment.
  • the 5GC NF or the AF who performs the member selection function e.g. the NEF, triggers the member/ UE(s) update procedures.
  • NEF an example of the 5GC NF or the AF who performs the member selection function in this embodiment.
  • the corresponding AF triggered the member/ UE(s) update procedures service is in member/ UE update service of this disclosure.
  • the AF may need to update the member/ UE(s)s during the AIML operation. For example, the AF modify the AIML model size, number of layers, updates the training data set, more or less or new member/ UE(s)s might be required to participate the service. Another example, some of the UEs may leave the AIML operation due to UE mobility, variation of link quality, QoS of the selected UEs cannot fulfil the requirement of the AIML services any longer, cannot perform the AIML operation as required etc..
  • the AF may need more UEs to participate in the AIML operation to achieve the expected service quality, or the AF may want to evaluate whether the UEs in the AIML operation can still fulfil requirements, or the AF would like to exclude/ deselect some existing UEs from the AIML operation etc..
  • the AF may require the NEF to perform the member/ UE(s) update procedures by invoking Nnef_UEMemberSelectionAssistance service operation, e.g. Nnef_UEMemberSelectionAssistance_update or Nnef_UEMemberSelectionAssistance_Subscribe.
  • Nnef_UEMemberSelectionAssistance_Subscribe the procedures are same as using a new service. Furthermore, for the member UE update, upon receiving the request from AF, the NEF correlates the update request to an existing subscription according to one or more of the following information in Nnef_UEMemberSelectionAssistance_Subscribe request to correlate the update request to an existing subscription/ event for the AIML service:
  • An explicit indication for update request e.g. member selection update etc.
  • the AF may include the following information in the update request (either Nnef_UEMemberSelectionAssistance_update or Nnef_UEMemberSelectionAssistance_Subscribe) (some of the information could be optional):
  • the NEF will use those ID to identify the update request corresponds to which AIML operation/ service or an existing subscription.
  • the UE list could be one or more of the following:
  • the AF may include the new UEs in the request. Those UEs are the potential candidates for the corresponding services but those UEs have not participate this service before.
  • the AF may would like to require the NEF to evaluate the performance of some old UEs based on the filtering criteria or requirements of the AIML service.
  • the AF may determine not to indicate the filtering criteria/ filter/ requirements to the NEF.
  • the NEF reuse the existing filtering criteria/ filter/ requirements for this AIML service.
  • the NEF may receive the filtering criteria/ filter/ requirements from the AF via Nnef_UEMemberSelectionAssistance_Subscribe or was indicated by the AF before this member/ UE(s) update procedure;
  • the AF may indicate the new filtering criteria/ filter/ requirements to the NEF, together with filtering criteria ID or other information that could assist the NEF to determine the update information corresponds to which existing filter;
  • the AF may indicate new filtering criteria/ filter/ requirements to the NEF. If the AF decided to deploy new filtering criteria/ filter/ requirements, the AF will include those information into the member/ UE(s) update procedure. The NEF will use the new filtering criteria/ filter/ requirements for member selection.
  • - AF may indicate all the new filtering criteria/ filter/ requirements related to this AIML service.
  • the NEF will flush the existing filtering criteria/ filter/ requirements, only use the information indicated by the AF in the member update request to perform the member/ UE(s) selection function.
  • Notification Target Address (+ Notification Correlation ID), Expiry time, time window(s) for selecting the candidate UEs, specific parameters depending on the UE member filtering criteria.etc.
  • the NEF will perform the member/ UE(s) selection update as instructed by the AF. And notify the updated UE list to the AF, e.g. by invoking Nnef_UEMemberSelectionAssistance_Notify or Nnef_UEMemberSelectionAssistance_Discard service operation.
  • the NEF may trigger to update the member/ UE(s)s during the AIML operation, e.g. some of the UE performance or the network performance cannot satisfy the service requirement/ filtering criteria based on the data collected by NEF from different 5GC NFs, AFs or OAM etc.
  • the member/UE update procedures might be trigger periodically.
  • the period/ the timer of the member/UE update might be indicated by the consumer, e.g. the AF, to the NEF.
  • the AF may also configure/ recommend the triggering condition of the member/UE update procedures to the NEF, e.g. a certain number of UEs cannot fulfil the requirement/ filtering criteria, a UE cannot fulfil one or more requirement/ filtering criteria for a certain number of time within a window etc.
  • the NF consumer requests the network to update the UEs of the list of services.
  • Inputs Optional: filtering criteria shown in Table 4.15.13.2-1, Application ID, Subscription Correlation ID (in the case of modification of the existing subscription), Expiry time, a set of UE member filtering criteria shown in the Table 4.15.13.2-1, time window(s) for selecting the candidate UEs, specific parameters depending on the UE member filtering criteria, Periodicity
  • discard the ue/member could mean one or more of the following: NEF or any NF performs the member/ UE
  • the NEF may (recommend to) remove/discard the UE from the potential candidate list from AF, or from the list of the UEs that have been selected by the NEF previously, or remove the UE from the service.
  • This service could be deployed during the initial or during the update of UE/member selection procedures.
  • the NEF determines the UE to be removed/discarded, e.g. based on requirement/ filtering criteria, and then informs the AF of those UE.
  • the NEF informs the AF of the reasons for discarding the UEs, e.g. UE performance cannot fulfil the filter/ filtering criteria for UE/member selection of a service, along with the filtering criteria ID.
  • the NEF deselect the UE/member and then inform the result to the AF, e.g. by involving the existing Nnef_UEMemberSelectionAssistance_Notify service operation or a another service Nnef_UEMemberSelectionAssistance_ Discard/ deselect/ remove service operation
  • the NEF may indicate either the recommended list(s) of selected or list(s) of deselected UEs, or recommended list(s) of selected and list(s) of deselected UEs.
  • the list(s) of selected or deselected could be optional.
  • the member/ UE update is triggered by NEF, only recommended list(s) of deselected UEs; if the member/ UE update is triggered by AF, only recommended list(s) of selected UEs are included; if Periodicity. Or any other conditions to determine whether the recommended list(s) of selected and/or list(s) of deselected UEs are included.
  • the Nnef_UEMemberSelectionAssistance_Notify service operation may include an indication to inform whether the list of UEs are recommended or not recommended to the AF. There are some possibilities:
  • the AF Upon receiving the list, the AF will determine which UEs are recommended by the NEF for the corresponding service, and which UEs are not recommended.
  • NEF reports the UE member selection assistance information to the consumer that has previously subscribed.
  • Notification Correlation Information either one or more list(s) of recommended selected UE(s) or one or more list(s) of recommended deselected UE(s).
  • Inputs Optional: Periodicity.
  • Nnef_UEMemberSelectionAssistance_Discard service operation (as an example).
  • NEF reports to the consumer of the UEs to be discarded from the list of the UEs that has previously selected. / NEF reports the UEs that are not recommended according to the requirements in the subscription to the consumer.
  • Notification Correlation Information one or more list(s) of recommended deselected UE(s).
  • Notification Correlation Information one or more list(s) of UE(s) that are not recommended according to the requirements indicated by the consumer.
  • Inputs Optional: reasons for discarding the corresponding UE(s) / cause of discarding the corresponding UE(s).
  • Nnef_UEMemberSelectionAssistance_Notify has been specified in 5.2.6.32.4 of TS 23.502. Periodicity or other relevant information could be included into the service to the consumer. The information is used to inform the consumer that the notify service is trigger by periodically member update procedure, or the time point when the member selection was made, or the time period the recommended UEs will be valid for etc.
  • NEF reports the UE member selection assistance information to the consumer that has previously subscribed.
  • Inputs Optional: Periodicity.
  • Nnef_UEMemberSelectionAssistance_Subscribe has been specified in 5.2.6.32.2 of TS 23.502. in order to inform the consumer the update requirements, the Periodicity (the periodicity of member update) could be included as the input to the consumer. If the AF determines to ask the consumer to update the UE/member selection periodically, the consumer use the Periodicity to perform the member/UE reselect/update procedure.
  • the NF consumer subscribes to receive the UE member selection assistance information, or the subscription is updated if the same subscription is already defined in NEF.
  • Inputs Required: Target of Member Selection Assistance Reporting (GPSI or a list of GPSIs), Notification Target Address (+ Notification Correlation ID), at least one filtering criteria shown in Table 4.15.13.2-1.
  • Inputs Optional: Application ID, Subscription Correlation ID (in the case of modification of the existing subscription), Expiry time, a set of UE member filtering criteria shown in the Table 4.15.13.2-1, time window(s) for selecting the candidate UEs, specific parameters depending on the UE member filtering criteria , Periodicity (the periodicity of member update).
  • the update of the members or filtering criteria might be required to ensure the members participating in the operation can fulfil requirements of the AIML service. For example, some UEs leave the FL due to mobility; therefore, the AF may require new UEs to join the FL operation. In another example, the AF may update the filtering criteria to improve the service quality, e.g. the AF may update the thresholds of the QoS requirements or E2E data volume transfer time to select the UEs can run the AIML service more efficiently. However, according to the current specified procedures and services, it is not clear how to update the members by the NEF.
  • the NEF may use the existing UE list and filtering criteria for initial member selection.
  • the AF include the UE list or the filtering criteria to be updated into Nnef_UEMemberSelectionAssistance_Subscribe.
  • the NEF correlates the update request to existing subscription according to Subscription Correlation ID and uses the updated and existing information for data collection and consolidation.
  • the number of UEs that cannot fulfil the filtering criteria is much less than those can satisfy that. Therefore, notifying the AF of the UEs that are not recommended can reduce the signaling size significantly. Therefore, the NEF should be allowed to send either the UEs that are recommended or the UEs that are not recommended for the FL service by invoking existing Nnef_UEMemberSelectionAssistance_Notify or Nnef_UEMemberSelectionAssistance_NotifyDiscard, respectively.
  • Figure 1c provides an example of 5GC assistance to UE member selection and update. The description below relates to Figure 1c.
  • AF subscribes the member selection assistance functionality by sending Nnef_UEMemberSelectionAssistance_subscribe request including a list of target UE(s), one or more UE member filtering criteria listed in the Table 4.15.13.2-1, and optionally, time window(s).
  • NEF verifies the authorization of the AF Request and identifies which information needs to be collected and executes the corresponding service operation based on the UE member filtering criteria provided by the AF, e.g. events, analytics and/or notifications.
  • NEF interacts with different 5GC network functions to collect the required information.
  • the set of interactions between NEF and among 5GC NFs are dependent on the UE member filtering criteria provided by the AF. See Table 4.15.13.2-1 for details.
  • NEF Based on the collected information from other 5GC NFs, NEF consolidates all the information collected from other 5GC NFs to derive the list(s) of candidate UE(s) which fulfil the UE member filtering criteria in the AF request.
  • NEF sends a Nnef_UEMemberSelectionAssistance_Notify request to the AF including the list(s) of candidate UE(s) and possibly additional information.
  • An update of the list(s) of candidate UE(s) may be performed periodically or requested by the AF via Nnef_UEMemberSelectionAssistance_Subscribe request including Subscription Correlation ID, and optionally, a list of target UE(s) to be updated, one or more UE member filtering criteria to be updated as listed in the Table 4.15.13.2-1, and time window(s) etc..
  • the NEF correlates the Nnef_UEMemberSelectionAssistance_Subscribe request to an existing subscription according to the Subscription Correlation ID. If the AF does not indicate a list of target UE(s) to the NEF in the request in Step 6, the NEF uses the target UEs received in Step 1 for the member UE update using the updated filtering criteria.
  • NEF sends a Nnef_UEMemberSelectionAssistance_Notify request and/or Nnef_UEMemberSelectionAssistance_NotifyDiscard request to the AF including the list(s) of the candidate UEs and/or list(s) of UEs that are not recommended, and possibly additional information.
  • Figure 2 shows a general procedure of 5GC assistance to UE/Member selection and update.
  • AF subscribes the member selection assistance functionality by sending Nnef_UEMemberSelectionAssistance_subscribe request including a list of target UE(s), one or more UE member filtering criteria listed in the Table 4.15.13.2-1, and optionally, time window(s).
  • NEF verifies the authorization of the AF Request and identifies which information needs to be collected and executes the corresponding service operation based on the UE member filtering criteria provided by the AF, e.g. events, analytics and/or notifications.
  • NEF interacts with different 5GC network functions to collect the required information.
  • the set of interactions between NEF and among 5GC NFs are dependent on the UE member filtering criteria provided by the AF. See Table 4.15.13.2-1 of TS 23.502 for details.
  • NEF Based on the collected information from other 5GC NFs, NEF consolidates all the information collected from other 5GC NFs to derive the list(s) of candidate UE(s) which fulfill the UE member filtering criteria in the AF request.
  • NEF sends a Nnef_UEMemberSelectionAssistance_Notify request to the AF including the list(s) of candidate UE(s) and possibly additional information. Or NEF sends a Nnef_UEMemberSelectionAssistance_Discard to the AF including the list(s) of discarded UE from the list of target UE(s) in step 1.
  • the UE Member selection update might be triggered by the AF or the NEF.
  • the AF sends Nnef_UEMemberSelectionAssistance_Update request including a list of target UE(s), and optionally one or more UE member filtering criteria listed in the Table 4.15.13.2-1, and time window(s). If the AF does not indicate filtering criteria to the NEF, the NEF uses the filtering criteria in Step 2 for member/ UE(s) update.
  • the NEF uses the a list of target UE(s) received in step 1 and filtering criteria in Step 2 for member/ UE(s) update.
  • NEF interacts with different 5GC network functions to collect the required information.
  • the set of interactions between NEF and among 5GC NFs are dependent on the UE member filtering criteria provided by the AF. See Table 4.15.13.2-1 of TS 23.502 for details.
  • NEF Based on the collected information from other 5GC NFs, NEF consolidates all the information collected from other 5GC NFs to derive the list(s) of candidate and/or discarded UE(s) which fulfil and/or cannot fill the UE member filtering criteria in the AF request.
  • NEF sends a Nnef_UEMemberSelectionAssistance_Notify or Nnef_UEMemberSelectionAssistance_Discard request to the AF including the list(s) of candidate UE(s) or discarded UE, and possibly additional information.
  • the filter/ filtering criteria including all of the filter/ filtering criteria detailed in this disclosure and also another filter/ filtering criteria, e.g. current location, historical location and direction, QoS filter etc. .
  • Figure 3 shows a detailed specific procedure for the 5GC assistance to member selection and update based on the various filter/ filtering criteria
  • AF subscribes the member selection assistance functionality by sending Nnef_UEMemberSelectionAssistance_subscribe request including a list of target UE(s), one or more UE member filtering criteria listed in the Table 4.15.13.2-1, and optionally, time window(s).
  • NEF verifies the authorization of the AF Request and identifies which information needs to be collected and executes the corresponding service operation based on the UE member filtering criteria provided by the AF, e.g. events, analytics and/or notifications.
  • NEF interacts with different 5GC network functions to collect the required information.
  • the set of interactions between NEF and among 5GC NFs are dependent on the UE member filtering criteria provided by the AF. See Table 4.15.13.2-1 of TS 23.502 for details.
  • the Consumer NF e.g. NEF
  • the NEF also provides the input information as specified in TS23.288 of the corresponding required analytics.
  • the NEF subscribes the required from AMF using Namf_EventExposure_Subscribe service, e.g. for collecting UE location(s) for a UE or a group of UEs, the UE CM, RM, RRC states etc.
  • Namf_EventExposure_Subscribe service e.g. for collecting UE location(s) for a UE or a group of UEs, the UE CM, RM, RRC states etc.
  • the NEF subscribes to service data from SMF by invoking Nsmf_EventExposure_Subscribe (Event ID, SUPI(s) or Application ID).
  • the NEF subscribes to information of the UE and may subscribe to N4 Session related input data from SMFs as required in the filtering criteria from the AF.
  • N4 related input data is provided by UPF to SMF.
  • NWDAF replies to the NEF of the required outputs of the analytics.
  • the AMF notifies the NEF the required data.
  • NEF Based on the collected information from other 5GC NFs, NEF consolidates all the information collected from other 5GC NFs to derive the list(s) of candidate UE(s) which fulfill the UE member filtering criteria in the AF request.
  • NEF sends a Nnef_UEMemberSelectionAssistance_Notify request to the AF including the list(s) of candidate UE(s) and possibly additional information. Or NEF sends a Nnef_UEMemberSelectionAssistance_Discard to the AF including the list(s) of discarded UE from the list of target UE(s) in step 1.
  • the UE Member selection update might be triggered by the AF or the NEF.
  • the AF sends Nnef_UEMemberSelectionAssistance_Update request including a list of target UE(s), and optionally one or more UE member filtering criteria listed in the Table 4.15.13.2-1, and time window(s). If the AF does not indicate filtering criteria to the NEF, the NEF uses the filtering criteria in Step 2 for member/ UE(s) update.
  • the NEF uses the a list of target UE(s) received in step 1 and filtering criteria in Step 2 for member/ UE(s) update.
  • NEF interacts with different 5GC network functions to collect the required information.
  • the set of interactions between NEF and among 5GC NFs are dependent on the UE member filtering criteria provided by the AF. See Table 4.15.13.2-1 of TS 23.502 for details.
  • the NEF may deploy some of the service from 7a. to 7h.
  • NEF Based on the collected information from other 5GC NFs, NEF consolidates all the information collected from other 5GC NFs to derive the list(s) of candidate and/or discarded UE(s) which fulfil and/or cannot fill the UE member filtering criteria in the AF request.
  • NEF sends a Nnef_UEMemberSelectionAssistance_Notify or Nnef_UEMemberSelectionAssistance_Discard request to the AF including the list(s) of candidate UE(s) or discarded UE, and possibly additional information.
  • examples of the present disclosure may be realized in the form of hardware, software or a combination of hardware and software.
  • Certain examples of the present disclosure may provide a computer program comprising instructions or code which, when executed, implement a method, system and/or apparatus in accordance with any aspect, example and/or embodiment disclosed herein.
  • Certain embodiments of the present disclosure provide a machine-readable storage storing such a program.
  • FIG 4 is a block diagram of an exemplary network entity/function that may be used in examples of the present disclosure, such as the techniques disclosed in relation to any of the preceding figures.
  • any of the network entities, network function etc. may be provided in the form of the network entity illustrated in Figure 4.
  • a network entity/function may be implemented, for example, as a network element on a dedicated hardware, as a software instance running on a dedicated hardware, and/or as a virtualised function instantiated on an appropriate platform, e.g. on a cloud infrastructure.
  • the entity 400 comprises a processor (or controller) 401, a transmitter 403 and a receiver 405.
  • the receiver 405 is configured for receiving one or more messages from one or more other network entities, for example as described above.
  • the transmitter 403 is configured for transmitting one or more messages to one or more other network entities, for example as described above.
  • the processor 401 is configured for performing one or more operations, for example according to the operations as described above.
  • Such an apparatus and/or system may be configured to perform a method according to any aspect, embodiment or example disclosed herein.
  • Such an apparatus may comprise one or more elements, for example one or more of receivers, transmitters, transceivers, processors, controllers, modules, units, and the like, each element configured to perform one or more corresponding processes, operations and/or method steps for implementing the techniques described herein.
  • an operation/function of X may be performed by a module configured to perform X (or an X-module).
  • the one or more elements may be implemented in the form of hardware, software, or any combination of hardware and software.
  • examples of the present disclosure may be implemented in the form of hardware, software or any combination of hardware and software. Any such software may be stored in the form of volatile or non-volatile storage, for example a storage device like a ROM, whether erasable or rewritable or not, or in the form of memory such as, for example, RAM, memory chips, device or integrated circuits or on an optically or magnetically readable medium such as, for example, a CD, DVD, magnetic disk or magnetic tape or the like.
  • volatile or non-volatile storage for example a storage device like a ROM, whether erasable or rewritable or not
  • memory such as, for example, RAM, memory chips, device or integrated circuits or on an optically or magnetically readable medium such as, for example, a CD, DVD, magnetic disk or magnetic tape or the like.
  • the storage devices and storage media are embodiments of machine-readable storage that are suitable for storing a program or programs comprising instructions that, when executed, implement certain examples of the present disclosure. Accordingly, certain examples provide a program comprising code for implementing a method, apparatus or system according to any example, embodiment and/or aspect disclosed herein, and/or a machine-readable storage storing such a program. Still further, such programs may be conveyed electronically via any medium, for example a communication signal carried over a wired or wireless connection.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

La divulgation concerne un système de communication 5G ou 6G permettant de prendre en charge un débit supérieur de transmission de données. Sont divulgués un procédé de sélection d'équipement utilisateur (UE) pour une opération de service dans un système de communication sans fil 3 GPP. Le procédé consiste à : recevoir d'une fonction d'application (AF), au niveau d'une fonction d'exposition réseau (NEF), une demande Nnef_UEMemberSelectionAssrésistance_Subscribe comprenant un ID de corrélation d'abonnement et un ou plusieurs critères de filtrage ; corréler, au moyen de la NEF, le message Nnef_UEMemberSelectionAssrésistance_Subscribe avec un abonnement existant selon l'ID de corrélation d'abonnement ; collecter, au moyen de la NEF, à partir d'une ou de plusieurs fonctions réseau (NF), des informations pour chaque UE d'une liste d'UE membres cibles d'après les critères de filtrage ; dériver, au moyen de la NEF, une liste d'un ou de plusieurs UE candidats associés à l'abonnement existant à l'aide des informations collectées ; et transmettre, de la NEF à l'AF, un message Nnef_UEMemberSelectionAssrésistance_Notify comprenant la liste d'UE candidats.
PCT/KR2024/004312 2023-04-06 2024-04-03 Procédé et appareil de sélection d'un ue pour un apprentissage fédéré WO2024210500A1 (fr)

Applications Claiming Priority (8)

Application Number Priority Date Filing Date Title
GB2305232.7 2023-04-06
GBGB2305232.7A GB202305232D0 (en) 2023-04-06 2023-04-06 Ue selection for federated learning
GBGB2305852.2A GB202305852D0 (en) 2023-04-06 2023-04-20 Ue selection for federated learning
GB2305852.2 2023-04-20
GBGB2307142.6A GB202307142D0 (en) 2023-04-06 2023-05-12 UE selection for federated learning
GB2307142.6 2023-05-12
GBGB2403129.6A GB202403129D0 (en) 2023-04-06 2024-03-04 UE selection for federated learning
GB2403129.6 2024-03-04

Publications (1)

Publication Number Publication Date
WO2024210500A1 true WO2024210500A1 (fr) 2024-10-10

Family

ID=86378824

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2024/004312 WO2024210500A1 (fr) 2023-04-06 2024-04-03 Procédé et appareil de sélection d'un ue pour un apprentissage fédéré

Country Status (2)

Country Link
GB (4) GB202305232D0 (fr)
WO (1) WO2024210500A1 (fr)

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020071887A1 (fr) * 2018-10-05 2020-04-09 Samsung Electronics Co., Ltd. Procédé de fourniture de paramètres de service à un ue et à un réseau dans un système 5g

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2020071887A1 (fr) * 2018-10-05 2020-04-09 Samsung Electronics Co., Ltd. Procédé de fourniture de paramètres de service à un ue et à un réseau dans un système 5g

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
"3rd Generation Partnership Project; Technical Specification Group Services and System Aspects; Procedures for the 5G System (5GS); Stage 2 (Release 18)", 3GPP STANDARD; 3GPP TS 23.502, 3RD GENERATION PARTNERSHIP PROJECT (3GPP), MOBILE COMPETENCE CENTRE ; 650, ROUTE DES LUCIOLES ; F-06921 SOPHIA-ANTIPOLIS CEDEX ; FRANCE, vol. SA WG2, no. V18.1.1, 5 April 2023 (2023-04-05), Mobile Competence Centre ; 650, route des Lucioles ; F-06921 Sophia-Antipolis Cedex ; France, pages 1 - 829, XP052284561 *
AIHUA LI, [LG ELECTRONICS, OPPO, TOYOTA, NOKIA, NOKIA SHANGHAI BELL, SAMSUNG, INTERDIGITAL], CMCC: "Assistance to Member Selection Functionality for Application Operation", 3GPP DRAFT; S2-2302853; TYPE CR; CR 3910; AIMLSYS, 3RD GENERATION PARTNERSHIP PROJECT (3GPP), MOBILE COMPETENCE CENTRE ; 650, ROUTE DES LUCIOLES ; F-06921 SOPHIA-ANTIPOLIS CEDEX ; FRANCE, vol. SA WG2, no. Athens, GR; 20230220 - 20230224, 10 February 2023 (2023-02-10), Mobile Competence Centre ; 650, route des Lucioles ; F-06921 Sophia-Antipolis Cedex ; France, XP052236146 *
SHABNAM SULTANA, ERICSSON: "AIMLsys - QoS filtering criteria in assistance to UE member selection", 3GPP DRAFT; S2-2303651; TYPE CR; CR 3893; AIMLSYS, 3RD GENERATION PARTNERSHIP PROJECT (3GPP), MOBILE COMPETENCE CENTRE ; 650, ROUTE DES LUCIOLES ; F-06921 SOPHIA-ANTIPOLIS CEDEX ; FRANCE, vol. SA WG2, no. Athens, GR; 20230220 - 20230224, 27 February 2023 (2023-02-27), Mobile Competence Centre ; 650, route des Lucioles ; F-06921 Sophia-Antipolis Cedex ; France, XP052249785 *
TRICCI SO, OPPO, NOKIA, NOKIA SHANGHAI BELL, APPLE, ETRI, TOYOTA, SONY, ORACLE, INTERDIGITAL. ERICSSON, SK TELECOM, MEDIATEK INC, : "AIMLsys: KI#7 NEF services description for assistance to member selection", 3GPP DRAFT; S2-2303874; TYPE CR; CR 3674; AIMLSYS, 3RD GENERATION PARTNERSHIP PROJECT (3GPP), MOBILE COMPETENCE CENTRE ; 650, ROUTE DES LUCIOLES ; F-06921 SOPHIA-ANTIPOLIS CEDEX ; FRANCE, vol. SA WG2, no. Athens, GR; 20230220 - 20230224, 27 February 2023 (2023-02-27), Mobile Competence Centre ; 650, route des Lucioles ; F-06921 Sophia-Antipolis Cedex ; France, XP052249963 *

Also Published As

Publication number Publication date
GB202305232D0 (en) 2023-05-24
GB202307142D0 (en) 2023-06-28
GB202403129D0 (en) 2024-04-17
GB202305852D0 (en) 2023-06-07

Similar Documents

Publication Publication Date Title
WO2022177347A1 (fr) Procédé et dispositif de découverte d'un serveur d'applications périphérique
WO2018066876A1 (fr) Procédé de prise en charge de communication v2x dans un système de communication sans fil
WO2021066353A1 (fr) Procédé et appareil d'exécution d'une autorisation pour un service de système aérien sans pilote dans un système de communication sans fil
WO2021091314A1 (fr) Procédé et appareil de gestion de tranches de réseau dans un système de communication sans fil
WO2024155091A1 (fr) Procédés et appareil d'enregistrement et de retransmission dans des déploiements de ntn
WO2021251559A1 (fr) Procédé pour amener une smf à effectuer efficacement une transmission redondante par amélioration de la fonction de nwdaf
WO2022203478A1 (fr) Procédé et appareil pour une minimisation de test mobile dans un système de communication sans fil
WO2022005207A1 (fr) Procédé de communication par multidiffusion
WO2021133011A1 (fr) Procédé et appareil de commutation de réseau
WO2023214806A1 (fr) Procédé et appareil pour la prise en charge d'un apprentissage fédéré dans un système de communications sans fil
WO2024210500A1 (fr) Procédé et appareil de sélection d'un ue pour un apprentissage fédéré
WO2022211519A1 (fr) Procédé de mesure de performance pour qos
WO2022149742A1 (fr) Procédé et dispositif pour minimiser le surdébit provoqué par une fonction étendue dans l'analyse à l'aide de nwdaf
WO2024128636A1 (fr) Procédés et appareil de prise en charge d'une gestion de cycle de vie de modèle ia/ml dans des réseaux de communication sans fil
WO2024186078A1 (fr) Procédés et appareil de gestion de conflit d'opérations ia/aa dans des systèmes de communication
WO2024010399A1 (fr) Gestion et/ou entraînement de modèles d'intelligence artificielle et d'apprentissage machine
WO2024151016A1 (fr) Procédé et appareil d'analyse de latence dans un système de communication
WO2024096386A1 (fr) Procédés et appareil de transfert de données ia/ml dans un système de communication sans fil
WO2024167241A1 (fr) Apprentissage fédéré
WO2023172094A1 (fr) Communication associée à des informations d'analyse
WO2023146253A1 (fr) Procédé destiné à favoriser efficacement la qs de services ia et aa
WO2024010230A1 (fr) Apprentissage fédéré
WO2024162763A1 (fr) Procédé et appareil de veille rrc et de mode inactif dans un système de communication
WO2023121357A1 (fr) Procédé et appareil destinés à un schéma de transmission d'informations de commande
WO2024096458A1 (fr) Procédé et dispositif de mesure de qoe dans un système de communication sans fil