WO2023185807A1 - 一种通信方法及装置 - Google Patents

一种通信方法及装置 Download PDF

Info

Publication number
WO2023185807A1
WO2023185807A1 PCT/CN2023/084300 CN2023084300W WO2023185807A1 WO 2023185807 A1 WO2023185807 A1 WO 2023185807A1 CN 2023084300 W CN2023084300 W CN 2023084300W WO 2023185807 A1 WO2023185807 A1 WO 2023185807A1
Authority
WO
WIPO (PCT)
Prior art keywords
terminal devices
status information
network element
network
network status
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/CN2023/084300
Other languages
English (en)
French (fr)
Inventor
封召
辛阳
王远
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Huawei Technologies Co Ltd
Original Assignee
Huawei Technologies 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 Huawei Technologies Co Ltd filed Critical Huawei Technologies Co Ltd
Priority to KR1020247032965A priority Critical patent/KR20240153394A/ko
Priority to EP23778172.9A priority patent/EP4475507A4/en
Publication of WO2023185807A1 publication Critical patent/WO2023185807A1/zh
Priority to US18/899,990 priority patent/US20250023795A1/en
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W88/00—Devices specially adapted for wireless communication networks, e.g. terminals, base stations or access point devices
    • H04W88/14—Backbone network devices
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/16—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks using machine learning or artificial intelligence
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/04—Processing captured monitoring data, e.g. for logfile generation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • H04L43/0805—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters by checking availability
    • H04L43/0817—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters by checking availability by checking functioning
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/16—Threshold monitoring
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
    • H04L41/14—Network analysis or design
    • H04L41/147—Network analysis or design for predicting network behaviour

Definitions

  • the embodiments of the present application relate to the field of communication technology, and in particular, to a communication method and device.
  • Federated learning is a machine learning framework that can effectively help multiple users perform data usage and machine learning modeling while meeting the requirements of user privacy protection, data security and government regulations.
  • federated learning can effectively solve the problem of data islands and perform joint modeling without sharing user data, thereby technically breaking the data islands and realizing artificial intelligence (AI) collaboration.
  • AI artificial intelligence
  • existing federated learning methods are not efficient.
  • Embodiments of the present application provide a communication method and device, in order to improve the efficiency of federated learning.
  • an embodiment of the present application provides a communication method, including: an application function network element sends a first request message, the first request message is used to request network status information of a terminal device within a candidate range; the application function The network element obtains a first response message, the first response message includes network status information of N1 terminal devices, the terminal devices within the candidate range include the N1 terminal devices, and N1 is a positive integer; the application function network Determine N2 terminal devices based on the network status information of the N1 terminal devices.
  • the N2 terminal devices are used to participate in the training of the federated learning model.
  • the N1 terminal devices include the N2 terminal devices, and N2 is Positive integer.
  • the network status information of the terminal device is taken into consideration to ensure the communication capabilities of the participants and improve the efficiency of federated learning.
  • the first request message includes information indicating the candidate range, where the candidate range includes a designated network area or a terminal device candidate list.
  • the value of N1 falls within a set quantity range.
  • the first request message may include first indication information, the first indication information being used to indicate that the value of N1 is within a set number range.
  • the first request message includes a value range of N1.
  • the value range of N1 may be the same as or different from the aforementioned set quantity range.
  • the network status information of the N1 terminal devices is greater than or equal to the first network status information threshold. It can be understood that the application function network element can also deduce that the network status information of terminal devices other than N1 terminal devices in the candidate range is less than the first network status information threshold. Through such a design, the signaling overhead of the first response message can be saved. .
  • the application function network element determines N2 terminal devices based on the network status information of the N1 terminal devices, including: the application function network element determines the N2 terminal devices based on the network status information of the N1 terminal devices. information and the application layer information of the N1 terminal devices to determine the N2 terminal devices.
  • the application side can use the network status information of the UE combined with the application layer information to select UEs to participate in federated learning model training or update, thereby optimizing the algorithm for selecting participants based on application layer information, thereby improving the efficiency of federated learning model training. efficiency.
  • the method further includes: the application function network element obtains the information of terminal devices with abnormal network status information among the N2 terminal devices; the application function network element obtains the network status information of N3 terminal devices , the candidate range includes the N3 terminal devices, the N3 terminal devices do not include the N2 terminal devices, and N3 is a positive integer; the application function network element is based on the network status information in the N2 terminal devices.
  • the abnormal terminal device information and the network status information of the N3 terminal devices determine N4 terminal devices.
  • the N4 terminal devices are used to participate in the update training of the federated learning model.
  • N4 is a positive integer, for example, N4 is equal to N2.
  • Such a design based on the terminal devices that have participated in federated learning, updates and determines the terminal devices to participate in subsequent rounds of federated learning. Compared with individually determining the terminal devices to participate in federated learning in each round, it is faster and more convenient, thus improving federated learning. s efficiency.
  • the N4 terminal devices include other terminal devices among the N2 terminal devices except for the terminal device with abnormal network status information.
  • the application function network element obtains the information of the terminal device with abnormal network status information among the N2 terminal devices, including: the application function network element sends a second message to the network data analysis function network element. Instruction information, the second instruction information is used to instruct the network status information of the N2 terminal devices to be monitored, the second instruction information includes a second network status information threshold, the second network status information threshold is used to Determine whether the service traffic information of the terminal device is abnormal; the application function network element obtains, from the network data analysis function network element, the information of the terminal device with abnormal network status information among the N2 terminal devices.
  • the communication method further includes: the application function network element and the N2 terminal devices training the federated learning model.
  • embodiments of the present application provide a communication method, which can be applied to access and mobility management function network elements or operation and maintenance management network elements.
  • the communication method includes:
  • the access and mobility management function network element obtains a first request message, where the first request message is used to request network status information of terminal devices within the candidate range;
  • the access and mobility management function network element sends a first response message.
  • the first response message includes network status information of N1 terminal devices.
  • the terminal devices within the candidate range include the N1 terminal devices.
  • N1 is positive. integer.
  • the network status information of N1 terminal devices is used to apply functional network elements to determine N2 terminal devices participating in federated learning model training.
  • the first request message includes information indicating the candidate range, where the candidate range includes a designated network area or a terminal device candidate list.
  • the value of N1 falls within a set quantity range.
  • the first request message may include first indication information, the first indication information being used to indicate that the value of N1 is within a set number range.
  • the first request message includes a value range of N1.
  • the value range of N1 may be the same as or different from the aforementioned set quantity range.
  • the network status information of the N1 terminal devices is greater than or equal to the first network status information threshold.
  • the communication method further includes: the access and mobility management function network element sends network status information of N3 terminal devices, the candidate range includes the N3 terminal devices, and the N3 terminal devices The terminal equipment does not include the N2 terminal equipment, and N3 is a positive integer.
  • inventions of the present application provide a communication method that can be applied to network elements with network data analysis functions.
  • the communication method includes:
  • the network data analysis function network element obtains a first request message, where the first request message is used to request network status information of terminal devices within the candidate range;
  • the network data analysis function network element obtains the network status information of the terminal equipment within the candidate range from the access and mobility management function network element and/or the operation and maintenance management network element;
  • the network data analysis function network element sends a first response message.
  • the first response message includes network status information of N1 terminal devices.
  • the terminal devices within the candidate range include the N1 terminal devices.
  • N1 is a positive integer. .
  • the network status information of N1 terminal devices is used to apply functional network elements to determine N2 terminal devices participating in federated learning model training.
  • the first request message includes information indicating the candidate range, where the candidate range includes a designated network area or a terminal device candidate list.
  • the value of N1 falls within a set quantity range.
  • the first request message may include first indication information, the first indication information being used to indicate that the value of N1 is within a set number range.
  • the first request message includes a value range of N1.
  • the value range of N1 may be the same as or different from the aforementioned set quantity range.
  • the network status information of the N1 terminal devices is greater than or equal to the first network status information threshold.
  • the method further includes: the network data analysis function network element receives second instruction information from the application function network element, the second instruction information is used to instruct the network status information of the N2 terminal devices to be processed. Monitoring, the second indication information includes a second network status information threshold, the second network status information threshold is used to determine whether the business traffic information of the terminal device is abnormal; the network data analysis function network element reports to the application The functional network element sends information about terminal devices with abnormal network status information among the N2 terminal devices.
  • embodiments of the present application provide a communication device, which can be applied to application function network elements, including: a communication module configured to send a first request message, where the first request message is used to request a terminal device within the candidate range. network status information; the communication module is used to obtain a first response message, the first response message includes network status information of N1 terminal devices, and the terminal devices within the candidate range include the N1 terminal devices, N1 is a positive integer; the processing module is used to determine N2 terminal devices according to the network status information of the N1 terminal devices.
  • the N2 terminal devices are used to participate in the training of the federated learning model.
  • the N1 terminal devices include For the N2 terminal devices, N2 is a positive integer.
  • the first request message includes information indicating the candidate range, where the candidate range includes a designated network area or a terminal device candidate list.
  • the first request message includes first indication information, and the first indication information is used to indicate that the value of N1 is within a set number range.
  • the first request message includes a value range of N1.
  • the network status information of the N1 terminal devices is greater than or equal to the first network status information threshold.
  • the processing module is specifically used for:
  • the N2 terminal devices are determined according to the network status information of the N1 terminal devices and the application layer information of the N1 terminal devices.
  • the communication module is also used to obtain information about terminal devices with abnormal network status information among the N2 terminal devices; and obtain network status information of N3 terminal devices, the candidate range Including the N3 terminal devices, the N3 terminal devices do not include the N2 terminal devices, and N3 is a positive integer;
  • the processing module is also used to determine the terminals with abnormal network status information among the N2 terminal devices.
  • the device information and the network status information of the N3 terminal devices determine N4 terminal devices.
  • the N4 terminal devices are used to participate in the update training of the federated learning model, and N4 is a positive integer.
  • the N4 terminal devices include other terminal devices among the N2 terminal devices except for the terminal device with abnormal network status information.
  • the communication module is further configured to: send second indication information to the network data analysis function network element, where the second indication information is used to indicate the network status information of the N2 terminal devices.
  • the second indication information includes a second network status information threshold, the second network status information threshold is used to determine whether the business traffic information of the terminal device is abnormal; obtain all the information from the network data analysis function network element Information about the terminal device with abnormal network status information among the N2 terminal devices.
  • the processing module is also used to train the federated learning model with the N2 terminal devices.
  • inventions of the present application provide a communication device, which can be applied to access and mobility management function network elements or operation and maintenance management network elements.
  • the communication device includes:
  • a communication module configured to obtain a first request message, where the first request message is used to request network status information of terminal devices within the candidate range;
  • a processing module used to determine the network status information of terminal devices within the candidate range
  • the communication module is also configured to send a first response message.
  • the first response message includes network status information of N1 terminal devices.
  • the terminal devices within the candidate range include the N1 terminal devices.
  • N1 is a positive integer. .
  • the network status information of N1 terminal devices is used to apply functional network elements to determine N2 terminal devices participating in federated learning model training.
  • the first request message includes information indicating the candidate range, where the candidate range includes a designated network area or a terminal device candidate list.
  • the value of N1 falls within a set quantity range.
  • the first request message may include first indication information, the first indication information being used to indicate that the value of N1 is within a set number range.
  • the first request message includes a value range of N1.
  • the value range of N1 may be the same as or different from the aforementioned set quantity range.
  • the network status information of the N1 terminal devices is greater than or equal to the first network status information threshold.
  • the communication module is also used to send network status information of N3 terminal devices, the candidate range includes the N3 terminal devices, and the N3 terminal devices do not include the N2 terminal equipment, N3 is a positive integer.
  • inventions of the present application provide a communication device that can be applied to a network element with a network data analysis function.
  • the communication device includes:
  • a communication module configured to obtain a first request message, where the first request message is used to request network status information of terminal devices within the candidate range;
  • the communication module is also used to obtain network status information of terminal devices within the candidate range from the access and mobility management function network element and/or the operation and maintenance management network element;
  • a processing module configured to send a first response message through the communication module.
  • the first response message includes network status information of N1 terminal devices.
  • the terminal devices within the candidate range include the N1 terminal devices.
  • N1 is Positive integer.
  • the network status information of N1 terminal devices is used to apply functional network elements to determine N2 terminal devices participating in federated learning model training.
  • the first request message includes information indicating the candidate range, where the candidate range includes a designated network area or a terminal device candidate list.
  • the value of N1 falls within a set quantity range.
  • the first request message may include first indication information, the first indication information being used to indicate that the value of N1 is within a set number range.
  • the first request message includes a value range of N1.
  • the value range of N1 may be the same as or different from the aforementioned set quantity range.
  • the network status information of the N1 terminal devices is greater than or equal to the first network status information threshold.
  • the communication module is also configured to: receive second instruction information from the application function network element, where the second instruction information is used to instruct the network status information of the N2 terminal devices to be monitored,
  • the second indication information includes a second network status information threshold, and the second network status information threshold is used to determine whether the service traffic information of the terminal device is abnormal; sending the N2 terminal devices to the application function network element Information about terminal devices with abnormal network status information.
  • embodiments of the present application provide a communication method, including:
  • the application function network element sends a second request message, where the second request message is used to request recommended terminal devices within the candidate range that participate in the training of the federated learning model;
  • the application function network element obtains a second response message, the second response message includes recommended N1 terminal devices, the terminal devices within the candidate range include the N1 terminal devices, and N1 is a positive integer;
  • the application function network element determines N2 terminal devices according to the second response message.
  • the N2 terminal devices are used to participate in the training of the federated learning model.
  • the N1 terminal devices include the N2 terminal devices.
  • N2 is a positive integer.
  • embodiments of the present application provide a communication method, including:
  • the network data analysis function network element receives a second request message, where the second request message is used to request recommended terminal devices within the candidate range that participate in the training of the federated learning model;
  • the network data analysis function network element determines recommended terminal devices within the candidate range that participate in the training of the federated learning model
  • the network data analysis function network element sends a second response message, the second response message includes recommended N1 terminal devices, the terminal devices within the candidate range include the N1 terminal devices, and N1 is a positive integer.
  • the network data analysis function network element determines recommended terminal devices within the candidate range that participate in the training of the federated learning model, including:
  • the network data analysis function network element obtains the network status information of the terminal equipment within the candidate range from the access and mobility management function network element and/or the operation and maintenance management network element;
  • the network data analysis function network element determines the recommended N1 devices based on the network status information of the terminal devices within the candidate range.
  • inventions of the present application provide a communication device that can be applied to application function network elements.
  • the communication device includes:
  • a communication module configured to send a second request message, where the second request message is used to request recommended terminal devices within the candidate range that participate in the training of the federated learning model;
  • the communication module is used to obtain a second response message, the second response message includes recommended N1 terminal devices, the terminal devices within the candidate range include the N1 terminal devices, and N1 is a positive integer;
  • a processing module configured to determine N2 terminal devices according to the second response message.
  • the N2 terminal devices are used to participate in the training of the federated learning model.
  • the N1 terminal devices include the N2 terminal devices.
  • N2 is Positive integer.
  • inventions of the present application provide a communication device that can be applied to network data analysis functional network elements.
  • the communication device includes:
  • a communication module configured to receive a second request message, the second request message being used to request recommended terminal devices within the candidate range that participate in the training of the federated learning model;
  • a processing module used to determine recommended terminal devices within the candidate range that participate in the training of the federated learning model
  • the communication module is configured to send a second response message, where the second response message includes recommended N1 terminal devices, the terminal devices within the candidate range include the N1 terminal devices, and N1 is a positive integer.
  • the processing module is specifically configured to: obtain the terminals within the candidate range from the access and mobility management function network element and/or the operation and maintenance management network element through the communication module. Network status information of the device; determine the recommended N1 devices based on the network status information of the terminal devices within the candidate range.
  • an embodiment of the present application provides a communication device.
  • the communication device includes a processor and is configured to implement the method described in the first aspect.
  • the processor is coupled to a memory, and the memory is used to store instructions and data.
  • the communication device may also include a memory; the communication device may also include a communication interface, which is used for the communication device to communicate with other devices.
  • the communication interface may be a transceiver, a circuit , bus, module, pin or other type of communication interface.
  • the communication device includes:
  • Memory used to store programs or instructions
  • a processor configured to use the communication interface to send a first request message, the first request message being used to request network status information of terminal devices within the candidate range; and to obtain a first response message, the first response message including N1 Network status information of terminal devices.
  • the terminal devices within the candidate range include the N1 terminal devices, and N1 is a positive integer;
  • the processor is also configured to determine N2 terminal devices based on the network status information of the N1 terminal devices,
  • the N2 terminal devices are used to participate in the training of the federated learning model, the N1 terminal devices include the N2 terminal devices, and N2 is a positive integer.
  • an embodiment of the present application provides a communication device.
  • the communication device includes a processor and is configured to implement the method described in the second aspect.
  • the processor is coupled to a memory, and the memory is used to store instructions and data.
  • the communication device may also include a memory; the communication device may also include a communication interface, which is used for the communication device to communicate with other devices.
  • the communication interface may be a transceiver, a circuit , bus, module, pin or other type of communication interface.
  • the communication device includes:
  • Memory used to store programs or instructions
  • a processor configured to use a communication interface to obtain a first request message, where the first request message is used to request network status information of a terminal device within the candidate range; determine the network status information of a terminal device within the candidate range; and use the communication interface Send a first response message, where the first response message includes network status information of N1 terminal devices, the terminal devices within the candidate range include the N1 terminal devices, and N1 is a positive integer.
  • an embodiment of the present application provides a communication device.
  • the communication device includes a processor and is configured to implement the method described in the third aspect.
  • the processor is coupled to a memory, and the memory is used to store instructions and data.
  • the communication device may also include a memory; the communication device may also include a communication interface, which is used for the communication device to communicate with other devices.
  • the communication interface may be a transceiver, a circuit , bus, module, pin or other type of communication interface.
  • the communication device includes:
  • Memory used to store programs or instructions
  • a processor configured to use a communication interface to obtain a first request message, the first request message being used to request network status information of a terminal device within the candidate range; using the communication interface to obtain a first request message from an access and mobility management function network element and/or In the operation and maintenance management network element, obtain network status information of terminal devices within the candidate range; and use the communication interface to send a first response message, where the first response message includes network status information of N1 terminal devices, and the candidate
  • the terminal devices within the range include the N1 terminal devices, and N1 is a positive integer.
  • an embodiment of the present application provides a communication device, which includes a processor for implementing the method described in the seventh aspect.
  • the processor is coupled to a memory, and the memory is used to store instructions and data.
  • the communication device may also include a memory; the communication device may also include a communication interface, which is used for the communication device to communicate with other devices.
  • the communication interface may be a transceiver, a circuit , bus, module, pin or other type of communication interface.
  • the communication device includes:
  • Memory used to store programs or instructions
  • a processor configured to use the communication interface to send a second request message, the second request message being used to request recommended terminal devices within the candidate range to participate in the training of the federated learning model; and to use the communication interface to obtain a second response message, the The second response message includes recommended N1 terminal devices, the terminal devices within the candidate range include the N1 terminal devices, and N1 is a positive integer;
  • the processor is further configured to determine N2 terminal devices according to the second response message, and the N2 terminal devices Prepared for participating in the training of the federated learning model, the N1 terminal devices include the N2 terminal devices, and N2 is a positive integer.
  • an embodiment of the present application provides a communication device.
  • the communication device includes a processor and is configured to implement the method described in the eighth aspect.
  • the processor is coupled to a memory, and the memory is used to store instructions and data.
  • the communication device may also include a memory; the communication device may also include a communication interface, which is used for the communication device to communicate with other devices.
  • the communication interface may be a transceiver, a circuit , bus, module, pin or other type of communication interface.
  • the communication device includes:
  • Memory used to store programs or instructions
  • a processor configured to use a communication interface to obtain a second request message, where the second request message is used to request a recommended terminal device within the candidate range to participate in the training of the federated learning model; and to determine the recommended terminal device within the candidate range to participate in the training of the federated learning model. terminal equipment; and use the communication interface to send a second response message, where the second response message includes recommended N1 terminal equipment, the terminal equipment within the candidate range includes the N1 terminal equipment, and N1 is a positive integer.
  • embodiments of the present application provide a communication system, including a communication device as described in the fourth or eleventh aspect; and a communication device as described in the fifth or twelfth aspect; or ,
  • It includes a communication device as described in the fourth or eleventh aspect, a communication device as described in the fifth or twelfth aspect, and a communication device as described in the sixth or thirteenth aspect.
  • embodiments of the present application provide a communication system, including a communication device as described in the ninth or fourteenth aspect; and a communication device as described in the tenth or fifteenth aspect.
  • inventions of the present application provide a communication system.
  • the communication system includes an application function network element, and the application function network element is used to execute the first aspect and any possible design solution of the first aspect.
  • the communication system may also include one or more of a network data analysis function network element, an access and mobility management function network element, and an operation and maintenance management network element.
  • the network data analysis function network element can communicate with the application function network element, the access and mobility management function network element, or the operation and maintenance management network element, and the access and mobility management function network element or the operation and maintenance management network element , can communicate with the application function network element or the network data analysis function network element.
  • embodiments of the present application further provide a computer program, which when the computer program is run on a computer, causes the computer to execute the above first to third aspects, the seventh aspect, and the eighth aspect. methods provided in either aspect.
  • embodiments of the present application further provide a computer program product, including instructions, which when the instructions are run on a computer, cause the computer to execute the above-mentioned first to third aspects, seventh aspects, and eighth aspects. methods provided by any of them.
  • embodiments of the present application further provide a computer-readable storage medium.
  • the computer-readable storage medium stores computer programs or instructions. When the computer program or instructions are run on a computer, it causes The computer executes the method provided in any one of the above first to third aspects, seventh aspect, and eighth aspect.
  • embodiments of the present application further provide a chip, which is used to read the computer program stored in the memory and execute any of the above-mentioned first to third aspects, seventh aspects, and eighth aspects. method provided on one hand.
  • embodiments of the present application also provide a chip system.
  • the chip system includes a processor and is used to support a computer device to implement any one of the first to third aspects, the seventh aspect, and the eighth aspect. methods provided.
  • the chip system further includes a memory, and the memory is used to store necessary programs and data of the computer device.
  • the chip system can be composed of chips or include chips and other discrete devices.
  • Figure 1 is a schematic diagram of a 5G network architecture
  • Figure 2 is a schematic diagram of data distribution of a kind of horizontal federated learning
  • Figure 3 is a schematic diagram of model training for horizontal federated learning
  • Figure 4 is one of the flow diagrams of the communication method provided by the embodiment of the present application.
  • FIG. 5 is one of the flow diagrams of the communication method provided by the embodiment of the present application.
  • Figure 6 is one of the flow diagrams of the communication method provided by the embodiment of the present application.
  • Figure 7 is one of the structural schematic diagrams of a communication device provided by an embodiment of the present application.
  • FIG. 8 is a schematic structural diagram of a communication device provided by an embodiment of the present application.
  • FIG. 1 is a schematic diagram of the 5G network architecture applicable to the embodiment of the present application.
  • the 5G network architecture shown in Figure 1 includes three parts, namely the terminal equipment part, the data network (DN) part and the operator network part. The following is a brief introduction to the functions of some of the network elements.
  • the operator's network may include one or more of the following network elements: Authentication Server Function (AUSF) network element, network exposure function (NEF) network element, Policy Control Function (Policy Control Function, PCF) network element, unified data management (UDM), unified database (Unified Data Repository, UDR), network storage function (Network Repository Function, NRF) network element, access and mobility management function (Access and Mobility Management Function (AMF) network elements, session management function (SMF) network elements, Radio Access Network (RAN) equipment, user plane function (UPF) network elements and networks Data analysis function (Network Data Analytics Function, NWDAF) network element, etc.
  • the part other than the wireless access network part may be called the core network part.
  • the operator network also includes application function (Application Function, AF) network elements and operation and maintenance management (operation administration and maintenance, OAM) network elements.
  • the terminal device in the embodiment of the present application may be a device used to implement wireless communication functions.
  • some examples of terminal equipment include: user equipment (UE), access terminal, terminal unit, terminal station, mobile station, mobile station, remote station, remote terminal, mobile device, wireless communication equipment, terminal agent or Terminal devices, etc.
  • the access terminal may be a cellular phone, a cordless phone, a session initiation protocol (SIP) phone, a wireless local loop (WLL) station, a personal digital assistant (PDA), a device with wireless communications Functional handheld device, computing device or other processing device connected to a wireless modem, vehicle-mounted device, wearable device.
  • SIP session initiation protocol
  • WLL wireless local loop
  • PDA personal digital assistant
  • the terminal device may be a terminal device in the Internet of Things, a virtual reality (VR) terminal device, a 5G network or a terminal device in a future evolved public land mobile network (PLMN), or augmented reality (Augmented Reality).
  • VR virtual reality
  • 5G 5G
  • PLMN public land mobile network
  • augmented reality Augmented Reality
  • AR AR
  • Wireless terminals in industrial control wireless terminals in self-driving, wireless terminals in remote medical, wireless terminals in smart grid, transportation safety ), wireless terminals in smart cities, wireless terminals in smart homes, household appliances, etc.
  • Terminals can be mobile or fixed.
  • the above-mentioned terminal device can establish a connection with the operator network through an interface (such as N1, etc.) provided by the operator network, and use data and/or voice services provided by the operator network.
  • the terminal device can also access the DN through the operator network and use the operator services deployed on the DN and/or services provided by third parties.
  • the above-mentioned third party can be a service provider other than the operator's network and terminal equipment, and can provide other data and/or voice services for the terminal equipment.
  • the specific manifestations of the above-mentioned third parties can be determined according to the actual application scenarios and are not limited here.
  • RAN is a subnetwork of the operator's network and an implementation system between service nodes and terminal equipment in the operator's network.
  • a terminal device To access the operator's network, a terminal device first passes through the RAN, and then can be connected to the service node of the operator's network through the RAN.
  • the RAN device in this application is a device that provides wireless communication functions for terminal devices, such as providing a connection between the terminal device and the core network.
  • the RAN device is also called an access network device.
  • the RAN equipment in this application includes but is not limited to: base station, evolved base station (evolved NodeB, eNodeB), transmission reception point (TRP), next-generation base station in 5G (g nodeB, gNB) , next-generation base stations in 6G mobile communication systems, base stations in future mobile communication systems or access nodes in wireless fidelity (WiFi) systems, evolved node B (evolved node B, eNB), wireless network control Radio network controller (RNC), node B (node B, NB), base station controller (BSC), base transceiver station (BTS), home base station (e.g., home evolved nodeB, or home node B, HNB), baseband unit (baseBand unit, BBU), transmitting point (transmitting point, TP), mobile switching center, etc.
  • base station evolved base station
  • TRP transmission reception point
  • next-generation base station in 5G g nodeB, gNB
  • OAM network elements mainly carry out functions such as daily network and business analysis, prediction, planning and configuration, as well as testing and fault management of the network and its services.
  • OAM network elements can also be called network management.
  • OAM can interact with the RAN to obtain information such as wireless channel conditions and wireless resource utilization on the RAN side.
  • AMF network elements mainly provide mobility management, legal interception, or access authentication/authorization functions for terminal devices. In addition, it is also responsible for transmitting user policies between UE and PCF.
  • the SMF network element mainly performs functions such as session management, execution of control policies issued by PCF, UPF selection, and UE Internet Protocol (IP) address allocation.
  • functions such as session management, execution of control policies issued by PCF, UPF selection, and UE Internet Protocol (IP) address allocation.
  • IP Internet Protocol
  • the UPF network element as the interface UPF with the data network, completes functions such as user plane data forwarding, session/flow-level accounting statistics, and bandwidth limitation.
  • the UDM network element is mainly responsible for managing contract data, user access authorization and other functions.
  • UDR is mainly responsible for the access function of contract data, policy data, application data and other types of data.
  • NEF network elements are mainly used to support the opening of capabilities and events.
  • AF network elements mainly convey the requirements of the application side to the network side, such as Quality of Service (QoS) requirements or user status event subscriptions.
  • QoS Quality of Service
  • AF can be a third-party functional entity or an application service deployed by the operator.
  • the NEF network element is mainly responsible for providing external 5G network capabilities and event opening, as well as receiving relevant external information.
  • the PCF network element is mainly responsible for policy control functions such as session and service flow level billing, QoS bandwidth guarantee and mobility management, and UE policy decision-making.
  • NRF network elements can be used to provide network element discovery functions and provide network element information corresponding to network element types based on requests from other network elements.
  • NRF also provides network element management services, such as network element registration, update, de-registration, network element status subscription and push, etc.
  • NWDAF network elements are mainly used to collect network data, such as relevant data from terminal equipment, RAN equipment, core networks, third-party business servers or network management systems. NWDAF network elements provide network data analysis services, can output data analysis results, and provide data analysis results to terminal equipment, RAN equipment, core networks, third-party service servers or network management systems. NWDAF can use machine learning models for data analysis. For example, the functions of NWDAF in 3GPP Release 17 are decomposed, including data collection function (or data collection logic function), model training function (or machine learning model training logic function ( machine learning model training logical function)) and model reasoning function (or analytics logical function). For example, the data analysis results can assist the network in selecting service quality parameters of the service, or assist the network in performing traffic routing, or assist the network in selecting background traffic transmission strategies, etc.
  • data collection function or data collection logic function
  • model training function or machine learning model training logic function ( machine learning model training logical function)
  • model reasoning function or analytics logical function
  • Nnwdaf, Nausf, Nnef, Npcf, Nudm, Naf, Namf, Nsmf, N1, N2, N3, N4, and N6 are interface serial numbers. The meaning of these interface serial numbers can be found in the meaning defined in the 3GPP standard protocol, and is not limited here.
  • the network data analysis function network element may be the NWDAF network element shown in Figure 1, or other network elements in the future communication system that have the functions of the NWDAF network element in the present application.
  • Mobile The performance management network element can be the AMF network element shown in Figure 1, or other network elements in the future communication system that have the functions of the AMF network element in this application;
  • the user plane network element can be the UPF network element shown in Figure 1 , or it can be other network elements in the future communication system that have the functions of the UPF network element in this application;
  • the session management function network element can be the SMF network element shown in Figure 1, or it can be the SMF network element in the future communication system that has the function of the UPF network element in this application.
  • the application function network element can be the AF network element shown in Figure 1, or other network elements in the future communication system that have the functions of the AF network element in this application
  • the access network equipment can It is the RAN equipment shown in Figure 1, or it can be other network elements in the future communication system that have the functions of the RAN equipment in this application
  • the operation and maintenance management network element can be the OAM network element shown in Figure 1, or it can be the future communication system.
  • the data analysis network element is the NWDAF network element
  • the mobility management network element is the AMF network element
  • the user plane network element is the UPF network element
  • the application function network element is the AF network element.
  • Operation and maintenance The management network element is an OAM network element and the access network device is a RAN device as an example for explanation.
  • the terminal device is a UE as an example for explanation.
  • NWDAF network element in the embodiment of the present application may also be abbreviated as NWDAF
  • the UPF network element may also be abbreviated as UPF
  • the AF network element may also be abbreviated as AF
  • the OAM network element may also be abbreviated as OAM.
  • Federated Learning is a machine learning framework in which nodes do not need to interact with data, but instead transfer intermediate results obtained during training, such as model parameters or gradients and other information that can characterize the model. That is to say, federated learning can perform machine learning modeling, that is, train AI models, while meeting the requirements of user privacy protection and data security. As a distributed machine learning paradigm, federated learning can effectively solve the problem of data islands, allowing nodes participating in federated learning to jointly model without sharing data, thereby technically breaking data islands and achieving AI collaboration.
  • federated learning can be divided into three categories according to the distribution of data sources among the participating parties: horizontal federated learning, vertical federated learning, and federated transfer learning.
  • horizontal federated learning refers to learning on multiple data sets (or understood as When the user characteristics of the sample set (sample set) overlap more and the users overlap less, the data set is divided horizontally (that is, the user dimension), and the model is trained based on the partial data with the same user characteristics but different users.
  • Vertical federated learning refers to splitting the data set vertically (i.e., feature dimension) when the users of multiple data sets overlap more but the user features have less overlap, based on the parts where the users are the same but the user features are not exactly the same. Data is used to train the model.
  • Federated transfer learning means that when there is little overlap between users and user features in multiple data sets, the data is not segmented, but transfer learning is used to overcome the situation of insufficient data or sample labels.
  • the following takes horizontal federated learning as an example to explain in detail the training process of the federated learning model.
  • Figure 2 is a schematic diagram of data distribution for horizontal federated learning, involving the data sets of operators A and B.
  • the intersection of the data sets of operators A and B is small in the user dimension, but the intersection in the user characteristic dimension is large.
  • data with the same characteristics can be selected from the data sets of operators A and B as training data.
  • the data set of operator A contains the data of (user 1, user 2, user 3, user 4), and each user's data contains features (feature 1, feature 2, feature 3, feature 4, feature 5),
  • the data set of operator B contains the data of (user 4, user 5, user 6, user 7, user 8), where each user's data contains features (feature 2, feature 3, feature 4, feature 5, feature 6 ).
  • the data of feature 2, feature 3, feature 4 and feature 5 can be selected from the data sets of operators A and B as training data.
  • Figure 3 is a schematic diagram of model training for horizontal federated learning.
  • Step 1 Multiple UEs participate in the training of the federated learning model, use local data to calculate the gradient corresponding to the local model, and report the gradient to the server used to train the federated learning model.
  • Figure 3 illustrates the horizontal federated learning model training process where intermediate results are transmitted by the 5G System (5GS).
  • k UEs can report gradients to the server through 5GS.
  • Step 2 The server performs gradient aggregation to determine or update the parameters of the federated learning model.
  • Step 3 The server distributes the gradient corresponding to the updated federated learning model to each participant. For example, the server sends the gradient corresponding to the updated federated learning model to each UE through 5GS.
  • Step 4 Each UE updates its own local model based on the gradient corresponding to the federated learning model obtained in step 3. Repeat steps 1 to 4 for multiple iterations until the federated learning model converges.
  • the number of UEs participating in federated learning model training may be very large, and all UEs participating in federated learning model training will occupy too much network bandwidth. Due to limited network bandwidth resources, when the total number of participants in horizontal federated learning model training is large, some UEs can be selected to participate in model training during each round of training or updating of the federated learning model. The quality of UE selection results will directly affect the efficiency of model training.
  • the heterogeneous clients in the federated learning FL environment There may be heterogeneous clients in the federated learning FL environment.
  • the data sets of different UEs are of different quality, and the storage, computing and communication capabilities of different UEs may also be quite different.
  • the random selection part will aggravate the problems in the federated learning FL environment.
  • Heterogeneity is not conducive to the training efficiency of the federated learning model.
  • the heterogeneity in the FL environment mainly includes: (1) data heterogeneity, the local data in each UE device may not be independently identically distributed; (2) storage and computing power heterogeneity , each UE may have a large difference in storage and computing capabilities.
  • the result is that when different UEs use local data to calculate the model gradient, the calculation time varies greatly; (3) Communication capability heterogeneity, the communication capabilities of different UEs There may also be a large difference, resulting in a large difference in the time for different UEs to upload the locally calculated gradients to the server.
  • the local calculation completion time of different UEs may be different, and the time to transmit the intermediate results to the server may also be different, and the server needs to wait until all UEs have completed transmitting the intermediate results before it can update the federated learning model, so the overall training cycle will become longer. . It can be seen that the heterogeneity caused by randomly selecting UEs will affect the training efficiency of the federated learning model.
  • the existing technology adopts a non-random client (or UE) selection scheme.
  • the server such as AF network element selects the client according to the UE's Application layer information, distribution information of user data sets, user transmission information volume, etc., select UEs to participate in federated learning model training.
  • the AF network element can select UEs that transmit a large amount of information to participate in the training of the federated learning model; or, the AF network element can use the experience of reinforcement learning to drive the federated learning framework and select UEs that participate in each round of training.
  • AF network elements use reinforcement learning modeling to prioritize data that has a greater effect on improving model accuracy and UEs that have the ability to quickly perform training.
  • this design only considers the application layer information of the UE. If the UE moves out of the service area, the UE's wireless signal becomes weak, or the connection between the UE and the network is interrupted, resulting in errors in federated learning model training and update, it will still affect federated learning. s efficiency.
  • embodiments of this application provide a participant selection scheme for federated learning.
  • the network status information of the UE is opened to facilitate the AF to promptly consider the network status information of the UE when selecting UEs to participate in the training or updating of the federated learning model, and avoid selecting UEs with poor network status information or unsuitable for federated learning. UE, thereby improving the efficiency of federated learning.
  • the network status information of the UE is used to indicate one or more of the following parameters: the connection status of the UE, such as the connection status of the UE is connected, idle or deactivated; the reachability of the UE, such as the UE Reachable or unreachable UE; UE's mobility, such as UE's mobility mode or mobility trend; UE's wireless channel conditions, or described as the condition of the wireless channel between UE and RAN equipment, such as the use of UE and RAN equipment.
  • SNR signal to interference plus noise ratio
  • RSRP reference signal received power
  • RSRQ reference signal received quality
  • the AF network element sends a first request message, where the first request message is used to request network status information of UEs within the candidate range.
  • the candidate range may be a designated network area or a terminal device candidate list.
  • the designated network area can be one or more network slices, and one network slice is identified by single network slice selection assistance information (single network slice selection assistance, S-NSSAI).
  • the designated network area can be one or more tracking areas, and one tracking area (tracking area, TA) corresponds to a tracking area identity (TAI).
  • the designated network area may be one or more cells, and one cell corresponds to a cell identity (cell ID).
  • the designated network area may correspond to one or more data networks.
  • a data network is identified by a data network name (DNN).
  • the UEs in the designated network area refer to the UEs connected to the data network.
  • the specified network area can correspond to one or more applications, and one application corresponds to an application identity (application ID).
  • the UE in the specified network area refers to the one or more applications identified by the ID that are running. Application UE.
  • the UE may move in the actual communication environment, and the UEs in the aforementioned designated network area may be different in different time periods.
  • the designated network area can also be a combination of the above identifiers.
  • the UE in the specified network area may refer to the UE in the area identified by TAl and within the slice identified by S-NSSAI1.
  • the UE within the specified network area may refer to the UE that is in the cell identified by cell2 and accesses the data network identified by DNN3.
  • the terminal device candidate list it can be understood as a preconfigured group of UEs corresponding to a group of UE identifiers (UE group ID or list of UE IDs).
  • UE group ID a group of UE identifiers
  • the list of Internet Protocol (IP) addresses of UE is represented as list of UE IP; or the list of user permanent identifiers (subscription permanent identifier, SUPI) is represented as list of SUPI; or the general public subscription identifier ( generic public subscription identifier, GPSI) list, represented as list of GPSI.
  • IP Internet Protocol
  • SUPI subscription permanent identifier
  • GPSI generic public subscription identifier
  • the AF network element may send the first request message to the network element that can provide the network status information of the UE according to the network status information it wishes to obtain. For example, if the AMF network element can provide the network status information that the AF network element wants to obtain, the AF network element can send a first request message to the AMF network element; if the OAM network element can provide the network status information that the AMF network element wants to obtain, the AF network element can The network element may send the first request message to the OAM network element. For a UE, the network status information that AF hopes to obtain comes from AMF and OAM as an example.
  • Figure 4 specifically shows in S401: S401a, the AF network element sends the first request message to the AMF; and S401b, the AF network Yuan sends the first request message to OAM.
  • the first request message may be a subscription message.
  • the AF can send a subscription message to the AMF through the Namf_EventExposure Subscribe service operation to obtain the network status information of the UE within the candidate range, such as the UE's connection status, the UE's reachability information, the UE's mobility information, etc.
  • the subscription message can carry the following information: event ID and information indicating the candidate range.
  • Event ID is used to identify the subscribed event type, indicating that AF wants to subscribe to and obtain the UE's network status information (specific type) from AMF.
  • Event ID UE reachability means that the AF subscribes to the UE's network status information from the AMF, including the UE's reachability information.
  • Event ID Location Reporting means that the AF subscribes to the UE's network status information from the AMF, including the UE's location change information.
  • the information used to indicate the candidate range may be the identifier of the candidate range, such as range of UEs representing a designated network area, and list of UE IDs representing a list of candidate terminal devices.
  • the AF network element may decide to request the number of terminal devices corresponding to the network request information.
  • the first request message may specifically request the network status information of some or all UEs within the candidate range.
  • the AF may include first indication information in the first request message.
  • the first indication information is used to indicate that the number of UEs corresponding to the AMF/OAM feedback network status information needs to be within a set number range.
  • N1 represent the number of UEs corresponding to AMF/OAM feedback network status information, and the value of N1 is within the set number range.
  • the set quantity range may be a quantity threshold, such as 1000, indicating that N1 ⁇ 1000, and N1 is a positive integer.
  • the set quantity range can also be a value interval, such as [200,3000], and N1 is a positive integer in [200,3000].
  • the AF can also indicate to the AMF/OAM the range of UE numbers corresponding to the network status information it requires.
  • the AF includes the range of UE number in the first request message, that is, The value range of N1.
  • the value range of N1 described here may be the same as or different from the aforementioned set quantity range.
  • the AF may request network status information of all UEs in the candidate terminal device list.
  • the AF network element may decide to request the network status information of the UE within the valid time range.
  • the AF may include a data validity indication in the first request message.
  • the data validity indication is used to indicate the valid time range corresponding to the valid data that the AF wishes to obtain.
  • the first request message including the data validity indication is specifically used for Request the network status information of the UE in the candidate range within the valid time range.
  • the valid time range can be a time range range, indicating that the network status information obtained by the data providing network element (AMF/OAM) within this time range is valid data.
  • the valid time range may be a cut-off time point, indicating that the data fed back by the data providing network element (AMF/OAM) before the cut-off time point is valid.
  • the AF network element can also add a validity joint indication in the first request message.
  • the validity joint indication is used to instruct the data provider network element to collect data within a certain time range and send it before the deadline.
  • the data obtained from the network element is the valid data.
  • the AF network element can also add performance indication information to the first request message to indicate that the AF network element has requirements for the performance of the requested UE.
  • the performance indication information is used to indicate the UE that the AF network element wants to request.
  • the network status information is greater than or equal to the first network status information threshold.
  • the first network status information threshold may be predefined or determined by the AF on its own, which is not limited in the embodiments of this application.
  • the first request message may include the first network status information threshold.
  • NEF controls the mapping relationship between AF identification and the Event IDs allowed to be obtained, as well as related inbound restrictions (i.e., limiting the Event IDs that AF can request) and outbound restrictions (i.e., limiting the Event IDs that can be notified to AF).
  • inbound restrictions i.e., limiting the Event IDs that AF can request
  • outbound restrictions i.e., limiting the Event IDs that can be notified to AF.
  • AMF first sends the network status information of the relevant UE to NEF. Then NEF sends it to the third-party AF.
  • AF can send a subscription message to OAM to obtain network status information of relevant UEs, such as wireless channel conditions, wireless resource utilization, etc. It should be noted that AF can obtain data directly from OAM. AF can also obtain data from OAM through NWDAF or other 5GC NF. For example, AF first sends a subscription message to NWDAF to subscribe to the RAN side data at OAM, then NWDAF subscribes and obtains relevant data from OAM, and finally NWDAF will obtain the data. The data is notified to AF through the service interface (such as Nnwdaf_AnalyticsSubscription_Notify service operation).
  • the service interface such as Nnwdaf_AnalyticsSubscription_Notify service operation.
  • the AF network element obtains a first response message, where the first response message includes network status information of N1 UEs.
  • N1 UEs are included in the candidate range, or it is described that the UEs in the candidate range include the N1 UEs, and the N1 UEs represent some or all UEs in the candidate range.
  • the definition of N1 UEs can be understood with reference to the description in S401.
  • the first request message sent by the corresponding AF network element includes the first indication information
  • the value of N1 is within the set number range; if the corresponding AF network element sends
  • the first request message indicates the quantity range, the value of N1 is within the quantity range indicated by the AF; if the first request message sent by the AF does not carry the first indication information and the indicated quantity range, the N1 UEs are included in the candidate range.
  • the data provider is able to collect all UEs with network status information; also, when the AF sends the first request message carrying performance indication information and/or the first network status information threshold, the network status information of N1 UEs is greater than or equal to the first network status information. Status information threshold.
  • the first response message may come from AMF and/or OAM.
  • Figure 4 illustrates in S402 S402a, the AMF network element sends the first response message to the AF network element; and, S402b, the OAM network element sends the first response message to the AF network element.
  • Figure 4 only shows that the AMF/OAM directly sends the first response message to the AF network element, while the relevant AMF/OAM indirectly sends the first response message to the AF network element through the intermediate network element (such as NEF, NWDAF, etc.)
  • the intermediate network element such as NEF, NWDAF, etc.
  • the AMF network element/OAM network element can carry the Event ID in the first response message sent to indicate the type of network status information of the UE in the first response message
  • the network status information included in the first response message by the AMF network element/OAM network element is the value of the network status information corresponding to the Event ID, or can also be described as the original value of the N1 UEs obtained by the AMF network element/OAM network element.
  • Network status information For example, when the type of network status information includes radio resource utilization, the radio resource utilization corresponding to the UE carried in the first response message may be a specific value such as 0.6.
  • the AF network element determines N2 UEs based on the network status information of the N1 UEs.
  • the N2 UEs are used to participate in the training of the federated learning model, the N2 UEs are included in the N1 UEs, or described as the N1 UEs include the N2 UEs, N2 is a positive integer, and N2 is less than or equal to N1.
  • the AF network element can determine the N2 UEs only based on the network status information of the N1 UEs through data analysis or a specific algorithm.
  • the AF network element can also determine the N2 UEs based on the network status information of the N1 UEs combined with the application layer information of the N1 UEs, such as using the network status information of the UEs to optimize the existing application layer algorithm, or combining the network status of the UEs.
  • a new algorithm is designed for information and application layer information, and N2 UEs are determined by the new algorithm.
  • the AF network element can also determine the N2 UEs based on the network status information of the N1 UEs combined with the application layer information of the N1 UEs.
  • the embodiments of this application provide the following specific implementation methods for determining the N2 UEs:
  • the AF network element has obtained the application layer information and network status information of 100 UEs, and is preparing to select 80 UEs to participate in the training of the federated learning model.
  • AF can score each UE according to its connection status. For example, "1" represents the connected state, and "0" represents the idle state.
  • scoring values for each parameter can also be set according to the actual situation. For example, for scoring the connection status of the UE, in addition to using "1" to represent the connected state and "0" to represent the idle state, it can also be Use "0.9” to represent the connected state, and "0.1” to represent the idle state. The embodiments of the present application do not limit this.
  • AF sets a weight for each parameter according to the importance of each parameter indicated by the network status information.
  • AF can Calculate the weighted scores of other UEs, then sort the weighted scores of all UEs, and select the 80 UEs with the highest scores. As a UE participating in federated learning model training.
  • S404 The AF network element performs federated learning model training with N2 UEs.
  • the N2 UEs participating in the model training in each round can be determined according to the above-mentioned S401 to S404; or, every certain round, the N2 UEs participating in the model training can be determined according to the above-mentioned steps.
  • S401 to S404 determine N2 UEs participating in model training in this round.
  • the AF can refer to S401 to S403 to determine the UE participating in the i-th round of model training in the I round.
  • i takes a positive integer from 1 to I.
  • i is a partial positive integer from 1 to I.
  • the difference between every two adjacent values of i is the same, such as I is 6 and the values of i are 1, 3, and 5. It can be understood that when I is 6 and the value of i is 1, 3, 5, it means that the UEs participating in model training in the second round and the first round are consistent, and the UEs participating in the model training in the fourth and third rounds are the same. The UEs remain consistent, as well as the UEs participating in model training in rounds 6 and 5. Another example is that the difference between every two adjacent values of i is not limited to the same.
  • the value of i can be random or determined by a relevant algorithm. For example, if I is 6, the value of i can be 1, 3, or 4. 6.
  • each round starting from the second round can update and determine the UEs participating in model training in this round based on the UEs determined in the previous round.
  • UE the first round of model training can be implemented according to the aforementioned S401 to S404, and the subsequent rounds of model training can be implemented according to the following S405 to S408. It can be understood that, assuming that the aforementioned S401 to S404 are recorded as the method of determining the participating UE in the i-th round of model training, then S405 to S408 indicated by dotted lines in Figure 4 can be understood as the method of determining the participating UE in the i+1 round of model training. way, i is a positive integer.
  • the AF network element obtains the information of the UEs with abnormal network status information among the N2 UEs.
  • the AF network element may send second indication information to the NWDAF network element, where the second indication information is used to instruct the network status information of the N2 UEs to be monitored.
  • S405b The NWDAF network element obtains the network status information of N2 UEs from the UPF network element.
  • S405c The NWDAF network element determines the UE with abnormal network status information among the N2 UEs.
  • S405d The NWDAF network element sends the information of the UE with abnormal network status information among the N2 UEs to the AF network element.
  • the information of the UE whose network status information is abnormal may include abnormality indication and network status abnormality information.
  • the exception indication is used to indicate the type of network status information abnormality, such as business traffic abnormality;
  • network status abnormality information may include business traffic abnormality information, such as UE quality of service (QoS) information, UE QoS information It can indicate UE abnormalities, such as the signal transmission delay corresponding to the UE being too large or too small, such as the packet loss rate corresponding to the UE being too large, etc.
  • the second indication information may include a second network status information threshold, and the second network status information threshold is used to determine whether the network status information of the UE is abnormal.
  • the second indication information in S405a may be implemented using a subscription request.
  • AF can send a subscription request to NWDAF through the Nnwdaf_AnalyticsSubscription_Subscribe service operation.
  • the subscription request carries the following parameters:
  • the Analytics ID is used to identify the subscription analysis type.
  • the Analytics ID can be "QoS information of abnormal UEs", which means to obtain the QoS information of abnormal UEs (such as UEs with too large or too small delay). .
  • Notification Indication is used to indicate the conditions for NWDAF feedback analysis results. For example, it can be periodic feedback, or feedback based on thresholds, such as feedback when the delay variance of the UE in UE ID list2 is greater than a certain threshold. It can be understood that the notification indication is an optional option, and the aforementioned subscription request may or may not include the notification indication.
  • This information may correspond to the aforementioned second network status information threshold.
  • the second network status information threshold corresponds to a specific threshold (such as 100ms), indicating that NWDAF needs to delay
  • the QoS information of the UE that is greater than the threshold is fed back to the AF; alternatively, the second network status information threshold can also correspond to a group threshold range represented by a group of thresholds, such as greater than 100ms or less than 10ms, indicating that the NWDAF needs to delay the delay within the group threshold.
  • the QoS information of the UEs within the range is fed back to the AF. It can be understood that Notification Threshold is an optional option, and the notification indication may or may not be included in the aforementioned subscription request.
  • the NWDAF network element can subscribe to the UPF network element through the Nupf_EventExposure_Subscribe service operation and obtain the QoS information of N2 UEs. After the UPF network element completes the information collection, it notifies the NWDAF network element of the collected QoS information through the Nupf_EventExposure_Notify service operation. Optionally, the NWDAF network element can also subscribe to the UPF network element through the SMF network element and obtain the QoS information of N2 UEs.
  • the NWDAF network element can determine whether the network status information of the N2 UEs is abnormal based on the second network status information threshold. Thus, the UE with abnormal network status information among the N2 UEs is determined. Alternatively, if the subscription request sent by the AF in S405a does not include the second network status information threshold, the NWDAF network element can determine the UE with abnormal network status information among the N2 UEs based on the preconfigured internal algorithm.
  • the second network status information threshold such as Notification Threshold
  • the NWDAF network element may send a notification message to the AF through the Nnwdaf_AnalyticsSubscription_Notify service operation to feed back the information of the UE with abnormal network status information among the N2 UEs.
  • the NWDAF network element can feed back the QoS information of the abnormal UE to the AF network element according to the Notification Indication. For example, it can determine periodic feedback based on the Notification Indication, or it can also determine based on the Notification Indication. threshold feedback.
  • the NWDAF network element can also determine by itself when to send the notification message to achieve feedback of QoS information. For example, when the NWDAF network element infers an abnormal UE, it can immediately feedback the QoS information of the abnormal UE. Give it AF.
  • the AF network element can also directly subscribe to and obtain the information of the UEs with abnormal network status information among the N2 UEs from the UPF network element without passing through the NWDAF network element.
  • the embodiments of the present application do not limit this.
  • the AF network element obtains network status information of N3 UEs.
  • the candidate range includes the N3 UEs, the N3 UEs do not include the N2 UEs, and N3 is a positive integer.
  • Figure 4 illustrates that the AF network element can subscribe to and obtain the network status information of N3 UEs from the AMF network element and the OAM network element. It can be understood that the N3 UEs refer to UEs within the candidate range that have not participated in the federated learning model training process described in S404.
  • the value of N3 may be determined by the AF.
  • the AF may determine the The weighted score selects the top N3 UEs with higher scores from the UEs that have not participated in the federated learning model training, or a group of UEs randomly selected by the AF from the UEs that have not participated in the federated learning model training.
  • the AF may include information indicating the value range of N3 in the subscription message sent to AMF and/or OAM, for example, indicating N3 UEs with UE ID list 1 (UE ID list1).
  • the AF may send the subscription message to AMF and/or OAM.
  • the UE ID list1 is included in the subscription message sent.
  • the quantity range corresponding to N3 may be predefined, or the AF may indicate the value range of N3 in the subscription message, and then AMF and/or OAM may determine the quantity range corresponding to N3 or The N3 value range indicated by the AF feeds back the network status information of the corresponding number of UEs to the AF.
  • AMF and/or OAM may determine the quantity range corresponding to N3 or The N3 value range indicated by the AF feeds back the network status information of the corresponding number of UEs to the AF.
  • the AF network element determines N4 UEs based on the information of the UEs with abnormal network status information among the N2 UEs and the network status information of the N3 UEs.
  • the N4 UEs are used to participate in the update of the federated learning model. Training, N4 is a positive integer.
  • the AF network element may refer to the scoring strategy described in S403, determine the weighted scores of the N3 UEs based on the network status information of the N3 UEs, and then determine the weighted scores of the N3 UEs from the N2 UEs based on the weighted scores of the N3 UEs.
  • N4 UEs are determined.
  • the N4 UEs include other UEs among the N2 UEs except the N5 UEs with abnormal network status information and the N6 UEs with the highest weighted scores among the N3 UEs.
  • N5 and N6 are positive integers.
  • the number of UEs participating in different rounds of federated learning may be the same, that is, N4 equals N2.
  • N6 may be equal to N5.
  • the UE ID list2 described in S405 contains 80 UEs, and the network status information of 10 UEs among the 80 UEs is abnormal.
  • the UE ID list1 described in S406 corresponds to the network status information of 20 UEs.
  • AF can use the same algorithm as in S403 to score the application layer information and network status information of the UE in UE ID list1, and then obtain the weighted score of each UE in UE ID list1.
  • AF can select the 70 non-abnormal UEs in UE ID list2 and the 10 UEs with the highest weighted scores in UE ID list1 as the N4 UEs participating in the federated learning model update training.
  • the number of UEs participating in different rounds of federated learning may be different, that is, N4 is not equal to N2.
  • N6 does not need to be equal to N5.
  • the UE ID list2 described in S405 contains 80 UEs, and the network status information of 10 UEs among the 80 UEs is abnormal.
  • the UE ID list1 described in S406 corresponds to the network status information of 20 UEs.
  • AF can use the same algorithm as in S403 to score the application layer information and network status information of the UE in UE ID list1, and then obtain the weighted score of each UE in UE ID list1.
  • AF can select the 70 non-abnormal UEs in UE ID list2 and the 15 UEs with the highest weighted scores in UE ID list1 as the N4 UEs participating in the federated learning model update training.
  • N4 is greater than N2; or AF can select UE ID
  • the 70 non-abnormal UEs in list2 and the 5 UEs with the highest weighted scores in UE ID list1 are used as the N4 UEs participating in the federated learning model update training.
  • N4 is smaller than N2.
  • the AF network element may refer to the scoring strategy described in S403, determine the weighted scores of the N3 UEs based on the network status information of the N3 UEs, and then combine the weighted scores of the N3 UEs with the N2 UEs.
  • UE determine N4 UEs.
  • the N4 UEs may include some UEs among the N2 UEs whose network status information is abnormal.
  • the UE ID list 2 described in S405 contains 80 UEs, and the network status information of 30 UEs among the 80 UEs is abnormal.
  • the UE ID list1 described in S406 corresponds to the network status information of 20 UEs.
  • AF can The same algorithm as in S403 is used to score the application layer information and network status information of the UE in the UE ID list 1, and then the weighted score of each UE in the UE ID list 1 is obtained.
  • UE ID list 1 corresponds to 15 UEs among 20 UEs whose weighted scores are greater than or equal to 0.6.
  • the AF can select 50 non-abnormal UEs in UE ID list 2, UEs with a weighted score exceeding 0.6 (such as 15 UEs) in UE ID list 1, and 15 UEs with the smallest delay difference among abnormal UEs in UE ID list 2 as participants in the federation.
  • the learning model updates the trained N4 UEs.
  • the delay difference refers to the absolute value of the difference between the actual signal transmission delay of the abnormal UE and the second network status information threshold corresponding to the signal transmission delay.
  • N4 is equal to N2.
  • N4 may not be equal to N2, and the embodiment of the present application does not limit this.
  • S408 The AF network element performs updated training of the federated learning model with N4 UEs.
  • the AF network element requests and obtains the network status information of the candidate UE from the 5GC NF (such as AMF) network element or OAM network element, and is applied to the federated learning environment to facilitate the application side to utilize the UE's network status information.
  • the network status information selects UEs that participate in federated learning model training or updating, optimizing the algorithm for selecting participants based on application layer information, thereby improving the efficiency of federated learning model training.
  • AF can also be called FL AF.
  • the method includes the following process.
  • the AF network element sends a first request message to the NWDAF network element, where the first request message is used to request network status information of UEs within the candidate range.
  • the first request message may be a subscription message.
  • the AF network element sends a subscription message to NWDAF through the Nnwdaf_AnalyticsSubscription_Subscribe service operation.
  • the parameters carried in the subscription message include the analytics ID.
  • Analytics ID UE Network Status Information, indicating that the AF network element wants to subscribe from NWDAF and obtain the network status information of the UE obtained through NWDAF analysis, such as the network status information of the UE and the set threshold or the network status information of other UEs. Comparative analysis results between.
  • the subscription message also includes information indicating the candidate range, which may be the identifier of the candidate range. For example, range of UEs represents the designated network area, and list of UE IDs represents the list of candidate terminal devices.
  • the subscription message may also carry a number range indication, such as the first indication information described in S401 or the range of UE number (range of UE number).
  • the subscription message may also carry a data validity indication or a joint validity indication as described in S401.
  • the subscription message may also carry the performance indication information and/or the first network status information threshold as described in S401.
  • NEF controls the mapping relationship between AF identities and the Analytics IDs allowed to be obtained, as well as related inbound restrictions (i.e., limiting the Analytics IDs that AF can request) and outbound restrictions (i.e., limiting the Analytics IDs that can be notified to AF).
  • inbound restrictions i.e., limiting the Analytics IDs that AF can request
  • outbound restrictions i.e., limiting the Analytics IDs that can be notified to AF.
  • AMF first sends the network status information of the relevant UE to NEF. Then NEF sends it to the third-party AF.
  • the NWDAF network element obtains the network status of the UE within the candidate range from the AMF network element and/or OAM network element. information.
  • the NWDAF may use a subscription method to subscribe to the AMF network element and/or the OAM network element and obtain the network status information of some or all UEs within the candidate range. For example, if the AMF network element/OAM network element can identify the UE in the non-connected state, then the AMF network element/OAM network element can provide the network status information of all UEs in the candidate range, where the network status information of the UE in the non-connected state is Indicates that the status of the UE is idle or deactivated.
  • the AMF network element/OAM network element can provide network status information of some UEs in the candidate range, or it can be understood that the AMF network element/OAM network element can Provide network status information of all UEs within the candidate range that can obtain network status information.
  • all UEs that can obtain network status information are in the connected state.
  • NWDAF which can obtain the network status information of the UE in a subscription manner, it can be implemented with reference to S401 to S402, which will not be described again in the embodiment of this application.
  • the NWDAF network element sends a first response message to the AMF network element.
  • the first response message includes the network status information of N1 UEs.
  • N1 UEs can be understood with reference to S402, which will not be described again in the embodiment of this application.
  • the NWDAF network element may analyze the network status information of N1 UEs obtained from the AMF network element/OAM network element, and carry the network status information of the N1 UEs analyzed by the NWDAF network element in the first response message. Or it can be described as, if the first request message carries the Analytics ID, the network status information of the N1 UEs carried in the first response message can be the network status information obtained by NWDAF network element analysis, such as the network status information of the UE and the device Comparative analysis results between certain thresholds or network status information of other UEs, such as network status information of UEs within a certain period of time or in a certain area.
  • NWDAF can organize the UE network status information historically obtained from AMF/OAM into a historical data set for statistical analysis, and obtain the statistical characteristics of the historical data set. NWDAF obtains predicted analysis results based on the statistical characteristics of the historical data set. NWDAF can carry the predicted analysis results in the first response message and send it to AF to assist AF in selecting UEs to participate in model training; or, NWDAF can use the obtained Use the historical data set to train an AI model, and use the AI model to reason and obtain the analysis results of the prediction.
  • NWDAF can provide statistical information on the UE's connection status, such as which UEs are in the connected state and which UEs are in the connected state within the range specified by the AF (such as area/slice/DNN/App ID) or within the specified time period.
  • NWDAF can provide prediction information of UE connection status, such as which UEs are currently in idle state, but will enter the connected state within the range specified by AF or within the specified time period.
  • NWDAF can provide UE reachability analysis results, such as which idle UEs are available in the AF specified range (such as area/slice/DNN/App ID) or within the specified time period. Which UEs are reachable and which ones are unreachable; NWDAF can provide prediction information of UE reachability, such as which UEs are currently unreachable, but will become reachable within the range specified by AF or within the specified time period.
  • NWDAF can provide statistical information on UE mobility, such as the UE’s movement mode and movement trend within the range specified by AF (such as area/slice/DNN/App ID) or within a specified time period; NWDAF can Provide prediction information of UE mobility, such as which UEs are about to move into the service area, which UEs are about to move out of the service area, which UEs will move more frequently, which UEs will become relatively fixed, etc. within the AF specified range or within the specified time period. .
  • NWDAF can provide statistical information on the UE's wireless channel conditions, such as in Which UEs receive better signal quality and which UEs receive average signal quality within the specified range of AF (such as area/slice/DNN/App ID) or within the specified time period; NWDAF can provide prediction information of UE wireless channel conditions, such as The connection status of which UEs will become relatively stable within the AF specified range or within the specified time period, and which UEs' connection stability will become poorer, etc.
  • NWDAF can provide statistical information on the wireless resource utilization of the UE, such as which UEs have higher performance in the range specified by the AF (such as area/slice/DNN/App ID) or within the specified time period. There are more available wireless time/frequency resources, which UEs are in busy cells and have fewer available wireless resources, etc.; NWDAF can provide prediction information of UE wireless resource utilization, such as which UEs are about to arrive within the range specified by AF or within the specified time period. There are more available wireless time/frequency resources, and the cells where the UEs are located are about to become busy (such as after work hours in the evening).
  • the first response message may be a notification message.
  • NWDAF sends a notification message to AF through the Nnwdaf_AnalyticsSubscription_Notify service operation.
  • the notification message carries: Analytics ID; network status information of N1 UEs.
  • the notification message can also be understood as the UE network status statistics or predicted analysis results (UE network status related analytics).
  • the notification message can also carry Collected UE network status information, which means that NWDAF will forward the original network status information obtained from AMF/OAM to AF.
  • the AF network element determines N2 UEs based on the network status information of the N1 UEs.
  • N2 UEs may be implemented with reference to the method described in S403, which will not be described again in this embodiment of the present application.
  • the AF network element can determine the N2 UEs only based on the network status information of the N1 UEs through data analysis or a specific algorithm.
  • the AF network element can also determine the N2 UEs based on the network status information of the N1 UEs combined with the application layer information of the N1 UEs, such as using the network status information of the UEs to optimize the existing application layer algorithm, or combining the network status of the UEs.
  • a new algorithm is designed for information and application layer information, and N2 UEs are determined by the new algorithm.
  • the network status information of N1 UEs is the statistical or predictive analysis result obtained by NWDAF analysis.
  • the AF network element Take the AF network element to determine N2 UEs based on the network status information of N1 UEs combined with the application layer information of N1 UEs.
  • the embodiments of this application provide specific implementation methods for determining N2 UEs as follows:
  • the AF network element has obtained the application layer information and network status information of 100 UEs, and is preparing to select 80 UEs to participate in the training of the federated learning model.
  • AF can score each UE according to its predicted connection state. For example, "0.9" represents the connected state, and "0.1" represents the idle state.
  • AF sets a weight for each parameter according to the importance of each parameter indicated by the network status information.
  • AF can Calculate the weighted scores of other UEs, then sort the weighted scores of all UEs, and select the 80 UEs with the highest scores as UEs participating in federated learning model training.
  • S505 The AF network element performs federated learning model training with N2 UEs.
  • the updated training of the federated learning model can be understood with reference to the description in Solution 1, which will not be described again in the embodiments of this application.
  • the dotted lines in Figure 5 illustrate the following S506 to S509, which reflect the update training process of the federated learning model.
  • the AF network element obtains the information of the UEs with abnormal network status information among the N2 UEs.
  • the AF network element subscribes to and obtains the network status information of N3 UEs from the AMF network element and/or OAM network element.
  • the candidate range includes the N3 UEs, and the N3 UEs do not include the N2 UEs.
  • N3 is a positive integer.
  • the AF network element determines N4 UEs based on the information of the UEs with abnormal network status information among the N2 UEs and the network status information of the N3 UEs.
  • the N4 UEs are used to participate in the update of the federated learning model. Training, N4 is a positive integer.
  • the AF requests and obtains the statistical or predicted analysis results related to the network status information of the candidate UE from the NWDAF, and applies it to the federated learning environment to facilitate the application side to use the network status information of the UE to choose to participate.
  • the UE trained or updated by the federated learning model can optimize the algorithm for selecting participants based on application layer information, thereby improving the efficiency of federated learning model training.
  • AF can also be called FL AF.
  • the method includes the following process.
  • the AF network element sends a second request message to the NWDAF network element, where the second request message is used to request recommended terminal devices within the candidate range that participate in the training of the federated learning model.
  • the second request message may be a subscription message.
  • the AF network element sends a subscription message to NWDAF through the Nnwdaf_AnalyticsSubscription_Subscribe service operation.
  • the subscription message also includes information indicating the candidate range, which may be an identifier of the candidate range, such as a range of UEs representing a designated network area, and a list of UE IDs representing a list of candidate terminal devices.
  • the subscription message may also carry a number range indication, such as the first indication information described in S401 or a range of UE number.
  • the subscription message may also carry a data validity indication or a joint validity indication as described in S401.
  • NEF controls the mapping relationship between AF identities and the Analytics IDs allowed to be obtained, as well as related inbound restrictions (i.e., limiting the Analytics IDs that AF can request) and outbound restrictions (i.e., limiting the Analytics IDs that can be notified to AF).
  • inbound restrictions i.e., limiting the Analytics IDs that AF can request
  • outbound restrictions i.e., limiting the Analytics IDs that can be notified to AF.
  • AMF first sends the network status information of the relevant UE to NEF. Then NEF sends it to the third-party AF.
  • the NWDAF network element obtains the network status information of the UE within the candidate range from the AMF network element and/or the OAM network element.
  • the NWDAF network element determines recommended N1 UEs based on obtaining network status information of UEs within the candidate range.
  • the UEs within the candidate range include the N1 UEs, and N1 is a positive integer.
  • the NWDAF network element performs statistical analysis based on obtaining the original network status information of UEs within the candidate range, and obtains statistical or predicted UE network status analysis results. Furthermore, the NWDAF network element can refer to the scoring strategy described in S504, set different weights for different network states of the UE, and then prioritize the UE with a high weighted score as the recommended UE based on the weighting and sorting.
  • the NWDAF network element sends a second response message to the AF network element, where the second response message includes information indicating the recommended N1 terminal devices.
  • the second response message is a notification message.
  • NWDAF can send a notification message to FL AF through the Nnwdaf_AnalyticsSubscription_Notify service operation.
  • the notification message can carry the following parameters: Analysis ID (Analytics ID).
  • Analytics ID Recommended UE Information; and the recommended UE list (Recommended UE list), used to indicate the N1 UEs recommended by NWDAF in S603, which can be indicated by a set of UE identifiers.
  • the NWDAF network element can also send the original network status information of the N1 UEs obtained in S602 to the AF.
  • the NWDAF network element can include Collected UE network status information in the notification message to represent the UE's original network status information.
  • the AF network element determines N2 UEs according to the second response message.
  • the N2 UEs are used to participate in the training of the federated learning model.
  • the N1 UEs include the N2 UEs, and N2 is a positive integer.
  • the AF network element can directly determine the N1 UEs recommended by the NWDAF as UEs participating in federated learning model training.
  • N1 is equal to N2.
  • the AF network element can select part of the N1 UEs based on the original network status information of the N1 UEs. Or all UEs participate in the training of the federated learning model. For example, the AF network element can select some UEs among the N1 UEs whose original network status information is better to participate in the training of the federated learning model, or the AF network element can select among the N1 UEs whose original network status information is greater than or equal to the first network status information threshold. The UE participates in the training of the federated learning model.
  • S606 The AF network element performs federated learning model training with N2 UEs.
  • the updated training of the federated learning model can be understood with reference to the description in Solution 1, which will not be described again in the embodiments of this application.
  • the dotted lines in Figure 6 illustrate the following S607 to S612, which reflect the update training process of the federated learning model.
  • the NWDAF network element obtains the information of the UE with abnormal network status information among the N2 UEs.
  • Figure 6 illustrates that the NWDAF network element obtains the information of the UE with abnormal network status information among the N2 UEs from the UPF network element.
  • the NWDAF may send second indication information to the UPF network element, where the second indication information is used to instruct the network status information of the N2 UEs to be monitored. Furthermore, when the UPF network element determines that there is a UE with abnormal network status information among the N2 UEs, it sends the information of the UE with abnormal network status information among the N2 UEs to the NWDAF network element.
  • the NWDAF network element subscribes to and obtains the network status information of N3 UEs from the AMF network element and/or OAM network element.
  • the candidate range includes the N3 UEs, and the N3 UEs do not include the N2 UEs.
  • N3 is a positive integer.
  • Figure 6 illustrates that the NWDAF network element obtains the network status information of N3 UEs from the AMF network element and/or the OAM network element.
  • the NWDAF network element determines a new recommended UE list based on the information of the UEs with abnormal network status information among the N2 UEs and the network status information of the N3 UEs.
  • the new recommended UE list includes multiple UEs.
  • the NWDAF network element may refer to the solution described in S407 to determine a new recommended UE list.
  • the new recommended UE list includes other UEs among the N2 UEs except the UEs with abnormal network status information, and some UEs among the N3 UEs.
  • the new recommended UE list includes some UEs among the N2 UEs and some UEs among the N3 UEs.
  • the new recommended UE list may include UEs with abnormal network status information.
  • the NWDAF network element sends information indicating a new recommended UE list to the AF network element.
  • the NWDAF can also send the network status information of each UE in the new recommended UE list to the AF network element.
  • the AF network element determines N4 UEs based on the new recommended UE list.
  • the N4 UEs are used to participate in the updated training of the federated learning model.
  • N4 is a positive integer.
  • the N4 UEs include some or all UEs in the recommended UE list.
  • the AF network element performs updated training of the federated learning model with N4 UEs.
  • the NWDAF can use the network status information of the UE to derive and determine the UE recommended to participate in federated learning model training or update.
  • AF can subscribe to recommended UEs from NWDAF, assisting the application side in selecting UEs to participate in federated learning model training or updating, and optimizing the algorithm for selecting participants based on application layer information, thereby improving the efficiency of federated learning model training. efficiency.
  • AF can also be called FL AF.
  • an embodiment of the present application provides a communication device 700 , which includes a processing module 701 and a communication module 702 .
  • the communication device 700 may be an AF network element, or may be a communication device applied to or matched with the AF network element, capable of implementing the communication method executed on the AF network element side; or, the communication device 700 may be an NWDAF network
  • the communication device 700 may be an AMF network element (or OAM network element, UPF network element, etc.), it can also be a communication device applied to or matched with the AMF network element and capable of implementing the communication method executed on the AMF network element side.
  • the communication module may also be called a transceiver module, a transceiver, a transceiver, or a transceiver device, etc.
  • the processing module may also be called a processor, a processing board, a processing unit, or a processing device.
  • the communication module is used to perform the sending and receiving operations of the relevant network elements in the above method.
  • the devices used to implement the receiving function in the communication module can be regarded as receiving units, and the devices used to implement the sending function in the communication module can be regarded as Considered as a sending unit, that is, the communication module includes a receiving unit and a sending unit.
  • the processing module 701 can be used to implement the processing function of the AF network element in the examples described in Figure 4, Figure 5 or Figure 6, and the communication module 702 can be used to implement the processing function of the AF network element in Figure 4 , the transceiver function of the AF network element in the example described in Figure 5 or Figure 6.
  • the communication device can also be understood with reference to the possible designs in the fourth aspect of the invention.
  • the processing module 701 can be used to implement the processing functions of the AMF or OAM network element in the examples described in Figure 4, Figure 5 or Figure 6, and the communication module 702 can be used to implement the processing functions of the AMF or OAM network element in Figure 4, Figure 5 or Figure 6. 5 or the sending and receiving functions of the AMF or OAM network element in the example shown in Figure 6.
  • the communication device can also be understood with reference to the possible design in the fifth aspect of the invention.
  • the processing module 701 can be used to implement the processing functions of the NWDAF network element in the examples described in Figure 4, Figure 5 or Figure 6, and the communication module 702 can be used to implement the processing functions of the NWDAF network element in Figure 4, Figure 5 or Figure 6
  • the sending and receiving functions of the NWDAF network element in the above example can also be understood with reference to the possible designs in the sixth aspect of the invention.
  • the aforementioned communication module and/or processing module can be implemented through a virtual module.
  • the processing module can be implemented through a software functional unit or a virtual device, and the communication module can be implemented through a software function or a virtual device.
  • the processing module or communication module can also be implemented by a physical device.
  • the communication device is implemented by a chip/chip circuit, the communication module can be an input/output circuit and/or a communication interface to perform input operations (corresponding to the aforementioned receiving operations) , output operation (corresponding to the aforementioned sending operation); the processing module is an integrated processor or microprocessor or integrated circuit.
  • each functional module in each example of the embodiment of the present application may be integrated into one In the processor, it can exist physically alone, or two or more modules can be integrated into one module.
  • the above integrated modules can be implemented in the form of hardware or software function modules.
  • an embodiment of the present application also provides a communication device 800.
  • the communication device 800 may be a chip or a chip system.
  • the chip system may be composed of chips, or may include chips and other discrete devices.
  • the communication device 800 can be used to implement the functions of any network element in the communication system described in the foregoing examples.
  • the communication device 800 may include at least one processor 810, which is coupled to a memory.
  • the memory may be located within the communication device.
  • the memory may be integrated with the processor.
  • the memory may also be located outside the communication device.
  • the communication device 800 may further include at least one memory 820.
  • the memory 820 stores the necessary computer programs, computer programs or instructions and/or data to implement any of the above examples; the processor 810 may execute the computer program stored in the memory 820 to complete the method in any of the above examples.
  • the communication device 800 may also include a communication interface 830, and the communication device 800 may interact with other devices through the communication interface 830.
  • the communication interface 830 may be a transceiver, a circuit, a bus, a module, a pin, or other types of communication interfaces.
  • the communication interface 830 in the communication device 800 can also be an input-output circuit, which can input information (or receive information) and output Information (or sending information)
  • the processor is an integrated processor or a microprocessor or an integrated circuit or a logic circuit, and the processor can determine the output information based on the input information.
  • the coupling in the embodiment of this application is an indirect coupling or communication connection between devices, units or modules, which may be in electrical, mechanical or other forms, and is used for information interaction between devices, units or modules.
  • the processor 810 may cooperate with the memory 820 and the communication interface 830.
  • the specific connection medium between the processor 810, the memory 820 and the communication interface 830 is not limited in the embodiment of the present application.
  • the bus 840 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc.
  • PCI peripheral component interconnect
  • EISA extended industry standard architecture
  • the bus can be divided into address bus, data bus, control bus, etc. For ease of presentation, only one thick line is used in Figure 8, but it does not mean that there is only one bus or one type of bus.
  • the processor may be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, a field programmable gate array or other programmable logic device, a discrete gate or transistor logic device, or a discrete hardware component, which may implement or Execute each method, step and logical block diagram disclosed in the embodiment of this application.
  • a general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods disclosed in conjunction with the embodiments of the present application can be directly implemented by a hardware processor for execution, or can be executed by a combination of hardware and software modules in the processor.
  • the memory may be a non-volatile memory, such as a hard disk drive (HDD) or a solid-state drive (SSD), etc., or it may be a volatile memory (volatile memory), such as Random-access memory (RAM).
  • Memory is, but is not limited to, any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer.
  • the memory in the embodiment of the present application can also be a circuit or any other device capable of realizing a storage function, used to store program instructions and/or data.
  • the communication device 800 can be applied to the AF network element.
  • the specific communication device 800 can be an AF network element, or can support the AF network element to implement the AF network element in any of the above-mentioned examples.
  • functional device The memory 820 stores computer programs (or instructions) and/or data that implement the functions of the AF network element in any of the above examples.
  • the processor 810 can execute the computer program stored in the memory 820 to complete the method executed by the AF network element in any of the above examples.
  • the communication interface in the communication device 800 can be used to interact with AMF, OAM or NWDAF network elements, send information to AMF, OAM or NWDAF network elements, or receive information from AMF, OAM or NWDAF network elements. information.
  • the communication device 800 can be applied to the NWDAF network element.
  • the specific communication device 800 can be the NWDAF network element, or can support the NWDAF network element to implement the NWDAF network element in any of the above-mentioned examples.
  • functional device The memory 820 stores computer programs (or instructions) and/or data that implement the functions of the NWDAF network element in any of the above examples.
  • the processor 810 can execute the computer program stored in the memory 820 to complete the method executed by the NWDAF network element in any of the above examples.
  • the communication interface in the communication device 800 can be used to interact with AMF, OAM or AF network elements, send information to AMF, OAM or AF network elements, or receive information from AMF, OAM or AF network elements. information.
  • the communication device 800 can be applied to an AMF/OAM network element.
  • the specific communication device 800 can be an AMF/OAM network element, or can support an AMF/OAM network element to achieve any of the above-related tasks.
  • the memory 820 stores computer programs (or instructions) and/or data that implement the functions of the AMF/OAM network element in any of the above examples.
  • Processor 810 executable memory 820 The stored computer program completes the method executed by the AMF/OAM network element in any of the above examples.
  • the communication interface in the communication device 800 can be used to interact with the AF network element or the NWDAF network element, send information to the AF network element or the NWDAF network element, or receive information from the AF network element or the NWDAF network. Yuan information.
  • embodiments of the present application provide a communication system, including UE, AF network element, NWDAF network element, AMF network element, and OAM network element.
  • UPF network elements are also included.
  • the AF network element, NWDAF network element, AMF network element, and OAM network element can implement the communication method provided in the examples shown in Figure 4, Figure 5, or Figure 6.
  • the technical solutions provided by the embodiments of this application can be implemented in whole or in part through software, hardware, firmware, or any combination thereof.
  • software When implemented using software, it may be implemented in whole or in part in the form of a computer program product.
  • the computer program product includes one or more computer instructions.
  • the computer program instructions When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part.
  • the computer may be a general-purpose computer, a special-purpose computer, a computer network, a UE, an AF network element, an NWDAF network element, an AMF network element, an OAM network element, or other programmable devices.
  • the computer instructions may be stored in or transmitted from one computer-readable storage medium to another, e.g., the computer instructions may be transferred from a website, computer, server, or data center Transmission to another website, computer, server or data center through wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) means.
  • the computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. that contains one or more available media integrated.
  • the available media may be magnetic media (eg, floppy disk, hard disk, tape), optical media (eg, digital video disc (digital video disc, DVD)), or semiconductor media, etc.
  • the examples may refer to each other.
  • the methods and/or terms between the method examples may refer to each other.
  • the functions and/or terms between the device examples may refer to each other.
  • Cross-references, for example, functions and/or terms between apparatus examples and method examples may refer to each other.

Landscapes

  • Engineering & Computer Science (AREA)
  • Signal Processing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Medical Informatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Databases & Information Systems (AREA)
  • Environmental & Geological Engineering (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Exchanges In Wide-Area Networks (AREA)
  • Mobile Radio Communication Systems (AREA)
  • Computer And Data Communications (AREA)

Abstract

本申请提供一种通信方法及装置，用于提升联邦学习的训练效率。该方法包括：应用功能网元发送第一请求消息，第一请求消息用于请求候选范围内的终端设备的网络状态信息；应用功能网元获取第一响应消息，第一响应消息包括N1个终端设备的网络状态信息，候选范围内的终端设备包括N1个终端设备，N1为正整数；应用功能网元根据N1个终端设备的网络状态信息，确定N2个终端设备，N2个终端设备用于参与联邦学习模型的训练，N1个终端设备包括N2个终端设备，N2为正整数。

Description

一种通信方法及装置
相关申请的交叉引用
本申请要求在2022年03月29日提交中华人民共和国知识产权局、申请号为202210324047.8、申请名称为“一种通信方法及装置”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请实施例涉及通信技术领域，尤其涉及一种通信方法及装置。
背景技术
联邦学习(federated learning，FL)是一个机器学习框架，能有效帮助多个用户在满足用户隐私保护、数据安全和政府法规的要求下，进行数据使用和机器学习建模。联邦学习作为分布式的机器学习范式，可以有效解决数据孤岛问题，在不共享用户数据的基础上进行联合建模，进而从技术上打破数据孤岛，实现人工智能(artificial intelligence，AI)协作。然而，现有的联邦学习的方法效率不高。
发明内容
本申请实施例提供一种通信方法及装置，以期提升联邦学习的效率。
第一方面，本申请实施例提供一种通信方法，包括：应用功能网元发送第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；所述应用功能网元获取第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；所述应用功能网元根据所述N1个终端设备的网络状态信息，确定N2个终端设备，所述N2个终端设备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
通过这样的设计，在选择参与联邦学习的终端设备时，引入对终端设备的网络状态信息的考虑，确保参与方的通信能力，能够提升联邦学习的效率。
在一种可能的设计中，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
在一种可能的设计中，N1的取值属于设定数量范围。可选的，所述第一请求消息可以包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
在一种可能的设计中，所述第一请求消息包括N1的取值范围。可选的，该N1的取值范围可以和前述设定数量范围相同也可以不相同。
通过这样的设计，便于数据提供方提供符合数量要求的终端设备的网络状态信息。
在一种可能的设计中，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。可以理解的是，应用功能网元也可以推导候选范围除N1个终端设备之外其他终端设备的网络状态信息小于第一网络状态信息阈值，通过这样的设计能够节省第一响应消息的信令开销。
在一种可能的设计中，所述应用功能网元根据所述N1个终端设备的网络状态信息，确定N2个终端设备，包括：所述应用功能网元根据所述N1个终端设备的网络状态信息以及所述N1个终端设备的应用层信息，确定所述N2个终端设备。通过这样的设计，应用侧可以利用UE的网络状态信息结合应用层信息选择参与联邦学习模型训练或更新的UE，实现对基于应用层信息选择参与方的算法的优化，从而提升联邦学习模型训练的效率。
在一种可能的设计中，还包括：所述应用功能网元获取所述N2个终端设备中网络状态信息异常的终端设备的信息；所述应用功能网元获取N3个终端设备的网络状态信息，所述候选范围包括所述N3个终端设备，所述N3个终端设备不包括所述N2个终端设备，N3为正整数；所述应用功能网元根据所述N2个终端设备中网络状态信息异常的终端设备的信息以及所述N3个终端设备的网络状态信息，确定N4个终端设备，所述N4个终端设备用于参与联邦学习模型的更新训练，N4为正整数，例如N4等于N2。
这样的设计，在已参与联邦学习的终端设备基础上，更新确定参与后续轮次联邦学习的终端设备，相较于每轮单独确定参与联邦学习的终端设备，更为快速便捷，从而提升联邦学习的效率。
在一种可能的设计中，所述N4个终端设备包括所述N2个终端设备中除网络状态信息异常的终端设备之外的其他终端设备。
在一种可能的设计中，所述应用功能网元获取所述N2个终端设备中网络状态信息异常的终端设备的信息，包括：所述应用功能网元向网络数据分析功能网元发送第二指示信息，所述第二指示信息用于指示对所述N2个终端设备的网络状态信息进行监控，所述第二指示信息包括第二网络状态信息阈值，所述第二网络状态信息阈值用于确定所述终端设备的业务流量信息是否异常；所述应用功能网元从所述网络数据分析功能网元获取所述N2个终端设备中网络状态信息异常的终端设备的信息。
在一种可能的设计中，所述通信方法还包括：所述应用功能网元与所述N2个终端设备进行所述联邦学习模型的训练。
第二方面，本申请实施例提供一种通信方法，可以应用于接入与移动性管理功能网元或者操作维护管理网元。以应用于接入与移动性管理功能网元为例，该通信方法包括：
接入与移动性管理功能网元获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
接入与移动性管理功能网元发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
对应第一方面，可以理解N1个终端设备的网络状态信息用于应用功能网元确定参与联邦学习模型训练的N2个终端设备。
在一种可能的设计中，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
在一种可能的设计中，N1的取值属于设定数量范围。可选的，所述第一请求消息可以包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
在一种可能的设计中，所述第一请求消息包括N1的取值范围。可选的，该N1的取值范围可以和前述设定数量范围相同也可以不相同。
在一种可能的设计中，所述N1个终端设备的网络状态信息大于或等于第一网络状态 信息阈值。
在一种可能的设计中，所述通信方法还包括：接入与移动性管理功能网元发送N3个终端设备的网络状态信息，所述候选范围包括所述N3个终端设备，所述N3个终端设备不包括所述N2个终端设备，N3为正整数。
第三方面，本申请实施例提供一种通信方法，可以应用于网络数据分析功能网元，该通信方法包括：
网络数据分析功能网元获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
所述网络数据分析功能网元从接入与移动性管理功能网元和/或操作维护管理网元中，获取所述候选范围内的终端设备的网络状态信息；
所述网络数据分析功能网元发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
对应第一方面，可以理解N1个终端设备的网络状态信息用于应用功能网元确定参与联邦学习模型训练的N2个终端设备。
在一种可能的设计中，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
在一种可能的设计中，N1的取值属于设定数量范围。可选的，所述第一请求消息可以包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
在一种可能的设计中，所述第一请求消息包括N1的取值范围。可选的，该N1的取值范围可以和前述设定数量范围相同也可以不相同。
在一种可能的设计中，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
在一种可能的设计中，还包括：网络数据分析功能网元接收来自应用功能网元的第二指示信息，所述第二指示信息用于指示对所述N2个终端设备的网络状态信息进行监控，所述第二指示信息包括第二网络状态信息阈值，所述第二网络状态信息阈值用于确定所述终端设备的业务流量信息是否异常；所述网络数据分析功能网元向所述应用功能网元发送所述N2个终端设备中网络状态信息异常的终端设备的信息。
第四方面，本申请实施例提供一种通信装置，可以应用于应用功能网元，包括：通信模块，用于发送第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；所述通信模块，用于获取第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；处理模块，用于根据所述N1个终端设备的网络状态信息，确定N2个终端设备，所述N2个终端设备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
在一种可能的设计中，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
在一种可能的设计中，所述第一请求消息包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
在一种可能的设计中，所述第一请求消息包括N1的取值范围。
在一种可能的设计中，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
在一种可能的设计中，所述处理模块，具体用于：
根据所述N1个终端设备的网络状态信息以及所述N1个终端设备的应用层信息，确定所述N2个终端设备。
在一种可能的设计中，所述通信模块，还用于获取所述N2个终端设备中网络状态信息异常的终端设备的信息；以及，获取N3个终端设备的网络状态信息，所述候选范围包括所述N3个终端设备，所述N3个终端设备不包括所述N2个终端设备，N3为正整数；所述处理模块，还用于根据所述N2个终端设备中网络状态信息异常的终端设备的信息以及所述N3个终端设备的网络状态信息，确定N4个终端设备，所述N4个终端设备用于参与联邦学习模型的更新训练，N4为正整数。
在一种可能的设计中，所述N4个终端设备包括所述N2个终端设备中除网络状态信息异常的终端设备之外的其他终端设备。
在一种可能的设计中，所述通信模块，还用于：向网络数据分析功能网元发送第二指示信息，所述第二指示信息用于指示对所述N2个终端设备的网络状态信息进行监控，所述第二指示信息包括第二网络状态信息阈值，所述第二网络状态信息阈值用于确定所述终端设备的业务流量信息是否异常；从所述网络数据分析功能网元获取所述N2个终端设备中网络状态信息异常的终端设备的信息。
在一种可能的设计中，所述处理模块，还用于与所述N2个终端设备进行所述联邦学习模型的训练。
第五方面，本申请实施例提供一种通信装置，可以应用于接入与移动性管理功能网元或者操作维护管理网元。该通信装置包括：
通信模块，用于获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
处理模块，用于确定候选范围内的终端设备的网络状态信息；
所述通信模块，还用于发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
对应第二方面，可以理解N1个终端设备的网络状态信息用于应用功能网元确定参与联邦学习模型训练的N2个终端设备。
在一种可能的设计中，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
在一种可能的设计中，N1的取值属于设定数量范围。可选的，所述第一请求消息可以包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
在一种可能的设计中，所述第一请求消息包括N1的取值范围。可选的，该N1的取值范围可以和前述设定数量范围相同也可以不相同。
在一种可能的设计中，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
在一种可能的设计中，所述通信装模块，还用于发送N3个终端设备的网络状态信息，所述候选范围包括所述N3个终端设备，所述N3个终端设备不包括所述N2个终端设备，N3为正整数。
第六方面，本申请实施例提供一种通信装置，可以应用于网络数据分析功能网元，该通信装置包括：
通信模块，用于获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
所述通信模块，还用于从接入与移动性管理功能网元和/或操作维护管理网元中，获取所述候选范围内的终端设备的网络状态信息；
处理模块，用于通过所述通信模块发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
对应第三方面，可以理解N1个终端设备的网络状态信息用于应用功能网元确定参与联邦学习模型训练的N2个终端设备。
在一种可能的设计中，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
在一种可能的设计中，N1的取值属于设定数量范围。可选的，所述第一请求消息可以包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
在一种可能的设计中，所述第一请求消息包括N1的取值范围。可选的，该N1的取值范围可以和前述设定数量范围相同也可以不相同。
在一种可能的设计中，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
在一种可能的设计中，通信模块，还用于：接收来自应用功能网元的第二指示信息，所述第二指示信息用于指示对所述N2个终端设备的网络状态信息进行监控，所述第二指示信息包括第二网络状态信息阈值，所述第二网络状态信息阈值用于确定所述终端设备的业务流量信息是否异常；向所述应用功能网元发送所述N2个终端设备中网络状态信息异常的终端设备的信息。
第七方面，本申请实施例提供一种通信方法，包括：
应用功能网元发送第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；
所述应用功能网元获取第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；
所述应用功能网元根据所述第二响应消息，确定N2个终端设备，所述N2个终端设备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
第八方面，本申请实施例提供一种通信方法，包括：
网络数据分析功能网元接收第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；
所述网络数据分析功能网元确定候选范围内推荐的参与联邦学习模型的训练的终端设备；
所述网络数据分析功能网元发送第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
在一种可能的设计中，所述网络数据分析功能网元确定候选范围内推荐的参与联邦学习模型的训练的终端设备，包括：
所述网络数据分析功能网元从接入与移动性管理功能网元和/或操作维护管理网元中，获取所述候选范围内的终端设备的网络状态信息；
所述网络数据分析功能网元根据所述候选范围内的终端设备的网络状态信息，确定所述推荐的N1个设备。
第九方面，本申请实施例提供一种通信装置，可以应用于应用功能网元，该通信装置包括：
通信模块，用于发送第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；
所述通信模块，用于获取第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；
处理模块，用于根据所述第二响应消息，确定N2个终端设备，所述N2个终端设备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
第十方面，本申请实施例提供一种通信装置，可以应用于网络数据分析功能网元，该通信装置包括：
通信模块，用于接收第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；
处理模块，用于确定候选范围内推荐的参与联邦学习模型的训练的终端设备；
所述通信模块，用于发送第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
在一种可能的设计中，所述处理模块，具体用于：通过所述通信模块从接入与移动性管理功能网元和/或操作维护管理网元中，获取所述候选范围内的终端设备的网络状态信息；根据所述候选范围内的终端设备的网络状态信息，确定所述推荐的N1个设备。
第十一方面，本申请实施例提供一种通信装置，所述通信装置包括处理器，用于实现上述第一方面所描述的方法。处理器与存储器耦合，存储器用于存储指令和数据，所述处理器执行所述存储器中存储的指令时，可以实现上述第一方面描述的方法。可选的，所述通信装置还可以包括存储器；所述通信装置还可以包括通信接口，所述通信接口用于该通信装置与其它设备进行通信，示例性的，通信接口可以是收发器、电路、总线、模块、管脚或其它类型的通信接口。
在一种可能的设备中，该通信装置包括：
存储器，用于存储程序或指令；
处理器，用于利用通信接口发送第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；以及获取第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；
所述处理器，还用于根据所述N1个终端设备的网络状态信息，确定N2个终端设备， 所述N2个终端设备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
第十二方面，本申请实施例提供一种通信装置，所述通信装置包括处理器，用于实现上述第二方面所描述的方法。处理器与存储器耦合，存储器用于存储指令和数据，所述处理器执行所述存储器中存储的指令时，可以实现上述第二方面描述的方法。可选的，所述通信装置还可以包括存储器；所述通信装置还可以包括通信接口，所述通信接口用于该通信装置与其它设备进行通信，示例性的，通信接口可以是收发器、电路、总线、模块、管脚或其它类型的通信接口。
在一种可能的设备中，该通信装置包括：
存储器，用于存储程序或指令；
处理器，用于利用通信接口获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；确定候选范围内的终端设备的网络状态信息；以及利用通信接口发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
第十三方面，本申请实施例提供一种通信装置，所述通信装置包括处理器，用于实现上述第三方面所描述的方法。处理器与存储器耦合，存储器用于存储指令和数据，所述处理器执行所述存储器中存储的指令时，可以实现上述第三方面描述的方法。可选的，所述通信装置还可以包括存储器；所述通信装置还可以包括通信接口，所述通信接口用于该通信装置与其它设备进行通信，示例性的，通信接口可以是收发器、电路、总线、模块、管脚或其它类型的通信接口。
在一种可能的设备中，该通信装置包括：
存储器，用于存储程序或指令；
处理器，用于利用通信接口获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；利用通信接口从接入与移动性管理功能网元和/或操作维护管理网元中，获取所述候选范围内的终端设备的网络状态信息；以及利用通信接口发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
第十四方面，本申请实施例提供一种通信装置，所述通信装置包括处理器，用于实现上述第七方面所描述的方法。处理器与存储器耦合，存储器用于存储指令和数据，所述处理器执行所述存储器中存储的指令时，可以实现上述第七方面描述的方法。可选的，所述通信装置还可以包括存储器；所述通信装置还可以包括通信接口，所述通信接口用于该通信装置与其它设备进行通信，示例性的，通信接口可以是收发器、电路、总线、模块、管脚或其它类型的通信接口。
在一种可能的设备中，该通信装置包括：
存储器，用于存储程序或指令；
处理器，用于利用通信接口发送第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；以及利用通信接口获取第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；
所述处理器，还用于根据所述第二响应消息，确定N2个终端设备，所述N2个终端设 备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
第十五方面，本申请实施例提供一种通信装置，所述通信装置包括处理器，用于实现上述第八方面所描述的方法。处理器与存储器耦合，存储器用于存储指令和数据，所述处理器执行所述存储器中存储的指令时，可以实现上述第八方面描述的方法。可选的，所述通信装置还可以包括存储器；所述通信装置还可以包括通信接口，所述通信接口用于该通信装置与其它设备进行通信，示例性的，通信接口可以是收发器、电路、总线、模块、管脚或其它类型的通信接口。
在一种可能的设备中，该通信装置包括：
存储器，用于存储程序或指令；
处理器，用于利用通信接口获取第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；确定候选范围内推荐的参与联邦学习模型的训练的终端设备；以及利用通信接口发送第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
第十六方面，本申请实施例提供了一种通信系统，包括如第四方面或第十一方面中所描述的通信装置；以及如第五方面或第十二方面所述的通信装置；或者，
包括如第四方面或第十一方面中所描述的通信装置；以及如第六方面或第十三方面所述的通信装置；或者，
包括如第四方面或第十一方面中所描述的通信装置、如第五方面或第十二方面所述的通信装置、以及如第六方面或第十三方面所述的通信装置。
第十七方面，本申请实施例提供了一种通信系统，包括如第九方面或第十四方面中所描述的通信装置；以及如第十方面或第十五方面所述的通信装置。
第十八方面，本申请实施例提供了一种通信系统，该通信系统包括应用功能网元，应用功能网元用于执行第一方面以及第一方面任一种可能的设计方案。该通信系统还可以包括网络数据分析功能网元，接入与移动性管理功能网元，操作维护管理网元中的一个或多个。其中，网络数据分析功能网元可以与所述应用功能网元、接入与移动性管理功能网元、或操作维护管理网元通信，接入与移动性管理功能网元或操作维护管理网元，可以与应用功能网元或网络数据分析功能网元通信。
第十九方面，本申请实施例还提供了一种计算机程序，当所述计算机程序在计算机上运行时，使得所述计算机执行上述第一方面至第三方面、第七方面、第八方面中任一方面提供的方法。
第二十方面，本申请实施例还提供了一种计算机程序产品，包括指令，当所述指令在计算机上运行时，使得计算机执行上述第一方面至第三方面、第七方面、第八方面中任一方面提供的方法。
第二十一方面，本申请实施例还提供了一种计算机可读存储介质，所述计算机可读存储介质中存储有计算机程序或指令，当所述计算机程序或者指令在计算机上运行时，使得所述计算机执行上述第一方面至第三方面、第七方面、第八方面中任一方面提供的方法。
第二十二方面，本申请实施例还提供了一种芯片，所述芯片用于读取存储器中存储的计算机程序，执行上述第一方面至第三方面、第七方面、第八方面中任一方面提供的方法。
第二十三方面，本申请实施例还提供了一种芯片系统，该芯片系统包括处理器，用于支持计算机装置实现上述第一方面至第三方面、第七方面、第八方面中任一方面提供的方法。在一种可能的设计中，所述芯片系统还包括存储器，所述存储器用于保存该计算机装置必要的程序和数据。该芯片系统可以由芯片构成，也可以包含芯片和其他分立器件。
附图说明
图1为一种5G网络架构示意图；
图2为一种横向联邦学习的数据分布示意图；
图3为一种横向联邦学习的模型训练示意图；
图4为本申请实施例提供的通信方法的流程示意图之一；
图5为本申请实施例提供的通信方法的流程示意图之一；
图6为本申请实施例提供的通信方法的流程示意图之一；
图7为本申请实施例提供的通信装置的结构示意图之一；
图8为本申请实施例提供的通信装置的结构示意图之一。
具体实施方式
参考图1，为本申请实施例所适用的5G网络架构示意图，图1所示的5G网络架构包括三部分，分别是终端设备部分、数据网络(data network，DN)部分和运营商网络部分。下面对其中的部分网元的功能进行简单介绍说明。
其中，运营商网络可包括以下网元中的一个或多个：鉴权服务器功能(Authentication Server Function，AUSF)网元、网络开放功能(network exposure function，NEF)网元、策略控制功能(Policy Control Function，PCF)网元、统一数据管理(unified data management，UDM)、统一数据库(Unified Data Repository，UDR)、网络存储功能(Network Repository Function，NRF)网元、接入与移动性管理功能(Access and Mobility Management Function，AMF)网元、会话管理功能(session management function，SMF)网元、无线接入网(Radio Access Network，RAN)设备、用户面功能(user plane function，UPF)网元以及网络数据分析功能(Network Data Analytics Function，NWDAF)网元等。上述运营商网络中，除无线接入网部分之外的部分可以称为核心网络部分。在一种可能的实现方法中，运营商网络中还包括应用功能(Application Function，AF)网元、操作维护管理(operation administration and maintenance，OAM)网元。
在具体实现中，本申请实施例中的终端设备，可以是用于实现无线通信功能的设备。其中，一些终端设备的举例包括：用户设备(user equipment，UE)、接入终端、终端单元、终端站、移动站、移动台、远方站、远程终端、移动设备、无线通信设备、终端代理或终端装置等。接入终端可以是蜂窝电话、无绳电话、会话启动协议(session initiation protocol，SIP)电话、无线本地环路(wireless local loop，WLL)站、个人数字助理(personal digital assistant，PDA)、具有无线通信功能的手持设备、计算设备或连接到无线调制解调器的其它处理设备、车载设备、可穿戴设备。终端设备可以是物联网中的终端设备、虚拟现实(virtual reality，VR)终端设备、5G网络或者未来演进的公共陆地移动网络(public land mobile network，PLMN)中的终端设备、增强现实(augmented reality，AR)终端设备、 工业控制(industrial control)中的无线终端、无人驾驶(self driving)中的无线终端、远程医疗(remote medical)中的无线终端、智能电网(smart grid)中的无线终端、运输安全(transportation safety)中的无线终端、智慧城市(smart city)中的无线终端、智慧家庭(smart home)中的无线终端，家用电器等。终端可以是移动的，也可以是固定的。
上述终端设备可通过运营商网络提供的接口(例如N1等)与运营商网络建立连接，使用运营商网络提供的数据和/或语音等服务。终端设备还可通过运营商网络访问DN，使用DN上部署的运营商业务，和/或第三方提供的业务。其中，上述第三方可为运营商网络和终端设备之外的服务方，可为终端设备提供其他数据和/或语音等服务。其中，上述第三方的具体表现形式，具体可根据实际应用场景确定，在此不做限制。
RAN作为接入网网元是运营商网络的子网络，是运营商网络中业务节点与终端设备之间的实施系统。终端设备要接入运营商网络，首先是经过RAN，进而可通过RAN与运营商网络的业务节点连接。本申请中的RAN设备，是一种为终端设备提供无线通信功能的设备，如提供终端设备和核心网络之间的连接，RAN设备也称为接入网设备。本申请中的RAN设备包括但不限于：基站(base station)、演进型基站(evolved NodeB，eNodeB)、发送接收点(transmission reception point，TRP)、5G中的下一代基站(g nodeB，gNB)、6G移动通信系统中的下一代基站、未来移动通信系统中的基站或无线保真(wireless fidelity，WiFi)系统中的接入节点、演进型节点B(evolved node B，eNB)、无线网络控制器(radio network controller，RNC)、节点B(node B，NB)、基站控制器(base station controller，BSC)、基站收发台(base transceiver station，BTS)、家庭基站(例如，home evolved nodeB，或home node B，HNB)、基带单元(baseBand unit，BBU)、发射点(transmitting point，TP)、移动交换中心等。
OAM网元，主要进行日常网络和业务的分析、预测、规划和配置、以及对网络及其业务的测试和故障管理等功能，OAM网元也可以称为网管。示例性，OAM可以和RAN交互，获取RAN侧的无线信道条件和无线资源利用率等信息。
AMF网元，主要为终端设备提供移动性管理、合法监听、或者接入鉴权/授权等功能。此外，还负责在UE与PCF间传递用户策略。
SMF网元，主要进行会话管理、PCF下发控制策略的执行、UPF的选择、UE互联网协议(internet protocol，IP)地址分配等功能。
UPF网元，作为和数据网络的接口UPF，完成用户面数据转发、基于会话/流级的计费统计，带宽限制等功能。
UDM网元，主要负责管理签约数据、用户接入授权等功能。
UDR，主要负责签约数据、策略数据、应用数据等类型数据的存取功能。
NEF网元，主要用于支持能力和事件的开放。
AF网元，主要传递应用侧对网络侧的需求，例如，服务质量(Quality of Service，QoS)需求或用户状态事件订阅等。AF可以是第三方功能实体，也可以是运营商部署的应用服务。
NEF网元，主要负责对外提供5G网络的能力和事件的开放，以及接收相关的外部信息。
PCF网元，主要负责针对会话、业务流级别进行计费、QoS带宽保障及移动性管理、UE策略决策等策略控制功能。
NRF网元，可用于提供网元发现功能，基于其他网元的请求，提供网元类型对应的网元信息。NRF还提供网元管理服务，如网元注册、更新、去注册以及网元状态订阅和推送等。
NWDAF网元，主要用于收集网络数据，例如收集来自终端设备、RAN设备、核心网、第三方业务服务器或者网管系统的相关数据等。NWDAF网元提供网络数据分析服务，可以输出数据分析结果，并向终端设备、RAN设备、核心网、第三方业务服务器或者网管系统提供数据分析结果。NWDAF可以利用机器学习模型进行数据分析，例如3GPP Release 17中NWDAF的功能被分解，包括数据收集功能(或者数据收集逻辑功能(data collection function))、模型训练功能(或者机器学习模型训练逻辑功能(machine learning model training logical function))以及模型推理功能(或者分析逻辑功能(analytics logical function))。示例性的，数据分析结果可协助网络选择业务的服务质量参数，或协助网络执行流量路由，或协助网络选择背景流量传输策略等。
图1中Nnwdaf、Nausf、Nnef、Npcf、Nudm、Naf、Namf、Nsmf、N1、N2、N3、N4，以及N6为接口序列号。这些接口序列号的含义可参见3GPP标准协议中定义的含义，在此不做限制。
需要说明的是，本申请实施例中，网络数据分析功能网元可以是图1所示的NWDAF网元，也可以是未来通信系统中具有本申请中NWDAF网元的功能的其它网元，移动性管理网元可以是图1所示的AMF网元，也可以是未来通信系统中具有本申请中AMF网元的功能的其它网元；用户面网元可以是图1所示的UPF网元，也可以是未来通信系统中具有本申请中UPF网元的功能的其它网元；会话管理功能网元可以是图1所示的SMF网元，也可以是未来通信系统中具有本申请中SMF网元的功能的其它网元；应用功能网元可以是图1所示的AF网元，也可以是未来通信系统中具有本申请中AF网元的功能的其它网元；接入网设备可以是图1所示的RAN设备，也可以是未来通信系统中具有本申请中RAN设备的功能的其它网元；操作维护管理网元可以是图1所示的OAM网元，也可以是未来通信系统中具有本申请中OAM网元的功能的其它网元。
为便于说明，本申请实施例中，以数据分析网元为NWDAF网元，移动性管理网元为AMF网元，用户面网元为UPF网元，应用功能网元为AF网元，操作维护管理网元为OAM网元，接入网设备为RAN设备为例进行说明。以及，以终端设备为UE为例进行说明。另外，本申请实施例中的NWDAF网元也可以简称为NWDAF,UPF网元也可以简称为UPF，AF网元也可简称为AF，OAM网元也可以简称为OAM。
为便于理解本申请实施例方案，下面先介绍与本申请实施例相关的技术。
联邦学习(Federated Learning，FL)是一个机器学习框架，在该框架中，节点之间无需涉及数据交互，而是传递训练中得到的中间结果，例如模型的参数或者梯度等能够表征模型的信息。即联邦学习可以在满足用户隐私保护、数据安全的要求，进行机器学习的建模，即训练AI模型。联邦学习作为分布式的机器学习范式，可以有效解决数据孤岛问题,让参与联邦学习的节点在不共享数据的基础上联合建模，从而在技术上打破数据孤岛,实现AI协作。
一般地，根据参与各方数据源分布的情况不同，联邦学习可以分为三类：横向联邦学习、纵向联邦学习、联邦迁移学习。其中，横向联邦学习指的是在多个数据集(或理解为 样本集)的用户特征重叠较多而用户重叠较少的情况下，把数据集按照横向(即用户维度)切分，并依据用户特征相同而用户不完全相同的部分数据进行模型的训练。纵向联邦学习指的是在多个数据集的用户重叠较多而用户特征重叠较少的情况下，把数据集按照纵向(即特征维度)切分，依据用户相同而用户特征不完全相同的部分数据进行模型的训练。联邦迁移学习指的是在多个数据集的用户与用户特征重叠都较少的情况下，不对数据进行切分，而是利用迁移学习克服数据或样本标签不足的情况。
下面以横向联邦学习为例，对联邦学习模型的训练过程进行详细说明。
图2为一种横向联邦学习的数据分布示意图，涉及运营商A和B的数据集。其中，运营商A和B的数据集在用户维度的交集较小，但在用户特征维度的交集较大。在进行横向联邦学习的模型训练时，可以从运营商A和B的数据集中挑选出相同特征的数据作为训练数据。例如，运营商A的数据集包含了(用户1，用户2，用户3，用户4)的数据，每个用户的数据包含特征(特征1，特征2，特征3，特征4，特征5)，运营商B的数据集包含了(用户4，用户5，用户6，用户7，用户8)的数据，其中每个用户的数据包含特征(特征2，特征3，特征4，特征5，特征6)。在进行横向联邦学习的模型训练时，可以从运营商A和B的数据集中挑选特征2，特征3，特征4和特征5的数据作为训练数据。
图3为一种横向联邦学习的模型训练示意图。步骤1，多个UE参与联邦学习模型的训练，利用本地数据计算本地模型对应的梯度，将梯度上报给用于训练联邦学习模型的服务器(server)。图3中示意出了由5G系统(5G System，5GS)进行中间结果传输的横向联邦学习模型训练过程，在步骤1中k个UE可通过5GS将梯度上报给服务器。步骤2，服务器进行梯度聚合，以确定或者更新联邦学习模型的参数。步骤3，服务器将更新后的联邦学习模型对应的梯度分发给各个参与方，如服务器通过5GS向各个UE发送更新后的联邦学习模型对应的梯度。步骤4，各UE根据步骤3获取的联邦学习模型对应的梯度，对自身的本地模型进行更新。重复步骤1～4，多轮迭代，直到联邦学习模型收敛。
在一些诸如海量物联网(massive internet of things，mIoT)的场景中，参与联邦学习模型训练的UE数量可能非常庞大，全部UE参与联邦学习模型训练会占用过多的网络带宽。由于网络带宽资源有限，在横向联邦学习模型训练参与方总数较大的情况下，每轮训练或更新联邦学习模型时，可以选出部分UE参与模型训练。UE选择结果的好坏将直接影响模型训练的效率。
在联邦学习FL环境中可能存在异构客户端，如不同UE的数据集优劣程度不同，不同UE的存储、计算和通信能力也可能相差较大，随机选择部分会加剧联邦学习FL环境中的异构性，不利于联邦学习模型的训练效率。示例性的，FL环境中的异构性主要包括：(1)数据异构性，各个UE设备中本地数据可能不是独立同分布(independent identically distributed)的；(2)存储、计算能力异构性，每个UE在存储、计算能力上可能相差较大，造成的结果就是不同UE在利用本地数据计算模型梯度时，计算时间相差较大；(3)通信能力异构性，不同UE的通信能力也可能相差较大，因而造成不同UE将本地计算完成的梯度上传到服务器的时间相差较大。不同UE的本地计算完成时间可能不同，将中间结果传输到服务器的时间也可能不同，而服务器又需要等到所有UE将中间结果均传输完成后才能更新联邦学习模型，所以会造成整体训练周期变长。由此可见，随机选择UE造成的异构性会影响联邦学习模型的训练效率。
已有技术采用一种非随机的客户端(或UE)选择方案，服务器如AF网元根据UE的 应用层信息，用户数据集的分布信息、用户的传输信息量等，选择参与联邦学习模型训练的UE。具体地，AF网元可以在每轮训练时，选择传输信息量较大的UE参与联邦学习模型的训练；或者，AF网元利用强化学习的经验驱动联邦学习框架，选择参与每轮训练的UE，以抵消不同UE之间非独立同分布数据引入的偏差；或者，AF网元利用强化学习建模，优先选取对提高模型准确度有较大作用的数据，又具备快速执行训练能力的UE。但这样的设计仅考虑UE的应用层信息，如果出现UE移出服务区、UE的无线信号变弱或UE与网络的连接中断等情况，导致联邦学习的模型训练和更新出错，仍然会影响联邦学习的效率。
基于此，本申请实施例提供一种联邦学习的参与方选择方案。在本申请实施例中，开放UE的网络状态信息，便于AF选择参与联邦学习模型的训练或更新的UE时，及时考虑UE的网络状态信息，避免选择网络状态信息不佳或者不适于联邦学习的UE，从而提升联邦学习的效率。
本申请实施例中，UE的网络状态信息用于指示如下一个或多个参数：UE的连接状态，如UE的连接状态为连接态、空闲态或去激活态；UE的可达性，如UE可达或者UE不可达；UE的移动性，如UE的移动模式或者移动趋势等；UE的无线信道条件，或描述为UE与RAN设备之间的无线信道的状况，如采用UE与RAN设备之间无线信道的信号干扰噪声比(signal to interference plus noise ratio，SNR)、UE接收参考信号的参考信号接收功率(reference signal received power，RSRP)或者参考信号接收质量(reference signal received quality，RSRQ)表现无线信道条件；UE对应的无线资源利用率，或描述为UE接入的RAN侧的无线资源利用率。
下面通过方案一至方案三，对本申请实施例结合UE的网络状态信息，确定参与联邦学习的UE的技术方案进行详细说明。
方案一
参见图4示意一种通信方法，该方法包括如下流程。
S401，AF网元发送第一请求消息，所述第一请求消息用于请求候选范围内的UE的网络状态信息。
具体地，该候选范围可以是指定的网络区域或者终端设备候选列表。
其中，一些指定的网络区域举例如下：例如，该指定的网络区域可以为一个或多个网络切片，一个网络切片由单网络切片选择辅助信息(single network slice selection assistance，S-NSSAI)标识。例如，该指定的网络区域可以为一个或多个跟踪区，一个跟踪区(tracking area，TA)对应一个跟踪区标识(tracking area identity，TAI)。例如，该指定的网络区域可以为一个或多个小区(cell)，一个小区对应一个小区标识(cell identity，cell ID)。例如，该指定的网络区域可以对应一个或多个数据网络，一个数据网络由数据网络名(data network name，DNN)标识，指定的网络区域内的UE指的是接入该数据网络的UE。例如，该指定的网络区域可以对应一个或多个应用，一个应用对应一个应用标识(application identity，application ID)，指定的网络区域内的UE指的是正在运行一个或多个application ID所标识的应用的UE。
可以理解的是，实际通信环境中UE可能发生移动，不同时段前述指定的网络区域内的UE可能不同。此外，指定的网络区域也可以是上述标识的组合。例如，指定的网络区域内的UE指的可以是处于TA1标识的区域，并且处于S-NSSAI1标识的切片内的UE。例 如，指定的网络区域内的UE指的可以是处于cell2标识的小区中，并且接入到DNN3标识的数据网络内的UE。
关于终端设备候选列表，可以理解为预配置的一组UE，对应一组UE标识(UE group ID或称为list of UE IDs)。例如，UE的网际协议(internet protocol，IP)地址列表，表示为list of UE IP；或者，用户永久标识(subscription permanent identifier，SUPI)列表，表示为list of SUPI；或者，一般公共订阅标识符(generic public subscription identifier，GPSI)列表，表示为list of GPSI。
具体地，AF网元可以根据自身希望获取的网络状态信息，向能够提供UE的网络状态信息的网元发送第一请求消息。例如，如果AMF网元可以提供AF网元希望获取的网络状态信息，AF网元可以向AMF网元发送第一请求消息；如果OAM网元可以提供AMF网元希望获取的网络状态信息，AF网元可以向OAM网元发送第一请求消息。以针对一个UE而言，AF希望获取的网络状态信息来自于AMF以及OAM为例，图4在S401中具体示意出了：S401a，AF网元向AMF发送第一请求消息；以及S401b，AF网元向OAM发送第一请求消息。
具体地，第一请求消息可以为订阅消息。例如AF可以通过Namf_EventExposure Subscribe服务操作向AMF发送订阅消息，以获取候选范围内UE的网络状态信息，如UE的连接状态、UE的可达性信息、UE的移动性信息等。该订阅消息可以携带如下信息：事件标识(Event ID)以及用于指示候选范围的信息。Event ID用于标识订阅的事件类型，表示AF想要从AMF订阅并获取UE的网络状态信息(的具体类型)。例如Event ID＝UE reachability，表示AF从AMF订阅UE的网络状态信息包括UE的可达性信息，如Event ID＝Location Reporting，表示从AF从AMF订阅UE的网络状态信息包括UE的位置变化信息。用于指示候选范围的信息可以为候选范围的标识，如以range of UEs表示指定的网络区域，以list of UE IDs表示候选终端设备列表。
可选的，AF网元可以决定请求网络请求信息对应终端设备的数量，如第一请求消息可以具体请求候选范围内部分或全部UE的网络状态信息。一种可选的实施方式中，AF可以在第一请求消息中包括第一指示信息，该第一指示信息用于指示AMF/OAM反馈网络状态信息对应UE数量需要位于设定数量范围之内，以N1表示AMF/OAM反馈网络状态信息对应的UE数量，N1的取值位于设定数量范围之内。示例性的，该设定数量范围可以为一个数量阈值，如1000，表示N1≤1000，N1为正整数。或者，该设定数量范围也可以为一个取值区间，如[200,3000]，N1为[200,3000]中的一个正整数。或者，一种可选的实施方式中，AF也可以向AMF/OAM指示其所需网络状态信息对应的UE数量范围，如AF在第一请求消息中包括UE数量范围(range of UE number)即N1的取值范围。可选的，这里描述的N1的取值范围可以与前述设定数量范围相同或者不同。或者，如果第一请求消息中不包括第一指示信息以及N1的取值范围，表示AF希望获取候选范围内全部UE的网络状态信息，或者表示AF不对获取多少UE的网络状态信息进行限定，进一步如果候选范围为候选终端设备列表，AF请求的可以是候选终端设备列表中的全部UE的网络状态信息。
可选的，AF网元可以决定请求有效时间范围内UE的网络状态信息。例如，AF可以在第一请求消息中包括数据有效性指示，该数据有效性指示用于指示AF希望获取的有效数据对应的有效时间范围，包括该数据有效性指示的第一请求消息具体用于请求候选范围的UE在有效时间范围内的网络状态信息。示例性的，该有效时间范围可以为一个时间范 围，指示数据提供网元(AMF/OAM)在该时间范围内获取的网络状态信息为有效数据。例如该时间范围为本日11点到12点，则该日11点之前或者12点之后收集的数据均是无效数据。或者，该有效时间范围可以为一个截止时间点，指示数据提供网元(AMF/OAM)在该截止时间点之前反馈的数据才是有效的。
可选的，AF网元也可以在第一请求消息中添加一个有效性联合指示，比如，有效性联合指示用于指示数据提供网元在一定时间范围内收集的数据并且在截止时间点之前发给数据获取网元的数据才是有效数据。
可选的，AF网元也可以在第一请求消息中添加用于表征AF网元对请求的UE的性能有要求的性能指示信息，例如性能指示信息用于指示AF网元想要请求的UE的网络状态信息大于或等于第一网络状态信息阈值。该第一网络状态信息阈值可以预先定义的或者AF自行决定的，本申请实施例对此不予限制。可选的，第一请求消息中可以包括该第一网络状态信息阈值。
此外，当AF是第三方应用功能时，出于安全性考虑，AF和AMF之间的交互需要经过NEF。NEF控制AF标识和允许获取的Event ID之间的映射关系，以及相关的入站限制(即限制AF可以请求的Event ID)和出站限制(即限制可以向AF通知的Event ID)。例如，当第三方AF需要从AMF订阅网络状态信息时，第三方AF需要先将订阅消息发送给NEF，再由NEF向AMF发出订阅消息，后续AMF先将相关UE的网络状态信息发送给NEF，再由NEF发送给第三方AF。
类似地，AF可以向OAM发送订阅消息，以获取相关UE的网络状态信息，如无线信道条件、无线资源利用率等。需要注意的是，AF可以直接从OAM获取数据。AF也可以通过NWDAF或者其他5GC NF从OAM获取数据，例如，AF先向NWDAF发送订阅消息，用于订阅OAM处的RAN侧数据，然后NWDAF从OAM订阅并获取相关数据，最后NWDAF再将获取的数据通过服务化接口(如Nnwdaf_AnalyticsSubscription_Notify服务操作)通知给AF。
S402，AF网元获取第一响应消息，所述第一响应消息包括N1个UE的网络状态信息。
其中，N1个UE包含于所述候选范围，或描述为所述候选范围内的UE包括所述N1个UE，N1个UE表示所述候选范围内的部分或全部UE。具体地，N1个UE的定义可以参照S401中的描述理解，如对应AF网元发送第一请求消息中包括第一指示信息时，N1的取值位于设定数量范围；如对应AF网元发送第一请求消息指示数量范围时，N1的取值位于AF指示的该数量范围；如AF发送第一请求消息中未携带第一指示信息以及指示数量范围时，N1个UE包括所述候选范围内数据提供方能够采集到网络状态信息的全部UE；又如AF发送第一请求消息中携带性能指示信息和/或第一网络状态信息阈值时，N1个UE的网络状态信息大于或等于第一网络状态信息阈值。具体地，对应S401中的描述，该第一响应消息可以来自于AMF和/或OAM。示例性，图4在S402中示意出了S402a，AMF网元向AF网元发送第一响应消息；以及，S402b，OAM网元向AF网元发送第一响应消息。作为示例，图4中仅示意出了AMF/OAM直接向AF网元发送第一响应消息，而有关AMF/OAM通过中间网元(如NEF、NWDAF等)间接向AF网元发送第一响应消息的方案可参照S401的描述理解，本申请实施例对此不再进行赘述。
具体地，对应S401第一请求消息中携带的Event ID，AMF网元/OAM网元可以在发送的第一响应消息中携带该Event ID，指示第一响应消息中UE的网络状态信息的类型， AMF网元/OAM网元在第一响应消息中包括的网络状态信息为Event ID对应类型的网络状态信息的取值，或还可以描述为AMF网元/OAM网元获取的N1个UE的原始网络状态信息。例如，网络状态信息的类型包括无线资源利用率时，第一响应消息中携带UE对应的无线资源利用率可以为0.6等具体的取值。
S403，AF网元根据所述N1个UE的网络状态信息，确定N2个UE。
其中，所述N2个UE用于参与联邦学习模型的训练，所述N2个UE包含于N1个UE，或描述为所述N1个UE包括所述N2个UE，N2为正整数，N2小于或者等于N1。
具体地，AF网元可以仅根据N1个UE的网络状态信息，通过数据分析或者特定的算法确定N2个UE。或者，AF网元也可以根据N1个UE的网络状态信息结合N1个UE的应用层信息，确定N2个UE，如利用UE的网络状态信息优化已有的应用层算法，或者结合UE的网络状态信息和应用层信息设计一个新的算法，并由新的算法确定N2个UE。
以AF网元也可以根据N1个UE的网络状态信息结合N1个UE的应用层信息，确定N2个UE为例，本申请实施例如下提供确定N2个UE的具体实施方式：
假设N1为100，N2为80。即AF网元获取了100个UE的应用层信息和网络状态信息，并且准备从中选出80个UE参与联邦学习模型的训练。首先，AF网元可以根据UE的应用层信息和设定的应用层算法给每个UE打分，得到100个UE对应的应用层打分：UE1_app＝0.7,UE2_app＝0.5,UE3_app＝0.9,…UE100_app＝0.75。AF可以根据UE的连接状态给每个UE打分，如“1”代表连接态，“0”代表空闲态，得到100个UE对应的连接状态打分：UE1_CMstate＝1,UE2_CMstate＝0,UE3_CMstate＝0,…UE100_CMstate＝1。AF根据UE的可达性信息给每个UE打分，如“1”代表可达，“0”代表不可达，可以理解的是处于空闲态的UE可能是可达的，得到100个UE对应的可达性打分：如UE1_reachable＝1,UE2_reachable＝1,UE3_reachable＝0,…UE100_reachable＝1。AF根据UE的移动性信息给每个UE打分，如AF根据UE所处的位置和移动速度等计算该UE对应移动性得分，得到100个UE对应的移动性打分：UE1_mobility＝0.65,UE2_mobility＝0.7,UE3_mobility＝0.3,…UE100_mobility＝0.95。AF根据UE的无线信道条件给每个UE打分，得到100个UE对应的无线信道条件打分：UE1_radiochannel＝0.85,UE2_radiochannel＝0.5,UE3_radiochannel＝0.6,…UE100_radiochannel＝0.7。AF根据UE的无线资源利用率给每个UE打分，得到100个UE对应的无线资源利用率打分：UE1_radioresource＝0.8,UE2_radioresource＝0.4,UE3_radioresource＝0.7,…UE100_radioresource＝0.55。
可以理解的是，前述对于各个参数打分的取值也可以根据实际情况设定，例如针对UE的连接状态的打分，除采用“1”代表连接态，“0”代表空闲态之外，也可以采用“0.9”代表连接态，“0.1”代表空闲态。本申请实施例对此不予限制。
其次，AF根据网络状态信息指示每个参数的重要程度给每个参数设定一个权重，比如，设定应用层信息的权重为UE_app_weight＝0.3，连接状态的权重为UE_CMstate_weight＝0.2,可达性的权重为UE_reachable_weight＝0.1，移动性的权重为UE_mobility_weight＝0.2，无线信道条件的权重为UE_radiochannel_weight＝0.1，无线资源利用率的权重为UE_radioresource_weight＝0.1。
最后，AF可以根据得分和权重计算每个UE的加权得分，如UE1_score＝0.7*0.3+1*0.2+1*0.1+0.65*0.2+0.85*0.1+0.8*0.1＝0.805，类似地，AF可以计算出其他UE的加权得分，然后将所有UE的加权得分排序，从中选出得分最高的80个UE， 作为参与联邦学习模型训练的UE。
S404，AF网元与N2个UE进行联邦学习模型的训练。
具体地，可参照相关技术中描述的步骤1～步骤4实施，本申请实施例对此不再进行赘述。
一种可选的实施方式中，对于联邦学习模型的多轮训练，每一轮均可以按照前述S401～S404确定每一轮参与模型训练的N2个UE；或者，每隔一定轮次可以按照前述S401～S404确定该轮参与模型训练的N2个UE。例如记联邦学习模型进行I轮训练达到收敛条件，AF可以参照S401～S403确定参与I轮中第i轮模型训练的UE。i取遍1至I的正整数。或者，i为1至I中的部分正整数，例如i的每两个相邻取值之间的差相同，如I为6，i取值为1，3，5。可以理解的是，当I为6，i取值为1，3，5，代表着第2轮和第1轮中参与模型训练的UE保持一致，第4轮和第3轮中参与模型训练的UE保持一致，以及第6轮和第5轮中参与模型训练的UE保持一致。又如不限定i的每两个相邻取值之间的差相同，i的取值可以是随机的或者由相关算法确定的，如I为6，i取值可以为1，3，4，6。可以理解的是，当I为6，i取值为1，3，4，6，代表着第2轮和第1轮中参与模型训练的UE保持一致，第5轮和第4轮中参与模型训练的UE保持一致。
另一种可选的实施方式中，对于联邦学习模型的多轮训练，从第二轮开始的每一轮均可以在上一轮确定参与模型训练的UE基础上，更新确定本轮参与模型训练的UE。例如，第一轮的模型训练可按照前述S401～S404实施，后续轮次的模型训练可以参照如下S405～S408实施。可以理解的是，假设将前述S401～S404记作第i轮模型训练中确定参与UE的方式，则图4中以虚线示意的S405～S408可以理解为第i+1轮模型训练中确定参与UE的方式，i为正整数。
S405，AF网元获取所述N2个UE中网络状态信息异常的UE的信息。
示例性，参见图4示意出了S405a，AF网元可以向NWDAF网元发送第二指示信息，所述第二指示信息用于指示对所述N2个UE的网络状态信息进行监控。S405b，NWDAF网元从UPF网元获取N2个UE的网络状态信息。S405c，NWDAF网元确定所述N2个UE中网络状态信息异常的UE。S405d，NWDAF网元向AF网元发送所述N2个UE中网络状态信息异常的UE的信息。
具体地，网络状态信息异常的UE的信息可以包括异常指示和网络状态异常信息。其中，该异常指示用于指示网络状态信息异常的类型，如业务流量异常；网络状态异常信息，可以包括业务流量异常信息，例如UE的服务质量(quality of service，QoS)信息，UE的QoS信息可以表征UE异常，如UE对应的信号传输时延过大或过小，如UE对应的丢包率过大等。可选的，所述第二指示信息可以包括第二网络状态信息阈值，所述第二网络状态信息阈值用于确定UE的网络状态信息是否异常。
具体地，在S405a中的第二指示信息可以采用订阅请求实现。例如，AF可以通过Nnwdaf_AnalyticsSubscription_Subscribe服务操作向NWDAF发出订阅请求，该订阅请求中携带如下参数：
分析标识(Analytics ID)，用于标识订阅的分析类型，本申请实施例中Analytics ID可以为"QoS information of abnormal UEs"，表示获取异常UE(如时延过大或过小UE)的QoS信息。
分析报告的目标(Target of analytics reporting)，如NWDAF要收集并分析N2个UE 的数据，可选的将该N2个UE的标识进行指示，将N2个UE的标识记作UE列表2(UE ID list2)，Target of analytics reporting＝UE ID list2。
通知指示(Notification Indication)，用于指示NWDAF反馈分析结果的条件，如可以是周期性反馈，或者基于阈值反馈，如UE ID list2中UE的时延方差大于一定阈值时反馈。可以理解的是，通知指示作为一个可选项，前述订阅请求中可以包括该通知指示也可以不包括该通知指示。
用于指示NWDAF通知的内容(Notification Threshold)的信息，该信息可以对应前述第二网络状态信息阈值，例如该第二网络状态信息阈值对应一个具体的阈值(如100ms)，指示NWDAF需要将时延大于该阈值的UE的QoS信息反馈给AF；或者，该第二网络状态信息阈值也可以对应一组阈值表示的组阈值范围，如大于100ms或者小于10ms，指示NWDAF需要将时延在该组阈值范围内的UE的QoS信息反馈给AF。可以理解的是，Notification Threshold作为一个可选项，前述订阅请求中可以包括该通知指示也可以不包括该Notification Threshold。
在S405b中，NWDAF网元可以通过Nupf_EventExposure_Subscribe服务操作向UPF网元订阅并获取N2个UE的QoS信息，UPF网元完成信息收集后，通过Nupf_EventExposure_Notify服务操作将收集的QoS信息通知给NWDAF网元。可选的，NWDAF网元也可以通过SMF网元向UPF网元订阅并获取N2个UE的QoS信息。
在S405c中，如果S405a中AF发送的订阅请求中包括第二网络状态信息阈值(如Notification Threshold)，NWDAF网元可以根据该第二网络状态信息阈值，判断N2个UE的网络状态信息是否异常，从而确定N2个UE中网络状态信息异常的UE。或者，如果S405a中AF发送的订阅请求中不包括第二网络状态信息阈值，NWDAF网元可以根据预配置的内部算法，确定N2个UE中网络状态信息异常的UE。
在S405d中，NWDAF网元可以通过Nnwdaf_AnalyticsSubscription_Notify服务操作向AF发送通知消息，以反馈N2个UE中网络状态信息异常的UE的信息。具体地，对应S405a订阅请求中携带的参数，Nnwdaf_AnalyticsSubscription_Notify服务操作中可以包括如下参数：Analytics ID；Target of analytics reporting＝UE ID list2；异常UE的QoS信息(QoS information of abnormal UEs)。此外可选的，如果S405a订阅请求中携带Notification Indication，NWDAF网元可以按照Notification Indication将异常UE的QoS信息反馈给AF网元，例如根据Notification Indication确定周期性反馈，也可以是根据Notification Indication确定基于阈值反馈。或者，如果S405a订阅请求中未携带Notification Indication，NWDAF网元也可以自己确定何时发送通知消息以实现QoS信息的反馈，例如当NWDAF网元推理得到异常UE后可以立即将异常UE的QoS信息反馈给AF。
此外可选的，AF网元也可以不通过NWDAF网元，直接从UPF网元订阅并获取所述N2个UE中网络状态信息异常的UE的信息。本申请实施例对此不予限制。
S406，AF网元获取N3个UE的网络状态信息，所述候选范围包括所述N3个UE，所述N3个UE不包括所述N2个UE，N3为正整数。
示例性的，图4中示意出AF网元可以从AMF网元和OAM网元中订阅并获取N3个UE的网络状态信息。可以理解的是，N3个UE指的是候选范围内未参与过S404中描述的联邦学习模型训练过程的UE。
一种可选的实施方式中，N3的取值可以由AF决定，例如AF可以根据S403中UE 的加权得分，从未参与联邦学习模型训练的UE中选出得分较高的前N3个UE，或者AF从未参与联邦学习模型训练的UE中随机选取的一组UE。AF可以在向AMF和/或OAM发送订阅消息中包括用于指示N3的取值范围的信息，例如以UE标识列表1(UE ID list1)指示N3个UE，AF可以在向AMF和/或OAM发送订阅消息中包括UE ID list1。
另一种可选的实施方式中，可以预先定义N3对应的数量范围，或者由AF在订阅消息中指示N3的取值范围，则AMF和/或OAM可以根据该预定义N3对应的数量范围或AF指示的N3取值范围，向AF反馈对应数量的UE的网络状态信息，具体的实施方式可以参见S401～S402执行，本申请实施例对此不再进行赘述。
S407，AF网元根据所述N2个UE中网络状态信息异常的UE的信息以及所述N3个UE的网络状态信息，确定N4个UE，该所述N4个UE用于参与联邦学习模型的更新训练，N4为正整数。
一种可选的实施方式中，AF网元可以参照S403中描述的打分策略，根据N3个UE的网络状态信息，确定N3个UE的加权得分，然后根据N3个UE的加权得分，从N2个UE中网络状态信息非异常的UE和N3个UE中，确定N4个UE。例如，N4个UE包括所述N2个UE中除网络状态信息异常的N5个UE之外的其他UE以及N3个UE中加权得分最高的N6个UE，N5，N6为正整数。
具体地，参与不同轮次联邦学习的UE的数量可以相同，即N4等于N2。该情况下，N6可以等于N5。
示例性的，假设S405中描述的UE ID list2包含80个UE，80个UE中10个UE的网络状态信息异常。S406中描述的UE ID list1对应包含20个UE的网络状态信息。AF可以利用和S403中相同的算法，对UE ID list1中UE的应用层信息和网络状态信息进行打分，进而得到UE ID list1中每个UE的加权得分。AF可以选择UE ID list2中非异常的70个UE和UE ID list1中加权得分最高的10个UE，作为参与联邦学习模型更新训练的N4个UE。
或者，参与不同轮次联邦学习的UE的数量可以不相同，即N4不等于N2。该情况下，N6可以不等于N5。
示例性的，假设S405中描述的UE ID list2包含80个UE，80个UE中10个UE的网络状态信息异常。S406中描述的UE ID list1对应包含20个UE的网络状态信息。AF可以利用和S403中相同的算法，对UE ID list1中UE的应用层信息和网络状态信息进行打分，进而得到UE ID list1中每个UE的加权得分。AF可以选择UE ID list2中非异常的70个UE和UE ID list1中加权得分最高的15个UE，作为参与联邦学习模型更新训练的N4个UE，此时N4大于N2；或者AF可以选择UE ID list2中非异常的70个UE和UE ID list1中加权得分最高的5个UE，作为参与联邦学习模型更新训练的N4个UE，此时N4小于N2。
另一种可选的实施方式中，AF网元可以参照S403中描述的打分策略，根据N3个UE的网络状态信息，确定N3个UE的加权得分，然后根据N3个UE的加权得分和N2个UE，确定N4个UE。该实施方式中，N4个UE可能包括所述N2个UE中网络状态信息异常的部分UE。
示例性的，假设S405中描述的UE ID list2包含80个UE，80个UE中30个UE的网络状态信息异常。S406中描述的UE ID list1对应包含20个UE的网络状态信息。AF可 以利用和S403中相同的算法，对UE ID list1中UE的应用层信息和网络状态信息进行打分，进而得到UE ID list1中每个UE的加权得分。假设UE ID list1对应包含20个UE中15个UE的加权得分大于或等于0.6。AF可以选择UE ID list2中非异常的50个UE和UE ID list1中加权得分超过0.6的UE(如15个UE)以及UE ID list2中异常UE中时延差最小的15个UE，作为参与联邦学习模型更新训练的N4个UE。其中，时延差指的是异常UE的实际信号传输时延与信号传输时延对应的第二网络状态信息阈值之间的差值的绝对值，此情况下N4等于N2。当然可以理解的是，N4也可以不等于N2，本申请实施例对此不予限制。
S408，AF网元与N4个UE进行联邦学习模型的更新训练。
具体地，可参照相关技术中描述的步骤1～步骤4实施，本申请实施例对此不再进行赘述。
在本申请实施例提供的上述方案一中，AF网元向5GC NF(如AMF)网元或OAM网元请求并获取候选UE的网络状态信息，应用于联邦学习环境，便于应用侧利用UE的网络状态信息选择参与联邦学习模型训练或更新的UE，实现对基于应用层信息选择参与方的算法的优化，从而提升联邦学习模型训练的效率。此外可以理解的是，在联邦学习环境中，AF也可以称为FL AF。
方案二
参见图5示意一种通信方法，该方法包括如下流程。
S501，AF网元向NWDAF网元发送第一请求消息，所述第一请求消息用于请求候选范围内的UE的网络状态信息。
关于候选范围的定义可参照S401中的描述理解，本申请实施例对此不再进行赘述。
具体地，第一请求消息可以为订阅消息。例如AF网元通过Nnwdaf_AnalyticsSubscription_Subscribe服务操作向NWDAF发送订阅消息，订阅消息中携带的参数包括分析标识(Analytics ID)。其中，Analytics ID＝UE Network Status Information，表示AF网元想要从NWDAF订阅并获取经由NWDAF分析得到的UE的网络状态信息，例如UE的网络状态信息与设定阈值或其他UE的网络状态信息之间的对比分析结果。订阅消息中还包括用于指示候选范围的信息可以为候选范围的标识，如以range of UEs表示指定的网络区域，以list of UE IDs表示候选终端设备列表。
可选的，该订阅消息中还可以携带数量范围指示，如S401描述的第一指示信息或者UE数量范围(range of UE number)。可选的，该订阅消息中还可以携带如S401描述的数据有效性指示或者有效性联合指示。可选的，该订阅消息中还可以携带如S401描述的性能指示信息和/或第一网络状态信息阈值。
此外，当AF是第三方应用功能时，出于安全性考虑，AF和NWDAF之间的交互需要经过NEF。NEF控制AF标识和允许获取的Analytics ID之间的映射关系，以及相关的入站限制(即限制AF可以请求的Analytics ID)和出站限制(即限制可以向AF通知的Analytics ID)。例如，当第三方AF需要从NWADF订阅网络状态信息时，第三方AF需要先将订阅消息发送给NEF，再由NEF向NWADF发出订阅消息，后续AMF先将相关UE的网络状态信息发送给NEF，再由NEF发送给第三方AF。
S502，NWDAF网元从AMF网元和/或OAM网元获取候选范围内的UE的网络状态 信息。
具体地，NWDAF可以采用订阅的方式，向AMF网元和/或OAM网元订阅并获取候选范围内的部分或全部UE的网络状态信息。示例性的，若AMF网元/OAM网元能够识别非连接态的UE，那么AMF网元/OAM网元可以提供候选范围内全部UE的网络状态信息，其中非连接态的UE的网络状态信息指示UE的状态为空闲态或去激活态。若AMF网元/OAM网元无法识别候选范围中非连接态的UE，那么AMF网元/OAM网元可以提供候选范围内部分UE的网络状态信息，或者理解为AMF网元/OAM网元可以提供候选范围内能够获取到网络状态信息的全部UE的网络状态信息，这里的能够获取到网络状态信息的全部UE为处于连接态。关于NWDAF可以采用订阅的方式获取UE的网络状态信息，可以参照S401～S402实施，本申请实施例对此不再进行赘述。
S503，NWDAF网元向AMF网元发送第一响应消息，该第一响应消息包括N1个UE的网络状态信息。
其中，N1个UE的定义可参照S402理解，本申请实施例对此不再进行赘述。
具体地，NWDAF网元可以对从AMF网元/OAM网元获取的N1个UE的网络状态信息进行分析，在第一响应消息中携带经由NWDAF网元分析过的N1个UE的网络状态信息。或者可以描述为，如果第一请求消息中携带Analytics ID，则第一响应消息中携带的N1个UE的网络状态信息可以是NWDAF网元分析得到的网络状态信息，例如UE的网络状态信息与设定阈值或其他UE的网络状态信息之间的对比分析结果，又如一定时间段内或者一定区域中UE的网络状态信息。
示例性的，NWDAF可以将历史从AMF/OAM获取的UE网络状态信息整理成历史数据集进行统计分析，得到该历史数据集的统计特征。NWDAF根据该历史数据集的统计特征，得到预测的分析结果，NWDAF可以将该预测的分析结果携带在第一响应消息发送给AF，辅助AF选择参与模型训练的UE；或者，NWDAF可以使用获取的历史数据集训练一个AI模型，利用该AI模型推理得到该预测的分析结果。
下面列举了一些由NWDAF分析得到的UE的网络状态信息，包括如下一个或多个分析结果。
UE的连接状态分析结果，NWDAF可以提供UE连接状态的统计信息，如在AF指定的范围(如区域/切片/DNN/App ID)或者指定的时间段内有哪些UE是处于连接态、哪些UE处于空闲态；NWDAF可以提供UE连接状态的预测信息，如哪些UE目前处于空闲态，但在AF指定的范围或者指定的时间段内将进入连接态。
UE的可达性分析结果，NWDAF可以提供UE可达性的统计信息，如在AF指定的范围(如区域/切片/DNN/App ID)或者指定的时间段内有哪些空闲态的UE是可达的、哪些是不可达的；NWDAF可以提供UE可达性的预测信息，如哪些UE目前是不可达的，但在AF指定的范围或者指定的时间段内将变为可达状态。
UE的移动性分析结果，NWDAF可以提供UE移动性的统计信息，如在AF指定的范围(如区域/切片/DNN/App ID)或者指定的时间段内UE的移动模式和移动趋势；NWDAF可以提供UE移动性的预测信息，如在AF指定的范围或者指定的时间段内有哪些UE即将移入服务区、哪些UE即将移出服务区、哪些UE将移动比较频繁、哪些UE将变得相对固定等。
UE的无线信道条件分析结果，NWDAF可以提供UE无线信道条件的统计信息，如在 AF指定的范围(如区域/切片/DNN/App ID)或者指定的时间段内有哪些UE接收信号质量较好、哪些UE接收信号质量一般；NWDAF可以提供UE无线信道条件的预测信息，如在AF指定的范围或者指定的时间段内有哪些UE的连接状态将会变得相对稳定，哪些UE的连接稳定性将变得较差等。
UE对应的无线资源利用率分析结果，NWDAF可以提供UE无线资源利用率的统计信息，如在AF指定的范围(如区域/切片/DNN/App ID)或者指定的时间段内有哪些UE有较多的可用无线时/频资源、哪些UE所在小区业务繁忙可用无线资源较少等；NWDAF可以提供UE无线资源利用率的预测信息，如在AF指定的范围或者指定的时间段内有哪些UE即将有较多的可用无线时/频资源、哪些UE所在小区即将变得业务繁忙(如晚上下班时间)。
具体地，第一响应消息可以为通知消息，例如NWDAF通过Nnwdaf_AnalyticsSubscription_Notify服务操作向AF发送通知消息，通知消息中携带：Analytics ID；N1个UE的网络状态信息，如当通知消息携带Analytics ID时，UE的网络状态信息还可以理解为UE的网络状态统计或预测的分析结果(UE network status related analytics)。可选的，通知消息中还可以携带Collected UE network status information，即表示NWDAF将从AMF/OAM中获取的原始网络状态信息转发给AF。
S504，AF网元根据所述N1个UE的网络状态信息，确定N2个UE。
其中，有关N2个UE的定义可参照S403描述的方式实施，本申请实施例对此不再进行赘述。
具体地，AF网元可以仅根据N1个UE的网络状态信息，通过数据分析或者特定的算法确定N2个UE。或者，AF网元也可以根据N1个UE的网络状态信息结合N1个UE的应用层信息，确定N2个UE，如利用UE的网络状态信息优化已有的应用层算法，或者结合UE的网络状态信息和应用层信息设计一个新的算法，并由新的算法确定N2个UE。
作为示例，假设N1个UE的网络状态信息为NWDAF分析得到统计或预测的分析结果，以AF网元根据N1个UE的网络状态信息结合N1个UE的应用层信息，确定N2个UE为例，本申请实施例如下提供确定N2个UE的具体实施方式：
假设N1为100，N2为80。即AF网元获取了100个UE的应用层信息和网络状态信息，并且准备从中选出80个UE参与联邦学习模型的训练。首先，AF网元可以根据UE的应用层信息和设定的应用层算法给每个UE打分，得到100个UE对应的应用层打分：UE1_app＝0.7,UE2_app＝0.5,UE3_app＝0.9,…UE100_app＝0.75。AF可以根据UE的预测连接状态给每个UE打分，如“0.9”代表连接态，“0.1”代表空闲态，得到100个UE对应的预测连接状态打分：UE1_CMstate＝0.9,UE2_CMstate＝0.1,UE3_CMstate＝0.1,…UE100_CMstate＝0.9。AF根据UE的预测可达性信息给每个UE打分，如“0.9”代表可达，“0.1”代表不可达，可以理解的是处于空闲态的UE可能是可达的，得到100个UE对应的预测可达性打分：如UE1_reachable＝0.9,UE2_reachable＝0.9,UE3_reachable＝0.1,…UE100_reachable＝0.9。AF根据UE的预测移动性给每个UE打分，如AF根据UE所处的位置和移动速度等计算该UE对应移动性得分，得到100个UE对应的预测移动性打分：UE1_mobility＝0.65,UE2_mobility＝0.7,UE3_mobility＝0.3,…UE100_mobility＝0.95。AF根据UE的预测无线信道条件给每个UE打分，得到100个UE对应的无线信道条件打分：UE1_radiochannel＝0.85,UE2_radiochannel＝0.5, UE3_radiochannel＝0.6,…UE100_radiochannel＝0.7。AF根据UE的预测无线资源利用率给每个UE打分，得到100个UE对应的无线资源利用率打分：UE1_radioresource＝0.8,UE2_radioresource＝0.4,UE3_radioresource＝0.7,…UE100_radioresource＝0.55。
其次，AF根据网络状态信息指示每个参数的重要程度给每个参数设定一个权重，比如，设定应用层信息的权重为UE_app_weight＝0.3，连接状态的权重为UE_CMstate_weight＝0.2,可达性的权重为UE_reachable_weight＝0.1，移动性的权重为UE_mobility_weight＝0.2，无线信道条件的权重为UE_radiochannel_weight＝0.1，无线资源利用率的权重为UE_radioresource_weight＝0.1。
最后，AF可以根据得分和权重计算每个UE的加权得分，如UE1_score＝0.7*0.3+0.9*0.2+0.9*0.1+0.65*0.2+0.85*0.1+0.8*0.1＝0.775，类似地，AF可以计算出其他UE的加权得分，然后将所有UE的加权得分排序，从中选出得分最高的80个UE，作为参与联邦学习模型训练的UE。
S505，AF网元与N2个UE进行联邦学习模型的训练。
具体地，可参照相关技术中描述的步骤1～步骤4实施，本申请实施例对此不再进行赘述。
进一步，有关联邦学习模型的更新训练可参照方案一中的描述理解，本申请实施例对此不再进行赘述。作为一种示例，图5中以虚线示意出如下S506～S509，体现联邦学习模型的更新训练过程。
S506，AF网元获取所述N2个UE中网络状态信息异常的UE的信息。
S507，AF网元从AMF网元和/或OAM网元中订阅并获取N3个UE的网络状态信息，所述候选范围包括所述N3个UE，所述N3个UE不包括所述N2个UE，N3为正整数。
S508，AF网元根据所述N2个UE中网络状态信息异常的UE的信息以及所述N3个UE的网络状态信息，确定N4个UE，该所述N4个UE用于参与联邦学习模型的更新训练，N4为正整数。
S509，AF网元与N4个UES608进行联邦学习模型的更新训练。
在本申请实施例提供的上述方案二中，AF向NWDAF请求并获取候选UE的网络状态信息相关的统计或预测的分析结果，应用于联邦学习环境，便于应用侧利用UE的网络状态信息选择参与联邦学习模型训练或更新的UE，实现对基于应用层信息选择参与方的算法的优化，从而提升联邦学习模型训练的效率。此外可以理解的是，在联邦学习环境中，AF也可以称为FL AF。
方案三
参见图6示意一种通信方法，该方法包括如下流程。
S601，AF网元向NWDAF网元发送第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备。
关于候选范围的定义可参照S401中的描述理解，本申请实施例对此不再进行赘述。
具体地，第二请求消息可以为订阅消息。例如AF网元通过Nnwdaf_AnalyticsSubscription_Subscribe服务操作向NWDAF发送订阅消息，订阅消息中携带的参数包括分析标识(Analytics ID)，Analytics ID＝Recommended UE Information，表示AF网元想要从NWDAF订阅并获取NWDAF推荐的参与联邦学习模型的训练的终端设备 标识。订阅消息中还包括用于指示候选范围的信息，可以为候选范围的标识，如以range of UEs表示指定的网络区域，以list of UE IDs表示候选终端设备列表。可选的，该订阅消息中还可以携带数量范围指示，如S401描述的第一指示信息或者UE数量范围(range of UE number)。可选的，该订阅消息中还可以携带如S401描述的数据有效性指示或者有效性联合指示。
此外，当AF是第三方应用功能时，出于安全性考虑，AF和NWDAF之间的交互需要经过NEF。NEF控制AF标识和允许获取的Analytics ID之间的映射关系，以及相关的入站限制(即限制AF可以请求的Analytics ID)和出站限制(即限制可以向AF通知的Analytics ID)。例如，当第三方AF需要从NWADF订阅网络状态信息时，第三方AF需要先将订阅消息发送给NEF，再由NEF向NWADF发出订阅消息，后续AMF先将相关UE的网络状态信息发送给NEF，再由NEF发送给第三方AF。
S602，NWDAF网元从AMF网元和/或OAM网元获取候选范围内的UE的网络状态信息。
具体地，可参照S502的实施方式执行，本申请实施例对此不再进行赘述。
S603，NWDAF网元根据获取候选范围内的UE的网络状态信息，确定推荐的N1个UE，所述候选范围内的UE包括所述N1个UE，N1为正整数。
具体地，NWDAF网元根据获取候选范围内的UE的原始网络状态信息，进行统计分析，得到统计或预测的UE网络状态分析结果。进而NWDAF网元可以参照S504中描述的打分策略，针对UE不同的网络状态设定不同的权重，然后根据加权和排序，将加权得分高的UE优先作为推荐的UE。
S604，NWDAF网元向AF网元发送第二响应消息，所述第二响应消息包括用于指示推荐的N1个终端设备的信息。
具体地，该第二响应消息为通知消息，例如NWDAF可以通过Nnwdaf_AnalyticsSubscription_Notify服务操作向FL AF发送通知消息，通知消息中可以携带如下参数：分析标识(Analytics ID)，本申请实施例中Analytics ID＝Recommended UE Information；以及推荐的UE列表(Recommended UE list)，用于指示S603中NWDAF推荐的N1个UE，可以由一组UE标识进行指示。
可选的，NWDAF网元也可以将S602中获取的N1个UE的原始网络状态信息发送给AF。例如，NWDAF网元可以在通知消息中包括Collected UE network status information，以表示UE的原始网络状态信息。
S605，AF网元根据所述第二响应消息，确定N2个UE，所述N2个UE用于参与联邦学习模型的训练，所述N1个UE包括所述N2个UE，N2为正整数。
一种可选的实施方式中，AF网元可以直接将NWDAF推荐的N1个UE，确定为参与联邦学习模型训练的UE，此情况下N1等于N2。
另一种可选的实施方式中，如果NWDAF在第二响应消息中包括N1个UE的原始网络状态信息，则AF网元可以根据N1个UE的原始网络状态信息，从N1个UE中选取部分或全部的UE参与联邦学习模型的训练。例如，AF网元可以选取N1个UE中原始网络状态信息较优的部分UE参与联邦学习模型的训练，或者AF网元可以选取N1个UE中原始网络状态信息大于或等于第一网络状态信息阈值的UE参与联邦学习模型的训练。
S606，AF网元与N2个UE进行联邦学习模型的训练。
具体地，可参照相关技术中描述的步骤1～步骤4实施，本申请实施例对此不再进行赘述。
进一步，有关联邦学习模型的更新训练可参照方案一中的描述理解，本申请实施例对此不再进行赘述。作为一种示例，图6中以虚线示意出如下S607～S612，体现联邦学习模型的更新训练过程。
S607，NWDAF网元获取所述N2个UE中网络状态信息异常的UE的信息。
具体地，图6中示意出NWDAF网元从UPF网元获取N2个UE中网络状态信息异常的UE的信息。
可选的，NWDAF可以向UPF网元发送第二指示信息，所述第二指示信息用于指示对所述N2个UE的网络状态信息进行监控。进而，UPF网元在确定N2个UE中存在网络状态信息异常的UE时，向NWDAF网元发送N2个UE中网络状态信息异常的UE的信息。
S608，NWDAF网元从AMF网元和/或OAM网元中订阅并获取N3个UE的网络状态信息，所述候选范围包括所述N3个UE，所述N3个UE不包括所述N2个UE，N3为正整数。
具体地，图6中示意出NWDAF网元从AMF网元和/或OAM网元获取N3个UE的网络状态信息。
S609，NWDAF网元根据所述N2个UE中网络状态信息异常的UE的信息以及所述N3个UE的网络状态信息，确定新的推荐UE列表，该新的推荐UE列表包括多个UE。
具体地，NWDAF网元可以参见S407中描述的方案，确定新的推荐UE列表。例如，新的推荐UE列表包括N2个UE中除网络状态信息异常的UE之外的其他UE，以及N3个UE中的部分UE。又如，新的推荐UE列表包括N2个UE中的部分UE以及N3个UE中的部分UE，新的推荐UE列表中可能包括网络状态信息异常的UE。
S610，NWDAF网元向AF网元发送用于指示新的推荐UE列表的信息。
可选的，NWDAF还可以向AF网元发送新的推荐UE列表中各UE的网络状态信息。
S611，AF网元根据新的推荐UE列表，确定N4个UE，N4个UE用于参与联邦学习模型的更新训练，N4为正整数，N4个UE包括推荐UE列表中的部分或者全部UE。
S612，AF网元与N4个UE进行联邦学习模型的更新训练。
在本申请实施例提供的上述方案二中，NWDAF可以利用UE的网络状态信息推导确定推荐参与联邦学习模型训练或更新的UE。应用于联邦学习环境，AF可以向NWDAF订阅推荐的UE，辅助应用侧选择参与联邦学习模型训练或更新的UE，实现对基于应用层信息选择参与方的算法的优化，从而提升联邦学习模型训练的效率。此外可以理解的是，在联邦学习环境中，AF也可以称为FL AF。
基于同一构思，参见图7，本申请实施例提供了一种通信装置700，该通信装置700包括处理模块701和通信模块702。该通信装置700可以是AF网元，也可以是应用于AF网元或者和AF网元匹配使用，能够实现AF网元侧执行的通信方法的通信装置；或者，该通信装置700可以是NWDAF网元，也可以是应用于NWDAF网元或者和NWDAF网元匹配使用，能够实现NWDAF网元侧执行的通信方法的通信装置；或者，该通信装置700可以是AMF网元(或者OAM网元、UPF网元等)，也可以是应用于AMF网元或者和AMF网元匹配使用，能够实现AMF网元侧执行的通信方法的通信装置。
其中，通信模块也可以称为收发模块、收发器、收发机、或收发装置等。处理模块也可以称为处理器，处理单板，处理单元、或处理装置等。可选的，通信模块用于执行上述方法中相关网元的发送操作和接收操作，可以将通信模块中用于实现接收功能的器件视为接收单元，将通信模块中用于实现发送功能的器件视为发送单元，即通信模块包括接收单元和发送单元。
示例性的，该通信装置700应用于AF网元时，处理模块701可用于实现图4、图5或图6所述示例中所述AF网元的处理功能，通信模块702可用于实现图4、图5或图6所述示例中所述AF网元的收发功能。或者也可以参照发明内容中第四方面中可能的设计理解该通信装置。
该通信装置700应用于AMF或OAM网元时，处理模块701可用于实现图4、图5或图6所述示例中AMF或OAM网元的处理功能，通信模块702可用于实现图4、图5或图6所述示例中AMF或OAM网元的收发功能。或者也可以参照发明内容中第五方面中可能的设计理解该通信装置。
该通信装置700应用于NWDAF网元时，处理模块701可用于实现图4、图5或图6所述示例中NWDAF网元的处理功能，通信模块702可用于实现图4、图5或图6所述示例中NWDAF网元的收发功能。或者也可以参照发明内容中第六方面中可能的设计理解该通信装置。
此外需要说明的是，前述通信模块和/或处理模块可通过虚拟模块实现，例如处理模块可通过软件功能单元或虚拟装置实现，通信模块可以通过软件功能或虚拟装置实现。或者，处理模块或通信模块也可以通过实体装置实现，例如若该通信装置采用芯片/芯片电路实现，所述通信模块可以是输入输出电路和/或通信接口，执行输入操作(对应前述接收操作)、输出操作(对应前述发送操作)；处理模块为集成的处理器或者微处理器或者集成电路。
本申请实施例中对模块的划分是示意性的，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，另外，在本申请实施例各个示例中的各功能模块可以集成在一个处理器中，也可以是单独物理存在，也可以两个或两个以上模块集成在一个模块中。上述集成的模块既可以采用硬件的形式实现，也可以采用软件功能模块的形式实现。
基于相同的技术构思，本申请实施例还提供了一种通信装置800。例如，该通信装置800可以是芯片或者芯片系统。可选的，在本申请实施例中芯片系统可以由芯片构成，也可以包含芯片和其他分立器件。
通信装置800可用于实现前述示例描述的通信系统中任一网元的功能。通信装置800可以包括至少一个处理器810，该处理器810与存储器耦合，可选的，存储器可以位于该通信装置之内，存储器可以和处理器集成在一起，存储器也可以位于该通信装置之外。例如，通信装置800还可以包括至少一个存储器820。存储器820保存实施上述任一示例中必要计算机程序、计算机程序或指令和/或数据；处理器810可能执行存储器820中存储的计算机程序，完成上述任一示例中的方法。
通信装置800中还可以包括通信接口830，通信装置800可以通过通信接口830和其它设备进行信息交互。示例性的，所述通信接口830可以是收发器、电路、总线、模块、管脚或其它类型的通信接口。当该通信装置800为芯片类的装置或者电路时，该通信装置800中的通信接口830也可以是输入输出电路，可以输入信息(或称，接收信息)和输出 信息(或称，发送信息)，处理器为集成的处理器或者微处理器或者集成电路或则逻辑电路，处理器可以根据输入信息确定输出信息。
本申请实施例中的耦合是装置、单元或模块之间的间接耦合或通信连接，可以是电性，机械或其它的形式，用于装置、单元或模块之间的信息交互。处理器810可能和存储器820、通信接口830协同操作。本申请实施例中不限定上述处理器810、存储器820以及通信接口830之间的具体连接介质。
可选的，参见图8，所述处理器810、所述存储器820以及所述通信接口830之间通过总线840相互连接。所述总线840可以是外设部件互连标准(peripheral component interconnect，PCI)总线或扩展工业标准结构(extended industry standard architecture，EISA)总线等。所述总线可以分为地址总线、数据总线、控制总线等。为便于表示，图8中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。
在本申请实施例中，处理器可以是通用处理器、数字信号处理器、专用集成电路、现场可编程门阵列或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件，可以实现或者执行本申请实施例中的公开的各方法、步骤及逻辑框图。通用处理器可以是微处理器或者任何常规的处理器等。结合本申请实施例所公开的方法的步骤可以直接体现为硬件处理器执行完成，或者用处理器中的硬件及软件模块组合执行完成。
在本申请实施例中，存储器可以是非易失性存储器，比如硬盘(hard disk drive，HDD)或固态硬盘(solid-state drive，SSD)等，还可以是易失性存储器(volatile memory)，例如随机存取存储器(random-access memory，RAM)。存储器是能够用于携带或存储具有指令或数据结构形式的期望的程序代码并能够由计算机存取的任何其他介质，但不限于此。本申请实施例中的存储器还可以是电路或者其它任意能够实现存储功能的装置，用于存储程序指令和/或数据。
在一种可能的实施方式中，该通信装置800可以应用于AF网元，具体通信装置800可以是AF网元，也可以是能够支持AF网元，实现上述涉及的任一示例中AF网元的功能的装置。存储器820保存实现上述任一示例中的AF网元的功能的计算机程序(或指令)和/或数据。处理器810可执行存储器820存储的计算机程序，完成上述任一示例中AF网元执行的方法。应用于AF网元，该通信装置800中的通信接口可用于与AMF、OAM或者NWDAF网元进行交互，向AMF、OAM或NWDAF网元发送信息，或者，接收来自AMF、OAM或NWDAF网元的信息。
在一种可能的实施方式中，该通信装置800可以应用于NWDAF网元，具体通信装置800可以是NWDAF网元，也可以是能够支持NWDAF网元，实现上述涉及的任一示例中NWDAF网元的功能的装置。存储器820保存实现上述任一示例中的NWDAF网元的功能的计算机程序(或指令)和/或数据。处理器810可执行存储器820存储的计算机程序，完成上述任一示例中NWDAF网元执行的方法。应用于NWDAF网元，该通信装置800中的通信接口可用于与AMF、OAM或者AF网元进行交互，向AMF、OAM或AF网元发送信息，或者，接收来自AMF、OAM或AF网元的信息。
在一种可能的实施方式中，该通信装置800可以应用于AMF/OAM网元，具体通信装置800可以是AMF/OAM网元，也可以是能够支持AMF/OAM网元，实现上述涉及的任一示例中AMF/OAM网元的功能的装置。存储器820保存实现上述任一示例中的AMF/OAM网元的功能的计算机程序(或指令)和/或数据。处理器810可执行存储器820 存储的计算机程序，完成上述任一示例中AMF/OAM网元执行的方法。应用于AMF/OAM网元，该通信装置800中的通信接口可用于与AF网元或NWDAF网元进行交互，向AF网元或NWDAF网元发送信息，或者，接收来自AF网元或NWDAF网元的信息。
基于以上示例，本申请实施例提供了一种通信系统，包括UE、AF网元、NWDAF网元、AMF网元、OAM网元。可选的，还包括UPF网元。其中，AF网元、NWDAF网元、AMF网元、OAM网元可以实现图4、图5、或者图6所示的示例中所提供的通信方法。
本申请实施例提供的技术方案可以全部或部分地通过软件、硬件、固件或者其任意组合来实现。当使用软件实现时，可以全部或部分地以计算机程序产品的形式实现。所述计算机程序产品包括一个或多个计算机指令。在计算机上加载和执行所述计算机程序指令时，全部或部分地产生按照本申请实施例所述的流程或功能。所述计算机可以是通用计算机、专用计算机、计算机网络、UE、AF网元、NWDAF网元、AMF网元、OAM网元或者其他可编程装置。所述计算机指令可以存储在计算机可读存储介质中，或者从一个计算机可读存储介质向另一个计算机可读存储介质传输，例如，所述计算机指令可以从一个网站站点、计算机、服务器或数据中心通过有线(例如同轴电缆、光纤、数字用户线(digital subscriber line，DSL))或无线(例如红外、无线、微波等)方式向另一个网站站点、计算机、服务器或数据中心进行传输。所述计算机可读存储介质可以是计算机可以存取的任何可用介质或者是包含一个或多个可用介质集成的服务器、数据中心等数据存储设备。所述可用介质可以是磁性介质(例如，软盘、硬盘、磁带)、光介质(例如，数字视频光盘(digital video disc，DVD))、或者半导体介质等。
在本申请实施例中，在无逻辑矛盾的前提下，各示例之间可以相互引用，例如方法示例之间的方法和/或术语可以相互引用，例如装置示例之间的功能和/或术语可以相互引用，例如装置示例和方法示例之间的功能和/或术语可以相互引用。
显然，本领域的技术人员可以对本申请实施例进行各种改动和变型而不脱离本申请实施例的范围。这样，倘若本申请实施例的这些修改和变型属于本申请实施例权利要求及其等同技术的范围之内，则本申请实施例也意图包含这些改动和变型在内。

Claims (57)

  1. 一种通信方法，其特征在于，包括：
    应用功能网元发送第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
    所述应用功能网元获取第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；
    所述应用功能网元根据所述N1个终端设备的网络状态信息，确定N2个终端设备，所述N2个终端设备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
  2. 如权利要求1所述的通信方法，其特征在于，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
  3. 如权利要求1或2所述的通信方法，其特征在于，所述第一请求消息包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
  4. 如权利要求1或2所述的通信方法，其特征在于，所述第一请求消息包括N1的取值范围。
  5. 如权利要求1或2所述的通信方法，其特征在于，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
  6. 如权利要求1-5任一项所述的通信方法，其特征在于，所述应用功能网元根据所述N1个终端设备的网络状态信息，确定N2个终端设备，包括：
    所述应用功能网元根据所述N1个终端设备的网络状态信息以及所述N1个终端设备的应用层信息，确定所述N2个终端设备。
  7. 如权利要求1-6任一项所述的通信方法，其特征在于，还包括：
    所述应用功能网元获取所述N2个终端设备中网络状态信息异常的终端设备的信息；
    所述应用功能网元获取N3个终端设备的网络状态信息，所述候选范围包括所述N3个终端设备，所述N3个终端设备不包括所述N2个终端设备，N3为正整数；
    所述应用功能网元根据所述N2个终端设备中网络状态信息异常的终端设备的信息以及所述N3个终端设备的网络状态信息，确定N4个终端设备，所述N4个终端设备用于参与联邦学习模型的更新训练，N4为正整数。
  8. 如权利要求7所述的通信方法，其特征在于，所述N4个终端设备包括所述N2个终端设备中除网络状态信息异常的终端设备之外的其他终端设备。
  9. 如权利要求7或8所述的通信方法，其特征在于，所述应用功能网元获取所述N2个终端设备中网络状态信息异常的终端设备的信息，包括：
    所述应用功能网元向网络数据分析功能网元发送第二指示信息，所述第二指示信息用于指示对所述N2个终端设备的网络状态信息进行监控，所述第二指示信息包括第二网络状态信息阈值，所述第二网络状态信息阈值用于确定所述终端设备的业务流量信息是否异常；
    所述应用功能网元从所述网络数据分析功能网元获取所述N2个终端设备中网络状态信息异常的终端设备的信息。
  10. 如权利要求1-9任一项所述的通信方法，其特征在于，还包括：
    所述应用功能网元与所述N2个终端设备进行所述联邦学习模型的训练。
  11. 一种通信方法，其特征在于，包括：
    接入与移动性管理功能网元获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
    所述接入与移动性管理功能网元发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
  12. 如权利要求11所述的通信方法，其特征在于，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
  13. 如权利要求11或12所述的通信方法，其特征在于，所述第一请求消息可以包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
  14. 如权利要求11或12所述的通信方法，其特征在于，所述第一请求消息包括N1的取值范围。
  15. 如权利要求11或12所述的通信方法，其特征在于，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
  16. 如权利要求11-15任一项所述的通信方法，其特征在于，还包括：
    所述接入与移动性管理功能网元发送N3个终端设备的网络状态信息，所述候选范围包括所述N3个终端设备，所述N3个终端设备不包括所述N2个终端设备，N3为正整数。
  17. 一种通信方法，其特征在于，包括：
    网络数据分析功能网元获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
    所述网络数据分析功能网元从接入与移动性管理功能网元和/或操作维护管理网元中，获取所述候选范围内的终端设备的网络状态信息；
    所述网络数据分析功能网元发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
  18. 如权利要求17所述的通信方法，其特征在于，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
  19. 如权利要求17或18所述的通信方法，其特征在于，所述第一请求消息可以包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
  20. 如权利要求17或18所述的通信方法，其特征在于，所述第一请求消息包括N1的取值范围。
  21. 如权利要求17或18所述的通信方法，其特征在于，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
  22. 如权利要求17-21任一项所述的通信方法，其特征在于，还包括：
    网络数据分析功能网元接收来自应用功能网元的第二指示信息，所述第二指示信息用于指示对所述N2个终端设备的网络状态信息进行监控，所述第二指示信息包括第二网络状态信息阈值，所述第二网络状态信息阈值用于确定所述终端设备的业务流量信息是否异常；
    所述网络数据分析功能网元向所述应用功能网元发送所述N2个终端设备中网络状态信息异常的终端设备的信息。
  23. 如权利要求1-22任一项所述的通信方法，其特征在于，所述第一请求消息包括用于指示所述网络状态信息的类型的事件标识，所述事件标识包括位置报告信息。
  24. 一种通信方法，其特征在于，包括：
    应用功能网元发送第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；
    所述应用功能网元获取第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；
    所述应用功能网元根据所述第二响应消息，确定N2个终端设备，所述N2个终端设备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
  25. 一种通信方法，其特征在于，包括：
    网络数据分析功能网元接收第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；
    所述网络数据分析功能网元确定所述候选范围内推荐的参与联邦学习模型的训练的终端设备；
    所述网络数据分析功能网元发送第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
  26. 如权利要求25所述的通信方法，其特征在于，所述网络数据分析功能网元确定候选范围内推荐的参与联邦学习模型的训练的终端设备，包括：
    所述网络数据分析功能网元从接入与移动性管理功能网元和/或操作维护管理网元中，获取所述候选范围内的终端设备的网络状态信息；
    所述网络数据分析功能网元根据所述候选范围内的终端设备的网络状态信息，确定所述推荐的N1个设备。
  27. 一种通信装置，其特征在于，包括：
    通信模块，用于发送第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
    所述通信模块，用于获取第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；
    处理模块，用于根据所述N1个终端设备的网络状态信息，确定N2个终端设备，所述N2个终端设备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
  28. 如权利要求27所述的通信装置，其特征在于，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
  29. 如权利要求27或28所述的通信装置，其特征在于，所述第一请求消息包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
  30. 如权利要求27或28所述的通信装置，其特征在于，所述第一请求消息包括N1的取值范围。
  31. 如权利要求27或28所述的通信装置，其特征在于，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
  32. 如权利要求27-31任一项所述的通信装置，其特征在于，所述处理模块，具体用于：
    根据所述N1个终端设备的网络状态信息以及所述N1个终端设备的应用层信息，确定所述N2个终端设备。
  33. 如权利要求27-32任一项所述的通信装置，其特征在于，
    所述通信模块，还用于获取所述N2个终端设备中网络状态信息异常的终端设备的信息；以及，获取N3个终端设备的网络状态信息，所述候选范围包括所述N3个终端设备，所述N3个终端设备不包括所述N2个终端设备，N3为正整数；
    所述处理模块，还用于根据所述N2个终端设备中网络状态信息异常的终端设备的信息以及所述N3个终端设备的网络状态信息，确定N4个终端设备，所述N4个终端设备用于参与联邦学习模型的更新训练，N4为正整数。
  34. 如权利要求33所述的通信装置，其特征在于，所述N4个终端设备包括所述N2个终端设备中除网络状态信息异常的终端设备之外的其他终端设备。
  35. 如权利要求33或34所述的通信装置，其特征在于，所述通信模块，还用于：
    向网络数据分析功能网元发送第二指示信息，所述第二指示信息用于指示对所述N2个终端设备的网络状态信息进行监控，所述第二指示信息包括第二网络状态信息阈值，所述第二网络状态信息阈值用于确定所述终端设备的业务流量信息是否异常；
    从所述网络数据分析功能网元获取所述N2个终端设备中网络状态信息异常的终端设备的信息。
  36. 如权利要求27-35任一项所述的通信装置，其特征在于，所述处理模块，还用于与所述N2个终端设备进行所述联邦学习模型的训练。
  37. 一种通信装置，其特征在于，包括：
    通信模块，用于获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
    处理模块，用于确定所述候选范围内的终端设备的网络状态信息；
    所述通信模块，还用于发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
  38. 如权利要求37所述的通信装置，其特征在于，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
  39. 如权利要求37或38所述的通信装置，其特征在于，所述第一请求消息可以包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
  40. 如权利要求37或38所述的通信装置，其特征在于，所述第一请求消息包括N1的取值范围。
  41. 如权利要求37或38所述的通信装置，其特征在于，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
  42. 如权利要求37-41任一项所述的通信装置，其特征在于，所述通信装模块，还用于发送N3个终端设备的网络状态信息，所述候选范围包括所述N3个终端设备，所述N3个终端设备不包括所述N2个终端设备，N3为正整数。
  43. 一种通信装置，其特征在于，包括：
    通信模块，用于获取第一请求消息，所述第一请求消息用于请求候选范围内的终端设备的网络状态信息；
    所述通信模块，还用于从接入与移动性管理功能网元和/或操作维护管理网元中，获取 所述候选范围内的终端设备的网络状态信息；
    处理模块，用于通过所述通信模块发送第一响应消息，所述第一响应消息包括N1个终端设备的网络状态信息，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
  44. 如权利要求43所述的通信装置，其特征在于，所述第一请求消息包括用于指示所述候选范围的信息，所述候选范围包括指定的网络区域，或者终端设备候选列表。
  45. 如权利要求43或44所述的通信装置，其特征在于，所述第一请求消息可以包括第一指示信息，所述第一指示信息用于指示N1的取值位于设定数量范围之内。
  46. 如权利要求43或44所述的通信装置，其特征在于，所述第一请求消息包括N1的取值范围。
  47. 如权利要求43或44所述的通信装置，其特征在于，所述N1个终端设备的网络状态信息大于或等于第一网络状态信息阈值。
  48. 如权利要求43-47任一项所述的通信装置，其特征在于，所述通信模块，还用于：
    接收来自应用功能网元的第二指示信息，所述第二指示信息用于指示对所述N2个终端设备的网络状态信息进行监控，所述第二指示信息包括第二网络状态信息阈值，所述第二网络状态信息阈值用于确定所述终端设备的业务流量信息是否异常；
    向所述应用功能网元发送所述N2个终端设备中网络状态信息异常的终端设备的信息。
  49. 如权利要求27-48任一项所述的通信装置，其特征在于，所述第一请求消息包括用于指示所述网络状态信息的类型的事件标识，所述事件标识包括位置报告信息。
  50. 一种通信装置，其特征在于，包括：
    通信模块，用于发送第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；
    所述通信模块，用于获取第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数；
    处理模块，用于根据所述第二响应消息，确定N2个终端设备，所述N2个终端设备用于参与联邦学习模型的训练，所述N1个终端设备包括所述N2个终端设备，N2为正整数。
  51. 一种通信装置，其特征在于，包括：
    通信模块，用于接收第二请求消息，所述第二请求消息用于请求候选范围内推荐的参与联邦学习模型的训练的终端设备；
    处理模块，用于确定所述候选范围内推荐的参与联邦学习模型的训练的终端设备；
    所述通信模块，用于发送第二响应消息，所述第二响应消息包括推荐的N1个终端设备，所述候选范围内的终端设备包括所述N1个终端设备，N1为正整数。
  52. 如权利要求51所述的通信装置，其特征在于，所述处理模块，具体用于：
    通过所述通信模块从接入与移动性管理功能网元和/或操作维护管理网元中，获取所述候选范围内的终端设备的网络状态信息；
    根据所述候选范围内的终端设备的网络状态信息，确定所述推荐的N1个设备。
  53. 一种通信装置，其特征在于，包括：处理器，所述处理器与存储器耦合，所述存储器用于存储程序或指令，所述处理器用于调用所述存储器存储的程序或指令，以执行如权利要求1-26任一项所述的方法。
  54. 一种芯片系统，其特征在于，包括：所述芯片系统包括至少一个处理器，和接口电 路，所述接口电路和所述至少一个处理器耦合，所述处理器通过运行指令，以执行权利要求1至26任一项所述的方法。
  55. 一种通信系统，其特征在于，包括：
    应用功能网元，所述应用功能网元用于执行如权利要求1-10、23和24中任一项所述的方法；以及
    用于与所述应用功能网元通信的网络数据分析功能网元和/或接入与移动性管理功能网元。
  56. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质上存储有指令，当所述指令在计算机上运行时，使得计算机执行如权利要求1-26任一项所述的方法。
  57. 一种计算机程序产品，其特征在于，包括指令，当所述指令在计算机上运行时，使得计算机执行如权利要求1-26任一项所述的方法。
PCT/CN2023/084300 2022-03-29 2023-03-28 一种通信方法及装置 Ceased WO2023185807A1 (zh)

Priority Applications (3)

Application Number Priority Date Filing Date Title
KR1020247032965A KR20240153394A (ko) 2022-03-29 2023-03-28 통신 방법 및 장치
EP23778172.9A EP4475507A4 (en) 2022-03-29 2023-03-28 Communication method and apparatus
US18/899,990 US20250023795A1 (en) 2022-03-29 2024-09-27 Communication method and apparatus

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210324047.8A CN116938747A (zh) 2022-03-29 2022-03-29 一种通信方法及装置
CN202210324047.8 2022-03-29

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US18/899,990 Continuation US20250023795A1 (en) 2022-03-29 2024-09-27 Communication method and apparatus

Publications (1)

Publication Number Publication Date
WO2023185807A1 true WO2023185807A1 (zh) 2023-10-05

Family

ID=88199128

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2023/084300 Ceased WO2023185807A1 (zh) 2022-03-29 2023-03-28 一种通信方法及装置

Country Status (5)

Country Link
US (1) US20250023795A1 (zh)
EP (1) EP4475507A4 (zh)
KR (1) KR20240153394A (zh)
CN (1) CN116938747A (zh)
WO (1) WO2023185807A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2025138173A1 (zh) * 2023-12-29 2025-07-03 北京小米移动软件有限公司 数据获取方法、装置及存储介质
WO2025179594A1 (en) * 2024-03-01 2025-09-04 Zte Corporation Systems and methods for federated learning
WO2025237135A1 (zh) * 2024-05-11 2025-11-20 华为技术有限公司 通信方法、系统及装置
WO2026011431A1 (zh) * 2024-07-12 2026-01-15 北京小米移动软件有限公司 信息处理方法、节点、通信设备、通信系统及存储介质

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN120238971A (zh) * 2023-12-28 2025-07-01 华为技术有限公司 通信方法和通信装置
CN120302295A (zh) * 2024-01-11 2025-07-11 华为技术有限公司 一种通信方法和通信装置

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2021032498A1 (en) * 2019-08-16 2021-02-25 Telefonaktiebolaget Lm Ericsson (Publ) Methods, apparatus and machine-readable media relating to machine-learning in a communication network
WO2021139892A1 (en) * 2020-01-10 2021-07-15 Telefonaktiebolaget Lm Ericsson (Publ) Distributed machine learning using network measurements
US20220038349A1 (en) * 2020-10-19 2022-02-03 Ziyi LI Federated learning across ue and ran
CN114079902A (zh) * 2020-08-13 2022-02-22 Oppo广东移动通信有限公司 联邦学习的方法和装置

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110309847B (zh) * 2019-04-26 2021-09-17 深圳前海微众银行股份有限公司 一种模型压缩方法及装置
CN110263921B (zh) * 2019-06-28 2021-06-04 深圳前海微众银行股份有限公司 一种联邦学习模型的训练方法及装置
EP4014436B1 (en) * 2019-08-16 2025-12-24 Telefonaktiebolaget LM Ericsson (publ) Methods, apparatus and machine-readable media relating to machine-learning in a communication network
EP3981218A1 (en) * 2019-12-23 2022-04-13 NEC Corporation Methods and devices of detection of misbehaving ues using data analysis
CN114981795A (zh) * 2020-01-14 2022-08-30 Oppo广东移动通信有限公司 资源调度方法、装置及可读存储介质
CN112512058B (zh) * 2020-05-24 2025-07-25 中兴通讯股份有限公司 网络优化方法、服务器、客户端设备、网络设备和介质
CN113283185B (zh) * 2021-07-23 2021-11-12 平安科技（深圳）有限公司 联邦模型训练、客户画像方法、装置、设备及介质
CN116866941A (zh) * 2022-03-28 2023-10-10 维沃移动通信有限公司 设备确定方法、装置及通信设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2021032498A1 (en) * 2019-08-16 2021-02-25 Telefonaktiebolaget Lm Ericsson (Publ) Methods, apparatus and machine-readable media relating to machine-learning in a communication network
WO2021139892A1 (en) * 2020-01-10 2021-07-15 Telefonaktiebolaget Lm Ericsson (Publ) Distributed machine learning using network measurements
CN114079902A (zh) * 2020-08-13 2022-02-22 Oppo广东移动通信有限公司 联邦学习的方法和装置
US20220038349A1 (en) * 2020-10-19 2022-02-03 Ziyi LI Federated learning across ue and ran

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4475507A4 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2025138173A1 (zh) * 2023-12-29 2025-07-03 北京小米移动软件有限公司 数据获取方法、装置及存储介质
WO2025179594A1 (en) * 2024-03-01 2025-09-04 Zte Corporation Systems and methods for federated learning
WO2025237135A1 (zh) * 2024-05-11 2025-11-20 华为技术有限公司 通信方法、系统及装置
WO2026011431A1 (zh) * 2024-07-12 2026-01-15 北京小米移动软件有限公司 信息处理方法、节点、通信设备、通信系统及存储介质

Also Published As

Publication number Publication date
EP4475507A1 (en) 2024-12-11
KR20240153394A (ko) 2024-10-22
EP4475507A4 (en) 2025-03-26
CN116938747A (zh) 2023-10-24
US20250023795A1 (en) 2025-01-16

Similar Documents

Publication Publication Date Title
WO2023185807A1 (zh) 一种通信方法及装置
EP4132066A1 (en) Communication method, apparatus and system
WO2020108003A1 (zh) 一种用户接入控制方法、信息发送方法及装置
KR20200039295A (ko) 5g 이동 통신 시스템에서 네트워크 분석 정보를 활용한 효율적 mico 모드 관리 방법
WO2021155579A1 (zh) 一种数据分析方法、装置及系统
WO2022141295A1 (zh) 一种通信方法和装置
WO2020108002A1 (zh) 一种传输策略确定方法、策略控制方法及装置
WO2022116061A1 (zh) 一种通信方法和装置
WO2021092937A1 (zh) 一种策略确定的方法、系统及装置
CN116208964B (zh) 一种业务传输及其控制方法、设备
US20230037402A1 (en) Communication method, apparatus, and system
WO2023078183A1 (zh) 一种数据收集方法及通信装置
WO2022116665A1 (zh) Tcp流的调整方法和系统
WO2026045458A1 (zh) 通信方法、装置和系统
CN113596909B (zh) 通信方法、装置及系统
EP4333538B1 (en) Communication method, system and apparatus
WO2024032031A1 (zh) 一种数据分析方法及装置
WO2022174780A1 (zh) DDoS攻击检测的方法和装置
WO2024032270A1 (zh) 一种策略确定方法、装置及存储介质
WO2024099016A1 (zh) 一种通信方法及装置
CN121357587A (zh) 模型更新方法、设备、装置及存储介质
WO2025000381A1 (zh) 资源管理方法、装置及系统
WO2024032197A1 (zh) 通信方法、装置、设备及存储介质
WO2024239785A1 (zh) 通信方法及装置
WO2025156910A1 (zh) 性能监控方法、装置、设备及存储介质

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23778172

Country of ref document: EP

Kind code of ref document: A1

DPE1 Request for preliminary examination filed after expiration of 19th month from priority date (pct application filed from 20040101)
WWE Wipo information: entry into national phase

Ref document number: 2023778172

Country of ref document: EP

ENP Entry into the national phase

Ref document number: 2023778172

Country of ref document: EP

Effective date: 20240904

ENP Entry into the national phase

Ref document number: 20247032965

Country of ref document: KR

Kind code of ref document: A

WWE Wipo information: entry into national phase

Ref document number: 1020247032965

Country of ref document: KR

NENP Non-entry into the national phase

Ref country code: DE