Method, device and equipment for determining candidate members

Abstract

The present application discloses a method, apparatus and device for determining a candidate member, which belongs to the field of communications technology. A method for determining a candidate member in an embodiment of the present application includes: a first network element receiving a request message sent by a second network element, the request message including screening information; the first network element determining one or more first devices as candidate members capable of participating in federated learning based on the screening information; and the first network element sending a response message to the second network element, the response message including an identifier of the candidate member, where the screening information includes at least one of a time period, an algorithm type, an accuracy threshold, a radio access method, a signal quality requirement, a traffic range, member type information, number information, area information and federated learning type information.

Description

[Technical field]

[0001] (CROSS REFERENCE TO RELATED APPLICATIONS) This application claims priority to a Chinese patent application proposed on March 28, 2022, with application number 202210314821.7 and titled "Method, Apparatus and Device for Determining Members to Participate in Federated Learning," and a Chinese patent application proposed on April 29, 2022, with application number 202210476433.9 and titled "Method, Apparatus and Device for Determining Candidate Members," the entire contents of which are incorporated by reference into this application.

[0002] The present application relates to the field of communications technology, and more particularly to a method, apparatus and device for determining candidate members. [Background technology]

[0003] Federated learning refers to a method of performing machine learning modeling by coordinating different participants (also called participants, or parties, data owners, or clients). In federated learning, participants do not need to disclose their own data to other participants and a coordinator (also called a coordinator, server, parameter server, or aggregation server). Therefore, federated learning is excellent in protecting user privacy and ensuring data security, and can solve the data island problem.

[0004] After applying federated learning to the field of communications, how to select a number of appropriate members to carry out federated learning and improve the efficiency of training is currently a problem that needs to be solved urgently. Summary of the Invention [Problem to be solved by the invention]

[0005] The embodiments of the present application provide a method, device and apparatus for determining candidate members, which can solve the problem that the participating members of federated learning do not meet the requirements, causing the training efficiency of federated learning to be relatively low. [Means for solving the problem]

[0006] A first aspect provides a method for determining candidate members, the method comprising: receiving, by a first network element, a request message sent by a second network element, the request message including screening information; The first network element determines one or more first devices as candidate members capable of participating in federated learning based on the screening information; the first network element sending a response message to the second network element, the response message including an identifier of the candidate member; wherein the screening information is A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0007] A second aspect provides a method for determining candidate members, the method comprising: A second network element sends a request message to a first network element, the request message including screening information, and the request message is used to instruct the first network element to determine one or more first devices as candidate members that can participate in federated learning based on the screening information; receiving, by the second network element, a response message sent by the first network element, the response message including an identifier of the candidate member; wherein the screening information is A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0008] A third aspect provides an apparatus for determining a candidate member, the apparatus comprising: a receiving module for receiving a request message sent by a second network element, the request message including screening information; a processing module for determining one or more first devices as candidate members capable of participating in federated learning based on the screening information; a sending module for sending a response message to the second network element, the response message including an identifier of the candidate member; wherein the screening information is A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0009] A fourth aspect provides an apparatus for determining a candidate member, the apparatus comprising: A sending module for sending a request message to a first network element, the request message including screening information, the request message being used to instruct the first network element to determine one or more first devices as candidate members capable of participating in federated learning based on the screening information; a receiving module for receiving a response message transmitted by the first network element, the response message including an identifier of the candidate member; wherein the screening information is A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0010] A fifth aspect provides a first network element, the first network element including a processor and a memory, the memory storing a program or instructions operable to run on the processor, the program or instructions, when executed by the processor, implementing steps of the method of the first aspect.

[0011] A sixth aspect provides a first network element, the first network element including a processor and a communication interface, where the processor is adapted to determine one or more first devices as candidate members capable of participating in federated learning based on screening information, and the communication interface is adapted to receive a request message sent by a second network element and to send a response message to the second network element, the response message including an identifier of the candidate member, where the request message includes screening information, and the screening information includes: A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0012] A seventh aspect provides a second network element, the network element including a processor and a memory, the memory storing a program or instructions operable to run on the processor, the program or instructions, when executed by the processor, implementing the steps of the method of the second aspect.

[0013] An eighth aspect provides a second network element, the second network element including a processor and a communication interface, wherein the communication interface is adapted to send a request message to a first network element, the request message including screening information, the request message instructing the first network element to determine one or more first devices as candidate members capable of participating in federated learning based on the screening information, and to receive a response message sent by the first network element, the response message including identifiers of the candidate members; wherein the screening information is A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0014] A ninth aspect provides a system for determining a candidate member, the system including a first network element and a second network element, the first network element being usable for performing steps of the method for determining a candidate member described in the first aspect, and the second network element being usable for performing steps of the method for determining a candidate member described in the second aspect.

[0015] A tenth aspect provides a readable storage medium having a program or instructions stored on the readable storage medium, the program or instructions performing steps of the method of the first aspect or performing steps of the method of the second aspect when executed by a processor.

[0016] An eleventh aspect provides a chip, the chip including a processor and a communication interface, the communication interface coupled to the processor, the processor running a program or instructions to implement the method of the first aspect or used to implement the method of the second aspect.

[0017] A twelfth aspect provides a computer program / program product, the computer program / program product being stored in a storage medium and executed by at least one processor to implement the steps of the candidate member determination method described in the first and second aspects. Effect of the Invention

[0018] In an embodiment of the present application, the first network element receives a request message sent by the second network element, and determines one or more first devices as candidate members that can participate in federated learning based on screening information included in the request message, and sends identifier information of the candidate members to the second network element, where the screening information includes at least one of a time period, an algorithm type, a precision threshold, a wireless access method, a signal quality requirement, a traffic range, member type information, number information, an area of ​​interest AOI, and a type of federated learning. The first network element can improve the training efficiency of federated learning by screening participating members that are suitable for performing federated learning through the screening information. [Brief description of the drawings]

[0019] [Figure 1] 1 is a block diagram of a wireless communication system to which an embodiment of the present application can be applied. [Diagram 2] 1 is a flowchart of a method for determining candidate members according to the present application. [Diagram 3] 2 is a flow chart of a method for determining another candidate member according to the present application. [Figure 4] FIG. 2 is a signaling diagram of a method for determining candidate members according to the present application. [Diagram 5] 1 is a structural schematic diagram of a candidate member determination device according to the present application; [Figure 6] 2 is a second structural schematic diagram of the candidate member determination device according to the present application. [Figure 7] 1 is a structural schematic diagram of a communication device according to an embodiment of the present application; [Figure 8] FIG. 2 is a hardware structure schematic diagram of a first network element according to an embodiment of the present application; [Figure 9] FIG. 2 is a hardware structural schematic diagram of a second network element according to an embodiment of the present application; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0020] The following clearly describes the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only some of the embodiments of the present application, and not all of the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present application are all within the scope of protection of the present application.

[0021] The terms "first," "second," etc. in the specification and claims of the present application are intended to distinguish between similar objects and are not intended to describe a particular order or sequence. It is to be understood that the terms so used are interchangeable where appropriate, such that the embodiments of the present application may be performed in other orders than those shown or described herein, and that the objects distinguished by "first" and "second" are generally of the same type and do not limit the number of objects. For example, the first object may be one or more. It is to be noted that "and / or" in the specification and claims represents at least one of the objects connected, and the character " / " generally represents an "or" relationship between the related objects.

[0022] It should be noted that the techniques described in the embodiments of the present application are not limited to Long Term Evolution (LTE) / LTE-Advanced (LTE-A) systems, but can also be applied to other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency Division Multiple Access (SC-FDMA), and other systems. The terms "system" and "network" in the embodiments of the present application are always used interchangeably, and the techniques described may be used in the above-mentioned systems and radio technologies, or in other systems and radio technologies. Although the following description describes a New Radio (NR) system for illustrative purposes and uses NR terminology in most of the description below, these techniques may be applied to applications other than NR system applications, such as 6th Generation (6G) communication systems.

[0023] 1 shows a block diagram of a wireless communication system to which an embodiment of the present application can be applied. The wireless communication system includes a terminal 11 and a network side device 12. In the present application, the terminal 11 may be referred to as a user equipment (UE). Here, the terminal 11 may be a terminal device such as a mobile phone, a tablet personal computer, a laptop computer (also called a notebook computer), a personal digital assistant (PDA), a palmtop computer, a netbook, an ultra-mobile personal computer (UMPC), a mobile Internet device (MID), an augmented reality (AR) / virtual reality (VR) device, a robot, a wearable device, a vehicle-mounted equipment (VUE), a pedestrian terminal (PUE), a smart home (home equipment having a wireless communication function, such as a refrigerator, a television, a washing machine, or furniture), a game console, a personal computer (PC), a deposit machine, or a self-service machine, and the wearable device includes a smart watch, a smart wristband, a smart earphone, a smart glasses, a smart accessory (a smart bracelet, a smart hand chain, a smart ring, a smart necklace, a smart ankle bracelet, a smart anklet, etc.), a smart band, a smart clothing, etc. It should be noted that the terminal 11 in the embodiment of the present application is not limited to a specific type. The network side equipment 12 may include an access network equipment or a core network equipment. Here, the access network equipment 12 may be called a radio access network equipment, a radio access network (Radio Access Network, RAN), a radio access network function or a radio access network unit.The access network equipment 12 may include a base station, a WLAN access point, or a WiFi node, and the base station may be called a Node B, an evolved Node B (eNB), an access point, a base transceiver station (BTS), a radio base station, a radio transceiver, a basic service set (BSS), an extended service set (ESS), a home B node, a home evolved B node, a transmitting receiving point (TRP), or any other suitable term in the art. As long as the same technical effect is achieved, the base station is not limited to a specific technical term, and it should be explained that in the embodiments of the present application, only base stations in an NR system are introduced as examples, and the specific type of the base station is not limited.The core network devices include a core network node, a core network function, a mobility management entity (MME), an access and mobility management function (AMF), a session management function (SMF), a user plane function (UPF), a policy control function (PCF), a policy and charging rules function (PCRF), an edge application server discovery function (EASDF), a unified data management (UDM), a unified data repository (UDR), a home subscriber server (HSS), a centralized network configuration (CNC), a network repository function (NRF), a network exposure function (NEF), a local NEF (Local NEF or L-NEF), a binding support function (BSF), an application function (Application Repository Function (ARF)), a network scrubbing function (SSR) and a network scrubbing function (SSR). It should be noted that the embodiments of the present application only take the core network device in the NR system as an example, and do not limit the specific type of the core network device.

[0024] The following describes in detail the method for determining candidate members according to the embodiments of the present application through several embodiments and application scenarios thereof in conjunction with the drawings.

[0025] Federated learning includes horizontal federated learning and vertical federated learning. Here, the essence of horizontal federated learning is the federation of samples, which is the same business type between participants but different reach clients. That is, it can be applied to scenarios when there is a lot of overlap in features and little overlap in samples, such as the same service (e.g., mobility management (MM), session management (SM) service, certain service) serving different users (e.g., each terminal, i.e., samples are different) in the core network (CN) domain and the radio access network (RAN) domain in the communication network. With the same data features of different samples of federated participants, horizontal federation obtains a better model by increasing the number of training samples.

[0026] The essence of vertical federation learning is feature federation, which can be applied to scenarios where samples have a lot of overlap and features have little overlap, such as different services (e.g., MM and SM services, i.e., features are different) serving the same user (e.g., terminal, i.e., samples are the same) in the CN domain and the RAN domain in a communication network. Due to the different data features of the joint samples of federation participants, vertical federation increases the feature dimension of the training sample and obtains a better model.

[0027] The federated learning system includes an administrator (server) and multiple participants. Here, the administrator is used to send a model to each participant, update the model based on the feedback results of each participant, and send the updated model to each participant again for the next model training. Here, each participant has its own data, and each participant uses the model sent by the administrator to train locally and then returns it to the administrator for model updating, so as not to send local data to others.

[0028] Illustratively, the process of model training includes: In step 1, each participant downloads the latest model from the administrator (server); Step 2: each participant uses local data to train the model and uploads the encrypted gradients to the administrator (server), which then aggregates the gradients of each participant to update the model parameters; Step 3: The administrator (server) returns the updated model to each participant; and step 4, in which each participant updates their respective model.

[0029] The above steps 2 to 4 involve multiple iterations / model updates, and model training is completed under certain conditions, for example, when a certain number of iterations are completed or the calculated value of the model's loss function falls below a preset value.

[0030] 5GS will introduce the network data analytic function (NWDAF), which will collect data from each network element of the core network, network management system, etc., perform big data statistics, analysis, or smart data analysis, and obtain analysis or prediction data for the network side, thereby assisting each network element to perform more effective control over terminal access based on the data analysis results.

[0031] In the method for determining candidate members according to the present application, the NWDAF can collect data of other network elements and analyze the data, thereby improving the efficiency of federated learning by screening participating members suitable for conducting federated learning based on the screening information and the data analysis results.

[0032] It should be explained that "determination of candidate members" in this application may be understood as "determination of candidate members who can participate in federated learning."

[0033] 2 is a flow chart of a method for determining candidate members according to the present application, and the method for determining candidate members according to an embodiment of the present application will be described below in conjunction with FIG 2. As shown in FIG 2, the method includes:

[0034] Step 201: A first network element receives a request message sent by a second network element.

[0035] Here, the execution body of the candidate member determination method according to the embodiment of the present application is a first network element, and the first network element may be implemented in various forms. For example, the first network element described in the embodiment of the present application may include an NWDAF, and of course may be a network element that can collect data from each network element of the core network, a network management system, etc., and perform big data statistics, analysis, or smart data analysis. The second network element may include a task consumer network element, and may be a network element or device such as an application function network element (AF), a base station, a terminal, etc., and it may even be a third-party server.

[0036] Here, the request message includes screening information, which includes at least one of a time period, an algorithm type, a precision threshold, a wireless access method, a signal quality requirement, a traffic range, member type information, number information, area information, and federated learning type information. By screening more suitable participating members based on different dimensions of screening information, the efficiency of federated learning can be improved.

[0037] Here, the time period is used to indicate a selective time period for performing the federated learning, and the time period may be a time period in the past or a time period in the future. As can be understood, when it is a future time period, the first network element predicts the network connection state of each member in the future time period based on the acquired historical data. The network connection state includes a state of already being connected to the network and a state of not being connected to the network. It should be understood that the time period is generally flexibly set according to the time period for actually performing the federated learning.

[0038] The algorithm type is used to indicate the model training algorithm type that needs to be supported to perform the federated learning, and includes algorithm types related to artificial intelligence (AI) data analysis tasks related to machine learning and the like supported by each member, such as "deep learning algorithm", "linear regression algorithm", etc. It should be understood that, generally, based on the function of the model, a member with an algorithm type matching this function is selected and supported as a participating member of the federated learning.

[0039] The accuracy threshold is used to indicate the accuracy requirements of model training that must be met to perform the federated learning, and may be understood as the accuracy value that the generated model can achieve after training each member, such as the accuracy rate of the model, which may be understood as the percentage of the model that correctly predicts or judges. It should be understood that in order to ensure the accuracy of the federated learning, some members with relatively high accuracy are generally selected as participating members in the federated learning.

[0040] The wireless access method is used to indicate the wireless access method that needs to be selected to perform federated learning, and includes a method connected to or accessed by each member's communication network, such as a WLAN (similar to connected to WiFi) connected to a non-3GPP, a 5g or 4g network connected to a 3GPP, etc. It should be understood that, in order to ensure the stability of federated learning and reduce traffic consumption, members connected to some WLAN or WiFi are generally selected as participating members in federated learning.

[0041] The signal quality requirement is used to indicate the wireless signal quality requirement when performing the federated learning. The signal quality requirement may include a network signal strength threshold and / or a stability threshold when connected to the communication network of each member. The signal quality requirement may include a WLAN network signal quality requirement, may include a 5G NR network signal quality requirement, and may further include a 4G Long Term Evolution (LTE) network signal quality requirement. The signal quality requirement may be understood as a minimum requirement for the time ratio that the signal strength is maintained above a certain value, for example, the signal strength in at least 90% of the time can achieve the threshold requirement. It should be understood that, in order to ensure the stability of the federated learning, some members with relatively strong signal quality (e.g., relatively strong network signal strength and relatively good stability) are generally selected as members participating in the federated learning.

[0042] Here, when connected to a WLAN, the signal quality can be expressed using the average value, variance, etc. of a received signal strength indication (RSSI) or the round-trip time (RTT) of the signal.

[0043] The traffic range is used to indicate the traffic usage range requirement of the candidate member, and is the traffic value requirement consumed by each member in a preset time period. For example, it is information such as the number of upload and download traffic, and may be, for example, the accumulated usage of the UE. It should be understood that, in order to reduce the pressure of each participating member when performing federated learning, generally, some members with stable network conditions and relatively low traffic consumption are selected as participants in federated learning.

[0044] The member type information is used to indicate the type request of a candidate member to participate in the federated learning, and the member type information may be understood as the type of the candidate member to participate in the current federated learning, such as a terminal, a core network element (e.g., NWDAF), or a base station.

[0045] An area of ​​interest (AOI) is used to indicate the area in which the candidate member is located, and the AOI may be the area in which the candidate member participating in the federated learning is located, or an area range of attention or interest, which may be longitude and latitude, or one or more cells / tracking areas (TAs), etc.

[0046] The number information is used to indicate the number requirement of the candidate members, which may be understood as the number of members participating in federated learning, i.e., the number of candidate members that the first network element needs to determine. The first network element can determine this number of candidate network elements. The number information may be the minimum number of required candidate members, i.e., the number of determined candidate members must not be smaller than the minimum number of required candidate members. The number information may also be the maximum number of required candidate members, i.e., the number of determined candidate members must not be larger than the maximum number of required candidate members.

[0047] The type information of federated learning is used to indicate whether the federated learning belongs to vertical federation or horizontal federation. Optionally, the request message may further include instruction information. This instruction information is used to instruct to perform a task of determining candidate members to participate in federated learning, i.e., the task is used for federated learning or member selection of federated learning. After receiving the request message, the first network element can know based on the request message or based on the instruction information in the request message, i.e., that it needs to perform a task of determining candidate members to participate in federated learning.

[0048] Step 202: The first network element determines one or more first devices as candidate members capable of participating in federated learning based on the screening information.

[0049] In this step, after receiving a request message, the first network element determines one or more first devices that satisfy the screening information by screening from multiple second devices based on the screening information in the request message, and determines these first devices as candidate members that can participate in federated learning.

[0050] Optionally, the first device includes at least one of a first device in a network connection state during the time period, a first device supporting federated learning during the time period, a first device supporting an algorithm type, a first device supporting federated learning using the algorithm type, a first device whose accuracy information of model training is greater than an accuracy threshold, a first device supporting a wireless access method, a first device in the wireless access method, a first device supporting federated learning in the wireless access method, a first device whose signal quality is greater than a signal quality threshold, a first device whose signal quality is greater than a signal quality threshold requirement, a first device whose consumed traffic is in a traffic range, a first device that meets the type requirement of the candidate member, a first device located within the AOI, and a first device whose type of federated learning is the same as the type of federated learning included in the screening information.

[0051] Here, when the screening information includes a time period, the first device is a first device that is in a network connection state during the time period, or the first device is a first device that supports federated learning during the time period; when the screening information includes an algorithm type, the first device is a first device that supports the algorithm type, or the first device is a first device that supports federated learning using the algorithm type; when the screening information includes an accuracy threshold, the first device is a first device whose accuracy information of model training is greater than the accuracy threshold; and when the screening information includes a wireless access method, the first device is a first device that supports the wireless access method. The first device is a first device in the wireless access system, or the first device is a first device supporting federated learning in the wireless access system, if the screening information includes a signal quality threshold, the first device is a first device whose signal quality is greater than the signal quality threshold, if the screening information includes a signal quality requirement, the first device is a first device whose signal quality is greater than the signal quality requirement, if the screening information includes a traffic range, the first device is a first device whose consumed traffic is in the traffic range, if the screening information includes member type information, the first device is a first device that satisfies the type requirement of the candidate member. As can be understood, the type of the first device is the same as the indicated member type information in the screening information, if the screening information includes an AOI, the first device is a first device located within the AOI, if the screening information includes federated learning type information, the type of federated learning of the first device is the same as the federated learning type information included in the screening information.

[0052] Also, if the screening information includes a device type, the first device is the first device of the device type, and if the screening information includes region information, the first device is the first device in the region information.

[0053] It should be understood that when the screening information includes at least two, the first device is a device that satisfies the at least two corresponding screening information. For example, when the screening information includes a time period and a supported algorithm type, the first device is a first device that is in a network-connected state during the time period and supports the algorithm type. When the screening information includes at least two other pieces of information, the information is similar to the information including a time period and a supported algorithm type, and is not further described here.

[0054] In the above embodiment, the screening information can be used to screen first devices that satisfy the screening information from at least one second device as candidate network elements, thereby enabling the selection of appropriate members that can participate in federated learning, thereby contributing to improving the efficiency of federated learning training.

[0055] Optionally, the first network element, when determining candidate members, can determine a data type corresponding to the screening information, obtain attribute information corresponding to each of the at least one device from the at least one third network element based on the data type, and determine candidate members based on the attribute information and the screening information.

[0056] Optionally, the data type includes at least one of network connection information and corresponding time period, algorithm type, wireless access method, signal quality, and traffic.

[0057] Specifically, when the screening information includes a time zone in which the screening information is located, the data type corresponding to the screening information is network connection information and a time zone corresponding to being in a network connection state; when the screening information includes a supported algorithm type, the data type corresponding to the screening information is an algorithm type; when the screening information includes an accuracy threshold, the data type corresponding to the screening information is accuracy information; when the screening information includes a wireless access method, the data type corresponding to the screening information is a wireless access method; when the screening information includes a signal quality threshold, the data type corresponding to the screening information is a signal quality; when the screening information includes a traffic range, the data type corresponding to the screening information is a traffic; when the screening information includes a member type, the data type corresponding to the screening information is a device type; when the screening information includes area information, the data type corresponding to the screening information is location information.

[0058] As can be understood, when the screening information includes a member type, the member type is different and the screening information included in the request message is different. That is, the screening information is related to the member type. For example, when the member type includes a terminal, the screening information may include at least one of a time period, a supported algorithm type, an accuracy threshold, a radio access scheme, a signal quality threshold, a traffic range, and area information. When the member type includes a network side device, such as a base station or a core network device, the screening information may include at least one of a supported algorithm type and an accuracy threshold.

[0059] After determining the data type, the first network element obtains attribute information corresponding to each of the at least one second device from the at least one third network element based on the data type. Here, the third network element includes a plurality of different network elements, and it should be understood that the attribute information that needs to be obtained is different. The first network element obtains the attribute information from the different third network elements. For example, the first network element obtains information such as the time when the terminal is connected to the WLAN and whether it is connected to the WLAN from a session management function (SMF) of the third network element. The first network element obtains information such as the signal quality of the terminal connected to the WLAN from the third network element. Here, the third network element may be a network management device, such as an operation administration and maintenance (OAM), and obtains the algorithms and accuracy supported by each candidate network element from a unified data management (UDM) or a data collection - application function (DC-AF) of the third network element, and obtains terminal traffic information, etc. from the user plane function (UPF) of the third network element, or the first network element obtains UE accumulated usage information from an SMF or a Charging Function (CHF).

[0060] As can be understood, when the data type includes network connection information and corresponding time period, the attribute information is the current network connection state corresponding to each of the at least one device and the time in the network connection state, such as whether it is connected to the network and in which time period. When the data type includes algorithm type, the attribute information is the currently supportable algorithm corresponding to each of the at least one device, such as whether it supports a deep learning algorithm, whether it supports a linear regression algorithm, etc. When the data type includes accuracy information, the attribute information is the currently supportable accuracy corresponding to each of the at least one device, such as the achievable accuracy of the model after each device trains on the model. When the data type includes wireless access method, the attribute information is the currently accessed network method type of the at least one device, such as whether the currently accessed is a WLAN or a 5G network. When the data type includes signal quality, the attribute information is the signal quality of the currently accessed network of the at least one second device, or the signal quality of the accessed network in a preset time period. When the data type includes traffic information, the attribute information is the traffic consumed in a preset time period corresponding to each of the at least one second device. When the data type includes a device type, the attribute information is a type corresponding to each of the at least one second device, such as a terminal, a base station, or a core network device. When the data type includes location information, the attribute information is a location where each of the at least one second device is located, such as a current location geographic location, or a cell or TA where the device is located.

[0061] After obtaining this attribute information, the attribute information corresponding to each of the at least one second device is matched with the screening information, and candidate members are determined based on the screening information, for example, a first device that satisfies the screening information can be determined as a candidate member.

[0062] For example, assume that the device type included in the screening information is a terminal, the area is cell A, time period A, and the radio access method is WiFi connected to non-3GPP. The first network element obtains corresponding attribute information according to these screening information, and selects a WiFi terminal that is in cell A, has a network connection state in time period A, and is connected to non-3GPP as a candidate member according to the obtained attribute information.

[0063] In this embodiment, the first network element obtains a data type corresponding to the screening information, and obtains attribute information of each second device from at least one third network element, and determines candidate members based on the attribute information and the screening information, and can determine candidate members suitable for participating in federated learning from at least one second device based on the preset screening information, thereby improving the efficiency of federated learning.

[0064] In one possible implementation manner, the first network element can directly determine devices that meet the screening information as candidate members, and by default, any candidate network elements that meet this screening information can participate in federated learning, and the first network element can send identifiers of any candidate members that meet this screening information to the second network element.

[0065] In another possible implementation manner, in order to screen more suitable participants, when the first network element determines candidate members based on the attribute information and the screening information, it can further screen based on the willingness information of each second device. Illustratively, the first network element can obtain first instruction information from the third network element, and determine one or more first devices as candidate members from the second devices based on the screening information and the first instruction information. Here, the first instruction information is used to represent the willingness information of each device to participate in federated learning, and the willingness information represents whether each device wants to participate in federated learning.

[0066] Here, the first device is a device that meets the screening information and wants to participate in federated learning.

[0067] For example, each device may correspond to its own first instruction information, and for a certain device, when the corresponding first instruction information is 0, it indicates that this device wishes to participate in federated learning, and when the first instruction information is 1, it indicates that this device does not wish to participate in federated learning, or when the first instruction information is 1, it indicates that this device wishes to participate in federated learning, and when the first instruction information is 0, it indicates that this device does not wish to participate in federated learning. Of course, the first instruction information may adopt other numerical values ​​to indicate the willingness information of each device to participate in federated learning.

[0068] After obtaining the first instruction information, the first network element matches the attribute information of each device with the screening information, and determines the devices that meet the screening information and are willing to participate in the federated learning instructed by the first instruction information as candidate members.

[0069] In this embodiment, the first network element meets the screening information based on the screening information and the first instruction information, and screens devices that wish to participate in federated learning as candidate members, thereby contributing to improving the efficiency of subsequent federated learning.

[0070] Optionally, when the first network element determines one or more first devices as candidate members from the second devices based on the screening information and the first instruction information, the first network element can further obtain capability information of each second device to participate in federated learning from a third network element, thereby determining candidate members based on the screening information, the first instruction information and the capability information.

[0071] Here, the capability information includes at least one of the radio access method participating in federated learning, the area participating in federated learning, the time for participating in federated learning, algorithm information that can support participation in federated learning, accuracy information that can be achieved by participation in federated learning, and the type of participation in federated learning.

[0072] Here, the type of participation in federated learning may include a desire to participate in horizontal federated learning or a desire to participate in vertical federated learning.

[0073] In one embodiment, the first network element determining one or more first devices as candidate members capable of participating in federated learning based on the screening information includes: The first network element acquires, from a third network element, information indicating that the first device intends to participate in the federated learning; The first network element determines, based on the intention information and the screening information, a first device that wishes to participate in the federated learning and matches the screening information as a candidate member that can participate in the federated learning.

[0074] In one embodiment, the first network element determining one or more first devices as candidate members capable of participating in federated learning based on the screening information includes: The first network element acquires capability information of federated learning of the first device, the capability information including at least one of a radio access method participating in federated learning, an area participating in federated learning, a time participating in federated learning, algorithm information that can support participation in federated learning, accuracy information that can be achieved by participation in federated learning, and a type participating in federated learning; The first network element determines candidate members that can participate in federated learning from the first device based on the capability information and the screening information, and the capability information of the candidate members matches the screening information.

[0075] Illustratively, a first network element may obtain federated learning capability information of the first device from a fourth network element, where the fourth network element may be at least one of an NRF, a UDM, a Data Collection Coordination Function (DCCF), and an AMF.

[0076] The ability information of the candidate member matches the screening information, The wireless access method in which the candidate member participates in federated learning is the same as the wireless access method included in the screening information; The area in which the candidate member participates in the federated learning is located within the AOI included in the screening information; The time when the candidate member participates in the federated learning is within the time period included in the screening information; The screening information includes an algorithm type that can support the candidate member's participation in federated learning; and The accuracy information achievable for the candidate member to participate in the federated learning is higher than the accuracy threshold included in the screening information; and the type of federated learning in which the candidate member participates is the same as the type of federated learning included in the screening information.

[0077] Optionally, the first network element can further obtain capability information of each device to participate in federated learning from the third network element, thereby matching the attribute information of each device with the screening information, and determining as candidate members devices that wish to participate in federated learning as instructed by the first instruction information and whose capability information to participate in federated learning satisfies the screening information, thereby improving the efficiency of subsequent federated learning.

[0078] Here, the ability information to participate in the associative learning satisfies the screening information. The wireless access system of the candidate member participating in the federated learning is the same as the wireless access system included in the screening information; The time when the candidate member participates in the federated learning is within the time period included in the screening information; The screening information includes an algorithm type that can support the candidate member's participation in federated learning; and The accuracy information achievable for the candidate member to participate in the federated learning is higher than the accuracy threshold included in the screening information; The type of federated learning in which the candidate member participates is the same as the type of federated learning included in the screening information.

[0079] For example, the screening information includes a supported algorithm type, and the supported algorithm type is a deep learning algorithm, and the first network element screens 80 devices that want to participate in federated learning based on the first instruction information, and the first network element can further screen the 80 devices based on the algorithm information that can be supported to participate in federated learning obtained from the third network element. For example, when it is obtained that 50 of the devices can not only support the deep learning algorithm but also support the linear regression algorithm when participating in federated learning, and 30 devices only support the deep learning algorithm when participating in federated learning, the first network element can further screen the first device that can not only support the deep learning algorithm but also support the linear regression algorithm as a candidate network element based on the algorithm type that can be supported when participating in federated learning.

[0080] It should be understood that in cases where the capability information includes other information, the implementation is similar to the implementation of algorithm types that can support participation in federated learning included in the capability information, and will not be described further here.

[0081] Here, the first network element can obtain the willingness information and capability information of each device to participate in federated learning from the UDM / DCAF / NRF / these third network elements, where the UDM / DCAF / NRF / are network elements with memory capability.

[0082] Optionally, the request message sent by the second network element further includes second indication information, and the second indication information is used to indicate a service type corresponding to the federated learning. When determining candidate members, the first network element may determine one or more first devices as candidate members from the plurality of second devices based on the screening information and the second indication information, where the first devices are devices that can support a service corresponding to the service type.

[0083] Specifically, the second indication information may be an analytic ID, which is used to indicate a service type corresponding to federated learning. In a data analysis task (i.e., other tasks handled by the NWDAF), the analytic ID may be as follows: "UE mobility" is used to indicate that the current task is an analysis task related to UE mobility, and "NF load" is used to indicate that the current task is an analysis task related to network element load.

[0084] Generally, when the service types corresponding to the federated learning are different, the selected candidate members may be different, so when the first network element selects the candidate members, it can further screen the first device capable of supporting the service corresponding to the service type from the plurality of second devices as the candidate members based on the service type indicated by the screening information and the analytic ID. For example, the second indication information is used to indicate that the service type corresponding to the federated learning is an analysis task related to terminal mobility, and the selected first device is a device capable of supporting the analysis task related to terminal mobility.

[0085] In this embodiment, the accuracy of the selected candidate members can be improved by further screening the candidate members based on the service type corresponding to the federated learning, thereby improving the training efficiency of the federated learning and the accuracy of the model after training.

[0086] In one embodiment, the first network element determining one or more first devices as candidate members capable of participating in federated learning based on the screening information includes: The first network element obtains network status information corresponding to the first device, the network status information comprising: type information of the first device; Location information of the first device; Wireless access method information in which the first device is located; and wireless signal quality information of the first device. The first network element determines candidate members that can participate in federated learning from the first device based on network status information corresponding to the first device and the screening information, and the network status information corresponding to the candidate members matches the screening information.

[0087] It should be explained that the first network element may obtain network status information corresponding to the first device from a third device, and the third device may be an AMF, a UDM, an NRF, a PCF, a RAN, or an operation, administration and maintenance device (OAM Operation Administration and Maintenance). For example, the first network element may obtain type information of the first device from a UDM or an NRF, or the first network element may obtain radio access method information in which the first device is located from an AMF or a PCF, or the first network element may obtain radio signal quality information of the first device from a RAN or an OAM, or the first network element may obtain UE accumulated usage information, etc. from an SMF or a CHF.

[0088] Wherein, the network status information corresponding to the candidate member matches the screening information, The wireless access system in which the candidate member is located is the same as the wireless access system included in the screening information; the wireless signal quality information of the candidate member is greater than the signal quality requirement included in the screening information; The position of the candidate member is located within an AOI included in the screening information; and and the type of the candidate member is included in member type information included in the screening information.

[0089] Optionally, the request message further includes sorting instruction information, which is used to instruct sorting the candidate members according to second information, where the second information includes at least one of signal quality of the candidate members, accuracy information of the candidate members, and traffic information of the candidate members, and when the request message includes the sorting instruction information, the identifier order of the candidate members in the response message is the order obtained after sorting according to the second information.

[0090] Specifically, when the request message includes sorting instruction information, the first network element, after determining the candidate members, sorts the determined candidate members in ascending or descending order according to the second information in the sorting instruction information, for example, sorting in ascending or descending order according to the signal quality of the candidate members, or sorting in ascending or descending order according to the accuracy information of the candidate members, or sorting in ascending or descending order according to the traffic information of the candidate members.

[0091] When the first network element sends a response message to the second network element, the identifier order of the candidate members in the response message is the order obtained after sorting according to the second information.

[0092] In this embodiment, the first network element sorts the determined candidate members according to the second information; thus, after receiving the identifier information of the candidate members sent by the first network element, the second network element can screen the candidate members that further meet its requirements based on the sorting information, and participate in subsequent federated learning, thereby improving the efficiency of federated learning.

[0093] Optionally, the request message further includes packet indication information, which is used to indicate the candidate members to packetize according to the third information, where the third information includes at least one of the following: an area where the candidate members are located, a time period during which the candidate members are in a network-connected state, an algorithm type supported by the candidate members, accuracy information of the candidate members, a radio access scheme of the candidate members, and a signal quality of the candidate members, and when the request message includes the packet indication information, the identifiers of the candidate members in the response message are the identifiers obtained after sorting according to the third information.

[0094] Specifically, when the request message includes packet indication information, the first network element determines the candidate members, and then packetizes the determined candidate members according to the third information in the packet indication information. For example, the packetization may be performed according to the area where the candidate members are located, for example, dividing the candidate members in the same area into the same group. Or the packetization may be performed according to the time period in which the candidate members are in a network connection state, for example, dividing the candidate members in a network connection state in the same time period into the same group, or the packetization may be performed according to the algorithm type supported by the candidate members, for example, dividing the network elements supporting the same algorithm type into the same group, or the packetization may be performed according to the accuracy information of the candidate members, for example, dividing the candidate members with the same accuracy information into the same group, or the packetization may be performed according to the radio access method of the candidate members, for example, dividing the candidate members with the same radio access method into the same group, or the packetization may be performed according to the signal quality of the candidate members, or the packetization may be performed according to the traffic information of the candidate members, or the packetization may be performed according to the time period in which the candidate members can perform federated learning, for example, dividing the candidate members with the same time period in which they can perform federated learning into the same group, for example, from 10:00 p.m. to 12:00 p.m.

[0095] When the first network element sends a response message to the second network element, the identifier of the candidate member in the response message is the identifier obtained after packetizing according to the third information.

[0096] In this embodiment, the first network element packets the determined candidate members according to the third information. In this way, after receiving the identifier information of the candidate members sent by the first network element, the second network element can further screen the candidate members that are more suitable for its own requirements based on the packet information, and participate in the subsequent federated learning, thereby improving the efficiency of federated learning.

[0097] Step 203: The first network element sends a response message to the second network element, where the response message includes an identifier of the candidate member.

[0098] In this step, after the first network element determines the candidate members that can participate in federated learning, it can carry the identifiers of these candidate members in a response message and send them to the second network element, where the identifiers of the candidate members may include the Subscriber Permanent Identifier (SUPI) or the Internet Protocol (IP) of the candidate members, and the identifiers of the candidate members may further be the Generic Public Subscription Identifier (GPSI), the International Mobile Subscriber Identity (IMSI), the AF specific UE ID or the UE IP address.

[0099] Optionally, the response message further includes at least one of the area in which the candidate member is located, the time period during which the candidate member is in a network connection state, the algorithm type supported by the candidate member, the accuracy information of the candidate member, the radio access method of the candidate member, the signal quality of the candidate member, the traffic information of the candidate member, and the time period during which the candidate member is available for federated learning.

[0100] Here, the region in which the candidate member is located may include the location of the candidate member, such as the longitude and latitude or the cell.

[0101] The time period during which the candidate members can perform federated learning is a predicted result analyzed by the first network element based on the intention information and capability information of each device.

[0102] The signal quality of a candidate member may be represented using the mean, variance or RTT of the RSSI when connected to the WLAN.

[0103] The traffic information of the potential member may include information such as traffic values ​​consumed during a preset time period, for example, the number of uploads and downloads.

[0104] Optionally, the response message may further include second indication information, which may be an analytic ID, which is used to indicate a service type corresponding to the federated learning.

[0105] Optionally, the response message may further include the coverage time percentage during which the network is connected, for example how much of the day the wifi is connected.

[0106] Optionally, the response message may further include address information of the candidate member, so that the second network element can find the candidate member based on the address information, thereby performing connection and association learning.

[0107] Optionally, the response message may further include number information, which may be understood as the number of members participating in the federated learning, i.e., the number of candidate members that the first network element needs to determine.

[0108] Furthermore, after receiving the response message sent by the first network element, the second network element selects target members to participate in the real federated learning according to the situation of performing the actual federated learning training.

[0109] For example, the second network element can determine the members to perform federated learning based on the signal quality of the candidate members in the response message, and can select the top 100 members with the best signal quality to perform federated learning. Of course, the method of determining the target member based on other information in the response message is similar to the method of determining the target member based on the signal quality, and will not be described further here.

[0110] Alternatively, when the request message sent by the second network element to the first network element includes number information, if the response message returned by the first network element includes a target number of candidate members, the second network element may directly select the target number of candidate members as target members. For example, when the request message includes a request for 50 federated learning members and the response message returned by the first network element carries 50 candidate members, the second network element may directly select the 50 members as target members.

[0111] Furthermore, after determining the target members, the second network element connects with these target members based on the identifier information of the target members to perform federated learning.

[0112] In this embodiment, the second network element can further screen target members to participate in true federated learning based on the candidate members returned by the first network element, so that the screened target members are more suitable to participate in federated learning and improve the efficiency of federated learning.

[0113] A method for determining a candidate member according to an embodiment of the present application includes receiving a request message sent by a second network element, determining one or more first devices as candidate members capable of participating in federated learning based on screening information included in the request message, and sending identifier information of the candidate members to the second network element, where the screening information includes at least one of a time period, an algorithm type, a precision threshold, a wireless access method, a signal quality requirement, a traffic range, member type information, number information, area information, and federated learning type information. The first network element can improve the efficiency of federated learning by screening participating members suitable for performing federated learning through the screening information.

[0114] 3 is a flowchart of another candidate member determination method according to the present application. In this embodiment, the execution body is a second network element. As shown in FIG. 3, the method includes:

[0115] Step 301: A second network element sends a request message to a first network element.

[0116] Here, the request message includes screening information, and the request message is used to instruct the first network element to determine one or more first devices as candidate members capable of participating in federated learning based on the screening information.

[0117] Here, the screening information is A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0118] Step 302: The second network element receives a response message sent by the first network element, where the response message includes an identifier of the candidate member.

[0119] The specific implementation process and beneficial effects in the embodiments of the present application may refer to the content of the embodiment shown in FIG. 2, and will not be further described here.

[0120] Optionally, the first device comprises: A first device that is in a network-connected state during the time period; A first device that supports performing federated learning during the time period; a first device that supports the algorithm type; a first device that supports federated learning using said algorithm type; A first device whose model training accuracy information is greater than the accuracy threshold; a first device in the wireless access system; A first device that supports federated learning in the wireless access scheme; a first device having a signal quality greater than a signal quality requirement; a first device whose consumed traffic is in the traffic range; a first device that satisfies the type requirements of the candidate member; A first device located within the AOI; and and a first device whose type of associative learning is the same as the type information of associative learning included in the screening information.

[0121] Optionally, the request message further includes instruction information, which is used to instruct the execution of a task of determining candidate members to participate in federated learning, or to indicate that the current task is for federated learning or member selection for federated learning.

[0122] Optionally, the request message further includes sorting instruction information, the sorting instruction information is used to instruct sorting the candidate members according to second information, the second information including at least one of the signal quality of the candidate members, the accuracy information of the candidate members, and the traffic information of the candidate members; The identifiers of the candidate members in the response message are the identifiers obtained after sorting according to the second information.

[0123] Optionally, the request message further includes packet indication information, and the packet indication information is used to instruct the candidate member to packetize according to third information, and the third information includes at least one of the following: an area where the candidate member is located, a time period when the candidate member is in a network connection state, an algorithm type supported by the candidate member, accuracy information of the candidate member, a radio access scheme of the candidate member, and a signal quality of the candidate member; The identifier of the candidate member in the response message is the identifier obtained after packetizing according to the third information.

[0124] Optionally, the response message further includes the area in which the candidate member is located, the time period during which the candidate member is in a network connected state, the algorithm type supported by the candidate member, the accuracy information of the candidate member, the radio access method of the candidate member, the signal quality of the candidate member, the traffic information of the candidate member, and the time period during which the candidate member is available for federated learning.

[0125] Optionally, the method further comprises: The method further includes the second network element determining, based on the identifiers of the candidate members, target members to participate in the federated learning.

[0126] A method for determining a candidate member according to an embodiment of the present application includes receiving a request message sent by a second network element, determining one or more first devices as candidate members capable of participating in federated learning based on screening information included in the request message, and sending identifier information of the candidate members to the second network element, where the screening information includes at least one of a time period, an algorithm type, a precision threshold, a wireless access method, a signal quality requirement, a traffic range, member type information, number information, area information, and federated learning type information. The first network element can improve the efficiency of federated learning by screening a participating member suitable for performing federated learning based on the screening information. In addition, the second network element can further screen based on the candidate members sent by the first network element to obtain a member suitable for participating in true federated learning, thereby improving the efficiency of federated learning.

[0127] The specific implementation process and technical effect of the method of this embodiment are similar to those of the embodiment of the method on the first network element side, and may refer to the detailed introduction in the embodiment of the method on the first network element side, and will not be described further here.

[0128] 4 is a signaling diagram of a method for determining candidate members according to the present application. As shown in FIG. 4, the method includes:

[0129] Step 401: A second network element sends a request message to a first network element.

[0130] Here, the request message includes screening information, which includes at least one of a time period, an algorithm type, an accuracy threshold, a radio access method, a signal quality requirement, a traffic range, member type information, number information, area information, and federated learning type information.

[0131] Step 402: The first network element determines one or more first devices as candidate members capable of participating in federated learning based on the screening information.

[0132] Step 403: The first network element sends a response message to the second network element, where the response message includes an identifier of the candidate member.

[0133] A method for determining a candidate member according to an embodiment of the present application includes receiving a request message sent by a second network element, determining one or more first devices as candidate members that can participate in federated learning based on screening information included in the request message, and sending identifier information of the candidate members to the second network element, where the screening information includes at least one of a time period, an algorithm type, a precision threshold, a wireless access method, a signal quality requirement, a traffic range, member type information, number information, an area of ​​interest AOI, and a type of federated learning. The first network element can improve the training efficiency of federated learning by screening participating members that are suitable for performing federated learning through the screening information.

[0134] The specific implementation process and technical effect of the method of this embodiment are similar to those of the embodiment of the method on the first network element side, and may refer to the detailed introduction in the embodiment of the method on the first network element side, and will not be described further here.

[0135] In the method for determining a candidate member according to the embodiment of the present application, the execution body may be a device for determining a candidate member. In the embodiment of the present application, the device for determining a candidate member executes the method for determining a candidate member as an example, and the device for determining a candidate member according to the embodiment of the present application will be described.

[0136] FIG. 5 is a structural diagram of a candidate member determination device according to the present application. As shown in FIG. 5, the candidate member determination device according to the present embodiment includes: a receiving module 11 for receiving a request message sent by a second network element, said request message including screening information; a processing module 12 for determining one or more first devices as candidate members capable of participating in federated learning based on the screening information; a sending module 13 for sending a response message to the second network element, the response message including an identifier of the candidate member; wherein the screening information is A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0137] In the device of this embodiment, the receiving module receives a request message sent by a second network element, the request message includes screening information, the processing module determines one or more first devices as candidate members that can participate in federated learning based on the screening information, and the sending module sends a response message to the second network element, the response message includes identifiers of the candidate members, where the screening information includes at least one of a time period, an algorithm type, a precision threshold, a wireless access method, a signal quality requirement, a traffic range, member type information, number information, an area of ​​interest AOI, and a type of federated learning. The candidate member determination device can improve the training efficiency of federated learning by screening participating members suitable for performing federated learning according to the screening information.

[0138] Optionally, the first device comprises: A first device that is in a network-connected state during the time period; A first device that supports performing federated learning during the time period; a first device that supports the algorithm type; a first device that supports federated learning using said algorithm type; A first device whose model training accuracy information is greater than the accuracy threshold; a first device in the wireless access system; A first device that supports federated learning in the wireless access scheme; a first device having a signal quality greater than a signal quality requirement; a first device whose consumed traffic is in the traffic range; a first device that satisfies the type requirements of the candidate member; A first device located within the AOI; and and a first device whose type of associative learning is the same as the type information of associative learning included in the screening information. Optionally, the processing module 12 includes: determining a data type corresponding to the screening information; Obtaining attribute information corresponding to each of the at least one device from the at least one third network element based on the data type; The candidate members are determined based on the attribute information and the screening information.

[0139] Optionally, the data type is: The information includes at least one of network connection information and corresponding time period, algorithm type, accuracy information, radio access method, signal quality, and traffic.

[0140] Optionally, the processing module 12 is Obtaining first indication information from a third network element for each second device to indicate its intention to participate in federated learning, the intention information indicating whether each second device wishes to participate in federated learning; Determining one or more first devices as candidate members based on the screening information and the first instruction information, the first devices being among the second devices that wish to participate in federated learning and that satisfy the screening information.

[0141] Optionally, the processing module 12 is acquiring, from a third network element, information indicating that the first device intends to participate in the federated learning; Based on the intention information and the screening information, a first device that wishes to participate in the federated learning and matches the screening information is used to determine as a candidate member that can participate in the federated learning.

[0142] Optionally, the processing module 12 is Acquiring capability information of federated learning of the first device, the capability information including at least one of a radio access method participating in federated learning, an area participating in federated learning, a time for participating in federated learning, algorithm information that can support participation in federated learning, accuracy information that can be achieved by participation in federated learning, and a type of participation in federated learning; Based on the capability information and the screening information, candidate members capable of participating in federated learning from the first device are determined, and the capability information of the candidate members is used to match the screening information.

[0143] Here, the matching of the candidate member's ability information with the screening information is The wireless access method in which the candidate member participates in federated learning is the same as the wireless access method included in the screening information; The area in which the candidate member participates in the federated learning is located within the AOI included in the screening information; The time when the candidate member participates in the federated learning is within the time period included in the screening information; The screening information includes an algorithm type that can support the candidate member's participation in federated learning; and The accuracy information achievable for the candidate member to participate in the federated learning is higher than the accuracy threshold included in the screening information; and the type of federated learning in which the candidate member participates is the same as the type of federated learning included in the screening information.

[0144] Optionally, the processing module 12 is obtaining network status information corresponding to the first device, the network status information comprising: type information of the first device; Location information of the first device; Wireless access method information in which the first device is located; and wireless signal quality information of the first device. Based on the network status information corresponding to the first device and the screening information, candidate members capable of participating in federated learning from the first device are determined, and the network status information corresponding to the candidate members is used to match the screening information.

[0145] Wherein, the network status information corresponding to the candidate member matches the screening information, The wireless access system in which the candidate member is located is the same as the wireless access system included in the screening information; the wireless signal quality information of the candidate member is greater than the signal quality requirement included in the screening information; The position of the candidate member is located within an AOI included in the screening information; and and the type of the candidate member is included in member type information included in the screening information.

[0146] Optionally, the processing module comprises: The present invention is used for obtaining capability information for each second device to participate in federated learning, the capability information including at least one of a radio access method to participate in federated learning, an area to participate in federated learning, a time to participate in federated learning, algorithm information that can support participation in federated learning, accuracy information achievable for participation in federated learning, and a type to participate in federated learning, and determining one or more first devices as candidate members based on the screening information, the first instruction information and the capability information, the first device being a device among the second devices that wishes to participate in federated learning and whose capability information for federated learning satisfies the screening information.

[0147] Here, the ability information to participate in the associative learning satisfies the screening information. The wireless access system of the candidate member participating in the federated learning is the same as the wireless access system included in the screening information; The time when the candidate member participates in the federated learning is within the time period included in the screening information; The screening information includes an algorithm type that can support the candidate member's participation in federated learning; and The accuracy information achievable for the candidate member to participate in the federated learning is higher than the accuracy threshold included in the screening information; and a type of the candidate member to participate in the federated learning is the same as a type of the federated learning included in the screening information. Optionally, the request message further includes second instruction information, the second instruction information being used to indicate a service type corresponding to the federated learning; The processing module includes: used to determine one or more first devices as the candidate members based on the screening information and the second instruction information; Here, the first device is a device capable of supporting a service corresponding to the service type.

[0148] Optionally, the request message further includes sorting instruction information, the sorting instruction information is used to instruct the candidate members to be sorted according to second information, the second information includes at least one of the signal quality of the candidate members, the accuracy information of the candidate members, and the traffic information of the candidate members; The order of the identifiers of the candidate members in the response message is the order obtained after sorting according to the second information.

[0149] Optionally, the request message further includes packet indication information, and the packet indication information is used to indicate that the candidate members are to be packetized according to third information, and the third information includes: The candidate member includes at least one of the following: an area where the candidate member is located; a time period when the candidate member is in a network connection state; an algorithm type supported by the candidate member; accuracy information of the candidate member; a radio access scheme of the candidate member; and a signal quality of the candidate member; The identifier of the candidate member in said response message is the identifier obtained after packetizing according to said third information.

[0150] Optionally, the response message further includes at least one of the area in which the candidate member is located, the time period during which the candidate member is in a network connection state, the algorithm type supported by the candidate member, the accuracy information of the candidate member, the radio access method of the candidate member, the signal quality of the candidate member, the traffic information of the candidate member, and the time period during which the candidate member is available for federated learning.

[0151] The device of this embodiment can be used to implement the method of any one of the embodiments of the method on the first network element side described above, and its specific implementation process and technical effect are similar to those in the embodiments of the method on the first network element side. For details, please refer to the detailed introduction in the embodiments of the method on the first network element side, and no further description will be given here.

[0152] FIG. 6 is a second structural diagram of the candidate member determination device according to the present application. As shown in FIG. 6, the candidate member determination device according to the present embodiment includes: A sending module 21 for sending a request message to a first network element, the request message including screening information, the request message being used to instruct the first network element to determine one or more first devices as candidate members that can participate in federated learning based on the screening information; a receiving module 22 for receiving a response message transmitted by the first network element, the response message including an identifier of the candidate member; wherein the screening information is A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0153] Optionally, the first device comprises: A first device that is in a network-connected state during the time period; A first device that supports performing federated learning during the time period; a first device that supports the algorithm type; a first device that supports federated learning using said algorithm type; A first device whose model training accuracy information is greater than the accuracy threshold; a first device in the wireless access system; A first device that supports federated learning in the wireless access scheme; a first device having a signal quality greater than a signal quality requirement; a first device whose consumed traffic is in the traffic range; a first device that satisfies the type requirements of the candidate member; A first device located within the AOI; and and a first device whose type of federated learning is the same as the type information of federated learning included in the screening information. Optionally, the request message further includes instruction information, which is used to instruct to execute a task of determining candidate members to participate in the federated learning, or to instruct that the current task is for federated learning or for selecting members of the federated learning.

[0154] Optionally, the request message further includes sorting instruction information, the sorting instruction information is used to instruct the candidate members to be sorted according to second information, the second information includes at least one of the signal quality of the candidate members, the accuracy information of the candidate members, and the traffic information of the candidate members; The identifiers of the potential members in the response message are the identifiers obtained after sorting according to the second information.

[0155] Optionally, the request message further includes packet indication information, which is used to instruct the candidate member to packetize according to third information, and the third information includes at least one of the following: the area where the candidate member is located, the time period when the candidate member is in a network connection state, the algorithm type supported by the candidate member, the accuracy information of the candidate member, the radio access scheme of the candidate member, and the signal quality of the candidate member; The identifier of the candidate member in said response message is the identifier obtained after packetizing according to said third information.

[0156] Optionally, the response message further includes the area in which the candidate member is located, the time period during which the candidate member is in a network connection state, the algorithm type supported by the candidate member, the accuracy information of the candidate member, the radio access method of the candidate member, the signal quality of the candidate member, the traffic information of the candidate member, and the time period during which the candidate member is available for federated learning.

[0157] Optionally, the apparatus further comprises a processing module 23, The processing module 23 is used to determine target members to participate in federated learning based on the identifiers of the candidate members.

[0158] The device of this embodiment can be used to implement the method of any one of the embodiments of the method on the second network element side described above, and its specific implementation process and technical effect are similar to those in the embodiments of the method on the second network element side. For details, please refer to the detailed introduction in the embodiments of the method on the second network element side, and no further description will be given here.

[0159] The candidate member determination device in the embodiment of the present application may be an electronic device, for example an electronic device having an operating system, or may be a component in an electronic device, for example an integrated circuit or chip. The electronic device may be a terminal or other device other than a terminal. Exemplarily, the terminal may include, but is not limited to, the types of terminal 11 listed above, and the other device may be a server, a network attached storage (NAS), etc., and the embodiment of the present application is not specifically limited.

[0160] The candidate member determination apparatus according to the embodiment of the present application can realize each process realized by the method embodiments of Figures 2 to 4 and achieve the same technical effects, and will not be described further here to avoid repetition of description.

[0161] Optionally, as shown in Fig. 7, an embodiment of the present application further provides a communication device 700, which includes a processor 701 and a memory 702, and the memory 702 stores a program or instruction that can run on the processor 701. For example, if the communication device 700 is a terminal, when the program or instruction is executed by the processor 701, each step of the embodiment of the method for determining a candidate member can be realized and the same technical effect can be achieved. If the communication device 700 is a network side device, when the program or instruction is executed by the processor 701, each step of the embodiment of the method for determining a candidate member can be realized and the same technical effect can be achieved, and in order to avoid repetition, no further description will be given here.

[0162] The embodiment of the present application further provides a first network element, which includes a processor and a communication interface, and the processor is used to determine one or more first devices as candidate members that can participate in federated learning based on screening information, and the communication interface is used to receive a request message sent by a second network element and send a response message to the second network element, where the response message includes an identifier of the candidate member. Wherein, the request message includes screening information, and the screening information includes at least one of a time period, an algorithm type, an accuracy threshold, a radio access method, a signal quality requirement, a traffic range, a member type information, a number information, an area information, and a type information of federated learning. This embodiment of the terminal corresponds to the embodiment of the method on the terminal side, and each implementation process and realization manner of the embodiment of the method can be applied to the embodiment of the terminal, and the same technical effect can be achieved. Specifically, FIG. 8 is a hardware structural schematic diagram for realizing the first network element of the embodiment of the present application.

[0163] The first network element 800 includes at least some components such as, but not limited to, a radio frequency unit 801, a network module 802, an audio output unit 803, an input unit 804, a sensor 805, a display unit 806, a user input unit 807, an interface unit 808, a memory 809 and a processor 810.

[0164] As can be understood by those skilled in the art, the first network element 800 may further include a power source (e.g., a battery) for powering each component, and the power source may be logically connected to the processor 810 by a power management system, so that the power management system can realize functions such as charging / discharging management and power consumption management. The first network element structure shown in Figure 8 does not constitute a limitation on the first network element, and the first network element may include more or less components than the number of components shown, or a combination of some components, or a different arrangement of components, and will not be further described here.

[0165] It should be understood that in the embodiment of the present application, the input unit 804 may include a graphics processing unit (GPU) 8041 and a microphone 8042, and the graphics processor 8041 processes image data of still or video images obtained by an image capture device (e.g., a camera) in a video capture mode or an image capture mode. The display unit 806 may include a display panel 8061, and the display panel 8061 may be configured in the form of a liquid crystal display, an organic light emitting diode, etc. The user input unit 807 includes at least one of a touch panel 8071 and other input devices 8072. The touch panel 8071 is also called a touch screen. The touch panel 8071 may include two parts: a touch detection device and a touch controller. The other input devices 8072 may include, but are not limited to, a physical keyboard, function keys (e.g., volume control buttons, switch buttons, etc.), a trackball, a mouse, and an operation lever, which will not be described further herein.

[0166] In the embodiment of the present application, after receiving downlink data from the network side device, the radio frequency unit 801 can transmit the downlink data to the processor 810 for processing, and the radio frequency unit 801 can transmit uplink data to the network side device. In general, the radio frequency unit 801 includes, but is not limited to, an antenna, an amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, etc.

[0167] The memory 809 may be used to store software programs or instructions and various data. The memory 809 may include a first storage area that mainly stores programs or instructions and a second storage area that stores data. Here, the first storage area can store an operating system, an application program or instructions required for at least one function (e.g., a sound playback function, an image playback function, etc.), and the like. Note that the memory 809 may include a volatile memory or a non-volatile memory, or the memory 809 may include both a volatile and a non-volatile memory. Here, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be Random Access Memory (RAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), Synch link Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM). Memory x09 in the embodiments of the present application includes, but is not limited to, these and any other suitable types of memory.

[0168] The processor 810 may include one or more processing units. Optionally, the processor 810 integrates an application processor and a modem processor, where the application processor mainly processes operations related to the operating system, user interface, application programs, etc., and the modem processor mainly processes wireless communication signals, such as a baseband processor. As can be understood, the modem processor does not have to be integrated into the processor 810.

[0169] Here, the radio frequency unit 801 is used to receive a request message sent by a second network element, the request message including screening information, and send a response message to the second network element, the response message including an identifier of the candidate member.

[0170] The processor 810 is used to determine one or more first devices as candidate members capable of participating in federated learning based on the screening information, the screening information comprising: A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to vertical federation or horizontal federation.

[0171] In the above embodiment, the first network element receives a request message sent by the second network element, and determines one or more first devices as candidate members that can participate in federated learning based on screening information included in the request message, and sends identifier information of the candidate members to the second network element, where the screening information includes at least one of a time period, an algorithm type, a precision threshold, a wireless access method, a signal quality requirement, a traffic range, member type information, number information, an area of ​​interest (AOI), and a type of federated learning. The first network element can improve the training efficiency of federated learning by screening participating members that are suitable for performing federated learning based on the screening information.

[0172] Optionally, the first device comprises: A first device that is in a network-connected state during the time period; A first device that supports performing federated learning during the time period; a first device that supports the algorithm type; a first device that supports federated learning using said algorithm type; A first device whose model training accuracy information is greater than the accuracy threshold; a first device in the wireless access system; A first device that supports federated learning in the wireless access scheme; a first device having a signal quality greater than a signal quality requirement; a first device whose consumed traffic is in the traffic range; a first device that satisfies the type requirements of the candidate member; A first device located within the AOI; and and a first device whose type of associative learning is the same as the type information of associative learning included in the screening information.

[0173] Optionally, the processor 810 further determines a data type corresponding to the screening information; Obtaining attribute information corresponding to each of the at least one second device from the at least one third network element based on the data type; The candidate members are determined based on the attribute information and the screening information.

[0174] Optionally, the data type is: The information includes at least one of network connection information and corresponding time period, algorithm type, accuracy information, radio access method, signal quality, and traffic.

[0175] Optionally, the processor 810 further obtains, from the third network element, first indication information for each second device to indicate its willingness to participate in federated learning, the willingness information indicating whether each device wishes to participate in federated learning; Determining one or more first devices as the candidate members based on the screening information and the first instruction information, the first devices being among the second devices that wish to participate in federated learning and that satisfy the screening information.

[0176] Optionally, processor 810 is used to obtain intention information of the first device to participate in the federated learning from a third network element, and based on the intention information and the screening information, determine a first device that wishes to participate in the federated learning and matches the screening information as a candidate member that can participate in the federated learning.

[0177] Optionally, processor 810 is used to obtain capability information of federated learning of the first device, the capability information including at least one of a radio access method participating in federated learning, an area participating in federated learning, a time for participating in federated learning, algorithm information that can support participation in federated learning, accuracy information achievable for participation in federated learning, and a type for participating in federated learning, and to determine candidate members that can participate in federated learning from the first device based on the capability information and the screening information, wherein capability information of the candidate members is matched to the screening information.

[0178] Here, the matching of the candidate member's ability information with the screening information is The wireless access method in which the candidate member participates in federated learning is the same as the wireless access method included in the screening information; The area in which the candidate member participates in the federated learning is located within the AOI included in the screening information; The time when the candidate member participates in the federated learning is within the time period included in the screening information; The screening information includes an algorithm type that can support the candidate member's participation in federated learning; and The accuracy information achievable for the candidate member to participate in the federated learning is higher than the accuracy threshold included in the screening information; and the type of federated learning in which the candidate member participates is the same as the type of federated learning included in the screening information.

[0179] Optionally, processor 810 is adapted to obtain network condition information corresponding to the first device, the network condition information being: type information of the first device; Location information of the first device; Wireless access method information in which the first device is located; and wireless signal quality information of the first device. Based on the network status information corresponding to the first device and the screening information, candidate members capable of participating in federated learning from the first device are determined, and the network status information corresponding to the candidate members is used to match the screening information.

[0180] Wherein, the network status information corresponding to the candidate member matches the screening information, The wireless access system in which the candidate member is located is the same as the wireless access system included in the screening information; the wireless signal quality information of the candidate member is greater than the signal quality requirement included in the screening information; The position of the candidate member is located within an AOI included in the screening information; and and the type of the candidate member is included in member type information included in the screening information.

[0181] Optionally, processor 810 is used to obtain capability information for each second device to participate in federated learning, the capability information including at least one of a radio access method to participate in federated learning, an area to participate in federated learning, a time to participate in federated learning, algorithm information that can support participation in federated learning, accuracy information achievable for participation in federated learning, and a type to participate in federated learning, and to determine one or more first devices as the candidate members based on the screening information, the first instruction information and the capability information, wherein the first device is a device among the second devices that wishes to participate in federated learning and whose capability information for federated learning satisfies the screening information.

[0182] Here, the ability information to participate in the associative learning satisfies the screening information. The wireless access system of the candidate member participating in the federated learning is the same as the wireless access system included in the screening information; The time when the candidate member participates in the federated learning is within the time period included in the screening information; The screening information includes an algorithm type that can support the candidate member's participation in federated learning; and The accuracy information achievable for the candidate member to participate in the federated learning is higher than the accuracy threshold included in the screening information; The type of federated learning in which the candidate member participates is the same as the type of federated learning included in the screening information.

[0183] Optionally, the request message further includes second indication information, and the second indication information is used to indicate a service type corresponding to the federated learning; The processor 810 is further adapted to determine one or more first devices as the candidate members based on the screening information and the second indication information; Here, the first device is a device capable of supporting a service corresponding to the service type.

[0184] Optionally, the request message further includes sorting instruction information, the sorting instruction information is used to instruct the candidate members to be sorted according to second information, the second information includes at least one of the signal quality of the candidate members, the accuracy information of the candidate members, and the traffic information of the candidate members; The order of the identifiers of the candidate members in the response message is the order obtained after sorting according to the second information.

[0185] Optionally, the request message further includes packet indication information, and the packet indication information is used to indicate that the candidate members are to be packetized according to third information, and the third information includes: The candidate member includes at least one of the following: an area where the candidate member is located; a time period when the candidate member is in a network connection state; an algorithm type supported by the candidate member; accuracy information of the candidate member; a radio access scheme of the candidate member; and a signal quality of the candidate member; The identifier of the candidate member in said response message is the identifier obtained after packetizing according to said third information.

[0186] Optionally, the response message further includes at least one of the area in which the candidate member is located, the time period during which the candidate member is in a network connection state, the algorithm type supported by the candidate member, the accuracy information of the candidate member, the radio access method of the candidate member, the signal quality of the candidate member, the traffic information of the candidate member, and the time period during which the candidate member is available for federated learning.

[0187] In the above embodiment, the first network element determines one or more first devices as candidate members that can participate in federated learning by receiving a request message sent by the second network element and based on the screening information included in the request message, and sends the identifier information of the candidate members to the second network element, where the screening information includes at least one of a time period, an algorithm type, a precision threshold, a wireless access method, a signal quality requirement, a traffic range, member type information, number information, an area of ​​interest (AOI), and a type of federated learning. The first network element can improve the training efficiency of federated learning by screening the participating members that are suitable for performing federated learning through the screening information.

[0188] Specifically, an embodiment of the present application further provides a second network element 900. As shown in Fig. 9, the second network element 900 includes a processor 901, a network interface 902, and a memory 903. Here, the network interface 902 is, for example, a common public radio interface (CPRI).

[0189] Specifically, the network side device 900 of the embodiment of the present application further includes instructions or programs stored in memory 903 and capable of running on the processor 901, and the processor 901 calls the instructions or programs in the memory 903 to execute the method performed by each module shown in FIG. 6, and achieves the same technical effect, which will not be described further here in order to avoid repetition of description.

[0190] An embodiment of the present application further provides a readable storage medium, in which a program or instruction is stored, which, when executed by a processor, can realize each process of the embodiment of the above-mentioned candidate member determination method and achieve the same technical effect, and will not be described further here in order to avoid repetition of description.

[0191] Wherein, the processor is the processor in the terminal according to the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read only memory ROM, a random access memory RAM, a magnetic disk, or an optical disk.

[0192] An embodiment of the present application further provides a chip, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run a program or instruction to realize each process of the embodiment of the above-mentioned candidate member determination method, and the same technical effect can be achieved, and in order to avoid repetition of description, no further description will be given here.

[0193] It should be understood that the chips referred to in the embodiments of the present application may be referred to as system level chips, system chips, chip systems, or systems on chips.

[0194] An embodiment of the present application further provides a computer program / program product, which is stored in a storage medium, and which can be executed by at least one processor to realize each process of the embodiment of the above-mentioned candidate member determination method and achieve the same technical effects, which will not be described further here in order to avoid repetition of description.

[0195] An embodiment of the present application further provides a candidate member determination system, including a first network element and a second network element, wherein the first network element may be used to perform steps of the candidate member determination method described above, and the second network element may be used to perform steps of the candidate member determination method described above.

[0196] It should be explained that in this specification, the terms "comprise", "include", "includes" or any other variants thereof are intended to cover the non-exclusive "comprise", whereby a process, method, article or apparatus that includes a set of elements includes not only those elements, but also other elements not expressly listed or inherent to such process, method, article or apparatus. Absent any further limitations, an element limited by the phrase "comprises one or more" does not preclude the presence of other identical elements in the process, method, article or apparatus that includes the element. It should be pointed out that the scope of the method and apparatus in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may include performing functions in an essentially simultaneous manner or in the reverse order based on the functions involved, for example, the described method can be performed in a different order than described, and various steps can be added, omitted or combined. Also, features described with reference to some examples can be combined in other examples.

[0197] From the above description of the embodiments, it is clear to those skilled in the art that the methods of the above embodiments can be realized in the form of software and a necessary general-purpose hardware platform. Of course, they can also be realized in hardware, but in many cases the former is a more preferred embodiment. In light of this understanding, the technical proposal of the present application may be substantially or the part that contributes to the prior art may be embodied in the form of a computer software product, which is stored in a storage medium (e.g., ROM / RAM, magnetic disk, optical disk) and includes some instructions for causing a terminal (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0198] The above describes the embodiments of the present application in conjunction with the drawings, but the present application is not limited to the above specific embodiments. The above specific embodiments are merely illustrative and not limiting. Those skilled in the art can take the suggestions of this application and make many forms without departing from the spirit and scope of the claims of this application, all of which belong to the protection scope of this application.

Claims

1. A method for determining candidate members, comprising the steps of: receiving, by a first network element, a request message sent by a second network element, the request message including screening information; The first network element determines one or more first devices as candidate members capable of participating in federated learning based on the screening information; the first network element sending a response message to the second network element, the response message including an identifier of the candidate member; The screening information includes: A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to a vertical federation or a horizontal federation.

2. The request message includes: The method for determining candidate members as described in claim 1, further comprising instruction information used to instruct the execution of a task for determining candidate members to participate in federated learning, or to indicate that the current task is for federated learning or for selecting members for federated learning.

3. The first device includes: A first device that is in a network-connected state during the time period; A first device that supports performing federated learning during the time period; a first device that supports the algorithm type; a first device that supports federated learning using said algorithm type; A first device whose model training accuracy information is greater than the accuracy threshold; a first device in the wireless access system; A first device supporting federated learning in the wireless access scheme; a first device having a signal quality greater than a signal quality requirement; a first device whose consumed traffic is in said traffic range; a first device that satisfies the type requirements of the candidate member; a first device located within the area of ​​interest AOI; The method for determining candidate members according to claim 1 or 2, further comprising at least one of: a first device having a type of federated learning that is the same as the type of federated learning included in the screening information;

4. The first network element determines one or more first devices as candidate members capable of participating in federated learning based on the screening information, The first network element acquires, from the third network element, first instruction information for indicating willingness information of each second device to participate in federated learning, the willingness information indicating whether each second device wishes to participate in federated learning; The method for determining candidate members as described in claim 3, comprising determining one or more first devices as the candidate members based on the screening information and the first instruction information, the first devices being devices among the second devices that wish to participate in federated learning and that satisfy the screening information.

5. The first network element determines one or more first devices as candidate members capable of participating in federated learning based on the screening information, The first network element acquires, from a third network element, information indicating that the first device intends to participate in the federated learning; A method for determining a candidate member as described in any one of claims 1 to 3, comprising: the first network element determining, based on the intention information and the screening information, a first device that wishes to participate in the federated learning and matches the screening information as a candidate member that can participate in the federated learning.

6. The first network element determines one or more first devices as candidate members capable of participating in federated learning based on the screening information, The first network element acquires capability information of federated learning of the first device, the capability information including at least one of a radio access method participating in federated learning, an area participating in federated learning, a time participating in federated learning, algorithm information that can support participation in federated learning, accuracy information that can be achieved by participation in federated learning, and a type participating in federated learning; A method for determining candidate members described in any one of claims 1 to 5, comprising: the first network element determining candidate members capable of participating in federated learning from the first device based on the capability information and the screening information, wherein the capability information of the candidate members matches the screening information.

7. The ability information of the candidate member matches the screening information, The wireless access method in which the candidate member participates in federated learning is the same as the wireless access method included in the screening information; The area in which the candidate member will participate in the federated learning is located within the AOI included in the screening information; The time when the candidate member participates in the federated learning is within the time period included in the screening information; The screening information includes an algorithm type that can support the candidate member's participation in federated learning; and The accuracy information achievable for the candidate member to participate in the federated learning is higher than the accuracy threshold included in the screening information; The method for determining a candidate member according to claim 6 , comprising at least one of: a type of federated learning in which the candidate member participates is the same as a type of federated learning included in the screening information.

8. The first network element determines one or more first devices as candidate members capable of participating in federated learning based on the screening information, The first network element obtains network status information corresponding to the first device, the network status information comprising: type information of the first device; Location information of the first device; Wireless access method information in which the first device is located; and wireless signal quality information of the first device. The method for determining candidate members according to any one of claims 1 to 7, comprising: the first network element determining candidate members capable of participating in federated learning from the first device based on network status information corresponding to the first device and the screening information, wherein the network status information corresponding to the candidate members matches the screening information.

9. Matching the network status information corresponding to the candidate member with the screening information; The wireless access system in which the candidate member is located is the same as the wireless access system included in the screening information; the wireless signal quality information of the candidate member is greater than the signal quality requirement included in the screening information; the location of the candidate member is located within an AOI included in the screening information; The method for determining a candidate member according to claim 8 , further comprising at least one of: the type of the candidate member being included in member type information included in the screening information.

10. determining one or more first devices as the candidate members based on the screening information and the first indication information; The first network element acquires capability information of each second device participating in federated learning, the capability information including at least one of a radio access method participating in federated learning, an area participating in federated learning, a time participating in federated learning, algorithm information that can support participation in federated learning, accuracy information that can be achieved by participation in federated learning, and a type participating in federated learning; The method for determining candidate members as described in claim 4, comprising determining one or more first devices as candidate members based on the screening information, the first instruction information and the capability information, wherein the first devices are among the second devices that wish to participate in federated learning and whose capability information for federated learning satisfies the screening information.

11. The request message further includes second indication information, the second indication information being used to indicate a service type corresponding to the federated learning; The first network element determines one or more first devices as candidate members capable of participating in federated learning based on the screening information, The first network element determines one or more first devices as the candidate members based on the screening information and the second indication information; The method for determining candidate members according to claim 1 , wherein the first device is a device capable of supporting a service corresponding to a service type.

12. The request message further includes sorting instruction information, the sorting instruction information is used to instruct sorting the candidate members according to second information, the second information including at least one of the signal quality of the candidate members, the accuracy information of the candidate members, and the traffic information of the candidate members; The method for determining candidate members according to claim 1 , wherein the order of identifiers of the candidate members in the response message is an order obtained after sorting according to the second information.

13. The request message further includes packet indication information, the packet indication information being used to indicate that the candidate members are to be packetized according to third information, and the third information includes: The candidate member includes at least one of the following: an area where the candidate member is located; a time period when the candidate member is in a network connection state; an algorithm type supported by the candidate member; accuracy information of the candidate member; a radio access scheme of the candidate member; and a signal quality of the candidate member; The method for determining a candidate member according to claim 1 , wherein the identifier of the candidate member in the response message is an identifier obtained after packetizing according to the third information.

14. The method for determining a candidate member according to any one of claims 1 to 10, wherein the response message further includes at least one of the following: an area in which the candidate member is located, a time period during which the candidate member is in a network connection state, an algorithm type supported by the candidate member, accuracy information of the candidate member, a radio access method of the candidate member, signal quality of the candidate member, traffic information of the candidate member, and a time period during which the candidate member can perform federated learning.

15. A method for determining candidate members, comprising the steps of: A second network element sends a request message to a first network element, the request message including screening information, and the request message is used to instruct the first network element to determine one or more first devices as candidate members that can participate in federated learning based on the screening information; receiving, by the second network element, a response message sent by the first network element, the response message including an identifier of the candidate member; The screening information includes: A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federated learning for indicating whether the federated learning belongs to a vertical federation or a horizontal federation.

16. The request message includes: The method for determining candidate members as described in claim 15, further comprising instruction information used to instruct the execution of a task for determining candidate members to participate in federated learning, or to indicate that the current task is for federated learning or for selecting members for federated learning.

17. The first device includes: A first device that is in a network-connected state during the time period; A first device that supports performing federated learning during the time period; a first device that supports the algorithm type; a first device that supports federated learning using said algorithm type; A first device whose model training accuracy information is greater than the accuracy threshold; a first device in the wireless access system; A first device supporting federated learning in the wireless access scheme; a first device having a signal quality greater than a signal quality requirement; a first device whose consumed traffic is in said traffic range; a first device that satisfies the type requirements of the candidate member; a first device located within the area of ​​interest AOI; The method for determining a candidate member according to claim 15 or 16, further comprising at least one of: a first device having a type of federated learning that is the same as the type of federated learning included in the screening information;

18. The request message further includes sorting instruction information, the sorting instruction information is used to instruct sorting the candidate members according to second information, the second information including at least one of the signal quality of the candidate members, the accuracy information of the candidate members, and the traffic information of the candidate members; The method for determining a candidate member according to any one of claims 15 to 17, wherein the identifier of the candidate member in the response message is an identifier obtained after sorting according to the second information.

19. The request message further includes packet indication information, which is used to instruct the candidate member to packetize according to third information, and the third information includes at least one of the following: an area where the candidate member is located, a time period when the candidate member is in a network connection state, an algorithm type supported by the candidate member, accuracy information of the candidate member, a radio access scheme of the candidate member, and a signal quality of the candidate member; The method for determining a candidate member according to any one of claims 15 to 17, wherein the identifier of the candidate member in the response message is an identifier obtained after packetizing according to the third information.

20. The method for determining a candidate member according to any one of claims 15 to 17, wherein the response message further includes the area in which the candidate member is located, the time period during which the candidate member is in a network connection state, the algorithm type supported by the candidate member, accuracy information of the candidate member, the radio access method of the candidate member, the signal quality of the candidate member, traffic information of the candidate member, and the time period during which the candidate member is able to perform federated learning.

21. The method for determining candidate members includes the steps of: The method for determining a candidate member according to claim 17 or 18, further comprising: the second network element determining a target member to participate in federated learning based on an identifier of the candidate member.

22. A candidate member determination device, a receiving module for receiving a request message sent by a second network element, the request message including screening information; a processing module for determining one or more first devices as candidate members capable of participating in federated learning based on the screening information; a sending module for sending a response message to the second network element, the response message including an identifier of the candidate member; The screening information includes: A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federation learning for indicating whether the federation learning belongs to a vertical federation or a horizontal federation.

23. A candidate member determination device, A sending module for sending a request message to a first network element, the request message including screening information, the request message being used to instruct the first network element to determine one or more first devices as candidate members capable of participating in federated learning based on the screening information; a receiving module for receiving a response message transmitted by the first network element, the response message including an identifier of the candidate member; The screening information includes: A time period for indicating a selective time period during which the associative learning is performed; an algorithm type for indicating a model training algorithm type that needs to be supported to perform the federated learning; an accuracy threshold for indicating a model training accuracy requirement that must be met to perform the federated learning; a radio access scheme for indicating a radio access scheme that needs to be selected to perform the federated learning; A signal quality requirement for indicating a wireless signal quality requirement when performing the joint learning; a traffic range for indicating a traffic usage range requirement of the potential member; Member type information for indicating type requirements of candidate members participating in federated learning; numerical information for indicating a numerical request for said candidate members; an area of ​​interest (AOI) for indicating an area in which the candidate members are located; and type information of the federation learning for indicating whether the federation learning belongs to a vertical federation or a horizontal federation.

24. A first network element comprising a processor and a memory, the memory storing a program or instructions operable on the processor, the program or instructions, when executed by the processor, implementing steps of a method for determining a candidate member as described in any one of claims 1 to 14.

25. A second network element comprising a processor and a memory, the memory storing a program or instructions operable on the processor, the program or instructions, when executed by the processor, implementing the steps of the method for determining a candidate member described in any one of claims 15 to 21.

26. A readable storage medium having a program or instructions stored thereon, the program or instructions, when executed by a processor, implementing a method for determining a candidate member as described in any one of claims 1 to 14, or implementing steps of a method for determining a candidate member as described in any one of claims 15 to 21.

Images