Model information acquisition method, transmission method, apparatus, node, and storage medium

Abstract

The application discloses a model information acquisition method, a sending method, a device, a node and a storage medium, and belongs to the technical field of communication. The model information acquisition method provided by the application comprises the following steps: a model training function node determines a federated learning (FL) server node; the model training function node sends a first request message to the FL server node, and the first request message is used for triggering the FL server node to perform federated learning to acquire a target model; and the model training function node receives information of the target model sent by the FL server node.

Description

Technical Field

[0001] This application belongs to the field of communication technology, specifically relating to a model information acquisition method, transmission method, device, node, and storage medium. Background Technology

[0002] Currently, some communication systems (such as 5G) have introduced artificial intelligence (AI) functions, which can be used to process network services through models. However, at present, only network nodes can learn the model independently. This may lead to network nodes being unable to train the model in some scenarios (such as when there is insufficient training data), resulting in poor model training performance of network nodes. Summary of the Invention

[0003] This application provides a model information acquisition method, transmission method, apparatus, node, and storage medium to address the problem of poor model training performance of network nodes.

[0004] Firstly, a method for acquiring model information is provided, characterized by comprising:

[0005] The model training function nodes determine the Federated Learning (FL) server nodes;

[0006] The model training function node sends a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain the target model;

[0007] The model training node receives the target model information sent by the FL server node.

[0008] Secondly, a method for sending model information is provided, including:

[0009] The federated learning FL server node receives a first request message sent by the model training function node. The first request message is used to trigger the FL server node to perform federated learning to obtain the target model.

[0010] The FL server node performs federated learning with the FL client node based on the first request message to obtain the target model;

[0011] The FL server node sends the target model information to the model training function node.

[0012] Thirdly, a model information acquisition device is provided, comprising:

[0013] The first determination module is used to determine the federated learning FL server node;

[0014] The first sending module is used to send a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain the target model;

[0015] The receiving module is used to receive information about the target model sent by the FL server node.

[0016] Fourthly, a model information transmission device is provided, comprising:

[0017] The receiving module is used to receive a first request message sent by the model training function node. The first request message is used to trigger the federated learning FL server node to perform federated learning to obtain the target model.

[0018] The learning module is used to perform federated learning with the FL client node based on the first request message to obtain the target model;

[0019] The sending module is used to send the information of the target model to the model training function node.

[0020] Fifthly, a model training function node is provided, including a processor and a memory, wherein the memory stores programs or instructions that can run on the processor, and when the programs or instructions are executed by the processor, they implement the steps of the model information acquisition method provided in the embodiments of this application.

[0021] In a sixth aspect, a model training function node is provided, including a processor and a communication interface, wherein the processor or the communication interface is used to determine the FL server node; the communication interface is used to send a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain a target model; and to receive information about the target model sent by the FL server node.

[0022] In a seventh aspect, a server node is provided, characterized in that it includes a processor and a memory, wherein the memory stores a program or instructions that can run on the processor, and the program or instructions, when executed by the processor, implement the steps of the model information sending method provided in the embodiments of this application.

[0023] Eighthly, a server node is provided, including a processor and a communication interface, wherein the communication interface is used to receive a first request message sent by a model training function node, the first request message being used to trigger a federated learning (FL) server node to perform federated learning to obtain a target model; the processor or communication interface is used to perform federated learning with an FL client node based on the first request message to obtain the target model; the communication interface is used to send information about the target model to the model training function node.

[0024] In a ninth aspect, a model information transmission system is provided, comprising: a model training function node and a server node, wherein the model training function node can be used to execute the steps of the model information acquisition method provided in the embodiments of this application, and the server node can be used to execute the steps of the model information transmission method provided in the embodiments of this application.

[0025] In a tenth aspect, a readable storage medium is provided, on which a program or instructions are stored, wherein when the program or instructions are executed by a processor, the program or instructions implement the steps of the model information acquisition method provided in the embodiments of this application, or implement the steps of the model information transmission method provided in the embodiments of this application.

[0026] Eleventhly, a chip is provided, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the model information acquisition method provided in the embodiments of this application, or to implement the model information transmission method provided in the embodiments of this application.

[0027] In a twelfth aspect, a computer program / program product is provided, which is stored in a storage medium, and the computer program / program product is executed by at least one processor using the steps of the model information acquisition method provided in the embodiments of this application, or the computer program / program product is executed by at least one processor using the model information transmission method provided in the embodiments of this application.

[0028] In this embodiment, the model training function node determines the FL server node; the model training function node sends a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain the target model; the model training function node receives the target model information sent by the FL server node. In this way, the model training function node can obtain the target model information for federated learning, thereby improving the model training performance of the network nodes. Attached Figure Description

[0029] Figure 1 This is a block diagram of a wireless communication system applicable to embodiments of this application;

[0030] Figure 2 This is a flowchart of a model information acquisition method provided in an embodiment of this application;

[0031] Figure 3 This is a flowchart of a model information sending method provided in an embodiment of this application;

[0032] Figure 4 This is a schematic diagram of a model information acquisition method provided in an embodiment of this application;

[0033] Figure 5 This is a schematic diagram of another model information acquisition method provided in an embodiment of this application;

[0034] Figure 6 This is a structural diagram of a model information acquisition device provided in an embodiment of this application;

[0035] Figure 7 This is a structural diagram of a model information sending device provided in an embodiment of this application;

[0036] Figure 8 This is a structural diagram of a communication device provided in an embodiment of this application;

[0037] Figure 9 This is a structural diagram of a network node provided in an embodiment of this application. Detailed Implementation

[0038] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0039] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0040] It is worth noting that the technologies described in this application are not limited to Long Term Evolution (LTE) / LTE-Advanced (LTE-A) systems, but can also be used in 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 this application are often used interchangeably, and the described technologies can be used in the systems and radio technologies mentioned above, as well as in other systems and radio technologies. The following description describes New Radio (NR) systems for illustrative purposes, and NR terminology is used in most of the following description; however, these technologies can also be applied to applications beyond NR systems, such as 6th Generation (6G) communication systems.

[0041] Figure 1This diagram illustrates a block diagram of a wireless communication system applicable to embodiments of this application. The wireless communication system includes a terminal 11 and a network-side device 12. Terminal 11 can be a mobile phone, tablet computer, laptop computer, personal digital assistant (PDA), handheld computer, netbook, ultra-mobile personal computer (UMPC), mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, vehicle-mounted device (VUE), pedestrian terminal (PUE), smart home (home devices with wireless communication capabilities, such as refrigerators, televisions, washing machines, or furniture), game console, personal computer (PC), ATM or self-service machine, etc. Wearable devices include: smartwatches, smart bracelets, smart headphones, smart glasses, smart jewelry (smart bracelets, smart chains, smart rings, smart necklaces, smart anklets, smart anklets, smart helmets, smart joysticks, etc.), smart wristbands, smart clothing, etc. It should be noted that the specific type of terminal 11 is not limited in the embodiments of this application.

[0042] Network-side equipment 12 may include access network equipment and core network equipment. Access network equipment may also be referred to as radio access network equipment, radio access network (RAN), radio access network function, or radio access network unit. Access network equipment may include base stations, WLAN access points, or WiFi nodes, etc. Base stations may be referred to as Node B, evolved Node B (eNB), access point, base transceiver station (BTS), radio base station, radio transceiver, Basic Service Set (BSS), Extended Service Set (ESS), home B node, home evolved B node, Transmitting Receiving Point (TRP), or any other suitable term in the field, as long as the same technical effect is achieved. The base station is not limited to specific technical terms. It should be noted that this application embodiment only uses a base station in an NR system as an example for description and does not limit the specific type of base station.

[0043] Core network equipment may include, but is not limited to, at least one of the following: core network node, core network function, Mobility Management Entity (MME), Access and Mobility Management Function (AMF), Session Management Function (SMF), User Plane Function (UPF), Policy Control Function (PCF), Policy and Charging Rules Function (PCRF), Edge Application Server Discovery Function (EASDF), Unified Data Management (UDM), Unified Data Repository (UDR), Home Subscriber Server (HSS), Centralized network configuration (CNC), Network Repository Function (NRF), Network Exposure Function (NEF), Local NEF (or L-NEF), and Binding Support. Core network functions (BSF), application functions (AF), and network data analytics functions (NWDAF), etc. It should be noted that this application embodiment only uses core network equipment in the NR system as an example for description, and does not limit the specific type of core network equipment.

[0044] In this embodiment, the model training function node can be a network node used to generate and train a model. The model training function node can refer to a regular model training function node (not an FL server node) or an FL client node. That is, the model training function node not being an FL server node can be understood as the model training function node not being able to serve as an FL server node for a specific model training task (corresponding to a certain analysis identifier, a certain AOI), but it is not limited to the fact that it can serve as an FL server node for other model training tasks (such as other analysis identifiers, other AOIs).

[0045] Model inference function nodes can be network nodes used to perform inference to generate predictive information, generate statistical information, or perform data analysis.

[0046] In some embodiments, a model training node can refer to a network element, terminal, or module in a communication network that has AI model training capabilities; a model inference node refers to a network element, terminal, or module in a communication network that has model inference capabilities. That is to say, of course, model training nodes and model inference nodes can also be called by other names.

[0047] In this embodiment, the model training and model inference nodes can be core network elements or internal modules of core network elements. For example, NWDAF may include model training and model inference nodes, the FL server node is a core network element with federated learning server capabilities, and the FL client node can be other core network elements or modules participating in federated learning. Alternatively, the model training and model inference nodes can be radio access network elements or internal modules of radio access network elements, that is, the model training and model inference nodes can be functions within the RAN. For example, the model training node can be a RAN device with model training capabilities. Specifically, the model training node can be a base station device or module with model training capabilities but not an FL server node, the FL server node can be a base station device or module with federated learning server capabilities or Operation Administration and Maintenance (OAM), and the FL client node can be other member base station devices or modules participating in federated learning.

[0048] In addition, model training function nodes and model inference function nodes can also be terminals or functions within terminals. For example, a model training function node can be a terminal with model training capabilities. Specifically, a model training function node can be a terminal with model training capabilities but not a FL server node. A FL server node can be a terminal or AF with federated learning server capabilities. Other FL client nodes can be other member terminals participating in federated learning.

[0049] It should be noted that, in the embodiments of this application, the model training function node and the model inference function node can be deployed independently as different network element devices, or co-deployed in the same network element device, such as two internal functional modules of the core network element, the wireless access network element, or the terminal. In this case, the core network element, the wireless access network element, or the terminal can provide both AI model training function and model inference function.

[0050] In this embodiment, the FL server node is a network element that coordinates or hosts federated learning. For example, the FL server node can be the FL central network element or the FL coordinator network element, or a central model training function node that can be used for FL operation. Specifically, the FL server node can be a core network element, a wireless access network element, a terminal, or an application server. The FL client node is a network element that participates in federated learning, which can be referred to as a network element that participates in FL operation. Specifically, the FL client node can be a core network element, a wireless access network element, a terminal, or an application server.

[0051] In some embodiments, the model training function node can be the Model Training Logical Function (MTLF), the model inference function node can be the Analytics Logical Function (AnLF), the FL server node can be the FL server, and the FL client node can be the FL client.

[0052] Please see Figure 2 , Figure 2 This is a flowchart of a model information acquisition method provided in an embodiment of this application, such as... Figure 2 As shown, the steps include:

[0053] Step 201: Determine the FL server node for the model training function node.

[0054] The determination of the FL server node can be achieved by the model training function node querying other network nodes for the FL server node, or by the model training function node selecting the FL server node based on pre-acquired configuration information.

[0055] Step 202: The model training function node sends a first request message to the FL server node. The first request message is used to trigger the FL server node to perform federated learning to obtain the target model.

[0056] The first request message can request the FL server node to perform federated learning to train the target model. Specifically, this can involve FL operations between the FL server and at least one FL client node, such as an iterative process of federated learning between the FL server and at least one FL client node, to obtain the target model. Since the target model is a model obtained through federated learning, it can also be called a federated learning model.

[0057] Step 203: The model training function node receives the target model information sent by the FL server node.

[0058] This step can be performed after the FL server node obtains the target model through federated learning, and then the FL server node sends the target model information to the model training function node.

[0059] The information about the target model mentioned above can be information used to determine the target model, such as the model file information of the target model, or the download address information of the model file, etc.

[0060] It should be noted that, in the embodiments of this application, the target model mentioned above can be a model used for communication service processing in a communication system, for example: it can be a model used for data analysis, or a model used for inference tasks, or a model used for channel estimation, or a model used for information prediction, etc.

[0061] In this embodiment, the above steps enable the model training node to obtain information about the target model for federated learning, thereby improving the model training performance of network nodes and resolving the issue of being unable to obtain the target model due to data privacy. For example, if the model training node needs to provide the target model for the AOI specified by the model inference node, but the model training node discovers that it may be unable to obtain all or part of the training data in the AOI due to data silos, it triggers federated learning to the FL server node.

[0062] As an optional implementation, the model training function node determines the FL server node, including:

[0063] The model training function node sends a node discovery request message to the network repository function element. The node discovery request message is used to request network nodes that participate in federated learning training.

[0064] The model training function node receives a response message sent by the network repository function element, and the response message includes information about the FL server node.

[0065] The node discovery request message mentioned above can be a network element discovery request (Nnrf_NFDiscovery_Request).

[0066] The aforementioned network warehouse function element can store information about one or more FL server nodes, so that after receiving the aforementioned request message, it returns the corresponding FL server node information to the aforementioned model training function node.

[0067] In this implementation, information about the FL server node can be obtained through the network warehouse function element.

[0068] It should be noted that in some implementations, the information of the FL server node may not need to be obtained from the network repository function element. For example, the information of the FL server node may be fixed in the model training function node.

[0069] Optionally, the node discovery request message mentioned above includes at least one of the following:

[0070] Analysis includes analytics ID, area of ​​interest (AOI) information, time of interest information, model description information, model shareability information, model performance information, model algorithm information, model training speed information, federated learning instruction information, federated learning type information, FL server node type instruction information, FL client node type instruction information, first service information, and second service information.

[0071] The aforementioned analysis identifier can be used to indicate that the network node requested by the request message needs to support the model training task corresponding to the analysis identifier.

[0072] The aforementioned AOI information can be used to indicate that the network node requested by the request message can serve the area corresponding to the AOI information, which may be at least one Tracking Area (TA), at least one cell, or other areas.

[0073] The aforementioned time of interest information can be used to indicate that the network node requested by the request message must support model training for the time period corresponding to the time of interest information.

[0074] The aforementioned model description information can be used to indicate that the network node requested by the request message must support a model representation based on the model description information corresponding to this information. This model description information can also be called model description requirement information or model description expectation information. For example, the model description information can be a model expression language represented by Open Neural Network Exchange (ONNX) or a model framework represented by TensorFlow or PyTorch.

[0075] The aforementioned model-shareable information can be used to indicate that the network node requested by the request message must be able to share the model with the model training function node. This model-shareable information can also be called model-shareable requirement information or model-shareable expectation information. Here, "shareable" means interoperable, or "shareable" means mutually understandable, and "shareable" means operational.

[0076] The aforementioned model performance information can be used to indicate that the network node requested by the request message must provide a model that meets the requirements of this model performance information. This model performance information can also be called model performance requirement information or model performance expectation information. Performance can refer to the model's accuracy value, mean absolute error (MAE), etc.

[0077] The aforementioned model algorithm information can be used to indicate that the network node requested by the request message must support training a model based on the model algorithm corresponding to the model algorithm information. This model algorithm information can also be called model algorithm requirement information or model algorithm expectation information.

[0078] The aforementioned model training speed information can be used to indicate that the network node requested by the request message must train the model at a speed that meets the model training speed requirements. This model training speed information can also be called model training speed requirement information or model training speed expectation information. The training speed can be expressed as the time taken for the model to reach convergence or a certain performance threshold.

[0079] The aforementioned federated learning instruction information can be used to indicate that the network node requested by the request message needs to support federated learning.

[0080] The above-mentioned federated learning type information can be used to ensure that the network node requested by the request message supports at least one of the following types of federated learning:

[0081] Horizontal federated learning type; Vertical federated learning type.

[0082] Specifically, the aforementioned horizontal federated learning type can be training using different training data samples with the same feature points; the aforementioned vertical federated learning type can be training using training data samples with different feature points of the same training samples.

[0083] The above FL server node type indication information can be used to indicate that the network node requested by the request message belongs to the FL server node type.

[0084] The above FL client node type indication information can be used to indicate that the network node requested by the request message belongs to the FL client node type.

[0085] The aforementioned first service information can be used to indicate that the network node requested by the request message needs to support the services of the federated learning server.

[0086] The aforementioned second service information can be used to indicate that the network node requested by the request message needs to support the services of federated learning members.

[0087] In this implementation, since the node discovery request message includes at least one of the above-mentioned information, the network repository function element can return information of the FL server node that meets the corresponding conditions to the model training function node, thereby enabling the FL server node to finally federate the target model required, satisfied, or expected by the above-mentioned model training function node.

[0088] Optionally, the response message includes information about N network nodes, including the FL server node, where N is a positive integer;

[0089] Each network node's information includes at least one of the following:

[0090] Fully Qualified Domain Name (FQDN), identification information, and address information.

[0091] In this implementation, the FL server node is determined by at least one of the FQDN, identification information, and address information.

[0092] In some implementations, the aforementioned N network nodes also include FL client nodes.

[0093] The N network nodes mentioned above include one or more FL client nodes that are participating in this federated learning process. These FL client nodes can be FL participants or FL members.

[0094] Since the response message also includes the FL client node, the model training function node can quickly identify the FL server node and the FL client node, and thus promptly request the FL server node and the FL client node to perform federated learning, thereby improving the efficiency of obtaining the target model.

[0095] Optionally, the information for each network node may also include:

[0096] Type information, which indicates the type of network node, which is either an FL server node or an FL client node.

[0097] In this embodiment, type information can be used to indicate whether a network node is an FL server node or an FL client node. Of course, in this embodiment, the type information may not be indicated. For example, the model training function node can identify itself as an FL server node or an FL client node through the identification information of these network nodes. Alternatively, if the response message includes FL server nodes and FL client nodes, the information of the FL server nodes and FL client nodes in the response message can be sorted, allowing the model training function node to identify the FL server nodes and FL client nodes through the sorting of information.

[0098] As an optional implementation, the first request message includes at least one of the following:

[0099] Federated learning instruction information and model identifiers;

[0100] The federated learning instruction information is used to request that the FL server node trigger federated learning to obtain the target model.

[0101] The model identifier is used to uniquely identify the target model.

[0102] The model identifier mentioned above can be obtained by the model training function node. For example, the method further includes: the model training function node obtains the model identifier, such as by generating it itself or obtaining it from the model identifier management network element.

[0103] In this implementation, since the first request message includes at least one of federated learning instruction information and model identifier, the FL server node can determine whether to perform federated learning and respond quickly to the request message. It should be noted that in some implementations, the federated learning instruction information and model identifier may not be included; for example, the model identifier can be obtained by the FL server node, and the FL server node will default to performing federated learning upon receiving the first request message.

[0104] In some implementations, the first request message may include at least one of the following:

[0105] Analysis includes identifiers, time of interest information, model description information, model shareability information, model performance information, model algorithm information, model training speed information, first model filtering information, first object information, and reporting information;

[0106] The analysis identifier, time of interest information, model description method information, model shareability information, model performance information, model algorithm information, and model training speed information mentioned above can be found in the corresponding descriptions of the above implementation methods, and will not be repeated here.

[0107] The first model filtering information mentioned above can be used to indicate at least one of the following: the regional scope, time scope, slice scope, and data network name (DNN) scope of federated learning.

[0108] The aforementioned first object information can be used to indicate that the object targeted by federated learning is a single terminal, multiple terminals, or any one of the terminals; for example, the first object information includes terminal information, such as terminal identifier (UE ID) or terminal group identifier (UE group ID).

[0109] The above-mentioned reporting information can be used to indicate: the reporting format and / or reporting conditions for sending the federated model.

[0110] In this embodiment, the first request message includes at least one of the above-mentioned pieces of information, which enables the FL server node to eventually federate the target model required, satisfied, or desired by the model training function node.

[0111] As an optional implementation, the first request message includes: information about the FL client nodes participating in federated learning.

[0112] The FL client nodes participating in federated learning can be one or more FL client nodes, and these clients may or may not include the model training function nodes mentioned above.

[0113] The information of the FL client nodes in the aforementioned federated learning can be obtained from the network repository function element or pre-configured by the model training function node. Since the first request message includes the information of the FL client nodes participating in federated learning, the FL server node can quickly perform federated learning with the FL client nodes, thereby improving the efficiency of acquiring the target model.

[0114] As an optional implementation, the information of the target model includes at least one of the following pieces of information corresponding to the target model:

[0115] Model identifier, federated instruction information, model file, and model file address information.

[0116] The above federated learning instruction information is used to indicate that the target model is a model obtained through federated learning;

[0117] The model identifier is used to uniquely identify the target model;

[0118] The aforementioned model file may include relevant file information of the target model, such as files containing the model's network structure, weight parameters, and input / output data.

[0119] The address information of the aforementioned model file is used to obtain the address information of the target model, such as indicating the storage address of the model file or the download address information of the model file.

[0120] It should be noted that, in this embodiment, the information of the target model may not include the model identifier, because the model training function node can obtain the identifier itself. The information of the target model may not include the federated instruction information, because the model training function node can assume that the target model sent by the FL server node is the model obtained by federated learning. The information of the target model may not include the model file, because the model training function node downloads the model file according to the address information. The information of the target model may not include the address information of the model file, because, in some implementations, after the model is trained, the model file can be stored in a pre-configured location, so that the model training function node can download the model file from that location.

[0121] In some implementations, the information of the target model may further include at least one of the following pieces of information corresponding to the target model:

[0122] Analysis includes identifiers, second model filtering information, and model validity information.

[0123] Among them, the analysis identifier can be used to indicate the task corresponding to the above target model;

[0124] The second model filtering information mentioned above can be used to indicate at least one of the following: the target model's regional range, time range, slice range, and data network name (DNN) range.

[0125] The above model validity information can be used to indicate: the model's validity time, validity area, and other valid information.

[0126] As an optional implementation, the method further includes:

[0127] The model training node sends the target model information to the model inference node.

[0128] In this implementation, the target model information can be sent proactively to the model inference function node, or it can be sent based on a request from the model inference function node.

[0129] For example, in the request sending method, the above-mentioned model training function node receives a model request message sent by the model inference function node, and the model request message may include at least one of the following:

[0130] Analysis includes identifiers, third-model filtering information, second-object information, and time information;

[0131] The fourth identifier can be used to indicate the data analysis task to which the model identifier requested by the model request message is targeted.

[0132] The aforementioned third model filtering information can be used to indicate the conditions that the model requested by the model request message needs to meet, such as the region of interest that the model needs to serve, Single Network Slice Selection Assistance Information (S-NSSAI), or DNN.

[0133] The aforementioned second object information can be used to indicate the training object of the model requested by the model request message. This training object can be for a single terminal, multiple terminals, or any terminal. For example, the second object information can be the terminal identifier or terminal group identifier of these terminals.

[0134] The time information is used to indicate at least one of the following: the applicable time of the model requested by the model request message and the reporting time of the model's information.

[0135] In this embodiment, since the model training function node sends the target model information to the model inference function node, the model inference function node may use the target model to improve the business performance of the model inference function node.

[0136] As an optional implementation, before the model training functional node determines the FL server node, the method further includes:

[0137] The model training function node determines that federated learning is needed to obtain the target model.

[0138] The determination that federated learning is needed to obtain the target model can be made when the model training function node cannot independently train the target model, thus requiring federated learning to obtain it. For example, the determination that federated learning is needed to obtain the target model by the model training function node includes:

[0139] If the model training function node determines that it is unable to obtain all or part of the training data used to generate the target model, the model training function node determines that it is necessary to obtain the target model through federated learning.

[0140] The inability to obtain all or part of the training data used to generate the target model may be due to reasons such as technical confidentiality or user privacy, which prevent the model training node from obtaining all or part of the training data used to generate the target model. For example, the model training node needs to provide a target model for the AOI specified by the model inference node, but if the model training node finds that it may not be able to obtain all or part of the training data in the AOI due to data silos, it will trigger federated learning from the FL server node. Another example is that the model training node needs to obtain models corresponding to certain terminals, but if the model training node cannot obtain the training data corresponding to these terminals due to data privacy issues, it can obtain the training data corresponding to these terminals by contacting the FL server node, or trigger federated learning to obtain the target model for the terminal.

[0141] In this implementation, when it is determined that the target model needs to be obtained through federated learning, the information of the target model can be obtained through the aforementioned FL server node, thereby improving the model training performance of the network node. For example, it can be implemented to obtain the target model when the model training function node does not have enough training data for the target model, or when the model training function node does not have enough training resources for the target model, thereby improving the model training performance of the network node.

[0142] In this embodiment, the model training function node determines the FL server node; the model training function node sends a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain the target model; the model training function node receives the target model information sent by the FL server node. In this way, the model training function node can obtain the target model information for federated learning, thereby improving the model training performance of the network nodes.

[0143] Please see Figure 3 , Figure 3 This is a flowchart of a model information sending method provided in the application embodiment, such as... Figure 3 As shown, it includes the following steps:

[0144] Step 301: The FL server node receives a first request message sent by the model training function node. The first request message is used to trigger the FL server node to perform federated learning to obtain the target model.

[0145] Step 302: The FL server node performs federated learning with the FL client node based on the first request message to obtain the target model.

[0146] Specifically, the federated learning between the FL server node and the FL client node based on the first request message can be triggered by the first request message to perform FL operations between the FL server node and the FL client node. In particular, the FL server node and the FL client node can perform an interactive iterative process of federated learning to train the target model.

[0147] Step 303: The FL server node sends the target model information to the model training function node.

[0148] Optionally, the first request message includes at least one of the following:

[0149] Federated learning instruction information and model identifiers;

[0150] The federated learning instruction information is used to request that the FL server node trigger federated learning to obtain the target model.

[0151] The model identifier is used to uniquely identify the target model.

[0152] In some implementations, the first request message may include at least one of the following:

[0153] Analysis includes identifiers, time of interest information, model description information, model shareability information, model performance information, model algorithm information, model training speed information, first model filtering information, first object information, and reporting information.

[0154] Optionally, the first request message includes: information about the FL client nodes participating in federated learning.

[0155] Optionally, the method further includes:

[0156] The FL server node determines the FL client nodes that participate in federated learning.

[0157] Specifically, the FL server node can determine the FL client nodes participating in federated learning by querying the network repository function element, or by determining the FL client nodes participating in federated learning based on pre-configured information.

[0158] Optionally, the FL server node determines the FL client nodes participating in federated learning, including:

[0159] The FL server node sends a node discovery request message to the network repository function element. The node discovery request message is used to request FL client nodes participating in federated learning.

[0160] The FL server node receives a response message sent by the network repository function element, and the response message includes information about the FL client nodes participating in federated learning.

[0161] Optionally, the information of the federated model includes at least one of the following pieces of information corresponding to the target model:

[0162] Model identifier, federation instruction information, model file, and model file address information;

[0163] The federated learning instruction information is used to indicate that the target model is a model obtained through federated learning.

[0164] The model identifier is used to uniquely identify the target model.

[0165] In some implementations, the information of the target model may further include at least one of the following pieces of information corresponding to the target model:

[0166] Analysis includes identifiers, second model filtering information, and model validity information.

[0167] Optionally, the model identifier is obtained by the model training function node for the target model.

[0168] Optionally, the method further includes:

[0169] The FL server node obtains the model identifier for the target model.

[0170] The model identifier obtained by the FL server node for the target model can be generated by the FL server node or obtained from the model identifier management network element.

[0171] It should be noted that this embodiment is used as a reference for... Figure 2 The implementation method of the corresponding FL server node in the illustrated embodiment can be found in the following documentation for details. Figure 2 To avoid repetition, the relevant descriptions of the embodiments shown will not be repeated in this embodiment.

[0172] The following examples, using MTLF as the model training node, AnLF as the model inference node, FL server as the FL server node, and FL client as the FL client node, illustrate the method provided in this application through multiple embodiments:

[0173] Example 1:

[0174] This embodiment uses the example of an FL server identifying an FL client to illustrate the concept. Specifically, as follows... Figure 4 As shown, it includes the following steps:

[0175] Step 0. Optionally, the Consumer Network Function (Consumer NF) requests data analysis results from AnLF, and the request message may carry at least one of the following:

[0176] The Analytics ID is used to identify the task type of a data analysis task. For example, analytics ID = UE mobility, which indicates a task requesting user movement trajectory data analysis.

[0177] Analytic filter information is used to indicate filtering information for data analysis results, such as including regions of interest (AOI), slice S-NSSAI, DNN, etc.

[0178] The target of analytic reporting is used to specify whether the data analysis is performed on a single terminal, multiple terminals, or all UEs.

[0179] Analytics reporting information indicates the reporting time and conditions for data analysis results.

[0180] Step 1: AnLF sends a model request message to MTLF1 to request the model corresponding to the analysis identifier.

[0181] Model request messages can be either model provision subscription (Nnwdaf_MLModelProvision_subscribe) messages or model information request messages (Nnwdaf_MLModelInfo_Request).

[0182] The model request message can carry the following information:

[0183] Analytic ID is used to identify a type of inference task, such as analytics ID = UE mobility, used to predict user movement trajectories.

[0184] Model filtering information indicates the conditions that the requested model needs to meet, such as region of interest (AOI), slice S-NSSAI, and DNN.

[0185] Model object information, used to indicate the training objects of the model, such as for a single terminal, multiple terminals, or any terminal.

[0186] Model time information is used to indicate the applicable time of the model or the reporting time of model information.

[0187] It should be noted that MTLF1 here is not the FL server. Specifically, MTLF1 is not the FL server for this target model, but it can be the FL server for other models. The FL server can be called a coordinator or a central MTLF for FL operation, but it is only an FL client (or a member participating in FL operation) or an MTLF that does not support FL capabilities.

[0188] Step 2: MTLF1 determines, based on the model request message, that federated learning is required to obtain the requested model.

[0189] Factors that MTLF1 considers to determine the need for federated learning may include:

[0190] There is insufficient training data in MTLF1 for this model training task. For example, due to data privacy issues, there is a lack of training data in other regions or other training entities, making it impossible to perform centralized model training to generate the requested model.

[0191] Optionally, MTLF1 can further determine that the type of federated learning is horizontal federated learning. For example, MTLF1 determines that horizontal federated learning is required based on the characteristic that the training task corresponds to different samples but has the same features.

[0192] Step 3a: Optionally, MTLF1 sends a node discovery request message to NRF to request network element devices that can perform federated learning training.

[0193] Specifically, the node discovery request message can use the discovery request (Nnrf_NFDiscovery_Request). This message may include:

[0194] The analytics ID is used to indicate that the requested network element needs to support the analytics ID.

[0195] An AOI can be at least one Tracking Area (TA), at least one cell, or other representation used to indicate that the requested network element must be able to serve the AOI.

[0196] The node discovery request message may also include at least one of the following:

[0197] Federated learning instruction information is used to instruct the requested network element to support federated learning. Furthermore, it can also instruct that horizontal federated learning must be supported.

[0198] The expected NF type is FL server, which indicates that the requested network element must be of type FL server.

[0199] The expected NF type is FL client, which indicates that the requested network element must be of type FL client.

[0200] The expected NF service name is the FL server (coordination) service, which indicates that the requested network element must support the Federated Learning Server (coordination) service.

[0201] The expected NF service name is FL client(particepant)service, which is used to indicate that the requested network element must support the services of federated learning members (participants).

[0202] In one implementation, the network element request message includes federated learning instruction information. The request message can specify an analysis identifier and expect the NF type to be MTLF (Expected NF type = MTLF) and AOI. In this case, MTLF1 is used for NRF requests to support all MTLFs for federated learning against AOI.

[0203] In one implementation, the network element request message includes the expected NF type as FL server (Expected NF type = FL server), and other information is similar to existing technologies (such as specifying analytics ID, Expected NF type = MTLF, AOI). In this case, it indicates that MTLF1 is used for NRF requests to support FL servers for federated learning.

[0204] In one implementation, the network element request message includes Expected NF type = FL server and FL client, and other information is similar to existing technologies (such as specifying analytics ID, Expected NF type = MTLF, AOI). In this case, it indicates that MTLF1 is used for NRF requests to support the FL server and FL client for federated learning.

[0205] Step 3b: NRF returns the device information that matches the request message initiated by the network element in step 3a to MTLF1.

[0206] The NRF sends the FQDN, identification information, address information, etc. of one or more MTLFs that meet the requirements to MTLF1. The feedback information can specify whether each MTLF is an FL server or an FL client.

[0207] Step 4: Optionally, MTLF1 assigns a model ID to the resulting federated learning model to uniquely identify the model.

[0208] Step 5: MTLF1 sends a Federated Learning Request message to the FL server, requesting the FL server to trigger the federated learning process. The Federated Learning Request message may include at least one of the following pieces of information:

[0209] A federated learning (FL) indication is used to instruct a request to proceed with the federated learning process.

[0210] An Analytics ID is used to indicate the type of task for which a federated learning process is requested.

[0211] Model ID is used to uniquely identify the model generated by federated learning.

[0212] Model filter information is used to define the scope of the federated learning process, such as the geographical scope, time scope, S-NSSAI, DNN, etc.

[0213] The model target is used to specify the objects that the federated learning process targets, such as one or more specific terminals, or all terminals.

[0214] Model reporting information indicates the reporting information of the generated federated learning model, such as reporting time (start time, end time, etc.) and reporting conditions (periodic triggering, event triggering, etc.).

[0215] Step 6: Optionally, if a model identifier is not received from MTLF1, the FL server may assign a model identifier to the upcoming federated learning model to uniquely identify the model.

[0216] Step 7a: The FL server determines at least one FL client to perform this federated learning process. Specifically, the FL server can query the network repository function elements to obtain FL clients that match the requirements of this federated learning process. See step 3a for reference.

[0217] One point to note is that the FL client here may or may not include MTLF1 itself.

[0218] Step 7: The FL server and FL client engage in an iterative interaction process of federated learning to obtain the federated learning model.

[0219] If the FL client does not include MTLF1 itself, then the interaction process here does not involve MTLF1.

[0220] Step 8: The FL server sends the target model information of the generated federated learning model to MTLF1. The target model information includes at least one of the following:

[0221] Model identifier;

[0222] Federal Learning Directive;

[0223] Model file (containing the model's network structure, weight parameters, input and output data, etc.);

[0224] Download address or storage address information for the model file (used to indicate the storage address of the model file, or where the model file can be downloaded);

[0225] Analysis of identifiers (indicating that the model is applicable to a certain type of reasoning task);

[0226] Model filtering information (used to limit the scope of the federated learning process, such as regional scope, time scope, S-NSSAI, DNN, etc.);

[0227] Valid area information (the area to which the model is applicable);

[0228] Effective time information (the time during which the model is applicable).

[0229] Step 9: MTLF1 sends the model information of the generated federated learning model to AnLF.

[0230] In this step, MTLF1 can send a notification message to the model (Nnwdaf_MLModelProvision_Notify) or a response message to the model information (Nnwdaf_MLModelInfo_Response). For details on the content to be sent, please refer to step 8.

[0231] Step 10: AnLF generates data analysis results based on the model.

[0232] Step 11: Optionally, AnLF sends the data analysis results to consumer NF.

[0233] Example 2:

[0234] This embodiment uses MTLF1 to determine the FL client as an example for illustration. Specifically, as follows... Figure 5 As shown, the difference from Embodiment 1 is that:

[0235] In step 3, MTLF1 obtains the FL server and FL client from NRF;

[0236] In step 5, optionally, MTLF1 simultaneously specifies the FL clients participating in this federated learning to the FL server. This eliminates the need for the FL server to query the NRF for FL clients.

[0237] In this embodiment, when facing data privacy issues and needing to generate models using federated learning, even if the model training node itself does not support federated learning capabilities or federated learning server capabilities, or if a specific analysis identifier or AOI does not support federated learning or federated learning server capabilities, the model training node can still trigger other devices to perform federated learning, thereby obtaining the model it needs. This expands the application scope of federated learning, thus solving data privacy problems on a larger scale.

[0238] Please see Figure 6 , Figure 6 This is a structural diagram of a model information acquisition device provided in an embodiment of this application, such as... Figure 6 As shown, it includes:

[0239] The first determination module 601 is used to determine the federated learning FL server node;

[0240] The first sending module 602 is used to send a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain the target model;

[0241] The receiving module 603 is used to receive the target model information sent by the FL server node.

[0242] Optionally, the determining module 601 is configured to send a node discovery request message to the network repository function element, the node discovery request message being used to request network nodes to participate in federated learning training; and to receive a response message sent by the network repository function element, the response message including information about the FL server node.

[0243] Optionally, the node discovery request message includes at least one of the following:

[0244] Analysis includes identifiers, region of interest (AOI) information, time of interest information, model description information, model shareability information, model performance information, model algorithm information, model training speed information, federated learning instruction information, federated learning type information, FL server node type instruction information, FL client node type instruction information, first service information, and second service information.

[0245] The federated learning instruction information is used to indicate that the network node requested by the request message needs to support federated learning.

[0246] The first service information is used to indicate that the network node requested by the request message needs to support the services of the federated learning server.

[0247] The second service information is used to indicate that the network node requested by the request message needs to support the services of federated learning members.

[0248] Optionally, the federated learning type information is used to indicate that the network node requested by the request message needs to support at least one of the following types of federated learning:

[0249] Horizontal federated learning type;

[0250] Vertical federated learning type.

[0251] Optionally, the response message includes information about N network nodes, including the FL server node, where N is a positive integer;

[0252] Each network node's information includes at least one of the following:

[0253] Fully Qualified Domain Name (FQDN), identification information, and address information.

[0254] Optionally, the N network nodes may also include FL client nodes.

[0255] Optionally, the information for each network node may also include:

[0256] Type information, which indicates the type of network node, which is either an FL server node or an FL client node.

[0257] Optionally, the first request message includes at least one of the following:

[0258] Federated learning instruction information and model identifiers;

[0259] in,

[0260] The federated learning instruction information is used to request that the FL server node trigger federated learning to obtain the target model.

[0261] The model identifier is used to uniquely identify the target model.

[0262] Optionally, the device further includes:

[0263] The acquisition module is used to acquire the model identifier.

[0264] Optionally, the first request message includes: information about the FL client nodes participating in federated learning.

[0265] Optionally, the information of the target model includes at least one of the following pieces of information corresponding to the target model:

[0266] Model identifier, federation instruction information, model file, and model file address information;

[0267] The federated learning instruction information is used to indicate that the target model is a model obtained through federated learning.

[0268] The model identifier is used to uniquely identify the target model.

[0269] Optionally, the device further includes:

[0270] The second sending module is used to send the target model information to the model inference function node.

[0271] Optionally, the device further includes:

[0272] The second determination module is used to determine whether the target model needs to be obtained through federated learning.

[0273] Optionally, the second determining module is used to determine that, if the model training function node determines that it is impossible to obtain all or part of the training data for generating the target model, it is necessary to obtain the target model through federated learning.

[0274] The aforementioned model information acquisition device can improve the model training performance of network nodes.

[0275] The model information acquisition device in this application embodiment can be an electronic device, such as an electronic device with an operating system, or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a core network device, a network-side device, or a terminal, or it can be any device other than a terminal.

[0276] The model information acquisition device provided in this application embodiment can achieve... Figure 2 The various processes implemented in the method embodiments shown achieve the same technical effects, and will not be described again here to avoid repetition.

[0277] Please see Figure 7 , Figure 7 This is a structural diagram of a model information sending device provided in an embodiment of this application, such as... Figure 7 As shown, it includes:

[0278] The receiving module 701 is used to receive a first request message sent by the model training function node. The first request message is used to trigger the federated learning FL server node to perform federated learning to obtain the target model.

[0279] Learning module 702 is used to perform federated learning with FL client nodes based on the first request message to obtain the target model;

[0280] The sending module 703 is used to send the information of the target model to the model training function node.

[0281] Optionally, the first request message includes at least one of the following:

[0282] Federated learning instruction information and model identifiers;

[0283] in,

[0284] The federated learning instruction information is used to request that the FL server node trigger federated learning to obtain the target model.

[0285] The model identifier is used to uniquely identify the target model.

[0286] Optionally, the first request message includes: information about the FL client nodes participating in federated learning.

[0287] Optionally, the device further includes:

[0288] The determination module is used to identify the FL client nodes participating in federated learning.

[0289] Optionally, the determining module is configured to send a node discovery request message to the network repository function element, the node discovery request message being used to request FL client nodes participating in federated learning; and to receive a response message sent by the network repository function element, the response message including information about the FL client nodes participating in federated learning.

[0290] Optionally, the information of the federated model includes at least one of the following pieces of information corresponding to the target model:

[0291] Model identifier, federation instruction information, model file, and model file address information;

[0292] The federated learning instruction information is used to indicate that the target model is a model obtained through federated learning.

[0293] The model identifier is used to uniquely identify the target model.

[0294] Optionally, the model identifier is obtained by the model training function node for the target model.

[0295] Optionally, the device further includes:

[0296] The acquisition module is used to acquire the model identifier for the target model.

[0297] The aforementioned model information transmission device can improve the model training performance of network nodes.

[0298] The model information transmitting device in this application embodiment can be an electronic device, such as an electronic device with an operating system, or a component in an electronic device, such as an integrated circuit or a chip. The electronic device can be a core network device, a network-side device, or a terminal, or it can be any device other than a terminal.

[0299] The model information sending device provided in this application embodiment can achieve... Figure 2 The various processes implemented in the method embodiments shown achieve the same technical effects, and will not be described again here to avoid repetition.

[0300] Optional, such as Figure 8 As shown, this application embodiment also provides a communication device 800, including a processor 801 and a memory 802. The memory 802 stores a program or instructions that can run on the processor 801. For example, when the communication device 800 is a first control network element, when the program or instructions are executed by the processor 801, they implement the various steps of the above-described model information acquisition method or the above-described model information transmission method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0301] This application embodiment also provides a model training function node element, including a processor and a communication interface, wherein the processor or communication interface is used to determine the FL server node; the communication interface is used to send a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain a target model; and to receive information about the target model sent by the FL server node.

[0302] This application embodiment also provides a server node, including a processor and a communication interface, wherein the communication interface is used to receive a first request message sent by a model training function node, the first request message being used to trigger a federated learning (FL) server node to perform federated learning to obtain a target model; the processor or communication interface is used to perform federated learning with an FL client node based on the first request message to obtain the target model; the communication interface is used to send information about the target model to the model training function node.

[0303] Specifically, embodiments of this application also provide a network node. For example... Figure 9 As shown, the network node 900 includes a processor 901, a network interface 902, and a memory 903. The network interface 902 is, for example, a common public radio interface (CPRI).

[0304] Specifically, the network node 900 in this embodiment further includes: instructions or programs stored in memory 903 and executable on processor 901, wherein processor 901 calls the instructions or programs in memory 903 to execute. Figures 3 to 5 The methods executed by each module shown achieve the same technical effect, and to avoid repetition, they will not be described in detail here.

[0305] In the above embodiment where the network node is a model training function node:

[0306] Processor 901 or network interface 902 is used to determine the federated learning (FL) server node;

[0307] Network interface 902 is used to send a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain the target model; and to receive information about the target model sent by the FL server node.

[0308] Optionally, determining the FL server node includes:

[0309] Send a node discovery request message to the network repository function element. The node discovery request message is used to request network nodes to participate in federated learning training.

[0310] The system receives a response message sent by the network warehouse function element, the response message including information about the FL server node.

[0311] Optionally, the node discovery request message includes at least one of the following:

[0312] Analysis includes identifiers, region of interest (AOI) information, time of interest information, model description information, model shareability information, model performance information, model algorithm information, model training speed information, federated learning instruction information, federated learning type information, FL server node type instruction information, FL client node type instruction information, first service information, and second service information.

[0313] The federated learning instruction information is used to indicate that the network node requested by the request message needs to support federated learning.

[0314] The first service information is used to indicate that the network node requested by the request message needs to support the services of the federated learning server.

[0315] The second service information is used to indicate that the network node requested by the request message needs to support the services of federated learning members.

[0316] Optionally, the federated learning type information is used to indicate that the network node requested by the request message needs to support at least one of the following types of federated learning:

[0317] Horizontal federated learning type;

[0318] Vertical federated learning type.

[0319] Optionally, the response message includes information about N network nodes, including the FL server node, where N is a positive integer;

[0320] Each network node's information includes at least one of the following:

[0321] Fully Qualified Domain Name (FQDN), identification information, and address information.

[0322] Optionally, the N network nodes may also include FL client nodes.

[0323] Optionally, the information for each network node may also include:

[0324] Type information, which indicates the type of network node, which is either an FL server node or an FL client node.

[0325] Optionally, the first request message includes at least one of the following:

[0326] Federated learning instruction information and model identifiers;

[0327] in,

[0328] The federated learning instruction information is used to request that the FL server node trigger federated learning to obtain the target model.

[0329] The model identifier is used to uniquely identify the target model.

[0330] Optionally, the processor 901 or network interface 902 is further configured to:

[0331] Obtain the model identifier.

[0332] Optionally, the first request message includes: information about the FL client nodes participating in federated learning.

[0333] Optionally, the information of the target model includes at least one of the following pieces of information corresponding to the target model:

[0334] Model identifier, federation instruction information, model file, and model file address information;

[0335] The federated learning instruction information is used to indicate that the target model is a model obtained through federated learning.

[0336] The model identifier is used to uniquely identify the target model.

[0337] Optionally, network interface 902 is also used for:

[0338] The model training node sends the target model information to the model inference node.

[0339] Optionally, before the model training function node determines the FL server node, the processor 901 is further configured to:

[0340] Federated learning is indeed necessary to obtain the target model.

[0341] Optionally, the determination requires obtaining the target model through federated learning, and the method includes:

[0342] If the model training function node determines that it is impossible to obtain all or part of the training data used to generate the target model, it is determined that federated learning is needed to obtain the target model.

[0343] In the above embodiment where the network node is an FL server node:

[0344] Network interface 902 is used to receive a first request message sent by the model training function node, the first request message being used to trigger the FL server node to perform federated learning to obtain a target model; based on the first request message, to perform federated learning with the FL client node to obtain the target model; and to send information about the target model to the model training function node.

[0345] Optionally, the first request message includes at least one of the following:

[0346] Federated learning instruction information and model identifiers;

[0347] in,

[0348] The federated learning instruction information is used to request that the FL server node trigger federated learning to obtain the target model.

[0349] The model identifier is used to uniquely identify the target model.

[0350] Optionally, the first request message includes: information about the FL client nodes participating in federated learning.

[0351] Optionally, the processor 901 or network interface 902 may also be used for:

[0352] Identify the FL client nodes participating in federated learning.

[0353] Optionally, determining the FL client nodes participating in federated learning includes:

[0354] Send a node discovery request message to the network repository function element. The node discovery request message is used to request FL client nodes participating in federated learning.

[0355] Receive a response message sent by the network repository function element, the response message including information about the FL client nodes participating in federated learning.

[0356] Optionally, the information of the federated model includes at least one of the following pieces of information corresponding to the target model:

[0357] Model identifier, federation instruction information, model file, and model file address information;

[0358] The federated learning instruction information is used to indicate that the target model is a model obtained through federated learning.

[0359] The model identifier is used to uniquely identify the target model.

[0360] Optionally, the model identifier is obtained by the model training function node for the target model.

[0361] Optionally, processor 90 is also used for:

[0362] Obtain the model identifier for the target model.

[0363] It should be noted that this embodiment uses the model training function node and the FL server node as core network elements for illustration.

[0364] This application also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the steps of the model information acquisition method provided in this application, or the steps of the model information transmission method provided in this application.

[0365] The processor is the processor in the terminal described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0366] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described model information acquisition method or model information sending method embodiments, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0367] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0368] This application also provides a computer program / program product, which is stored in a storage medium and executed by at least one processor to implement the various processes of the above-described model information acquisition method or model information transmission method embodiments, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0369] This application also provides a model information transmission system, including: a model training function node and a server node. The model training function node can be used to execute the steps of the model information acquisition method provided in this application, and the server node can be used to execute the steps of the model information sending method provided in this application.

[0370] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0371] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0372] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A method for acquiring model information, characterized in that, include: The model training function node determines the federated learning (FL) server node, and the model training function node includes the model training logic function MTLF. The model training function node sends a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain the target model; The model training function node receives the target model information sent by the FL server node; The model training function node determines the FL server node, including: The model training function node sends a node discovery request message to the network repository function element. The node discovery request message is used to request network nodes that participate in federated learning training. The model training function node receives a response message sent by the network repository function element, and the response message includes information about the FL server node.

2. The method as described in claim 1, characterized in that, The node discovery request message includes at least one of the following: Analysis includes identifiers, region of interest (AOI) information, time of interest information, model description information, model shareability information, model performance information, model algorithm information, model training speed information, federated learning instruction information, federated learning type information, FL server node type instruction information, FL client node type instruction information, first service information, and second service information. The federated learning instruction information is used to indicate that the network node requested by the request message needs to support federated learning. The first service information is used to indicate that the network node requested by the request message needs to support the services of the federated learning server. The second service information is used to indicate that the network node requested by the request message needs to support the services of federated learning members.

3. The method as described in claim 2, characterized in that, The federated learning type information is used to indicate that the network node requested by the request message needs to support at least one of the following types of federated learning: Horizontal federated learning type; Vertical federated learning type.

4. The method as described in claim 1, characterized in that, The response message includes information about N network nodes, including the FL server node, where N is a positive integer; Each network node's information includes at least one of the following: Fully Qualified Domain Name (FQDN), identification information, and address information.

5. The method as described in claim 4, characterized in that, The N network nodes also include FL client nodes.

6. The method as described in claim 4 or 5, characterized in that, The information for each network node also includes: Type information, which indicates the type of network node, which is either an FL server node or an FL client node.

7. The method according to any one of claims 1 to 5, characterized in that, The first request message includes at least one of the following: Federated learning instruction information and model identifiers; in, The federated learning instruction information is used to request that the FL server node trigger federated learning to obtain the target model. The model identifier is used to uniquely identify the target model.

8. The method according to claim 7, characterized in that, The method further includes: The model training function node obtains the model identifier.

9. The method according to any one of claims 1 to 5, characterized in that, The first request message includes information about the FL client nodes participating in federated learning.

10. The method according to any one of claims 1 to 5, characterized in that, The information of the target model includes at least one of the following: Model identifier, federation instruction information, model file, and model file address information; The federated learning instruction information is used to indicate that the target model is a model obtained through federated learning. The model identifier is used to uniquely identify the target model.

11. The method according to any one of claims 1 to 5, characterized in that, The method further includes: The model training node sends the target model information to the model inference node.

12. The method according to any one of claims 1 to 5, characterized in that, Before the model training functional node determines the FL server node, the method further includes: The model training function node determines that federated learning is needed to obtain the target model.

13. The method as described in claim 12, characterized in that, The model training function node determines that federated learning is needed to obtain the target model, including: If the model training function node determines that it is unable to obtain all or part of the training data used to generate the target model, the model training function node determines that it is necessary to obtain the target model through federated learning.

14. A method for transmitting model information, characterized in that, include: The federated learning FL server node receives a first request message sent by the model training function node. The first request message is used to trigger the FL server node to perform federated learning to obtain the target model. The model training function node includes the model training logic function MTLF. The FL server node performs federated learning with the FL client node based on the first request message to obtain the target model; The FL server node sends the target model information to the model training function node. Wherein, the FL server node is the FL server node indicated by the response message sent by the network repository function element to the model training function node. The response message is a response message to the node discovery request message sent by the model training function node to the network repository function element. The node discovery request message is used to request network nodes to participate in federated learning training.

15. The method as described in claim 14, characterized in that, The first request message includes at least one of the following: Federated learning instruction information and model identifiers; in, The federated learning instruction information is used to request that the FL server node trigger federated learning to obtain the target model. The model identifier is used to uniquely identify the target model.

16. The method as described in claim 14 or 15, characterized in that, The first request message includes information about the FL client nodes participating in federated learning.

17. The method as described in claim 14 or 15, characterized in that, The method further includes: The FL server node determines the FL client nodes that participate in federated learning.

18. The method as described in claim 17, characterized in that, The FL server node determines the FL client nodes participating in federated learning, including: The FL server node sends a node discovery request message to the network repository function element. The node discovery request message is used to request FL client nodes participating in federated learning. The FL server node receives a response message sent by the network repository function element, and the response message includes information about the FL client nodes participating in federated learning.

19. The method as described in claim 14 or 15, characterized in that, The information of the target model includes at least one of the following: Model identifier, federation instruction information, model file, and model file address information; The federated learning instruction information is used to indicate that the target model is a model obtained through federated learning. The model identifier is used to uniquely identify the target model.

20. The method as described in claim 15, characterized in that, The model identifier is obtained by the model training function node for the target model.

21. The method as described in claim 19, characterized in that, The method further includes: The FL server node obtains the model identifier for the target model.

22. A model information acquisition device, characterized in that, The model training function node corresponding to the device includes the Model Training Logic Function (MTLF), which includes: The first determination module is used to determine the federated learning FL server node; The first sending module is used to send a first request message to the FL server node, the first request message being used to trigger the FL server node to perform federated learning to obtain the target model; The receiving module is used to receive the target model information sent by the FL server node; The first determining module is used to send a node discovery request message to the network repository function element, the node discovery request message being used to request network nodes to participate in federated learning training; and to receive a response message sent by the network repository function element, the response message including information about the FL server node.

23. A model information transmitting device, characterized in that, include: The receiving module is used to receive a first request message sent by the model training function node. The first request message is used to trigger the federated learning (FL) server node to perform federated learning to obtain the target model. The model training function node includes the model training logic function (MTLF). The learning module is used to perform federated learning with the FL client node based on the first request message to obtain the target model; The sending module is used to send the target model information to the model training function node; Wherein, the FL server node is the FL server node indicated by the response message sent by the network repository function element to the model training function node. The response message is a response message to the node discovery request message sent by the model training function node to the network repository function element. The node discovery request message is used to request network nodes to participate in federated learning training.

24. A model training function node, characterized in that, It includes a processor and a memory, the memory storing a program or instructions that can run on the processor, the program or instructions being executed by the processor to implement the steps of the model information acquisition method as described in any one of claims 1 to 13.

25. A server node, characterized in that, It includes a processor and a memory, the memory storing a program or instructions that can run on the processor, the program or instructions being executed by the processor to implement the steps of the model information transmission method as described in any one of claims 14 to 21.

26. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of the model information acquisition method as described in any one of claims 1 to 13, or the steps of the model information transmission method as described in any one of claims 14 to 21.

Images