Associative learning methods, apparatus, related equipment, and storage media

Abstract

The present disclosure discloses a federated learning method, apparatus, related device, and storage medium, wherein the method includes a first network function sending a first request to a network control device corresponding to a type of federated learning node, the first request being used to request optimization of network connections between the federated learning nodes.

Description

[Technical Field]

[0001] (Cross-reference to Related Applications) The present disclosure is proposed based on the Chinese patent application with application number 202310118232.6 and filing date of February 10, 2023, claims priority based on the Chinese patent application, and the entire content of the Chinese patent application is incorporated into the present disclosure by reference.

[0002] The present disclosure relates to the technical field of artificial intelligence technology, and in particular to a federated learning method, an apparatus, related equipment and a storage medium. [Background Art]

[0003] In order to solve the problems of data fragmentation and data silos caused by insufficient data samples in a single domain and the inability to share data between different domains, related technologies have proposed the concept of federated learning. Federated learning is a typical representative of collaborative learning, which improves the effect of machine learning modeling by performing distributed federated modeling without leaving own data locally and without leaking data privacy. However, the execution efficiency of existing federated learning strategies is low. [Summary of the Invention] [Problem to be Solved by the Invention]

[0004] In order to solve related technical problems, embodiments of the present disclosure provide a federated learning method, an apparatus, related equipment and a storage medium. [Means for Solving the Problem]

[0005] The technical solution according to an embodiment of the present disclosure is implemented as follows. An embodiment of the present disclosure provides a federated learning method applied to a first network function, the method comprising:[1] comprising transmitting a first request to a network control device corresponding to a type of a federated learning node, wherein the first request is used for requesting to optimize network connections between federated learning nodes.

[0006] In the above embodiment, transmitting the first request to the network control device corresponding to the type of federated learning node means: The process includes sending a first request to a network control device corresponding to the type of federated learning node if the network connectivity quality between federated learning nodes is below a predetermined threshold.

[0007] In the above embodiment, the first request is accompanied by relevant parameters for optimizing the network connection.

[0008] In the above embodiment, if the federated learning node is a federated learning node within a carrier network, the network control device is a Software Defined Network (SDN) controller.

[0009] In the above embodiment, if the federated learning node is an external node of the carrier network, the network control device is a corresponding connected node of the external network.

[0010] In the above embodiment, when the federated learning node is a terminal (UE, User Equipment), the network control device is a policy control function (PCF, Policy Control Function).

[0011] In the above embodiment, the first requirement is: The Internet Protocol (IP) address of the federated learning node associated with the network connection to be optimized, It must include at least one of the following requirements: Quality of Service (QoS) requirements for network connectivity between federated learning nodes.

[0012] In the above embodiment, the first requirement is: UE identifier and It must include at least one of the UE bearer flow QoS requirements.

[0013] Embodiments of this disclosure further provide a federated learning method applicable to network control devices, the method being: The first request is to receive a first request transmitted from the first network function, the first request being used to request optimization of the network connection between federated learning nodes, and the types of federated learning nodes and network control devices correspond to each other. This includes optimizing the network connection based on the first request received.

[0014] In the above embodiment, the first request is accompanied by relevant parameters for optimizing the network connection.

[0015] In the above embodiment, if the federated learning node is a federated learning node within the carrier network, the network control device is an SDN controller.

[0016] In the above embodiment, optimizing the network connection based on the received first request is: The process includes calculating an optimal route based on a first received request and transmitting the optimal route to the associated routing device, the optimal route being used to transmit data between federated learning nodes.

[0017] In the above embodiment, if the federated learning node is the UE, the network control device is the PCF.

[0018] In the above embodiment, optimizing the network connection based on the received first request is: This includes sending policy control and charging (PCC) policies to the user plane function (UPF) via the session management function (SMF).

[0019] In the above embodiment, the PCC policy is: Resource allocation priority, a guaranteeable speed, a maximum speed, and a maximum packet loss rate.

[0020] In the above aspect, when the federated learning node is an external node of an operator network, the network control device is a corresponding connection node of an external network.

[0021] In the above aspect, optimizing network connection based on the received first request comprises: controlling a network control node corresponding to the corresponding connection node of the external network to configure a related route.

[0022] In the above aspect, the first request: carries at least one of an IP address of a federated learning node associated with a network connection to be optimized, and a QoS requirement for a network connection between federated learning nodes.

[0023] In the above aspect, the first request: carries at least one of an identifier of a UE, and a QoS requirement for a bearer flow of the UE.

[0024] In the above aspect, the method further comprises: deleting the received first request when federated learning is completed.

[0025] An embodiment of the present disclosure further provides a federated learning device comprising a first transmitting unit, wherein the first transmitting unit is configured to transmit a first request to a network control device corresponding to a type of a federated learning node, and the first request is used for requesting to optimize a network connection between federated learning nodes.

[0026] An embodiment of the present disclosure further provides a federated learning device comprising a first receiving unit and a first processing unit, The first receiving unit is configured to receive a first request transmitted from a first network function, the first request being used to request optimization of network connectivity between federated learning nodes, and the types of federated learning nodes and network control devices correspond to each other. The first processing unit is configured to optimize the network connection based on the first request it receives.

[0027] Embodiments of this disclosure further provide a first network function including a first processor and a first communication interface, The first communication interface is configured to send a first request to a network control device corresponding to the type of federated learning node, the first request being used to request optimization of the network connection between the federated learning nodes.

[0028] The embodiments of this disclosure further provide network control equipment including a second processor and a second communication interface. The second communication interface is configured to receive a first request transmitted from the first network function, the first request being used to request optimization of the network connection between federated learning nodes, and the types of federated learning nodes and network control devices correspond to each other. The second processor is configured to optimize the network connection based on the first request it receives.

[0029] Embodiments of the present disclosure further provide a first network function including a first processor and a first memory for storing a computer program executable on the first processor. Here, the first processor is configured to execute one of the steps on the first network function side when executing the computer program.

[0030] Embodiments of the present disclosure further provide a network control device including a second processor and a second memory for storing a computer program executable on the second processor. Here, the second processor is configured to execute one of the steps on the network control device side when executing the computer program.

[0031] Embodiments of the present disclosure further provide a storage medium on which a computer program is stored, which implements a step of either the first network function side or a step of either the network control device side when the computer program is executed by a processor. [Effects of the Invention]

[0032] According to the federated learning method, apparatus, related equipment, and storage medium of the embodiments of this disclosure, a first network function transmits a first request to a network control device corresponding to the type of federated learning node, the first request is used to request optimization of the network connection between the federated learning nodes, the network control device receives the first request transmitted from the first network function and optimizes the network connection based on the received first request. In the above invention, the type of federated learning node and the network control device correspond to each other, and during the process of federated learning, the network control device can improve the network connection quality between the federated learning nodes, further improve the data transmission efficiency between the federated learning nodes, and improve the execution efficiency of federated learning by dynamically optimizing the network connection between the federated learning nodes. [Brief explanation of the drawing]

[0033] [Figure 1] This is an illustrative diagram of a federated learning system architecture applicable to embodiments of the present disclosure. [Figure 2] This is a schematic diagram illustrating the implementation flow of the federated learning method according to the embodiment of this disclosure. [Figure 3] This is a schematic diagram illustrating the implementation flow of the federated learning method according to the embodiment of this disclosure. [Figure 4] This is a schematic diagram of the interaction flow of the federated learning method according to the embodiment of this disclosure. [Figure 5]This is a schematic diagram of the interaction flow of the federated learning method according to the embodiment of this disclosure. [Figure 6] This is a schematic diagram of the interaction flow of the federated learning method according to the embodiment of this disclosure. [Figure 7] This is a schematic diagram of the configuration of a federated learning device according to an embodiment of the present disclosure. [Figure 8] This is a schematic diagram of the configuration of a federated learning device according to an embodiment of the present disclosure. [Figure 9] This is a schematic diagram of the structure of the first network function according to an embodiment of the present disclosure. [Figure 10] This is a schematic diagram of the structure of a network control device according to an embodiment of the present disclosure. [Modes for carrying out the invention]

[0034] Associative learning is a subset or paradigm of distributed machine learning (ML), also known as distributed associative learning. Currently, support for associative learning is being introduced in network intelligence applications. This involves distributing intelligent computing nodes within a network and supporting associative learning among them. Intelligent computing nodes can be understood as nodes possessing computing or data processing capabilities.

[0035] While conventional technical solutions for implementing federative learning primarily focus on optimizations of the federative learning algorithm itself, such as node selection, parameter aggregation, and parameter compression, solutions where a network supports federative learning lack a strategy to optimize data transmission efficiency between federative learning nodes by combining the advantages of the network. Since transmission efficiency is a crucial factor affecting the efficiency of federative learning execution, this leads to the low execution efficiency of existing federative learning strategies.

[0036] Based on this, in each embodiment of the present disclosure, the first network function transmits a first request to a network control device corresponding to the type of federated learning node, the first request is used to request optimization of the network connection between the federated learning nodes, the network control device receives the first request transmitted from the first network function and optimizes the network connection based on the received first request. In the above technical proposal, the type of federated learning node and the network control device correspond to each other, and in the process of federated learning, the network control device can improve the network connection quality between the federated learning nodes, further improve the data transmission efficiency between the federated learning nodes, and improve the execution efficiency of federated learning by dynamically optimizing the network connection between the federated learning nodes.

[0037] The present disclosure will be described in more detail below with reference to the drawings and embodiments.

[0038] First, to better illustrate the federated learning method according to the embodiments of this disclosure, Figure 1 shows an example of a federated learning system architecture applied to the embodiments of this disclosure. In the system architecture shown in Figure 1, consumers using ML services can request ML services from a first network function via the network function (NF) of an ML underlay network, and the NF of the ML underlay network can be understood as the operator's basic network function. The first network function mainly provides correlation services to federated learning, and the services provided include receiving federated learning tasks published by consumers using ML services, analyzing the federated learning tasks, determining federated learning nodes, distributing the federated learning tasks to the federated learning nodes, monitoring the network connection status between the federated learning nodes, and transmitting a first request to a network control device, where the first request is used to request optimization of the network connection between the federated learning nodes. A federated learning node can be understood as a network node or intelligent computing node that executes a federated learning task or participates in federated learning. A network node or intelligent computing node may be a Training Servers Function or a Data Servers Function. Associative learning tasks can be understood as distributed associative learning tasks or associative learning requirements.

[0039] The first network function may be an artificial intelligence (AI) service orchestration function in a carrier network, or an ML service orchestration function in a carrier network, for example, a machine learning function orchestration (MLFO) service function in a carrier network, i.e., an MLFO Services Function. Consumers using ML services include internal consumers and external consumers, with external consumers interacting with the NF of the ML underlay network via an outward-facing public interface of the ML underlay network. Internal consumers include the NF of the ML underlay network, and external consumers include consumers of ML services from external systems or carrier external networks.

[0040] A joint learning node may include a parameter aggregation node (DJL server), a distributed execution node (DJL client), and a distributed data processing node (Data service). Here, DJL is an abbreviation for Distributed Joint Learning. Both the parameter aggregation node and the distributed execution node must be registered with the first network function in advance, as overall orchestration control is performed by the first network function. The parameter aggregation node can be understood as a joint learning server. The distributed execution node can be understood as a training client or client execution node. The distributed data processing node is associated with the client node and is also called a data processing node or data service node. The training service function within the operator network can function as a parameter aggregation node and / or a distributed execution node.

[0041] Parameter aggregation nodes and distributed execution nodes may be located within the operator network, or within at least one of the following: UE, application function (AF), and external enterprise IT system, or within an external operator network such as another operator network. Other operator networks may include International Mobile Telecommunications (IMT) networks, such as the IMT-2020 network. In other words, parameter aggregation nodes and distributed execution nodes may be internal nodes of the operator network or external nodes of the operator network (nodes in an external operator network). Parameter aggregation nodes and distributed execution nodes located in an external operator network can access the ML underlay network via an outward-facing public interface of the ML underlay network and interact with the first network function and signaling.

[0042] Network control equipment may include at least one of the following: a controller responsible for lower-layer routing equipment (Router), a PCF, and corresponding connection nodes for the external network. The external network may be understood as the carrier's external network. The controller responsible for lower-layer routing equipment may include a Software Defined Network (SDN) controller. The SDN controller and Router together constitute the lower-layer transmission network of the carrier network, providing network connectivity for network nodes within the carrier network, where network nodes include federated learning nodes.

[0043] An embodiment of the present disclosure provides a federated learning method applicable to a first network function in the system architecture of Figure 1, the method including the following step 201, as shown in Figure 2.

[0044] In step 201, a first request is sent to the network control device corresponding to the type of federated learning node. Here, the first request is used to request that the network connection between the federated learning nodes be optimized.

[0045] Here, the first network function receives a federated learning task published by a consumer using the ML service, determines a federated learning node based on the received federated learning task, determines the type of federated learning node, determines the network control device corresponding to the type of federated learning node, and sends a first request to the determined network control device. The first request can be understood as a request for optimization of the network connection. The first network function may send the first request to a network controller corresponding to some or all of the types of federated learning nodes, where the network controller corresponding to some or all of the types of federated learning nodes may be a network controller corresponding to the type of federated learning node associated with the network connection to be optimized.

[0046] The implementation process for determining a federative learning node based on a received federative learning task may involve analyzing the received federative learning task to obtain analysis results, and then selecting a federative learning node from the intelligent computing nodes based on these analysis results. When analyzing the received federative learning task, it is possible to analyze the needs of the federative learning task and the characteristics of the associated data, as well as the characteristics of the data of the intelligent computing nodes, and the distance between the location of the intelligent computing nodes and the consumers using the ML service.

[0047] The type of federated learning node can be configured according to the actual situation, and a correspondence or relationship exists between the type of federated learning node and the network control device. The network control device corresponding to a type of federated learning node can be understood as the network control device responsible for controlling the network connectivity of that type of federated learning node.

[0048] Furthermore, since network connectivity control between different types of federated learning nodes is handled by different types of network control devices, sending the first request to the network control device corresponding to the type of federated learning node, rather than sending the first request to all network control devices in the network, reduces the signaling overhead of the first network function, saves network controller resources unrelated to the network connectivity to be optimized, and improves the success rate of network connectivity optimization.

[0049] In one embodiment, the type of federated learning node includes at least one of the first, second, and third types. Here, the first type is used to indicate that the federated learning node is a federated learning node within a carrier network, and includes at least one of a parameter aggregation node, a distributed execution node, and a distributed data processing node, where the federated learning node within the carrier network can be understood as the training service function and / or data service function within the carrier network in Figure 1. The second type is used to indicate that the federated learning node is a UE. The third type is used to indicate that the federated learning node is an external node of the carrier network, where the external node of the carrier network includes a parameter aggregation node and / or a distributed execution node.

[0050] During the federated learning process, if the nodes participating in federated learning change due to factors such as the state of the federated learning nodes, network conditions, or user needs, the first network function needs to re-select or adjust the nodes that will perform the federated learning task. Furthermore, if the federated learning nodes change, the network connections between the federated learning nodes may also change, requiring the first network function to re-determine the first requirement. Based on this, in one embodiment, the method further includes: This includes re-determining the first request if the federated learning node changes.

[0051] Here, the first network function is The state information of the associative learning node, Network status and The position information of the federated learning node, Characteristics of the data in the associative learning node, Based on at least one of the needs information for the associative learning task, it is possible to decide whether to modify or adjust the associative learning node.

[0052] When a federated learning node is modified or adjusted, the network connectivity between the federated learning nodes is re-determined based on the modified or adjusted federated learning node, and the first request is re-determined based on the re-determined network connectivity between the federated learning nodes.

[0053] Here, the state information of a federated learning node may be used to indicate whether the federated learning node is available or whether a failure has occurred. Network status, whether it is available, is understood to be used to indicate whether the network connection status is stable or whether an anomaly exists. Data characteristics of a federated learning node may be understood to include the data types and data processing methods that the federated learning node supports. Needs information for a federated learning task may include capability needs and / or data processing needs for the intelligent computing node.

[0054] To save signaling overhead for the first network function and conserve data processing resources for the network control device, the first network function can request the network control device to optimize the network connection between federated learning nodes when the network connection quality between federated learning nodes is poor. Based on this, in one embodiment, sending the first request to the network control device corresponding to the type of federated learning node is: The process includes sending a first request to a network control device corresponding to the type of federated learning node if the network connectivity quality between federated learning nodes is below a predetermined threshold.

[0055] Here, the first network function detects the network connectivity quality between federated learning nodes executing a distributed federated learning task. If the network connectivity quality between federated learning nodes is below a predetermined threshold, the network connectivity between the federated learning nodes needs to be optimized, and the first network function sends a first request to the network control device corresponding to the type of federated learning node. Here, network connectivity quality can be measured or evaluated by indicators such as packet loss rate and / or delay.

[0056] Furthermore, if the network connection quality between federated learning nodes is greater than a predetermined threshold, the first request does not need to be sent to the network control device corresponding to the type of federated learning node.

[0057] In order to improve the efficiency and success rate of optimizing network connections, in one embodiment, the first requirement is accompanied by relevant parameters for optimizing network connections.

[0058] Since network connectivity control between different types of federated learning nodes is handled by different types of network control devices, in one embodiment, if the federated learning node is a federated learning node within a carrier network, the network control device is an SDN controller, in order to accurately optimize the network connectivity that should be optimized.

[0059] In order to precisely optimize the network connections that need to be optimized, in one embodiment, if the federated learning node is the UE, the network control device is the PCF.

[0060] In order to accurately optimize the network connections that need to be optimized, in one embodiment, if the federated learning node is an external node of the carrier network, the network control device is the corresponding connection node of the external network.

[0061] Here, the external network can be understood as the carrier's external network. The corresponding connected nodes of the external network include at least one of the following: AF, the IT systems of an external company, or the MLFO (MLFO of the external network) service function of another carrier network.

[0062] If the federated learning node is a federated learning node within the carrier network or an external node within the carrier network, the first request is: The IP address of the federated learning node associated with the network connection to be optimized, It includes at least one of the QoS requirements for network connectivity between federated learning nodes.

[0063] Here, if the network connection to be optimized is associated with federated learning nodes within the carrier network, the SDN controller is responsible for controlling the network connection between the federated learning nodes within the carrier network, and therefore the first network function sends a first request to the SDN controller.

[0064] When the network connection to be optimized is associated with an external node of the carrier network (a federated learning node of the carrier's external network), the network connection control of the carrier's external node is handled by the corresponding connection node of the external network. Since the network control device connected to correspond to the first network function is the corresponding connection node of the external network, the first network function sends the first request to the corresponding connection node of the external network via a predetermined interface. Whether or not the network connection of the carrier's external network can be optimized depends on the functionality of the carrier's external network.

[0065] The first requirement includes at least one of the following: the IP addresses of the federated learning nodes associated with the network connection to be optimized, and the QoS requirements for the network connection between the federated learning nodes. Here, the IP addresses of the federated learning nodes are used by the network control device to determine the network connection to be optimized. The federated learning nodes associated with the network connection to be optimized include at least one of the following: parameter aggregation nodes, distributed execution nodes, and distributed data processing nodes. The QoS requirements for the network connection include at least one of the following: maximum latency, minimum bandwidth, maximum packet loss rate, and maximum jitter.

[0066] If the federated learning node is a UE, in one embodiment, the first request is: UE identifier and It must include at least one of the UE bearer flow QoS requirements.

[0067] Here, if a network connection to be optimized is associated with a UE, the first network function sends a first request to the PCF, which is responsible for user plane control of the UE. In this case, the first request can be understood as a user plane optimization request. The first request is accompanied by at least one of the UE identifier and the QoS requirements for the UE's bearer flow. The UE identifier includes the User Permanent Identifier (SUPI) and / or the International Mobile Subscriber Identity (IMSI). The QoS requirements for the UE's bearer flow include at least one of the maximum latency, maximum packet loss rate, and maximum jitter. The UE's bearer flow can be understood as the UE's Protocol Data Unit (PDU) session channel, also known as the PDU session channel.

[0068] To correspond to the above, embodiments of the present disclosure further provide a federated learning method applicable to network control devices, the method comprising the following steps 301 and 302, as shown in Figure 3.

[0069] In step 301, a first request is received from the first network function. The first request is used to request optimization of the network connection between federated learning nodes, and the types of federated learning nodes and network control devices correspond to each other.

[0070] In step 302, the network connection is optimized based on the first request received.

[0071] Here, the network control device determines which network connection should be optimized based on the first request it receives, and then optimizes that network connection.

[0072] In order to improve the efficiency of optimizing network connectivity, in one embodiment, the first requirement is accompanied by relevant parameters for optimizing network connectivity.

[0073] In one embodiment, if the federated learning node is a federated learning node within a carrier network, the network control device is an SDN controller.

[0074] In one embodiment, if the federated learning node is an external node of the carrier network, the network control device is a corresponding connected node of the external network.

[0075] In one embodiment, when the federated learning node is the UE, the network control device is the PCF.

[0076] If the network control device is an SDN controller or a corresponding connection node of an external network, in one embodiment, the first requirement is: The IP address of the federated learning node associated with the network connection to be optimized, It includes at least one of the QoS requirements for network connectivity between federated learning nodes.

[0077] Here, the network control device can determine the federated learning node based on the IP address of the federated learning node associated with the network connection to be optimized, and determine the network connection associated with the federated learning node to obtain the network connection to be optimized.

[0078] When the network control device is a PCF, in one embodiment, the first requirement is: UE identifier and It must include at least one of the UE bearer flow QoS requirements.

[0079] If the federated learning node associated with the network connection to be optimized is a federated learning node within the carrier network, the network control device is an SDN controller, and the first network function sends a first request to the SDN controller. Based on this, in one embodiment, in order to optimize the network connection between federated learning nodes within the carrier network, optimizing the network connection based on the received first request is: The process includes calculating an optimal route based on a first received request and transmitting the optimal route to the associated routing device, the optimal route being used to transmit data between federated learning nodes.

[0080] Here, when the SDN controller receives a first request, it calculates the optimal route between federated learning nodes based on the information attached to the received first request, and sends the optimal route to the routing device associated with that route. The associated routing device then transmits the relevant data of the federated learning nodes based on the received optimal route, thereby improving data transmission efficiency.

[0081] For example, the SDN controller determines which network connection to optimize based on the IP addresses of the federated learning nodes associated with that connection, and then calculates the optimal path between the federated learning nodes associated with the network connection to optimize based on the network topology structure of the carrier network and the QoS requirements of the IP addresses of the federated learning nodes and / or the network connections between the federated learning nodes.

[0082] If the federated learning node associated with the network connection to be optimized is a UE, the network control device is a PCF, and the first network function sends a first request to the PCF. Based on this, in one embodiment, in order to optimize the network connection associated with the UE, optimizing the network connection based on the received first request is: This includes sending PCC policies to UPF via SMF.

[0083] Here, the PCF determines a PCC policy based on the information attached to the first request and sends the PCC policy to the SMF, which then forwards the received PCC policy to the UPF. In this way, the UPF can optimize the transmission policy for the UE's relevant data based on the received PCC policy, thereby improving the transmission efficiency of the UE's relevant data. Here, the PCC policy may be pre-configured or may be determined by the PCF based on the UE's bearer flow QoS requirements attached to the first request.

[0084] In one embodiment, the PCC policy is: Resource allocation priority, Guaranteed speed and Maximum speed and This includes at least one of the following: maximum packet loss rate.

[0085] When the federated learning node associated with the network connection to be optimized is an external node of the carrier network, the network control device is the corresponding connection node of the external network, and the first network function sends a first request to the corresponding connection node of the external network. Based on this, in one embodiment, in order to optimize the network connection of the external node of the carrier network, optimizing the network connection based on the received first request is: This includes controlling a network control node corresponding to a corresponding connection node of the external network to configure related paths.

[0086] Here, in order to improve the efficiency of data transmission between external nodes of the carrier network, when a corresponding connection node of the external network receives a first request transmitted from a first network function, it can forward the first request to a network control node corresponding to the corresponding connection node of the external network, and can control the network control node to configure a data transmission path between associated federated learning nodes based on the information attached to the first request.

[0087] Once federated learning is complete, in order to conserve the memory resources of the network control device, in one embodiment, the method further... When associative learning is complete, this includes deleting the first request that was received.

[0088] At this point, once federated learning is complete, the network controller no longer needs to focus on the network connections between the federated learning nodes and deletes the first request it received.

[0089] The following will provide a more detailed explanation of this disclosure with reference to application examples and a schematic diagram of the interaction flow.

[0090] [Application example 1] The federated learning node associated with the network connection to be optimized is a federated learning node within the carrier network, and the network control device is an SDN controller. As shown in Figure 4, the federated learning method includes the following steps 1 to 8.

[0091] In Step 1, the first network function receives federative learning tasks published by consumers using the ML service.

[0092] Here, the associative learning task can also be understood as an associative learning request. The first network function may be the MLFO of the carrier network.

[0093] In step 2, the first network function determines federative learning nodes based on the received federative learning task.

[0094] Here, the first network function analyzes the received federative learning task to obtain analysis results, and based on the analysis results, it can select federative learning nodes from intelligent computing nodes, for example, parameter aggregation nodes, distributed execution nodes, and distributed data processing nodes to participate in federative learning.

[0095] Here, when analyzing the received associative learning tasks, it is possible to analyze the needs of the associative learning tasks and the characteristics of the associated data, as well as the characteristics of the data of the intelligent computing nodes, and the distance between the location of the intelligent computing nodes and the consumers using the ML service.

[0096] Here, the first network function further, The state information of the associative learning node, Network status and The position information of the federated learning node, Characteristics of the data in the associative learning node, Based on at least one of the needs information for the associative learning task, the associative learning nodes can be re-determined or adjusted.

[0097] In step 3, the first network function sends a federated learning task to the determined federated learning node.

[0098] Here, the federative learning node performs federative learning based on the federative learning task it received.

[0099] In step 4, the first network function detects the network connectivity quality between federated learning nodes.

[0100] In step 5, if the network connectivity quality between federated learning nodes is below a predetermined threshold, and the federated learning node associated with the network connectivity to be optimized is a federated learning node within the carrier network, the first network function sends a first request to the SDN controller.

[0101] Here, if the network connection quality between federated learning nodes is greater than a predetermined threshold, the first network function does not need to send the first request. Here, the first request is: The IP address of the federated learning node associated with the network connection to be optimized, The QoS requirements include at least one of the following: maximum latency, minimum bandwidth, maximum packet loss rate, and maximum jitter.

[0102] Here, the first network function can re-determine the first request if the federated learning node changes.

[0103] In step 6, the SDN controller receives the first request and calculates the optimal route based on the received first request.

[0104] In one embodiment, once federated learning is complete, the SDN controller can delete the first request that it has received.

[0105] In step 7, the SDN controller sends the optimal route to the associated routing device.

[0106] In step 8, the routing device transmits data between federated learning nodes based on the received optimal route.

[0107] [Application example 2] The federated learning node associated with the network connection to be optimized is the UE, and the network control device is the PCF. As shown in Figure 5, the federated learning method includes the following steps 1 to 8.

[0108] In Step 1, the first network function receives federative learning tasks published by consumers using the ML service.

[0109] The implementation process of steps 1 to 4 in Figure 5 can be found in the related explanations in Application Example 1, and will therefore be omitted here.

[0110] In step 2, the first network function determines federative learning nodes based on the received federative learning task.

[0111] In step 3, the first network function sends a federated learning task to the determined federated learning node.

[0112] In step 4, the first network function detects the network connectivity quality between federated learning nodes.

[0113] In step 5, if the network connection quality between federated learning nodes is below a predetermined threshold, and the federated learning node associated with the network connection to be optimized is a UE, the first network function sends a first request to the PCF.

[0114] Here, the first requirement is: UE identifier and It must include at least one of the UE bearer flow QoS requirements.

[0115] Here, the UE identifier includes SUPI and / or IMSI. The UE's bearer flow QoS requirements include at least one of maximum delay, maximum packet loss rate, and maximum jitter.

[0116] In step 6, the PCF receives the first request sent from the first network function and determines the PCC policy based on the received first request.

[0117] Here, the PCC policy is: Resource allocation priority, Guaranteed speed and Maximum speed and This includes at least one of the following: maximum packet loss rate.

[0118] In one embodiment, once associative learning is complete, the PCF can delete the first request it has received.

[0119] In step 7, the PCF sends the determined PCC policy to the SMF.

[0120] In step 8, the SMF forwards the received PCC policy to the UPF, allowing the UPF to optimize the transmission policy for the relevant UE data based on the received PCC policy.

[0121] [Application Example 3] The federated learning node associated with the network connection to be optimized is an external node of the carrier network, and the network control device is the corresponding connection node of the external network. As shown in Figure 6, the federated learning method includes the following steps 1 to 6.

[0122] In Step 1, the first network function receives federative learning tasks published by consumers using the ML service.

[0123] The implementation process of steps 1 to 4 in Figure 6 can be found in the related explanations in Application Example 1, and will therefore be omitted here.

[0124] In step 2, the first network function determines federative learning nodes based on the received federative learning task.

[0125] In step 3, the first network function sends a federated learning task to the determined federated learning node.

[0126] In step 4, the first network function detects the network connectivity quality between federated learning nodes.

[0127] In step 5, if the network connection quality between federated learning nodes is below a predetermined threshold, and the federated learning node associated with the network connection to be optimized is an external node of the carrier network, the first network function sends a first request to the corresponding connection node of the external network.

[0128] Here, the corresponding connection node in the external network can be the MLFO of the external network.

[0129] The first request is, The IP address of the federated learning node associated with the network connection to be optimized, The QoS requirements include at least one of the following: maximum latency, minimum bandwidth, maximum packet loss rate, and maximum jitter.

[0130] In step 6, the corresponding connection node of the external network controls the network control node corresponding to the external network connection node to configure the relevant route based on the received first request.

[0131] Here, in order to improve the efficiency of data transmission between external nodes of the carrier network, a corresponding connected node of the external network can forward the received first request to the corresponding network control node, and can instruct the network control node to configure a data transmission path between associated federated learning nodes based on the information attached to the first request.

[0132] According to the federated learning method, apparatus, related equipment, and storage medium of the embodiments of this disclosure, a first network function transmits a first request to a network control device corresponding to the type of federated learning node, the first request is used to request optimization of the network connection between the federated learning nodes, the network control device receives the first request transmitted from the first network function and optimizes the network connection based on the received first request. In the above invention, the type of federated learning node and the network control device correspond to each other, and during the process of federated learning, the network control device can improve the network connection quality between the federated learning nodes, further improve the data transmission efficiency between the federated learning nodes, and improve the execution efficiency of federated learning by dynamically optimizing the network connection between the federated learning nodes.

[0133] To realize the federated learning method of the embodiment of the present disclosure, the embodiment of the present disclosure further provides a federated learning device provided in a first network function, which includes a first transmission unit 701, as shown in Figure 7.

[0134] The first transmission unit 701 is configured to send a first request to a network control device corresponding to the type of federated learning node, the first request being used to request optimization of the network connection between the federated learning nodes.

[0135] In one embodiment, the first transmission unit 701 is specifically configured to transmit a first request to a network control device corresponding to the type of federated learning node when the network connection quality between federated learning nodes is below a predetermined threshold.

[0136] In one embodiment, the first request is accompanied by relevant parameters for optimizing the network connection.

[0137] In one embodiment, if the federated learning node is a federated learning node within a carrier network, the network control device is an SDN controller.

[0138] In one embodiment, if the federated learning node is an external node of the carrier network, the network control device is a corresponding connected node of the external network.

[0139] In one embodiment, when the federated learning node is the UE, the network control device is the PCF.

[0140] In one embodiment, the first requirement is: The IP address of the federated learning node associated with the network connection to be optimized, It includes at least one of the QoS requirements for network connectivity between federated learning nodes.

[0141] In one embodiment, the first requirement is: UE identifier and It must include at least one of the UE bearer flow QoS requirements.

[0142] In actual application, the first transmission unit 701 may be realized by a combination of a processor and a communication interface within the federated learning device.

[0143] In the above embodiment, when the federated learning device performs federated learning, only the division of each program module was used as an example for explanation. However, in actual application, the above processing can be assigned to different program modules as needed and completed; that is, the internal structure of the device can be divided into different program modules to complete all or part of the processing described above. Furthermore, the federated learning device according to the above embodiment belongs to the same concept as the federated learning method embodiment, and the specific implementation process can be found in the method embodiment, so the explanation is omitted here.

[0144] To realize the federated learning method of the embodiment of the present disclosure, the embodiment of the present disclosure further provides a federated learning device provided in a network control device, which, as shown in Figure 8, includes a first receiving unit 801 and a first processing unit 802.

[0145] The first receiving unit 801 is configured to receive a first request transmitted from a first network function, which is used to request optimization of the network connection between federated learning nodes, and the types of federated learning nodes and network control devices correspond to each other.

[0146] The first processing unit 802 is configured to optimize the network connection based on the first request it receives.

[0147] In one embodiment, the first request is accompanied by relevant parameters for optimizing the network connection.

[0148] In one embodiment, if the federated learning node is a federated learning node within a carrier network, the network control device is an SDN controller.

[0149] In one embodiment, the first processing unit 802 is specifically configured to calculate an optimal route based on a received first request and to transmit the optimal route to an associated routing device, the optimal route being used to transmit data between federated learning nodes.

[0150] In one embodiment, when the federated learning node is the UE, the network control device is the PCF.

[0151] In one embodiment, the first processing unit 802 is specifically configured to send the PCC policy to the UPF via the SMF.

[0152] In one embodiment, the PCC policy is: Resource allocation priority, Guaranteed speed and Maximum speed and This includes at least one of the following: maximum packet loss rate.

[0153] In one embodiment, if the federated learning node is an external node of the carrier network, the network control device is a corresponding connected node of the external network.

[0154] In one embodiment, the first processing unit 802 is specifically configured to control a network control node corresponding to a corresponding connection node of the external network in order to configure the associated path.

[0155] In one embodiment, the first requirement is: The IP address of the federated learning node associated with the network connection to be optimized, It includes at least one of the QoS requirements for network connectivity between federated learning nodes.

[0156] In one embodiment, the first requirement is: UE identifier and It must include at least one of the UE bearer flow QoS requirements.

[0157] In one embodiment, the first processing unit 802 is further configured to delete the received first request when federated learning is completed.

[0158] In actual application, the first receiving unit 801 and the first processing unit 802 can be realized by a combination of a processor and a communication interface within the federated learning device, or by a processor within the federated learning device.

[0159] In the above embodiment, when the federated learning device performs federated learning, only the division of each program module was used as an example for explanation. However, in actual application, the above processing can be assigned to different program modules as needed and completed; that is, the internal structure of the device can be divided into different program modules to complete all or part of the processing described above. Furthermore, the federated learning device according to the above embodiment belongs to the same concept as the federated learning method embodiment, and the specific implementation process can be found in the method embodiment, so the explanation is omitted here.

[0160] Based on the hardware implementation of the above-mentioned program module, in order to implement the first network function method according to the embodiment of the present disclosure, the embodiment of the present disclosure further provides a first network function, and as shown in Figure 9, the first network function 900 includes a first communication interface 901 and a first processor 902.

[0161] The first communication interface 901 can interact with other network nodes to exchange information. The first processor 902 is connected to the first communication interface 901 and is used to perform information interaction with other network nodes and to execute one or more technical solutions on the first network function side when executing a computer program. The computer program is stored in the first memory 903.

[0162] Specifically, the first communication interface 901 is configured to send a first request to a network control device corresponding to the type of federated learning node, and the first request is used to request that the network connection between the federated learning nodes be optimized.

[0163] In one embodiment, the first communication interface 901 is specifically configured to send a first request to a network control device corresponding to the type of federated learning node when the network connection quality between federated learning nodes is below a predetermined threshold.

[0164] In one embodiment, the first request is accompanied by relevant parameters for optimizing the network connection.

[0165] In one embodiment, if the federated learning node is a federated learning node within a carrier network, the network control device is an SDN controller.

[0166] In one embodiment, if the federated learning node is an external node of the carrier network, the network control device is a corresponding connected node of the external network.

[0167] In one embodiment, when the federated learning node is the UE, the network control device is the PCF.

[0168] In one embodiment, the first requirement is: The IP address of the federated learning node associated with the network connection to be optimized, It includes at least one of the QoS requirements for network connectivity between federated learning nodes.

[0169] In one embodiment, the first requirement is: UE identifier and It must include at least one of the UE bearer flow QoS requirements.

[0170] The specific processing steps of the first processor 902 and the first communication interface 901 can be understood by referring to the method described above.

[0171] Of course, in actual application, each component within the first network function 900 is connected via the bus system 904. It is understood that the bus system 904 is used to enable connection communication between these components. In addition to the data bus, the bus system 904 further includes a power bus, a control bus, and a status signal bus. However, for clarity, in Figure 9, the various buses are denoted as the bus system 904.

[0172] In embodiments of this disclosure, the first memory 903 stores various types of data to support the operation of the first network function 900. Examples of this data include any computer program to run on the first network function 900.

[0173] The methods according to the embodiments of the present disclosure described above may be applied within or implemented by the first processor 902. The first processor 902 may be an integrated circuit chip having signal processing capabilities. In the implementation process, each step of the method may be completed by hardware integrated logic circuits or software-based instructions within the first processor 902. The first processor 902 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The first processor 902 may implement or execute each method, step and logic block diagram according to the embodiments of the present disclosure. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods according to the embodiments of the present disclosure may be combined and directly implemented as completion by a hardware decoding processor, or completion by a combination of hardware and software modules within the decoding processor. The software module can be placed in a storage medium, which is located in a first memory 903, and the first processor 902 reads the information in the first memory 903 and combines its hardware to complete the steps of the above method.

[0174] In exemplary embodiments, the first network function 900 may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontroller units (MCUs), microprocessors, or other electronic components to perform the above method.

[0175] Based on the hardware implementation of the above-mentioned program module, in order to implement the network control device method according to the embodiment of the present disclosure, the embodiment of the present disclosure further provides a network control device, which, as shown in Figure 10, includes a second communication interface 1001 and a second processor 1002.

[0176] The second communication interface 1001 can interact with other network nodes to exchange information. The second processor 1002 is connected to the second communication interface 1001 and is used to perform information interaction with other network nodes and to execute one or more technical solutions on the network control device side when executing a computer program. The computer program is stored in the second memory 1003.

[0177] Specifically, the second communication interface 1001 is configured to receive a first request transmitted from the first network function, which is used to request the optimization of network connectivity between federated learning nodes, and the types of federated learning nodes and network control devices correspond to each other.

[0178] The second processor 1002 is configured to optimize the network connection based on the first request it receives.

[0179] In one embodiment, the first request is accompanied by relevant parameters for optimizing the network connection.

[0180] In one embodiment, if the federated learning node is a federated learning node within a carrier network, the network control device is an SDN controller.

[0181] In one embodiment, the second processor 1002 is specifically configured to calculate an optimal route based on a first request received and to transmit the optimal route to an associated routing device, the optimal route being used to transmit data between federated learning nodes.

[0182] In one embodiment, when the federated learning node is the UE, the network control device is the PCF.

[0183] In one embodiment, the second processor 1002 is specifically configured to send PCC policies to the UPF via the SMF.

[0184] In one embodiment, the PCC policy is: Resource allocation priority, Guaranteed speed and Maximum speed and This includes at least one of the following: maximum packet loss rate.

[0185] In one embodiment, if the federated learning node is an external node of the carrier network, the network control device is a corresponding connected node of the external network.

[0186] In one embodiment, the second processor 1002 is specifically configured to control a network control node corresponding to a corresponding connection node of the external network in order to configure the associated path.

[0187] In one embodiment, the first requirement is: The IP address of the federated learning node associated with the network connection to be optimized, It includes at least one of the QoS requirements for network connectivity between federated learning nodes.

[0188] In one embodiment, the first requirement is: UE identifier and It must include at least one of the UE bearer flow QoS requirements.

[0189] In one embodiment, the second processor 1002 is further configured to delete the received first request when federated learning is completed.

[0190] The specific processing steps of the second processor 1002 and the second communication interface 1001 can be understood by referring to the method described above.

[0191] Of course, in actual application, each component within the network control device 1000 is connected via the bus system 1004. It is understood that the bus system 1004 is used to enable connection communication between these components. In addition to the data bus, the bus system 1004 further includes a power bus, a control bus, and a status signal bus. However, for clarity, in Figure 10, the various buses are denoted as the bus system 1004.

[0192] In embodiments of this disclosure, the second memory 1003 stores various types of data to support the operation of the network control device 1000. Examples of this data include any computer program to run on the network control device 1000.

[0193] The methods according to the embodiments of the present disclosure described above may be applied within or implemented by the second processor 1002. The second processor 1002 may be an integrated circuit chip having signal processing capabilities. In the implementation process, each step of the method may be completed by hardware integrated logic circuits or software-based instructions within the second processor 1002. The second processor 1002 may be a general-purpose processor, a DSP, or other programmable logic device, discrete gate or transistor logic device, discrete hardware component, etc. The second processor 1002 may implement or execute each method, step and logic block diagram according to the embodiments of the present disclosure. The general-purpose processor may be a microprocessor or any conventional processor, etc. The steps of the methods according to the embodiments of the present disclosure may be combined and directly implemented as completion by a hardware decoding processor, or completion by a combination of hardware and software modules within the decoding processor. The software module can be placed in a storage medium, which is located in a second memory 1003, and the second processor 1002 reads the information in the second memory 1003 and combines its hardware to complete the steps of the above method.

[0194] In exemplary embodiments, the network control device 1000 may be implemented by one or more ASICs, DSPs, PLDs, CPLDs, FPGAs, general-purpose processors, controllers, MCUs, microprocessors, or other electronic components to perform the above method.

[0195] It should be understood that the memories (first memory 903, second memory 1003) in the embodiments of this disclosure may be volatile memory or non-volatile memory, and may include both volatile and non-volatile memory. Here, non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM®), flash memory, magnetic surface memory, optical disk, or read-only optical disk (CD-ROM, Compact Disc Read-Only Memory). Magnetic surface memory may be magnetic disk memory or magnetic tape memory. Volatile memory may be random access memory (RAM) used as an external cache. Many forms of RAM are available, although this is illustrative and not an exhaustive description.For example, these may include static random access memory (SRAM), synchronous static random access memory (SSRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDRSDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synclink dynamic random access memory (SLDRAM), and direct memory bus random access memory (DRRAM). The memories described in the embodiments of this disclosure are intended to include, but are not limited to, these and any other suitable types of memory.

[0196] In exemplary embodiments, embodiments of the present disclosure further provide a storage medium which is a computer storage medium, specifically a computer-readable storage medium. For example, a first memory 903 for storing a computer program may be provided, which can be executed by a first processor 902 of a first network function 900 to complete the steps of the first network function side method. Another example is a second memory 1003 for storing a computer program, which can be executed by a second processor 1002 of a network control device 1000 to complete the steps of the network control device side method. The computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, surface-mount memory, optical disc, or CD-ROM.

[0197] Note that terms like "1st," "2nd," etc., do not indicate a specific order or priority, but are simply used to distinguish similar objects.

[0198] The term "and / or" as used herein simply describes a related relationship that describes related objects, indicating that three relationships exist. For example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone. The term "at least one" as used herein means any combination of one or at least two of a plurality. For example, "including at least one of A, B, and C" can mean any one or more elements selected from the set consisting of A, B, and C.

[0199] Furthermore, the technical solutions described in the embodiments of this disclosure may be combined in any way if they do not conflict with each other.

[0200] The foregoing description is merely a preferred embodiment of the present disclosure and is not intended to limit the scope of protection of the present disclosure.

Claims

1. A federated learning method applied to a first network function, A federated learning method comprising sending a first request to a network control device responsible for controlling the connectivity of federated learning nodes, wherein the first request is used to request that the network connectivity between federated learning nodes be optimized.

2. Sending the first request to the network control device responsible for controlling the connection of the aforementioned federated learning nodes means The method according to claim 1, further comprising sending a first request to a network control device responsible for controlling the connection of federated learning nodes when the network connection quality between federated learning nodes is below a predetermined threshold.

3. The method according to claim 1, wherein the first request is accompanied by relevant parameters for optimizing the network connection.

4. If the federated learning node is a federated learning node within the carrier network, the network control device is a software-defined network SDN controller. The first requirement is, The Internet Protocol IP address of the federated learning node associated with the network connection to be optimized, The method according to claim 1, further comprising at least one of the following requirements: a quality of service (QoS) requirement for network connectivity between federated learning nodes.

5. If the federated learning node is an external node of the carrier network, the network control device is the corresponding connected node of the external network. The first requirement is, The Internet Protocol IP address of the federated learning node associated with the network connection to be optimized, The method according to claim 1, further comprising at least one of the following requirements: a quality of service (QoS) requirement for network connectivity between federated learning nodes.

6. When the federated learning node is a terminal UE, the network control device is a policy control function PCF. The first requirement is, UE identifier and The method according to claim 1, further comprising at least one of the UE bearer flow QoS requirements.

7. A federated learning method applicable to network control devices, The first network function receives a first request, which is used to request the optimization of network connections between federated learning nodes, and the network control device is responsible for controlling the connections of the federated learning nodes. A federated learning method, which includes optimizing a network connection based on a first received request.

8. The first request includes relevant parameters for optimizing the network connection, Here, the first requirement includes at least one of the IP addresses of federated learning nodes associated with the network connection to be optimized, and the QoS requirements for the network connection between the federated learning nodes, Or, The method according to claim 7, wherein the first requirement is accompanied by at least one of an identifier for a UE and a QoS requirement for the bearer flow of a UE.

9. If the federated learning node is a federated learning node within the carrier network, the network control device is the SDN controller. Optimizing the network connection based on the first request received is The method according to claim 7, comprising calculating an optimal route based on a first received request and transmitting the optimal route to an associated routing device, wherein the optimal route is used to transmit data between federated learning nodes.

10. If the federated learning node is a UE, the network control device is a PCF. Optimizing the network connection based on the first request received is The method according to claim 7, comprising transmitting policy control and charging PCC policies to the user plane function UPF via the session management function SMF.

11. The PCC policy is, Resource allocation priority, Guaranteed speed and Maximum speed and The method according to claim 10, comprising at least one of the maximum packet loss rate.

12. If the federated learning node is an external node of the carrier network, the network control device is the corresponding connected node of the external network. Optimizing the network connection based on the first request received is The method according to claim 7, comprising controlling a network control node corresponding to a corresponding connection node of the external network to configure related paths.

13. The method according to claim 7, further comprising deleting the received first request when the associative learning is completed.

14. A first network function including a first processor and a first memory for storing a computer program executable on the first processor, Herein, the first processor is configured to perform the steps of the method according to any one of claims 1 to 6 when executing the computer program, a first network function.

15. A network control device comprising a second processor and a second memory for storing computer programs executable on the second processor, Herein, the second processor is configured to perform the steps of the method according to any one of claims 7 to 13 when executing the computer program, a network control device.

Images