Communication method and apparatus

Abstract

The present application relates to the technical field of communications, and provides a communication method and apparatus, used for improving the efficiency of implementing federated learning by multiple nodes in a communication system. The method comprises: determining first indication information corresponding to at least one first apparatus, the first indication information being used for indicating time information and / or frequency information of a first apparatus sending machine learning intermediate data, and the first apparatus being a participation node and / or a candidate participation node of a machine learning task; and sending the first indication information.

Description

A communication method and apparatus

[0001] This application claims priority to Chinese Patent Application No. 202410875303.1, filed with the State Intellectual Property Office of China on June 29, 2024, entitled "A Communication Method and Apparatus", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of communication technology, and in particular to a communication method and apparatus. Background Technology

[0003] To address the increasing complexity, diversified business needs, and personalized service experiences of future communication networks, Artificial Intelligence (AI) can be introduced into communication networks to provide on-demand services and improve network resource utilization. For example, Network Data Analytics Function (NWDAF) can be used to analyze and process various types of network data, such as collecting network operation data from network function nodes, obtaining terminal and network-related statistical data from Operation Administration and Maintenance (OAM) systems, or acquiring application data from third-party Application Functions (AFs). This data can then be used for data training or model inference.

[0004] In communication systems, multiple network nodes (such as NWDAF) can be used to implement distributed machine learning (such as federated learning). Multiple nodes collaboratively model the data, improving the effectiveness of AI models. For example, in federated learning among multiple network nodes, the initiating node can request data collection from multiple participating nodes, which can then collect the data and send it sequentially to the initiating node. Therefore, improving the performance of distributed learning among network nodes is a pressing issue that needs to be addressed. Summary of the Invention

[0005] This application provides a communication method and apparatus for improving the performance of distributed learning across multiple nodes in a communication system.

[0006] Firstly, a communication method is provided, which can be executed by a communication device. The communication device can be a network device, a module (such as a chip, chip system, or circuit) within the network device, or a module or software capable of implementing all or part of the functions of the network device. For example, the communication device can be an initiating node or coordinating node for machine learning. The method includes: determining first indication information corresponding to at least one first device; wherein the first indication information is used to instruct the first device to send time information and / or frequency information of intermediate data for machine learning, and the first device is a participating node and / or a candidate participating node in the machine learning task; and sending the first indication information.

[0007] In the above embodiments, in scenarios where multiple nodes in a communication system perform distributed machine learning, the time and / or frequency information of data reported by participating nodes in machine learning can be flexibly determined, and the participating nodes can be instructed on the time and / or frequency information of data reporting. This enables fine management or control of the data transmitted by participating nodes, allowing them to report data according to the indicated time and / or frequency. This avoids model deviations caused by inaccurate data reporting or excessive data reporting that leads to excessive transmission resource consumption, thereby improving the performance of machine learning.

[0008] In one implementation, determining the first indication information includes: determining the first indication information based on attribute information corresponding to the first device, wherein the attribute information includes at least one of the following: location information corresponding to the first device, information of at least one network function (NF) associated with the first device, data type associated with the first device, or capability information of the first device; wherein the capability information includes at least one of the following: machine learning algorithms supported by the first device, data processing capabilities of the first device, or performance indicators of the first device.

[0009] In the above implementation, the time and / or frequency of data reporting by the participating node can be flexibly determined based on the node attribute information. The determination method is flexible and can meet the fine management or control of data reporting for different machine learning needs, thereby improving machine learning performance.

[0010] In one implementation, determining the first indication information based on the attribute information corresponding to the first device includes at least one of the following methods: determining the type information of the first device based on the location information of the first device, and determining the first indication information corresponding to the first device based on the type information of the first device; wherein the type information includes edge nodes or non-edge nodes; determining the first indication information corresponding to the first device based on the number of network functions (NFs) associated with the first device; or, determining the first indication information corresponding to the first device based on the federated learning type supported by the first device.

[0011] In the above embodiments, the communication device can flexibly determine the time and / or frequency of the data corresponding to different participating nodes based on their location information, the number of NFs managed by each participating node, or the federated learning type supported by each participating node. For example, based on location information, data in core locations is more important; therefore, participating nodes in core locations can be set to report data at a faster frequency, while those in peripheral locations can be set to report data at a slower frequency, thereby improving the machine learning effect. Alternatively, based on the number of NFs managed by a participating node, participating nodes with a larger number of NFs can be set to report data at a faster frequency, while those with a smaller number of NFs can be set to report data at a slower frequency, thereby achieving fine-grained management of data reporting, avoiding data congestion, and improving machine learning performance. Furthermore, based on the federated learning type supported by a participating node, participating nodes supporting vertical federated learning can be set to report data at a faster frequency, while those not supporting vertical federated learning can be set to report data at a slower frequency, thereby improving the efficiency of data reporting and enhancing machine learning performance.

[0012] In one embodiment, the method further includes: receiving attribute information corresponding to the first device.

[0013] In one implementation, attribute information is carried in network function registration request messages, network function update messages, status update messages, or context update messages. In the above implementation, the communication device can flexibly obtain attribute information corresponding to different participating nodes during the network function registration process, network function update process, status update process with network elements, or context update interaction with terminals. This allows the device to determine the time and / or frequency of data reported by the participating node based on the attribute information, enabling fine-grained management or control of the data transmitted by the participating nodes and improving machine learning performance.

[0014] In one implementation, the time information and / or frequency information includes at least one of the following: the time, time interval, time range, transmission rate, transmission frequency, or indicator information indicating the speed of transmission when the first device sends intermediate data for machine learning to the second device, wherein the second device is the initiating node of the machine learning.

[0015] In the above embodiments, the time information and / or frequency information of the data reported by the participating nodes determined by the communication device are flexible in form and are not limited to a specific indication method. The indication method can be flexibly configured according to the needs of machine learning, which facilitates the fine management or control of the data reported by the participating nodes in various scenarios or with different needs, thereby improving the performance of machine learning.

[0016] In one embodiment, the method further includes: sending a first message to a third device, including the requirements of the machine learning task, wherein the first message is used to obtain information on candidate participating nodes required by the machine learning task. In the above embodiments, candidate participating nodes that meet the requirements of the current machine learning task can be selected, thereby improving the accuracy of the data source for machine learning and enhancing learning performance.

[0017] In one embodiment, the method is applied to a coordinating node of the machine learning task, wherein the aforementioned communication device is the coordinating node, and the method further includes: determining at least one candidate participating node of the machine learning task based on the requirements of the machine learning task, capability information and / or contract information of at least one network node; and sending information of the at least one candidate participating node to a second device; wherein the at least one candidate participating node includes the first device, and the second device is the initiating node of the machine learning task.

[0018] In the above embodiments, the coordinating node of the machine learning task can be used to determine the time and / or frequency of data reporting by participating nodes. Specifically, the coordinating node can select candidate participating nodes that meet the requirements of the machine learning task, the node's capability information and / or contract information, and send the information of the selected candidate participating nodes to the initiating node. Thus, the coordinating node achieves data isolation, meets the requirements of data transmission security in the communication system, and improves the performance of machine learning.

[0019] In one embodiment, the method is applied to a coordination node of the machine learning task, wherein the aforementioned communication device is the coordination node. The method further includes receiving a second message, the second message being used to request first indication information corresponding to each participating node in the machine learning task. In the above embodiment, after the coordination node determines the candidate participating nodes, it can determine the time and / or frequency of the data reported by the participating nodes requested by the initiating node according to the second message, thereby reducing unnecessary data processing (such as determining the time and / or frequency of data reported by unnecessary candidate participating nodes) and transmission overhead (transmitting the aforementioned unnecessary information), and improving the performance of machine learning.

[0020] In one embodiment, the method is applied to the initiating node of the machine learning task, wherein the aforementioned communication device is the initiating node, and the determination of the first indication information includes: receiving second indication information from a fourth device, wherein the second indication information is used to instruct at least one participating node and / or candidate participating node of the machine learning task to send time information and / or frequency information of the intermediate data of the machine learning; wherein the fourth device is the coordinating node of the machine learning task; and determining the first indication information corresponding to the first device based on the second indication information.

[0021] In the above embodiments, the initiating node of the machine learning task can be used to determine the time and / or frequency of data reported by participating nodes based on the second indication information determined by the coordinating node. That is, a two-level time indication can be used: the coordinating node first determines the first-level time indication (such as the second indication information), and then the initiating node further determines the second-level time indication (such as the first indication information) based on the second indication information, thereby obtaining the final time and / or frequency of data reported by the participating node. This achieves flexible configuration of time indications, with flexible indication methods and reduced indication signaling overhead, thus improving the performance of machine learning.

[0022] In one embodiment, the method further includes: determining third indication information corresponding to at least one participating node in the machine learning task, the third indication information being used to indicate the time information and / or frequency information for updating the participating node to send intermediate data for machine learning; and sending the third indication information.

[0023] In the above embodiments, considering that the demand for data reported by participating nodes may be updated during the machine learning process, the communication device can flexibly update the time and / or frequency of the data reported by a certain participating node through the third indication information, thereby realizing fine management or control of the data transmitted by the participating nodes and improving the performance of machine learning.

[0024] Secondly, a communication method is provided, which can be executed by a first device. The first device can be a network device, a module (such as a chip, chip system, or circuit) within the network device, or a module or software capable of implementing all or part of the functions of the network device. Alternatively, the first device can be a terminal, a module (such as a chip, chip system, or circuit) on the terminal, or a module or software capable of implementing all or part of the terminal's functions. The method includes: receiving first indication information, wherein the first indication information is used to instruct the first device to send time information and / or frequency information of intermediate data for machine learning; the first device is a participating node in the machine learning task; and sending the intermediate data for machine learning according to the time information and / or frequency information.

[0025] In one implementation, the time information and / or frequency information includes at least one of the following: the time, time interval, time range, transmission rate, transmission frequency, or indicator information indicating the speed of transmission when the first device sends intermediate data for machine learning to the second device, wherein the second device is the initiating node of the machine learning.

[0026] In one embodiment, the method includes: sending attribute information of the first device to a third device, the attribute information including at least one of the following: location information corresponding to the first device, information of at least one network function (NF) associated with the first device, data type associated with the first device, or capability information of the first device; wherein the capability information includes at least one of the following: machine learning algorithms supported by the first device, data processing capabilities of the first device, or performance indicators of the first device.

[0027] In one implementation, attribute information is carried in a network function registration request message, a network function update message, a status update message, or a context update message.

[0028] In one embodiment, the method further includes: receiving third indication information, the third indication information being used to indicate updating the time information and / or frequency information of the participating nodes sending intermediate machine learning data; and sending the intermediate machine learning data according to the updated time information and / or frequency information.

[0029] Thirdly, a communication device is provided for implementing the above-described method. This communication device may be an apparatus for performing the method of the first or second aspect, or a node or device containing the aforementioned apparatus, or a module within the aforementioned apparatus, such as a chip, chip system, or circuit, or a logic node, logic module, or software capable of implementing some or all of the functions.

[0030] The apparatus includes modules, units, or means that implement the methods described above. These modules, units, or means can be implemented in hardware, software, or by hardware executing corresponding software. The hardware or software includes one or more modules or units corresponding to the functions described above.

[0031] In one possible implementation, the device may include a processing module and a transceiver module. The processing module can be used to implement the processing functions in any of the above aspects and any possible implementations thereof. The processing module may be, for example, a processor. The transceiver module, also referred to as a transceiver unit, is used to implement the sending and / or receiving functions in any of the above aspects and any possible implementations thereof. The transceiver module may include: transceiver circuitry, transceiver, transceiver unit, or communication interface.

[0032] In one possible implementation, the transceiver module includes a sending module and a receiving module, which are used to implement the sending and receiving functions in any of the above aspects and any possible implementations.

[0033] Fourthly, a communication device is provided, comprising: a processor; the processor being coupled to a memory and, after reading instructions from the memory, executing the method as described in any of the preceding aspects according to the instructions. The communication device may be a device as described in the first or second aspect, or a node or device containing the aforementioned device, or a module of the aforementioned device, such as a chip, chip system, or circuit, or a logic node, logic module, or software capable of implementing some or all of the functions.

[0034] In one possible implementation, the communication device further includes a memory for storing program instructions and / or data. Optionally, the memory and processor are integrated together.

[0035] In one possible implementation, the communication device is a chip or a chip system. Optionally, when the communication device is a chip system, it can be composed of chips or may include chips and other discrete components.

[0036] Fifthly, a communication device is provided, comprising: a processor and an interface circuit; the interface circuit is configured to receive a computer program or instructions and transmit them to the processor; the processor is configured to execute the computer program or instructions to cause the communication device to perform the method described in any of the preceding aspects. The communication device may be a device as described in the first or second aspect, or a node or device containing the aforementioned device, or a module of the aforementioned device, such as a chip, chip system, or circuit, or a logic node, logic module, or software capable of implementing some or all of the functions.

[0037] In one possible implementation, the communication device is a chip or a chip system. Optionally, when the communication device is a chip system, it can be composed of chips or may include chips and other discrete components.

[0038] In a sixth aspect, a computer-readable storage medium is provided that stores instructions which, when executed on a computer, cause the computer to perform the methods described in any of the preceding aspects.

[0039] In a seventh aspect, a computer program product containing instructions is provided, which, when run on a computer, enables the computer to perform the methods described in any of the preceding aspects.

[0040] Eighthly, a communication system is provided, comprising a communication device performing the method as described in the first aspect, and a first device performing the method as described in the second aspect. Exemplarily, the communication device performing the method as described in the first aspect may be a machine learning initiating node or a coordinating node.

[0041] In one possible implementation, the communication system includes a coordinating node, an initiating node, and a first means for performing the method described in the first aspect above, as well as the method described in the second aspect above.

[0042] The technical effects of any of the possible implementations of aspects two through eight can be found in the technical effects of the different possible implementations of aspect one above, and will not be repeated here.

[0043] Understandably, provided that the solutions do not contradict each other, the solutions in the above aspects can be combined. Attached Figure Description

[0044] Figure 1 is a schematic diagram of the architecture of a communication system provided in an embodiment of this application;

[0045] Figure 2 is a schematic diagram of the interaction process of a data analysis service provided in an embodiment of this application;

[0046] Figure 3 is a schematic diagram of the architecture of a communication device provided in an embodiment of this application;

[0047] Figure 4 is a flowchart illustrating a communication method provided in an embodiment of this application;

[0048] Figure 5 is a schematic diagram of a machine learning architecture provided in an embodiment of this application;

[0049] Figure 6 is a flowchart illustrating another communication method provided in an embodiment of this application;

[0050] Figure 7 is a flowchart illustrating another communication method provided in an embodiment of this application;

[0051] Figure 8 is a schematic diagram of the structure of a communication device provided in an embodiment of this application. Detailed Implementation

[0052] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this embodiment, unless otherwise stated, "a plurality of" means two or more.

[0053] It should be noted that, in this application, the terms "exemplary" or "for example" are used to indicate that something is being described as an example, illustration, or illustration. Any embodiment or design described as "exemplary" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design solutions. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0054] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0055] First, a brief introduction will be given to the implementation environment and application scenarios of the embodiments of this application.

[0056] The communication method provided in this application can be applied to fourth-generation (4G) communication systems, such as long-term evolution (LTE) systems, as well as fifth-generation (5G) communication systems, such as 5G new radio (NR) systems, or various communication systems evolving after 5G, and future communication systems. Figure 1 illustrates a network architecture to which this application applies. Specifically, Figure 1 uses the network service architecture of a fifth-generation (5G) mobile communication system as an example to illustrate the interaction relationships between network functions (NFs) and entities, as well as the corresponding interfaces.The 3rd generation partnership project (3GPP) service-based architecture (SBA) for 5G systems includes the following network functions and entities: user equipment (UE), at least one access network (AN) or radio access network (RAN) node, user plane function (UPF), data network (DN), access and mobility management function (AMF), session management function (SMF), policy control function (PCF), application function (AF), unified data management (UDM), network exposure function (NEF), unified data repository (UDR), authentication server function (AUSF), network repository function (NRF), network slice selection function (NSSF), and network slice specific authentication and authorization (NSSAAF). Functions such as Network Data Analytics Function (NWDAF) and Network Data Analytics Function (NWDAF) are also included.

[0057] In this context, the UE, radio access network (RAN) node, UPF, and DN are generally referred to as user plane network functions and entities (or user plane network elements), while the others are generally referred to as control plane network functions and entities (or control plane network elements). Control plane network elements, defined by 3GPP, define the processing functions within a network. They possess 3GPP-defined functional behaviors and interfaces. An NF can function as a network element running on proprietary hardware, a software instance running on proprietary hardware, or a virtual function instantiated on a suitable platform, such as a cloud infrastructure.

[0058] The main functions of each network function are described in detail below.

[0059] The user plane network functions in the communication system include:

[0060] (R)AN Node: A (R)AN can be an AN, a RAN, or an access network device, RAN entity, or access node, etc., forming part of the communication system to help terminal devices access the communication network. For example, a (R)AN can be various types of base stations, such as macro base stations, micro base stations, radio controllers, relay stations, access points, or network equipment in vehicle-mounted devices, wearable devices, or future Public Land Mobile Networks (PLMNs). The (R)AN is primarily responsible for air interface-side radio resource management, quality of service management, data compression, and encryption.

[0061] In addition, (R)AN nodes can also be access nodes in open RAN (O-RAN or ORAN), cloud radio access network (CRAN), or wireless fidelity (WiFi) systems, or access nodes in communication systems that integrate two or more of the above systems.

[0062] In one possible scenario, a RAN node can be a base station, an evolved NodeB (eNodeB), an access point (AP), a transmission reception point (TRP), a next-generation NodeB (gNB) in a future mobile communication system, or an access node in a WiFi system. A RAN node can be a macro base station, a micro base station, an indoor station, a relay node, a donor node, or a radio controller in a CRAN scenario. Optionally, a RAN node can also be a server, a wearable device, a vehicle, or in-vehicle equipment. For example, the access network equipment in vehicle-to-everything (V2X) technology can be a roadside unit (RSU). All or part of the functions of the RAN node in this application can also be implemented through software functions running on hardware, or through virtualization functions instantiated on a platform (e.g., a cloud platform). The RAN node in this application can also be a logical node, logical module, or software capable of implementing all or part of the RAN node functions.

[0063] In another possible scenario, multiple RAN nodes collaborate to assist terminal devices in achieving wireless access, with different RAN nodes each implementing some of the base station's functions. For example, RAN nodes can be central units (CUs), distributed units (DUs), CU-control plane (CPs), CU-user plane (UPs), or radio units (RUs), etc. CUs and DUs can be set up separately or included in the same network element, such as a baseband unit (BBU). RUs can be included in radio frequency equipment or radio frequency units, such as remote radio units (RRUs), active antenna units (AAUs), or remote radio heads (RRHs).

[0064] In different systems, CU (or CU-CP and CU-UP), DU, or RU may have different names, but those skilled in the art will understand their meaning. For example, in an ORAN system, CU can also be called O-CU (open CU), DU can also be called O-DU, CU-CP can also be called O-CU-CP, CU-UP can also be called O-CU-UP, and RU can also be called O-RU. For ease of description, this application uses CU, CU-CP, CU-UP, DU, and RU as examples. Any of the units among CU (or CU-CP, CU-UP), DU, and RU in this application can be implemented through software modules, hardware modules, or a combination of software and hardware modules.

[0065] UE: Also known as a terminal, terminal device, mobile station (MS), mobile terminal (MT), etc., it is a device used to provide voice or data connectivity to users, or an Internet of Things (IoT) device. For example, terminal devices include handheld devices with wireless connectivity, in-vehicle devices, etc. Currently, terminal devices can include: mobile phones, tablets, laptops, PDAs, mobile internet devices (MIDs), wearable devices (such as smartwatches, smart bracelets, pedometers, smart glasses, etc.), in-vehicle devices (such as cars, bicycles, electric vehicles, airplanes, ships, trains, high-speed trains, etc.), satellite terminals, virtual reality (VR) devices, augmented reality (AR) devices, point-of-sale (POS) machines, customer-premises equipment (CPE), light user equipment (UE), reduced capability user equipment (REDCAP UE), wireless terminals in industrial control, smart home devices (such as refrigerators, televisions, air conditioners, electricity meters, etc.), intelligent robots, robotic arms, workshop equipment, wireless terminals in autonomous driving, wireless terminals in telemedicine, wireless terminals in smart grids, wireless terminals in transportation safety, wireless terminals in smart cities, or wireless terminals in smart homes, and flying equipment (such as intelligent robots, hot air balloons, drones, airplanes), etc. Terminal devices can also be vehicle devices, such as vehicle devices, vehicle modules, vehicle chips, on-board units (OBUs) or telematics boxes (T-BOXs). Terminal devices can also be other devices with terminal functions. For example, a terminal device can also be a device that performs terminal functions in D2D communication.

[0066] The embodiments of this application do not limit the form of the terminal device. The device used to implement the functions of the terminal device can be the terminal device itself, or it can be a device that supports the terminal device in implementing the functions, such as a chip system. The device can be installed in the terminal device or used in conjunction with the terminal device. In the embodiments of this application, the chip system can be composed of chips, or it can include chips and other discrete devices. All or part of the functions of the terminal device in this application can also be implemented by software functions running on hardware, or by virtualization functions instantiated on a platform (such as a cloud platform).

[0067] UPF: Primarily responsible for forwarding and receiving user plane data. The UPF can receive downlink data from the DN and then transmit that downlink data to the UE via (R)AN. The UPF can also receive uplink data from the UE via (R)AN and then forward that uplink data to the DN.

[0068] DN: For example, DN can be a carrier service network, Internet access, or a third-party service network. DN can exchange information with UE through PDU sessions. PDU sessions can be of various types, such as Internet Protocol version 4 (IPv4) and IPv6.

[0069] In addition, the control plane network functions in the communication system include:

[0070] AMF (Automatic Management Function): Primarily responsible for processing control plane messages and managing the mobility of terminal devices, including mobility state management, assigning temporary user identities, and authenticating and authorizing users. Examples include access control, mobility management, registration and deregistration, and network element selection.

[0071] SMF: Primarily used for session management, session establishment, allocation and management of UE's (private) IP address, responsible for session establishment, modification and release, and quality of service (QoS) control, etc.

[0072] UDM (User Authentication and Authorization Manager): Primarily used for authentication and credit processing, it manages subscription data, user identification, access authorization, registration / mobility management, subscription management, and SMS management. For example, when a user's subscription data is modified, the UDM is responsible for notifying the relevant network elements.

[0073] NEF: Primarily used to provide corresponding security guarantees to ensure the security of external applications to the communication network, providing functions such as opening up QoS customization capabilities for external applications, subscription to mobility state events, and distribution of AF requests.

[0074] NRF: Primarily used to provide internal / external addressing functions, etc.

[0075] AUSF: Primarily used for authentication processing functions, enabling two-way authentication between terminals and networks.

[0076] AF: Primarily used to send data routing information affecting applications to the network side, and to perform policy control through interaction between network open function elements and the policy framework.

[0077] NSSAAF: Primarily responsible for network slice authentication and authorization, it can interact with the authentication, authorization, and accounting server (AAA-S) through the authentication, authorization, and accounting proxy (AAA-P).

[0078] NWDAF (Network Data Analyzer): Primarily used for analyzing various types of network data. This network data can specifically include: network operation data collected from NF (Network Functions), terminal and network-related statistical data obtained from OAM (Operating Analysis Center), application data obtained from third-party AF (Application Analyzer), and terminal and / or user information obtained from terminals. Subsequently, NWDAF can feed back the generated analysis results to nodes such as NF, OAM, or third-party AF, allowing NF, OAM, or AF to perform various optimization operations using the NWDAF analysis results.

[0079] For example, the specific steps of NWDAF may include several processes such as requesting data analysis, subscribing to analysis, collecting data, and providing feedback on analysis results. The following is a brief introduction to the NWDAF workflow, using the interactive diagram in Figure 2 as an example.

[0080] As shown in Figure 2, the process by which a consumer of the NWDAF service requests data analysis from NWDAF may include the following steps 1-2.

[0081] Step 1: The consumer sends a data analysis request (such as Nnwdaf_AnalyticsInfo_Request) message to NWDAF to request data analysis services from NWDAF.

[0082] Alternatively, in one implementation, data analysis services can be implemented through a subscription service. For example, the data analysis request message in step 1 can be a subscription request message, such as the Nnwdaf_AnalyticsSubscription_Subscribe message.

[0083] The parameters carried in the data analysis request message may include:

[0084] One or more data analysis identifiers (Analytics IDs): used to define the type of analysis requested, such as network slicing analysis, NF load analysis, etc.

[0085] Analytics Filter Information: Indicates the analytical information that needs to be reported, such as Single Network Slice Selection Assistance Information (S-NSSAI), NF identifiers, etc., to narrow down the scope of analytical data.

[0086] Target of Analytics Reporting: Indicates the objectives of the data analysis.

[0087] Notification Target Address: For subscription services, this can include the notification address for related events. For example, NF-1 subscribes to the data analysis service from NWDAF, indicating that the analysis results corresponding to the data analysis can be notified to NF-2.

[0088] Analytics Reporting Information: Indicates the analysis results corresponding to the requested data analysis.

[0089] Step 2: NWDAF sends a data analysis response (such as Nnwdaf_AnalyticsInfo_Request Reponse) message to the consumer to provide feedback on the analysis results.

[0090] For example, for a subscription service, the data analytics response message can specifically be an Nnwdaf_AnalyticsSubscription_Notify message.

[0091] Subsequently, NWDAF can collect data from other nodes in the communication network based on the parameters in the data analysis request, such as NF, OAM, AF, and terminals.

[0092] For example, as shown in Figure 2, the NWDAF data collection process may include steps 3-4. In the following examples, the nodes for which NWDAF requests data collection can be collectively referred to as data nodes. This application does not limit the type of data nodes; any node in the communication system that can provide the data required for NWDAF data analysis can be called a data node.

[0093] Step 3: NWDAF sends a data request message to the data node.

[0094] If the data node can be an NF, NWDAF sends a data request message to the NF, such as an Nnf_EventExposure_Subscribe message, to subscribe to data from the NF. Optionally, the data request message may include subscription parameters required by NWDAF to subscribe to data.

[0095] Optionally, for subscription services, step 3-1 may also be included:

[0096] Step 3-1: The data node notifies NWDAF whether the data subscription was successful or failed.

[0097] Optionally, for subscription services, step 3-2 may also be included:

[0098] Step 3-2: Data nodes prepare data.

[0099] Step 4: The data node sends the collected data to the NWDAF.

[0100] For example, NF can return subscription data by sending an Nnf_EventExposure_Notify message.

[0101] Optionally, this message can also be used to notify NWDAF that the subscription was successful.

[0102] Alternatively, NF can report data to NWDAF at regular intervals or after collecting a certain amount of data, based on the subscription parameters in the data request message.

[0103] For example, NWDAF can collect data via the File Transfer Protocol (FTP).

[0104] In one application scenario, through the aforementioned interaction process, 5G mobility management-related functions can request NWDAF to predict the terminal's movement trajectory. For example, NWDAF can request OAM to collect data to obtain the terminal's historical location information. The NWDAF can then analyze the collected data to generate a mobility prediction model for the terminal. Subsequently, NWDAF can provide the 5G mobility management-related functions with the terminal's mobility prediction information (such as the mobility prediction model) based on the terminal's current location, enabling the 5G mobility management-related functions to formulate more accurate network policies and complete optimized mobility management operations. Examples include registration area allocation based on statistical patterns of terminal location, handover decision-making assistance based on terminal location prediction information, and mobility anchor point selection based on the terminal's mobility trajectory.

[0105] However, it is difficult for a single NWDAF to collect data distributed across different regions. Multiple NWDAFs are usually deployed in a communication network. Therefore, data collection and analysis can be achieved through distributed processing. For example, an NWDAF distributed in one region can collect data for that region, and subsequently, this NWDAF can share its model or data analysis with other NWDAFs.

[0106] It should be understood that the embodiments of this application do not limit the network nodes that perform data collection and / or data analysis, and NWDAF is used as an example in the above example.

[0107] In one implementation, multiple network nodes (such as NWDAF) can perform distributed machine learning, such as federated learning (FL). Multiple nodes collaboratively model the data, improving the effectiveness of the AI ​​model. For example, in federated learning among multiple network nodes, the initiating node can request data collection from multiple participating nodes, which can then collect the data and send it sequentially to the initiating node. Therefore, improving the efficiency and effectiveness of federated learning among network nodes is a pressing issue that needs to be addressed.

[0108] It should be noted that the functions of the other network elements included in Figure 1 can be found in the relevant descriptions in conventional technologies, and will not be repeated here. The network architecture shown in Figure 1 is for illustrative purposes only and is not intended to limit the technical solutions of this application. Those skilled in the art should understand that in specific implementations, other network elements or devices may be included, and the number of access network devices, terminal devices, and / or core network devices can be determined according to specific needs.

[0109] Optionally, each network element shown in Figure 1 can be a device, a functional module within a device, or a logical functional unit. It is understood that the above functions can be network components in hardware devices, such as communication chips in mobile phones, software functions running on dedicated hardware, or virtualization functions instantiated on a platform (e.g., a cloud platform).

[0110] It is understood that devices or network elements in the communication systems shown in Figure 1 or Figure 2 can communicate directly or through forwarding by other devices. This application does not specifically limit this.

[0111] It is understood that Figure 1 or Figure 2 above are merely schematic diagrams and do not constitute a limitation on the applicable scenarios of the technical solutions provided in this application. Those skilled in the art should understand that in specific implementation processes, the communication system may include fewer devices or network elements than those shown in Figure 1 or Figure 2, or the communication system may also include other devices or other network elements, and the number of devices or network elements in the communication system can be determined according to specific needs.

[0112] For example, each network element in Figure 1 or Figure 2 can be implemented using the communication device 300 in Figure 3. Figure 3 shows a schematic diagram of the hardware structure of a communication device applicable to embodiments of this application. The communication device 300 includes at least one processor 301, a communication line 302, and at least one communication interface 304. Optionally, the communication device 300 may also include a memory 303.

[0113] Processor 301 can be one or more Central Processing Units (CPUs). If the processor is a CPU, it can be a single-core CPU or a multi-core CPU. The processor can be a general-purpose processor, digital signal processor, application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware component. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. A general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as being executed by a hardware processor, or executed by a combination of hardware and software modules within the processor.

[0114] Optionally, the processor may include one or more of the following: a central processing unit (CPU), an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a microprocessor unit (MPU), a microcontroller unit (MCU), a graphics processing unit (GPU), a field-programmable gate array (FPGA), an artificial intelligence processor (AI processor), or a neural processing unit (NPU).

[0115] Communication line 302 may include a path for transmitting information between the aforementioned components, such as a bus.

[0116] Communication interface 304 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet interface, RAN interface, wireless local area network (WLAN) interface, etc.

[0117] The memory 303 may include, but is not limited to, cache, read-only memory (ROM), random access memory (RAM), synchronous dynamic random access memory (SDRAM), hard disk drive (HDD) or solid-state drive (SSD), erasable programmable read-only memory (EPROM), or compact disc read-only memory (CD-ROM), etc. Memory is any other medium capable of carrying or storing desired program code in the form of instructions or data structures that can be accessed by a computer, but is not limited to this. The memory in the embodiments of this application may also be a circuit or any other device capable of implementing storage functions for storing computer programs or instructions, and / or data. Optionally, the memory may exist independently and be connected to the processor via communication line 302. Optionally, the memory may also be integrated with the processor. The memory provided in the embodiments of this application can generally be non-volatile. The memory 303 stores computer execution instructions involved in the present application, and the processor 301 controls the execution of these instructions. The processor 301 executes the computer execution instructions stored in the memory 303 to implement the method provided in the embodiments of the present application.

[0118] Optionally, the computer execution instructions in the embodiments of this application may also be referred to as application code, and the embodiments of this application do not specifically limit this.

[0119] In a specific implementation, as one embodiment, processor 301 may include one or more CPUs, such as CPU0 and CPU1 in FIG3.

[0120] In a specific implementation, as one embodiment, the communication device 300 may include multiple processors, such as processor 301 and processor 307 in FIG. 3. Each of these processors may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0121] In a specific implementation, as one embodiment, the communication device 300 may further include an output device 305 and an input device 306. The output device 305 communicates with the processor 301 and can display information in various ways. For example, the output device 305 may be a liquid crystal display (LCD), a light-emitting diode (LED) display device, a cathode ray tube (CRT) display device, or a projector, etc. The input device 306 communicates with the processor 301 and can receive user input in various ways. For example, the input device 306 may be a mouse, keyboard, touchscreen device, or sensing device, etc.

[0122] The communication device 300 described above can be a general-purpose device or a dedicated device. In specific implementations, the communication device 300 can be a portable computer, a web server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, an embedded device, or a device with a similar structure to that shown in Figure 3. This application does not limit the type of communication device 300.

[0123] The communication method provided in the embodiments of this application will be described in detail below.

[0124] It should be noted that the message names between network elements or the names of parameters in the messages in the following embodiments of this application are just examples. Other names may be used in the specific implementation. This application does not limit them in this respect.

[0125] It is understood that some or all of the steps in the embodiments of this application are merely examples, and other steps or variations thereof may also be performed in the embodiments of this application. Furthermore, the steps may be performed in different orders as presented in the embodiments of this application, and it is not necessary to perform all the steps in the embodiments of this application.

[0126] This application provides a communication method that determines the time information or feedback frequency of feedback data corresponding to multiple participating nodes in a machine learning task, and instructs one or more participating nodes to do so. This allows participating nodes to report data to the initiating node of the machine learning task at a specified time or frequency, preventing offsets caused by asynchronous data feedback, improving the correctness and effectiveness of model training or updates, and thus improving the performance of machine learning.

[0127] For example, in machine learning among multiple network nodes, the initiating node can request data collection from multiple participating nodes, which can then collect data and send it sequentially to the initiating node. However, data from a remote area may have geographical limitations, and frequent reporting of data from that region could significantly skew the machine learning results, leading to a decline in performance. The implementation method provided in this application allows the initiating node to finely manage or control the data reporting behavior of multiple participating nodes, thereby improving machine learning performance.

[0128] As shown in Figure 4, this application provides a communication method that may include the following steps.

[0129] 401: The communication device determines first indication information corresponding to at least one first device.

[0130] The first device is a participating node and / or a candidate participating node for the machine learning task.

[0131] In one implementation, the communication device can be an initiating node or a coordinating node for machine learning. For example, the initiating node or participating node can determine the participating nodes required for a machine learning task from a plurality of candidate participating nodes. For instance, the initiating node or participating node can select one or more participating nodes from a plurality of subsequent participating nodes to provide the intermediate data required for the current machine learning task.

[0132] In one implementation, as shown in Figure 5, a machine learning architecture in a communication network may include: a machine learning initiating node and one or more machine learning participating nodes. The initiating node initiates machine learning to the participating nodes, and the participating nodes collect data according to the needs of the machine learning and send the collected data or intermediate data to the initiating node.

[0133] Intermediate data may include locally trained models, model parameters, weight data, or gradient data.

[0134] For example, as shown in Figure 5, the initiating node of machine learning can be the central NWDAF, also known as the server NWDAF. The participating nodes of machine learning can be client NWDAFs in different regions, such as NWDAF-1 in region 1, NWDAF-2 in region 2, ... and NWDAF-n in region n, etc.

[0135] In this context, the client-side NWDAF in each region can collect data from the nodes associated with or managed in this region, such as from nodes like NF, external AF, OAM, or terminals.

[0136] For example, machine learning can specifically be federated learning.

[0137] Federated learning, also known as federated machine learning, is a potential solution for addressing issues such as data privacy and security, and model training. Based on the type of data, federated learning can be categorized into horizontal federated learning, vertical federated learning (VFL), and federated transfer learning.

[0138] Among them, horizontal federated learning is suitable for scenarios where data features have high overlap and samples have little overlap, such as data from banks or hospitals in different regions. Their businesses are similar (i.e., the data features are similar), but their users are different (the data samples are different).

[0139] Vertical federated learning, also known as longitudinal federated learning, is suitable for scenarios where there is a lot of sample overlap but little overlap in data features. For example, data from hospitals, banks, or insurance institutions in the same region may be similar because they serve users in a large area. However, pharmacies and banks have different businesses, resulting in data with different characteristics.

[0140] Furthermore, in vertical federated learning, the local datasets in different VFL participating nodes used for local model training have different feature spaces for the same samples (e.g., the same user). Considering data security, nodes can avoid transmitting raw data and instead share local models or intermediate data, thereby improving data security and reducing the risk of data leakage. For example, in a multi-level NWDAF architecture, the NWDAF may co-locate with NFs (e.g., UPF, SMF), and raw data cannot be shared due to information security and performance reasons. In this case, federated learning can be used, where multiple local NWDAFs collect data within the region and train models. The local NWDAFs can then share the obtained models to the central NWDAF, which performs model aggregation to obtain the global model, optimal model, or model parameters, and sends them to client NWDAFs for data inference and other data analysis tasks.

[0141] The model aggregation method refers to the process whereby, after the initiating node receives models from several participating nodes (such as NWDAFs in various regions), it can combine these models to generate a single global model. This application does not limit the specific implementation of model aggregation; however, any of the following aggregation models can be used as examples:

[0142] Simple averaging: This method averages the models of all participating nodes. For example, for models with the same structure, the global model can be obtained by averaging the model parameters. For instance, if the initiating node receives Model 1, Model 2, and Model 3 from three participating nodes, it can obtain a global model through simple averaging. The model parameters of this global model can be the average of the model parameters of the three models mentioned above.

[0143] Weighted averaging: Before averaging the models, a weighted average can be performed based on the quality of the models or the amount of training data. For example, according to pre-configured rules, weights corresponding to each model can be labeled, and the average can be multiplied by the weight coefficients to obtain the global model.

[0144] Federated average: An algorithm for calculating weighted averages. It obtains a weighted average by weighting data points according to a certain weight. For example, at the beginning of each round of model aggregation, a random small proportion of participating nodes are selected. Then, the initiating node can send the current global algorithm state (current model parameters) to each selected participating node. After repeating the selection process for multiple rounds, the aggregated global model is obtained by averaging the results using a pre-configured algorithm.

[0145] Hybrid approach: Combining multiple model aggregation techniques mentioned above.

[0146] The first indication information is used to instruct the first device to send intermediate machine learning data at specific times and / or at varying frequencies. In other words, the communication device can determine the timing and / or frequency of intermediate machine learning data sent by multiple participating nodes and / or candidate participating nodes to the machine learning initiating node.

[0147] It should be understood that the first indication information corresponding to different participating nodes can be different; that is, the time or frequency at which different participating nodes report intermediate data can be different. The communication device can determine the first indication information corresponding to each of the multiple participating nodes separately.

[0148] In one embodiment, the time information and / or frequency information indicated by the first indication information may include at least one of the following: the time, time interval, time range, transmission rate, transmission frequency, or indicator information indicating the speed of transmission when the first device transmits intermediate data for machine learning.

[0149] For example, the first indication information can indicate the specific time when the first device sends intermediate data, such as indicating T1, T2, etc.; or, it can indicate the time range for sending data, such as indicating that data is sent within T3 to T4; or, it can indicate the time interval for sending data, such as sending intermediate data once every 3 seconds, or sending intermediate data once every 3 seconds; or, it can indicate the frequency of sending data, such as indicating that intermediate data is sent five times per minute; or, it can indicate the data transmission rate, such as indicating that 100 kbit / s means 100 kilobits of data code are transmitted per second; or, it can be indicated by predefined indicators of transmission frequency, such as predefined transmission frequencies as fast, relatively fast, relatively slow, slow, etc., and each indicator corresponding to a specific time indication and / or frequency indication, so that one of the indicators can be indicated by the first indication information to indicate the time indication and / or frequency indication corresponding to the data transmission, saving indication signaling overhead.

[0150] The embodiments of this application do not limit the specific form of the first instruction information.

[0151] In one possible implementation, when the communication device is the initiating node, determining the first indication information includes: receiving second indication information from the fourth device, and determining the first indication information based on the second indication information. The second indication information is used to instruct at least one participating node and / or candidate participating node of the machine learning task to send time information and / or frequency information of the intermediate data of the machine learning, and the fourth device is the coordinating node of the machine learning task.

[0152] 402: The communication device sends the first instruction information.

[0153] Specifically, the communication device can send the first instruction information corresponding to the first device to the first device respectively.

[0154] Correspondingly, the first device receives the first instruction information and can send intermediate machine learning data based on time information and / or frequency information.

[0155] Through the above implementation method, in a scenario where multiple nodes in a communication system perform distributed machine learning, the time and / or frequency information of the data reported by the participating nodes can be flexibly determined according to the needs of machine learning. The participating nodes are instructed on the time and / or frequency information of the data to be reported, so that the participating nodes can report data according to the indicated time and / or frequency, avoiding model deviation caused by inappropriate data reporting, or transmission resource occupation caused by unsuitable data reporting, thereby improving the performance of machine learning.

[0156] In one embodiment, the communication device determines the first indication information, which may specifically include: the communication device determining the first indication information based on the attribute information corresponding to the first device.

[0157] The attribute information corresponding to the first device includes at least one of the following: the location information corresponding to the first device, the information of at least one network function (NF) associated with the first device, the data type associated with the first device, or the capability information of the first device.

[0158] The capability information includes at least one of the following: the machine learning algorithms supported by the first device, the data processing capabilities of the first device, or the performance metrics of the first device. For example, the machine learning algorithms may include different types of federated learning, such as vertical federated learning, horizontal federated learning, or federated transfer learning. The data processing capabilities of the first device may include metrics or parameters characterizing its data processing strength, such as the performance of the Central Processing Unit (CPU), such as CPU clock speed, number of cores, and cache size. The performance metrics of the first device may include metrics or parameters characterizing its transmission performance, such as bandwidth, packet loss rate, throughput, and frame rate.

[0159] For example, the communication device can determine the type information of the first device based on the location information of the first device, and determine the first indication information corresponding to the first device based on the type information of the first device. The type information can refer to the node type determined based on the location information, such as including edge nodes or non-edge nodes, or including center nodes or non-center nodes (or edge nodes).

[0160] It should be noted that, in the embodiments of this application, the node type information, including edge nodes, is only an example. For instance, an edge node can refer to a Mobile Edge Computing (MEC) node; or, edge nodes are only a general classification of node types. For example, nodes can be divided into core nodes (i.e., non-edge nodes) and edge nodes based on whether their corresponding location is an important or core location in the communication network. For example, node type information may also include Visited PLMN (V-PLMN), Home PLMN (H-PLMN), H-core node, H-edge node, V-core node, or V-edge node, etc. This application does not specifically limit the node type information.

[0161] In other words, the communication device can determine the type of participating nodes based on location information, such as whether they are edge nodes, and thus further determine the time indication. For example, if the first device is determined to be an edge node, the collected data may not be representative, may be of low importance, and the frequency and speed of data reporting may be low; if the first device is a core node, the collected data may be of high importance, and the frequency and speed of data reporting may be high.

[0162] In one implementation, the location information can be the specific location of the participating node, such as latitude and longitude information, service cell information, or electronic fence information. The communication device can then determine the timing indication for the participating node to transmit intermediate data based on this location information. For example, if the location information of the area where the participating node is located is determined to be of low importance, the node can report data more frequently and at a faster speed; otherwise, the node can report data less frequently and at a slower speed.

[0163] In another example, the communication device can determine the first indication information based on the NF information associated with the first device, wherein the NF information associated with the first device can specifically be the number of NFs. That is, the communication device can determine the time or frequency of data reporting by the participating node based on the number of NFs associated with or managed by the participating node.

[0164] For example, if the area where the first device (such as NWADF) is located has a large number of network elements such as AMF and SMF, and if the number of network elements reaches or exceeds a preset threshold, it is considered that the first device has a large number of NFs associated with it. In order to avoid data transmission congestion or increased transmission delay, the communication device can determine that the data feedback speed corresponding to the first device is faster or the feedback frequency is higher, so as to improve the data transmission efficiency of machine learning. Conversely, if the number of NFs associated with the first device is smaller, the communication device can determine that the data feedback speed corresponding to the first device is slower or the feedback frequency is lower.

[0165] In another example, the communication device can determine the first indication information based on the data type associated with (or managed by) the first device. That is, the communication device can determine the timing indication for the node to send machine learning data based on the priority of the data types associated with or managed by the first device. For example, the first device is an external access point (AF) of the communication network, and the data type collected by the AF is typically third-party business data, such as multimedia data; or, for example, the first device is a terminal, and the data type collected by the terminal is typically user data. Different data type priorities can be pre-configured or defined. For example, if the priority of user data is higher than that of third-party business data, the communication device can determine that the data feedback speed of the external AF is slower or the feedback frequency is lower, while the data feedback speed of the terminal can be faster or the feedback frequency is higher, based on the data type associated with the participating nodes.

[0166] In another example, the communication device determines the first instruction information corresponding to the first device based on the capability information of the first device. Specifically, the communication device may determine the time or frequency of its data reporting based on the federated learning type supported by the first device.

[0167] For example, if the first device (such as NWADF) supports vertical federated learning, or if the first device has strong computing power, the communication device can determine that the data feedback speed corresponding to the first device is fast, or the feedback frequency is high.

[0168] It should be understood that in the above embodiments, the different strategies used by the communication device to determine the first indication information (such as time information or frequency information of the reported data) corresponding to the node can be used individually or in combination. Considering that the feedback speed obtained according to different determination strategies may be different, conflict resolution methods between different strategies can be pre-configured. For example, it can be configured such that if the result of one strategy is slow feedback, the final feedback speed is set to slow; or, it can be configured such that if the result of one strategy is fast feedback, the final feedback speed is set to fast; or, it can be configured such that the result with the larger number of results among multiple strategies is determined as the final result; or, it can be configured such that the results of multiple strategies correspond to 5, 7, and 9 data reports per minute, respectively, and the average can be taken, i.e., 7 data reports per minute. Alternatively, it can also be configured to determine the final feedback speed according to the priority of the type of feedback data, etc., which is not specifically limited in this application.

[0169] In one implementation, the embodiment shown in Figure 4 above can be applied to the initiator node of a machine learning task. It can be used to initiate machine learning tasks, perform model aggregation, or filter participating nodes (members) for machine learning tasks. For example, the initiator node can be an NWDAF, AF, or other network element. That is, in this application, the initiator node of the machine learning task can determine the time or frequency of data reporting for each participating node or candidate participating node. Further optionally, the initiator node can instruct each participating node separately, or send the determined time or frequency information to an intermediate node (or coordinating node), which then instructs each participating node separately.

[0170] Optionally, if the node initiating the machine learning task has its own data processing capabilities, it can also be a participating node in federated learning.

[0171] Furthermore, considering data security, a coordinator node can be introduced into the implementation process of federated learning tasks. The coordinator node can be used to coordinate data exchange between nodes and perform functions such as security authorization; it can also be called an intermediate node. For example, the coordinator node can be a network element such as NWDAF, NEF, Model Training Logical Function (MTLF), or Analytics Logical Function (AnLF), or it can be a single network element.

[0172] Optionally, the initiating node and coordinating node of the machine learning task can be deployed on the same network element, in which case the first device can have the functions of both an initiating node and a coordinating node. Alternatively, the initiating node and coordinating node of the machine learning task can be deployed independently, in which case the first device has either the function of an initiating node or the function of a coordinating node; this application does not limit this.

[0173] In addition, the machine learning process also includes participating nodes (members) of the machine learning task: these are nodes selected by the initiating node to join the machine learning group. Their main function is to collect data and, through data analysis, generate intermediate machine learning data such as gradient information and weight information, which are then sent to the initiating node for model aggregation. Participating nodes can specifically be NWDAF, external AF, AMF, SMF, or endpoints, etc.

[0174] In another implementation, the embodiment shown in Figure 4 can also be applied to the coordination node of a machine learning task. That is, the coordination node determines the time or frequency of data to be reported by each participating node or candidate participating node. Optionally, the coordination node can provide instructions to each participating node individually. Alternatively, the coordination node can send the determined time or frequency information of the reported data to the initiating node, which then instructs the participating nodes.

[0175] Optionally, in the aforementioned time or frequency information (Information 1) for data reported by participating nodes determined by the coordinating node, and the time or frequency information (Information 2) for data reporting indicated by the initiating node to the participating nodes, Information 1 and Information 2 may be the same or different. That is, the initiating node can process Information 1 as needed and generate Information 2 to instruct the participating nodes. For example, the coordinating node can determine that a participating node reports data at a relatively high frequency and instruct the initiating node. The initiating node can then instruct the participating node on the specific time or frequency information for data reporting (e.g., reporting every few seconds) based on the current data aggregation process and network resource conditions for data transmission, thereby achieving flexible scheduling of data reporting.

[0176] The following section will introduce several possible implementation methods based on a specific interaction process. This embodiment involves network elements such as a data analysis consumer node, a machine learning initiating node, a coordinating node, and at least one participating node. As shown in Figure 6, the communication method may include the following steps.

[0177] 601: The consumer node sends a first request message to the machine learning initiating node, and the initiating node receives the first request message, which carries the requirements of the machine learning task.

[0178] The first request message is used to request a data processing task, and can trigger the machine learning task by carrying the requirement information of the machine learning task. Optionally, the initiating node can determine whether to trigger the machine learning task based on the consumer's requirements, such as triggering federated learning.

[0179] For example, the first request message can specifically be a machine learning request, such as a federated learning request, a model request, an inference request, etc., all of which can be used to trigger machine learning.

[0180] 602: The initiating node of the machine learning task sends a second request message to the coordinating node, which receives the second request message. This second request message is used to obtain information about the participating nodes required for the machine learning task.

[0181] After receiving the first request message, the initiating node determines, based on the required information it carries, that federated machine learning needs to be triggered, and can then send a second request message to the coordinating node.

[0182] The second request message may carry the requirements of the machine learning task mentioned in step 601 above. That is, the second request message is used to request the coordinator to determine the participating nodes or candidate participating nodes for machine learning based on the requirements.

[0183] For example, the second request message may specifically be a federated learning prepare request message, used to request the coordinating node to provide participating nodes or candidate participating nodes for federated learning.

[0184] Optionally, the following steps may also be included.

[0185] 602-1: The coordinating node of the machine learning sends a first message to the third device, and the third device receives the first message, wherein the first message is used to obtain information on candidate participating nodes required for the machine learning task.

[0186] For example, the third device may be a network storage function or a data management function. For instance, the third device may be an NRF or a UDM.

[0187] The first message may include the requirements of the machine learning task, which is used by the third device to screen candidate participating nodes that can participate in the machine learning process for this machine learning task according to the requirements.

[0188] For example, the coordinating node sends a first message to the network storage function. The first message can be a network element discovery request (Nnrf_NFDiscovery_Request) message, used to obtain the currently registered machine learning nodes or the nodes that can currently participate in machine learning tasks from the network storage function.

[0189] 602-2: The third device determines the candidate participating nodes.

[0190] For example, the network storage function can select a registered node in the network as a candidate participating node from a candidate node list (such as a federated learning member list).

[0191] In one possible implementation, the network storage function can select and determine candidate participating nodes (such as members in federated learning) based on the needs of the machine learning task of the initiating node. For example, if the current machine learning task requires nodes that support vertical federated learning, the network storage function can select nodes that support vertical federated learning as candidate participating nodes for this machine learning task based on the algorithms supported by the nodes.

[0192] For example, if current machine learning requires the participation of network elements with the node type SMF, the network storage function can select network elements with the node type SMF as candidate participating nodes based on the required node type.

[0193] 602-3: The third device sends information about candidate participating nodes to the coordinating node, and the coordinating node receives the information about candidate participating nodes.

[0194] For example, the network storage function can send a network function discovery response (Nnrf_NFDiscovery_response) message to the coordinating node, which includes a list of candidate nodes, such as {candidate member list for federated learning}, which may include one or more candidate member nodes.

[0195] The information of the candidate participating nodes may include the identification information of at least one candidate node.

[0196] In some possible implementations, the information of candidate participating nodes may also include the node's corresponding contract information, such as attribute information and capability information. This contract information is used to assist in the selection of machine learning participating nodes. Additionally, the coordinating node can obtain the attribute information of (candidate) participating nodes through other means, such as requesting the attribute information of participating nodes from the data management function. This will be described below and will not be elaborated upon here.

[0197] Optionally, the coordinating node may obtain attribute and / or capability information of candidate participating nodes from a third device.

[0198] 603: The coordinating node of the machine learning determines the candidate participating nodes and / or participating nodes, and sends a first response message to the initiating node, which receives the first response message.

[0199] In one possible implementation, the coordinating node determines candidate participating nodes based on the requirements of this machine learning task and sends a first response message to the initiating node. The first response message is used to indicate the information of the determined candidate participating nodes.

[0200] In another possible implementation, the coordinating node determines the participating nodes based on the requirements of the current machine learning task and sends a first response message to the initiating node. The first response message indicates the information of the determined participating nodes. Optionally, the coordinating node determines candidate participating nodes based on the requirements of the current machine learning task, and then determines the participating nodes from the candidate participating nodes; or, the coordinating node determines the participating nodes based on at least one of the following: the requirements of the current machine learning task, the capabilities of the nodes, and the time and / or frequency information corresponding to the nodes.

[0201] In one possible implementation, the initiating node's pre-defined strategy is to determine all candidate participating nodes reported by the coordinating node as the final participating nodes. In this case, the candidate participating nodes and the participating nodes are the same.

[0202] The following description uses the information in the first response message used to indicate candidate participating nodes as an example:

[0203] In one possible implementation, the machine learning coordinating node can determine at least one candidate participating node for the machine learning task based on the requirements of the current machine learning task and the capability information and / or contract information of at least one network node. The aforementioned first response message includes information about at least one candidate participating node. For example, at least one candidate participating node includes the first device in the embodiment shown in FIG4 above.

[0204] For example, the coordinating node in a machine learning task determines candidate participating nodes needed for the current task based on relevant information about nodes in the stored communication network. Alternatively, the coordinating node can request subscription information or capability information of nodes in the communication network from other network elements to further determine candidate participating nodes needed for the current machine learning task.

[0205] In one possible implementation, the coordinating node can determine the time information and / or frequency information of at least one candidate participating node according to the implementation shown in Figure 4 above.

[0206] In one possible implementation, the coordinating node can determine the time and / or frequency information of the candidate participating nodes based on the attribute information of the nodes.

[0207] Specifically, after receiving the information of the candidate participating nodes fed back by the third device (such as the network storage function) in the aforementioned step 602-3, the coordination node can determine the attribute information of the candidate participating nodes based on the information of the candidate participating nodes (such as the identifier ID), and then determine the time information and / or frequency information corresponding to the candidate participating nodes based on the attribute information.

[0208] The attribute information includes, but is not limited to, the following: the location information of the node, the associated (or managed) NF information, the associated (or managed) data type, capability information, or the time information of the node's registration with the network. The NF information may include the number of NFs or the amount of data that an NF can collect. For example, the number of AMFs, SMFs, or terminals within the area associated with the NWDAF, or the number of service NFs that the NWDAF can collect data from, or the number of PDUs that the NF can collect data from.

[0209] In one implementation, attribute information can be carried in a network function registration request message, a network function update message, a status update message, or a context update message.

[0210] The coordinating node obtains the attribute information corresponding to the candidate participating nodes based on the information of the candidate participating nodes. For example, it can be obtained in the following ways.

[0211] In one possible implementation, when the candidate participating node is an NF in the core network, the coordinating node can obtain the attribute information corresponding to the candidate participating node by requesting the network storage function based on the information of the candidate participating node. When the candidate participating node requests registration from the network storage function, it can carry its own attribute information in the network function registration request message. Therefore, the network storage function can store the attribute information corresponding to the participating node in the network function registration request message.

[0212] Another possible implementation involves the coordinating node, when the candidate participating node is an AF (Agent Frontier) or a terminal, requesting attribute information corresponding to the candidate participating node from the data management function. The contracted data of nodes such as AFs and terminals is stored in the data management function; therefore, the coordinating node can send a node contracted data query request to the data management function to obtain the node's corresponding attribute information. This could include, for example, the node's location information, the NF (Network Frontier) information associated with (or managed by) the node, the data type associated with (or managed by) the node, or capability information.

[0213] Another possible implementation involves the coordinating node pre-configuring the location information, associated (or managed) data types, or capability information of the candidate participating nodes locally. The coordinating node then uses the information of currently registered candidate participating nodes sent by the network storage function to query its locally pre-configured information to determine the attribute information corresponding to a particular node.

[0214] Another possible implementation is that the coordinating node can request the attribute information of a candidate participating node, such as querying location information, associated data types, or capability information.

[0215] Another possible implementation is that if the candidate participating node is a terminal, the coordinating node can obtain the attribute information corresponding to the terminal from the data management function through the context update message.

[0216] It should be noted that the above attribute information can be obtained through one or more data sources. For example, the coordinating node can obtain the location information of the node from the UDM side (or request it from the node), query the associated (or managed) NF information from the network storage function, and obtain the node's capability information from the local provisioning information, etc.

[0217] The coordinating node can determine the time information and / or frequency information corresponding to the candidate participating node based on the attribute information of the candidate participating node, which can be used in the implementation method shown in Figure 4.

[0218] For example, the coordinating node determines the time information corresponding to a certain participating node. This time information is used to describe the rate at which the participating node feeds back intermediate data such as weights and gradients to the initiating node, or the length of the feedback interval, or the speed of the response.

[0219] For example, the coordinating node determines that the feedback time for participating node 1 is 3 seconds (s), meaning that participating node 1 will report intermediate data once within 3 seconds. Using this time indication, the initiating node of the federated learning can determine a local policy: accepting only one data report from participating node 1 within 3 seconds, discarding any extra data received by the receiving side (such as the initiating node). The above local policy is merely an example, and this application does not limit it.

[0220] For example, the coordinating node determines the timing information of participating node 2 as follows: the feedback rate is 1 time per second. Based on this timing indication, the federated learning initiating node can determine a local policy that the participating node 2 will provide intermediate data of federated learning at most once per second.

[0221] For example, the coordinating node can determine the timing information of participating node 3 as "high response speed". Based on this timing information, the initiating node of federated learning can determine a local policy, pre-configuring a high response speed as faster than once per second, which allows participating node 3 to report data once every 0.5 seconds.

[0222] In one implementation, the first response message can be used to indicate the node identifier of the candidate participating nodes in this machine learning process, and the indication information #1 corresponding to each candidate participating node (used to indicate time information and / or frequency information). For example, the first response message can be a federated learning prepare request response message. This indication information #1 can correspond to the second indication information sent by the fourth device to the communication device when the communication device is the initiating node and the fourth device is the coordinating node in the embodiment of Figure 4.

[0223] The node identifier (ID) can be used to identify a network element in a message. For example, the node identifier can be the identifier of an AF, an IP address, or the ID of an NF instance, etc., and this application does not limit it.

[0224] Optionally, the first response message may also carry capability information corresponding to the candidate participating nodes, which can be used by the initiating node to further filter participating nodes from the candidate participating nodes based on the capability information. For example, the first response message may carry {federated learning member ID, capability information, time indication}.

[0225] It should be noted that the coordinating node may send the capability information of the candidate participating node or the time indication of the candidate participating node to the initiating node in a single message or separately; this application does not impose any restrictions on this.

[0226] For example, the first response message may include a list of candidate participating nodes, including the identifier ID of at least one candidate participating node. Optionally, the first response message may also include capability information corresponding to each ID, time indication, etc.

[0227] Alternatively, in another implementation, to save on the overhead of the response message indication, the first response message sent by the coordinating node to the initiating node may only carry the identifiers of the candidate participating nodes for this machine learning, without carrying the indication information #1 corresponding to each candidate participating node. Subsequently, the initiating node may request the time indication of the participating nodes from the coordinating node, so that the coordinating node can indicate the corresponding time indication to the initiating node based on the node identifier carried in the request.

[0228] The above description uses the example of the coordinating node determining candidate participating nodes and the first response message indicating the information of the candidate participating nodes. The specific process of the first response message indicating the information of the participating nodes is similar to the above content. For example, "candidate participating nodes" can be replaced with "participating nodes", which will not be elaborated here.

[0229] In one possible implementation, the coordinating node determines at least one participating node and corresponding indication information #2 for each participating node, and sends the indication information #2 to the at least one participating node. The indication information #2 is used to indicate the time information and / or frequency information corresponding to the participating node. This indication information #2 may correspond to the first indication information determined when the communication device is the coordinating node in the embodiment of FIG4.

[0230] Optionally, the following steps may also be included.

[0231] 604: The machine learning initiating node determines the participating nodes and the corresponding instruction information #3.

[0232] In one implementation, the initiating node determines the participating nodes for this machine learning task and the time indications corresponding to the participating nodes, such as indication information #3, based on the information of the participating nodes carried in the first response message in step 603 above.

[0233] Optionally, the indication information #3 in step 604 can correspond to the first indication information determined when the communication device is the initiating node in the embodiment of FIG4.

[0234] In another implementation, the initiating node can select the final participating node based on the information of the candidate participating nodes in the first response message in step 603, according to the needs of this machine learning. Optionally, the initiating node can also select a suitable participating node based on the node's capability information, etc.

[0235] For example, if the candidate participating nodes include multiple participating nodes, the initiating node can select one or more nodes to participate in this federated learning according to actual needs.

[0236] For example, the initiating node may select participating nodes in the following ways:

[0237] 1. Select participating nodes based on the requirements of the machine learning task in step 601 above. For example, if the consumer node requires rapid learning, a large number of nodes need to be selected to provide data. All candidate participating nodes can be determined as the final participating nodes for this machine learning.

[0238] 2. Select participating nodes based on node capability information. For example, if the initiating node needs to perform vertical federated learning, then nodes that support vertical federated learning can be selected as participating nodes based on their capability information. In other words, nodes that do not support vertical federated learning can be removed from the candidate participating nodes.

[0239] 3. Select participating nodes based on the time and / or frequency information corresponding to the nodes. For example, the current round of federated learning of the initiating node needs nodes that can quickly provide feedback data. Therefore, suitable participating nodes can be determined based on the time information of the feedback intermediate data corresponding to the nodes.

[0240] It should be noted that the above method can also be used to determine the participating nodes from the candidate participating nodes in step 603, and the specific details will not be repeated here.

[0241] Optionally, if the first response message in step 603 does not include the candidate participating node or the time indication corresponding to the participating node, the initiating node may request the time indication corresponding to the participating node from the coordinating node. Specifically, this may include the following steps.

[0242] 604-1: The initiating node of the machine learning request the time indication corresponding to the participating nodes from the coordinating node.

[0243] The initiating node sends a second message to the coordinating node, requesting time and / or frequency information (such as indication information #3) for the participating nodes in the machine learning task. This second message may carry the identifier of at least one participating node identified by the initiating node. The coordinating node returns the indication information #3 corresponding to the participating node to the initiating node.

[0244] Optionally, the initiating node can determine the participating nodes for this machine learning based on the list of candidate participating nodes indicated by the coordinating node. Then, the initiating node can send a second message to the coordinating node.

[0245] 605: The initiating node of the machine learning sends a machine learning request message to the participating nodes, carrying instruction information #4.

[0246] The machine learning request message may carry the identification information of this machine learning task, and / or the data analysis task ID requested by this consumer node, as well as the time and / or frequency information of the intermediate data reported by the participating node, such as indication information #4. For example, a data analysis task may include one or more machine learning tasks.

[0247] Optionally, the data analysis task ID is used to identify a round of machine learning tasks, and it can be generated by the initiating node or the coordinating node.

[0248] It should be understood that in the embodiments of this application, in order to distinguish the time indications determined in the two stages, a first indication information and a second indication information can be used for differentiation. The first indication information and the second indication information can be the same or different. The first indication information can be the aforementioned indication information #3 or indication information #4.

[0249] Optionally, the indication information #4 in step 605 can correspond to the first indication information determined when the communication device is the initiating node in the embodiment of FIG4.

[0250] In another example, corresponding to the embodiment of Figure 4 where the communication device is a coordinating node, the indication information #3 in step 604 can be the second indication information received by the initiating node from the coordinating node. Thus, the initiating node can further obtain indication information #4 based on indication information #3, where indication information #4 corresponds to the first indication information determined by the initiating node in the embodiment of Figure 4.

[0251] For example, the coordinating node determines the time indication corresponding to participating node 1 as indication information #3, which could specifically be "high data reporting frequency"; based on the received indication information #3, the initiating node can generate the time indication corresponding to participating node 1 as indication information #4, which could specifically be "data reporting frequency is twice per second". In other words, the initiating node can further generate specific time information and / or frequency information based on the time indication determined by the coordinating node, such as the index information of the frequency of data transmission.

[0252] 606: Participating nodes send intermediate data to the initiating node based on time and / or frequency information.

[0253] Correspondingly, when a participating node receives a machine learning request message, it obtains time information and / or frequency information, and then sends intermediate data, such as gradient data or weight data, to the initiating node based on the time information and / or frequency information.

[0254] 607: The initiating node of machine learning performs model aggregation and sends a second response message to the consumer node.

[0255] For example, after the initiating node completes the aggregation of models in federated learning, it can send a second response message to the consumer node, which may carry the model obtained from the aggregation process.

[0256] Considering the flexibility of data reporting during machine learning, in one implementation, the communication device can further determine third indication information corresponding to at least one participating node in the machine learning task. This third indication information is used to update the time and / or frequency information for sending intermediate machine learning data corresponding to that participating node. As in the aforementioned embodiment, the communication device can be a coordinating node or an initiating node, and can send the third indication information to the participating nodes. Thus, the participating nodes can send intermediate data to the initiating node based on the received third indication information and the latest time and / or frequency information. Therefore, during machine learning, the time or frequency of data reporting by each participating node can be flexibly adjusted according to processing needs, improving the performance and flexibility of machine learning.

[0257] In another implementation, the participating nodes in machine learning can send their own attribute information to the network storage function through the registration process after network element initialization. Thus, in steps 602-1 to 602-3 of the above implementation process, in the scenario where the third device is the network storage function, the initiating node can obtain the attribute information of the candidate participating nodes through the network storage function, and determine the first indication information corresponding to the participating node based on the attribute information.

[0258] The following section will describe the possible implementation methods described above, using a specific interaction process as an example. This embodiment involves network elements such as a data analysis consumer node, a machine learning initiating node, a coordinating node, and at least one participating node. As shown in Figure 7, the communication method may include the following steps.

[0259] 701: Participating nodes send network element registration request messages to the network storage function, carrying attribute information.

[0260] For example, participating nodes send a network element registration request (such as an NF registration request) message to the network storage function to request the registration process after the network element is initialized.

[0261] The request message may include some or all of the following information: NF identifier, location information (e.g., whether it is an MEC node), and associated (or managed) NF information (e.g., the number of AMFs, SMFs, or terminals).

[0262] 702: The network storage function sends a network element registration response message to the participating node to indicate that the participating node has successfully registered.

[0263] 703: The consumer node sends the first request message to the machine learning initiating node, carrying the requirements of the machine learning task.

[0264] 704: The initiating node of the machine learning task sends a second request message to the coordinating node to obtain information about the participating nodes required for the machine learning task.

[0265] For specific steps 703 to 704, please refer to the aforementioned steps 601 to 602.

[0266] 705: The machine learning coordinating node sends the first message to the network storage function to obtain information on candidate participating nodes required for the machine learning task.

[0267] For example, the first message can specifically be a network element discovery request (Nnrf_NFDiscovery_Request), which can carry the node type of the request to indicate the type of the selected participating node NF. For example, NF type = FL member indicates that the selected participating node type can be a participating node in federated learning. For example, when the participating node is NWDAF, NF type = nwdaf. It should be noted that the type of NF can be an actual existing node type, or any existing node type, or a new custom type; this application does not limit this.

[0268] Optionally, the first message can also carry requirement information corresponding to this machine learning task, indicating the selection criteria for participating nodes (NFs). For example, the first message can provide requirement information indicating that the required participating nodes are MEC nodes, so that the network storage function can select nodes based on this requirement.

[0269] It should be understood that the execution order of steps 701-702 and 703-705 in the foregoing embodiments is not limited. For example, steps 701-702 and 703-705 can be executed simultaneously, or the above steps can be executed sequentially. This application does not limit this.

[0270] 706: The network storage function sends a third response message to the coordinating node, carrying information about the candidate participating nodes.

[0271] Optionally, if the first message carries the requirement information corresponding to this machine learning task, the network storage function can screen suitable candidate participating nodes for the coordinating node and feed them back to the coordinating node through the first response.

[0272] For example, the third response message can be a network element discovery response (Nnrf_NFDiscovery_Response) message.

[0273] Alternatively, if the first message does not carry the requirement information corresponding to this machine learning task, the network storage function can send the attribute information of the candidate participating nodes to the coordinating node through the third response message, and the coordinating node can select suitable participating nodes based on the attribute information.

[0274] For example, the third response message may include some or all of the following information: NF profile, location information, associated (or managed) NF information, etc.

[0275] It should be understood that the attribute information corresponding to the participating nodes mentioned above can be carried in the NF profile field of the response message, or it can be not carried in the response message, but fed back to the coordinating node through a separate message.

[0276] In one implementation, if the network storage function has a new service or a new node registers, the initiating node can first obtain the node's NF information (NF profile) from the network storage function through the discovery service, and then send a request to the network storage function based on the NF profile to obtain the aforementioned attribute information of the node.

[0277] As mentioned earlier, after the coordinating node obtains the attribute information of the candidate participating nodes, it can filter the participating nodes according to the attribute information and determine the first indication information corresponding to the participating nodes, thereby giving instructions to each participating node.

[0278] The subsequent process can be referred to the corresponding description of steps 603-607 in the aforementioned embodiments, and will not be repeated here.

[0279] The various embodiments mentioned above in this application can be combined without contradiction, and no limitation is imposed.

[0280] The above mainly describes the solution provided in this application from the perspective of interaction between various network nodes. Accordingly, this application also provides a communication device, which can be one of the communication devices or nodes in the above method embodiments, or a component such as a chip that can be used in the above communication devices or nodes.

[0281] It is understood that, in order to achieve the aforementioned functions, the communication device includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the unit and algorithm operations of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0282] It should be understood that the above description is merely an example illustrating the interactions between various network element nodes. In reality, the processing performed by the aforementioned communication devices or nodes is not limited to being performed by a single network element.

[0283] This application can divide the communication device into functional modules based on the above method examples. For example, each function can be divided into its own functional modules, or two or more functions can be integrated into one processing module. The integrated modules can be implemented in hardware or as software functional modules. It is understood that the module division in this application is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0284] For example, when the functional modules are divided in an integrated manner, Figure 8 shows a schematic diagram of the structure of a communication device 800. The communication device 800 includes a processing module 801 and an interface module 802.

[0285] In some embodiments, the communication device 800 may further include a storage module (not shown in FIG8) for storing program instructions and data.

[0286] For example, the communication device 800 can be used to implement the functions of the communication device in the embodiment of FIG4 above. The communication device 800 is, for example, the coordinating node or initiating node described in the various embodiments of FIG6-7 above.

[0287] The processing module 801 can be used to determine first indication information corresponding to at least one first device; wherein the first indication information is used to indicate the time information and / or frequency information of the first device sending intermediate data for machine learning, and the first device is a participating node and / or a candidate participating node of the machine learning task.

[0288] The interface module 802 can be used to send the first indication information.

[0289] In one embodiment, the processing module 801 can be used to determine first indication information based on the attribute information corresponding to the first device. The attribute information includes at least one of the following: location information corresponding to the first device, information of at least one network function (NF) associated with the first device, data type associated with the first device, or capability information of the first device; wherein the capability information includes at least one of the following: machine learning algorithms supported by the first device, data processing capability of the first device, or performance indicators of the first device.

[0290] In one embodiment, the processing module 801 can be used to determine the type information of the first device based on the location information of the first device, and determine the first indication information corresponding to the first device based on the type information of the first device; wherein, the type information includes edge nodes or non-edge nodes; determine the first indication information corresponding to the first device based on the number of network functions (NFs) associated with the first device; or, determine the first indication information corresponding to the first device based on the federated learning type supported by the first device.

[0291] In one embodiment, the time information and / or frequency information includes at least one of the following: the time, time interval, time range, transmission rate, transmission frequency, or indicator information indicating the speed of transmission when the first device sends intermediate data for machine learning to the second device, wherein the second device is the initiating node of the machine learning.

[0292] In one implementation, the interface module 802 can also be used to send a first message to a third device, including the requirements of the machine learning task, wherein the first message is used to obtain information on candidate participating nodes required by the machine learning task.

[0293] In one embodiment, the interface module 802 can also be used to receive attribute information corresponding to the first device.

[0294] In one implementation, attribute information is carried in a network function registration request message, a network function update message, a status update message, or a context update message.

[0295] In one embodiment, the communication device 800 can be a coordinating node for the machine learning task, and the processing module 801 can be used to determine at least one candidate participating node for the machine learning task based on the requirements of the machine learning task, the capability information and / or contract information of at least one network node; the interface module 802 can be used to send the information of the at least one candidate participating node to a second device; wherein, the at least one candidate participating node includes the first device, and the second device is the initiating node of the machine learning task.

[0296] In one embodiment, the communication device 800 can be a coordinating node for the machine learning task, and the interface module 802 can be used to receive a second message, which is used to request first indication information corresponding to each participating node in the machine learning task.

[0297] In one embodiment, the communication device 800 can be the initiating node of the machine learning task, and the interface module 802 can be used to receive second indication information from the fourth device. The second indication information is used to instruct at least one participating node and / or candidate participating node of the machine learning task to send time information and / or frequency information of the intermediate data of the machine learning; wherein, the fourth device is an intermediate node (coordinating node) of the machine learning task; the processing module 801 can be used to determine the first indication information corresponding to the first device according to the second indication information.

[0298] In one embodiment, the processing module 801 can be used to determine third indication information corresponding to at least one participating node in the machine learning task, the third indication information being used to indicate the time information and / or frequency information for updating the intermediate data of the machine learning sent by the participating node; the interface module 802 can be used to send the third indication information.

[0299] Additionally, the communication device 800 can be used to implement the function of the participating node in the machine learning in the above embodiments. The communication device 800 is, for example, the participating node described in the various embodiments of FIG6-FIG7.

[0300] The interface module 802 is used to receive first indication information, wherein the first indication information is used to instruct the first device to send intermediate data for machine learning, and / or frequency information; the first device is a participating node in the machine learning task.

[0301] The processing module 801 is used to send intermediate machine learning data through the interface module 802 based on the time information and / or frequency information.

[0302] In one embodiment, the time information and / or frequency information includes at least one of the following: the time, time interval, time range, transmission rate, transmission frequency, or indicator information indicating the speed of transmission when the first device sends intermediate data for machine learning to the second device, wherein the second device is the initiating node of the machine learning.

[0303] In one embodiment, the interface module 802 is used to send attribute information of the first device to the third device. The attribute information includes at least one of the following: location information corresponding to the first device, information of at least one network function (NF) associated with the first device, data type associated with the first device, or capability information of the first device; wherein the capability information includes at least one of the following: machine learning algorithms supported by the first device, data processing capabilities of the first device, or performance indicators of the first device.

[0304] In one implementation, the attribute information is carried in a network function registration request message, a network function update message, a status update message, or a context update message.

[0305] In one embodiment, the interface module 802 can be used to receive third indication information, which is used to indicate the time information and / or frequency information for updating the intermediate data of machine learning sent by the participating node; the processing module 801 can be used to send the intermediate data of machine learning according to the updated time information and / or frequency information.

[0306] In summary, when the communication device 800 is used to implement the functions performed by the communication device or node in the above embodiments, other functions that the communication device 800 can implement can be referred to the relevant descriptions of any of the embodiments shown above, and will not be elaborated further.

[0307] In a simplified embodiment, those skilled in the art will recognize that the communication device 800 can take the form shown in FIG3. For example, the processor 301 in FIG3 can invoke computer execution instructions stored in memory 303 to cause the communication device 300 to execute the method described in the above-described method embodiment.

[0308] For example, the function / implementation process of the processing module 801 in Figure 8 can be implemented by the processor 301 in Figure 3.

[0309] For example, the function / implementation process of the interface module 802 in Figure 8 can be implemented through the communication interface 304 in Figure 3.

[0310] It is understood that one or more of the above modules or units can be implemented by software, hardware, or a combination of both. When any of the above modules or units are implemented by software, the software exists as computer program instructions and is stored in memory. The processor can be used to execute the program instructions and implement the above method flow. The processor can be built into a SoC (System-on-Chip) or ASIC, or it can be a separate semiconductor chip. In addition to the core that executes software instructions for computation or processing, the processor may further include necessary hardware accelerators, such as field-programmable gate arrays (FPGAs), programmable logic devices (PLDs), or logic circuits that implement dedicated logic operations.

[0311] When the above modules or units are implemented in hardware, the hardware can be any one or any combination of a CPU, microprocessor, digital signal processing (DSP) chip, microcontroller unit (MCU), artificial intelligence processor, ASIC, SoC, FPGA, PLD, application-specific digital circuit, hardware accelerator, or non-integrated discrete device, which can run the necessary software or perform the above method flow independently of software.

[0312] In one possible implementation, this application also provides a chip system, including: at least one processor and an interface, wherein the at least one processor is coupled to a memory via the interface, and when the at least one processor executes a computer program or instructions in the memory, the method in any of the above method embodiments is executed. In one possible implementation, the chip system further includes a memory. Optionally, the chip system may be composed of chips or may include chips and other discrete devices; this application does not specifically limit this.

[0313] Optionally, this application also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be implemented by a computer program instructing related hardware. This program can be stored in the aforementioned computer-readable storage medium. When executed, the program can include the processes described in the above method embodiments. The computer-readable storage medium can be an internal storage unit of the communication device in any of the foregoing embodiments, such as the hard disk or memory of the communication device. The aforementioned computer-readable storage medium can also be an external storage device of the communication device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the communication device. Further, the aforementioned computer-readable storage medium can include both internal storage units and external storage devices of the communication device. The aforementioned computer-readable storage medium is used to store the aforementioned computer program and other programs and data required by the communication device. The aforementioned computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0314] In one possible implementation, this application also provides a computer program product. All or part of the processes in the above method embodiments can be executed by a computer program instructing related hardware. This program can be stored in the above computer program product, and when executed, it can include the processes of the above method embodiments.

[0315] In one possible implementation, this application also provides computer instructions. All or part of the processes in the above method embodiments can be executed by computer instructions instructing related hardware (such as a computer, processor, network device, or terminal device). The program can be stored in the aforementioned computer-readable storage medium or the aforementioned computer program product.

[0316] In one possible implementation, this application also provides a communication system, including: the first device and the communication device in the above embodiments. For example, the communication device can be a machine learning initiating node or a coordinating node.

[0317] In one possible implementation, the communication system may include the first device, the machine learning initiating node, and the coordinating node in the above embodiments.

[0318] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0319] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0320] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0321] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0322] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A communication method, characterized in that, The method includes: First indication information corresponding to at least one first device is determined; wherein the first indication information is used to indicate the time information and / or frequency information of the first device sending intermediate data for machine learning, and the first device is a participating node and / or a candidate participating node of the machine learning task. Send the first instruction information.

2. The method according to claim 1, characterized in that, The determination of the first indication information includes: The first indication information is determined based on the attribute information corresponding to the first device, wherein the attribute information includes at least one of the following: The location information corresponding to the first device, the information of at least one network function (NF) associated with the first device, the data type associated with the first device, or the capability information of the first device; The capability information includes at least one of the following: the machine learning algorithms supported by the first device, the data processing capability of the first device, or the performance indicators of the first device.

3. The method according to claim 2, characterized in that, Determining the first indication information based on the attribute information corresponding to the first device includes at least one of the following methods: The type information of the first device is determined based on the location information of the first device, and the first indication information corresponding to the first device is determined based on the type information of the first device; wherein, the type information includes edge nodes or non-edge nodes; Based on the number of network functions (NFs) associated with the first device, determine the first indication information corresponding to the first device; or... Based on the federated learning type supported by the first device, determine the first instruction information corresponding to the first device.

4. The method according to claim 2 or 3, characterized in that, The method further includes: Receive the attribute information corresponding to the first device.

5. The method according to any one of claims 2-4, characterized in that, The attribute information is carried in network function registration request messages, network function update messages, status update messages, or context update messages.

6. The method according to any one of claims 1-5, characterized in that, The timing and / or frequency information for sending intermediate machine learning data includes at least one of the following: The first device sends the intermediate data of the machine learning to the second device at the time, time interval, time range, sending rate, sending frequency, or an indicator of the sending frequency. The second device is the initiating node of the machine learning.

7. The method according to any one of claims 1-6, characterized in that, The method further includes: A first message is sent to a third device. The first message includes the requirements of the machine learning task and is used to obtain information on candidate participating nodes required by the machine learning task.

8. The method according to any one of claims 1-7, characterized in that, The method is applied to the coordination node of the machine learning task, and the method further includes: Based on the requirements of the machine learning task, the capability information and / or contract information of at least one network node, at least one candidate participating node for the machine learning task is determined; The information of the at least one candidate participating node is sent to the second device; wherein the at least one candidate participating node includes the first device, and the second device is the initiating node of the machine learning.

9. The method according to any one of claims 1-8, characterized in that, The method is applied to the coordination node of the machine learning task, and the method further includes: Receive a second message, which is used to request first indication information corresponding to each participating node in the machine learning task.

10. The method according to any one of claims 1-8, characterized in that, The method is applied to the initiating node of the machine learning task, and determining the first indication information includes: The system receives a second instruction from a fourth device, the second instruction being used to instruct at least one participating node and / or candidate participating node of the machine learning task to send time information and / or frequency information of the intermediate data of the machine learning; wherein, the fourth device is the coordinating node of the machine learning task; The first instruction information corresponding to the first device is determined based on the second instruction information.

11. The method according to any one of claims 1-10, characterized in that, The method further includes: A third indication information is determined for at least one participating node in the machine learning task, the third indication information being used to indicate the time information and / or frequency information for updating the intermediate data sent by the participating node for machine learning; Send the third instruction information.

12. A communication method, characterized in that, Applied to a first device, the method includes: Receive first indication information, wherein the first indication information is used to instruct the first device to send time information and / or frequency information of intermediate data for machine learning; the first device is a participating node in the machine learning task; Intermediate data for machine learning is sent based on the time information and / or frequency information.

13. The method according to claim 12, characterized in that, The timing and / or frequency information for sending intermediate machine learning data includes at least one of the following: The first device sends the intermediate data of the machine learning to the second device at the time, time interval, time range, sending rate, sending frequency, or an indicator of the sending frequency. The second device is the initiating node of the machine learning.

14. The method according to claim 12 or 13, characterized in that, The method includes: Send attribute information of the first device to the third device, the attribute information including at least one of the following: The location information corresponding to the first device, the information of at least one network function (NF) associated with the first device, the data type associated with the first device, or the capability information of the first device; The capability information includes at least one of the following: the machine learning algorithms supported by the first device, the data processing capability of the first device, or the performance indicators of the first device.

15. The method according to claim 14, characterized in that, The attribute information is carried in network function registration request messages, network function update messages, status update messages, or context update messages.

16. The method according to any one of claims 12-15, characterized in that, The method further includes: Receive third indication information, the third indication information being used to indicate the time information and / or frequency information for updating the intermediate data for machine learning sent by the participating node; Intermediate data for machine learning is sent based on the updated time and / or frequency information.

17. A communication device, characterized in that, It includes at least one module or unit for implementing the method as described in any one of claims 1-11; or, it includes at least one module or unit for implementing the method as described in any one of claims 12-16.

18. A communication device, characterized in that, include: A processor coupled to a memory for storing a program or instructions which, when executed by the processor, cause the method as claimed in any one of claims 1-11 to be performed; or cause the method as claimed in any one of claims 12-16 to be performed.

19. A communication device, characterized in that, Includes a processor for running a computer program or instructions to cause the communication device to perform the method as described in any one of claims 1-11; or, the method as described in any one of claims 12-16.

20. The communication device according to claim 19, characterized in that, The communication device further includes a memory for storing the computer program or instructions.

21. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, When the computer program or instructions are executed, the method as described in any one of claims 1-11 is performed; or, the method as described in any one of claims 12-16 is performed.

22. A computer program product, wherein the computer program product includes a computer program, characterized in that, When the computer program is run on a computer, it causes the method as described in any one of claims 1-11 to be performed; or causes the method as described in any one of claims 12-16 to be performed.

Images