Communication method, apparatus and system

Abstract

Provided in the present application are a communication method, apparatus and system. The method provides a solution capable of effectively improving the accuracy of a model obtained by means of federated learning. Specifically, the method comprises: before using a local data set to train a first model from a first network element, a second network element may perform data processing on the local data set according to an instruction of the first network element, may use the local data set that has been subjected to the data processing to train the first model, so as to obtain a second model, and may send the second model to the first network element, wherein the data processing is used for making the local data set meet a data reliability requirement. In this way, by means of performing, before model training, data processing on a local data set used for model training, the reliability of the local data set can be effectively improved, and thus the accuracy of a model obtained by means of federated learning can be effectively improved.

Description

Communication method, device and system

[0001] This application claims priority to the Chinese patent application filed with the State Intellectual Property Office of China on July 18, 2023, with application number 202310883716.X and application name “Communication Methods, Devices and Systems”, the entire contents of which are incorporated by reference into this application. Technical Field

[0002] The present application relates to the field of communication technology, and more specifically, to a communication method, device, and system. Background Art

[0003] Federated learning (FL) is an emerging basic technology for artificial intelligence. Its design goal is to carry out efficient machine learning among multiple participants or multiple computing nodes while ensuring information security during big data exchange, protecting terminal data and personal data privacy, and ensuring legality.

[0004] Specifically, the server-side network element sends the model to be trained to the client-side network element. The client-side network element trains the model based on the local training data set it has collected, obtaining the model parameters of the trained model. The client-side network element sends the model parameters to the server-side network element, which aggregates the model parameters to complete the model training.

[0005] However, the accuracy of the models trained using federated learning may be questionable, which may prevent them from being well applied in some scenarios. Therefore, how to effectively improve the accuracy of the models trained using federated learning is a technical problem that needs to be solved urgently.

[0006] Summary of the Invention

[0007] The present application provides a communication method, device, and system that can effectively improve the accuracy of the model obtained through federated learning.

[0008] In a first aspect, a communication method is provided, including: a first network element sends a first model and first indication information to a second network element, the first indication information is used to instruct the second network element to perform data processing on a first data set, and the first data set is used by the second network element to train the first model; the first network element receives a second model sent by the second network element, the second model is obtained by the second network element training the first model using the second data set, and the second data set is a data set obtained after the second network element performs the data processing on the first data set.

[0009] The first data set is a data set obtained by the second network element for training the first model from the first network element. The second network element may obtain or collect the first data set from other network elements.

[0010] Specifically, the first network element sending the first model to the second network element may be: the first network element sending the model architecture of the first model and the initial parameters within the model architecture to the second network element. The first network element receiving the second model may be: the first network element receiving the model architecture of the second model and the parameters within the model architecture sent by the second network element. The model architecture of the first model may be the same as or different from the model architecture of the second model, and this is not limited.

[0011] It should be noted that the above-mentioned data processing is used to make the first data set meet the data reliability requirements. Among them, the above-mentioned data processing can be data cleaning processing, which can be used to delete data with security risks in the first data set, or it can be used to eliminate invalid data or incomplete data or duplicate data in the first data set, or it can also be used to supplement the incomplete data in the first data set, etc., so that the reliability of the data set used for model training can be guaranteed. In addition, the embodiment of the present application does not specifically limit the type or specific content of the above-mentioned data processing. Among them, the "data reliability requirements" can be: the first data set includes less duplicate data, or the first data set includes less incomplete data, or the first data set includes less dangerous data, etc.

[0012] In addition, the above-mentioned first indication information can also be used to instruct the second network element to use the first data set after data processing to train the first model. In this way, the second network element can process the first data set according to the first indication information to obtain the first data set after data processing, that is, the second data set.

[0013] Specifically, by performing the above-mentioned data processing on the first data set, the reliability of the data set used for the machine learning model can be effectively guaranteed, so that the machine learning model is trained based on a secure data set, thereby effectively improving the accuracy of the model obtained through federated learning.

[0014] Specifically, the second network element may, according to the instruction of the first network element, first process the first data set to obtain a second data set before using the first data set to train the first model, and then use the second data set to train the first model. In this way, before training the first model, the data set used to train the first model is first processed, which can effectively ensure the reliability of the data set used for machine learning models, thereby effectively improving the accuracy of the model obtained through federated learning.

[0015] In a possible implementation, the method further includes: the first network element sending second indication information to the second network element, where the second indication information is used to indicate an algorithm, and the algorithm is used by the second network element to perform the data processing on the first data set.

[0016] Specifically, the second network element may use the algorithm indicated by the first network element to process the first data set, thereby improving the efficiency of performing the above data processing on the first data set.

[0017] In one possible implementation, the method further includes: the first network element sends a data volume threshold to the second network element, where the data volume threshold is used to instruct the second network element to train the machine learning model when the data volume of the second data set is greater than or equal to the data volume threshold.

[0018] In this way, the machine learning model can be trained based on a data set that meets a certain data volume, thereby effectively improving the accuracy of the model obtained through federated learning.

[0019] In a possible implementation, the method further includes: the first network element receiving first feedback information sent by the second network element, where the first feedback information is used to indicate that the second network element has performed the data processing on the first data set.

[0020] In this way, the first network element can determine that the model parameters of the trained machine learning model are obtained after the second network element trains the model using the data set after data processing.

[0021] In a possible implementation, the method further includes: the first network element receiving the data volume of the first data set and the data volume of the second data set sent by the second network element.

[0022] In this way, the first network element can observe the data quality when the second network element collects data, and then determine whether to continue to let the second network element perform model training.

[0023] In a possible implementation, the method further includes: the first network element receiving second feedback information sent by the second network element, where the second feedback information is used to indicate that the second network element has used the algorithm to perform the data processing on the first data set.

[0024] In this way, the first network element can determine that the second network element uses the algorithm indicated by the first network element to perform the above data processing on the first data set.

[0025] In a possible implementation, the method further includes: the first network element sends a request message to the second network element, where the request message is used to request the second network element to train the first model, and the request message includes the first model and the first indication information.

[0026] Specifically, after receiving the request information from the first network element, the second network element can trigger or start training the first model based on the request information.

[0027] In a possible implementation manner, the request information further includes second indication information.

[0028] By carrying the second indication information in the same message, signaling overhead can be effectively reduced.

[0029] In a possible implementation, the request information further includes the data volume threshold.

[0030] By carrying the data volume threshold in the same message, the signaling overhead can be effectively reduced.

[0031] In a second aspect, a communication method is provided, including: a second network element receives a first model and a first indication information sent by a first network element, the first indication information being used to instruct the second network element to perform data processing on a first data set; the second network element performs the data processing on the first data set according to the first indication information to obtain a second data set; the second network element trains the first model according to the second data set to obtain a second model; and the second network element sends the second model to the first network element.

[0032] In a possible implementation, the method further includes: the second network element receiving second indication information sent by the first network element, where the second indication information is used to indicate an algorithm, and the algorithm is used by the second network element to perform the data processing on the first data set.

[0033] In one possible implementation, the method further includes: the second network element receiving a data volume threshold sent by the first network element, where the data volume threshold is used to instruct the second network element to train the first model when the data volume of the second data set is greater than or equal to the data volume threshold.

[0034] In a possible implementation, the method further includes: the second network element sending first feedback information to the first network element, where the first feedback information is used to indicate that the second network element has performed the data processing on the first data set.

[0035] In a possible implementation, the method further includes: the second network element sending the data volume of the first data set and the data volume of the second data set to the first network element.

[0036] In a possible implementation, the method further includes: the second network element sending second feedback information to the first network element, where the second feedback information is used to indicate that the second network element has used the algorithm to perform the data processing on the first data set.

[0037] In a possible implementation, the method further includes: the second network element receives a request message sent by the first network element, the request message is used to request the second network element to train the first model, and the request message includes the first model and the first indication information.

[0038] In a possible implementation manner, the request information further includes second indication information.

[0039] In a possible implementation, the request information further includes the data volume threshold.

[0040] In a third aspect, a communication device is provided. The communication device may be a network device, or a device or module for performing the functions of a network device.

[0041] In one possible implementation, the communication device may include a module or unit corresponding to each of the methods / operations / steps / actions described in the first aspect. The module or unit may be a hardware circuit, software, or a combination of hardware circuit and software.

[0042] In a fourth aspect, a communication device is provided. The communication device may be a terminal device, or a device or module for executing the functions of the terminal device.

[0043] In one possible implementation, the communication device may include a module or unit corresponding to each of the methods / operations / steps / actions described in the second aspect. The module or unit may be a hardware circuit, software, or a combination of hardware circuit and software.

[0044] In a fifth aspect, a communication device is provided, comprising a processor, wherein the processor is configured to, by executing a computer program or instruction, or by a logic circuit, enable the communication device to execute the first aspect and any possible method described in the first aspect; or enable the communication device to execute the second aspect and any possible method described in the second aspect.

[0045] In a possible implementation, the communication device further includes a memory for storing the computer program or instruction.

[0046] In a possible implementation, the communication device further includes a communication interface, which is used to input and / or output signals.

[0047] In the sixth aspect, a communication device is provided, comprising a logic circuit and an input / output interface, the input / output interface being used to input and / or output signals, the logic circuit being used to execute the method described in the first aspect and any possible embodiment of the first aspect; or the logic circuit being used to execute the method described in the second aspect and any possible embodiment of the second aspect.

[0048] In the seventh aspect, a computer-readable storage medium is provided, on which a computer program or instruction is stored. When the computer program or the instruction is run on a computer, the method described in the first aspect and any possible method of the first aspect is executed; or, the method described in the second aspect and any possible method of the second aspect is executed.

[0049] In an eighth aspect, a computer program product is provided, comprising instructions, which, when executed on a computer, cause the method described in the first aspect and any possible method of the first aspect to be executed; or cause the method described in the second aspect and any possible method of the second aspect to be executed.

[0050] In the ninth aspect, a communication system is provided, including: a first network element and a second network element; the first network element is used to send a first model and a first indication information to the second network element, the first indication information is used to instruct the second network element to perform data processing on the first data set, and the first data set is used for the second network element to train the first model; the second network element is used to receive the first model and the first indication information; the second network element is used to perform the data processing on the first data set according to the first indication information to obtain a second data set; the second network element is used to train the first model according to the second data set to obtain a second model; the second network element is used to send the second model to the first network element; the first network element is used to receive the second model.

[0051] The functions performed by the first network element and the second network element can be referred to the relevant descriptions of the first aspect and the second aspect, and will not be repeated here.

[0052] The description of the advantageous effects of any of the second to ninth aspects etc. may refer to the description of the advantageous effects of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] FIG1 is a schematic diagram of a communication system 100 applicable to an embodiment of the present application.

[0054] FIG2 is a schematic diagram of a 5G network architecture 200 according to an embodiment of the present application.

[0055] FIG3 is a schematic diagram of an interaction flow of a communication method 300 according to an embodiment of the present application.

[0056] FIG4 is a schematic diagram of an interaction flow of a communication method 400 according to an embodiment of the present application.

[0057] FIG5 is a schematic diagram of an interaction flow of a communication method 500 according to an embodiment of the present application.

[0058] FIG6 is a schematic block diagram of a communication device 600 according to an embodiment of the present application.

[0059] FIG7 is a schematic block diagram of a communication device 700 according to an embodiment of the present application.

[0060] FIG8 is a schematic block diagram of a communication device 800 according to an embodiment of the present application. DETAILED DESCRIPTION

[0061] In order to facilitate understanding of the embodiments of the present application, the following points are first explained.

[0062] 1. In the various embodiments of this application, unless otherwise specified or there is a logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced by each other. The technical features of different embodiments can be combined to form new embodiments based on their internal logical relationships.

[0063] 2. The various numerical numbers involved in this application are only used for the convenience of description and are not used to limit the scope of this application. The size of the serial numbers involved in this application does not mean the order of execution. The order of execution of each process should be determined by its function and internal logic. For example, the terms "first", "second", "third", "fourth" and other various terminology labels (if any) in the specification and claims of this application and the drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate so that the embodiments described here can be implemented in an order other than what is illustrated or described here.

[0064] 3. The terms "comprise" and "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such process, method, product or apparatus.

[0065] 4. In this application, "used to indicate" can be understood as "enabling," and "enabling" can include direct enabling and indirect enabling. When describing that certain information is used to enable A, it can include that the information directly enables A or indirectly enables A, and does not necessarily mean that the information contains A.

[0066] The information enabled by the information is called information to be enabled. In the specific implementation process, there are many ways to enable the enabled information, such as but not limited to, directly enabling the information to be enabled, such as the information to be enabled itself or the index of the information to be enabled. The information to be enabled can also be indirectly enabled by enabling other information, wherein there is an association between the other information and the information to be enabled. It is also possible to enable only a part of the information to be enabled, while the other parts of the information to be enabled are known or agreed in advance. For example, it is also possible to enable specific information with the help of the arrangement order of each piece of information agreed in advance (such as specified in the protocol), thereby reducing the enabling overhead to a certain extent. At the same time, it is also possible to identify the common parts of each piece of information and enable them uniformly to reduce the enabling overhead caused by enabling the same information separately.

[0067] 5. "Storage" or "saving" as used in this application may refer to storage in one or more memories. The one or more memories may be provided separately or integrated into an encoder or decoder, a processor, or a communication device. The one or more memories may also be provided in part separately and in part integrated into a decoder, a processor, or a communication device. The memory may be any type of storage medium and is not limited in this application.

[0068] 6. The “protocol” referred to in this application may refer to a standard protocol in the field of communications, such as the fourth generation (4G) network, the fifth generation (5G) network protocol, the new radio (NR) protocol, the 5.5G network protocol, the sixth generation (6 th generation, 6G) network protocols and related protocols used in future communication systems, which are not limited in this application.

[0069] 7. The arrows or boxes indicated by dotted lines in the schematic diagrams in the accompanying drawings of this application specification represent optional steps or optional modules.

[0070] 8. Any embodiment or design described in this application as "exemplary" or "for example" should not be construed as preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner to facilitate understanding.

[0071] 9. In this application, unless otherwise specified, “ / ” indicates that the objects associated with each other are in an “or” relationship. For example, A / B can represent A or B. “And / or” in this application is only a description of the association relationship between associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural.

[0072] To better understand the technical solution of the present application, the following description is made from the following aspects: a communication system, a communication method, and a communication device.

[0073] 1. Communication System

[0074] In order to solve the technical problems mentioned in the background technology part, the present application provides an applicable communication system 100. As shown in Figure 1, the communication system 100 includes: a first network element (which can be a server-side network element) and a second network element (which can be a client-side network element).

[0075] Specifically, the first network element and the second network element can exchange model training task content, for example, the first network element is used to send the model to be trained to the second network element, and the second network element is used to collect a local data set and use the local data set to train the model sent by the first network element. The first network element can also be used to send instruction information to the second network element for instructing the second network element to perform data processing on the local data set it has obtained.

[0076] In the communication system 100, the first network element and the second network element can be data function network elements (network elements with data processing functions). For example, the first network element can be a network data analytics function network element (NWDAF) on the server side in the 5G network, and the second network element can be a client side NWDAF in the 5G network.

[0077] In addition, the communication system 100 may also include multiple second network elements, that is, the first network element and multiple second network elements perform interactive model training tasks, etc., thereby jointly completing the model training task. For ease of description, the embodiment of the present application describes the communication system 100 including one second network element as an example, but is not limited to the scenario of including more second network elements. The functions or tasks performed by each second network element are roughly the same.

[0078] The following description uses the communication system 100 applied to a 5G network architecture as an example. FIG2 is a schematic diagram of the communication system 100 applied to a 5G network architecture. As shown in FIG2(a), the 5G network architecture 200 may include:

[0079] 1. Access and mobility management function (AMF) network element.

[0080] AMF mainly performs functions such as mobility management or access authentication / authorization. In addition, AMF is also responsible for transmitting user policies between terminal devices and policy control function (PCF) network elements. In addition, AMF can receive non-access stratum (NAS) signaling (including mobility management (MM) signaling and session management (SM) signaling) of terminal devices and related signaling of access network devices (for example, base station-level next generation (NG) 2 interface signaling that interacts with AMF), complete the user registration process and forwarding of SM signaling and mobility management.

[0081] 2. Session management function (SMF) network element.

[0082] SMF is primarily used for session management, allocation and management of Internet Protocol (IP) addresses for terminal devices, selection of endpoints for manageable user plane functions, policy control, and charging interfaces, or downlink data notification. SMF can also be used to complete processes related to establishing, releasing, or updating protocol data unit (PDU) sessions.

[0083] 3. Policy control function (PCF) network element.

[0084] The PCF may be responsible for user policy management, including both mobility-related policies and PDU session-related policies, such as QoS policies or charging policies.

[0085] 4. Unified data repository (UDR).

[0086] UDR mainly includes the following functions: 1) Unified data management (UDM) stores subscription data or reads subscription data; 2) PCF stores policy data or reads policy data; 3) stores exposed data or reads exposed data from it.

[0087] The UDR and the network function (NF) accessing it have the same public land mobile network (PLMN), that is, they are in the same network, that is, the Nudr interface is a PLMN internal interface.

[0088] 5. UDM.

[0089] UDM mainly includes the following functions: unified data management, support for authentication credentials processing in 3GPP authentication and key negotiation mechanism, user identity processing, access authorization, registration and mobility management, contract management and short message management.

[0090] 6. Application function network element (AF).

[0091] AF mainly includes the following functions: interacting with the 3GPP core network to provide business or services, including: interacting with NEF, policy architecture interaction, etc.

[0092] 7. User plane function (UPF) network element.

[0093] As the interface with the data network, the UPF performs functions such as user plane data forwarding, session / flow-level billing and statistics, and bandwidth limiting. This includes packet routing and forwarding, as well as QoS processing for user plane data.

[0094] 8. (Radio) access network equipment (R)AN).

[0095] (R)AN can manage wireless resources, provide access services for terminal devices, and complete the forwarding of terminal device data between the terminal device and the core network. (R)AN can also be understood as a base station. For details, please refer to the previous description.

[0096] 9. Data network element (DN).

[0097] DN is used to provide, for example, operator services, Internet access, or third-party services, and includes a server that implements video source encoding and rendering.

[0098] 10. Network repository function (NRF)

[0099] NRF is responsible for maintaining network element service information, supporting the acceptance of network element registration services, and returning target network element service information based on network element discovery requests. Because it can have interfaces with all service-oriented network elements in the above architecture (such as AMF, SMF, PCF, etc.), it is not shown in the figure.

[0100] 11. NWDAF.

[0101] The NWDAF is responsible for providing network data analysis services. Specifically, the NWDAF can communicate with other NFs using existing service-based interfaces. The NWDAF can also use service-based interfaces to expose data analysis results to any consuming NF. Interactions between NFs and the NWDAF occur within the PLMN. The NWDAF can also interact with the NRF and PCF, among others.

[0102] In the above description, the network element can be a network element in a hardware device, a software function running on dedicated hardware, or a virtualized function instantiated on a platform (e.g., a cloud platform). The functional network element can be divided into one or more services, and further, there may be services that exist independently of the network function. An instance of the functional network element, an instance of a service included in the functional network element, or an instance of a service that exists independently of the network function can be referred to as a service instance.

[0103] As can be seen from (a) of Figure 2, the terminal device accesses 5GS through the access network device, the terminal device communicates with AMF through the NG1 interface (N1 for short), the access network device communicates with AMF through the NG2 interface (N2 for short), the access network device communicates with UPF through the NG3 interface (N3 for short), AMF communicates with SMF through the NG11 interface (N11 for short), AMF communicates with UDM through the NG8 interface (N8 for short), AMF communicates with PCF through the NG15 interface (N15 for short), SMF communicates with PCF through the NG7 interface (N7 for short), SMF communicates with UPF through the NG4 interface (N4 for short), UPF accesses DN through the NG6 interface (N6 for short), NWDAF communicates with PCF through NG23 (N23 for short), UDM communicates with UDR through the NG35 interface (N35 for short), PCF communicates with UDR through the NG36 interface (N36 for short), etc.

[0104] The interfaces between the control plane network elements in Figure 2(a) are point-to-point interfaces. In actual implementation, the interfaces between the control plane network elements can also be service-based interfaces as shown in Figure 2(b). In Figure 2(b), Npcf, Nudr, Nudm, Naf, Namf, Nsmf, and Nnwdaf are service-based interfaces provided by the PCF, UDR, UDM, AF, AMF, SMF, and NWDAF, respectively, for invoking corresponding service-based operations.

[0105] The naming of each network element shown in Figure 2 (a) or Figure 2 (b) is only a name, and the name does not limit the function of the network element itself. In 5G networks and other future networks, the above-mentioned network elements may also have other names, and the embodiments of this application do not specifically limit this. For example, in a 6G network, some or all of the above-mentioned network elements may use the terminology in 5G, or may have other names, etc., which are uniformly explained here and will not be repeated below.

[0106] The 5G network structure shown in Figure 2 is only an example description. The 5G network structure shown in Figure 2 may also include other network elements not mentioned, such as an authentication server function (AUSF), a network slice selection function (NSSF), a network open function network element (NEF), etc.

[0107] In addition, the "network element" in this article may also be referred to as a network function instance, NF, equipment, device or module, etc., which is not specifically limited in this application. In addition, the above naming is only defined to facilitate the distinction between different functions and should not constitute any limitation. This application does not exclude the possibility of adopting other naming in 5G networks and other networks in the future. For example, in a 6G network, some or all of the above-mentioned networks may continue to use the terminology in 5G, or may adopt other names, etc. The interface name between the above-mentioned network elements is only an example. The name of the interface in the specific implementation may be other names, and there is no specific limitation on this. In addition, the name of the message (or signaling) transmitted between the above-mentioned network elements is only an example and does not constitute any limitation on the function of the message itself.

[0108] The following further describes the interaction between the first network element (taking NWDAF1 as an example) and the second network element (taking NWDAF2 as an example) with reference to FIG. 3 .

[0109] 2. Communication method

[0110] FIG3 is a schematic diagram of an interaction flow of a communication method 300 according to an embodiment of the present application. The method 300 involves interaction between NWDAF1 and NWDAF2. NWDAF1 and NWDAF2 are examples of the aforementioned data function network elements, but are not limited thereto. The method 300 includes:

[0111] S301 , NWDAF1 sends model 1 and indication information 1 (such as first indication information) to NWDAF2 , where indication information 1 is used to instruct data processing 1 to be performed on data set 1 (such as the first data set), and data set 1 is used by NWDAF2 to train model 1 .

[0112] Accordingly, NWDAF2 receives model 1 and instruction information 1. Further, NWDAF2 determines that data processing 1 needs to be performed on data set 1 according to instruction information 1.

[0113] Specifically, Model 1 can be any model that a function consumer requests NWDAF1 to train. In other words, the function consumer can subscribe to NWDAF1's federated learning service for Model 1. Furthermore, NWDAF1 initiates and leads the federated learning process based on Model 1. Model 1 can be a deep neural network or a random forest model, for example.

[0114] Specifically, NWDAF1 sends Model 1 to NWDAF2, which can be: NWDAF1 sends the model architecture of Model 1 and the initial parameters in the model architecture to NWDAF2. Further, NWDAF2 is used to specifically perform the model training process for Model 1.

[0115] In order to ensure the data reliability of dataset 1, NWDAF1 sends instruction information 1 to NWDAF2, which is used to instruct data processing 1 to be performed on dataset 1. NWDAF2 performs data processing 1 on dataset 1 according to instruction information 1, so that dataset 1 can meet preset requirements, such as data reliability requirements. In other words, data processing 1 is used to make dataset 1 meet data reliability requirements.

[0116] In the embodiment of the present application, data processing 1 can be data cleaning processing, which can be used to delete data that poses a security risk in the data set 1, or it can be used to eliminate invalid data, incomplete data, or duplicate data in the data set 1, etc., so as to ensure the reliability and / or security of the data set used for the machine learning model. In addition, the embodiment of the present application does not specifically limit the type or specific content of the above-mentioned data processing. Among them, the "data reliability requirement" can be: the data set 1 includes less duplicate data, or the data set 1 includes less incomplete data, or the data set 1 includes less dangerous data, etc.

[0117] In one possible implementation, instruction information 1 can be used to instruct NWDAF2 to use processed dataset 1 to train model 1. In this way, NWDAF2 can process dataset 1 based on instruction information 1 to obtain processed dataset 1, i.e., dataset 2. Therefore, instruction information 1 does not necessarily need to explicitly instruct NWDAF2 to perform data processing 1 on dataset 1; it can implicitly instruct NWDAF2 to perform data processing 1 on dataset 1 to obtain dataset 2.

[0118] In one possible implementation, the indication information 1 may indicate whether NWDAF2 uses the data set 1 after data processing to train the model 1. For example, the indication information 1 may be indicated by one bit. Exemplarily, when the value of the indication information 1 is 1, it is used to indicate that data processing 1 is performed on the data set 1. When the value of the indication information 1 is 0, it is not used to indicate that data processing 1 is performed on the data set 1. In an optional implementation, the indication information 1 may indicate whether NWDAF2 performs data processing on the data set 1. The indication information 1 may indicate that data processing 1 is performed on the data set 1 through the indication information 1. It may also support that the indication information 1 is not used to indicate that data processing 1 is performed on the data set 1. For example, when the indication information 1 is specified or configured as a special value (for example, 0), it is not used to indicate that data processing 1 is performed on the data set 1. Therefore, the above content is to be understood as an example only and not as a final limitation.

[0119] S302 , NWDAF 2 performs data processing 1 on data set 1 according to instruction information 1 to obtain data set 2 (eg, a second data set).

[0120] Specifically, Dataset 1 is a dataset obtained by NWDAF2 for training Model 1 from NWDAF1. NWDAF2 can obtain Dataset 1 from other network elements. For example, NWDAF2 can obtain Dataset 1 that can be used by NWDAF2 to train Model 1 from NRF or other NFs that provide data.

[0121] Furthermore, NWDAF2 performs data processing 1 on dataset 1 to obtain dataset 2.

[0122] S303 , NWDAF2 trains model 1 according to data set 2 to obtain model 2.

[0123] For details about how NWDAF2 uses dataset 2 to train model 1, please refer to the existing process and will not be repeated here.

[0124] S304 : NWDAF2 sends model 2 to NWDAF1 .

[0125] Correspondingly, NWDAF1 receives model 2.

[0126] Specifically, NWDAF1 receives Model 2, which may be: NWDAF1 receives the model architecture and parameters within the model architecture of Model 2 sent by NWDAF2. The model architecture of Model 1 may be the same as or different from the model architecture of Model 2, which is not limited.

[0127] In the above solution, NWDAF2 can, according to NWDAF1's instructions, process Dataset 1 to obtain Dataset 2 before using Dataset 1 to train Model 1, and then use Dataset 2 to train Model 1. In this way, the data set used for model training is processed before training Model 1. This can effectively ensure the reliability of the data set used for machine learning models, thereby effectively improving the accuracy of the model obtained through federated learning.

[0128] Specifically, when data processing 1 is used to eliminate dangerous data in dataset 1, NWDAF2 can train model 1 based on safe data or reliable data, and model 2 obtained by training based on safe data or reliable data can be more accurately applied to specific scenarios. When data processing 1 is used to eliminate incomplete or incomplete data in dataset 1, NWDAF2 can train model 1 based on complete data or non-incomplete data to obtain model 2, and model 2 obtained by training based on complete data or non-incomplete data can be more accurately applied to specific scenarios. When data processing 1 is used to delete redundant data in dataset 1, NWDAF2 can train model 1 based on non-redundant data, which can effectively optimize the algorithm for model training, thereby making the results more accurate, and thus improving the accuracy of the model obtained through federated learning.

[0129] The method shown in FIG3 is further described below in conjunction with FIG4 .

[0130] FIG4 is a schematic diagram of an interaction flow of a communication method 400 according to an embodiment of the present application. As shown in FIG4 , the method 400 includes:

[0131] S401 . NWDAF1 sends request information 1 to NWDAF2 . Request information 1 is used to request NWDAF2 to train model 1 . Request information 1 includes model 1 and instruction information 1 .

[0132] Correspondingly, NWDAF2 receives the request information 1, and further determines to train the model 1 based on the request information 1.

[0133] Specifically, the request information 1 can be used to trigger or notify NWDAF2 to start a federated learning process based on model 1.

[0134] S402 , NWDAF2 obtains dataset 1.

[0135] Specifically, after NWDAF2 receives the request information 1 , NWDAFA2 interacts with NRF to obtain the dataset 1 for training the model 1 .

[0136] S403 . NWDAF 2 performs data processing 1 on data set 1 according to instruction information 1 to obtain data set 2 .

[0137] For the description of S403 , please refer to the description of S302 and will not be repeated here.

[0138] In a possible embodiment, the request information 1 may also include indication information 2 (such as second indication information), which is used to indicate algorithm 1, and algorithm 1 is used by NWDAF2 to perform data processing 1 on data set 1. In other words, NWDAF2 can use algorithm 1 to perform data processing 1 on data set 1 to obtain data set 2.

[0139] Specifically, NWDAF2 can use the algorithm indicated by NWDAF1 to process dataset 1, thereby improving the efficiency of the above data processing on dataset 1, and enabling dataset 2 to better meet NWDAF1's requirements for datasets used for machine learning models.

[0140] In one possible embodiment, request information 1 may further include a data volume threshold, which instructs NWDAF 2 to train model 1 only when the data volume of dataset 2 is greater than or equal to the data volume threshold. This allows model 1 to be trained based on a certain amount of data, effectively improving the accuracy of the model obtained through federated learning.

[0141] S404. NWDAF2 trains model 1 according to data set 2 to obtain model 2.

[0142] For the description of S404, please refer to the description of S303 and will not be repeated here.

[0143] S405 . NWDAF2 sends response information 1 to NWDAF1 . Response information 1 includes model 2 .

[0144] Accordingly, NWDFA1 receives response information 1 and obtains model 2 based on response information 1 .

[0145] In one possible embodiment, the response information 1 may further include feedback information 1, which is used to instruct NWDAF2 to perform data processing 1 on the data set 1. In this way, NWDFA1 can determine that the model parameters of the trained model 1 are obtained after NWDFA2 performs training using the data set after data processing.

[0146] In a possible embodiment, the response information 1 may further include feedback information 2, where the feedback information 2 is used to indicate that the NWDAF 2 has performed data processing 1 on the dataset 1 using the algorithm 1. In this way, the NWDFA 1 may determine that the NWDFA 2 performs the above-mentioned data processing 1 on the dataset 1 using the algorithm indicated by the NWDFA 1.

[0147] In a possible embodiment, the response information 1 may further include the data volume of the dataset 1 and the data volume of the dataset 2. In this way, the NWDFA1 can observe the data quality when the NWDFA2 collects data, and then determine whether to continue to let the NWDFA2 perform model training.

[0148] It should be noted that the above description uses the example of request information 1 including a data volume threshold and indication information 2. However, this does not necessarily require the data volume threshold and indication information 2 to be included in the same message as indication information 1 and model 1. When request information 1 includes both the data volume threshold and indication information 2, this can effectively reduce signaling overhead.

[0149] In addition, the above description uses the example of response information 1 including feedback information 1 and feedback information 2. However, feedback information 1 and feedback information 2 are not necessarily required to be in the same message as model 2. When response information 1 includes both feedback information 1 and feedback information 2, this can effectively reduce signaling overhead.

[0150] In the above solution, NWDAF2 can, according to NWDAF1's instructions, process Dataset 1 to obtain Dataset 2 before using Dataset 1 to train Model 1, and then use Dataset 2 to train Model 1. In this way, the data set used for model training is processed before training Model 1, which can effectively ensure the reliability of the data set used for model training and thus effectively improve the accuracy of the model obtained through federated learning.

[0151] In method 400 , when NWDAF2 does not use algorithm 1 to perform data processing 1 on dataset 1 and uses algorithm 2 to perform data processing 1 on dataset 1, NWDAF2 may indicate algorithm 2 to NWDAF1 . In this way, NWDAF1 may determine the specific algorithm name of data processing 1 implemented by NWDAF2 .

[0152] It should be noted that when method 400 involves multiple NWDAF2s, NWDAF1 can perform model aggregation processing on the models obtained after training from multiple NWDAF2s, so as to better complete the model training.

[0153] In the embodiments of the present application, the above method embodiments (method 300 and method 400) mainly describe that NWDAF1 sends content to NWDAF2 to instruct data processing of the data set used for model training, and NWDAF2 first processes the data set used for model training before training the model, so that the data set used for model training can meet the reliability requirements, thereby effectively improving the accuracy of the model obtained through federated learning.

[0154] The information names used or employed in the above method embodiments are intended for illustrative purposes only and are not intended to be definitive. For example, indication information 1 may be parameter 1, which indicates that data processing 1 should be performed on data set 1. Another example is indication information 1 may be indicator. Furthermore, the above information may be compatible with information names or communication processes in 5G or future communication architectures. For example, NWDAF1 may send indication information 1 and indication information 2 via a service interface, and this is not a limitation.

[0155] The method shown in FIG. 4 will be further described below in conjunction with FIG. 5 .

[0156] FIG5 is a schematic diagram of an interaction flow of a communication method 500 according to an embodiment of the present application. As shown in FIG5 , the method 500 includes:

[0157] S501. A function consumer sends a request message to a server-side NWDAF.

[0158] Accordingly, the server NWDAF receives a request from the consumer requesting to subscribe to the federated learning service. Furthermore, the server NWDAF can determine the need to initiate a federated learning process based on the request. The request can also include the initial federated learning model to be trained.

[0159] S502: The server NWDAF sends initial federated learning parameter provisions to n client NWDAFs.

[0160] Correspondingly, the n client NWDAFs receive the initial federated learning parameters provisions, which can also be understood as the initial federated learning model.

[0161] It should be noted that the n client NWDAFs are selected by the server NWDAF to participate in federated learning training. These N client NWDAFs are responsible for carrying out the federated learning process. The initial federated learning parameters provisions include an initial federated learning model, which is the model to be trained.

[0162] S503: n client NWDAFs obtain local training datasets from NRF.

[0163] Specifically, each of the n client NWDAFs may collect data from the NRF (or from other NFs) and form a local training dataset, which is used for federated learning training.

[0164] Furthermore, each of the n client NWDAFs trains the initial federated learning model according to the local training dataset obtained by each client to obtain a trained federated learning model.

[0165] S504: Each of the n client NWDAFs sends the trained federated learning model to the server NWDAF.

[0166] Correspondingly, the server NWDAF receives the trained federated learning model sent by each of the n client NWDAFs.

[0167] S505 , server NWDAF performs model aggregation processing.

[0168] After the server NWDAF receives the trained federated learning models sent by n client NWDAFs, the server NWDAF performs model aggregation processing on the received n trained federated learning models, thereby completing the training of the initial federated learning model.

[0169] In the above process, if the server NWDAF does not instruct the n client NWDAFs to process the local training dataset, the accuracy of the trained federated learning model reported by the n client NWDAFs may not be high. Therefore, in S502, this application supports the server NWDAF also sending an indicator to the n client NWDAFs to instruct them to process the data. This can improve the accuracy of the trained federated learning model reported by the n client NWDAFs to a certain extent.

[0170] In addition, the content shown in method 500 can be compatible with the content shown in Figures 3 and 4. For specific descriptions, please refer to the above descriptions and will not be repeated here.

[0171] 3. Communication Device

[0172] To implement the various functions in the above method, the first network element and the second network element may each include hardware structures and / or software modules, and implement the aforementioned functions in the form of hardware structures, software modules, or a combination of hardware structures and software modules. Whether a particular one of the aforementioned functions is implemented in the form of hardware structures, software modules, or a combination of hardware structures and software modules depends on the specific application and design constraints of the technical solution.

[0173] Figure 6 is a schematic block diagram of a communication device 600 according to an embodiment of the present application. The communication device 600 includes a processor 610 and a communication interface 620, which are interconnected via a bus 630. The communication device 600 can be a first network element or a second network element.

[0174] Optionally, the communication device 600 further includes a memory 640 .

[0175] Specifically, the memory 640 includes but is not limited to random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), or portable read-only memory (CD-ROM), and the memory 640 is used for related instructions and data.

[0176] The processor 610 may be one or more central processing units (CPUs). In the case where the processor 610 is a CPU, the CPU may be a single-core CPU or a multi-core CPU.

[0177] When the communication device 600 is a first network element, illustratively, the processor 610 is configured to perform the following operations: sending model 1 and indication information 1; and receiving model 2.

[0178] When the communication device 600 is a second network element, illustratively, the processor 610 is configured to perform the following operations: receive model 1 and indication information 1; send model 2, etc.

[0179] The above contents are merely exemplary descriptions. When the communication device 600 is the first network element or the second network element, it will be responsible for executing the methods or steps related to the first network element and the second network element in the above method embodiments.

[0180] The above description is merely exemplary. For details, please refer to the contents of the above method embodiments. Furthermore, the implementation of each operation in FIG6 may also correspond to the corresponding description of the method embodiments shown in FIG3 to FIG5.

[0181] Figure 7 is a schematic block diagram of a communication device 700 according to an embodiment of the present application. Communication device 700 may be a first network element or a second network element, or a chip or module within the first network element or the second network element. Communication device 700 includes a transmitting unit 710 and a receiving unit 720. The transmitting unit 710 and the receiving unit 720 are described below as examples.

[0182] For example, when the communication device 700 is the first network element, the sending unit 710 is used to send model 1 and indication information 1, and the receiving unit 720 is used to receive model 2.

[0183] For example, when the communication device 700 is a second network element, the sending unit 710 is used to send model 2, and the receiving unit 720 is used to receive model 1 and indication information 1, etc.

[0184] Optionally, the communication device 700 may further include a processing unit 730, which is used to execute the content of the second network element involving processing, coordination and other steps.

[0185] In addition, the communication device 700 may also include a storage unit 740 (regardless of whether it is a network warehouse function network element, a first network element or a second network element), which is used to store a program or code for executing the aforementioned method.

[0186] The above description is provided for illustrative purposes only. When communication device 700 is a first network element or a second network element, it will be responsible for executing the methods or steps associated with the first network element or the second network element in the aforementioned method embodiments. The implementation of each operation in FIG. 7 can also be described with reference to the corresponding description of the methods shown in the aforementioned embodiments, and will not be repeated here.

[0187] Figure 8 is a schematic block diagram of a communication device 800 according to an embodiment of the present application. Communication device 800 can be used to implement the functions of the first network element or the second network element in the above-described method. Communication device 800 can be a chip within the first network element or the second network element. Communication device 800 includes an input / output interface 820 and a processor 810. Input / output interface 820 can be an input / output circuit. Processor 810 can be a signal processor, a chip, or other integrated circuit capable of implementing the method of the present application. Input / output interface 820 is used for inputting or outputting signals or data.

[0188] For example, when the communication device 800 is a first network element, the input / output interface 820 is used to send model 1 and indication information 1, and also to receive model 2. The processor 810 is used to execute part or all of the steps of any method provided in this application.

[0189] For example, when the communication device 800 is a second network element, the input / output interface 820 is used to send model 2 and receive model 1 and indication information 1, etc.

[0190] In one possible implementation, the processor 810 implements the functions of the network warehouse function network element, the first network element, or the second network element by executing instructions stored in the memory.

[0191] Optionally, the communication device 800 further includes a memory.

[0192] Optionally, the processor and memory are integrated together.

[0193] Optionally, the memory is outside the communication device 800 .

[0194] In one possible implementation, the processor 810 may be a logic circuit, which inputs / outputs messages or signals through the input / output interface 820. The logic circuit may be a signal processor, a chip, or other integrated circuit that can implement the method of the embodiment of the present application.

[0195] The above description of the device in FIG. 8 is merely an exemplary description. The device can be used to execute the method described in the above embodiment. For details, please refer to the description of the above method embodiment.

[0196] The present application also provides a chip, including a processor, for calling and executing instructions stored in a memory from the memory, so that a communication device equipped with the chip executes the methods in the above examples.

[0197] The present application also provides another chip, comprising: an input interface, an output interface, and a processor, wherein the input interface, the output interface, and the processor are connected via an internal connection path, and the processor is configured to execute code in a memory. When the code is executed, the processor is configured to execute the methods in the above examples. Optionally, the chip also includes a memory, which is configured to store computer programs or code.

[0198] The present application also provides a processor for coupling with a memory, and for executing the methods and functions involving a network device or a terminal device in any of the above embodiments.

[0199] In another embodiment of the present application, a computer program product including instructions is provided. When the computer program product is run on a computer, the method of the above embodiment is implemented.

[0200] The present application also provides a computer program. When the computer program is executed in a computer, the method of the aforementioned embodiment is implemented.

[0201] In another embodiment of the present application, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a computer, the method described in the above embodiment is implemented.

[0202] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel 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.

[0203] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0204] In the several embodiments provided in this application, the disclosed systems, devices, and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is merely a logical functional division. In actual implementation, other division methods may be used, such as combining or integrating multiple units or components into another system, or ignoring or not implementing certain features.

[0205] On the other hand, the mutual coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interface, device or unit, which may be electrical, mechanical or other forms.

[0206] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0207] Each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0208] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the various embodiments of the present application. The aforementioned storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a ROM, a RAM, a magnetic disk, or an optical disk.

[0209] The above is only a specific implementation of the embodiment of the present application, but the scope of protection of the embodiment of the present application is not limited to this. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the embodiment of the present application, and they should be included in the scope of protection of the embodiment of the present application. Therefore, the scope of protection of the embodiment of the present application should be based on the scope of protection of the claims.

Claims

1. A communication method, characterized in that: include: The first network element sends a first model and first indication information to the second network element, where the first indication information is used to instruct the second network element to perform data processing on a first data set, where the first data set is a data set acquired by the second network element, and the first data set is used by the second network element to train the first model; The first network element receives a second model sent by the second network element, where the second model is a model obtained after the second network element trains the first model using a second data set, and the second data set is a data set obtained after the second network element performs the data processing on the first data set.

2. The method according to claim 1, characterized in that The method further comprises: The first network element sends second indication information to the second network element, where the second indication information is used to indicate an algorithm, and the algorithm is used by the second network element to perform the data processing on the first data set.

3. The method according to claim 1 or 2, characterized in that: The method further comprises: The first network element sends a data volume threshold to the second network element, and the data volume threshold is used to instruct the second network element to train the machine learning model when the data volume of the second data set is greater than or equal to the data volume threshold.

4. The method according to any one of claims 1 to 3, characterized in that The method further comprises: The first network element receives first feedback information sent by the second network element, where the first feedback information is used to indicate that the second network element has performed the data processing on the first data set.

5. The method according to any one of claims 1 to 4, characterized in that The method further comprises: The first network element receives the data volume of the first data set and the data volume of the second data set sent by the second network element.

6. The method according to any one of claims 2 to 5, characterized in that The method further comprises: The first network element receives second feedback information sent by the second network element, where the second feedback information is used to indicate that the second network element has performed the data processing on the first data set using the algorithm.

7. The method according to any one of claims 1 to 6, characterized in that The method further comprises: The first network element sends a request message to the second network element, where the request message is used to request the second network element to train the first model, and the request message includes the first model and the first indication information.

8. A communication method, characterized in that: include: The second network element receives the first model and first indication information sent by the first network element, where the first indication information is used to instruct the second network element to perform data processing on a first data set, where the first data set is a data set acquired by the second network element, and the first data set is used by the second network element to train the first model; The second network element performs the data processing on the first data set according to the first indication information to obtain a second data set; The second network element trains the first model according to the second data set to obtain a second model; The second network element sends the second model to the first network element.

9. The method according to claim 8, characterized in that The method further comprises: The second network element receives second indication information sent by the first network element, where the second indication information is used to indicate an algorithm, and the algorithm is used by the second network element to perform the data processing on the first data set.

10. The method according to claim 8 or 9, characterized in that: The method further comprises: The second network element receives a data volume threshold sent by the first network element, where the data volume threshold is used to instruct the second network element to train the first model when the data volume of the second data set is greater than or equal to the data volume threshold.

11. The method according to any one of claims 8 to 10, characterized in that The method further comprises: The second network element sends first feedback information to the first network element, where the first feedback information is used to indicate that the second network element has performed the data processing on the first data set.

12. The method according to any one of claims 8 to 11, characterized in that The method further comprises: The second network element sends the data volume of the first data set and the data volume of the second data set to the first network element.

13. The method according to any one of claims 9 to 12, characterized in that The method further comprises: The second network element sends second feedback information to the first network element, where the second feedback information is used to indicate that the second network element has performed the data processing on the first data set using the algorithm.

14. The method according to any one of claims 8 to 13, characterized in that The method further comprises: The second network element receives request information sent by the first network element, where the request information is used to request the second network element to train the first model, and the request information includes the first model and the first indication information.

15. A communication device, characterized in that: comprising a processor configured to, by executing a computer program or instructions, or, by means of a logic circuit, enabling the communication device to perform the method according to any one of claims 1 to 7; or, The communication device is enabled to execute the method according to any one of claims 8 to 14.

16. The communication device according to claim 15, characterized in that: The communication device further comprises a memory for storing the computer program or instructions.

17. The communication device according to claim 15 or 16, characterized in that: The communication device further comprises a communication interface, and the communication interface is used for inputting and / or outputting signals.

18. A communication device, characterized in that: It includes a logic circuit and an input / output interface, wherein the input / output interface is used to input and / or output signals. The logic circuit is used to execute the method according to any one of claims 1 to 7; or, The logic circuit is configured to execute the method according to any one of claims 8 to 14 .

19. A computer-readable storage medium, characterized in that: The computer readable storage medium stores a computer program or instruction. When the computer program or instruction is executed on a computer, so that the method according to any one of claims 1 to 7 is performed, or The method according to any one of claims 8 to 14 is performed.

20. A computer program product, characterized in that Contains instructions that, when executed on a computer, so that the method according to any one of claims 1 to 7 is performed, or The method according to any one of claims 8 to 14 is performed.

21. A communication system, characterized in that: include: a first network element and a second network element; The first network element is used to send a first model and first indication information to a second network element, where the first indication information is used to instruct the second network element to perform data processing on a first data set, where the first data set is a data set acquired by the second network element, and the first data set is used by the second network element to train the first model; The second network element is used to receive the first model and the first indication information; The second network element is configured to perform the data processing on the first data set according to the first indication information to obtain a second data set; The second network element is used to train the first model according to the second data set to obtain a second model; The second network element is used to send the second model to the first network element; The first network element is used to receive the second model.

Images