Federal learning-based network data analysis processing method and related device

Abstract

The embodiment of the invention discloses a federated learning-based network data analysis processing method and a related device. An AF network element respectively sends information identifiers to a first NWDAF network element and a second NWDAF network element, then receives intersection identifiers respectively transmitted by the first NWDAF network element and the second NWDAF network element, executes information alignment processing according to the obtained intersection identifiers, and executes training processing of a local model according to an aligned data set. And obtaining intermediate information representing model loss information, and executing model training corresponding to federated learning according to the intermediate information obtained by executing local model training by each NWDAF network element through the aligned data set and the intermediate information of the NWDAF network element. Therefore, the embodiment of the invention provides a technical scheme of performing federated learning between the AF network element and the plurality of NWDAF network elements, and even in a scene that the AF network element or the NWDAF does not support sharing of own information, based on a federated learning mode, the AF network element can still perform information exchange with different NWDAF network elements.

Description

Technical Field

[0001] The present application relates to the field of communication technology, and in particular to a network data analysis and processing method based on federated learning and related devices. Background Art

[0002] The 3rd Generation Partnership Project (3GPP) defined the architectural enhancements of the Roaming Exchange Network Data Analytics Function (NWDAF) network element in its 18th technical specification version (R18), which is used to support data or analysis exchange between operators.

[0003] However, in a scenario where an application function (AF) network element or an NWDAF network element does not support sharing its own information, how to exchange information between the AF network element and different NWDAF network elements is still a technical problem to be solved. Summary of the Invention

[0004] To solve the above technical problems, the embodiments of the present application provide a network data analysis and processing method and device based on federated learning, an electronic device, a computer-readable storage medium, and a computer program product.

[0005] In the first aspect, an embodiment of the present application provides a network data analysis and processing method based on federated learning, which is applied to an application function network element, and the method includes: sending information identifiers to a first network data analysis network element and a second network data analysis network element respectively; receiving intersection identifiers transmitted respectively by the first network data analysis network element and the second network data analysis network element; wherein the intersection identifier is determined by each network data analysis network element performing information alignment processing according to the information identifier; performing information alignment processing according to the obtained intersection identifier to obtain an aligned data set, and performing local model training processing through the aligned data set to obtain intermediate information characterizing model loss information; performing model training corresponding to federated learning according to the intermediate information obtained by each network data analysis network element through training the local model through its own aligned data set, as well as its own intermediate information.

[0006] In the second aspect, an embodiment of the present application provides a network data analysis and processing device based on federated learning, which is applied to an application function network element, and the device includes: a sending module, configured to send information identifiers to a first network data analysis network element and a second network data analysis network element respectively; a first receiving module, configured to receive intersection identifiers transmitted respectively by the first network data analysis network element and the second network data analysis network element; wherein the intersection identifier is determined by each network data analysis network element performing information alignment processing according to the information identifier; a first training module, configured to perform information alignment processing according to the obtained intersection identifier to obtain an aligned data set, and perform training processing of a local model through the aligned data set to obtain intermediate information representing model loss information; a second training module, configured to perform model training corresponding to federated learning based on the intermediate information obtained by each network data analysis network element training the local model through its own aligned data set, as well as its own intermediate information.

[0007] On the third aspect, an embodiment of the present application provides another network data analysis and processing method based on federated learning, which is applied to a network data analysis network element, and the method includes: receiving an information identifier from an application function network element; performing information alignment processing on a local data set according to the information identifier to obtain an intersection identifier and an aligned data set; transmitting the intersection identifier to the application function network element, so that the application function network element performs information alignment processing according to the intersection identifiers from different network data analysis network elements, and performs local model training processing according to the data set obtained by performing the information alignment processing; performing local model training processing according to the aligned data set to obtain intermediate information representing model loss information, and sending the intermediate information to the application function network element, so that the application function network element performs model training corresponding to federated learning according to the intermediate information of each network data analysis network element and its own intermediate information.

[0008] In a fourth aspect, an embodiment of the present application provides another network data analysis and processing device based on federated learning, which is applied to a network data analysis network element, and the device includes: a second receiving module, configured to receive an information identifier from an application function network element; an alignment module, configured to perform information alignment processing on a local data set according to the information identifier, to obtain an intersection identifier and an aligned data set; a first transmission module, configured to transmit the intersection identifier to the application function network element, so that the application function network element performs information alignment processing according to the intersection identifiers from different network data analysis network elements, and performs local model training processing according to the data set obtained by performing the information alignment processing; a third training module, configured to perform local model training processing according to the aligned data set, to obtain intermediate information characterizing model loss information, and to send the intermediate information to the application function network element, so that the application function network element performs model training corresponding to federated learning according to the intermediate information of each network data analysis network element and its own intermediate information.

[0009] In the fifth aspect, an embodiment of the present application provides another network data analysis and processing method based on federated learning, which is applied to a network open function network element. The method includes: transmitting the information identifier sent by the application function network element to the first network data analysis network element and the second network data analysis network element; transmitting the intersection identifier sent by each network data analysis network element to the application function network element, so that the application function network element performs information alignment processing according to the received intersection identifier, and performs local model training processing according to the data set obtained by performing the information alignment processing, and correspondingly obtains intermediate information representing the model loss information, and executes the corresponding model of federated learning based on the intermediate information of each network data analysis network element and its own intermediate information; wherein, the intermediate information of each network data analysis network element is obtained by each network data analysis network element training the local model through its own aligned data set.

[0010] In the sixth aspect, an embodiment of the present application provides another network data analysis and processing device based on federated learning, which is applied to a network open function network element, and the device includes: a second transmission module, configured to transmit the information identifier sent by the application function network element to the first network data analysis network element and the second network data analysis network element; a third transmission module, configured to transmit the intersection identifier sent by each network data analysis network element to the application function network element, so that the application function network element performs information alignment processing according to the received intersection identifier, and performs local model training processing according to the data set obtained by performing the information alignment processing, and accordingly obtains intermediate information representing the model loss information, and performs model training corresponding to federated learning according to the intermediate information of each network data analysis network element and its own intermediate information; wherein, the intermediate information of each network data analysis network element is obtained by each network data analysis network element training the local model through its own aligned data set.

[0011] In the seventh aspect, an embodiment of the present application provides an electronic device, comprising: one or more processors; a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device implements the network data analysis and processing method based on federated learning as described above.

[0012] In an eighth aspect, an embodiment of the present application provides a computer-readable storage medium having computer-readable instructions stored thereon. When the computer-readable instructions are executed by a processor of a computer, the computer executes the network data analysis and processing method based on federated learning as described above.

[0013] In a ninth aspect, an embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the network data analysis and processing method based on federated learning as described above.

[0014] In the technical solution provided in the embodiments of the present application, a technical solution for performing federated learning between the AF network element and multiple NWDAF network elements is proposed. Even in a scenario where the AF network element or NWDAF does not support sharing its own information, the federated learning model can still enable the AF network element to exchange information with different NWDAF network elements, thereby meeting business needs.

[0015] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a diagram of an exemplary dataset corresponding to horizontal federated learning;

[0017] Figure 2This is a diagram of an exemplary dataset corresponding to vertical federated learning;

[0018] Figure 3 This is an interactive flow chart for implementing information exchange between an AF network element and different NWDAF network elements based on a federated learning framework, proposed in an exemplary embodiment of the present application;

[0019] Figure 4 This is a schematic diagram of an exemplary service architecture of an NWDAF network element;

[0020] Figure 5 is a flowchart of a network data analysis and processing method based on federated learning, shown in an exemplary embodiment of the present application;

[0021] Figure 6 is a flowchart of a network data analysis and processing method based on federated learning shown in another exemplary embodiment of the present application;

[0022] Figure 7 is a flowchart of a network data analysis and processing method based on federated learning shown in another exemplary embodiment of the present application;

[0023] Figure 8 is a block diagram of a network data analysis and processing device based on federated learning, shown in an exemplary embodiment of the present application;

[0024] Figure 9 is a block diagram of a network data analysis and processing device based on federated learning, shown in another exemplary embodiment of the present application;

[0025] Figure 10 is a block diagram of a network data analysis and processing device based on federated learning, shown in another exemplary embodiment of the present application;

[0026] Figure 11 A schematic diagram of the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0027] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. When the following description refers to the drawings, identical numerals in different figures represent identical or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments consistent with the present application. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present application, as detailed in the appended claims.

[0028] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. That is, these functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0029] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be combined or partially combined. Therefore, the actual execution order may vary depending on the actual situation.

[0030] In this application, "plurality" refers to two or more. "And / or" describes the relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can mean: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally indicates that the related objects are in an "or" relationship.

[0031] The terms "first," "second," "third," and "fourth," etc., in the specification and claims of this application and the accompanying drawings are used to distinguish different objects, not to describe a specific order. The terms "including," "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus comprising a series of steps or elements is not limited to the listed steps or elements, but may optionally include steps or elements not listed, or may optionally include other steps or elements inherent to the process, method, product, or apparatus.

[0032] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program that has a predetermined function and works together with other related parts to achieve a predetermined goal, and can be implemented in whole or in part by using software, hardware (such as processing circuits or memories) or a combination thereof. Similarly, a processor (or multiple processors or memories) can be used to implement one or more modules or units. In addition, each module or unit can be part of an overall module or unit that includes the function of the module or unit.

[0033] First, it's important to note that federated learning (FL) is a machine learning method used to train a model so that it can learn from data from multiple parties without centralizing the data in a single location. This approach is suitable for situations involving privacy-sensitive data because it allows model training without sharing the original data. In a federated learning model, the model is iteratively trained across multiple parties, with each party sharing only updates to the model parameters, not the original data. Therefore, this approach helps protect individual privacy while achieving global improvements to the model.

[0034] According to the distribution of training data in the data feature space and sample identity document (ID) space among different participants, federated learning can be divided into two methods: horizontal federated learning (HFL) and vertical federated transfer learning (FTL).

[0035] The data sets of each parameter party used in horizontal federated learning have the same feature space and different sample spaces, so horizontal federated learning is also called federated learning divided by samples. Figure 1 This is a diagram of an exemplary dataset corresponding to horizontal federated learning. Figure 1 As can be seen, the training data used in horizontal federated learning is data features that are aligned across all participants. Horizontal federated learning is suitable for scenarios where the datasets of the participants have significant overlap in feature dimensions. For example, participants A and B are two banks in different regions. Their user groups are distinct and have little overlap, but their businesses are similar. Therefore, the recorded data features are essentially the same, with significant overlap in feature dimensions.

[0036] The data sets of each parameter party used in vertical federated learning have different feature spaces but the same sample space, so vertical federated learning is also called feature-divided federated learning. Figure 2 This is a diagram of an exemplary dataset corresponding to vertical federated learning. Figure 2 As can be seen, the training data used in vertical federated learning consists of data samples aligned across all participants. Vertical federated learning is suitable for scenarios where the datasets of the various participants overlap significantly in the sample dimension. For example, participant C is a bank in a certain region, and participant D is an e-commerce company in the same region. Their respective user groups likely include the majority of residents in that region, resulting in a significant user overlap. However, due to the significant differences between banking and e-commerce businesses, the data characteristics recorded by them differ significantly.

[0037] Therefore, it can be understood more simply that horizontal federated learning refers to multiple participants having different sample IDs but the same characteristics; vertical federated learning refers to multiple participants having the same sample ID but different characteristics.

[0038] In order to solve the problem of how to exchange information between AF network elements and different NWDAF network elements in a scenario where AF network elements or NWDAF network elements do not support sharing their own information, the embodiment of the present application proposes a technical solution for realizing information exchange between AF network elements and different NWDAF network elements based on the framework of federated learning.

[0039] See first Figure 3 , Figure 3 This is an interactive flow chart of information exchange between an AF network element and different NWDAF network elements based on a federated learning framework proposed in an exemplary embodiment of the present application.

[0040] It should be understood that in this embodiment, the NWDAF-1 and NWDAF-2 network elements, acting as the first and second participants in federated learning, respectively, do not share raw data externally. The NWDAF-1 and NWDAF-2 network elements represent two networks, which can be managed by the same operator or by different operators, without limitation. One of the NWDAF-1 and NWDAF-2 network elements may contain label information, which is also not a limitation.

[0041] It is also important to understand that although Figure 3 The diagram only illustrates information interaction between the AF network element and two NWDAF network elements, but does not limit the AF network element to only be able to interact with two NWDAF network elements. In actual application scenarios, information interaction between two or more NWDAF network elements and the AF network element may be supported.

[0042] like Figure 3As shown, the AF network element first sends a request to the Network Exposure Function (NEF) network element to represent the subscription of analysis information to the NWDAF-1 network element and the NWDAF-2 network element. Exemplarily, the AF network element can initiate a call to the NEF call service to implement the request to send the subscription analysis information to the NEF network element. The NEF call service is, for example, Nnef_AnalyticsExposure_Subscribe or Nnef_AnalyticsExposure_Unsubscribe. The analysis information can be understood as including the intersection identifier determined after the NWDAF-1 network element and the NWDAF-2 network element perform information alignment processing, and the intermediate information obtained by the NWDAF-1 network element and the NWDAF-2 network element training the local model based on the data set obtained by performing the information alignment processing. The meaning of the intersection identifier and the intermediate information will be introduced in detail in the subsequent content and will not be repeated here.

[0043] It can be understood that in the technical solution proposed in this application, the NEF network element is responsible for controlling the analysis and open mapping between the AF network element and the NWDAF network element. The AF network element will not request the original information of the network. At the same time, the AF network element will not expose its own original information to the network, thereby ensuring that the AF network element and the NWDAF network element can exchange information while protecting privacy.

[0044] After receiving the request from the AF network element to subscribe to analytical information, the NEF network element authorizes the AF network element and verifies whether the user has access rights to the system. If the user is found to have the corresponding rights, the NEF network element subscribes to the analytical information in the NWDAF-1 network element and the NWDAF-2 network element. For example, the NEF network element can subscribe to the analytical information in the NWDAF-1 network element and the NWDAF-2 network element by calling the Nnwdaf_AnalyticsSubscription_Subscribe or Nnwdaf_AnalyticsSubscription_Unsubscribe service.

[0045] When the NEF network element subscribes to the analysis information, the NWDAF-1 network element and the NWDAF-2 network element will return a notification message indicating that the subscription is successful to the NEF network element. For example, the NWDAF-1 network element and the NWDAF-2 network element can return the notification message to the NEF network element by calling the Nnwdaf_AnalyticsSubscription_Notify service.

[0046] When the NEF network element receives notification messages from the NWDAF-1 network element and the NWDAF-2 network element, the NEF network element returns a notification message indicating that the subscription to the analysis information is successful to the AF network element. For example, the NEF network element can send a notification message to the AF by calling the Nnwdaf_AnalyticsSubscription_Notify service.

[0047] The NEF network element then sends requests for obtaining analysis information to the NWDAF-1 network element and the NWDAF-2 network element respectively. For example, the Nnwdaf_AnalyticsInfo_Reqest service can be called to request the NWDAF-1 network element and the NWDAF-2 network element to obtain analysis information, so that the NWDAF-1 network element and the NWDAF-2 network element start data collection.

[0048] The NWDAF-1 and NWDAF-2 network elements collect data to obtain information required for federated learning. This application does not limit the specific type of information required for federated learning. For example, the information may include Chartered Quality Institute (CQI) information or other network information defined in the 3GPP R18 specification.

[0049] NWDAF-1 network elements and NWDAF-2 network elements can collect data from other network elements or designated communication services, and this application does not limit the source of data collection. For example, in the specifications defined by 3GPP R18, an NWDAF network element can collect information from other NWDAF network elements, subject to legal provisions. For example, the law stipulates that NWDAF network elements can collect information from other NWDAF network elements in the same area, but cannot collect information from other NWDAF network elements in different areas. For another example, NWDAF network elements can also collect information from network elements such as NEF, Session Management Function (SMF), Access and Mobility Management Function (AMF), and Policy Control Function (PCF). Alternatively, NWDAF network elements can also collect information from Operation Administration and Maintenance (OAM) network elements. Alternatively, when the user terminal (UE) is related to the data analysis business, the NWDAF network element can also collect information from OTT (Over-The-Top) services. It can be understood that OTT services refer to value-added services provided by content service providers on the Internet, rather than directly provided by Internet service providers.

[0050] Next, the AF network element sends an information identifier to the NWDAF-1 network element and the NWDAF-2 network element, respectively. The information identifier can include at least one of a sample identifier and a feature identifier. The sample identifier is also known as the sample ID, and the feature identifier can be information such as a feature name. The information identifier is understood to be the information identifier contained in the local dataset of the AF network element. For example, assuming that the local dataset of the AF network element consists of feature data of 100 user samples, the information identifier includes the sample IDs corresponding to these 100 user samples and the feature names corresponding to each feature data item.

[0051] The NWDAF-1 network element and the NWDAF-2 network element each perform information alignment processing based on the information identifier sent by the AF network element. The information alignment processing performed by each NWDAF network element means finding the data subset to be used for federated learning from the local data set. This data subset can be called the aligned data set, and the information identifier contained in the aligned data set is the intersection identifier, that is, the intersection identifier refers to the information identifier that intersects in different data sets. It should be noted that the local data sets of the NWDAF-1 network element and the NWDAF-2 network element are obtained by the data collection performed by the NWDAF-1 network element and the NWDAF-2 network element respectively.

[0052] Information alignment processing includes at least one of sample alignment processing and feature alignment processing. Sample alignment processing corresponds to the vertical federated learning mode, and feature alignment processing corresponds to the horizontal federated learning mode. Exemplarily, the NWDAF network element performs sample alignment processing, which is to find a data subset that matches the sample ID from the local data set as the aligned data set; the NWDAF network element performs feature alignment processing, which is to find a data subset that matches the feature name from the local data set as the aligned data set. It can also be understood that information alignment is a privacy intersection process, which aligns the information identifier sent by the AF network element for federated learning with its own data set, eliminates unnecessary samples or features, and leaves only the data set used for federated learning.

[0053] The NWDAF-1 network element and the NWDAF-2 network element each perform information alignment processing. After obtaining the aligned data set, they determine the intersection identifier by finding the sample IDs and feature names in the aligned data set, and transmit the determined intersection identifier to the AF network element.

[0054] After receiving the intersection identifiers returned by NWDAF-1 and NWDAF-2, the AF network element also needs to perform information alignment. Similarly, the AF network element performs information alignment by matching its own dataset with the intersection identifier, thereby eliminating unnecessary data and retaining the dataset required for federated learning.

[0055] The following processing can be understood as the process of executing federated learning, including:

[0056] The AF network element uses the dataset obtained from the information alignment process to train its local model, and accordingly obtains intermediate information. Furthermore, the NWDAF-1 network element uses its own aligned dataset to train its local model, and accordingly obtains intermediate information. Furthermore, the NWDAF-2 network element uses its own aligned dataset to train its local model, and accordingly obtains intermediate information. It is understood that intermediate information, such as gradient information, private key encryption information, and other information representing model loss information, is also information required in the federated learning process and is not limited here.

[0057] Figure 4 This is a schematic diagram of an exemplary NWDAF network element service architecture. Figure 4It can be seen that the NWDAF network element includes: Analytics Logical Function (AnLF), Model Training Logical Function (MTLF), Data Collection Coordination and Delivery Function (DCCF), Messaging Framework Adaptor Function (MFAF), and Analytics Data Repository Function (ADRF). It can be understood that AnLF is responsible for model reasoning and provides NWDAF service interfaces such as "Nnwdaf_AnalyticsSubscription" and "Nnwdaf_AnalyticsInfo" to the outside world. It can generate analysis results based on the request of the consumer network element. The analysis results include static statistical data and dynamic reasoning results. MTLF is responsible for model training and can provide the trained model to AnlF. It can also be understood that AnLF is the only consumer network element provided by MTLF. Therefore, the NWDAF-1 network element and NWDAF-2 network element in this application can use the MTLF module to execute model training-related processes.

[0058] The AF network element sends intermediate information acquisition requests to the NWDAF-1 network element and the NWDAF-2 network element respectively through the NEF network element to request to obtain the intermediate information obtained by the NWDAF-1 network element and the NWDAF-2 network element respectively performing local model training.

[0059] After receiving the intermediate information acquisition request from the AF network element, the NWDAF-1 network element and the NWDAF-2 network element transmit their own intermediate information to the AF network element.

[0060] After receiving the intermediate information sent by NWDAF-1 and NWDAF-2, the AF network element performs the model training corresponding to federated learning based on its own intermediate information and the intermediate information sent by NWDAF-1 and NWDAF-2. After obtaining the trained model, the AF can execute the model deduction process.

[0061] From the above, it can be seen that in the technical architecture of the federated learning between the AF network element and multiple NWDAF network elements proposed in this application, even in the scenario where the AF network element or NWDAF does not support sharing its own information, by Figure 3The process of information exchange between the AF network element and different NWDAF network elements shown enables the AF network element to obtain a model trained jointly with the NWDAF-1 network element and the NWDAF-2 network element, thereby meeting service requirements.

[0062] based on Figure 3 The proposed technical architecture for performing federated learning between an AF network element and multiple NWDAF network elements, an exemplary embodiment of the present application also proposes the following Figure 5 The network data analysis and processing method based on federated learning is shown in FIG. Figure 5 As shown, the exemplary network data analysis and processing method based on federated learning is applied to AF network elements, including S510-S540, and is described in detail as follows:

[0063] S510: Send information identifiers to the first network data analysis network element and the second network data analysis network element respectively.

[0064] First of all, it should be noted that the first network data analysis network element mentioned in this embodiment is equivalent to Figure 3 The schematic NWDAF-1 network element, the second network data analysis network element is equivalent to Figure 3 The illustrated NWDAF-2 network element. This embodiment does not limit the number of the first network data analysis network element and the second network data analysis network element, that is, the number of the first network data analysis network element and the second network data analysis network element is at least one. The first network data analysis network element and the second network data analysis network element do not share original data externally and have high data privacy. One of the first network data analysis network element and the second network data analysis network element has label data. The first network data analysis network element and the second network data analysis network element correspond to two different networks. The two networks can be managed by the same operator or by different operators. This embodiment does not limit this.

[0065] This embodiment is based on a federated learning model. First, the AF network element sends an information identifier to the first network data analysis network element and the second network data analysis network element, respectively, so that the first network data analysis network element and the second network data analysis network element perform information alignment processing based on the received information identifier. Exemplarily, the information identifier includes at least one of a sample identifier and a feature identifier.

[0066] It should be noted that the information exchange between the AF network element and the first network data analysis network element and the second network data analysis network element can be implemented through the NEF network element. It is understandable that the NEF network element is responsible for controlling the analysis and open mapping between the AF network element and the NWDAF network element. The AF network element will not request the original information of the network, and at the same time, the AF network element will not expose its own original information to the network, thereby ensuring that the AF network element and the NWDAF network element can exchange information while protecting privacy.

[0067] In addition, the AF network element can also request the NEF network element to subscribe to the analysis information in the first network data analysis network element and the second network data analysis network element, so that the NEF network element can perform the subscription processing of the analysis information in each network data analysis network element respectively. After the AF network element receives the notification message returned by the NEF network element indicating that the subscription to the analysis information is successful, it can be indicated that the AF network element has established a connection with the first network data analysis network element and the second network data analysis network element through the NEF network element. Therefore, the AF network element can subsequently exchange information with the first network data analysis network element and the second network data analysis network element through the NEF network element.

[0068] As another exemplary implementation, correspondingly, the AF network element may unsubscribe from the analysis information in the first network data analysis network element and the second network data analysis network element in the same manner, and this process will not be described in detail here.

[0069] S520, receiving intersection identifiers transmitted respectively by the first network data analysis network element and the second network data analysis network element; wherein the intersection identifier is determined by each network data analysis network element performing information alignment processing according to the information identifier.

[0070] The first network data analysis network element and the second network data analysis network element perform information alignment based on the received information identifier. This means that each network data analysis network element finds a data subset that matches the information identifier from the local data set as the aligned data set. The aligned data set is also the data set to be used for federated learning. It can also be understood that information alignment is a privacy intersection process. Each network data analysis network element aligns its own data set based on the information identifier sent by the AF network element for federated learning, eliminating unnecessary samples or features, and leaving only the data set used for federated learning.

[0071] Information alignment processing includes at least one of sample alignment processing and feature alignment processing. Sample alignment processing corresponds to the vertical federated learning model, and feature alignment processing corresponds to the horizontal federated learning model. Exemplarily, each network data analysis network element performs sample alignment processing to find a data subset that matches the sample ID from its respective local data set as the aligned data set; each network data analysis network element performs feature alignment processing to find a data subset that matches the feature name from its respective local data set as the aligned data set.

[0072] The first network data analysis network element and the second network data analysis network element can determine the intersection identifier based on their own aligned data sets and send the intersection identifier to the AF network element. The intersection identifier can be understood as the information identifier contained in the aligned data set. For example, each network data analysis network element finds each sample ID and feature name from the aligned data set and sends these sample IDs and feature names as the intersection identifier to the AF network element.

[0073] S530: Perform information alignment processing according to the obtained intersection identifier to obtain an aligned data set, and perform local model training processing through the aligned data set to obtain intermediate information representing model loss information.

[0074] The AF network element performs information alignment processing locally based on the intersection identifier sent by the first network data analysis network element and the second network data analysis network element, and then performs local model training processing based on the data set obtained by performing the information alignment processing, and accordingly obtains intermediate information representing the model loss information. The process of the AF network element performing information alignment processing is similar to the process of the first network data analysis network element and the second network data analysis network element performing information alignment processing, and will not be repeated here.

[0075] Similarly, after obtaining the data set obtained by executing the information alignment process, the first network data analysis network element and the second network data analysis network element also use the data set to execute the training process of the local model, and will also obtain the intermediate information accordingly.

[0076] The first network data analysis network element and the second network data analysis network element can actively send their own intermediate information to the AF network element, or they can send their own intermediate information to the AF network element after receiving the intermediate information acquisition request from the AF network element. This embodiment does not limit this.

[0077] S540, performing model training corresponding to federated learning based on the intermediate information obtained by each network data analysis network element through training the local model through its own aligned data set, as well as its own intermediate information.

[0078] The AF network element performs the process of federated learning deduction based on the intermediate information of the first network data analysis network element and the second network data analysis network element, as well as its own intermediate information, that is, the process of updating the parameters of the local model by combining these three intermediate information.

[0079] Therefore, for the AF network element, based on the federated learning model, it combines the first network data analysis network element and the second network data analysis network element to complete the training of its own model. During this training process, the original data of the first network data analysis network element and the second network data analysis network element are not leaked, which not only ensures the data privacy of the first network data analysis network element and the second network data analysis network element, but also realizes business needs.

[0080] based on Figure 3 The proposed technical architecture for performing federated learning between an AF network element and multiple NWDAF network elements, another exemplary embodiment of the present application also proposes the following Figure 6 The network data analysis and processing method based on federated learning is shown in FIG. Figure 6 As shown, this exemplary federated learning-based network data analysis and processing method is applied to NWDAF network elements, including S610-S640, and is detailed as follows:

[0081] S610: Receive an information identifier from an application function network element.

[0082] First, it should be noted that in the technical solution provided in this application, the AF network element can exchange information with at least two NWDAF network elements. Each NWDAF network element has a high degree of data privacy and does not share original data externally. The method provided in this embodiment can be executed by any NWDAF network element.

[0083] As described in the above embodiment, the AF network element will send an information identifier to the NWDAF network element with which it exchanges information. Therefore, the NWDAF network element will correspondingly receive the information identifier from the AF network element.

[0084] It should also be noted that information exchange between the AF network element and each NWDAF network element can be achieved through the NEF network element. As can be understood, the NEF network element is responsible for controlling the analysis and open mapping between the AF network element and the NWDAF network element. The AF network element does not request the network's original information, and at the same time, the AF network element does not expose its own original information to the network. This ensures that the AF network element and the NWDAF network element can exchange information while protecting privacy.

[0085] In addition, the AF network element also requests the NEF network element to subscribe to the analysis information in each NWDAF network element, so that the NEF network element can perform the subscription processing of the analysis information in each NWDAF network element respectively. In response to the NEF network element's request to subscribe to the analysis information, the NWDAF network element performs the subscription processing accordingly and returns a notification message indicating that the subscription is successful. The NEF network element also returns a notification message indicating that the subscription is successful to the AF network element accordingly. After the AF network element receives the notification message returned by the NEF network element, it means that the AF network element has established a connection with the corresponding NWDAF network element through the NEF network element. Therefore, the AF network element can subsequently exchange information with each NWDAF network element through the NEF network element. Subsequently, the NEF network element requests each NWDAF network element to obtain the analysis information, so that each NWDAF network element performs data collection accordingly, thereby obtaining a local data set.

[0086] Exemplarily, each NWDAF network element, in response to a request from an NEF network element to obtain analysis information, collects data through at least one preset method to obtain a local data set. The preset methods mentioned herein include, for example, data collection through other NWDAF network elements, data collection through OAM network elements, and data collection through designated communication services (such as OTT services), which are not limited here.

[0087] S620: Perform information alignment processing on the local data set according to the information identifier to obtain an intersection identifier and an aligned data set.

[0088] The NWDAF network element performs information alignment processing on the collected local data set according to the information identifier from the AF network element, and accordingly obtains an aligned data set, and determines the intersection identifier according to the aligned data set.

[0089] It should be noted that the information identifier can include at least one of a sample identifier or a feature identifier. The sample identifier is also known as the sample ID, and the feature identifier can be information such as a feature name. Information alignment processing includes at least one of sample alignment processing and feature alignment processing. Sample alignment processing corresponds to the vertical federated learning model, while feature alignment processing corresponds to the horizontal federated learning model.

[0090] S630, transmit the intersection identifier to the application function network element, so that the application function network element performs information alignment processing according to the intersection identifiers from different network data analysis network elements, and performs local model training processing according to the data set obtained by performing the information alignment processing.

[0091] The NWDAF network element first transmits the intersection identifier to the AF network element, so that the AF network element performs information alignment processing based on the intersection identifiers from different NWDAF network elements, and performs local model training processing based on the data set obtained by performing the information alignment processing, and accordingly obtains intermediate information representing the model loss information.

[0092] S640, performs training processing of the local model according to the aligned data set, obtains intermediate information representing the model loss information, and sends the intermediate information to the application function network element, so that the application function network element performs model training corresponding to federated learning based on the intermediate information of each network data analysis network element and its own intermediate information.

[0093] The NWDAF network element also performs local model training processing based on the aligned data set, obtains corresponding intermediate information, and sends the intermediate information to the AF network element, so that the AF network element updates the parameters of the local model based on the intermediate information of each NWDAF network element and its own intermediate information, thereby obtaining a model trained by multiple NWDAF network elements.

[0094] The NWDAF network element may actively send its own intermediate information to the AF network element, or may transmit its own intermediate information to the AF network element after receiving an intermediate information acquisition request from the AF network element, and this embodiment does not limit this.

[0095] Therefore, for the NWDAF network element, even if it does not share the original data externally, it can still exchange information with the AF network element based on the federated learning model. This enables the AF network element to complete the training of its own model in conjunction with multiple NWDAF network elements, thereby meeting business needs while ensuring the data privacy of each NWDAF network element.

[0096] based on Figure 3 The proposed technical architecture for performing federated learning between an AF network element and multiple NWDAF network elements, another exemplary embodiment of the present application also proposes the following Figure 7 The network data analysis and processing method based on federated learning is shown in FIG. Figure 7 As shown, the exemplary network data analysis and processing method based on federated learning is applied to NEF network elements, including S710-S720, and is described in detail as follows:

[0097] S710: Transmit the information identifier sent by the application function network element to the first network data analysis network element and the second network data analysis network element.

[0098] As mentioned above, in the federated learning model, each participant needs to first perform information alignment processing to obtain the training set required for federated learning. The AF network element needs to send the information identifier used for information alignment processing to the first network data analysis network element and the second network data analysis network element through the NEF network element. Therefore, for the NEF network element, the information identifier sent by the AF network element will be transmitted to the first network data analysis network element and the second network data analysis network element accordingly.

[0099] S720, transmits the intersection identifier sent by each network data analysis network element to the application function network element, so that the application function network element performs information alignment processing according to the received intersection identifier, and performs local model training processing according to the data set obtained by performing the information alignment processing, and accordingly obtains intermediate information representing the model loss information, and performs model training corresponding to federated learning based on the intermediate information of each network data analysis network element and its own intermediate information.

[0100] The first network data analysis network element and the second network data analysis network element each perform information alignment processing based on the information identifier from the AF network element, and correspondingly obtain an aligned data set and an intersection identifier, and return the intersection identifier to the AF network element through the NEF network element. It should be noted that after obtaining the aligned data set, the first network data analysis network element and the second network data analysis network element also use their respective aligned data sets to perform local model training processing, and correspondingly obtain intermediate information representing model loss information.

[0101] The NEF network element transmits the intersection identifiers sent by each network data analysis network element to the AF network element, so that the AF network element also performs information alignment processing based on the received intersection identifier, and trains the local model according to the data set obtained by performing the alignment processing, and also obtains the intermediate information representing the model loss information accordingly. After that, the AF network element performs the model training corresponding to the federated learning based on the intermediate information of each network data analysis network element and its own intermediate information, thereby obtaining a model trained based on the federated learning mode.

[0102] It should be noted that, in another exemplary embodiment, before S710, the NEF network element also responds to the request of the AF network element to subscribe to the analysis information, and subscribes to the analysis information in each network data analysis network element respectively, and after receiving the notification message indicating the successful subscription returned by each network data analysis network element, requests each network data analysis network element to obtain the analysis information, so that each network data analysis network element performs data collection to obtain the local data set required for performing information alignment processing. As a result, the NEF network element acts as a bridge for the AF network element to exchange information with the first network data analysis network element and the second network data analysis network element. In the process of uniting the AF network element with the first network data analysis network element and the second network data analysis network element, the NEF network element discovers potential network data analysis network elements for federated learning.

[0103] In another exemplary embodiment, after receiving the request for subscription analysis information initiated by the AF network element, the NEF network element also performs verification processing of the access rights corresponding to the AF network element. Only after the verification is passed does it respond to the request for subscription analysis information initiated by the AF network element to ensure system stability.

[0104] It should also be noted that more detailed details involved in this embodiment have been recorded in the aforementioned embodiments, so they will not be repeated in this embodiment.

[0105] Figure 8 FIG is a block diagram of a network data analysis and processing device based on federated learning, shown as an exemplary embodiment of the present application. Figure 8 As shown, the device is applied to an AF network element, and the device includes:

[0106] A sending module 810 is configured to send an information identifier to the first network data analysis network element and the second network data analysis network element respectively;

[0107] The first receiving module 820 is configured to receive the intersection identifiers transmitted by the first network data analysis network element and the second network data analysis network element respectively; wherein the intersection identifier is determined by each network data analysis network element performing information alignment processing according to the information identifier;

[0108] A first training module 830 is configured to perform information alignment processing based on the obtained intersection identifier to obtain an aligned data set, and perform local model training processing using the aligned data set to obtain intermediate information representing model loss information;

[0109] The second training module 840 is configured to perform model training corresponding to federated learning based on the intermediate information obtained by each network data analysis network element through training the local model through its own aligned data set, as well as its own intermediate information.

[0110] In another exemplary embodiment, the sending module 810 is further configured to: transmit the information identifier to each network data analysis network element respectively through the network open function network element.

[0111] In another exemplary embodiment, the apparatus further includes a request module, wherein the request module is configured to:

[0112] Requesting the network open function network element to subscribe to analysis information in the first network data analysis network element and the second network data analysis network element, so that the network open function network element performs subscription processing for the analysis information in each network data analysis network element respectively;

[0113] Receive the notification message of successful subscription to the characterization analysis information returned by the network open function network element.

[0114] In another exemplary embodiment, the second training module 840 is further configured to:

[0115] Request intermediate information from each network data analysis element;

[0116] Receive the intermediate information returned by each network data analysis network element, and update the parameters of the local model based on the received intermediate information and its own intermediate information.

[0117] In another exemplary embodiment, the information identification includes at least one of a sample identification or a feature identification, and the information alignment processing includes at least one of a sample alignment processing or a feature alignment processing.

[0118] Figure 9 FIG. 1 is a block diagram of a network data analysis and processing device based on federated learning, shown in another exemplary embodiment of the present application. Figure 9 As shown, the device is applied to the NWDAF network element, and the device includes:

[0119] A first receiving module 910 is configured to receive an information identifier from an application function network element;

[0120] An alignment module 920 is configured to perform information alignment processing on the local data set according to the information identifier to obtain an intersection identifier and an aligned data set;

[0121] The first transmission module 930 is configured to transmit the intersection identifier to the application function network element, so that the application function network element performs information alignment processing according to the intersection identifiers from different network data analysis network elements, and performs local model training processing according to the data set obtained by performing the information alignment processing;

[0122] The third training module 940 is configured to perform training processing of the local model based on the aligned data set, obtain intermediate information representing the model loss information, and send the intermediate information to the application function network element, so that the application function network element performs federated learning deduction based on the intermediate information of each network data analysis network element and its own intermediate information.

[0123] In another exemplary embodiment, the apparatus further includes a first response module, wherein the first response module is configured to:

[0124] In response to the request of the network open function network element to subscribe to the analysis information, the network open function network element performs the subscription processing accordingly and returns a notification message indicating that the subscription is successful;

[0125] In response to the request of the network open function network element to obtain analysis information, data collection is performed accordingly to obtain a local data set.

[0126] In another exemplary embodiment, the first response module is further configured to: in response to a request from a network open function network element to obtain analysis information, collect data through at least one preset method to obtain a local data set; wherein the preset method includes: collecting data through other network data analysis network elements, collecting data through operation management and maintenance network elements, and collecting data through designated communication services.

[0127] In another exemplary embodiment, the information identifier includes at least one of a sample identifier or a feature identifier; the alignment module 920 is further configured to: perform at least one of sample alignment or feature alignment on the local dataset according to the information identifier to obtain an intersection identifier and an aligned dataset.

[0128] In another exemplary embodiment, the first transmission module 930 is further configured to: after receiving an intermediate information acquisition request from the application function network element, transmit the obtained intermediate information to the application function network element.

[0129] Figure 10 FIG. 1 is a block diagram of a network data analysis and processing device based on federated learning, as shown in another exemplary embodiment of the present application. Figure 10 As shown, the device is applied to a NEF network element and includes:

[0130] The second transmission module 1010 is configured to transmit the information identifier sent by the application function network element to the first network data analysis network element and the second network data analysis network element;

[0131] The third transmission module 1020 is configured to transmit the intersection identifier sent by each network data analysis network element to the application function network element, so that the application function network element performs information alignment processing according to the received intersection identifier, and performs local model training processing according to the data set obtained by performing the information alignment processing, and accordingly obtains intermediate information representing the model loss information, and performs model training corresponding to federated learning based on the intermediate information of each network data analysis network element and its own intermediate information; wherein, the intermediate information of each network data analysis network element is obtained by each network data analysis network element training the local model through its own aligned data set.

[0132] In another exemplary embodiment, the apparatus further includes a second response module, wherein the second response module is configured to:

[0133] In response to the request of the application function network element to subscribe to the analysis information, subscribe to the analysis information in each network data analysis network element respectively;

[0134] After receiving notification messages indicating successful subscription returned by each network data analysis network element, a request is made to each network data analysis network element to obtain analysis information, so that each network data analysis network element performs data collection.

[0135] In another exemplary embodiment, the second response module is further configured to: after receiving a request to subscribe to analysis information, perform verification processing of the access rights corresponding to the application function network element, and after the verification passes, execute a response to the request to subscribe to analysis information. It should be noted that the network data analysis and processing device based on federated learning provided in the above embodiment and the network data analysis and processing method based on federated learning provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs the operation has been described in detail in the method embodiment and will not be repeated here. In actual applications, the network data analysis and processing device based on federated learning provided in the above embodiment can allocate the above functions to different functional modules as needed, that is, divide the internal structure of the device into different functional modules to complete all or part of the functions described above, and this is not limited here.

[0136] The network data analysis and processing device based on federated learning provided in the above embodiment, based on the federated learning model, combines the AF network element and multiple NWDAF network elements to complete the training of its own model. During this training process, the original data of each NWDAF network element is not leaked, which not only ensures the data privacy of each NWDAF network element, but also realizes business needs.

[0137] An embodiment of the present application also provides an electronic device, comprising: one or more processors; and a memory for storing one or more programs. When the one or more programs are executed by the one or more processors, the electronic device implements the network data analysis and processing method based on federated learning provided in the above-mentioned embodiments.

[0138] Figure 11 The following is a schematic diagram showing the structure of a computer system suitable for implementing an electronic device according to an embodiment of the present application. Figure 11 The computer system 1100 of the electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0139] like Figure 11 As shown, the computer system 1100 includes a central processing unit (CPU) 1101, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 1102 or the program loaded from the storage part 1108 into the random access memory (RAM) 1103, such as executing the method described in the above embodiment. Various programs and data required for system operation are also stored in the RAM 1103. The CPU 1101, ROM 1102 and RAM 1103 are connected to each other via a bus 1104. An input / output (I / O) interface 1105 is also connected to the bus 1104.

[0140] The following components are connected to the I / O interface 1105: an input section 1106 including a keyboard, a mouse, and the like; an output section 1107 including devices such as a cathode ray tube (CRT), a liquid crystal display (LCD), and a speaker; a storage section 1108 including a hard disk; and a communication section 1109 including a network interface card such as a LAN (Local Area Network) card or a modem. The communication section 1109 performs communication processing via a network such as the Internet. A drive 1110 is also connected to the I / O interface 1105 as needed. Removable media 1111, such as a magnetic disk, an optical disk, a magneto-optical disk, or a semiconductor memory, is installed in the drive 1110 as needed, so that computer programs read from the removable media can be installed in the storage section 1108 as needed.

[0141] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a computer program for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via the communication section 1109, and / or installed from a removable medium 1111. When the computer program is executed by the central processing unit (CPU) 1101, the various functions defined in the system of the present application are executed.

[0142] It should be noted that the computer-readable medium shown in the embodiment of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a flash memory, an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. The computer program contained in the computer-readable medium can be transmitted using any appropriate medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0143] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. Among them, each box in the flowchart or block diagram can represent a module, program segment, or part of the code, and the above-mentioned module, program segment, or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0144] The units involved in the embodiments described in this application may be implemented by software or hardware, and the units described may also be set in a processor. In some cases, the names of these units do not constitute limitations on the units themselves.

[0145] Another aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for analyzing and processing network data based on federated learning. The computer-readable storage medium may be included in the electronic device described in the above embodiments, or may exist independently and not be incorporated into the electronic device.

[0146] Another aspect of the present application provides a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the network data analysis and processing method based on federated learning provided in each of the above embodiments.

[0147] The above content is only a preferred exemplary embodiment of the present application and is not intended to limit the implementation scheme of the present application. Ordinary technicians in this field can easily make corresponding changes or modifications based on the main ideas and spirit of the present application. Therefore, the scope of protection of the present application shall be based on the scope of protection required by the claims.

[0148] It is understandable that in the specific implementation of this application, when it involves information identification, intermediate information, network information and other related data, when the above embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

Claims

1. A network data analysis and processing method based on federated learning, characterized in that: Applied to an application function network element, the method includes: Sending information identifiers to the first network data analysis network element and the second network data analysis network element respectively; Receiving intersection identifiers transmitted respectively by the first network data analysis network element and the second network data analysis network element; wherein the intersection identifier is determined by each network data analysis network element performing information alignment processing according to the information identifier; Perform information alignment processing based on the obtained intersection identifier to obtain an aligned data set, and perform local model training processing on the aligned data set to obtain intermediate information representing model loss information; Based on the intermediate information obtained by each network data analysis element through training the local model through its own aligned data set, as well as its own intermediate information, the model training corresponding to federated learning is performed.

2. The method according to claim 1, characterized in that The sending of information identifiers to the first network data analysis network element and the second network data analysis network element respectively includes: The information identifier is transmitted to each network data analysis network element respectively through the network open function network element.

3. The method according to claim 2, characterized in that Before sending the information identifier to the first network data analysis network element and the second network data analysis network element respectively, the method further includes: Requesting the network open function network element to subscribe to analysis information in the first network data analysis network element and the second network data analysis network element, so that the network open function network element performs subscription processing for analysis information in each network data analysis network element respectively; Receive a notification message of successful subscription to the characterization analysis information returned by the network open function network element.

4. The method according to claim 1, wherein The method of performing model training corresponding to federated learning based on intermediate information obtained by each network data analysis network element through training the local model with its own aligned data set, as well as its own intermediate information, includes: Request intermediate information from each network data analysis element; Receive the intermediate information returned by each network data analysis network element, and update the parameters of the local model based on the received intermediate information and its own intermediate information.

5. The method according to any one of claims 1 to 4, characterized in that The information identification includes at least one of a sample identification or a feature identification, and the information alignment processing includes at least one of a sample alignment processing or a feature alignment processing.

6. A network data analysis and processing method based on federated learning, characterized in that: Applied to a network data analysis network element, the method includes: Receive information identification from application function network element; Performing information alignment processing on the local data set according to the information identifier to obtain an intersection identifier and an aligned data set; Transmitting the intersection identifier to the application function network element, so that the application function network element performs information alignment processing according to the intersection identifiers from different network data analysis network elements, and performs local model training processing according to the data set obtained by performing the information alignment processing; The local model training process is performed according to the aligned data set to obtain intermediate information representing the model loss information, and the intermediate information is sent to the application function network element, so that the application function network element performs the model training corresponding to the federated learning based on the intermediate information of each network data analysis network element and its own intermediate information.

7. The method according to claim 6, characterized in that The method further comprises: In response to the request of the network open function network element to subscribe to the analysis information, the network open function network element performs the subscription processing accordingly and returns a notification message indicating that the subscription is successful; In response to the request of the network open function network element to obtain analysis information, data collection is performed accordingly to obtain the local data set.

8. The method according to claim 7, characterized in that The step of responding to the request of the network open function network element to obtain analysis information and performing data collection accordingly to obtain the local data set includes: In response to the request of the network open function network element to obtain analysis information, data collection is performed through at least one preset method to obtain the local data set; wherein, the preset method includes: data collection through other network data analysis network elements, data collection through operation management and maintenance network elements, and data collection through designated communication services.

9. The method according to claim 6, characterized in that The information identifier includes at least one of a sample identifier and a feature identifier; performing information alignment processing on the local data set according to the information identifier to obtain an intersection identifier and an aligned data set includes: At least one of sample alignment and feature alignment is performed on the local data set according to the information identifier to obtain an intersection identifier and an aligned data set.

10. The method according to claim 6, characterized in that The method further comprises: After receiving the intermediate information acquisition request from the application function network element, the obtained intermediate information is transmitted to the application function network element.

11. A network data analysis and processing method based on federated learning, characterized in that: Applied to a network open function network element, the method includes: Transmitting the information identifier sent by the application function network element to the first network data analysis network element and the second network data analysis network element; The intersection identifier sent by each network data analysis network element is transmitted to the application function network element, so that the application function network element performs information alignment processing according to the received intersection identifier, and performs local model training processing according to the data set obtained by performing the information alignment processing, and correspondingly obtains intermediate information representing the model loss information, and performs model training corresponding to federated learning based on the intermediate information of each network data analysis network element and its own intermediate information; wherein, the intermediate information of each network data analysis network element is obtained by each network data analysis network element training the local model through its own aligned data set.

12. The method according to claim 11, characterized in that Before transmitting the information identifier sent by the application function network element to the first network data analysis network element and the second network data analysis network element, the method further includes: In response to the request of the application function network element to subscribe to the analysis information, subscribe to the analysis information in each network data analysis network element respectively; After receiving notification messages indicating successful subscription returned by each network data analysis network element, a request is made to each network data analysis network element to obtain analysis information, so that each network data analysis network element performs data collection.

13. The method according to claim 12, characterized in that The method further comprises: After receiving the request to subscribe to the analysis information, performing a verification process of the access rights corresponding to the application function network element; After the verification is passed, a response to the request for subscribing to analysis information is executed.

14. A network data analysis and processing device based on federated learning, characterized in that: Applied to an application function network element, the device includes: A sending module configured to send an information identifier to the first network data analysis network element and the second network data analysis network element respectively; A first receiving module is configured to receive intersection identifiers transmitted respectively by the first network data analysis network element and the second network data analysis network element; wherein the intersection identifier is determined by each network data analysis network element performing information alignment processing according to the information identifier; A first training module is configured to perform information alignment processing based on the obtained intersection identifier to obtain an aligned data set, and perform local model training processing using the aligned data set to obtain intermediate information representing model loss information; The second training module is configured to perform model training corresponding to federated learning based on the intermediate information obtained by each network data analysis network element through training the local model through its own aligned data set, as well as its own intermediate information.

15. A network data analysis and processing device based on federated learning, characterized in that: Applied to network data analysis network elements, the device includes: A second receiving module is configured to receive an information identifier from an application function network element; an alignment module configured to perform information alignment processing on the local data set according to the information identifier to obtain an intersection identifier and an aligned data set; a first transmission module configured to transmit the intersection identifier to the application function network element, so that the application function network element performs information alignment processing according to the intersection identifiers from different network data analysis network elements, and performs local model training processing according to the data set obtained by performing the information alignment processing; The third training module is configured to perform training processing of the local model according to the aligned data set, obtain intermediate information representing the model loss information, and send the intermediate information to the application function network element, so that the application function network element performs model training corresponding to federated learning based on the intermediate information of each network data analysis network element and its own intermediate information.

16. A network data analysis and processing device based on federated learning, characterized in that: Applied to a network open function network element, the device includes: a second transmission module configured to transmit the information identifier sent by the application function network element to the first network data analysis network element and the second network data analysis network element; The third transmission module is configured to transmit the intersection identifier sent by each network data analysis network element to the application function network element, so that the application function network element performs information alignment processing according to the received intersection identifier, and performs local model training processing according to the data set obtained by performing the information alignment processing, and accordingly obtains intermediate information representing the model loss information, and performs model training corresponding to federated learning based on the intermediate information of each network data analysis network element and its own intermediate information; wherein, the intermediate information of each network data analysis network element is obtained by each network data analysis network element training the local model through its own aligned data set.

17. An electronic device, characterized in that: include: one or more processors; A memory for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the network data analysis and processing method based on federated learning as described in any one of claims 1 to 13.

18. A computer-readable storage medium, characterized in that Computer-readable instructions are stored thereon, and when the computer-readable instructions are executed by a processor of a computer, the computer is caused to execute the network data analysis and processing method based on federated learning according to any one of claims 1 to 13.

19. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the network data analysis and processing method based on federated learning as described in any one of claims 1 to 13 is implemented.

Images