Network data analysis method and device, electronic equipment and storage medium

Abstract

Embodiments of the present application disclose a network data analysis method and device, equipment and a storage medium, applied to a network artificial intelligence network element, comprising: receiving an artificial intelligence analysis request, the artificial intelligence analysis request comprising a corresponding service type; inputting the service type into a preset service feature extraction model in response to the artificial intelligence analysis request to obtain service features, the service feature extraction model being obtained by training a first training service data contained in a target application layer network element as training data in a federated learning manner, the service type corresponding to the first training service data being the same as the service type included in the artificial intelligence analysis request; matching the service features with user plane data information obtained from a user plane network element, obtaining an analysis result of network data according to the obtained matching result, and sending the analysis result to a consumption network element. The embodiments of the present application can provide more accurate perception analysis for services while protecting the privacy data of the application layer network element.

Description

Technical Field

[0001] This application relates to the field of computer technology, and more specifically, to a method, apparatus, electronic device, and storage medium for analyzing network data. Background Technology

[0002] With the rapid development of internet applications, the number of internet users and the volume of business have increased significantly. However, low-value services occupy a large amount of air interface resources for extended periods, even leading to network congestion, while some high-value services cannot receive timely service due to limited air interface resources, seriously affecting the user experience. How to implement differentiated service quality management for different users and different services, make full use of limited wireless network resources, and improve user experience has become a focus of industry attention. Summary of the Invention

[0003] To address the aforementioned technical problems, embodiments of this application provide a method for analyzing network data, a device for analyzing network data, an electronic device, and a storage medium, which can provide more accurate perception and analysis for services while protecting the privacy data of application layer network elements.

[0004] Other features and advantages of this application will become apparent from the following detailed description, or may be learned in part from practice of this application.

[0005] According to one aspect of the embodiments of this application, a method for analyzing network data is provided, applied to a network artificial intelligence (AI) network element, comprising: receiving an AI analysis request, the AI ​​analysis request including a corresponding service type; responding to the AI ​​analysis request by inputting the service type into a preset service feature extraction model to obtain service features corresponding to the service type, wherein the service feature extraction model is trained using first training service data contained in the target application layer network element as training data and employing a federated learning approach, wherein the service type corresponding to the first training service data is the same as the service type included in the AI ​​analysis request; matching the service features with user plane data information obtained from a user plane network element, obtaining the analysis result of the network data based on the obtained matching result, and sending it to a consumer network element.

[0006] In an exemplary embodiment, before inputting the service type into a preset service feature extraction model in response to an artificial intelligence analysis request, the method further includes: determining a target application layer network element that meets preset conditions; using the first training service data included in the target application layer network element that meets the preset conditions as training data, training the service feature extraction model to be trained using a federated learning approach to obtain a trained service feature extraction model, wherein the first training service data is obtained from the user equipment network element connected to the target application layer network element.

[0007] In an exemplary embodiment, determining a target application layer network element that meets preset conditions includes: obtaining registration information corresponding to the application layer network element, the registration information including sharing parameters used to characterize whether the application layer network element supports service data sharing; and selecting the application layer network element corresponding to the sharing parameters that characterize whether the application layer network element supports service data sharing as the target application layer network element that meets the preset conditions.

[0008] In an exemplary embodiment, the service feature extraction model includes a sub-service feature extraction model constructed by a target application layer network element, and the service feature extraction model includes model parameters of the sub-service feature extraction model. The process involves using first training service data from the target application layer network element that meets preset conditions as training data, and training the service feature extraction model to be trained using federated learning to obtain the trained service feature extraction model. This includes: obtaining the values ​​of the model parameters sent by the target application layer network element, where the values ​​of the model parameters are obtained by the target application layer network element iteratively training the sub-service feature extraction model using the first training service data as training data; inputting the values ​​of the model parameters into the service feature extraction model to be trained, and iteratively updating the values ​​of the model parameters using the loss value until the loss value meets a preset threshold condition.

[0009] In an exemplary embodiment, the process of inputting the values ​​of model parameters into the business feature extraction model to be trained, and iteratively updating the values ​​of model parameters using the loss value until the loss value meets a preset threshold condition includes: inputting the values ​​of model parameters into the business feature extraction model to be trained; inputting second training business data into the business feature extraction model after inputting the values ​​of model parameters, and iteratively updating the values ​​of model parameters using the loss value until the loss value meets a preset threshold condition, wherein the second training business data is obtained from the application layer network element corresponding to the shared parameters that represent the support for business data sharing.

[0010] In an exemplary embodiment, before inputting the service type into a preset service feature extraction model in response to an artificial intelligence analysis request, the method further includes: if it is detected that the shared parameter representation of the user equipment network element connected to the target application layer network element does not support the sharing of training service data, then the training service data included in the user equipment network element is used as training data, and an initial service feature extraction model is trained using a multi-layer federated learning approach to obtain the service feature extraction model.

[0011] In an exemplary embodiment, matching service features with user plane data information obtained from user plane network elements, obtaining network data analysis results based on the matching results, and sending them to consumer network elements includes: matching service features with user plane data information obtained from user plane network elements to obtain target service features that match the service features; obtaining service information corresponding to the target service features; and sending the service information and target service features as network data analysis results to consumer network elements.

[0012] According to one aspect of the embodiments of this application, a network data analysis apparatus is provided, comprising: a receiving module for receiving an artificial intelligence analysis request, the artificial intelligence analysis request including a corresponding service type; a response module for responding to the artificial intelligence analysis request by inputting the service type into a preset service feature extraction model to obtain service features corresponding to the service type, the service feature extraction model being trained using first training service data contained in a target application layer network element as training data and trained using a federated learning method, wherein the service type corresponding to the first training service data is the same as the service type included in the artificial intelligence analysis request; and a matching module for matching the service features with user plane data information obtained from a user plane network element, obtaining network data analysis results based on the obtained matching results, and sending them to a consuming network element.

[0013] According to one aspect of the embodiments of this application, an electronic device is provided, including a processor and a memory, wherein computer-readable instructions are stored in the memory, and when the computer-readable instructions are executed by the processor, the above-described method for analyzing network data is implemented.

[0014] According to one aspect of the embodiments of this application, a computer-readable storage medium is provided, on which computer-readable instructions are stored, which, when executed by a computer's processor, cause the computer to perform the network data analysis method provided above.

[0015] According to one aspect of the embodiments of this application, a computer program product or computer program is provided, 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 method provided in the various alternative embodiments described above.

[0016] In the technical solution provided in the embodiments of this application, since the service feature extraction model is trained by using the training service data contained in the target application layer network element as training data and using federated learning, the process of obtaining the service features corresponding to the service type in this application can both utilize the service data of application layer network elements that do not intend to share service data and protect the privacy service data of application layer network elements. Through the cross-domain federated learning method of network artificial intelligence network elements and application layer network elements, more accurate perception and analysis of the same type of service can be provided.

[0017] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0018] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application. It is obvious that the drawings described below are merely some embodiments of this application, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort. In the drawings:

[0019] Figure 1 This is a structural diagram of the implementation environment of the network data analysis method illustrated in an exemplary embodiment of this application;

[0020] Figure 2 This is a flowchart illustrating a method for analyzing network data according to an exemplary embodiment of this application;

[0021] Figure 3 exist Figure 2 A flowchart of an exemplary embodiment of step S103 in the embodiments;

[0022] Figure 4 exist Figure 2 A flowchart illustrating a method for analyzing network data, based on the embodiments proposed in this example;

[0023] Figure 5 exist Figure 4 The flowchart illustrating step S201 in the illustrated embodiment is shown.

[0024] Figure 6 exist Figure 4 The flowchart illustrating step S202 in the illustrated embodiment is shown.

[0025] Figure 7 exist Figure 6 The flowchart illustrating step S402 in the illustrated embodiment is shown.

[0026] Figure 8 This is a schematic diagram illustrating a method for analyzing network data according to an exemplary embodiment of this application;

[0027] Figure 9 This is a block diagram of a network data analysis apparatus illustrated in an exemplary embodiment of this application;

[0028] Figure 10 A schematic diagram of the structure of a computer system suitable for implementing the electronic device of the present application is shown. Detailed Implementation

[0029] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0030] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0031] The flowcharts shown in the accompanying drawings are merely illustrative and do not necessarily include all content and operations / steps, nor do they necessarily have to be performed in the described order. For example, some operations / steps can be broken down, while others can be combined or partially combined; therefore, the actual execution order may change depending on the specific circumstances.

[0032] It should also be noted that "multiple" as mentioned in this application refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. The character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0033] It should be noted that the network elements involved in the embodiments of this application may also be referred to as functions or functional entities, and this application does not impose any limitations. For example, the access and mobility management function network element may also be referred to as an access and mobility management function or an access and mobility management function entity, and the session management function network element may be referred to as a session management function or a session management function entity, etc. The names of each network element are not limited in this application, and those skilled in the art can replace the names of the above network elements with other names to perform the same function, all of which fall within the scope of protection of this application.

[0034] It is understood that the network elements or functions or functional entities involved in this embodiment can be network components in hardware devices, software functions running on dedicated hardware, or virtualization functions instantiated on a platform (e.g., cloud platform), and this application does not impose any restrictions.

[0035] To better understand and explain the solutions of the embodiments of this application, the technical terms involved in the embodiments of this application will be briefly explained below.

[0036] NWDAF (Network Data Analytics Function) network element is an intelligent network element defined by 3GPP. It can utilize big data and artificial intelligence technologies to automatically sense and analyze the network based on network data, and participate in the entire lifecycle of network planning, construction, operation and maintenance, network optimization, and operation. This makes the network easier to maintain and control, improves the efficiency of network resource utilization, and enhances the user service experience.

[0037] PFDF (Packet Flow Description Function) network element.

[0038] AF (Application Function) network element: a function used to process service-related signaling and media data, and to interact with the network side for QoS authorization.

[0039] UE: User equipment, user terminal.

[0040] UPF (User Plane Function) network element, mainly supports the routing and forwarding of service data of UE network element, data and service identification, action and policy execution, etc.

[0041] PCF (Policy Control Function) network element: Used for user subscription data management, policy control, billing policy control, and service quality control.

[0042] In existing technologies, when NWDAF network elements perform service experience-related analyses or application detection, the current 3GPP specification defines a method that obtains relevant data from the application layer for model training. This method has two problems: one is that the application layer (e.g., third-party applications) is unwilling to provide its own business data or user data due to privacy or business operation reasons; the other is that each application layer is a specific business, such as targeting a specific region or certain user groups. If it is necessary to perform perception analysis of one type of business or multiple business or user combinations, data bias is inevitable, which cannot meet the generalization requirements of model training, thereby reducing the accuracy of model training.

[0043] The network data analysis method provided in this embodiment finds a group of application layer network elements of the same type and trains the model together through federated learning based on the requirements of perception analysis. On the one hand, it can protect the privacy of applications and their users. On the other hand, through the cross-domain federated learning method of core network, application layer network elements and users, it can provide more accurate perception analysis for a type of service or a combination of services or users.

[0044] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0045] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0046] Machine learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and learn-by-doing.

[0047] To address at least the aforementioned problems in the prior art, the network data analysis method, network data analysis device, electronic device, and computer-readable storage medium proposed in the embodiments of this application involve artificial intelligence technology and machine learning technology. These embodiments will be described in detail below.

[0048] See Figure 1 , Figure 1 This is a structural diagram of the implementation environment of a network data analysis method illustrated in an exemplary embodiment of this application, such as... Figure 1As shown, the implementation environment of the network data analysis method provided in this application includes several UE network elements, application layer network elements, network artificial intelligence network elements, consumer network elements, and user plane network elements.

[0049] Consumer network elements, such as PFDF network elements and core network elements, are used to send AI analysis requests, which include the corresponding service types. Target application layer network elements, such as AF network elements, are used to obtain first training service data from UE network elements, train an initial service feature extraction model based on the first training service data using federated learning, and send the model parameter values ​​of the initial service feature extraction model to the network AI network element.

[0050] User plane network elements can be UPF network elements, used to send user plane data information to network artificial intelligence network elements.

[0051] The network AI element can be an NWDAF element, used to receive AI analysis requests, which include the corresponding service type; in response to the AI ​​analysis request, it inputs the service type into a pre-set service feature extraction model to obtain service features corresponding to the service type. The service feature extraction model is trained using the first training service data contained in the target application layer network element as training data and employing federated learning. The service type corresponding to the first training service data is the same as the service type included in the AI ​​analysis request; the service features are matched with user plane data information obtained from the user plane network element, and the analysis results of the network data are obtained based on the matching results and sent to the consumer network element.

[0052] See Figure 2 , Figure 2 This is a flowchart illustrating a network data analysis method according to an exemplary embodiment of this application. The network data analysis method provided in this embodiment is applied to... Figure 1 The network artificial intelligence element shown is an example. Figure 2 As shown, the network data analysis method provided in this embodiment includes steps S101-S103, which are described in detail below:

[0053] Step S101: Receive an artificial intelligence analysis request.

[0054] In this embodiment, the AI ​​analysis request includes corresponding business types. Business types include, but are not limited to, video services, payment services, autonomous driving services, vertical services, etc.

[0055] For example, the AI ​​analytics request may also include an application ID, which identifies the application sending the AI ​​analytics request.

[0056] Step S102: In response to the artificial intelligence analysis request, input the business type into the preset business feature extraction model to obtain the business features corresponding to the business type.

[0057] In this embodiment, the service feature extraction model is trained using the first training service data contained in the target application layer network element as training data and trained using a federated learning approach.

[0058] Federated Learning (FL), also known as federated machine learning, consortium learning, etc., is a machine learning framework that effectively helps multiple organizations use data and perform machine learning modeling while meeting user privacy, data security, and government regulations. As a distributed machine learning paradigm, federated learning effectively solves the data silo problem, allowing participants to jointly model data without sharing it, technically breaking down data silos and enabling AI collaboration. Federated learning defines a machine learning framework within which virtual models are designed to address the problem of collaboration among different data owners without exchanging data. The virtual model is the optimal model that aggregates data from all parties, and each region uses the model to serve its local goals.

[0059] In this embodiment, network AI network elements serve as the main service nodes, and target application layer network elements serve as edge nodes to construct a federated machine learning framework for federated learning of business types, namely a business feature extraction model. It can be understood that the target application layer network elements included in the federated machine learning framework constructed in this embodiment can be one or more depending on the actual application situation, and no specific limitation is made here.

[0060] In this embodiment, the first training service data is the service data corresponding to the historical services executed by the target application layer network element, including the service type and the corresponding service characteristics. The service type corresponding to the first training service data is the same as the service type included in the artificial intelligence analysis request, so as to ensure that the service feature extraction model trained by the first training service data can accurately analyze the service type included in the artificial intelligence analysis request.

[0061] Business characteristics include, but are not limited to, uplink and downlink rates (maximum, minimum, average), typical usage periods, latency, data burst characteristics (e.g., Maximum Data Burst Volume, rate, duration window, etc.), bit error rate, and other auxiliary information such as application ID, slice ID, DNN, stream description, URL, domain name information, etc.

[0062] This embodiment does not limit the type of business feature extraction model, such as CNN (Convolutional Neural Network), DNN (Deep Neural Network), LSTM (Long Short-Term Memory), ResNet (Residual Network), etc.

[0063] For example, the first training service data included in the target application layer network element is obtained from the user equipment network element. For example, the target application layer network element stores registration information received during the registration process of the user equipment network element, wherein the registration information includes shared parameters for the user equipment network element, wherein the shared parameters are used to characterize whether the user equipment network element supports training service data sharing.

[0064] For example, if it is detected that the shared parameter representation of the user equipment network element connected to the target application layer network element does not support the sharing of training service data, then the training service data included in the user equipment network element is used as training data, and the initial service feature extraction model is trained using a multi-layer federated learning approach to obtain the service feature extraction model.

[0065] In other words, the network AI element and the target application layer element form the first-layer federated learning framework, namely the service feature extraction model. The target application layer element and the connected user equipment elements form the second-layer federated learning framework. The second-layer federated learning framework is actually a sub-service feature extraction model of the service feature extraction model, except that the sub-service feature extraction model is divided into multiple modules. Each module is distributed in the various user equipment elements connected to the target application layer element. During the training process of the multi-layer federated learning framework composed of the first-layer and second-layer federated learning frameworks, the user equipment element uses its training service data to train one module of the second-layer federated learning framework. After obtaining the model parameters, it transmits them to the target application layer element it is connected to. After the target application layer element obtains the values ​​of multiple parameters sent by each user equipment element, it adjusts the values ​​of the parameters in the second-layer federated learning framework again to obtain the final parameter values. The target application layer element sends the final parameter values ​​to the network AI element so that the network AI element can adjust the parameters of the service feature extraction model.

[0066] This embodiment can flexibly train the service feature extraction model based on whether the application layer network element and the user equipment network element support the sharing of training service data. While ensuring the privacy of the service data of the application layer network element or the user equipment network element, it can make full use of the training service data to train the service feature extraction model, improve the generalization ability of the model, and provide more accurate perception analysis based on the trained service feature extraction model.

[0067] It is understood that if this embodiment includes more than one target application layer network element that meets the preset conditions, and some of the target application layer network elements can obtain the first training service data from the user equipment network element connected to them, while other network elements cannot obtain the first training service data from the user equipment network element connected to them, then the target application layer network element that can obtain the first training service data and the user equipment network element that is communicatively connected to the target application layer network element that cannot obtain the first training service data serve as the edge node of the federated learning framework, and the network artificial intelligence network element serves as the main service node.

[0068] Step S103: Match the service characteristics with the user plane data information obtained from the user plane network element, obtain the analysis results of the network data based on the matching results, and send them to the consumer network element.

[0069] In this embodiment, user plane data information is first obtained from the user plane network element. This user plane data information includes, but is not limited to, uplink and downlink rates, latency, transmission time period, duration, and corresponding server IP address, port number, protocol, etc. For example, the user plane network element can be a UPF network element. If it is detected that the UPF network element and the network AI network element have not established a communication connection, or that the communication connection established between the UPF network element and the network AI network element is disconnected, then the user plane data information is obtained from the UPF network element through the SMF network element.

[0070] In this embodiment, since the user plane network element or the network artificial intelligence network element cannot distinguish the service type corresponding to the user plane data information, this embodiment uses the service feature extraction model to output the service feature corresponding to the service type to distinguish the target service feature in the user plane data information that corresponds to the service type.

[0071] Specifically, the service characteristics are matched with user plane data information obtained from user plane network elements, and the analysis results of the network data are obtained based on the matching results. For example, service characteristics similar to the service characteristics are determined from the user plane data information, and service information is obtained from the user plane data information based on these similar service characteristics and sent as the analysis results of the network data to the consumer network element. It is understood that service characteristics matching the service characteristics are periodically determined from the user plane data information within a preset time period. If a matching service characteristic cannot be obtained within the preset time period, a notification indicating network data analysis failure is returned to the consumer network element.

[0072] In this embodiment, the analysis results of network data include, but are not limited to, service type, application ID, service MOS (Mean Opinion Score) value, protocol server IP address, port number, URL, domain name, and data flow characteristics.

[0073] The network data analysis method provided in this embodiment receives an artificial intelligence (AI) analysis request, which includes a corresponding service type. In response to the AI ​​analysis request, the service type is input into a preset service feature extraction model to obtain service features corresponding to the service type. The service feature extraction model is trained using first training service data contained in the target application layer network element as training data, employing a federated learning approach. The service type corresponding to the first training service data is the same as the service type included in the AI ​​analysis request. The service features are matched with user plane data information obtained from the user plane network element. Based on the matching result, the analysis result of the network data is obtained and sent to the consumer network element. Since the service feature extraction model is trained using training service data contained in the target application layer network element as training data, the process of obtaining service features corresponding to a service type in this application can utilize service data from application layer network elements that do not intend to share service data, while also protecting the privacy of application layer network element service data. Through cross-domain federated learning between network AI network elements and application layer network elements, more accurate perception and analysis of the same type of service is provided.

[0074] See Figure 3 , Figure 3 exist Figure 2 A flowchart of an exemplary embodiment of step S103 in the embodiments is shown below. Figure 3 As shown, step S103 includes steps S701-S703, which are described in detail below:

[0075] Step S701: Match the service features with the user plane data information obtained from the user plane network elements to obtain the target service features that match the service features.

[0076] In this embodiment, the user plane data information includes multiple services being processed by the user plane network element, the channel characteristics corresponding to each service, and the server IP address, port number, protocol, and other information corresponding to each service. The channel characteristics include, but are not limited to, uplink / downlink rates, latency, transmission time period, and duration. In fact, what the consumer network element most wants to know during network data analysis is the server IP address, port number, protocol, and other information corresponding to a particular service type.

[0077] For example, the similarity between service features and user plane data information is calculated to obtain target service features that match the service features. For instance, the similarity between multiple channel features corresponding to multiple services included in the user plane network element and the service features is calculated respectively, and the channel feature with the highest similarity to the service features is obtained as the target service feature.

[0078] Step S702: Obtain the business information corresponding to the target business characteristics.

[0079] For example, the target service corresponding to the channel feature with the greatest similarity to the service feature is determined, and then the server IP address, port number, protocol and other information corresponding to the target service are obtained and used as service information.

[0080] Step S703: Send the business information and target business characteristics as network data analysis results to the consumer network element.

[0081] In this embodiment, business information and target business characteristics are sent to the consumer network element as network data analysis results, so that the consumer at the consumer network element can understand the business information corresponding to the business type included in the artificial intelligence analysis request based on the analysis results.

[0082] See Figure 4 , Figure 4 exist Figure 2 A flowchart illustrating a method for analyzing network data, based on the embodiments proposed, is shown below. Figure 4 As shown, before step S102, the network data analysis method provided in this embodiment further includes steps S201-S202, which are described in detail below:

[0083] Step S201: Determine the target application layer network element that meets the preset conditions.

[0084] To reduce computational complexity and improve the efficiency of network data analysis, this embodiment selects application layer network elements that meet preset conditions from all application layer network elements that communicate with network artificial intelligence network elements as target application layer network elements.

[0085] For example, application layer network elements that have processed services with the same service type as the AI ​​analysis request within a preset time period are designated as target application layer network elements.

[0086] See Figure 5 , Figure 5 exist Figure 4 The flowchart illustrating step S201 in the illustrated embodiment is as follows: Figure 5 As shown, step S201 includes steps S301-S302, which are described in detail below:

[0087] Step S301: Obtain the registration information corresponding to the application layer network element.

[0088] In this embodiment, the registration information includes sharing parameters used to characterize whether the application layer network element supports service data sharing. The registration information corresponding to the application layer network element is sent to the network AI network element and stored by the network AI network element when the corresponding application layer network element applies for registration.

[0089] For example, the registration information includes multiple service types and shared parameters corresponding to each service type, wherein the shared parameters corresponding to each service type are used to characterize whether the application layer network element supports service data sharing for the corresponding service type.

[0090] Step S302: Select the application layer network element corresponding to the shared parameter that does not support service data sharing as the target application layer network element that meets the preset conditions.

[0091] When it is detected that the shared parameter representation of the application layer network element does not support business data sharing, it means that the corresponding application layer network element has no intention of sending business data to the network artificial intelligence network element. Therefore, it is impossible to train the business feature extraction model in the network artificial intelligence network element based on the business data. Therefore, in order to avoid the inability to use the business data of the corresponding application layer network element to train the business feature extraction model, which would result in a decrease in the generalization of the business feature extraction model, the corresponding application layer network element is used as the target application layer network element.

[0092] For example, if the shared parameter representation of the application layer network element is detected to support business data sharing, it means that the corresponding application layer network element is willing to send business data to the network artificial intelligence network element. In this case, the business data of the corresponding application layer network element can be directly obtained for training the business feature extraction model.

[0093] Step S202: Use the first training service data of the target application layer network element that meets the preset conditions as training data, and use federated learning to train the service feature extraction model to be trained, so as to obtain the trained service feature extraction model.

[0094] In this embodiment, the first training service data is obtained from the user equipment network element connected to the target application layer network element, and the first training service data includes service features corresponding to the service type.

[0095] In this embodiment, an untrained business feature extraction model is first constructed. Then, the business feature extraction model to be trained is trained using the first training business data as training data.

[0096] See Figure 6 , Figure 6 exist Figure 4 The flowchart illustrating step S202 in the illustrated embodiment is as follows: Figure 5 As shown, step S202 includes steps S401-S402, and a detailed description is provided below:

[0097] Step S401: Obtain the values ​​of the model parameters sent by the target application layer network element.

[0098] In this embodiment, the values ​​of the model parameters are obtained by the target application layer network element iteratively training the sub-service feature extraction model using the first training service data as training data.

[0099] In this embodiment, the service feature extraction model includes a sub-service feature extraction model constructed by the target application layer network element, and the service feature extraction model includes model parameters of the sub-service feature extraction model. For example, if there are two target application layer network elements, which can be designated as a first target application layer network element and a second application layer network element, their respective sub-service feature extraction models are designated as a first sub-service feature extraction model and a second sub-service feature extraction model. The first sub-service feature extraction model includes first model parameters, and the second sub-service feature extraction model includes second model parameters. Therefore, the service feature extraction model includes both first and second model parameters.

[0100] In this embodiment, the sub-business feature extraction model is iteratively trained using the first training business data as training data. The model parameters are adjusted by optimizing the loss function to finally obtain model parameter values ​​that make the loss value meet the preset threshold conditions. This embodiment does not limit the type of loss function, such as Mean Squared Loss (MSL), Mean Absolute Error Loss (MAEL), Quantile Loss (QL), Cross Entropy Loss (CEL), etc.

[0101] Step S402: Input the values ​​of the model parameters into the business feature extraction model to be trained, and use the loss value to iteratively update the values ​​of the model parameters until the loss value meets the preset threshold condition.

[0102] In this embodiment, since the service feature extraction model includes the same model parameters as the sub-service feature model constructed by the target application layer network element, after obtaining the values ​​of the model parameters of the sub-service feature model, the values ​​of the model parameters are directly substituted into the model parameters corresponding to the service feature extraction model.

[0103] In this embodiment, if the number of target application layer network elements is one, the values ​​of the model parameters can be directly substituted into the service feature extraction model to obtain the trained service feature extraction model.

[0104] If there is more than one target application layer network element, the values ​​of multiple model parameters need to be input into the business feature extraction model, and the loss value is used to update the values ​​of the model parameters again until the loss value meets the preset threshold condition, and then the trained business feature extraction model can be obtained.

[0105] See Figure 7 , Figure 7 exist Figure 6 The flowchart illustrating step S402 in the illustrated embodiment is as follows: Figure 7 As shown, step S402 includes steps S501-S502, which are described in detail below:

[0106] Step S501: Input the values ​​of the model parameters into the business feature extraction model.

[0107] In this embodiment, if there is more than one target application layer network element, and if the model parameters in the sub-service feature extraction models constructed by the target application layer network elements are the same, and the values ​​of the model parameters obtained after iterative training on their respective training data are different, then the different values ​​of the model parameters are respectively substituted into the service feature extraction model, and the values ​​of the model parameters are iteratively updated again using the loss value until the loss value meets the preset threshold condition, and the trained service feature extraction model can be obtained.

[0108] In this embodiment, the values ​​of the model parameters are substituted into the business feature extraction model to obtain the business feature extraction model after substituted with the values ​​of the model parameters.

[0109] Step S502: Input the second training business data into the business feature extraction model after the model parameters are input, and use the loss value to iteratively update the model parameter value until the loss value meets the preset threshold condition.

[0110] In this embodiment, the second training service data corresponds to the service type included in the artificial intelligence analysis request. The second training service data is obtained from the application layer network element corresponding to the shared parameters that represent support for service data sharing. In this embodiment, if the shared parameters corresponding to the application layer network element are detected to represent support for service data sharing, it indicates that the corresponding application layer network element intends to send service data to the network artificial intelligence network element. In this case, the second training service data is obtained from the corresponding application layer network element to train the service feature extraction model and improve the generalization ability of the trained service feature extraction model.

[0111] In this embodiment, when training the business feature extraction model, it can train the model using first training business data from target application layer network elements that do not intend to send business data to the network AI network element through federated learning, and it can also train the model using second training business data from application layer network elements that intend to send business data to the network AI network element. While ensuring the privacy of business data, it can further improve the generalization ability of the trained business feature extraction model and provide more accurate perception and analysis for the same type of business.

[0112] See Figure 8 , Figure 8 This is a schematic diagram illustrating a network data analysis method according to an exemplary embodiment of this application. The network data analysis method provided in this embodiment is applied to... Figure 1 The network artificial intelligence element shown is an example. Figure 8 As shown, the network data analysis method provided in this embodiment includes steps S1-S8, which are described in detail below:

[0113] Step S1: The NWDAF network element obtains the application type of the AF network element.

[0114] In this embodiment, registration information corresponding to the AF network element is obtained. The registration information includes sharing parameters used to characterize whether the AF network element supports service data sharing. For example, the application type of the AF network element includes: an AF network element whose sharing parameters characterize that the application layer network element supports service data sharing, or an AF network element whose sharing parameters characterize that the application layer network element does not support service data sharing.

[0115] This embodiment obtains the application types of AF1 network element, AF2 network element and NEF network element.

[0116] Step S2: Consumers subscribe to or request application-aware analytics.

[0117] In this embodiment, the consumer can be a consumer network element, such as a PFDF network element or an NF network element, without specific limitations. For example, the consumer sends an artificial intelligence analysis request to subscribe to or request application perception analysis. For example, the artificial intelligence analysis request includes the corresponding service type. Service types include, but are not limited to, video services, payment services, autonomous driving services, vertical services, etc.

[0118] Step S3: The NWDAF network element selects the analysis method according to its requirements.

[0119] In this embodiment, the selected analysis method is to perform federated learning in conjunction with AF network elements, and extract the service features corresponding to the service type based on the obtained federated learning framework.

[0120] Step S4: The NWDAF network element starts federated learning.

[0121] Step S5: The AF network element interacts with the UE network element to initiate the training of the service feature extraction model or the collection of training data.

[0122] The training data consists of business features corresponding to the business type.

[0123] Step S6: NWDAF network elements extract and analyze application layer features.

[0124] Application layer features include business features corresponding to business types. In this embodiment, the business type is input into a trained business feature extraction model, which then outputs the business features corresponding to the business type.

[0125] Step S7: The NWDAF network element obtains user plane data and related feature information from the SMF network element or the UPF network element.

[0126] In this embodiment, user plane data and related feature information include, but are not limited to, uplink and downlink rates, latency, transmission period, duration, and corresponding server IP address / port number, protocol, and other information.

[0127] Step S8: The NWDAF network element returns the result to the consumer.

[0128] See Figure 9 , Figure 9 This is a block diagram of a network data analysis apparatus illustrated in an exemplary embodiment of this application, such as... Figure 9 As shown, the network data analysis device 600 includes a receiving module 601, a response module 602, and a matching module 603.

[0129] The receiving module 601 receives an artificial intelligence analysis request, which includes a corresponding service type. The response module 602 responds to the artificial intelligence analysis request by inputting the service type into a preset service feature extraction model to obtain service features corresponding to the service type. The service feature extraction model is trained using federated learning with the first training service data contained in the target application layer network element as training data. The service type corresponding to the first training service data is the same as the service type included in the artificial intelligence analysis request. The matching module 603 matches the service features with user plane data information obtained from the user plane network element, obtains the analysis results of the network data based on the matching results, and sends them to the consumer network element.

[0130] In another exemplary embodiment, the network data analysis apparatus 600 further includes a determination module and a training module, wherein the determination module is used to determine target application layer network elements that meet preset conditions; the training module is used to use the first training service data included in the target application layer network elements that meet the preset conditions as training data, and to train the service feature extraction model to be trained using a federated learning approach to obtain the trained service feature extraction model.

[0131] In another exemplary embodiment, the determining module includes a first obtaining unit and a second obtaining unit, wherein the first obtaining unit is used to obtain registration information corresponding to the application layer network element, the registration information including sharing parameters used to characterize whether the application layer network element supports service data sharing; the second obtaining unit is used to select the application layer network element corresponding to the sharing parameters that characterize whether the application layer network element supports service data sharing as the target application layer network element that meets the preset conditions.

[0132] In another exemplary embodiment, the training module includes a third acquisition unit and an iterative update unit. The third acquisition unit is used to acquire the values ​​of model parameters sent by the target application layer network element. The values ​​of the model parameters are obtained by the target application layer network element iteratively training the sub-service feature extraction model using the first training service data as training data. The iterative update unit is used to input the values ​​of the model parameters into the service feature extraction model and iteratively update the values ​​of the model parameters using the loss value until the loss value meets a preset threshold condition. The service feature extraction model includes the sub-service feature extraction model constructed by the target application layer network element, and the service feature extraction model includes the model parameters of the sub-service feature extraction model.

[0133] In another exemplary embodiment, the iterative update unit includes an input subunit and an iterative update subunit, wherein the input subunit is used to input the values ​​of the model parameters into the business feature extraction model to be trained; the iterative update subunit is used to input the second training business data into the business feature extraction model after inputting the values ​​of the model parameters, and iteratively update the values ​​of the model parameters using the loss value until the loss value meets the preset threshold condition, wherein the second training business data is obtained from the application layer network element corresponding to the shared parameters that represent the support for business data sharing.

[0134] In another exemplary embodiment, the network data analysis apparatus 600 further includes a training module, which is used for...

[0135] If the shared parameter representation of the user equipment network element connected to the target application layer network element does not support the sharing of training service data, then the training service data included in the user equipment network element is used as training data, and the initial service feature extraction model is trained using a multi-layer federated learning approach to obtain the service feature extraction model.

[0136] In another exemplary embodiment, the matching module 603 includes a matching unit, a fourth acquisition unit, and a sending unit. The matching unit is used to match service features with user plane data information obtained from user plane network elements to obtain target service features that match the service features. The fourth acquisition unit is used to acquire service information corresponding to the target service features. The sending unit is used to send the service information and the target service features as network data analysis results to the consumer network element.

[0137] It should be noted that the apparatus provided in the above embodiments and the method provided in the above embodiments belong to the same concept, and the specific way in which each module and unit performs operations has been described in detail in the method embodiments, and will not be repeated here.

[0138] In another exemplary embodiment, this application provides an electronic device including a processor and a memory, wherein the memory stores computer-readable instructions that, when executed by the processor, implement the aforementioned policy and billing control method.

[0139] Figure 10 A schematic diagram of the structure of a computer system suitable for implementing the electronic device of the present application is shown.

[0140] It should be noted that, Figure 10 The computer system 1000 of the electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0141] like Figure 10As shown, the computer system 1000 includes a Central Processing Unit (CPU) 1001, which can perform various appropriate actions and processes based on programs stored in Read-Only Memory (ROM) 1002 or programs loaded from storage portion 1008 into Random Access Memory (RAM) 1003, such as executing the information recommendation method described in the above embodiments. Various programs and data required for system operation are also stored in RAM 1003. The CPU 1001, ROM 1002, and RAM 1003 are interconnected via bus 1004. An Input / Output (I / O) interface 1005 is also connected to bus 1004.

[0142] The following components are connected to I / O interface 1005: an input section 1006 including a keyboard, mouse, etc.; an output section 1007 including a cathode ray tube (CRT), liquid crystal display (LCD), etc., and speakers, etc.; a storage section 1008 including a hard disk, etc.; and a communication section 1009 including a network interface card such as a LAN (Local Area Network) card, modem, etc. The communication section 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to I / O interface 1005 as needed. Removable media 1011, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 1010 as needed so that computer programs read from them can be installed into storage section 1008 as needed.

[0143] Specifically, according to embodiments of this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program including a computer program for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 1009, and / or installed from removable medium 1011. When the computer program is executed by central processing unit (CPU) 1001, it performs various functions defined in the system of this application.

[0144] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. The transmitted data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0145] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0146] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0147] Another aspect of this application provides a computer-readable storage medium having computer-readable instructions stored thereon, which, when executed by a processor, implement a network data analysis method as described in any of the preceding embodiments.

[0148] Another aspect of this application provides a computer program product or computer program including 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 method provided in the various embodiments described above.

[0149] It should be noted that the computer-readable medium shown in the embodiments of this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium can be, for example, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this application, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying a computer-readable computer program. The transmitted data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The computer program contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wired, etc., or any suitable combination thereof.

[0150] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. Each block in a flowchart or block diagram may represent a module, segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram or flowchart, and combinations of blocks in a block diagram or flowchart, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0151] The units described in the embodiments of this application can be implemented in software or hardware, and the described units can also be located in a processor. The names of these units do not necessarily limit the specific unit itself.

[0152] The above description is merely a preferred exemplary embodiment of this application and is not intended to limit the implementation of this application. Those skilled in the art can easily make corresponding modifications or alterations based on the main concept and spirit of this application. Therefore, the scope of protection of this application should be determined by the scope of protection claimed in the claims.

Claims

1. A method for analyzing network data, applied to network artificial intelligence elements, characterized in that, include: Receive an artificial intelligence analysis request, wherein the artificial intelligence analysis request includes the corresponding business type; In response to the AI ​​analysis request, the service type is input into a preset service feature extraction model to obtain service features corresponding to the service type. The service feature extraction model includes a sub-service feature extraction model constructed by the target application layer network element, and the service feature extraction model includes model parameters of the sub-service feature extraction model. The service feature extraction model is trained in the following way: the values ​​of the model parameters sent by the target application layer network element are obtained, and the values ​​of the model parameters are obtained by iteratively training the sub-service feature extraction model using the first training service data contained in the target application layer network element as training data. The values ​​of the model parameters are input into the business feature extraction model to be trained, and the values ​​of the model parameters are iteratively updated using the loss value until the loss value meets the preset threshold condition. The business type corresponding to the first training business data is the same as the business type included in the artificial intelligence analysis request. The target application layer network element is the application layer network element corresponding to the shared parameter that does not support business data sharing. The service characteristics are matched with user plane data information obtained from user plane network elements. Based on the matching results, the analysis results of the network data are obtained and sent to the consumer network elements.

2. The method according to claim 1, characterized in that, Before inputting the business type into a preset business feature extraction model in response to the artificial intelligence analysis request, the method further includes: The application layer network element that meets the preset conditions is taken as the target application layer network element; The first training service data included in the target application layer network element is used as training data, and the service feature extraction model to be trained is trained using federated learning to obtain the trained service feature extraction model. The first training service data is obtained from the user equipment network element connected to the target application layer network element.

3. The method according to claim 2, characterized in that, The step of selecting application layer network elements that meet preset conditions as the target application layer network elements includes: Obtain the registration information corresponding to the application layer network element, wherein the registration information includes sharing parameters used to characterize whether the application layer network element supports service data sharing; The application layer network element corresponding to the shared parameters that do not support business data sharing is taken as the target application layer network element.

4. The method according to claim 1, characterized in that, The step of feeding the values ​​of the model parameters into the business feature extraction model to be trained, and iteratively updating the values ​​of the model parameters using the loss value until the loss value meets a preset threshold condition includes: The values ​​of the model parameters are then fed into the business feature extraction model to be trained. The second training service data is input into the service feature extraction model after the model parameters are input, and the model parameters are iteratively updated using the loss value until the loss value meets the preset threshold condition. The second training service data is obtained from the application layer network element corresponding to the shared parameters that represent the support for service data sharing.

5. The method according to claim 1, characterized in that, Before inputting the business type into a preset business feature extraction model in response to the artificial intelligence analysis request, the method further includes: If it is detected that the shared parameter representation of the user equipment network element connected to the target application layer network element does not support the sharing of training service data, then the training service data included in the user equipment network element is used as training data, and an initial service feature extraction model is trained using a multi-layer federated learning approach to obtain the service feature extraction model.

6. The method according to claim 1, characterized in that, The step of matching the service characteristics with user plane data information obtained from the user plane network element, obtaining network data analysis results based on the matching results, and sending them to the consumer network element includes: The service features are matched with user plane data information obtained from user plane network elements to obtain target service features that match the service features. Obtain the business information corresponding to the target business feature; The business information and target business characteristics are sent to the consumer network element as the network data analysis results.

7. A network data analysis device, characterized in that, include: A receiving module is used to receive artificial intelligence analysis requests, wherein the artificial intelligence analysis requests include corresponding business types; A response module is configured to respond to the artificial intelligence analysis request by inputting the service type into a preset service feature extraction model to obtain service features corresponding to the service type. The service feature extraction model includes a sub-service feature extraction model constructed by the target application layer network element, and the service feature extraction model includes model parameters of the sub-service feature extraction model. The service feature extraction model is trained in the following manner: obtaining the values ​​of the model parameters sent by the target application layer network element, wherein the values ​​of the model parameters are obtained by iteratively training the sub-service feature extraction model using the first training service data contained in the target application layer network element as training data. The values ​​of the model parameters are input into the business feature extraction model to be trained, and the values ​​of the model parameters are iteratively updated using the loss value until the loss value meets the preset threshold condition. The business type corresponding to the first training business data is the same as the business type included in the artificial intelligence analysis request. The target application layer network element is the application layer network element corresponding to the shared parameter that does not support business data sharing. The matching module is used to match the service features with user plane data information obtained from user plane network elements, obtain the analysis results of network data based on the matching results, and send them to the consumer network elements.

8. An electronic device, characterized in that, include: Memory, which stores computer-readable instructions; A processor reads computer-readable instructions stored in memory to perform the method described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, It stores computer-readable instructions that, when executed by the processor of a computer, cause the computer to perform the method described in any one of claims 1-6.

Images