Single sign-on network optimization method and device, electronic equipment and storage medium

Abstract

The application discloses a one-key login network state optimization method and device, electronic equipment and storage medium, and belongs to the technical field of communication. The method comprises the following steps: acquiring one-key login multi-modal data containing base station information, terminal device network state data, operator core network node information and one-key login operation logs; constructing a multi-element heterogeneous graph according to the base station, the terminal device and the operator core network node; inputting the multi-element heterogeneous graph into a graph neural network for data fusion to obtain network state fusion data; and generating an optimal network switching strategy based on the network state fusion data. The application aims to solve the problem that the one-key login has a low success rate and a poor experience in a weak network environment due to the inability to comprehensively perceive the network state in the prior art, and achieves the technical effect of significantly improving the service stability and success rate of the one-key login.

Description

Technical Field

[0001] This invention relates to the field of network communication technology, and in particular to a method, apparatus, electronic device, and storage medium for optimizing network status through one-click login. Background Technology

[0002] In mobile internet applications, the "one-click login" function has become a mainstream solution for improving user registration and login experience due to its convenience. This function heavily relies on mobile cellular networks to obtain users' mobile phone numbers; therefore, the quality of the network environment directly determines the stability and success rate of its service.

[0003] Currently, the mobile network environment is extremely complex. Base station deployment density, signal coverage strength, and network load vary significantly across different regions. For example, in remote areas, base stations are sparse, resulting in severe signal attenuation; in urban core areas, user density is high, leading to network congestion; and in special scenarios such as subways, signal shielding effects are strong. These factors can all cause issues such as response timeouts and failed number retrieval for the "one-click login" service, not only reducing user experience but also potentially leading to user churn.

[0004] To address the network connectivity optimization problem, several solutions have been proposed in the prior art. One prior art method discloses a method for determining the target base station based on a graph neural network. This method constructs a heterogeneous graph based on base station and terminal information and trains a model to determine the optimal handover target. However, the inventors discovered in their research that the aforementioned prior art solutions still have the following technical shortcomings:

[0005] 1. Limited data dimensions, failing to fully reflect network status: This solution only considers information from the wireless access network side consisting of terminals and base stations, ignoring the load, congestion, and other statuses of the operator's core network (such as the Service Gateway (SGW), Packet Data Network Gateway (PGW), etc.). The performance bottleneck of the core network is also a significant cause of service timeouts.

[0006] 2. Lack of performance feedback specific to the business: The optimization goal of this solution is the universal "handover smoothness," without optimizing for specific businesses such as "one-click login," which have extremely high requirements for instantaneous network quality within 5 seconds. Its model training does not depend on the final success or failure of the business, resulting in a disconnect between optimization and actual business needs.

[0007] 3. Insufficient model generalization ability and difficulty in adapting to dynamic network changes: The scheme does not clearly define how to deal with dynamic changes in network topology. For example, when a new base station comes online or an old base station goes offline due to a fault, its graph neural network model that relies on fixed meta-paths may experience a decline in prediction performance because it cannot handle unseen nodes.

[0008] In view of this, how to comprehensively consider information from multiple dimensions such as terminals, base stations, and core networks, and combine it with performance feedback from specific services to build an intelligent optimization system that can dynamically adapt to changes in network topology, so as to improve the success rate of one-click login in weak network environments, is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0009] The main objective of this invention is to provide a one-click login network status optimization method, device, electronic device, and storage medium, aiming to overcome the technical problems of low one-click login success rate caused by single data dimension, lack of business feedback, and poor model generalization ability in the prior art.

[0010] To achieve the above objectives, the present invention provides a one-click login network status optimization method, the method comprising: Acquire one-click login multimodal data, which includes: base station information, network status data of terminal devices connected to the base station, operator core network node information, and one-click login operation logs; The one-click login multimodal data is preprocessed to obtain preprocessed multimodal data; A multi-element heterogeneous graph is constructed based on the base station, the terminal device, and the operator core network node, wherein the base station, the terminal device, and the operator core network node are nodes, and the connection relationships between nodes are edges. The multi-modal graph is input into a preset graph neural network, and the preprocessed multimodal data is fused through the graph neural network to obtain network state fusion data. Based on the network status fusion data, an optimal network switching strategy is generated, which is used to instruct the terminal device to switch networks when performing the one-click login service.

[0011] Furthermore, the preprocessing includes: data cleaning, data integration, feature extraction, and normalization.

[0012] Furthermore, the characteristics of the edges and / or nodes of the multi-dimensional heterogeneous graph include at least one of the following: signal strength, network latency, packet loss rate, transmission rate, one-click login success rate, one-click login response time, transmission bandwidth, signaling traffic, congestion index, handover success rate, and link reliability.

[0013] Further, the step of inputting the multi-modal heterogeneous graph into the preset graph neural network, and fusing the preprocessed multimodal data through the graph neural network, includes: The multivariate heterogeneous graph is processed by a heterogeneous graph pre-trained model to obtain the first fusion feature; The heterogeneous graph is learned by a heterogeneous graph meta-learner, and meta-paths suitable for the current one-click login optimization task are dynamically generated and selected. Based on the selected meta-path, information is aggregated from the multi-dimensional heterogeneous graph to obtain the second fusion feature; An attention mechanism is used to fuse the first fusion feature and the second fusion feature in parallel.

[0014] Furthermore, the method further includes: training the graph neural network using a hybrid training mode, wherein the hybrid training mode includes at least one of supervised learning, incremental learning, transfer learning, and online learning.

[0015] Furthermore, the step of training the graph neural network using online learning includes: pushing the dynamic network handover scheme generated by the optimal network handover strategy to the operator's core network node for execution by the terminal device; collecting the login results after the terminal device performs the handover as feedback data; and optimizing the graph neural network in real time based on the feedback data.

[0016] In another aspect, the present invention also proposes a one-click login network status optimization device, the device comprising: The data acquisition and preprocessing module is used to acquire one-click login multimodal data, which includes: base station information, network status data of terminal devices connected to the base station, operator core network node information, and one-click login operation logs; and to preprocess the one-click login multimodal data to obtain preprocessed multimodal data. The heterogeneous graph construction module is used to construct a multi-dimensional heterogeneous graph based on the base station, the terminal device, and the operator core network node, wherein the base station, the terminal device, and the operator core network node are nodes, and the connection relationship between nodes is an edge. The data fusion module is used to input the multi-modal graph into a preset graph neural network, and fuse the preprocessed multimodal data through the graph neural network to obtain network state fusion data. The strategy generation module is used to generate an optimal network switching strategy based on the network status fusion data. The optimal network switching strategy is used to instruct the terminal device to switch networks when performing the one-click login service.

[0017] Furthermore, the data fusion module is specifically used for: The multivariate heterogeneous graph is processed by a heterogeneous graph pre-trained model to obtain the first fusion feature; The heterogeneous graph is learned by a heterogeneous graph meta-learner, and meta-paths suitable for the current one-click login optimization task are dynamically generated and selected. Based on the selected meta-path, information is aggregated from the multi-dimensional heterogeneous graph to obtain the second fusion feature; An attention mechanism is used to fuse the first fusion feature and the second fusion feature in parallel.

[0018] In another aspect, the present invention provides an electronic device comprising: a processor and a memory; the memory for storing computer execution instructions, the processor being connected to the memory via a bus; when the electronic device is in operation, the processor executes the computer execution instructions stored in the memory to cause the electronic device to perform the method described in any of the preceding claims.

[0019] In another aspect, the present invention provides a computer-readable storage medium having instructions stored thereon that, when run on a computer, cause the computer to perform any of the methods described above.

[0020] Compared with the prior art, the technical solution provided by the present invention has at least the following beneficial effects: 1. Improved the comprehensiveness and accuracy of network status awareness: By innovatively incorporating "operator core network nodes" into the heterogeneous graph and integrating "one-click login operation logs", this invention can build an end-to-end network status view from the access network, core network to upper-layer services, solving the problem that existing technologies cannot fully assess network quality based solely on wireless side data, thereby improving the accuracy of subsequent decisions.

[0021] 2. Enhanced adaptability and generalization ability of the model to dynamic network environments: This invention introduces a "task-oriented heterogeneous primitive learner" to achieve dynamic generation and selection of meta-paths, enabling the model to handle nodes not seen during training (such as newly added or decommissioned base stations), and possessing inductive learning capabilities. This solves the fundamental problem of poor generalization ability of existing models that rely on fixed meta-paths in real dynamic networks.

[0022] 3. Achieved closed-loop adaptive network optimization, significantly improving service success rate: This invention adopts a hybrid training mode, particularly through online learning, using real one-click login success / failure results as feedback to continuously and in real-time optimize the model. This business outcome-oriented closed-loop optimization mechanism ensures that network strategies always evolve in the direction of improving the success rate of specific services, significantly improving the stability and success rate of the "one-click login" service in complex network environments. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a flowchart of a one-click login network state optimization method based on a heterogeneous graph neural network according to an embodiment of the present invention.

[0025] Figure 2 This is a schematic diagram of a multi-element heterogeneous graph constructed according to an embodiment of the present invention.

[0026] Figure 3 This is a schematic diagram of a feature fusion architecture proposed according to an embodiment of the present invention.

[0027] Figure 4 This is a structural block diagram of an optimization device provided according to an embodiment of the present invention.

[0028] Figure 5 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0029] To make the technical problems, solutions, and advantages of this invention clearer, the invention will be described in further detail below with reference to the accompanying drawings and specific embodiments. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it.

[0030] This invention provides a one-click login network status optimization system based on heterogeneous graph neural networks. The core idea of ​​this system is to comprehensively collect multi-dimensional data from terminals, base stations, operator core networks, and the one-click login service itself. Through an advanced graph neural network model capable of adapting to dynamic changes in network topology, it performs deep data fusion and analysis to accurately predict the one-click login success rate under different network paths and generate optimal network switching strategies. This significantly improves the success rate of one-click login and user experience in complex environments such as weak networks.

[0031] The solution of this invention can be applied to 4G / 5G mobile communication networks and deployed by operators or third-party service providers. In a typical application scenario, when a user clicks the "One-Click Login" button on a mobile app, the optimization system described in this invention is triggered. This system replaces the traditional fixed network connection process, dynamically selecting an optimal network link for this login request within a very short time (e.g., within 5 seconds). This may involve switching to a neighboring base station with a better signal, and in the future, enabling operator network sharing, i.e., a China Unicom account using a China Telecom operator node, to maximize the success rate of the login.

[0032] Please see Figure 1 The diagram illustrates a flowchart of a one-click login network state optimization method based on a heterogeneous graph neural network according to an embodiment of the present invention. Figure 1 As shown, the method may specifically include the following steps:

[0033] S101: Obtain one-click login multimodal data, which includes: base station information, network status data of terminal devices connected to the base station, operator core network node information, and one-click login operation log; preprocess the one-click login multimodal data to obtain preprocessed multimodal data.

[0034] In this step, the optimization system acts as the central hub for data acquisition, retrieving all data related to the one-click login service from multiple heterogeneous data sources in real-time or near real-time. Multimodal data is necessary because a single data source cannot fully capture the complex reasons leading to login failures.

[0035] Specifically, the one-click login multimodal data includes, but is not limited to: One-click login operation log: This is the most direct data reflecting the final effect of the business. It may include: the anonymized identifier of the user's mobile phone number, public IP address, private IP address, login initiation time, total login time, login result (success / failure / timeout), etc.

[0036] Base station information reflects the physical status of the radio access network. This may include: the base station's global cell identifier (LAC+CELLID), signal strength (RSRP), signal quality (RSRQ), signal-to-noise ratio (SINR), historical handover success rate, link reliability, etc.

[0037] Carrier core network node information: This is a key data dimension that distinguishes this invention from existing technologies, reflecting the status of the data transmission backbone link. The carrier core network nodes generally refer to network elements such as the MME (Mobility Management Entity), SGW (Serving Gateway), and PGW (Packet Data Network Gateway) in the 4G (LTE) core network, and the AMF (Access and Mobility Management Function), SMF (Session Management Function), and UPF (User Plane Function) in the 5G core network. This information may include: the IP address of the connected core network node, network type (4G / 5G), transmission bandwidth, real-time signaling traffic, congestion index, etc.

[0038] Terminal device network status data: Reflects the user's network experience and may include: the terminal's geographical location information, network connection latency, packet loss rate, instantaneous transmission rate, etc.

[0039] Historical data: One-click login to the server records the response request information. The server uses the known information to query the required information from different data logs, thereby completing the pre-training, i.e. the basic model building stage.

[0040] Raw data often contains noise and has inconsistent formats, requiring preprocessing to improve the efficiency and effectiveness of model training. In one embodiment, this step may include:

[0041] Data cleaning: Processing base station information to remove duplicate, erroneous, or incomplete base station data records.

[0042] Data standardization / normalization: Normalize numerical features in terminal device data, such as signal strength and network latency (e.g., map to the [0,1] interval) to eliminate the influence of different feature units.

[0043] Feature engineering: Process login logs to extract useful derived features, such as converting login timestamps into category features of "whether it was a peak period", and performing one-hot encoding on the operator (China Mobile, China Unicom, China Telecom) and login results.

[0044] Integration and dimensionality reduction: Integrate preprocessed data from different sources, and use methods such as principal component analysis (PCA) to reduce dimensionality and remove redundant information, as appropriate.

[0045] S102: Construct a multi-element heterogeneous graph based on the base station, the terminal device, and the operator core network node, wherein the base station, the terminal device, and the operator core network node are nodes, and the connection relationships between nodes are edges.

[0046] Graph structures are fundamental to GNN models. This invention constructs a multivariate heterogeneous graph G=(V, E) to describe the various entities in the network and their complex relationships. Please refer to [link to relevant documentation]. Figure 2This is a schematic diagram of a heterogeneous graph.

[0047] Node (V): The diagram contains three core types of nodes: terminal device nodes (U), base station nodes (B), and operator core network nodes (C).

[0048] Edge (E): An edge represents a connection between nodes, and must include at least the following: Terminal-Base Station (UB) Connection: Indicates the base station where the terminal is currently camped or can access.

[0049] Base station-core network node (BC) connection: This refers to the core network element to which the base station belongs or is connected.

[0050] Base station-to-base station (BB) connection: refers to a pair of base stations that are geographically adjacent or have a handover relationship.

[0051] Node and edge features: The preprocessed multimodal data is used as attribute features for nodes and edges. For example:

[0052] The characteristics of the UB side can include: signal strength, network latency, historical one-click login success rate, etc.

[0053] The characteristics of the BC edge can be: transmission bandwidth, congestion index, etc.

[0054] The characteristics of the BB side can include: historical handover success rate, link reliability, etc.

[0055] S103: Input the multi-modal heterogeneous graph into a preset graph neural network, and fuse the preprocessed multimodal data through the graph neural network to obtain network state fusion data; This is the core processing step of the present invention. Leveraging the powerful relation learning capabilities of graph neural networks, discrete multidimensional data is fused into a vector representation capable of characterizing the overall state of the network. In a preferred embodiment, the internal implementation architecture of this step is as follows: Figure 3 As shown, the model includes two parallel paths: heterogeneous graph pre-training and a task-oriented meta-learner. The heterogeneous graph pre-training model extracts features through node content transformation, intra-meta-path aggregation, and inter-meta-path aggregation. The meta-path is defined as follows:

[0056]

[0057] Abbreviated as ( A path in the form of ) describes the node type and The composite relationship between them: , where ∘ represents a compound operator on relations.

[0058] Node content transformation: Different nodes have different feature vector dimensions. This method maps the original feature vectors to the same space, that is, the mapped feature vectors have the same dimension.

[0059] Meta-path aggregation: Extract semantic information from each meta-path and sum it using weights. The specific process is as follows: Use an encoder to compress all node features of a meta-path instance into a single vector; use a Gaussian Attention (GAT) mechanism to assign different weights to different meta-path instances, where different weights represent different contributions to the target node; finally, perform weighted summation to obtain the node representation under that meta-path.

[0060] Meta-path aggregation: Information from different meta-paths is combined and merged into a final node representation. A global summary (average + linear transformation) is performed on the output of each meta-path; an attention mechanism is used again to determine which meta-path is more important; the results of all meta-paths are weighted and merged, and finally a linear layer and activation function are added to output the final embedding.

[0061] The heterogeneous primitive learner generates a series of meta-path sequences through a meta-generation module, then selects the most suitable meta-path for feature extraction through parameter learning. Finally, the features obtained from the two paths are weighted using an attention mechanism to further optimize the features. This architecture aims to address the performance degradation problem of traditional GNN models when nodes change dynamically. Its fusion process is as follows:

[0062] 1. Heterogeneous graph pre-training (e.g.) Figure 3 (Upper half) This path is responsible for learning general, high-quality node and relation embeddings from large-scale, relatively stable historical heterogeneous graphs. It first maps the original feature vectors (which may have different dimensions) of different types of nodes (U, B, C) to the same feature space through node content transformation.

[0063] Then, based on a set of artificially defined meta-paths that can capture important semantics (e.g., UBC indicates that a user connects to a core network via a base station), intra-meta-path aggregation (e.g., using an attention mechanism to aggregate path instances) and inter-meta-path aggregation (e.g., using an attention mechanism to aggregate the outputs of different meta-paths) are performed to generate the first fused feature.

[0064] Given information: Nodes are divided into three categories: U, B, and C, which represent terminals, base stations, and operator core network nodes, respectively. V represents the union of the three categories of nodes.

[0065] Node content transformation: All original node features have been projected into a unified latent space to obtain the initial node embeddings. .

[0066] The aggregation process within a metapath is as follows: For a predefined set of generic metapaths: The model aggregates each meta-path instance through an attention mechanism to obtain the nodes under that meta-path. A specific representation of. For example, for metapaths. :

[0067]

[0068] in, It is a node Along the metapath The neighborhood group, It is a metapath instance encoder (such as a mean, linear, or relational rotary encoder). These are normalized weights calculated using an attention mechanism.

[0069] The aggregation process between metapaths is as follows: Node Representation under all predefined metapaths The general pre-trained embeddings are then aggregated again using an attention mechanism:

[0070] in, Through attention mechanisms (e.g., computation) With learnable attention vectors The weights obtained by calculating similarity and performing Softmax normalization represent the metapaths. right The importance of.

[0071] The calculation process for the β parameter is as follows: For nodes Each metapath-specific representation The model will calculate it with The "similarity" or "alignment" between them. The higher the similarity, the more important the metapath is for generating a general node representation.

[0072] The specific calculation method is as follows: First, After a linear transformation (e.g.) ,in (It is a learnable weight matrix), resulting in a... Dimension matching representation. Then, calculate the transformed representation and... The dot product, or a scalar value, is calculated via a small neural network (such as a single-layer perceptron) to represent the unnormalized attention score.

[0073] Finally, the scores of all meta-paths are normalized using the Softmax function to obtain the final attention weight for each meta-path. .

[0074] For example:

[0075]

[0076] in It is an activation function. It is a learnable weight matrix.

[0077] 2. Task-oriented heterogeneous primitive learners ( Figure 3 (Lower half) This path is the key innovation of this invention, responsible for dynamically adapting to network changes. It is an "intelligent meta-path discovery algorithm." When the network changes (such as when a new base station comes online) or when optimization is needed for a specific task, the meta-generation module in the meta-learner dynamically generates or selects the most suitable meta-path sequence from a large candidate set for the current network state and task.

[0078] Subsequently, information is aggregated based on these dynamically selected meta-paths to generate a second fusion feature. This feature captures specific, immediate information about the current "snapshot" network.

[0079] 3. Parallel fusion of attention: Finally, the system uses a top-level attention mechanism to weightedly fuse the general features (first fusion features) output by the pre-trained model and the immediate features (second fusion features) output by the meta-learner. This allows the model to utilize stable knowledge learned over a long period of time while also responding quickly to changes in the network, resulting in highly accurate network state fusion data.

[0080] The two-path feature aggregation process is as follows:

[0081]

[0082]

[0083]

[0084] in, and These are the features pre-trained from the heterogeneous graph and the features extracted by the heterogeneous graph primitive path learner, respectively.

[0085] It is a learnable attention vector, with a size equal to... and same.

[0086] and There are two scores, indicating and Importance.

[0087] Softmax guarantees and According to the probability distribution, it satisfies .

[0088] S104: Based on the network status fusion data, generate an optimal network switching strategy, which is used to instruct the terminal device to switch networks when performing one-click login service.

[0089] After obtaining the fused data, the system generates decisions through a downstream prediction module.

[0090] Predicting login success rate: Input network state fusion data into a prediction model (such as a simple multilayer perceptron MLP or a module with an attention mechanism), which will predict the success rate of a terminal performing one-click login through different potential paths (e.g., connecting to neighboring cell base station A, neighboring cell base station B) under the current network state.

[0091] Optimal Strategy Generation: The system selects the network path with the highest prediction success rate as the optimal path and generates the corresponding network handover strategy. For example, the strategy could be "suggest that the terminal handover from the current base station X to the neighboring base station Y", or "suggest that the operator's core network node assign the current terminal a public IP address = xxx.xxx.xxx.xxx".

[0092] S105: The model is trained using a hybrid training mode.

[0093] To ensure the model maintains high performance in complex and ever-changing environments, this invention employs a hybrid training mode: Supervised learning: Train an initial, well-performing baseline model using a large amount of historical labeled data (including network status and corresponding login success results).

[0094] Incremental learning: For newly arriving streaming data, instead of completely retraining the model, the parameters are fine-tuned based on the baseline model to adapt to recent changes in network features, saving computational resources.

[0095] Transfer learning: When the system needs to be deployed to a new city or region, the backbone network parameters of the model can be frozen, and only the strategy modules related to the new region can be fine-tuned to achieve rapid knowledge transfer and cross-domain adaptation.

[0096] Online learning: This is key to achieving closed-loop optimization. After the system issues and executes the generated switching strategy, it immediately receives the actual login result (success or failure). This result serves as a new training sample, and the policy generation module is dynamically adjusted in real time using online learning algorithms (such as Q-learning or policy gradient in reinforcement learning) to quickly respond to sudden network problems.

[0097] Generate and execute dynamic network switching schemes, and provide feedback on the results for continuous optimization.

[0098] This is a closed-loop execution and optimization process.

[0099] 1. Based on the optimal network handover strategy generated in step S104, the system forms a specific and executable dynamic network handover scheme.

[0100] 2. This solution is pushed to the operator's core network nodes through a standardized interface.

[0101] 3. The core network node sends instructions to the target terminal to perform network handover operations (such as switching cells).

[0102] 4. After the terminal switches, it continues the one-click login process. The final login result is collected by the feedback module and sent back to the training system as input for online learning, used to optimize the model in the next round of iterations.

[0103] Through the above steps, this invention constructs a network state optimization system capable of comprehensive perception, intelligent decision-making, dynamic adaptation, and closed-loop optimization, thereby effectively solving the problem of the success rate of "one-click login" in weak network environments.

[0104] Please see Figure 4 The diagram illustrates a structural block diagram of an optimization device 400 according to an embodiment of the present invention. The device 400 can implement any of the aforementioned method embodiments. Specifically, the device 400 may include:

[0105] Data Acquisition and Preprocessing Module 401: Responsible for executing S101, comprehensively collecting all multimodal data related to the one-click login service from multiple heterogeneous data sources, including base station information, terminal network status data, operator core network node information, and one-click login operation logs. It performs cleaning, standardization, and feature engineering operations on the collected raw data.

[0106] Heterogeneous graph construction module 402: responsible for executing S102, constructing a multi-dimensional heterogeneous graph containing terminals, base stations, and operator core network nodes based on the preprocessed data.

[0107] Data fusion module 403: Responsible for executing S103, it is the core of the device. It integrates a preset graph neural network model. In a preferred embodiment, this unit can be further divided into a heterogeneous graph pre-training sub-module and a task-oriented meta-learner sub-module, and includes an attention fusion unit for fusing the features output by the two to generate the final network state fusion data.

[0108] Strategy generation module 404: responsible for executing S104. It receives network status fusion data output by data fusion module 404, and predicts the login success rate of different network paths through its internal prediction model, and generates the optimal network switching strategy accordingly.

[0109] Training module 405: responsible for executing S105, using a hybrid training mode to train and continuously optimize the models in data fusion unit 404 and policy generation unit 404.

[0110] Please see Figure 5 This diagram illustrates the hardware structure of an electronic device 500 according to an embodiment of the present invention. The electronic device 500 may be a server, workstation, or dedicated network device. Figure 5 As shown, the electronic device 500 includes a processor 501, a memory 502, and a bus 503 connecting them.

[0111] The memory 502 stores a computer program (i.e., computer execution instructions), which, when executed by the processor 501, can implement the steps in any of the aforementioned method embodiments.

[0112] The processor 501 may be a central processing unit (CPU), a graphics processing unit (GPU), or a dedicated AI chip (such as an ASIC or FPGA). The memory 502 may include random access memory (RAM) and / or non-volatile memory (NVM) such as a hard disk drive (HDD) or a solid-state drive (SSD).

[0113] The present invention also provides a computer-readable storage medium (such as a USB flash drive, optical disc, or hard disk) storing computer instructions that, when executed on a computer, cause the computer to perform any of the aforementioned method embodiments.

[0114] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for optimizing network status through one-click login, characterized in that, include: One-click login multimodal data is obtained, which includes: base station information, network status data of terminal devices connected to the base station, operator core network node information, and one-click login operation log; the one-click login multimodal data is preprocessed to obtain preprocessed multimodal data; A multi-element heterogeneous graph is constructed based on the base station, the terminal device, and the operator core network node, wherein the base station, the terminal device, and the operator core network node are nodes, and the connection relationships between nodes are edges. The multivariate heterogeneous graph is input into a preset graph neural network, and the preprocessed multimodal data is fused through the graph neural network to obtain network state fusion data; including: processing the multivariate heterogeneous graph through a heterogeneous graph pre-trained model to obtain a first fusion feature; learning the multivariate heterogeneous graph through a heterogeneous graph meta-learner to dynamically generate and select meta-paths suitable for the current one-click login optimization task; aggregating information from the multivariate heterogeneous graph based on the selected meta-paths to obtain a second fusion feature; and using an attention mechanism to fuse the first fusion feature and the second fusion feature in parallel. Based on the network status fusion data, an optimal network switching strategy is generated, which is used to instruct the terminal device to switch networks when performing the one-click login service.

2. The method according to claim 1, characterized in that, The preprocessing includes: data cleaning, data integration, feature extraction, and normalization.

3. The method according to claim 1, characterized in that, The characteristics of the edges and / or nodes of the multi-dimensional heterogeneous graph include at least one of the following: signal strength, network latency, packet loss rate, transmission rate, one-click login success rate, one-click login response time, transmission bandwidth, signaling traffic, congestion index, handover success rate, and link reliability.

4. The method according to claim 1, characterized in that, The method further includes: The graph neural network is trained using a hybrid training mode, which includes at least one of supervised learning, incremental learning, transfer learning, and online learning.

5. The method according to claim 4, characterized in that, The step of training the graph neural network using online learning includes: The dynamic network switching scheme generated by the optimal network switching strategy is pushed to the operator's core network node for the terminal device to execute; Collect the login result after the terminal device performs the switch as feedback data; The graph neural network is optimized in real time based on the feedback data.

6. A one-click login network status optimization device, characterized in that, include: The data acquisition and preprocessing module is used to acquire one-click login multimodal data, which includes: base station information, network status data of terminal devices connected to the base station, operator core network node information, and one-click login operation logs; and to preprocess the one-click login multimodal data to obtain preprocessed multimodal data. The heterogeneous graph construction module is used to construct a multi-dimensional heterogeneous graph based on the base station, the terminal device, and the operator core network node, wherein the base station, the terminal device, and the operator core network node are nodes, and the connection relationship between nodes is an edge. A data fusion module is used to input the multivariate heterogeneous graph into a preset graph neural network, and fuse the preprocessed multimodal data through the graph neural network to obtain network state fusion data; including: processing the multivariate heterogeneous graph through a heterogeneous graph pre-training model to obtain a first fusion feature; learning the multivariate heterogeneous graph through a heterogeneous graph meta-learner to dynamically generate and select meta-paths suitable for the current one-click login optimization task; aggregating information from the multivariate heterogeneous graph based on the selected meta-paths to obtain a second fusion feature; and using an attention mechanism to fuse the first fusion feature and the second fusion feature in parallel; The strategy generation module is used to generate an optimal network switching strategy based on the network status fusion data. The optimal network switching strategy is used to instruct the terminal device to switch networks when performing the one-click login service.

7. The apparatus according to claim 6, characterized in that, The data fusion module is specifically used for: The multivariate heterogeneous graph is processed by a heterogeneous graph pre-trained model to obtain the first fusion feature; The heterogeneous graph is learned by a heterogeneous graph meta-learner, and meta-paths suitable for the current one-click login optimization task are dynamically generated and selected. Based on the selected meta-path, information is aggregated from the multi-dimensional heterogeneous graph to obtain the second fusion feature; An attention mechanism is used to fuse the first fusion feature and the second fusion feature in parallel.

8. An electronic device, characterized in that, include: Processor, and memory; The memory is used to store computer execution instructions, and the processor is connected to the memory via a bus; When the electronic device is running, the processor executes the computer execution instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-5.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1-5.

Images