Opportunity network link prediction method based on time sequence diagram neural network

Abstract

The invention discloses an opportunity network link prediction method based on a time sequence diagram neural network, and particularly relates to the technical field of the time sequence diagram neural network, and the method comprises the following steps: S1, data collection and network snapshot generation; s2, weighting feature coding clusters; s3, GAT and LSTM combined processing embedding is carried out; s4, the probability is calculated through a multi-layer perceptron; s5, constructing a loss function according to the link; s6, drawing sets and training parameters; and S7, multi-index evaluation of the optimal model. The method accurately captures the spatial-temporal characteristics of nodes, collaboratively operates embedding, coding and decoding layers, improves the accuracy and stability of link prediction, stabilizes the reliability of network communication and the success rate of data transmission, innovates a loss function and an optimization strategy, solves the problems of data sparseness and over-fitting, enhances the generalization ability, adapts the model to a multi-field network, and improves the data transmission efficiency. And links are accurately predicted in scenes such as intelligent traffic and disaster rescue, resource configuration and communication efficiency are optimized, and innovative development of multi-industry network application is powerfully promoted.

Description

Technical Field

[0001] The present invention relates to the technical field of time-series graph neural networks, and in particular to an opportunistic network link prediction method based on time-series graph neural networks. Background Art

[0002] With the vigorous development of wireless communication technology and equipment, although the traditional wireless communication network that relies on base stations is widely used, it is vulnerable. MANET was born, which can self-organize communication, but the early routing protocol has limitations in dynamic scenarios. In order to break through this dilemma, DTN and its derivative opportunistic networks emerged. The latter has been widely used in many fields with its unique advantages, and its position in the military field is becoming increasingly prominent. Link prediction is of great significance to opportunistic networks, but it faces many difficult challenges, such as the dynamic nature of the network causing the failure of traditional methods, sparse data that makes it difficult to support prediction, feature extraction is difficult due to a variety of complex factors, the design of the prediction model needs to take into account both efficiency and generality, and data fusion is limited by format and information conflicts. Many problems need to be overcome.

[0003] Chinese patent announcement number CN106934489B discloses a method for predicting time links for complex networks. It uses the evolution information of the network to perform high-precision link prediction, and designs the core steps of the invention based on the overall synchronous parallel computing model. The prediction method of the invention has good universality and can be applied to the prediction of time links in a variety of social networks; and the prediction method of the invention has good scalability and can be applied to the prediction of time links in distributed environments;

[0004] However, the above methods only mine shallow features based on interaction time and frequency, and the algorithms lack flexibility and are difficult to adapt to diverse networks. Summary of the invention

[0005] The main purpose of the present invention is to provide an opportunistic network link prediction method based on a temporal graph neural network, which can effectively solve the problems of shallow feature mining, relying only on interaction time and frequency, and lack of algorithm flexibility.

[0006] To achieve the above object, the technical solution adopted by the present invention is:

[0007] A method for predicting opportunistic network links based on a time-series graph neural network comprises the following steps:

[0008] S1: Comprehensively collect multi-channel data within the opportunity network, organize the data into a series of network snapshots in an orderly manner according to the predetermined time span rules, construct a weighted adjacency matrix to reflect the details of node connection interactions, and after normalization, consolidate the underlying data architecture for link prediction.

[0009] S2: After receiving the weighted adjacency matrix and node feature materials, the richness of feature dimensions is improved through encoding fusion operations, and the key features of the nodes are accurately focused on by means of attention calculation. The embedding matrix is ​​generated by weighted aggregation to capture the structural characteristics of the node and transfer the encoding process to deeply extract the feature connotation.

[0010] S3: Taking node embedding materials as the starting point, the spatial structure between nodes is mined with the help of graph attention network, and the dynamic evolution trajectory of the node historical feature sequence is analyzed using long short-term memory network. The hidden states are spliced ​​and condensed into spatiotemporal embedding representations. The self-attention mechanism is used to dynamically adjust the weights according to the node coreness and time context, and the weights are transmitted to the decoding link after optimization.

[0011] S4: Start the multi-layer perceptron architecture, introduce the spatiotemporal embedding results of the coding layer into the input layer, use the deep abstract conversion function of the hidden layer, rely on the sophisticated processing of the activation function, and through multi-layer progressive conversion, map the features into a link generation probability matrix, clarify the probability of node emergence to the link, and provide a core decision-making reference for link prediction.

[0012] S5: Build the core architecture of the loss function around the predicted and real links, formulate a dynamic weighting strategy to address the link sparsity dilemma, integrate complexity management technology and label optimization methods, and use the cross-validation mechanism to fine-tune the weight parameters to achieve an ideal balance between model prediction accuracy and generalization breadth, and consolidate the model performance foundation.

[0013] S6: Cut the data set into training, validation, and testing modules according to reasonable proportions, and integrate simulated annealing to expand the search breadth and stochastic gradient descent to accelerate convergence to drive the training process. Flexible adjustment of the learning rate based on the dynamic rhythm of the network, multiple rounds of iterations to accurately refresh the model weight parameters and improve the prediction performance.

[0014] S7: Use multiple indicators to evaluate the prediction effect and compare it with the baseline, develop a visual interface to display links, node trajectories, topology evolution and indicator trends, optimize the model based on feedback, and intuitively present and improve performance.

[0015] Preferably, the S1 step includes a multi-source data acquisition system, data preprocessing software, a network snapshot generation program and an adjacency matrix construction tool.

[0016] Preferably, the S2 step includes a feature encoding and fusion module, a dimensionality enhancement subsystem, a multi-head graph attention calculation engine and an embedding matrix generation software.

[0017] Preferably, the S3 step includes a GAT-LSTM network model, a hidden state concatenation program, a multi-head self-attention allocation system, and a feature optimization and output module.

[0018] Preferably, the S4 step includes a multi-layer perceptron architecture, a hidden layer processing component and a Softmax conversion unit.

[0019] Preferably, the S5 step includes a loss function builder, an adaptive weighting strategy, an L2 regularization and label smoothing tool, and a cross-validation optimizer.

[0020] Preferably, the S6 step includes a data set partitioning tool, a hybrid optimization training system, a learning rate adaptive adjustment module and a model updating program.

[0021] Preferably, the S7 step includes a multi-index evaluation system, a baseline comparison software, a visualization development tool and a model feedback optimizer.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] 1. The present invention uses a unique TGNN model architecture to accurately capture the spatiotemporal characteristics between nodes in the opportunistic network. The GAT of the embedding layer accurately analyzes the node structure association, the LSTM of the encoding layer and the multi-head self-attention work together to capture the temporal evolution, the decoding layer efficiently maps the features to the probability space, and the multi-mechanism fusion constructs a strong prediction capability model, providing solid support for network communication planning, greatly enhancing the accuracy and stability of link prediction, and improving the reliability of network communication and the success rate of data transmission.

[0024] 2. The present invention effectively resolves the challenges of opportunistic networks through innovative loss functions and optimization strategies. The adaptive weighting strategy resolves the problem of data sparsity, L2 regularization and label smoothing suppress overfitting and improve generalization, and multiple optimization algorithms work together to ensure that the model is robust and efficient in dynamic and complex environments. The model can flexibly adapt to the characteristics of multiple domain networks, and show strong adaptability in scenarios such as intelligent transportation and disaster relief. It accurately predicts links according to scenario requirements, optimizes resource allocation and communication efficiency, and effectively promotes the development and innovation of network applications in multiple industries. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 The overall process of the present invention Figure 1 ;

[0026] Figure 2 The process of the present invention Figure 2 ;

[0027] Figure 3 The process of the present invention Figure 3 ;

[0028] Figure 4 The process of the present invention Figure 4 ;

[0029] Figure 5 The process of the present invention Figure 5 ;

[0030] Figure 6 The process of the present invention Figure 6 ;

[0031] Figure 7 The process of the present invention Figure 7 ;

[0032] Figure 8 The process of the present invention Figure 8 . DETAILED DESCRIPTION

[0033] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0034] See also Figure 1-8 , an opportunistic network link prediction method based on a temporal graph neural network, comprising the following steps:

[0035] A method for predicting opportunistic network links based on a time-series graph neural network, the steps of which include:

[0036] S1: Comprehensively collect multi-channel data within the opportunity network, organize the data into a series of network snapshots in an orderly manner according to the predetermined time span rules, construct a weighted adjacency matrix to reflect the details of node connection interactions, and after normalization, consolidate the underlying data architecture for link prediction.

[0037] S2: After receiving the weighted adjacency matrix and node feature materials, the richness of feature dimensions is improved through encoding fusion operations, and the key features of the nodes are accurately focused on by means of attention calculation. The embedding matrix is ​​generated by weighted aggregation to capture the structural characteristics of the node and transfer the encoding process to deeply extract the feature connotation.

[0038] S3: Taking node embedding materials as the starting point, the spatial structure between nodes is mined with the help of graph attention network, and the dynamic evolution trajectory of the node historical feature sequence is analyzed using long short-term memory network. The hidden states are spliced ​​and condensed into spatiotemporal embedding representations. The self-attention mechanism is used to dynamically adjust the weights according to the node coreness and time context, and the weights are transmitted to the decoding link after optimization.

[0039] S4: Start the multi-layer perceptron architecture, introduce the spatiotemporal embedding results of the coding layer into the input layer, use the deep abstract conversion function of the hidden layer, rely on the sophisticated processing of the activation function, and through multi-layer progressive conversion, map the features into a link generation probability matrix, clarify the probability of node emergence to the link, and provide a core decision-making reference for link prediction.

[0040] S5: Build the core architecture of the loss function around the predicted and real links, formulate a dynamic weighting strategy to address the link sparsity dilemma, integrate complexity management technology and label optimization methods, and use the cross-validation mechanism to fine-tune the weight parameters to achieve an ideal balance between model prediction accuracy and generalization breadth, and consolidate the model performance foundation.

[0041] S6: Cut the data set into training, validation, and testing modules according to reasonable proportions, and integrate simulated annealing to expand the search breadth and stochastic gradient descent to accelerate convergence to drive the training process. Flexible adjustment of the learning rate based on the dynamic rhythm of the network, multiple rounds of iterations to accurately refresh the model weight parameters and improve the prediction performance.

[0042] S7: Use multiple indicators to evaluate the prediction effect and compare it with the baseline, develop a visual interface to display links, node trajectories, topology evolution and indicator trends, optimize the model based on feedback, and intuitively present and improve performance.

[0043] See also Figure 2 The S1 step includes a multi-source data acquisition system, data preprocessing software, a network snapshot generation program and an adjacency matrix construction tool.

[0044] Specifically, the multi-source data acquisition system can collect data from multiple data sources such as sensors, communication nodes, logging devices, etc. in the opportunistic network, such as communication timestamps between nodes, signal strength, node geographic location coordinate information, node type or functional attribute identification, etc.

[0045] Furthermore, data preprocessing software can clean the collected data, remove obviously erroneous or abnormal data points, fill in missing data through interpolation or mean filling, and uniformly convert data in different formats into a format suitable for subsequent processing.

[0046] Furthermore, the network snapshot generation program slices the preprocessed data according to a preset time period (such as every 5 minutes or 10 minutes, etc.) to form network snapshots. Each snapshot represents the network status within the time period and records information such as the connection relationship and interaction frequency between nodes.

[0047] Furthermore, the adjacency matrix construction tool constructs the initial adjacency matrix based on the node connections in the network snapshot. If node i and node j are connected in the snapshot, the element at the corresponding position is set to 1, and if there is no connection, it is set to 0. t = {V,E t ,W t The weight W of the edge in t It can be expressed as the total duration of the connection between nodes in the snapshot graph divided by the number of connections. The formula is as follows:

[0048]

[0049] in, f ij is the number of connections of node (i, j) on the tth network snapshot, ∑d ij is the total connection duration of node (i, j) on the tth network snapshot, and Δt represents the slice duration.

[0050] Implementation method: First, start the multi-source data collection system to collect data from various devices and data sources of the opportunity network. After the collection is completed, the data is input into the data preprocessing software for processing to ensure the quality and format consistency of the data. Then, run the network snapshot generation program to generate a network snapshot sequence according to the predetermined time granularity. Finally, use the adjacency matrix construction tool to construct the initial adjacency matrix, calculate the edge weights and perform normalization to obtain the weighted adjacency matrix, which provides a data basis for subsequent steps.

[0051] See also Figure 3 , the S2 step includes a feature encoding and fusion module, a dimensionality enhancement subsystem, a multi-head graph attention calculation engine and an embedding matrix generation software.

[0052] Specifically, the feature encoding and fusion module fuses the node's one-hot encoding (used to represent the node's unique identity) and position encoding (reflecting the node's location information in space) through a specific algorithm to generate an initial feature vector that can comprehensively represent the node's identity and position.

[0053] Furthermore, the dimensionality enhancement subsystem uses a learnable linear transformation matrix to process the initial feature vector and increase the dimension of the feature vector so as to better capture the complex relationship of node features.

[0054] Furthermore, the multi-head graph attention calculation engine uses the multi-head graph attention mechanism to set multiple attention heads, each of which focuses on different subspaces of the feature vector. The attention coefficient calculation formula can be referred to:

[0055]

[0056]

[0057] Among them, a T represents a shared attention mechanism, θ represents a shared linear transformation matrix, and then an activation function LeakyReLU function is used for activation. The above formula describes the importance of the feature of node j to the updated feature of node i. Finally, the softmax function is used to update e ij After normalization, n(i) represents the first-order neighbor node set of node i.

[0058] Secondly, after introducing the multi-head attention mechanism, the calculation formula for calculating the average feature value between nodes is as follows:

[0059]

[0060] By independently calculating and averaging multiple attention networks, the model's feature extraction capabilities are enriched and the learning process is stabilized, and attention coefficients are learned from multiple mapping subspaces to improve model performance.

[0061] Furthermore, the embedding matrix generation software can perform weighted aggregation of the attention coefficients calculated by each attention head and the corresponding node features to obtain the embedding matrix of the node. The formula for mapping the attention coefficient to the snapshot adjacency matrix is ​​as follows:

[0062]

[0063] Among them, α ij In order to take into account the attention coefficients obtained by the multi-head attention mechanism, a set of attention coefficient matrices can be obtained through each snapshot of the opportunity network. The dynamic opportunity network spatial features can be represented by a set of three-dimensional variables, as follows:

[0064]

[0065] in, T is the number of snapshots, and N is the number of opportunistic network nodes.

[0066] By automatically learning the relationship between nodes through the attention network, the attention coefficient matrix calculated can intuitively show the spatial correlation between nodes, and then analyze the evolution of the attention coefficient matrix and observe the temporal characteristics between nodes.

[0067] Implementation method: After receiving the weighted adjacency matrix and the initial feature vector of the node, the feature encoding and fusion module first performs encoding and fusion operations on the node features to generate the initial feature vector. Then, the dimensionality enhancement subsystem performs dimensionality enhancement processing on the vector. Then, the multi-head graph attention calculation engine calculates the attention coefficient between nodes according to the above formula, and multiple attention heads calculate the attention of different subspaces in parallel. Finally, the embedding matrix generation software performs weighted aggregation on the results of each attention head, generates a node embedding matrix, and passes it to the encoding layer.

[0068] See also Figure 4 The S3 step includes a GAT-LSTM network model, a hidden state concatenation procedure, a multi-head self-attention allocation system, and a feature optimization and output module.

[0069] Specifically, the GAT-LSTM network model combines the advantages of the graph attention network (GAT) and the long short-term memory network (LSTM). When processing the link prediction task of the opportunistic network, it works based on the weighted adjacency matrix and the node feature vector, and accurately calculates the node attention coefficient through the multi-head graph attention mechanism. This mechanism can mine the complex structural associations between nodes from multiple angles, and then perform weighted aggregation operations on these coefficients to generate an embedding matrix. This embedding matrix accurately captures the structural characteristics of the node and provides a high-quality node representation foundation for subsequent LSTM processing.

[0070] Furthermore, the forward and reverse hidden states generated by the LSTM module in GAT-LSTM when processing the node embedding sequence are spliced ​​and integrated to achieve the convergence of multi-directional features. The formula for GAT-LSTM to extract the temporal features of network evolution and obtain the spatiotemporal embedding representation is as follows:

[0071]

[0072] Furthermore, the multi-head self-attention allocation system is based on the multi-head self-attention mechanism. According to the importance of the node in the network and the time dependency, it dynamically allocates attention weights to different feature dimensions of the node. The formula for defining the output of the multi-head self-attention mechanism is as follows:

[0073]

[0074] MultiHead(Q,K,V)=(head1;...;head i )

[0075] wherehead i =Attention(QW i Q ,KW i K ,VW i V )

[0076] Furthermore, the feature optimization and output module further optimizes the features after multi-head self-attention weighted update, such as feature compression or feature enhancement, to improve the expressiveness and effectiveness of the features, and then outputs the optimized spatiotemporal embedding representation to the decoding layer.

[0077] Implementation method: The GAT-LSTM network model processing flow is started with the node embedding matrix as input. GAT uses the weighted adjacency matrix and node feature vector to enable the multi-head graph attention mechanism, carefully calculate the node attention coefficient, deeply explore the multi-subspace structure association, and weighted aggregation to build the embedding matrix, which is the cornerstone for the subsequent delivery of high-quality features. After receiving, LSTM conducts a two-way analysis of the node historical feature sequence, accurately extracts the forward and reverse hidden states, and splices them into a spatiotemporal embedding representation according to a rigorous formula to keenly capture the dynamic evolution law. Then, the multi-head self-attention allocation system calculates the attention coefficient according to the above formula and performs weighted updates on the features. Finally, the feature optimization and output module optimizes the weighted features and outputs them to the decoding layer.

[0078] See also Figure 5 , the S4 step includes a multi-layer perceptron architecture, a hidden layer processing component and a Softmax conversion unit.

[0079] Specifically, the multi-layer perceptron architecture is a neural network architecture consisting of an input layer, several hidden layers, and an output layer. The input layer receives the spatiotemporal embedding representation from the encoding layer, and performs multi-layer feature abstraction and transformation through the hidden layer to gradually extract high-level features related to link prediction.

[0080] Furthermore, the hidden layer processing component uses the rectified linear unit (ReLU) as the activation function, which can increase the nonlinear expression ability of the network, speed up the model training and reduce the gradient vanishing problem. By stacking multiple hidden layers, the input spatiotemporal embedding representation is deeply abstracted.

[0081] Furthermore, the Softmax conversion unit uses the Softmax function in the output layer of the multi-layer perceptron to convert the features output by the last hidden layer into a link generation probability matrix between node pairs.

[0082] The calculation method of each layer of the neural network is as follows:

[0083]

[0084]

[0085]

[0086] Among them, h is the output result of the attention layer and They represent the weight matrix and bias coefficient matrix of the kth layer of the neural network respectively. The activation function of the hidden layer adopts the sigmoid function, and the activation function of the output layer adopts the softmax function. The above formula clarifies the calculation method of each layer in the fully connected neural network structure, including the linear transformation from the input through the hidden layer to the output layer and the activation function processing to map the time series features to the probability space output link generation probability.

[0087] Implementation method: The input layer of the multi-layer perceptron architecture receives the spatiotemporal embedding representation from the encoding layer. The data enters the hidden layer processing component. Each neuron in the hidden layer first performs a linear transformation on the input, and then performs a nonlinear transformation through the ReLU activation function. After layer-by-layer processing of multiple hidden layers, more advanced features are extracted. Finally, the Softmax conversion unit converts the output of the last hidden layer into a link generation probability matrix through the Softmax function. The matrix represents the predicted probability of the existence of a link between each pair of nodes in the network.

[0088] See also Figure 6 The S5 step includes a loss function builder, an adaptive weighting strategy, an L2 regularization and label smoothing tool, and a cross-validation optimizer.

[0089] Specifically, the loss function builder constructs a loss function based on the predicted link probability matrix and the actual link label to measure the difference between the model prediction results and the actual situation.

[0090] Furthermore, the adaptive weighting strategy takes into account the sparse links in opportunistic networks. For accurate prediction, dynamic weights are set for positive samples (samples with real links). The influence of the samples in the loss function is increased according to their characteristics (such as proportion of quantity and prediction difficulty), thereby improving the model's sensitivity to positive examples.

[0091] Furthermore, L2 regularization and label smoothing tools: L2 regularization adds regularization terms to the loss function to constrain model complexity and prevent overfitting from causing generalization degradation; label smoothing fine-tunes the true label value, eases model confidence, and enhances generalization. The two work together to stabilize the model and improve performance.

[0092] Furthermore, the cross-validation optimizer uses cross-validation technology to split the data set into multiple subsets, cyclically use the subsets for validation and the rest for training, optimize the weight parameters according to the validation set indicators (loss value, accuracy, etc.), and use search strategies (grid, random search) to find the optimal combination in the parameter domain to ensure good model generalization.

[0093] Among them, the formula for improving the loss function calculation method is as follows:

[0094]

[0095] Aiming at the sparsity of opportunity networks, a penalty value is introduced to improve the loss function based on the L2 norm. By processing the difference between the predicted value and the true value and setting the penalty value in different situations, the model bias is avoided to ensure the accuracy of zero-element prediction and overfitting.

[0096] For each training process, if Then p = 1; if Then p = β>1, β is a positive value, indicating the penalty coefficient. This penalty matrix imposes a greater penalty on non-zero elements, thereby avoiding overfitting of the model to a certain extent. At the same time, this paper introduces the regularization term L based on the L2 norm. reg , the regularization loss is determined by calculating the sum of the squared norms of all weights in the model to ensure that the inputs between neural network layers have the same uniform distribution. Finally, a hybrid loss function is used to train and optimize the model, and the formula is as follows:

[0097] L total =L+αL reg

[0098] Among them, α is the proportion of regularization loss in the loss function.

[0099] Furthermore, an adaptive weighting strategy is used: in view of the sparsity of links in opportunistic networks, dynamically increased weights are set for positive samples (i.e., samples with real links).

[0100] Implementation method: The loss function builder constructs the core of the loss function based on the predicted and real link matrices, and selects cross entropy and other methods. The adaptive weighting strategy calculates weights based on the features of the positive sample and applies them to the loss function. The L2 regularization and label smoothing tools respectively add regularization terms and smooth labels to improve the loss function. The cross-validation optimizer divides the data set, conducts multiple rounds of training and verification, and uses algorithms and search strategies such as gradient descent to find the best parameter group within the parameter range to reduce loss and improve generalization.

[0101] See also Figure 7 The S6 step includes a data set partitioning tool, a hybrid optimization training system, a learning rate adaptive adjustment module and a model updating program.

[0102] Specifically, the dataset partitioning tool splits the opportunity network dataset into training set, validation set, and test set according to a specific ratio (such as 60% training, 20% validation, and 20% testing) to ensure that the data distribution reflects the overall characteristics, lays the foundation for model evaluation, prevents overfitting and underfitting, and balances model learning and evaluation.

[0103] Furthermore, the hybrid optimization training system integrates simulated annealing and stochastic gradient descent algorithms to optimize training. Simulated annealing allows acceptance of inferior solutions in the initial stage to avoid local optimality and expand the search domain; stochastic gradient descent uses sample gradients to quickly update parameters to promote convergence, and collaborates during iteration. Stochastic gradient descent is used first to obtain gradient updates, and then simulated annealing is used to determine acceptance, thereby improving training efficiency and effectiveness.

[0104] Furthermore, the learning rate adaptive adjustment module dynamically adjusts the learning rate based on node mobility (speed, direction change) and network dynamics (topology frequency change, link duration distribution) indicators. When the network is turbulent, the learning rate is increased to accelerate model adaptation; when it is stable, the learning rate is reduced to stabilize training and fine-tune parameters to ensure that the model accurately tracks network characteristics.

[0105] The model update procedure training iteration updates the model weight parameters according to the gradient and learning rate.

[0106] Implementation method: The dataset partitioning tool divides the dataset according to proportion to prepare data for model training and evaluation. The hybrid optimization training system uses algorithms to collaboratively update parameters in the training loop. The learning rate adaptive adjustment module monitors network characteristics and calculates the learning rate. The model update program iteratively updates the weights with gradients and learning rates. Multiple rounds of optimization are performed until the model performance steadily increases and adapts to network dynamics.

[0107] See also Figure 8 The S7 step includes a multi-index evaluation system, a baseline comparison software, a visualization development tool and a model feedback optimizer.

[0108] Specifically, the multi-index evaluation system uses multiple indicators such as accuracy, recall, F1 value, mean square error and AUC-ROC curve to quantitatively evaluate the model link prediction performance from the dimensions of prediction accuracy, comprehensiveness, error and classification performance, and analyze the advantages and disadvantages and generalization ability of the model from multiple angles.

[0109] Furthermore, the baseline comparison software compares the prediction results of this model with those of the baseline model (traditional statistics, simple machine learning model), explains the advantages and disadvantages of the model based on indicator values ​​(accuracy, recall rate difference, etc.), verifies its advancedness in opportunistic network link prediction, and helps researchers identify optimization directions.

[0110] Furthermore, visualization development tools create visualization interfaces that intuitively present predicted links, node trajectories, topology changes, and indicator trends in the form of graphics (node-link diagrams, trajectory diagrams, topology evolution animations, and indicator trend line charts), helping operation and maintenance personnel gain insight into network performance and prediction status, and providing a visualization basis for decision optimization.

[0111] Furthermore, the model feedback optimizer digs deep into the model's deficiencies based on visual feedback and evaluation results. For example, if the link prediction in a certain area is poor, it will analyze the processed data and adjust the parameters in detail. It will optimize the training strategy (variable learning rate strategy, data set ratio) according to the indicator trend, and cyclically optimize to improve the model's performance and adaptability, thereby achieving accurate link prediction.

[0112] Implementation method: The multi-indicator evaluation system calculates the indicator values ​​of the model test set, the baseline comparison software compares the model with the baseline indicators to find the gap, the visualization development tool visualizes the prediction, node, topology, and indicator information, and the model feedback optimizer focuses on the problem area based on the feedback results, adjusts the model data processing, parameters, and training strategies according to the indicator trends, and improves performance through multiple cycles to adapt to the network link prediction needs.

[0113] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for predicting opportunistic network links based on a time-series graph neural network, comprising the following steps: S1: Collect multi-channel data within the opportunistic network, organize the data into a series of network snapshots according to the predetermined time span rules, build a matrix to reflect the details of node connection interactions, and after normalization, consolidate the underlying data architecture for link prediction. S2: After receiving the weighted adjacency matrix and node feature materials, the richness of feature dimensions is improved through encoding fusion operations, and the key features of the nodes are accurately focused on by means of attention calculation. The weighted aggregation generates an embedding matrix to capture the structural characteristics of the node and transfer the encoding process. S3: Taking node embedding materials as the starting point, the spatial structure between nodes is mined with the help of graph attention network, and the dynamic evolution trajectory of the node historical feature sequence is analyzed using long short-term memory network. The hidden states are spliced ​​and condensed into spatiotemporal embedding representations. The self-attention mechanism is used to dynamically adjust the weights according to the node coreness and time context, and the weights are transmitted to the decoding link after optimization. S4: Start the multi-layer perceptron architecture, introduce the spatiotemporal embedding results of the coding layer into the input layer, use the deep abstract conversion function of the hidden layer, rely on the sophisticated processing of the activation function, and through progressive conversion, map the features into a link generation probability matrix, clarify the probability of node emergence to the link, and provide a core decision-making reference for link prediction. S5: Build the core architecture of the loss function around the predicted and real links, formulate a dynamic weighting strategy to address the link sparsity dilemma, integrate complexity management technology and label optimization methods, use the cross-validation mechanism to fine-tune the weight parameters, and consolidate the model performance foundation. S6: Divide the data set into training, validation, and testing modules, integrate simulated annealing to expand the search breadth and stochastic gradient descent to accelerate convergence to drive the training process, flexibly adjust the learning rate according to the dynamic rhythm of the network, and iteratively refresh the model weight parameters. S7: Use multiple indicators to evaluate the prediction effect and compare it with the baseline, develop a visual interface to display the links, node trajectories, topology evolution and indicator trends, optimize the model based on feedback, and intuitively present and improve performance.

2. The method for predicting opportunistic network links based on a time-series graph neural network according to claim 1, characterized in that: The S1 step includes a multi-source data acquisition system, data preprocessing software, a network snapshot generation program and an adjacency matrix construction tool.

3. The method for predicting opportunistic network links based on a time-series graph neural network according to claim 1, characterized in that: The S2 step includes a feature encoding and fusion module, a dimensionality enhancement subsystem, a multi-head graph attention calculation engine and an embedding matrix generation software.

4. The method for predicting opportunistic network links based on a time-series graph neural network according to claim 1, characterized in that: The S3 step includes a GAT-LSTM network model, a hidden state concatenation procedure, a multi-head self-attention allocation system, and a feature optimization and output module.

5. The method for predicting opportunistic network links based on a time-series graph neural network according to claim 1, characterized in that: The S4 step includes a multi-layer perceptron architecture, a hidden layer processing component and a Softmax conversion unit.

6. The method for predicting opportunistic network links based on a time-series graph neural network according to claim 1, characterized in that: The S5 step includes a loss function builder, an adaptive weighting strategy, an L2 regularization and label smoothing tool, and a cross-validation optimizer.

7. The method for predicting opportunistic network links based on a time-series graph neural network according to claim 1, characterized in that: The S6 step includes a data set partitioning tool, a hybrid optimization training system, a learning rate adaptive adjustment module and a model updating program.

8. The method for predicting opportunistic network links based on a time-series graph neural network according to claim 1, characterized in that: The S7 step includes a multi-index evaluation system, a baseline comparison software, a visualization development tool and a model feedback optimizer.

Images