Metacosm resource scheduling method, device, equipment, medium and product

Abstract

The invention discloses a meta-universe resource scheduling method and device, equipment, a medium and a product, and the method comprises the steps: carrying out the resource demand prediction through a pre-trained graph neural network model, and making a first resource allocation strategy based on a generated resource demand prediction value; and analyzing resource use data by using the graph neural network model, dynamically adjusting the first resource allocation strategy according to an analysis result, and updating to obtain a second resource allocation strategy. According to the method, the self-learning and self-adaptive capabilities of the graph neural network are utilized, analysis and adjustment are carried out according to the dynamic resource use data monitored in real time, the resource allocation strategy is dynamically adjusted, and it is ensured that the stable performance of the system can still be maintained when the resource demand suddenly increases or changes.

Description

Technical Field

[0001] The present application relates to the field of electronic digital data processing technology, and in particular to methods, devices, equipment, media and products for scheduling metaverse resources. Background Art

[0002] With the development of IT cloud-native technology, the complexity of application systems has increased exponentially, requiring continuous investment in various types of operations, maintenance, and developers to participate in business and platform stability assurance, but discrete personnel deployment requires a new collaborative model to efficiently complete work tasks. As a virtual digital space, the Metaverse is increasingly becoming an important part of people's lives and work.

[0003] In the metaverse environment, there are a large number of nodes and edges. These nodes include users, virtual items, and servers, while edges include user interactions, item transactions, and server connections. Traditional resource scheduling methods are usually based on static rules or predefined optimization strategies, such as polling, priority, or simple load balancing algorithms. These strategies cannot adapt to the dynamically changing relationships between nodes and edges in the metaverse. In the metaverse, the relationship between users, virtual items, and service nodes may change at any time. Traditional methods find it difficult to dynamically adjust resource allocation strategies to cope with this complexity and variability. Summary of the invention

[0004] The main purpose of this application is to provide a metaverse resource scheduling method, device, equipment, medium and product, aiming to solve the technical problem of untimely dynamic resource scheduling existing in traditional resource scheduling methods.

[0005] To achieve the above objectives, the present application provides a metaverse resource scheduling method, which comprises the following steps:

[0006] Using a pre-trained graph neural network model to predict resource demand, and formulating a first resource allocation strategy based on the generated resource demand prediction value;

[0007] The graph neural network model is used to analyze the resource usage data, and the first resource allocation strategy is dynamically adjusted according to the analysis results to update the second resource allocation strategy.

[0008] In one embodiment, the step of analyzing the resource usage data using the graph neural network model, dynamically adjusting the first resource allocation strategy according to the analysis result, and updating the second resource allocation strategy further includes:

[0009] According to the pre-calculated priority index, the second resource allocation strategy is optimized through the graph neural network model to generate a resource allocation plan.

[0010] In one embodiment, the step of optimizing the second resource allocation strategy through the graph neural network model according to the pre-calculated priority index and generating a resource allocation solution further includes:

[0011] The resource scheduling operation is performed according to the resource allocation scheme, and the resource configuration is dynamically adjusted.

[0012] In one embodiment, the step of constructing graph structure data in a metaverse environment, optimizing the second resource allocation strategy through the graph neural network model according to a pre-calculated priority index, and generating a resource allocation solution comprises:

[0013] Calculate the importance parameters and resource utilization of nodes and edges in the Metaverse environment;

[0014] Calculating the priority indexes of nodes and edges respectively according to the user demand parameter, the importance parameter and the resource utilization rate;

[0015] The graph neural network model is used to optimize the second resource allocation strategy according to the priority indicators of nodes and edges to generate a resource allocation plan.

[0016] In one embodiment, the step of using a pre-trained graph neural network model to predict resource demand and formulating a first resource allocation strategy based on the generated resource demand prediction value includes:

[0017] Use the pre-trained graph neural network model to predict the resource demand of nodes and edges in the pre-built graph structure data, and obtain the resource demand prediction values ​​of nodes and edges respectively;

[0018] According to the predicted values ​​of resource requirements of nodes and edges, the resource allocation of nodes and edges is calculated respectively.

[0019] In one embodiment, the step of using a pre-trained graph neural network model to predict resource requirements for nodes and edges in pre-constructed graph structure data to obtain resource requirement prediction values ​​for nodes and edges respectively further includes:

[0020] Generate a graph dataset by extracting feature vectors of nodes and edges from pre-constructed graph structure data;

[0021] The pre-built graph neural network model is trained using the graph dataset.

[0022] In one embodiment, the step of training the pre-built graph neural network model using the graph dataset includes:

[0023] Inputting the graph dataset into a pre-built graph neural network model;

[0024] Using a back-propagation algorithm, the gradient of the loss function with respect to the parameters of the graph neural network model is calculated;

[0025] Using a gradient descent algorithm, updating the parameters of the graph neural network model according to the gradient;

[0026] The loss function is minimized through iterative training until the loss function converges to a preset threshold, thereby obtaining an optimized graph neural network model.

[0027] In one embodiment, the step of analyzing the resource usage data using the graph neural network model, dynamically adjusting the first resource allocation strategy according to the analysis result, and updating the second resource allocation strategy includes:

[0028] Monitor the resource usage data of nodes and edges through the pre-deployed resource monitoring module, and dynamically analyze the resource usage data to calculate the resource usage rate;

[0029] Analyze the input resource usage data using the graph neural network model to calculate the resource usage change trend of nodes and edges;

[0030] Using the graph neural network model to identify bottleneck nodes and bottleneck edges, and based on the resource utilization rate, calculating the resource stress of the bottleneck nodes and the bottleneck edges;

[0031] According to the resource usage change trend and the resource tension, the first resource allocation strategy is dynamically adjusted through the graph neural network model to update the second resource allocation strategy.

[0032] In addition, to achieve the above-mentioned purpose, the present application also provides a metaverse resource scheduling device, the metaverse resource scheduling device comprising:

[0033] A first strategy module, configured to use a pre-trained graph neural network model to predict resource demand and formulate a first resource allocation strategy based on the generated resource demand prediction value;

[0034] The second strategy module is used to analyze the resource usage data using the graph neural network model, dynamically adjust the first resource allocation strategy according to the analysis results, and update the second resource allocation strategy.

[0035] In addition, to achieve the above-mentioned purpose, the present application also provides a terminal device, which includes a memory, a processor, and a metaverse resource scheduling program stored in the memory and executable on the processor, and when the metaverse resource scheduling program is executed by the processor, the steps of the metaverse resource scheduling method as described above are implemented.

[0036] In addition, to achieve the above-mentioned purpose, the present application also provides a computer-readable storage medium, on which a Metaverse resource scheduling program is stored, and when the Metaverse resource scheduling program is executed by a processor, the steps of the Metaverse resource scheduling method as described above are implemented.

[0037] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the metaverse resource scheduling method described above.

[0038] One or more technical solutions proposed in this application have at least the following technical effects:

[0039] This application uses a pre-trained graph neural network model to predict resource demand, and formulates a first resource allocation strategy based on the generated resource demand prediction value, uses the graph neural network model to analyze resource usage data, dynamically adjusts the first resource allocation strategy according to the analysis results, and updates the technical solution of the second resource allocation strategy. By utilizing the self-learning and adaptive capabilities of the graph neural network, analysis and adjustment are performed based on real-time monitored dynamic resource usage data. It can respond to changes in resource demand in a timely manner and dynamically adjust resource allocation strategies to ensure that the stable performance of the system can be maintained when resource demand suddenly increases or changes. This real-time dynamic scheduling capability greatly improves the response speed and flexibility of the system. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 This is a flow chart of the first exemplary embodiment of the method for scheduling metaverse resources of the present application;

[0041] Figure 2 This is a flow chart of the second exemplary embodiment of the method for scheduling metaverse resources of the present application;

[0042] Figure 3 This is a flow chart of the second exemplary embodiment of the method for scheduling metaverse resources of the present application;

[0043] Figure 4 A flowchart of the third exemplary embodiment of the method for scheduling metaverse resources in this application

[0044] Figure 5 This is a schematic diagram of the module structure of the metaverse resource scheduling device in an embodiment of the present application;

[0045] Figure 6 Schematic diagram of the device structure of the hardware operating environment involved in the metaverse resource scheduling method in the embodiment of the present application.

[0046] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0047] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0048] This application actually takes into account that traditional resource scheduling methods usually use rule-based or simple optimization algorithm strategies to manage the allocation of computing resources, storage resources, and network resources. These methods are often applicable to traditional computing environments, such as data centers or traditional network services, but when faced with complex virtual environments such as the Metaverse, there are significant applicability defects. In addition, traditional resource scheduling methods are often based on static rules or predefined optimization strategies. These optimization strategies may be based on polling, priority, or simple load balancing algorithms, but cannot adapt to the complex dynamic changes in the Metaverse.

[0049] Based on this, the first embodiment of the present application proposes a solution, which uses a pre-trained graph neural network model to predict resource demand, and formulates a first resource allocation strategy based on the generated resource demand prediction value; uses the graph neural network model to analyze resource usage data, dynamically adjusts the first resource allocation strategy based on the analysis results, and updates to obtain a second resource allocation strategy. Through this method, the self-learning and adaptive capabilities of the graph neural network are utilized to analyze and adjust the dynamic resource usage data monitored in real time, and dynamically adjust the resource allocation strategy to ensure that the system's stable performance can be maintained when resource demand suddenly increases or changes.

[0050] Specifically, the following are the detailed steps of the first exemplary embodiment of the Metaverse resource scheduling method:

[0051] Reference Figure 1 , Figure 1 This is a flow chart of the first exemplary embodiment of the Metaverse resource scheduling method of the present application. In this embodiment, the Metaverse resource scheduling method includes steps S10 to S20:

[0052] Step S10, using a pre-trained graph neural network model to predict resource demand, and formulating a first resource allocation strategy based on the generated resource demand prediction value;

[0053] Specifically, in this embodiment, historical resource usage data is first collected, including but not limited to CPU usage, memory usage, network bandwidth, storage I / O, etc., as well as the relationship between these resource usage and business load, time (such as weekdays and weekends, daytime and night), and then the historical resource usage data is used to build a data set, including cleaning and standardizing the data, processing missing values, outliers, etc.

[0054] Furthermore, a graph neural network model is constructed and trained, and the data set obtained based on the resource prediction demand is input into the trained graph neural network model. The graph neural network model is used to predict the resource demand in the future and output the prediction result, that is, the resource demand prediction value is obtained.

[0055] Based on the resource demand forecast, we can combine business needs and resource constraints, and also consider resource priority, scalability, cost-effectiveness and other factors to reasonably allocate resources to meet business needs while maintaining effective resource utilization. Based on this, we formulate a preliminary resource allocation plan, namely the first resource allocation strategy.

[0056] Step S20, using the graph neural network model to analyze the resource usage data, dynamically adjusting the first resource allocation strategy according to the analysis result, and updating the second resource allocation strategy;

[0057] Specifically, in this embodiment, resource usage data is first monitored and collected, including but not limited to the real-time consumption of computing resources, storage resources, network resources, etc., and the resource usage data is deeply analyzed using a graph neural network model to identify abnormal patterns, trend changes, and potential performance bottlenecks in resource usage. According to the analysis results, the deficiencies in the first resource allocation strategy are identified, such as uneven resource allocation, over-reservation, or insufficient resources. Based on this, the first resource allocation strategy is dynamically adjusted, such as adjusting the resource allocation ratio, priority, schedule, etc., to optimize resource utilization efficiency. The adjusted resource allocation plan is integrated into the second resource allocation strategy.

[0058] Furthermore, resource scheduling is performed according to the second resource allocation strategy, and resource usage and business performance are continuously monitored. Feedback data is collected for further optimization of the graph neural network model and resource allocation strategy. Based on the feedback data, the graph neural network model is updated and optimized to improve prediction accuracy. Based on the continuously optimized graph neural network model, the resource allocation strategy is iteratively updated to ensure optimal utilization of resources.

[0059] The present application also considers that traditional resource scheduling methods are inefficient when processing large-scale graph structure data. The nodes and edges in the metaverse environment constitute a large and complex graph structure, and traditional methods may require a lot of computing time and resources to process these complex relationships and data volumes. For example, traditional load balancing algorithms may only be able to simply allocate resources based on the current load, and cannot use the potential information in the graph structure for more accurate resource optimization.

[0060] In addition, traditional methods have bottlenecks in achieving multi-dimensional resource scheduling. Resource scheduling in the Metaverse environment involves not only the allocation of computing resources, but also the coordinated management of storage resources and network resources. Traditional methods often focus on the scheduling of a single resource type and cannot effectively balance and optimize between multi-dimensional resources. When a certain resource type becomes a bottleneck, the overall system performance and user experience will be affected.

[0061] Based on this, the second embodiment of the present application proposes a solution: generate a graph data set by extracting feature vectors of nodes and edges in pre-constructed graph structure data; use the graph data set to train a pre-constructed graph neural network model; use the pre-trained graph neural network model to predict resource requirements for nodes and edges in the pre-constructed graph structure data, and obtain resource requirement prediction values ​​for the nodes and edges respectively; calculate the resource allocation amounts of the nodes and edges respectively according to the resource requirement prediction values ​​of the nodes and edges.

[0062] The second embodiment of the present application effectively learns and captures the complex relationships between nodes and edges in the metaverse through a pre-built and trained graph neural network model. Compared with traditional resource scheduling methods, graph neural networks have stronger processing capabilities and can identify important features and patterns in large and complex graph structure data, thereby achieving more accurate resource demand prediction and allocation.

[0063] Secondly, by comprehensively considering the scheduling of computing resources, storage resources, and network resources, the multi-dimensional analysis of the graph neural network model can coordinate and manage various resources. This avoids the bottleneck problem caused by the scheduling of a single resource in traditional methods, thereby improving the overall resource utilization efficiency of the system.

[0064] Specifically, the following are detailed steps of the second exemplary embodiment of the Metaverse resource scheduling method:

[0065] Please refer to Figure 2 , Figure 2 This is a flow chart of a second exemplary embodiment of the metaverse resource scheduling method of the present application. In this embodiment, step S10: using a pre-trained graph neural network model to predict resource demand, and formulating a first resource allocation strategy based on the generated resource demand prediction value includes steps S11 to S12:

[0066] Step S11, using a pre-trained graph neural network model to predict resource requirements for nodes and edges in pre-constructed graph structure data, and obtaining resource requirement prediction values ​​for nodes and edges respectively;

[0067] Step S12, calculating the resource allocation amounts of the nodes and edges respectively according to the predicted values ​​of the resource requirements of the nodes and edges.

[0068] In a feasible implementation manner, step S11 may include steps S01 to S02:

[0069] Step S01: Generate a graph data set by extracting feature vectors of nodes and edges in pre-constructed graph structure data;

[0070] Specifically, in this embodiment, graph structure data is pre-constructed, and the steps of constructing the graph structure data are as follows:

[0071] First, construct the node set V and edge set E in the metaverse environment.

[0072] Specifically, all key elements in the metaverse are identified and classified as nodes and edges. For example, the node set V may include user nodes, virtual item nodes, and server nodes. Correspondingly, the edge set E includes user interaction edges, item transaction edges, and server connection edges.

[0073] Further, define the node feature vector X v and edge eigenvector X e , where X v ={x1,x2,…,x n}, x i represents the feature of the i-th node, X e ={e1,e2,…,e m},e j Represents the feature of the jth edge. A feature vector is a numerical array used to describe specific attributes of a node or edge. For a node, a feature vector can include the node's activity level, user level, value of virtual items, server load capacity, etc. For example, a user node's feature vector can include its login frequency, number of social relationships, consumption records, etc. Similarly, an edge feature vector describes specific attributes of an edge, such as the frequency of user interactions, the amount of item transactions, the data transmission rate between servers, etc.

[0074] Furthermore, node type labels and edge type labels are constructed: node type labels may include user type labels, virtual item type labels, and server type labels, which are represented by T u 、T v and T s Indicates that edge type labels can include user interaction type labels, item transaction type labels, and server connection type labels, which are represented by R u , R t , R s Representation. Type labels are identifiers that classify nodes and edges, helping the model better understand the structure and meaning of the data.

[0075] Finally, the node set V and the edge set E are constructed into graph structure data G, and corresponding feature vectors and type labels are added to each node and edge. The formula of G is defined as follows:

[0076]

[0077] Through the definition of graph structure data G, the graph structure in the metaverse environment is constructed. The graph structure data is expressed as:

[0078] G(V,E,X v ,X e ,T,R);

[0079] T={T(v i )|v i ∈V}; R = {R(e ij )|e ij ∈E};

[0080] Among them, T(v i ) represents the type label of the i-th node, R(e ij ) represents the type label of the ijth edge.

[0081] For example, in the graph structure data G, a user node consists of its unique identifier, feature vector (such as login frequency, number of social relationships, etc.), and user type label (such as "advanced user"). Similarly, a user interaction edge may also contain its unique identifier, feature vector (such as interaction frequency, relationship strength, etc.) and user interaction type label (such as "friend relationship").

[0082] Based on the above steps, graph structure data can be obtained.

[0083] Based on the graph structure data obtained in the above steps, the feature vectors of nodes and edges are extracted to generate a graph dataset for training. The specific implementation is as follows:

[0084] First, construct the graph structure data G(V,E,X v ,X e ,T,R) to clean the data, remove redundant and invalid data, and then calculate the node feature vector X v and edge eigenvector X e Perform standardization to convert the eigenvalues ​​to a uniform scale. Then, extract the eigenvectors of nodes and edges. The node eigenvector X v Expressed as Edge eigenvector X e Expressed as Finally, a graph dataset for training is generated, which includes the preprocessed node set V, edge set E, node feature vector X vand edge eigenvector X e , the graph dataset is represented as: (V,E,X v ,X e ).

[0085] For example, suppose there is a user node A and a server node B in the metaverse environment, which are connected by a user interaction edge C. When extracting feature vectors, the feature vector of user node A is [user type, current high load, frequent interaction frequency], the feature vector of server node B is [server type, current low load, average interaction frequency], and the feature vector of user interaction edge C is [high bandwidth, low latency, user interaction edge type]. These feature vectors will be used to generate a graph dataset containing nodes A, B and edge C.

[0086] Step S02: Use the graph dataset to train the pre-built graph neural network model.

[0087] In a feasible implementation, step S02 may include steps S021 to S024:

[0088] Step S021, inputting the graph data set into a pre-built graph neural network model;

[0089] It should be noted that before implementing step S021, the graph neural network model can be constructed in the following ways:

[0090] Specifically, the graph neural network model includes an input layer, multiple improved graph convolutional layers and an output layer.

[0091] The input layer receives node feature vectors and edge feature vectors and performs preliminary processing to adapt to the calculation of the graph convolution layer.

[0092] The improved graph convolution layer is used to extract the complex relationship between nodes and edges. The following formula is the calculation formula for each improved graph convolution to obtain the representation of the node:

[0093]

[0094] Among them, H (l) represents the node representation of the lth layer, N(i) represents the set of neighbor nodes of node i, and W (l) represents the weight matrix of the lth layer, represents the self-loop weight matrix of the lth layer, c ij is a normalization constant, σ represents the activation function, P(i) represents the set of meta-paths associated with node i, is the meta-path weight matrix, g(e ik ) is the characteristic function associated with the meta-path, α ijRepresents the scheduling priority weight between node i and node j.

[0095] Scheduling priority weight α ij Calculated by the priority function specific to Metaverse resource scheduling, α ij The calculation formula is as follows:

[0096]

[0097] Among them, ψ is a priority function specific to Metaverse resource scheduling, and ψ is defined as:

[0098] ψ(h i ,h j ,r ij )=a T ReLU(W a [h i ∥h j ∥r ij ]);

[0099] Among them, h i and h j Represent the feature representation vectors of node i and node j respectively.

[0100] Among them, a is a learnable priority vector, W a is a learnable weight matrix, ∥ represents the concatenation operation of the vector, ReLU represents the activation function, and r ij is the feature vector related to the resource scheduling of the Metaverse calculated based on the following formula, r ij The calculation formula is as follows:

[0101] r ij =[load(i),load(j),bandwidth(e ij ),latency(e ij ),interaction(i,j)];

[0102] Where load(i) and load(j) represent the current load of node i and node j respectively, bandwidth(e ij ) represents edge e ij Bandwidth, latency(e ij ) represents edge e ij The delay of, interaction(i,j) represents the interaction frequency between node i and node j;

[0103] Furthermore, based on the above implementation, a dynamic adaptive learning mechanism is introduced, an adaptive learning algorithm is designed, and the structure and parameters of the graph neural network are dynamically adjusted to adapt to changing resource requirements. The adaptive learning mechanism can be expressed by the following formula:

[0104]

[0105] Where η is the learning rate, is the loss function, Y is the actual resource demand, and λ is the regularization parameter. By dynamically adjusting the learning rate and regularization parameter, the adaptive adjustment of the network structure and parameters is achieved;

[0106] Furthermore, the output layer generates resource demand prediction results, using the final representation H of nodes and edges (l) Resource demand prediction is performed, including the demand values ​​of computing resources, storage resources, and network resources. The output layer generates prediction results through a fully connected layer and a Softmax function.

[0107] In summary, after the above-mentioned graph neural network model is constructed, the following steps are implemented: the graph dataset is input into the pre-built graph neural network model.

[0108] Step S022, using a back propagation algorithm to calculate the gradient of the loss function with respect to the parameters of the graph neural network model;

[0109] Specifically, the gradient of the model parameters is calculated layer by layer through the back propagation algorithm Among them, L represents the loss function, W (l) represents the weight matrix of the lth layer.

[0110] The loss L function is defined as the resource demand prediction error, that is, the difference between the resource demand value predicted by the model and the actual resource demand value. The definition formula is as follows:

[0111]

[0112] in, represents the resource demand value of the i-th sample predicted by the model, y i represents the actual resource requirement value of the i-th sample, and N represents the total number of samples.

[0113] The basic principle of the above back-propagation algorithm is to calculate the gradient of the loss function with respect to the network parameters in order to update the parameters and minimize the loss function. Specifically, for each neuron in the output layer, the partial derivative of the loss function with respect to the output value of the neuron is calculated. By using the chain rule, the gradient of the loss function with respect to the activation value of the output layer is decomposed into the product of several factors. These factors are used to calculate the gradient of the loss function with respect to the weights and biases of the output layer. For each hidden layer, the gradient of the loss function with respect to the activation value of the layer is calculated layer by layer in the opposite order of the forward propagation, and the gradient of the loss function with respect to the weights and biases of the layer is calculated using the chain rule. In the process of back-propagation, the error is gradually back-propagated from the output layer to the input layer, and the gradient of the corresponding weights and biases is updated at each step.

[0114] Step S023, using a gradient descent algorithm to update the parameters of the graph neural network model according to the gradient;

[0115] The gradient descent algorithm is used to update the parameters of the graph neural network model. This step can be expressed by the following formula:

[0116]

[0117] This formula shows that by (l) Subtracting the product of the gradient and the learning rate η, we can get the new parameter value W (l+1) .

[0118] Step S024, minimizing the loss function through iterative training until the loss function converges to a preset threshold, thereby obtaining an optimized graph neural network model.

[0119] Specifically, based on the gradient obtained in the above steps, the parameters of the model are gradually adjusted according to a predetermined learning rate to minimize the loss function.

[0120] Finally, the loss function L is minimized through iterative training until the loss function L converges to the preset threshold. This means that the prediction error of the model has been reduced to an acceptable range and the model has been fully trained and optimized.

[0121] In summary, based on the above steps S01 to S02, the graph neural network model is constructed and trained. Based on this, step S11 is implemented: using the pre-trained graph neural network model to predict the resource demand of the nodes and edges in the pre-constructed graph structure data, and obtaining the resource demand prediction values ​​of the nodes and edges respectively;

[0122] Specifically, in this embodiment, the preprocessed graph data set is input into the trained graph neural network model, and the graph neural network model is used to predict the resource requirements of nodes and edges in the metaverse environment to obtain the computing resource requirement value of the node. Storage resource requirements and network resource requirements For each node v i ∈V, according to its node feature vector x i and the characteristics of neighboring nodes, and calculate the predicted value of their computing resource requirements The calculation formula is as follows:

[0123]

[0124] Among them, W c is the computing resource demand weight matrix, which represents the weight used for computing resource demand prediction, W x is the node feature weight matrix, which represents the weight of the node feature on the computing resource demand, W n is the neighbor node feature weight matrix, which represents the weight of the neighbor node features on the computing resource requirements, b c is the computing resource demand bias vector, which represents the bias of computing resource demand prediction, α ij is the scheduling priority weight, which represents the scheduling priority between node i and node j, σ is the activation function, which represents the nonlinear activation function, ReLU is the rectified linear unit, which represents the nonlinear function used for activation, and N(i) is the set of neighbor nodes of node i, which represents the set of nodes directly connected to node i.

[0125] Furthermore, we calculate each node v i ∈V storage resource demand prediction value and network resource demand forecast The calculation formula is as follows:

[0126]

[0127] Among them, W s is the storage resource demand weight matrix, which represents the weight used for storage resource demand prediction, b s is the storage resource demand bias vector, which represents the bias of storage resource demand prediction, g s,i is the storage resource feature vector, which represents the specific characteristics of storage resource requirements, W n is the network resource demand weight matrix, which represents the weight used for network resource demand prediction, b n is the network resource demand bias vector, which represents the bias of network resource demand prediction, g n,i is the network resource feature vector, which represents the specific characteristics of network resource requirements.

[0128] Furthermore, for each edge e ij ∈E, according to its edge feature vector e ij and the characteristics of the connected nodes, and calculate the predicted value of their computing resource requirements Forecast of storage resource demand and network resource demand forecast The calculation formula is as follows:

[0129]

[0130] W e Represents the edge feature vector e ij When calculating resource demand prediction, W e Used to map edge feature vectors to a new feature space. xi and W xj represents the feature vector x for nodes i and j i and x j When calculating resource demand prediction, W xi and W xj It is used to map the node feature vector to a new feature space and perform comprehensive calculations in combination with the edge features.

[0131] Among them, W ce is the computing resource demand weight matrix, which represents the weights used for computing resource demand prediction, b ce is the computing resource demand bias vector, which represents the bias of computing resource demand prediction, g ce,ij is the computing resource feature vector, which represents the specific characteristics of computing resource requirements, W se is the storage resource demand weight matrix, which represents the weight used for storage resource demand prediction, b se is the storage resource demand bias vector, which represents the bias of storage resource demand prediction, g se,ij is the storage resource feature vector, which represents the specific characteristics of storage resource requirements, W ne is the network resource demand weight matrix, which represents the weight used for network resource demand prediction, b ne is the network resource demand bias vector, which represents the bias of network resource demand prediction, g ne,ij is the network resource feature vector, which represents the specific characteristics of network resource requirements.

[0132] In summary, we get the resource demand prediction values ​​of nodes and edges Based on this, step S12 is implemented.

[0133] Step S12, calculating the resource allocation amounts of the nodes and edges respectively according to the predicted values ​​of the resource requirements of the nodes and edges.

[0134] Specifically, based on the resource demand prediction values ​​of nodes and edges Calculate the total resource requirements for each node and edge.

[0135] The total resource demand of a node is calculated as follows:

[0136]

[0137] The total resource demand of an edge is calculated as follows:

[0138]

[0139] Furthermore, the resource allocation vector R i and Respectively represent the resource allocation of nodes and edges, R i and The definition formula is as follows:

[0140] R i =[R c,i ,R s,i ,R n,i ];

[0141]

[0142] Among them, R c,i is the computing resource allocated to node i, R s,i is the storage resource allocated to node i, R n,i is the network resource allocated to node i, Assigned to edge e ij of computing resources, Assigned to edge e ij storage resources, Assigned to edge e ij network resources.

[0143] Specifically, the following formula is used to preliminarily allocate resources to nodes:

[0144]

[0145] Among them, R i Represents the total amount of resources allocated to node i.

[0146] Specifically, the following formula is used to preliminarily allocate resources to edges:

[0147]

[0148] in, represents the assignment to edge e ij total resources.

[0149] Please refer to Figure 3 , Figure 3 This is a flow chart of a second exemplary embodiment of the metaverse resource scheduling method of the present application. In this embodiment, step S20: using the graph neural network model to analyze the resource usage data, dynamically adjusting the first resource allocation strategy according to the analysis results, and updating the second resource allocation strategy includes steps S21 to S24:

[0150] Step S21, monitoring the resource usage data of nodes and edges through a pre-deployed resource monitoring module, and dynamically analyzing the resource usage data to calculate the resource usage rate;

[0151] Step S22, using the graph neural network model to analyze the input resource usage data and calculate the resource usage change trend of nodes and edges;

[0152] Step S23, using the graph neural network model to identify bottleneck nodes and bottleneck edges, and calculating the resource tension of the bottleneck nodes and the bottleneck edges based on the resource utilization rate;

[0153] Step S24: According to the resource usage change trend and the resource tension, the first resource allocation strategy is dynamically adjusted through the graph neural network model to update the second resource allocation strategy.

[0154] It should be noted that before implementing step S21, a resource monitoring module can be deployed in the metaverse environment to monitor the usage of computing resources, storage resources and network resources in real time, and then collect resource usage data.

[0155] Specifically, we deploy a resource monitoring module to monitor the computing resource usage of each node and edge in real time, collect data such as CPU usage, memory usage, and computing task load of each node and edge, and generate a computing resource usage dataset D c ;

[0156] The resource monitoring module also monitors the storage resource usage of each node and edge in real time, collects data such as storage space usage, data read and write rate, and storage task load of each node and edge, and generates a storage resource usage data set D s ;

[0157] The resource monitoring module also monitors the network resource usage of each node and edge in real time, collects data such as bandwidth usage, network delay, and network traffic of each node and edge, and generates a network resource usage dataset D n .

[0158] Furthermore, the resource usage data of nodes and edges are monitored by a pre-deployed resource monitoring module, and the resource usage data is dynamically analyzed to calculate the resource usage rate.

[0159] Specifically, the resource monitoring module is used to dynamically analyze the collected usage data of computing resources, storage resources, and network resources, and calculate the real-time resource utilization rate U c,i , U s,i , U n,i , in:

[0160] U c,i : Calculate resource utilization, indicating the CPU utilization of node i at a certain moment. The formula is:

[0161]

[0162] Among them, CPU usage is the CPU resource currently used by node i, and total CPU capacity is the maximum CPU resource that node i can provide.

[0163] U s,i : Storage resource utilization rate, which indicates the storage space usage of node i at a certain moment. The formula is:

[0164]

[0165] The storage space usage is the storage resource currently used by node i, and the total storage space is the maximum storage resource that node i can provide.

[0166] U n,i : Network resource utilization rate, which indicates the bandwidth utilization rate of node i at a certain moment. The formula is:

[0167]

[0168] The bandwidth utilization rate is the network bandwidth resource currently used by node i, and the total bandwidth capacity is the maximum network bandwidth resource that node i can provide.

[0169] Compute resource utilization, representing edge e ij The CPU usage at a certain moment. The formula is:

[0170]

[0171] Among them, CPU usage is the edge ij The CPU resources currently used, the total CPU capacity is edge e ij The maximum CPU resources that can be provided.

[0172] Storage resource utilization, representing edge e ij The amount of storage space used at a certain moment. The formula is:

[0173]

[0174] Among them, the storage space usage is the edge e ij The total storage space currently used is e ij The maximum storage resources that can be provided.

[0175] Network resource utilization, indicating edge e ij The bandwidth usage at a certain moment. The formula is:

[0176]

[0177] The bandwidth utilization rate is ij The network bandwidth resources currently used, the total bandwidth capacity is edge e ij The maximum network bandwidth resources that can be provided.

[0178] Furthermore, the graph neural network model is used to analyze the input resource usage data to calculate the resource usage change trends of nodes and edges.

[0179] Specifically, the dynamic resource usage data D monitored in real time c , D s , D n The data is input into the graph neural network model, which analyzes the input data and calculates the resource usage change trend ΔU of each node and edge. c,i , ΔU s,i , ΔU n,i , The formula for calculating the change trend is as follows:

[0180] ΔU c,i =U c,i (t)-U c,i (t-1);

[0181] ΔU s,i =U s,i (t)-U s,i (t-1);

[0182] ΔU n,i =U n,i (t)-U n,i (t-1);

[0183]

[0184] Among them, U c,i (t) represents the computing resource usage of node i at time point t, U s,i represents the storage resource usage of node i at time point t, Un,i represents the network resource utilization rate of node i at time point t, Represents edge e ij The usage of computing resources, storage resources, and network resources at time point t.

[0185] Further, the graph neural network model is used to identify bottleneck nodes and bottleneck edges in resource usage, and based on the resource usage rate, the resource tension of the bottleneck nodes and the bottleneck edges is calculated;

[0186] Specifically, the graph neural network model is used to identify potential bottlenecks in resource usage and calculate the resource intensity T of bottleneck nodes and bottleneck edges. i and The tension calculation formula is as follows:

[0187]

[0188] Furthermore, according to the resource usage change trend and the resource tension, the first resource allocation strategy is dynamically adjusted through the graph neural network model to update the second resource allocation strategy.

[0189] Specifically, according to the resource tension of each node and edge calculated in the previous step (including computing resource tension, storage resource tension, and network resource tension), as well as the corresponding resource usage change trend, the graph neural network model can use the optimization technology in the deep learning algorithm, such as gradient descent or its variants, to iteratively update the resource allocation vectors of nodes and edges. These resource allocation vectors represent the allocation of each node and edge in different resource dimensions. The update process aims to maximize the overall resource utilization efficiency while ensuring that the resource requirements of key tasks or bottleneck nodes are met. Finally, the graph neural network model outputs the updated resource allocation vector, and updates the resource allocation vector R of the node and edge. i and

[0190] The present application further takes into account that in traditional resource scheduling methods, due to failure to fully consider the differences in node importance, changes in resource utilization and the real-time nature of user needs, the resource needs of critical tasks and important users cannot be met in a timely manner, resulting in waste of resources and degradation of overall system performance.

[0191] Based on this, the third embodiment of the present application proposes a solution: use a pre-trained graph neural network model to predict resource demand, and formulate a first resource allocation strategy based on the generated resource demand prediction value, use the graph neural network model to analyze resource usage data, dynamically adjust the first resource allocation strategy according to the analysis results, and update the second resource allocation strategy. According to the pre-calculated priority index, the second resource allocation strategy is optimized through the graph neural network model to generate a resource allocation plan.

[0192] The embodiment of the present application comprehensively considers factors such as the importance of nodes and edges, resource utilization, and user needs, and generates an intelligent resource scheduling solution, so as to identify high-priority nodes and edges, give priority to meeting their resource needs, and ensure important tasks and user experience. At the same time, through intelligent optimization, resource waste is reduced and resource utilization is improved.

[0193] Specifically, the following are detailed steps of the third exemplary embodiment of the Metaverse resource scheduling method:

[0194] Reference Figure 3 , Figure 3 This is a flow chart of the third exemplary embodiment of the Metaverse resource scheduling method of the present application. In this embodiment, the Metaverse resource scheduling method includes steps S10 to S40:

[0195] Step S10, using a pre-trained graph neural network model to predict resource demand, and formulating a first resource allocation strategy based on the generated resource demand prediction value;

[0196] Step S20, using the graph neural network model to analyze the resource usage data, dynamically adjusting the first resource allocation strategy according to the analysis result, and updating the second resource allocation strategy;

[0197] Step S30, optimizing the second resource allocation strategy through the graph neural network model according to the pre-calculated priority index to generate a resource allocation plan;

[0198] Step S40: executing resource scheduling operations according to the resource allocation scheme, and dynamically adjusting resource configuration.

[0199] It should be noted that, in this embodiment, the specific implementation of steps S10 to S20 can be referred to the above description and will not be described in detail later.

[0200] In a feasible implementation, step S30 may include steps S31 to S33:

[0201] Step S31, calculating the importance parameters and resource utilization of nodes and edges in the Metaverse environment;

[0202] Specifically, calculate the importance of each node I i and the importance of each edge The importance of a node is determined by its role and connectivity, and the importance of an edge is determined by the importance and interaction frequency of the nodes it connects:

[0203]

[0204] Among them, w j represents the importance weight of neighbor nodes, ∥N(i)∥ represents the number of neighbor nodes of node i;

[0205]

[0206] Among them, f(e ij ) represents edge e ij Frequency of interaction;

[0207] Calculate the resource utilization U of each node and edge i and Resource utilization includes computing resource utilization, storage resource utilization, and network resource utilization:

[0208]

[0209] Step S32, calculating the priority indexes of nodes and edges respectively according to the user demand parameter, the importance parameter and the resource utilization rate;

[0210] Specifically, the priority P of each node and edge is calculated according to user requirements. i and The priority is determined based on the urgency and importance of the user's needs, and the calculation formula is as follows:

[0211] P i =d i I i ·U i ;

[0212]

[0213] Among them, d i and Represent the user demand weights of nodes and edges respectively.

[0214] It should be noted that user demand values ​​can be obtained through market research, user interviews, questionnaires, and data analysis. These methods aim to gain an in-depth understanding of users' actual needs, preferences, and the importance they place on different functions and features. Specifically, a detailed questionnaire can be designed to assess the urgency and importance of different aspects of a product or service, such as functional requirements, performance requirements, price sensitivity, and user experience. By collecting and analyzing these feedback data, the value of user demand can be quantified, which can then be used to calculate the priority and sum of each node and edge.

[0215] During the calculation process, each user requirement is assigned a weight that reflects the importance of the requirement to the user. Urgency can be evaluated by the time required to meet the requirement expressed by the user. Combining these weights with the urgency factor, the priority and of each node and edge can be calculated according to the formula.

[0216] Step S33, optimizing the second resource allocation strategy according to the priority indicators of nodes and edges through the graph neural network model to generate a resource allocation plan.

[0217] Specifically, the priorities of nodes and edges are input into the graph neural network model for resource scheduling optimization to generate the final resource allocation plan. The final resource allocation plan includes the reallocation of computing resources, storage resources, and network resources to ensure optimal resource utilization and maximum satisfaction of user needs.

[0218] Based on this, step S40 is executed: performing resource scheduling operations according to the resource allocation scheme, and dynamically adjusting resource configuration.

[0219] Specifically, the system performs resource scheduling operations according to the final resource allocation plan, dynamically adjusts the resource allocation of each node and edge, allocates computing resources, storage resources, and network resources, and ensures that the resource requirements of each node are met while ensuring the stability and efficiency of system performance.

[0220] In addition, the present application also proposes a metaverse resource scheduling device, the metaverse resource scheduling device comprising:

[0221] The first strategy module 10 uses a pre-trained graph neural network model to predict resource demand and formulates a first resource allocation strategy based on the generated resource demand prediction value;

[0222] The second strategy module 20 uses the graph neural network model to analyze the resource usage data, dynamically adjusts the first resource allocation strategy according to the analysis results, and updates the second resource allocation strategy.

[0223] The Metaverse Resource Scheduling Device provided by the present application adopts the Metaverse Resource Scheduling Method in the above-mentioned embodiment, aiming to solve the technical problem of untimely dynamic resource scheduling in the traditional resource scheduling method, and realize efficient resource management and system performance optimization in the Metaverse environment. Compared with the prior art, the beneficial effects of the Metaverse Resource Scheduling Device provided by the present application are the same as the beneficial effects of the Metaverse Resource Scheduling Method provided by the above-mentioned embodiment, and the other technical features in the Metaverse Resource Scheduling Device are the same as the features disclosed in the above-mentioned embodiment method, which will not be repeated here.

[0224] The present application provides a metaverse resource scheduling device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the metaverse resource scheduling method in the above-mentioned embodiment one.

[0225] Reference below Figure 6 , which shows a schematic diagram of the structure of the Metaverse resource scheduling device used to implement the embodiment of the present application. The Metaverse resource scheduling device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The metaverse resource scheduling device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0226] like Figure 6As shown, the Metaverse Resource Scheduling Device may include a processing device 1001 (e.g., a central processing unit, a graphics processor, etc.), which may perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM 1004, various programs and data required for the operation of the Metaverse Resource Scheduling Device are also stored. The processing device 1001, ROM 1002, and RAM 1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems may be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touchpad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the Metaverse Resource Scheduling Device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a Metaverse Resource Scheduling Device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0227] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0228] The Metaverse Resource Scheduling Device provided by this application adopts the Metaverse Resource Scheduling Method in the above-mentioned embodiment, aiming to solve the technical problem of untimely dynamic resource scheduling in the traditional resource scheduling method, and realize efficient resource management and system performance optimization in the Metaverse environment. Compared with the prior art, the beneficial effects of the Metaverse Resource Scheduling Device provided by this application are the same as the beneficial effects of the Metaverse Resource Scheduling Method provided by the above-mentioned embodiment, and the other technical features in the Metaverse Resource Scheduling Device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0229] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0230] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0231] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the metaverse resource scheduling method in the above-mentioned embodiment.

[0232] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM: Random Access Memory), a read-only memory (ROM: Read Only Memory), an erasable programmable read-only memory (EPROM: Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM: CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency: Radio Frequency), etc., or any suitable combination of the above.

[0233] The above-mentioned computer-readable storage medium may be included in the Metaverse resource scheduling device; or it may exist independently without being assembled into the Metaverse resource scheduling device.

[0234] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0235] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0236] The modules involved in the embodiments described in the present application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0237] The readable storage medium provided in this application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned metaverse resource scheduling method, aiming to solve the technical problem of untimely dynamic resource scheduling in traditional resource scheduling methods, and realize efficient resource management and system performance optimization in the metaverse environment. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as the beneficial effects of the metaverse resource scheduling method provided in the above-mentioned embodiment, which will not be repeated here.

[0238] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned metaverse resource scheduling method when executed by a processor.

[0239] The computer program product provided in this application is intended to solve the technical problem of untimely dynamic resource scheduling in traditional resource scheduling methods, and to achieve efficient resource management and system performance optimization in the metaverse environment. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the metaverse resource scheduling method provided in the above embodiment, and will not be elaborated here.

[0240] Compared with the prior art, the metaverse resource scheduling method, device, equipment, medium and computer product proposed in the embodiment of the present application extracts the business feature information of the target business, performs data standardization processing on the business feature information to obtain standard feature data, performs hash processing on the standard feature data to obtain unique feature data, performs numerical processing and splicing processing on the unique feature data, obtains the first business feature value, accumulates the first business feature value of the target business, obtains the target business feature value, and finally compares the target business feature value with the feature value set to obtain the metaverse resource scheduling result. Compared with the traditional method of identifying duplicate businesses by generating a unique key value or a continuous serial number for each business, it is more efficient, flexible and reliable. Based on the solution of the present application, the business of complex scenarios is transformed into a comparison of two numbers through a series of simple transformations, so that the comparison process is very intuitive and efficient. The system only needs to simply compare whether the two values ​​are equal to quickly determine whether the two businesses are exactly the same.

[0241] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0242] The serial numbers of the above-mentioned embodiments of the present application are for description only and do not represent the advantages or disadvantages of the embodiments.

[0243] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as above, and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, a controlled terminal, or a network device, etc.) to execute the method of each embodiment of the present application.

[0244] The above are only preferred embodiments of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for scheduling metaverse resources, characterized in that: The metaverse resource scheduling method comprises: Using a pre-trained graph neural network model to predict resource demand, and formulating a first resource allocation strategy based on the generated resource demand prediction value; The graph neural network model is used to analyze the resource usage data, and the first resource allocation strategy is dynamically adjusted according to the analysis results to update the second resource allocation strategy.

2. The method for scheduling metaverse resources according to claim 1, characterized in that: After the step of analyzing the resource usage data by using the graph neural network model and dynamically adjusting the first resource allocation strategy according to the analysis result to update the second resource allocation strategy, the following step further includes: According to the pre-calculated priority index, the second resource allocation strategy is optimized through the graph neural network model to generate a resource allocation plan.

3. The method for scheduling metaverse resources according to claim 2, characterized in that: After the step of optimizing the second resource allocation strategy by the graph neural network model according to the pre-calculated priority index and generating a resource allocation solution, the step further includes: The resource scheduling operation is performed according to the resource allocation scheme, and the resource configuration is dynamically adjusted.

4. The method for scheduling metaverse resources according to claim 2, characterized in that: The step of optimizing the second resource allocation strategy through the graph neural network model according to the pre-calculated priority index to generate a resource allocation solution includes: Calculate the importance parameters and resource utilization of nodes and edges in the Metaverse environment; Calculating the priority indexes of nodes and edges respectively according to the user demand parameter, the importance parameter and the resource utilization rate; The graph neural network model is used to optimize the second resource allocation strategy according to the priority indicators of nodes and edges to generate a resource allocation plan.

5. The method for scheduling metaverse resources according to claim 1, characterized in that: The steps of using a pre-trained graph neural network model to predict resource demand and formulating a first resource allocation strategy based on the generated resource demand prediction value include: Use the pre-trained graph neural network model to predict the resource demand of nodes and edges in the pre-built graph structure data, and obtain the resource demand prediction values ​​of nodes and edges respectively; According to the predicted values ​​of resource requirements of nodes and edges, the resource allocation of nodes and edges is calculated respectively.

6. The method for scheduling metaverse resources according to claim 5, characterized in that: The step of using the pre-trained graph neural network model to predict resource requirements for nodes and edges in the pre-constructed graph structure data to obtain resource requirement prediction values ​​for the nodes and edges respectively further includes: Generate a graph dataset by extracting feature vectors of nodes and edges from pre-constructed graph structure data; The pre-built graph neural network model is trained using the graph dataset.

7. The method for scheduling metaverse resources according to claim 6, characterized in that: The step of using the graph dataset to train the pre-built graph neural network model includes: Inputting the graph dataset into a pre-built graph neural network model; Using a back-propagation algorithm, the gradient of the loss function with respect to the parameters of the graph neural network model is calculated; Using a gradient descent algorithm, updating the parameters of the graph neural network model according to the gradient; The loss function is minimized through iterative training until the loss function converges to a preset threshold, thereby obtaining an optimized graph neural network model.

8. The method for scheduling metaverse resources according to claim 1, characterized in that: The step of analyzing the resource usage data by using the graph neural network model, dynamically adjusting the first resource allocation strategy according to the analysis result, and updating the second resource allocation strategy includes: Monitor the resource usage data of nodes and edges through the pre-deployed resource monitoring module, and dynamically analyze the resource usage data to calculate the resource usage rate; Analyze the input resource usage data using the graph neural network model to calculate the resource usage change trend of nodes and edges; Using the graph neural network model to identify bottleneck nodes and bottleneck edges, and based on the resource utilization rate, calculating the resource stress of the bottleneck nodes and the bottleneck edges; According to the resource usage change trend and the resource tension, the first resource allocation strategy is dynamically adjusted through the graph neural network model to update the second resource allocation strategy.

9. A metaverse resource scheduling device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the metaverse resource scheduling method according to any one of claims 1 to 8.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the metaverse resource scheduling method as described in any one of claims 1 to 8 are implemented.

11. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the metaverse resource scheduling method according to any one of claims 1 to 8 are implemented.

Images