Digital economy resource scheduling and management method based on cloud-edge collaboration

Abstract

The present invention relates to the field of resource management technology, and in particular to a cloud-edge collaborative digital economic resource scheduling and management method. The method comprises the following steps: acquiring edge computing data, and constructing an edge relationship network based on the edge computing data to obtain edge computing network data; performing edge center extraction and edge sub-center extraction based on the edge computing network data to obtain edge center data and edge sub-center data; performing data update voting based on the edge center data and edge sub-center data to obtain edge data update voting data; and sending the edge data update voting data to the cloud to perform digital economic resource scheduling operations. Based on the collaborative relationship and task dependency between edge computing nodes, the present invention can provide efficient and reliable decision support in a complex network environment, and significantly improve the overall performance and resource utilization efficiency of the system.

Description

Technical Field

[0001] The present invention relates to the field of resource management technology, and in particular to a cloud-edge collaborative digital economy resource scheduling and management method. Background Art

[0002] In the context of the rapid development of the current digital economy, the combination of edge computing and cloud computing has become an important means to improve computing efficiency and resource utilization. With the popularization of IoT devices and the explosive growth of data volume, relying solely on cloud computing for data processing and resource management has been unable to meet the needs of real-time and efficiency. Edge computing can significantly reduce latency and improve the real-time performance of data processing by deploying computing and storage resources at the edge of the network close to the data source. However, edge computing nodes usually have limited computing power and uneven resource distribution. Therefore, they need to work together with cloud computing to make full use of the powerful computing and storage capabilities of the cloud to achieve efficient scheduling and management of resources.

[0003] In the process of edge computing and cloud computing collaboration, resource scheduling and management are one of the key links. Traditional resource scheduling methods are mainly based on static preset rules or simple algorithms, which are difficult to adapt to complex and changing network environments, resulting in low resource utilization and low task execution efficiency. In addition, existing resource scheduling methods often lack in-depth analysis of the functional coupling relationship between nodes, and fail to fully consider the collaborative relationship and task dependency between nodes, resulting in unsatisfactory results in task scheduling and resource allocation. Summary of the invention

[0004] In order to solve the above-mentioned technical problems, the present invention proposes a cloud-edge collaborative digital economic resource scheduling and management method to solve at least one of the above-mentioned technical problems.

[0005] The present application provides a cloud-edge collaborative digital economy resource scheduling and management method, the method comprising:

[0006] S1. Obtain edge computing data, and construct an edge relationship network based on the edge computing data to obtain edge computing network data;

[0007] S2. Extract edge centers and edge sub-centers according to edge computing network data to obtain edge center data and edge sub-center data;

[0008] S3, perform data update voting according to the edge center data and the edge sub-center data to obtain edge data update voting data;

[0009] S4, send the edge data updated voting data to the cloud to perform digital economic resource scheduling operations.

[0010] The edge relationship network constructed in the present invention can reflect the functional relationship and data flow between each node, which helps to optimize the network topology and improve network efficiency. By extracting edge centers and sub-centers, key nodes and important paths in the network can be identified, which helps to concentrate resources for optimization and management. Through the voting mechanism, consensus can be reached among multiple nodes to ensure the consistency and accuracy of the data. Combining the data and opinions of multiple nodes, the voting results are more objective and scientific, reducing the deviation caused by single-point decision-making. Cloud scheduling can dynamically adjust resource allocation based on real-time updates of edge data, improve resource utilization and service quality.

[0011] Optionally, the present application further provides a cloud-edge collaborative digital economic resource scheduling and management system, which is used to execute the cloud-edge collaborative digital economic resource scheduling and management method as described above, and the cloud-edge collaborative digital economic resource scheduling and management system includes:

[0012] The edge relationship network construction module is used to obtain edge computing data and construct an edge relationship network based on the edge computing data to obtain edge computing network data;

[0013] The edge center and secondary center extraction module is used to extract the edge center and the edge secondary center according to the edge computing network data to obtain the edge center data and the edge secondary center data;

[0014] The edge data update voting module is used to perform data update voting based on the edge center data and the edge sub-center data to obtain edge data update voting data;

[0015] The digital economy resource scheduling module is used to send edge data update voting data to the cloud to perform digital economy resource scheduling operations.

[0016] The purpose of the present invention is that through high-frequency data updates, the system can reflect the status and changes of edge computing nodes in real time, thereby improving the response speed and accuracy of the system. Rapidly build an edge relationship network to ensure that the system can quickly adapt to changes in the network environment and improve the timeliness of resource scheduling. In a test environment with 100 edge nodes, the construction time of the edge relationship network is less than 5 seconds, ensuring the rapid construction and response of the network topology. High-accuracy edge center and sub-center extraction ensures the accurate identification of key nodes and provides a reliable data basis for resource optimization. The extraction process is fast and efficient, ensuring that the system can quickly adapt to network changes and improving the response speed of resource scheduling. In the test environment, the main traffic center and sub-center nodes are accurately identified through edge computing network data with an accuracy of more than 95%. By combining comprehensive voting decisions in different dimensions, the scientificity and reliability of data updates are ensured, high-frequency data interactions are reduced, the interaction frequency of reliable data is increased, and the overall performance of the system is improved. The average response time for resource scheduling in the cloud is 5 seconds, and resource utilization is increased by 15%. Efficient resource scheduling improves resource utilization and overall system performance, ensuring the efficient operation of digital economic applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings:

[0018] Figure 1 A flowchart showing a method for scheduling and managing digital economic resources in a cloud-edge collaborative manner according to an embodiment of the present invention is shown;

[0019] Figure 2 A flowchart showing a method for constructing an edge relationship network according to an embodiment of the present invention is shown;

[0020] Figure 3 A flowchart showing a method for constructing an edge computing network topology according to an embodiment is shown;

[0021] Figure 4 A flowchart showing a method for extracting edge centers and edge sub-centers according to an embodiment of the present invention is shown;

[0022] Figure 5 A flowchart of the steps of an edge data update voting method according to an embodiment is shown;

[0023] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0024] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0025] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different network and / or processor methods and / or microcontroller methods.

[0026] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0027] See also Figures 1 to 5 , the present application provides a cloud-edge collaborative digital economy resource scheduling and management method, the method comprising:

[0028] S1. Obtain edge computing data, and construct an edge relationship network based on the edge computing data to obtain edge computing network data;

[0029] Specifically, data is obtained from edge nodes, including computing load, network traffic, node status and other information. An edge relationship network is constructed based on the acquired data to form edge computing network data. The nodes and edges in the network represent computing resources and data transmission paths. Number of edge nodes: 100, data collection frequency: 10 times per second, edge relationship network construction time: 5 seconds.

[0030] S2. Extract edge centers and edge sub-centers according to edge computing network data to obtain edge center data and edge sub-center data;

[0031] Specifically, based on the network topology and traffic data, the nodes with the highest traffic are identified as edge centers. According to the importance of the nodes, the secondary center nodes are extracted. There are 2 edge center nodes (node2, node3) and 2 edge secondary center nodes (node1, node2).

[0032] S3, perform data update voting according to the edge center data and the edge sub-center data to obtain edge data update voting data;

[0033] Specifically, each node votes to select a data update plan, and the plan with the most votes is selected. Weighted voting is performed based on the importance of the node in the network. Voting is performed based on the confidence of the node to obtain the final update result.

[0034] S4, send the edge data updated voting data to the cloud to perform digital economic resource scheduling operations.

[0035] Specifically, the voting results are sent to the cloud data center. The cloud performs resource scheduling and optimizes resource allocation based on the voting results. Data transmission delay: 50 milliseconds, resource scheduling response time: 5 seconds, and resource utilization rate improvement: 15%.

[0036] Specifically, node A: load 0.6, connection [B, C]; node B: load 0.7, connection [A, D]; node C: load 0.5, connection [A, E]; node D: load 0.4, connection [B]; node E: load 0.3, connection [C]; edge center data: edge center such as [A, B, C]; edge sub-center such as [D, E]; edge data update voting data such as voting results: {'A': 2.1, 'B': 1.6, 'C': 1.4, 'D': 0.4, 'E': 0.3}, the importance score of each node represents the comprehensive influence and confidence of the node in the entire edge computing network, which is obtained by integrating the voting results of each node, reflecting the role and importance of the node in the network; cloud scheduling operations are performed according to edge nodes A, B, and C. Cloud scheduling operations such as increasing the resource quota of node A, reducing the load of node B, and optimizing the connection between node C and node D.

[0037] The edge relationship network constructed in the present invention can reflect the functional relationship and data flow between each node, which helps to optimize the network topology and improve network efficiency. By extracting edge centers and sub-centers, key nodes and important paths in the network can be identified, which helps to concentrate resources for optimization and management. Through the voting mechanism, consensus can be reached among multiple nodes to ensure the consistency and accuracy of the data. Combining the data and opinions of multiple nodes, the voting results are more objective and scientific, reducing the deviation caused by single-point decision-making. Cloud scheduling can dynamically adjust resource allocation based on real-time updates of edge data, improve resource utilization and service quality.

[0038] Optionally, S1 includes:

[0039] S11: Obtain edge computing data;

[0040] Specifically, data is collected from each edge node, including computing load, network traffic, node status and other information.

[0041] S12: construct a network topology according to the edge computing data to obtain edge computing network topology data;

[0042] Specifically, based on the collected edge computing data, a network topology is constructed to represent each node and its connection relationship.

[0043] S13: extracting the functional coupling relationship according to the edge computing data to obtain edge computing functional coupling relationship data;

[0044] Specifically, the functional coupling relationship is extracted from the edge computing data to represent the functional dependency and collaborative relationship between the nodes. The computing task of node A depends on the output of node B, indicating that there is a computing task dependency between node A and node B. The computing task dependency can be obtained by analyzing the task flow graph (TFG). The data transmission and information interaction between nodes are frequent, indicating that there is a communication traffic dependency between them. The communication traffic dependency can be obtained by analyzing the communication traffic matrix (TM). The nodes work together frequently, indicating that there is a collaborative strength relationship between them. The collaborative strength can be obtained by analyzing the collaborative work records (such as collaborative computing tasks, data sharing frequency, etc.).

[0045] S14: constructing a local coupling relationship network according to the edge computing function coupling relationship data and the edge computing data to obtain edge local coupling relationship network data;

[0046] Specifically, the functional coupling relationship and edge computing data are used to construct a local coupling relationship network to represent the nodes and relationships in the local area. Determine the local area in the edge computing network and divide the area based on factors such as functional zoning. Extract the functional coupling relationship between nodes in the local area from the edge computing functional coupling relationship data. According to the extracted local functional coupling relationship, construct a local coupling relationship network to represent the nodes and relationships in the local area. Node: represents the computing node in the local area. Edge: represents the functional coupling relationship between nodes, and the weight of the edge reflects the functional coupling strength. The local coupling relationship network is represented by the graph data structure, and the relationship between nodes and edges is represented by the nodes and edges of the graph. According to the functional coupling relationship data, the weight of each node and edge is calculated. The node weight is comprehensively determined based on factors such as the node's computing load and the number of connections. The edge weight is determined by taking the strength value in the functional coupling relationship data. Obtain the edge local coupling relationship network data.

[0047] S15: Construct a global coupling relationship network according to the edge computing network topology data and the edge local coupling relationship network data to obtain edge computing network data.

[0048] Specifically, the edge computing network topology data and the local coupling relationship network data are integrated to construct a global coupling relationship network, which represents the nodes and relationships of the entire edge computing network.

[0049] Create an empty global coupling network graph structure. Extract node information from edge computing network topology data and local coupling network data, and add it to the global coupling network. Extract physical connection relationships from edge computing network topology data, and add functional coupling relationships based on local coupling network data. Calculate the weight of each edge based on physical connection and functional coupling relationships, such as weighted average. Organize the global coupling network data into a format that is easy to use and analyze.

[0050] The present invention ensures the real-time and accuracy of data, and provides basic data for the subsequent network and functional relationship construction. The calculation data of each edge node is fully collected to ensure the comprehensiveness and representativeness of the data. By constructing the network topology, the connection relationship and data flow path between the nodes can be clarified, which helps to optimize the network structure and improve network efficiency. After constructing the topology map, the key paths and bottlenecks in the network can be identified to provide a basis for subsequent optimization. By extracting the functional coupling relationship, the functional dependency between the nodes can be deeply understood, which helps to optimize task allocation and resource scheduling. By analyzing the functional coupling relationship, task coordination can be optimized, unnecessary data transmission and calculation can be reduced, and overall efficiency can be improved. By constructing a local coupling relationship network, the local network can be finely managed and optimized, and the performance and reliability of the local network can be improved. Through local optimization, the changes and needs of the local network can be quickly responded to, and the flexibility and adaptability of the network can be improved. Through the construction of the global coupling relationship network, the entire network can be optimized and managed from a global perspective, and the performance and stability of the overall network can be improved. Global optimization can effectively allocate and schedule network resources, avoid resource waste and overload, and improve resource utilization.

[0051] Optionally, the network topology construction includes:

[0052] S121: Register the node according to the edge computing data to obtain edge node registration data;

[0053] Specifically, basic information such as node ID, computing power, storage capacity, etc. is collected from each edge node to perform node registration.

[0054] S122: Initialize the node according to the edge node registration data to obtain edge node initialization data;

[0055] Specifically, based on the registration data, each node is initialized and configured, including initializing computing load, network connection status, etc.

[0056] S123: constructing an initial topology according to the edge node initialization data to obtain edge node topology data;

[0057] Specifically, based on the initialization data, a network topology is constructed to represent each node and its connection relationship.

[0058] S124: collecting edge node topology data in real time to obtain edge node topology real-time data;

[0059] Specifically, the status data of each node is monitored and collected in real time, including load, traffic, connection status, etc.

[0060] S125: Generate adjustment data according to the real-time data of the edge node topology to obtain edge node adjustment data;

[0061] Specifically, features are extracted from real-time data and the adjustment strategy is generated using the Deep Q-Network (DQN) algorithm.

[0062] S126: updating edge node topology data in real time according to edge node adjustment data to obtain edge computing network topology data;

[0063] Specifically, according to the adjustment strategy, the network topology is updated in real time to optimize the node load and connection status.

[0064] The adjustment data generation is specifically as follows:

[0065] Extract features based on the real-time data of edge node topology to obtain edge node topology feature data;

[0066] Specifically, features are extracted from real-time data of edge node topology, including node degree, inter-node distance, node load, and traffic intensity.

[0067] Data is generated based on the edge node topology feature data through the deep Q network algorithm to obtain edge node adjustment data.

[0068] Specifically, the extracted feature data is represented as the state input of DQN. The action space of DQN is defined, including actions such as load balancing, traffic adjustment, and connection optimization. A reward function is defined to guide DQN strategy generation. The reward function is based on load balancing, communication delay, and traffic optimization effects. The DQN model is trained using feature data, and the optimal strategy is selected through Q value update. The optimal strategy generated by DQN is executed to adjust the edge node topology and obtain edge node adjustment data.

[0069] The present invention optimizes resource allocation and maximizes the use of network resources through dynamic monitoring and intelligent adjustment strategies. Real-time monitoring and adjustment of topology can timely discover and solve network problems and improve network stability and reliability. Intelligent strategy generation and real-time topology update ensure that the network is always in an efficient operating state, reduce communication delays, and improve transmission efficiency. Adjustments can be made according to the real-time status of the network, and different application scenarios and demand changes can be flexibly responded to, improving the adaptability and response speed of the system. Combined with the deep Q network algorithm, decisions are made in a data-driven manner, reducing human intervention and subjective judgment, and improving the scientificity and objectivity of decision-making.

[0070] Optionally, the functional coupling relationship extraction is specifically as follows:

[0071] Extract computing flow characteristics based on edge computing data to obtain computing flow characteristic data;

[0072] Specifically, the flow characteristics of computing tasks are extracted from edge computing data, including data transmission path, transmission volume, transmission time, etc.

[0073] Node path (Path): records the transmission path of each computing task between nodes. Each node in the path is connected in sequence to form a complete task flow. Transmission volume (Volume): records the amount of data transmitted between nodes in each task in the path. Transmission time (Time): records the time required for the task to be transmitted between nodes in the path. The transmission path, data volume and time of each task are read from the edge computing data to form the computing flow feature data.

[0074] Dependency feature extraction and synergy strength feature extraction are performed based on edge computing data to obtain dependency feature data and synergy strength feature data;

[0075] Specifically, the dependency relationship between nodes (the degree of the node is used to represent its dependency degree) and the collaborative strength (the number of interactions between nodes is counted as the collaborative strength) are extracted, including task dependency, collaborative frequency and strength.

[0076] Task dependencies: record the tasks of each node that depend on other nodes. Dependencies can be unidirectional, that is, the tasks of one node depend on the tasks of another node. Collaboration Strength: record the frequency and strength of task collaboration between each node and other nodes, reflecting the closeness of the collaborative work between nodes. Read the task dependencies and collaboration strength of each node from the edge computing data to form task collaboration feature data. Collaboration strength can be a value between 0 and 1. The larger the value, the closer the collaboration.

[0077] A calculation flow recursive forest is constructed according to the calculation flow characteristic data to obtain calculation flow recursive forest data;

[0078] Specifically, based on the computational flow feature data, a computational flow recursive tree is constructed to represent the computational task flow between nodes. A nested dictionary structure is used to represent the recursive tree, and each node contains the computational task flow of its child nodes. For each node, find the target node to which it sends the computational task. Check whether the current node has child nodes (i.e., whether it is in the computational flow feature data). If there are child nodes, add the node to the tree and recursively process its child nodes. If there are no child nodes, set the node to a leaf node. Obtain computational flow recursive tree forest data. Node A sends tasks to nodes B and C, node B sends tasks to node D, node C sends tasks to nodes D and E, and so on, eventually forming a recursive tree structure.

[0079] Prune the computational flow recursive forest data according to the task coordination feature data to obtain computational flow optimized forest data;

[0080] Specifically, the recursive tree is pruned using task collaboration feature data to optimize the computing task flow. A threshold is set, and only the computing task flow between nodes with collaboration strength greater than or equal to the threshold can be retained. For each node in the tree, check whether its child nodes meet the collaboration strength threshold. If so, retain the child node and its recursive subtree; otherwise, prune the child node. Recursively process all child nodes to eventually form a pruned computing flow recursive tree. Starting from the root node of the tree, recursively traverse all nodes and perform pruning according to the collaboration strength.

[0081] The synergy strength between nodes A and B is 0.9, the synergy strength between nodes A and C is 0.8, the synergy strength between nodes B and D is 0.7, the synergy strength between nodes C and D is 0.6, the synergy strength between nodes C and E is 0.5, the synergy strength between nodes D and F is 0.4, and the synergy strength between nodes E and F is 0.3;

[0082] The synergy strength between nodes E and G is 0.2. Since the synergy strength threshold is 0.5, only the computing task flow between nodes with synergy strength greater than or equal to 0.5 is retained. Therefore, the connection between node A and node B and node C is retained, and the connection between node B and node D, node C and node D, and node C and node E is retained. Other connections that do not meet the threshold are pruned, such as the connection between node D and node F, node E and node F, and node E and node G.

[0083] According to the computing flow, the forest data is optimized to extract the single data flow and obtain the edge computing function coupling relationship data.

[0084] Specifically, the computational flow optimization forest data represents the computational flow recursive tree after pruning. The relationship between each node and its child nodes represents the flow of the computing task. Create a data structure to store the edge computing function coupling relationship data, using a dictionary or graph data structure to represent it. According to the computational flow optimization forest data, a computational flow matrix is ​​constructed. Each element in the matrix represents the computational task flow relationship between nodes. Traverse the optimized recursive tree and record the relationship from each parent node to the child node. Convert these relationships into a matrix form, where the rows and columns of the matrix represent the nodes, and the element values ​​in the matrix represent the flow relationship (1 indicates the existence of a computational flow, and 0 indicates the absence of a computational flow). According to the task collaboration feature data, a task collaboration relationship matrix is ​​constructed. Each element in the matrix represents the collaboration strength between nodes. According to the collaboration relationship between nodes, fill in the matrix, where the rows and columns of the matrix represent the nodes, and the element values ​​in the matrix represent the collaboration strength. Extract the singular data flow based on the computational flow matrix and the task collaboration relationship matrix. Traverse the computational flow matrix, and for each node pair with a computational flow (i.e., a node pair with a matrix element value of 1), check the synergy strength value in the corresponding task synergy relationship matrix. If the synergy strength value is greater than 0, record the node pair and synergy strength as a single data flow.

[0085] The present invention can more effectively utilize computing resources by extracting and optimizing computing flows and dependencies. It can optimize task scheduling and coordination, improve task execution efficiency and overall system performance. It can reduce redundant computing and data transmission in the network and improve network stability and reliability by pruning and extracting single data flows. It can provide scientific decision support and improve the intelligence level of the system by detailed data analysis and extraction of functional coupling relationships.

[0086] Optionally, the functional coupling relationship extraction further includes the following steps:

[0087] Perform computation flow feature extraction and task collaboration feature extraction based on edge computing data to obtain computation flow feature data and task collaboration feature data;

[0088] Specifically, based on the edge computing data, the computing flow feature data between the nodes is extracted. Based on the edge computing data, the task collaboration feature data between the nodes is extracted.

[0089] Constructing a calculation flow matrix according to the calculation flow characteristic data to obtain calculation flow matrix data;

[0090] Specifically, a calculation flow matrix is ​​constructed according to the calculation flow feature data.

[0091] Constructing task coordination relationships according to task coordination feature data to obtain task coordination relationship data;

[0092] Specifically, a task collaboration relationship matrix is ​​constructed based on task collaboration feature data.

[0093] Specifically, a single data flow is extracted from the optimized computation flow recursive forest to form functional coupling relationship data.

[0094] Specifically, create a data structure to store edge computing function coupling relationship data, represented by a graph data structure. Extract a single data flow based on the computing flow matrix and the task collaboration matrix. The single data flow represents the direct and unique computing task flow between nodes. Traverse the computing flow matrix, and for each node pair with a computing flow (i.e., a node pair with a matrix element value of 1), check the collaboration strength value in the corresponding task collaboration relationship matrix. If the collaboration strength value is greater than 0, record the node pair and the collaboration strength as a single data flow.

[0095] The present invention improves resource utilization and reduces redundancy in computing and transmission by optimizing the computing task flow and task coordination relationship. The optimized task flow and coordination relationship reduce unnecessary computing paths and data transmission, significantly improving the efficiency of task execution. By extracting a single data flow, redundant computing and data transmission in the network are reduced, and the stability and reliability of the network are improved. Through detailed data analysis and functional coupling relationship extraction, scientific decision support is provided to improve the intelligence level of the system. This method can dynamically adjust the computing flow and task coordination relationship, flexibly adapt to complex and changeable network environments, and improve the adaptability and flexibility of the system.

[0096] Optionally, S2 includes:

[0097] S21: extracting the traffic center according to the edge computing network data to obtain edge center data;

[0098] Specifically, the traffic center in the edge computing network is extracted, the nodes that process the most traffic are identified, and the edge center data is formed.

[0099] S22: extract edge computing dependency relationships according to edge computing network data to obtain edge computing dependency relationship data;

[0100] Specifically, the dependencies between nodes are extracted, the task dependencies and data transmission relationships between nodes are identified, and edge computing dependency data is formed.

[0101] S23: sorting the edge computing dependency data by importance to obtain edge importance sorting data;

[0102] Specifically, based on the dependency data, the importance of nodes is sorted, key nodes and sub-key nodes are identified, and edge importance sorting data is formed. Weights are assigned to each node according to the dependency data to reflect the importance of the node in the dependency relationship, and the edge importance sorting data is obtained by sorting according to the weights.

[0103] S24: Select nodes according to the edge importance ranking data to obtain edge secondary center data.

[0104] Specifically, according to the importance ranking data, select the top-ranked nodes as the secondary center nodes to form the edge secondary center data. For example, calculate the change rate according to the edge importance ranking data to obtain the average change rate data; calculate the adjacent change rate from high to low according to the edge importance ranking data to obtain the adjacent node change rate data; select the adjacent node change rate data corresponding to the nodes after the preset node threshold data in the edge importance ranking data and compare them with the average change rate data one by one until it is determined that the adjacent node change rate data is greater than or equal to the average change rate data, and the position of the node corresponding to the adjacent node change rate data corresponding to the determination in the edge importance ranking data is used as the selected segmentation point; the edge importance ranking data before the selected segmentation point is determined as the edge secondary center data.

[0105] According to the edge importance sorting data, calculate the importance change rate of each node. Find the average of all change rates to obtain the average change rate data. According to the edge importance sorting data, sort from high to low. Calculate the importance change rate between every two adjacent nodes to obtain the adjacent node change rate data. According to the preset node threshold, select the nodes after the threshold. Compare the adjacent node change rate data of these nodes with the average change rate data. Compare these nodes one by one until a node whose adjacent node change rate data is greater than or equal to the average change rate data is found. The position of the found node in the edge importance sorting data is determined as the selected split point. All nodes before the selected split point are determined as edge secondary center data.

[0106] In the present invention, by extracting traffic centers and sub-centers, key nodes in the network can be identified and optimized to improve overall network performance. Dependency extraction and importance sorting help optimize task scheduling and resource allocation, ensuring that key tasks and nodes receive priority support. The multi-level node optimization structure, including traffic centers and sub-centers, improves the redundancy and fault tolerance of the network, and enhances the stability and reliability of the network. Through systematic analysis and optimization, the overall efficiency of the edge computing network can be significantly improved, and resource waste and task conflicts can be reduced.

[0107] Optionally, S3 includes:

[0108] S31: Perform majority voting based on the edge center data and the edge sub-center data to obtain first edge data update voting data;

[0109] Specifically, according to the data of the edge center and the secondary center nodes, a majority vote is performed to select the scheme with the most votes to form the first edge data update voting data. If there are three nodes: M1, M2, and M3, where M1 and M2 send data update instructions and M3 does not send data update instructions; then the data update instruction is sent through majority voting.

[0110] S32: Perform graph weighted voting according to the edge center data and the edge sub-center data to obtain the second edge data to update the voting data;

[0111] Specifically, an interactive relationship graph is constructed, graph feature data is extracted according to the graph convolution network, edge center weight data is generated, and weighted voting is performed. Based on the graph feature data, weight data for each node is generated. The weight data reflects the importance of the node in the network and can be calculated based on the eigenvalue and connection strength. The eigenvalue of each node is obtained from the graph feature data after Gaussian convolution or maximum value convolution. The number of connections of each node is counted, that is, the degree of the node. The connection strength is calculated based on the connection weight of the node. Weight = degree*0.4+strength*0.4+feature_sum*0.2, the weight of each node is calculated based on the eigenvalue feature_sum, the node degree and the connection strength strength.

[0112] S33: Perform classified confidence voting according to the edge center data and the edge sub-center data to obtain the third edge data to update the voting data;

[0113] Specifically, the classified confidence voting method is used to perform weighted voting according to the confidence of the voting results of the nodes. Classification processing is performed according to the edge center data and the edge sub-center data to obtain edge center classification data, and confidence parameters are generated according to the edge center classification data to obtain classification confidence parameter data, and confidence calculation is performed according to the classification confidence parameters to obtain edge confidence data; after weighted voting according to the edge confidence data, the third edge data update voting data is obtained.

[0114] confidences = {'A': 0.9, 'B': 0.6, 'C': 0.7}, the confidence value of each node is between 0 and 1, reflecting the credibility of its voting results. Initialize the counter confidence_votes = Counter() to record the weighted number of votes for each node. Weight according to confidence for node in nodes:confidence_votes[node] + = confidences.get(node, 1), traverse all nodes and perform weighted counting according to their confidence. Count the voting results most_common = confidence_votes.most_common(1)[0][0], find the node with the most votes after weighting, that is, the final voting result.

[0115] S34: performing weighted average processing on the voting data updated by the first edge data, the voting data updated by the second edge data, and the voting data updated by the third edge data to obtain voting weighted average data;

[0116] Specifically, the results of different voting methods are weighted averaged, and each voting result is considered comprehensively. The sum of the weights should be 1. The weighted average is to calculate the weighted average data based on each voting result and its corresponding weight.

[0117] There are three voting results: the first edge data updates the voting data: vote1 = {A1: 0.8, B1: 0.2}, the second edge data updates the voting data: vote2 = {A2: 0.6, B2: 0.4}, and the third edge data updates the voting data: vote3 = {A3: 0.9, B3: 0.1}.

[0118] Use the vote1 data and the total data for weighted average (such as A1': 0.8 / (0.8+0.6+0.9)=0.3478, B1': 0.2 / (0.2+0.4+0.1)=0.2857), the result is the weighted average of vote1. Use the vote2 data and the total data for weighted average, the result is the weighted average of vote2 (A2': 0.2609, B2': 0.5714). Use the vote3 data and the total data for weighted average, the result is the weighted average of vote3. Get the weighted average voting data (A3': 0.3913, B3': 0.1429).

[0119] S35: Perform weighted confidence processing on the weighted average voting data to obtain edge data to update the voting data.

[0120] Specifically, the weighted confidence processing is performed on the weighted averaged data to comprehensively determine the final data update voting results. #Weighted average voting data (such as A1': 0.3478, B1': 0.2857), (A2': 0.2609, B2': 0.5714), (A3': 0.3913, B3': 0.1429). #Confidence c1 = 0.9, c2 = 0.8, c3 = 0.85. The result data is calculated according to the confidence level to obtain the edge data update voting data (A≈0.3344, B≈0.3277). A=[(A1'×c1)+(A2'×c2)+(A2'×c2)] / (c1+c2+c3)≈0.3344; B=[(B1'×c1)+(B2'×c2)+(B2'×c2)] / (c1+c2+c3)≈0.3277; Since A>B, the data is updated according to A.

[0121] The present invention uses a majority voting method to vote based on the edge center and sub-center data to determine the data update plan. Majority voting means that each node casts one vote and selects the plan with the most votes. Each node has the same voting rights, which ensures the fairness of decision-making. The graph weighted voting method can take into account the importance of the node in the network, and the nodes with larger weights have a greater impact on the decision. Through weighted voting, the influence of a single node can be reduced and the reliability of the decision can be improved. The classified confidence voting method can take into account the confidence of the node, and the nodes with high confidence have a greater impact on the decision. Through confidence voting, the reliability and accuracy of the decision can be improved. Weighted average processing can comprehensively consider the results of different voting methods and reduce the deviation of a single method. Through weighted average processing, the accuracy and reliability of the decision can be improved. Weighted confidence processing can further optimize the comprehensive decision-making results, so that the voting update methods in the three modes are processed under the condition of relatively average weights, avoiding the three voting results from being inconsistent in form or dimension.

[0122] Optionally, graph weighted voting includes:

[0123] An interactive relationship graph is constructed according to the edge center data and the edge sub-center data to obtain edge center interactive relationship graph data;

[0124] Specifically, the interaction relationship between nodes is extracted from the edge center and sub-center data to construct an interaction relationship graph. The nodes in the graph represent edge centers and sub-centers, and the edges represent the interaction relationship between nodes (such as data transmission, task collaboration, etc.).

[0125] Perform convolution processing on the edge-center interaction relationship graph data to obtain edge-center graph feature data;

[0126] Specifically, a graph convolutional network (GCN) is used to perform convolution processing on the interaction relationship graph to extract the graph feature data of the nodes. The graph feature data includes information such as the location of the node in the network and the connection strength.

[0127] Specifically, information entropy calculation is performed based on the edge center interaction graph data to obtain information entropy data of the interaction graph; it is determined whether the information entropy data of the interaction graph is greater than or equal to the preset information entropy threshold data of the interaction graph; when it is determined that the information entropy data of the interaction graph is greater than or equal to the preset information entropy threshold data of the interaction graph, Gaussian convolution calculation is performed on the edge center interaction graph data to obtain edge center graph feature data; when it is determined that the information entropy data of the interaction graph is greater than or equal to the preset information entropy threshold data of the interaction graph, the edge center interaction graph data is subjected to maximum value convolution calculation to obtain edge center graph feature data. The Gaussian convolution kernel is a two-dimensional matrix whose value is determined by the Gaussian function. A 5x5 Gaussian convolution kernel is generated with a standard deviation σ of 1.0. Gaussian convolution kernel: [[0.48086467,0.46705673,0.51004103,0.51546636,0.5020206],[0.47400654,0.46026373,0.46375933,0.5306258,0.5137221], ... [0.4844092,0.47751617,0.46968204,0.52240825,0.51109314]].

[0128] Generate edge center weights according to edge center graph feature data to obtain edge center weight data;

[0129] Specifically, based on the graph feature data, the weight data of each node is generated. The weight data reflects the importance of the node in the network and can be calculated based on the eigenvalue and connection strength. The eigenvalue of each node is obtained from the graph feature data after Gaussian convolution or maximum convolution. The number of connections of each node is counted, that is, the degree of the node. The connection strength is calculated based on the connection weight of the node. Weight = degree*0.4+strength*0.4+feature_sum*0.2, the weight of each node is calculated based on the eigenvalue feature_sum, the node degree and the connection strength strength.

[0130] Weighted voting is performed according to the edge center weight data, the edge center data and the edge sub-center data to obtain the second edge data to update the voting data.

[0131] Specifically, weighted voting is performed based on the node weight data and the edge center and sub-center data to determine the data update plan. The voting weight of each node is obtained based on the edge center weight data.

[0132] The interaction relationship diagram in the present invention can intuitively represent and analyze the relationship and interaction between nodes, providing a basis for subsequent optimization. By extracting deep feature data through a graph convolutional network, the attributes and interaction relationships of nodes can be better represented, improving the representation ability of the model. Through weight generation, the importance of each node is quantified, providing a scientific basis for resource allocation and voting decisions. Weighted voting combines node weight data to improve the scientificity and accuracy of decision-making, optimize resource allocation and system performance. By dynamically adjusting node weights and voting mechanisms, the adaptability and flexibility of the system are enhanced, and it can respond to changes in network status in a timely manner.

[0133] Optionally, the categorical confidence vote includes:

[0134] Collect update frequency according to edge center data and edge sub-center data to obtain edge center update frequency data and edge sub-center update frequency data;

[0135] Specifically, the update frequency is collected from the edge center and sub-center node data, and the update frequency data of each node is recorded.

[0136] Perform clustering calculation based on the edge center update frequency data and the edge sub-center update frequency data to obtain edge center clustering feature data;

[0137] Specifically, cluster analysis is performed using update frequency data to identify patterns and features of update frequency and form edge-center cluster feature data.

[0138] Extract cluster area features based on edge center cluster feature data to obtain cluster area feature data;

[0139] Specifically, the area characteristics of each cluster are extracted, the distribution range and density of the cluster are quantified, and cluster area characteristic data are formed.

[0140] The cluster confidence is calculated based on the cluster area feature data to obtain the edge center cluster confidence data;

[0141] Specifically, the confidence of each cluster is calculated using the cluster area feature data, the reliability of the cluster is evaluated, and the edge center cluster confidence data is formed. The area of ​​each cluster is calculated, and the cluster area is determined according to the distribution and coverage of the nodes. The node density is calculated based on the number of nodes and the cluster area of ​​each cluster. The stability and importance of the cluster center node are evaluated, which can usually be calculated based on historical data and the importance score of the node. The ratio of edge center nodes to secondary center nodes in each cluster is calculated. Based on the above calculation results, the confidence of each cluster is calculated comprehensively. The weighted average method can be used to combine the various indicators. The weight distribution is as follows: cluster area weight: 0.3, node density weight: 0.2, cluster center stability weight: 0.3, edge center to secondary center ratio weight: 0.2.

[0142] Confidence voting is performed based on the edge center clustering confidence data, the edge center data, and the edge sub-center data to obtain the third edge data to update the voting data.

[0143] Specifically, confidence voting is performed based on cluster confidence data, edge center and sub-center data to determine the data update plan. Voting weight of edge center node: determined based on its cluster confidence, the higher the confidence, the greater the weight. Voting weight of edge sub-center node: given a lower weight based on its secondary position in the network. Calculation of voting results: The voting results of edge center nodes and sub-center nodes are combined to obtain the final third edge data update voting data.

[0144] In the present invention, the update frequency is extracted from the edge center and sub-center data, and the update frequency data of each node is recorded, so as to perform confidence calculation on the data update frequency dimension at the central node to increase the credibility of the confidence, rather than the preset weight or preset confidence calculation at the central node, so as to increase the practicality of the present invention in different scenarios or uses, and reduce the problem that the use of the preset weight is too single due to a single scenario. Through clustering calculation, the pattern and characteristics of the node update frequency can be identified. Through area feature extraction, the range and density of the cluster are quantified, so as to reflect the distribution of the node mapped to the data update frequency in the central dimension, so that the fit of the confidence calculation can be better reflected in the voting stage. The confidence of each cluster is calculated using the cluster area feature data, and the reliability of the cluster is evaluated, so that the data regularity of the central node in the update frequency dimension can be better mined, and the confidence data brought by the update frequency dimension can be mined based on this. Through the confidence voting mechanism, combined with the data update frequency dimension and the deep learning algorithm, the data reliability of the node confidence is further improved, and the reliability and accuracy of the decision are improved.

[0145] Optionally, the present application further provides a cloud-edge collaborative digital economic resource scheduling and management system, which is used to execute the cloud-edge collaborative digital economic resource scheduling and management method as described above, and the cloud-edge collaborative digital economic resource scheduling and management system includes:

[0146] The edge relationship network construction module is used to obtain edge computing data and construct an edge relationship network based on the edge computing data to obtain edge computing network data;

[0147] The edge center and secondary center extraction module is used to extract the edge center and the edge secondary center according to the edge computing network data to obtain the edge center data and the edge secondary center data;

[0148] The edge data update voting module is used to perform data update voting based on the edge center data and the edge sub-center data to obtain edge data update voting data;

[0149] The digital economy resource scheduling module is used to send edge data update voting data to the cloud to perform digital economy resource scheduling operations.

[0150] The purpose of the present invention is that through high-frequency data updates, the system can reflect the status and changes of edge computing nodes in real time, thereby improving the response speed and accuracy of the system. Rapidly build an edge relationship network to ensure that the system can quickly adapt to changes in the network environment and improve the timeliness of resource scheduling. In a test environment with 100 edge nodes, the construction time of the edge relationship network is less than 5 seconds, ensuring the rapid construction and response of the network topology. High-accuracy edge center and sub-center extraction ensures the accurate identification of key nodes and provides a reliable data basis for resource optimization. The extraction process is fast and efficient, ensuring that the system can quickly adapt to network changes and improving the response speed of resource scheduling. In the test environment, the main traffic center and sub-center nodes are accurately identified through edge computing network data with an accuracy of more than 95%. By combining comprehensive voting decisions in different dimensions, the scientificity and reliability of data updates are ensured, high-frequency data interactions are reduced, the interaction frequency of reliable data is increased, and the overall performance of the system is improved. The average response time for resource scheduling in the cloud is 5 seconds, and resource utilization is increased by 15%. Efficient resource scheduling improves resource utilization and overall system performance, ensuring the efficient operation of digital economic applications.

[0151] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0152] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A cloud-edge collaborative digital economy resource scheduling and management method, characterized in that: The method comprises: S1. Obtain edge computing data, and construct an edge relationship network based on the edge computing data to obtain edge computing network data; S2. Extract edge centers and edge sub-centers according to edge computing network data to obtain edge center data and edge sub-center data; S3, perform data update voting according to the edge center data and the edge sub-center data to obtain edge data update voting data; S4, sending edge data updated voting data to the cloud for digital economic resource scheduling operations; S1 includes: Get edge computing data; Construct network topology according to edge computing data to obtain edge computing network topology data; Extract the functional coupling relationship based on the edge computing data to obtain the edge computing functional coupling relationship data; According to the edge computing function coupling relationship data and edge computing data, a local coupling relationship network is constructed to obtain edge local coupling relationship network data; According to the edge computing network topology data and the edge local coupling relationship network data, a global coupling relationship network is constructed to obtain edge computing network data; The functional coupling relationship extraction is specifically as follows: Extract computing flow characteristics based on edge computing data to obtain computing flow characteristic data; Dependency feature extraction and synergy strength feature extraction are performed based on edge computing data to obtain dependency feature data and synergy strength feature data; A calculation flow recursive forest is constructed according to the calculation flow characteristic data to obtain calculation flow recursive forest data; Prune the computational flow recursive forest data according to the task coordination feature data to obtain computational flow optimized forest data; According to the computing flow, the forest data is optimized to extract the single data flow and obtain the edge computing function coupling relationship data.

2. The method according to claim 1, characterized in that Network topology construction includes: Register nodes according to edge computing data to obtain edge node registration data; Perform node initialization according to edge node registration data to obtain edge node initialization data; Perform initial topology construction according to edge node initialization data to obtain edge node topology data; Real-time data collection is performed based on the edge node topology data to obtain the edge node topology real-time data; Adjust data generation based on real-time data of edge node topology to obtain edge node adjustment data; The edge node topology data is updated in real time according to the edge node adjustment data to obtain the edge computing network topology data; The adjustment data generation is specifically as follows: Extract features based on the real-time data of edge node topology to obtain edge node topology feature data; Data is generated based on the edge node topology feature data through the deep Q network algorithm to obtain edge node adjustment data.

3. The method according to claim 1, characterized in that Functional coupling relationship extraction also includes the following steps: Perform computation flow feature extraction and task collaboration feature extraction based on edge computing data to obtain computation flow feature data and task collaboration feature data; Constructing a calculation flow matrix according to the calculation flow characteristic data to obtain calculation flow matrix data; Constructing task coordination relationships according to task coordination feature data to obtain task coordination relationship data; Based on the computing flow matrix data and task coordination relationship data, a single data flow is extracted to obtain the edge computing function coupling relationship data.

4. The method according to claim 1, characterized in that S2 include: Extract the traffic center based on the edge computing network data to obtain the edge center data; Extract edge computing dependencies based on edge computing network data to obtain edge computing dependency data; Sort the edge computing dependency data by importance to obtain edge importance sorting data; Nodes are selected based on the edge importance sorting data to obtain edge secondary center data.

5. The method according to claim 1, characterized in that S3 include: Perform majority voting based on edge center data and edge sub-center data to obtain first edge data to update voting data; Perform graph weighted voting based on edge center data and edge secondary center data to obtain second edge data to update voting data; Perform classified confidence voting based on edge center data and edge sub-center data to obtain third edge data to update voting data; Perform weighted average processing on the voting data updated by the first edge data, the voting data updated by the second edge data, and the voting data updated by the third edge data to obtain voting weighted average data; Perform weighted confidence processing on the weighted average voting data to obtain edge data to update the voting data.

6. The method according to claim 5, characterized in that Graph weighted voting includes: An interactive relationship graph is constructed according to the edge center data and the edge sub-center data to obtain edge center interactive relationship graph data; Perform convolution processing on the edge-center interaction relationship graph data to obtain edge-center graph feature data; Generate edge center weights according to edge center graph feature data to obtain edge center weight data; Weighted voting is performed according to the edge center weight data, the edge center data and the edge sub-center data to obtain the second edge data to update the voting data.

7. The method according to claim 5, characterized in that Categorical confidence voting includes: Collect update frequency according to edge center data and edge sub-center data to obtain edge center update frequency data and edge sub-center update frequency data; Perform clustering calculation based on the edge center update frequency data and the edge sub-center update frequency data to obtain edge center clustering feature data; Extract cluster area features based on edge center cluster feature data to obtain cluster area feature data; The cluster confidence is calculated based on the cluster area feature data to obtain the edge center cluster confidence data; Confidence voting is performed based on the edge center clustering confidence data, the edge center data, and the edge sub-center data to obtain the third edge data to update the voting data.

8. A cloud-edge collaborative digital economy resource scheduling and management system, characterized in that: Used to execute the cloud-edge collaborative digital economic resource scheduling and management method according to claim 1, the cloud-edge collaborative digital economic resource scheduling and management system includes: The edge relationship network construction module is used to obtain edge computing data and construct an edge relationship network based on the edge computing data to obtain edge computing network data; The edge center and secondary center extraction module is used to extract the edge center and the edge secondary center according to the edge computing network data to obtain the edge center data and the edge secondary center data; The edge data update voting module is used to perform data update voting based on the edge center data and the edge sub-center data to obtain edge data update voting data; The digital economy resource scheduling module is used to send edge data update voting data to the cloud to perform digital economy resource scheduling operations.

Images