A cross-department collaborative multi-link communication method and system

Abstract

The present invention discloses a cross-departmental collaborative multi-link communication method and system, which relates to the field of electrical communication technology, including collecting network performance data from links of various departments through monitoring equipment to construct a cross-departmental network topology map; training using a graph neural network algorithm according to the cross-departmental network topology map to generate a cross-departmental link scheduling strategy; performing link redundancy configuration of a multi-path TCP protocol according to the cross-departmental link scheduling strategy to optimize link data flow distribution; real-time monitoring of link failures, switching data flows to spare links according to the multi-path TCP protocol for failed links, and updating the cross-departmental network topology map to perform fault location and recovery. The present invention improves the reliability, stability and efficiency of the network in cross-departmental collaboration through intelligent and dynamic link scheduling and flow distribution optimization.

Description

Technical Field

[0001] The present invention relates to the field of electrical communication technology, and in particular to a cross-departmental collaborative multi-link communication method and system. Background Art

[0002] With the rapid development of information and digital technologies, the demand for cross-departmental collaboration is increasing, especially in enterprises and government agencies. Large-scale data sharing and information exchange between departments are becoming more and more complex. Multi-link communication technology has therefore become one of the key technologies to support cross-departmental collaboration. In traditional cross-departmental communication systems, the networks of each department are usually independent, and the selection of links and traffic scheduling rely on static configuration and simple routing algorithms, which are difficult to adapt to dynamic network environments and complex business needs. In recent years, with the introduction of the multi-path TCP (MPTCP) protocol, the link redundancy of the network has been effectively improved, which can achieve data flow transmission through multiple paths and improve the reliability and stability of the network. At the same time, as an emerging deep learning algorithm, graph neural network (GNN) has shown great potential in graph structure data processing, and can mine complex correlations in multi-dimensional and multi-source data, thereby providing more intelligent support for link scheduling and traffic optimization.

[0003] However, the existing technology still has certain shortcomings in practical applications. First, traditional link scheduling methods generally rely on preset static topology maps and simple scheduling algorithms, and cannot respond in real time to dynamic changes in network status, resulting in low link utilization and long fault recovery time. Secondly, the existing multi-path TCP protocol mainly implements link redundancy configuration through static path selection, but in the face of large-scale and diverse cross-departmental network environments, it is impossible to perform effective traffic scheduling, resulting in unbalanced data flow and waste of network resources. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a cross-departmental collaborative multi-link communication method to solve the problem of low link utilization and inability to perform effective traffic scheduling.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, the present invention provides a cross-departmental collaborative multi-link communication method, which includes collecting network performance data from the links of each department through monitoring equipment to construct a cross-departmental network topology map; using a graph neural network algorithm for training based on the cross-departmental network topology map to generate a cross-departmental link scheduling strategy; based on the cross-departmental link scheduling strategy, performing link redundancy configuration of the multi-path TCP protocol to optimize link data flow distribution; real-time monitoring of link failures, switching the data flow to the backup link according to the multi-path TCP protocol for the failed link, and updating the cross-departmental network topology map to perform fault location and recovery.

[0008] As a preferred solution of the cross-departmental collaborative multi-link communication method of the present invention, wherein: the network performance data is collected from the links of each department through the monitoring device to construct a cross-departmental network topology map, and the specific steps are:

[0009] Deploy network monitoring equipment and configure it to collect network performance data for each link;

[0010] Preprocess the collected network performance data;

[0011] transmitting the pre-processed network performance data to a central data processing unit;

[0012] Build a cross-department network topology map based on the collected network performance data;

[0013] Through visualization means, the cross-departmental network topology is graphically displayed.

[0014] As a preferred solution of the cross-departmental collaborative multi-link communication method of the present invention, wherein: the cross-departmental network topology diagram is trained using a graph neural network algorithm to generate a cross-departmental link scheduling strategy, and the specific steps are:

[0015] According to the cross-department network topology graph, the input of the graph neural network is defined as the node feature matrix X and the adjacency matrix A of the graph, and the output is defined as the cross-department link scheduling strategy S;

[0016] Select the GCN-based graph neural network model to build the graph model;

[0017] High-order graph convolution is introduced to enhance the learning ability of the graph model through multi-order propagation of the weighted adjacency matrix, and multi-scale information fusion is used to adjust the adjacency matrix. The expression is: ;

[0018] in, is the node representation after graph convolution operation, is the node representation of the previous layer, is the normalized adjacency matrix Power, is the weight matrix of each order convolution, σ is the ReLU activation function, is the layer coefficient, is the maximum order in high-order graph convolution, is the power coefficient of the adjacency matrix;

[0019] An adaptive adjacency matrix is ​​introduced to adaptively adjust the adjacency matrix structure based on the real-time network status. The expression is: ;

[0020] in, is the adaptively adjusted adjacency matrix, A is the original adjacency matrix, D is the degree matrix, It is a dynamic adjustment item based on real-time network status. is the Frobenius norm, is the normalization function;

[0021] After multi-layer graph convolution operations, node feature representation is obtained , the node representation is mapped to the cross-department link scheduling strategy S through the fully connected layer

[0022] Design loss function by combining network performance and bandwidth allocation optimization;

[0023] Use the Adam optimizer to optimize the loss function and train the graph neural network model;

[0024] After completing the model training, the cross-department link scheduling strategy S is generated.

[0025] As a preferred solution of the cross-departmental collaborative multi-link communication method of the present invention, wherein: the link redundancy configuration of the multi-path TCP protocol is performed according to the cross-departmental link scheduling strategy, and the specific steps are:

[0026] Enable MPTCP support on both client and server side;

[0027] Initialize the sub-flow parameters of MPTCP;

[0028] According to the cross-department link scheduling strategy S, the main path and the backup path are distinguished to complete the sub-flow link allocation.

[0029] As a preferred solution of the cross-departmental collaborative multi-link communication method of the present invention, wherein: the optimization link data flow allocation is specifically performed as follows:

[0030] Based on the cross-department link scheduling strategy S generated by the graph neural network and the real-time network performance data of the link, the allocation of link data flow is dynamically adjusted. The expression is: ;

[0031] in, For Link The distribution flow, For Link The scheduling strategy, is the link priority adjustment factor, For Link performance index, M is the number of links involved in the transmission, For the link that is currently distributing traffic, is the index coefficient of the number of links M participating in the transmission;

[0032] Distribute traffic based on links , configure the traffic ratio between the main link and the backup link, and allocate multi-path sub-flows.

[0033] As a preferred solution of the cross-departmental collaborative multi-link communication method of the present invention, wherein: the real-time monitoring of link failures, for the failed link, switching the data stream to the backup link according to the multi-path TCP protocol, the specific steps are:

[0034] Using monitoring equipment to continuously collect link performance indicators and transmit the link performance indicators to a central data processing unit;

[0035] Set the fault judgment threshold according to the link performance indicators;

[0036] According to the link performance indicators collected in real time, the system compares them with the set fault judgment threshold, performs fault detection, and marks the faulty link;

[0037] For failed links, MPTCP is used to switch to backup links.

[0038] After the backup link is activated, the traffic distribution data of each link is recalculated according to the new network topology, and the data flow distribution is adjusted according to the real-time situation of the link.

[0039] As a preferred solution of the cross-departmental collaborative multi-link communication method of the present invention, wherein: the updating of the cross-departmental network topology map, fault location and recovery, the specific steps are:

[0040] Update the cross-department network topology map based on the links marked as faulty;

[0041] Update the graphical display of cross-department network topology through visualization;

[0042] Generate fault alarms to notify network management personnel or automated fault handling modules to perform fault repairs;

[0043] After the link failure is restored, the restored link is reintroduced into the network and data flow distribution is adjusted according to the link performance status.

[0044] In the second aspect, the present invention provides a cross-departmental collaborative multi-link communication system, including a data acquisition module, a model building module, a link configuration module and a fault handling module; the data acquisition module is used to collect network performance data from the links of each department through monitoring equipment to construct a cross-departmental network topology map; the model building module is used to train using a graph neural network algorithm based on the cross-departmental network topology map to generate a cross-departmental link scheduling strategy; the link configuration module is used to perform link redundancy configuration of the multi-path TCP protocol based on the cross-departmental link scheduling strategy, and optimize the link data flow distribution; the fault handling module is used to monitor link failures in real time, switch the data flow to the backup link according to the multi-path TCP protocol for the faulty link, and update the cross-departmental network topology map to perform fault location and recovery.

[0045] In a third aspect, the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the cross-departmental collaborative multi-link communication method as described in the first aspect of the present invention is implemented.

[0046] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the cross-departmental collaborative multi-link communication method as described in the first aspect of the present invention.

[0047] The beneficial effects of the present invention are as follows: the present invention deploys monitoring equipment in the links of each department, collects network performance data and constructs a cross-department network topology map, and generates a link scheduling strategy in combination with a graph neural network algorithm, thereby realizing link redundancy configuration and data flow optimization based on a multi-path TCP protocol. This method can dynamically adjust the link data flow distribution to improve network efficiency and stability. When a link failure occurs, the system ensures uninterrupted communication by real-time monitoring and rapid switching of data flows to backup links, and quickly locates the source of the fault and restores network functions by updating the network topology map and visual display. Through intelligent and dynamic link scheduling and traffic distribution optimization, the reliability, stability and efficiency of the network in cross-departmental collaboration are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0049] Figure 1 This is a flow chart of the cross-departmental collaborative multi-link communication method in Example 1.

[0050] Figure 2 This is a module diagram of the cross-departmental collaborative multi-link communication system in Example 1. DETAILED DESCRIPTION

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0053] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0054] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and provides a cross-departmental collaborative multi-link communication method, including:

[0055] S1: Collect network performance data from links in each department through monitoring equipment and build a cross-department network topology map.

[0056] Specifically, the following steps are included:

[0057] S1.1: Deploy network monitoring equipment and configure it to collect network performance data for each link.

[0058] Specifically, network monitoring equipment needs to be deployed at the link entrances and key nodes of each department, including routers, switches, gateways and other network equipment of each department.

[0059] Monitoring equipment should be able to support multiple protocols (such as SNMP, NetFlow, sFlow, etc.) and high-bandwidth monitoring to ensure efficient data collection.

[0060] After the installation is complete, the device is configured for network so that it can be seamlessly connected to the upper-level central data processing unit.

[0061] After the configuration is completed, perform equipment debugging to ensure that the monitoring equipment can work stably.

[0062] Among them, network performance data includes but is not limited to: bandwidth, latency, packet loss rate, throughput, traffic load, link utilization, maximum transmission rate of the link, etc.

[0063] The acquisition cycle is set to every second to ensure real-time performance.

[0064] S1.2: Preprocess the collected network performance data.

[0065] It should be noted that the data preprocessing method is:

[0066] The collected data is cleaned to remove noise, duplication and invalid data. The cleaning process also includes data standardization, unifying the different data formats and units of the links of each department.

[0067] Process missing data that may appear during the collection process. Use mean filling, interpolation, and data interpolation algorithms to fill in missing values.

[0068] The moving average smoothing technique is used to remove instantaneous fluctuations and extract stable data trends to better reflect the long-term changes in link performance.

[0069] Key features are extracted from the collected network performance data, including bandwidth, latency, packet loss rate, throughput, traffic load, link utilization, maximum link transmission rate, etc. These features will be used for subsequent network topology modeling and analysis.

[0070] S1.3: Transmit the pre-processed network performance data to the central data processing unit.

[0071] Specifically, MQTT is selected as the data transmission protocol to ensure that data can be transmitted from the monitoring equipment to the central data processing unit in real time and stably.

[0072] In the process of data transmission, in order to ensure the security of data, the SSL encryption algorithm is used to encrypt the transmitted data. At the same time, the gzip data compression algorithm is used to reduce the bandwidth consumption of data transmission.

[0073] Considering the fluctuation of the network and the load of the equipment, a cache mechanism is introduced to ensure that even if there is a temporary interruption in the data transmission process, there will be no data loss. After the data is cached, it will be uploaded to the central data processing unit in batches at regular intervals.

[0074] S1.4: Build a cross-departmental network topology map based on the collected network performance data.

[0075] Specifically, the network devices of each department (such as routers, switches, terminal devices, etc.) are used as nodes in the graph, and the links are used as edges between nodes. The network is modeled using topology modeling software.

[0076] Define corresponding attributes for each node (network device), such as bandwidth, latency, etc.; define performance characteristics for each link, such as packet loss rate, throughput, traffic load, link utilization, maximum transmission rate, etc.

[0077] As network status and performance data change, the network topology map is dynamically updated to promptly reflect topology changes such as link disconnection, device addition or failure.

[0078] S1.5: Graphically display the cross-departmental network topology through visualization.

[0079] Specifically, based on the establishment of the topology map, visualization tools such as D3.js are used to display the network structure.

[0080] Design a graphical interface for the network topology diagram so that it can clearly show the connection relationship between departments. Use different colors, sizes, shapes and other elements to represent different types of equipment or link status (such as normal, faulty, low latency, etc.). Add interactive functions to the graphical display to allow users to view detailed information on each node and link, such as bandwidth, latency and other performance indicators, and support operations such as zooming in, zooming out, and dragging, so that users can analyze network status from different angles.

[0081] Ensure that the network topology diagram can be synchronized with the central data processing unit in real time to reflect the latest network status changes. If a link failure occurs or a device goes offline, the system can instantly update the graphical display and highlight the faulty node.

[0082] Preferably, by deploying monitoring equipment and collecting network performance data of links in each department, real-time monitoring of network status can be achieved, providing accurate basic data for subsequent analysis. Data preprocessing ensures the accuracy and consistency of collected data, eliminates noise and anomalies, and extracts stable performance trends. At the same time, the encryption, compression, and caching mechanisms during data transmission ensure data security and transmission efficiency, and ensure data integrity even when the network is unstable. Based on these data, the cross-department network topology map constructed can intuitively display the network structure and performance, helping administrators to understand the health status of the network in real time and respond quickly when a fault occurs. Graphical display and interactive functions further enhance the visualization of network management, enabling network administrators to efficiently locate and solve network problems, thereby improving the stability, reliability, and optimization capabilities of the network.

[0083] S2: Based on the cross-department network topology, the graph neural network algorithm is used for training to generate a cross-department link scheduling strategy.

[0084] Specifically, the following steps are included:

[0085] S2.1: According to the cross-department network topology graph, the input of the graph neural network is defined as the node feature matrix X and the adjacency matrix A of the graph, and the output is defined as the cross-department link scheduling strategy S.

[0086] In the node feature matrix X, each node represents a device in the network (such as a router, switch, etc.), and its features include network performance data such as bandwidth, latency, packet loss rate, link utilization, etc. These features will be dynamically updated according to the current performance status of each node.

[0087] The adjacency matrix A represents the connection relationship in the network. The elements of A are =1, indicating that node e and node There is a connection between =0, indicating that node e and node There is no connection between them. The adjacency matrix can be directly obtained through the cross-department network topology graph.

[0088] The cross-department link scheduling strategy S is the scheduling strategy of the network, including the traffic allocation, bandwidth allocation, path selection, etc. of each link.

[0089] S2.2: Select a GCN-based graph neural network model to build a graph model.

[0090] It should be understood that the graph convolutional network (GCN) is chosen as the basic model because GCN can effectively capture the dependencies between nodes when processing graph structured data and is suitable for network topology modeling.

[0091] S2.3: Introduce high-order graph convolution, enhance the learning ability of the graph model through multi-order propagation of the weighted adjacency matrix, and use multi-scale information fusion to adjust the adjacency matrix. The expression is: ;

[0092] in, is the node representation after graph convolution operation, is the node representation of the previous layer, is the normalized adjacency matrix Power, is the weight matrix of each order convolution, σ is the ReLU activation function, is the layer coefficient, is the maximum order in high-order graph convolution, is the power coefficient of the adjacency matrix.

[0093] Preferably, by introducing high-order graph convolution, the model can capture more complex interaction information between distant nodes. This has not been widely used in traditional GCN, especially in cross-departmental network scheduling, where the relationship between distant nodes is often ignored, which directly affects the accuracy and effectiveness of the scheduling strategy. Through multi-order graph convolution, the model can consider more network topology information, thereby generating a more accurate scheduling strategy.

[0094] S2.4: Introduce an adaptive adjacency matrix to adaptively adjust the adjacency matrix structure based on the real-time network status. The expression is: ;

[0095] in, is the adaptively adjusted adjacency matrix, A is the original adjacency matrix, D is the degree matrix, It is a dynamic adjustment item based on real-time network status. is the Frobenius norm, is the normalization function.

[0096] Specifically, It is a dynamic adjustment item based on the real-time network status. It represents the dynamic changes in the network, such as changes caused by factors such as increase in network traffic, link failure, link bandwidth change, etc. This adjustment item is calculated based on the current network status and updates the adjacency matrix in real time.

[0097] Preferably, in existing GCNs, the normalization of the adjacency matrix usually uses a simple degree normalization: ;

[0098] This formula is sufficient for most graphs, but in a specific network, the structure of the adjacency matrix may not be fixed. The link status of the cross-departmental network will change with time and data traffic. Therefore, using a fixed adjacency matrix for graph convolution cannot fully capture the changes in the dynamic network. By introducing an adaptive adjacency matrix, the network can respond to changing bandwidth, latency, link failures and other issues in real time, making the graph neural network more adaptable to the actual cross-departmental network scheduling needs.

[0099] S2.5: After multi-layer graph convolution operations, node feature representation is obtained , the node representation is mapped to the cross-department link scheduling strategy S through a fully connected layer.

[0100] Specifically, node feature representation Contains the optimization information of each network device (node). The generated cross-department link scheduling strategy S satisfies the following formula: ;

[0101] in, is a nonlinear mapping function, which consists of a fully connected layer and a ReLU activation function. Finally, the cross-department link scheduling strategy S is obtained. The matrix represents the slave node To Node The bandwidth allocation ratio of the link.

[0102] S2.6: Design loss function by combining network performance and bandwidth allocation optimization.

[0103] Specifically, the loss function is designed from three aspects: bandwidth utilization, delay and packet loss rate. The expression is: ;

[0104] in, is the loss function of bandwidth utilization, is the delay loss function, is the packet loss rate loss function, S is the cross-department link scheduling strategy; θ is the weight coefficient of the bandwidth utilization loss function, β is the weight coefficient of the delay loss function, and γ is the weight coefficient of the packet loss rate loss function, which is used to balance the contribution of different loss terms to the total contribution.

[0105] Furthermore, The expression is: ;

[0106] in, Yes Link To Link The actual transmission delay, Yes Link To Link The maximum allowed delay.

[0107] The expression is: ;

[0108] in, Yes Link To Link The packet loss rate.

[0109] S2.7: Use the Adam optimizer to optimize the loss function and train the graph neural network model.

[0110] Specifically, during the training process, gradients are calculated and parameters are updated through back propagation until the loss function converges.

[0111] S2.8: Complete model training and generate the cross-department link scheduling strategy S.

[0112] Preferably, by introducing adaptive adjacency matrix and high-order graph convolution, it can respond to changes in network status in real time, accurately capture the dependencies between distant nodes, and generate more accurate scheduling strategies. The multi-dimensional loss function optimizes bandwidth utilization, latency, and packet loss rate to improve network performance, and uses the Adam optimizer for efficient training to ensure rapid convergence and stability of the model. Compared with traditional methods, the automated and intelligent features of this solution improve the accuracy and efficiency of scheduling strategies, have certain adaptability and scalability, and ensure efficient allocation and optimization of network resources.

[0113] S3: Based on the cross-department link scheduling strategy, perform link redundancy configuration of the multi-path TCP protocol and optimize link data flow distribution.

[0114] Specifically, the following steps are included:

[0115] S3.1: Enable MPTCP support on both the client and server sides.

[0116] Specifically, MCTCP support is enabled by modifying the kernel configuration file (for example, the sysctl command in the Linux kernel), enabling the TCP extension function and setting related parameters.

[0117] S3.2: Initialize the sub-flow parameters of MPTCP.

[0118] Specifically, initial transmission parameters (such as congestion window, maximum transmission unit, etc.) are set for each sub-flow, and specific network parameters are configured for each sub-flow, such as link priority, maximum throughput, etc., to ensure that the collaboration between sub-flows can flexibly adapt under different network conditions.

[0119] S3.3: Based on the cross-department link scheduling strategy S, the primary path and the backup path are distinguished to complete the sub-flow link allocation.

[0120] Specifically, the links are sorted according to the performance indicators of each link, such as bandwidth, delay, packet loss rate, etc. The link with the best performance is assigned as the main path, and the other links are used as backup paths. Subflows are assigned to the main path and the backup path respectively.

[0121] S3.4: Based on the cross-department link scheduling strategy S generated by the graph neural network and the real-time network performance data of the link, the allocation of link data flow is dynamically adjusted, and the expression is: ;

[0122] in, For Link The distribution flow, For Link The scheduling strategy, is the link priority adjustment factor, For Link performance index, M is the number of links involved in the transmission, For the link that is currently distributing traffic, is the index coefficient of the number of links M participating in the transmission.

[0123] Preferably, traffic is allocated to each link , according to the link scheduling strategy and link performance indicators Dynamic adjustment. This allows traffic distribution to be optimized based on the real-time status of the link (such as bandwidth, latency, etc.) to avoid overloaded and high-latency paths. Whenever the network status changes (such as bandwidth changes, link failures, latency fluctuations, etc.), the network topology and adjacency matrix are updated in real time, and the scheduling strategy is recalculated. and the priority adjustment factor Further optimize link traffic distribution.

[0124] S3.5: Distribute traffic based on links , configure the traffic ratio between the main link and the backup link, and allocate multi-path sub-flows.

[0125] It should be understood that more traffic is allocated to the primary link to ensure that the primary path preferentially carries most of the data traffic; when the primary link cannot carry the traffic, the backup link takes over the traffic.

[0126] Preferably, by enabling the MPTCP protocol and combining cross-department link scheduling strategies, implementing link redundancy configuration and data flow distribution optimization, the stability, reliability and performance of the network can be improved. By dynamically adjusting link traffic distribution, bandwidth utilization is optimized based on real-time network performance data to ensure that the main link carries most of the traffic, and the backup link automatically takes over when the main link load is too high, avoiding overload and excessive delay. Adaptive link priority adjustment and multi-path load balancing help improve throughput, reduce latency, and ensure that the network can be flexibly adjusted in a dynamically changing environment, providing efficient and reliable cross-department data transmission.

[0127] S4: Monitor link failures in real time. For failed links, switch data streams to backup links according to the multipath TCP protocol, and update the cross-department network topology map to locate and recover from the failure.

[0128] Specifically, the following steps are included:

[0129] S4.1: Use monitoring equipment to continuously collect link performance indicators and transmit the link performance indicators to a central data processing unit.

[0130] It should be noted that the collected performance indicators include but are not limited to: bandwidth, latency, packet loss rate, throughput, traffic load, link utilization, maximum link transmission rate and other data.

[0131] S4.2: Set the fault judgment threshold according to the link performance index.

[0132] The threshold is set according to the service requirements and link performance requirements. The fault judgment threshold is set as follows:

[0133] Delay threshold: If the link delay exceeds 200ms, the link is considered to be faulty.

[0134] Packet loss rate threshold: When the packet loss rate exceeds 5%, the link is considered unstable.

[0135] Bandwidth utilization threshold: When the throughput is lower than 50% bandwidth utilization, the link performance is considered to be degraded.

[0136] For some links with frequent fluctuations, the delay or packet loss rate threshold can be appropriately increased.

[0137] S4.3: Based on the link performance indicators collected in real time, the indicators are compared with the set fault judgment thresholds to perform fault detection and mark the faulty links.

[0138] Specifically, a threshold detection method is used to compare data to determine whether a link is faulty. Once a link fault is detected, the faulty link is marked and a fault event log is stored. The faulty link will be marked as "failed" and temporarily removed from the link usage pool to prevent it from being used.

[0139] S4.4: For a failed link, MPTCP is used to switch to a backup link.

[0140] Specifically, after detecting a link failure, MPTCP will select a suitable backup link based on the real-time status information of the link (such as the bandwidth and delay of the backup link). The data flow will automatically switch to the backup link without manual intervention. The MPTCP protocol dynamically adjusts the load of each link to ensure maximum bandwidth utilization. The data packet transmission during the switching process will be seamless, avoiding packet loss and increased delay.

[0141] S4.5: After the standby link is activated, the traffic distribution data of each link is recalculated according to the new network topology, and the data flow distribution is adjusted according to the real-time situation of the link.

[0142] S4.6: Update the cross-department network topology map based on the links marked as faulty.

[0143] Specifically, the faulty link will be deleted from the current network topology to form a new topology. The updated topology will reflect the activation status of the backup link and the network path after the faulty link is removed.

[0144] S4.7: Update the graphical display of the cross-departmental network topology through visualization.

[0145] S4.9: Generate a fault alarm to notify the network management personnel or the automated fault handling module to repair the fault.

[0146] It should be understood that the alarm content includes detailed information of the faulty link (such as link ID, fault type, occurrence time, etc.) and the activation status of the standby link.

[0147] S4.10: After the link failure is restored, the restored link is reintroduced into the network, and the data flow distribution is adjusted according to the link performance status.

[0148] Preferably, by real-time monitoring of link performance and using multipath TCP protocol to implement link redundancy switching, it can automatically switch to the backup link when a link failure occurs to ensure that the data flow is not interrupted, while updating the network topology map and adjusting the traffic distribution. Through fault detection, performance threshold comparison, real-time topology update and alarm mechanism, the fault response speed is optimized, manual intervention is reduced, and the reliability, stability and automated management capabilities of the network are improved. After the link is restored, the traffic distribution is automatically reintroduced and intelligently optimized, further enhancing the redundancy and resource utilization efficiency of the network, and improving user experience and operation and maintenance efficiency.

[0149] This embodiment also provides a cross-departmental collaborative multi-link communication system, including:

[0150] The data collection module is used to collect network performance data from links in each department through monitoring equipment and build a cross-department network topology map;

[0151] The model building module is used to generate cross-department link scheduling strategies based on the cross-department network topology diagram using the graph neural network algorithm for training;

[0152] The link configuration module is used to configure link redundancy of the multi-path TCP protocol and optimize the link data flow distribution according to the cross-department link scheduling strategy;

[0153] The fault handling module is used to monitor link failures in real time. For failed links, it switches the data stream to the backup link according to the multi-path TCP protocol, and updates the cross-department network topology map to locate and recover the fault.

[0154] This embodiment also provides a computer device, which is suitable for the case of a cross-department collaborative multi-link communication method, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the cross-department collaborative multi-link communication method proposed in the above embodiment.

[0155] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0156] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for realizing cross-departmental collaborative multi-link communication proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0157] In summary, the present invention achieves link redundancy configuration and data flow optimization based on the multi-path TCP protocol by: deploying monitoring equipment in the links of each department, collecting network performance data and constructing a cross-department network topology map, and generating a link scheduling strategy in combination with a graph neural network algorithm. This method can dynamically adjust the link data flow distribution to improve network efficiency and stability. When a link failure occurs, the system ensures uninterrupted communication by real-time monitoring and rapid switching of data flows to backup links, and quickly locates the source of the fault and restores network functions by updating the network topology map and visual display. Through intelligent and dynamic link scheduling and traffic distribution optimization, the reliability, stability and efficiency of the network in cross-departmental collaboration are improved.

[0158] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of the cross-departmental collaborative multi-link communication method is provided.

[0159] The experiment built a cross-departmental multi-link network with 5 departments (department A, B, C, D, E), which are connected to each other through several links. The network performance data is collected through the network monitoring tool, the above performance data is integrated into a node feature matrix, and the adjacency matrix is ​​constructed according to the link connection relationship. The network is trained by the graph convolutional neural network (GCN). The goal of the network is to generate an optimized link scheduling strategy based on the collected network performance data to ensure efficient transmission of network traffic. Based on the link scheduling strategy, the allocated traffic for each link is calculated. The traffic allocation of each link needs to consider factors such as bandwidth, latency, and throughput to avoid network bottlenecks. The monitoring system regularly evaluates the health of the link and detects whether indicators such as packet loss rate and throughput exceed the threshold. When a link fails, it automatically switches to the backup link. Update the topology map and traffic allocation, and generate a fault alarm.

[0160] The network performance data is collected through the network monitoring tool, as shown in Table 1 below:

[0161] Table 1 Network performance data table

[0162]

[0163] Integrate the collected network performance data into the node feature matrix X:

[0164] ;

[0165] Construct the adjacency matrix A according to the link connection relationship:

[0166] ;

[0167] The network is trained through a graph convolutional neural network (GCN) to obtain the link scheduling strategy S, as shown in Table 2 below:

[0168] Table 2 Link scheduling strategy table

[0169]

[0170] The rows represent source nodes (starting from a certain node) and the columns represent target nodes (reaching a certain node). The values ​​in the matrix represent the To Node The diagonal elements represent their own bandwidth allocation percentage.

[0171] Based on the link scheduling strategy S, the allocated flow of each link is calculated as follows: ;

[0172] Design priority adjustment factors based on bandwidth and latency ,Specifically, the link with larger bandwidth and smaller delay has a higher priority.,Therefore, the following priority adjustment factor is assigned to each link.

[0173] For bandwidth and delay , priority adjustment factor The expression is: ;

[0174] Calculate the priority adjustment factor of each link according to the formula:

[0175] A to B: ;

[0176] B to C: ;

[0177] C to D: ;

[0178] D to E: ;

[0179] A to E: ;

[0180] Using link throughput as a performance indicator , first calculate: ;

[0181] The throughput ( ) Bring in: ;

[0182] Using Link Scheduling Policy Table , calculate the allocated flow of each link according to the formula:

[0183] For link A to B: ;

[0184] For link B to C: ;

[0185] For link C to D: ;

[0186] For link D to E: ;

[0187] For link A to E: ;

[0188] The traffic distribution results are shown in Table 3 below:

[0189] Table 3 Link traffic distribution table

[0190]

[0191] According to the traffic distribution of the links in Table 4, the traffic ratio of the main link and the backup link is configured, and the multi-path sub-flows are allocated to complete the optimized allocation of data flows.

[0192] For fault monitoring, a set of fault data monitored by the network sensor is shown in Table 4 below:

[0193] Table 3 Network performance data table (faults exist)

[0194]

[0195] Compare with the set fault judgment threshold to perform fault detection

[0196] The fault thresholds are:

[0197] Delay threshold: If the link delay exceeds 200ms, the link is considered to be faulty.

[0198] Packet loss rate threshold: When the packet loss rate exceeds 5%, the link is considered unstable.

[0199] Bandwidth utilization threshold: When the throughput is lower than 50% bandwidth utilization, the link performance is considered to be degraded.

[0200] The fault detection process is:

[0201] A to B:

[0202] Delay: 21ms<200ms (normal)

[0203] Packet loss rate: 1.8%<5% (normal)

[0204] Bandwidth utilization: 85 / 95×100≈89.47%>50% (normal)

[0205] B to C:

[0206] Delay: 28ms<200ms (normal)

[0207] Packet loss rate: 0.6%<5% (normal)

[0208] Bandwidth utilization: 130 / 145×100≈89.66%>50% (normal)

[0209] C to D:

[0210] Delay: 539ms>200ms (possible failure)

[0211] Packet loss rate: 1.2%<5% (normal)

[0212] Bandwidth utilization: 98 / 110×100≈89.09%>50% (normal)

[0213] D to E:

[0214] Delay: 45ms<200ms (normal)

[0215] Packet loss rate: 3.0%<5% (normal)

[0216] Bandwidth utilization: 52 / 78×100≈66.67%>50% (normal)

[0217] A to E:

[0218] Delay: 56ms<200ms (normal)

[0219] Packet loss rate: 0.9%<5% (normal)

[0220] Bandwidth utilization: 68 / 85×100≈80>50% (normal)

[0221] According to the above analysis, only the delay of the link from C to D exceeds 200ms, so this link may be faulty. Other links do not exceed the fault threshold, so they are considered to be working normally.

[0222] The MPTCP protocol is used to switch the fault line, mark and disable the C to D link, and switch to the backup link D to E, update the topology map and traffic distribution, and generate a fault alarm to notify the network management personnel or the automated fault handling module to repair the fault.

[0223] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A cross-departmental collaborative multi-link communication method, characterized in that: include, Through monitoring equipment, network performance data is collected from links in each department to build a cross-department network topology map; Based on the cross-department network topology, the graph neural network algorithm is used for training to generate cross-department link scheduling strategies; Based on the cross-department link scheduling strategy, link redundancy configuration of multi-path TCP protocol is performed to optimize link data flow distribution; Monitor link failures in real time, switch data streams to backup links based on multipath TCP protocol for failed links, and update cross-department network topology maps to locate and recover from faults; The monitoring device is used to collect network performance data from links in each department and construct a cross-department network topology map. The specific steps are: Deploy network monitoring equipment and configure it to collect network performance data for each link; Preprocess the collected network performance data; transmitting the pre-processed network performance data to a central data processing unit; Build a cross-department network topology map based on the collected network performance data; Through visualization, the cross-department network topology is graphically displayed; The cross-department network topology diagram is trained using a graph neural network algorithm to generate a cross-department link scheduling strategy. The specific steps are: According to the cross-department network topology graph, the input of the graph neural network is defined as the node feature matrix of the graph and the adjacency matrix , the output is defined as the cross-department link scheduling strategy ; Select the GCN-based graph neural network model to build the graph model; Introducing high-order graph convolution, enhancing the learning ability of the graph model through multi-order propagation of the weighted adjacency matrix, and using multi-scale information fusion to adjust the adjacency matrix; Introducing an adaptive adjacency matrix to adaptively adjust the adjacency matrix structure based on real-time network status; After multiple layers of graph convolution operations, node feature representation is obtained, and the node representation is mapped to the cross-department link scheduling strategy through the fully connected layer; Design loss function by combining network performance and bandwidth allocation optimization; Use the Adam optimizer to optimize the loss function and train the graph neural network model; Complete model training and generate cross-department link scheduling strategies; the optimization of link data flow allocation, the specific steps are: Cross-department link scheduling strategy based on graph neural network generation , and the real-time network performance data of the link, dynamically adjust the allocation of link data flow; According to the allocated traffic of the link, the traffic ratio of the main link and the backup link is configured, and the multi-path sub-flows are allocated.

2. The cross-departmental collaborative multi-link communication method according to claim 1, characterized in that: The cross-department network topology diagram is trained using a graph neural network algorithm to generate a cross-department link scheduling strategy. The specific steps are: High-order graph convolution is introduced to enhance the learning ability of the graph model through multi-order propagation of the weighted adjacency matrix, and multi-scale information fusion is used to adjust the adjacency matrix. The expression is: ； in, is the node representation after graph convolution operation, is the node representation of the previous layer, is the normalized adjacency matrix Power, is the weight matrix of each order convolution, is the ReLU activation function, is the layer coefficient, is the maximum order in high-order graph convolution, is the power coefficient of the adjacency matrix; An adaptive adjacency matrix is ​​introduced to adaptively adjust the adjacency matrix structure based on the real-time network status. The expression is: ； in, is the adjacency matrix after adaptive adjustment, is the original adjacency matrix, is the degree matrix, It is a dynamic adjustment item based on real-time network status. is the Frobenius norm, is the normalization function; Combining network performance and bandwidth allocation optimization, the loss function is designed and expressed as: ； in, is the loss function of bandwidth utilization, is the delay loss function, is the packet loss rate loss function, is the weight coefficient of the bandwidth utilization loss function, is the weight coefficient of the delay loss function, is the weight coefficient of the packet loss rate loss function, Scheduling strategy for cross-department links.

3. The cross-departmental collaborative multi-link communication method according to claim 2, characterized in that: The link redundancy configuration of the multi-path TCP protocol is performed according to the cross-department link scheduling strategy, and the specific steps are: Enable MPTCP support on both client and server side; Initialize the sub-flow parameters of MPTCP; Based on cross-department link scheduling strategy , distinguish between the main path and the backup path, and complete the sub-flow link allocation.

4. The cross-departmental collaborative multi-link communication method according to claim 3, characterized in that: The optimization link data flow allocation comprises the following specific steps: Cross-department link scheduling strategy based on graph neural network generation , and the real-time network performance data of the link, dynamically adjust the allocation of link data flow, the expression is: ； in, For Link The distribution flow, For Link The scheduling strategy, is the link priority adjustment factor, For Link performance indicators, is the number of links involved in the transmission, For the link that is currently distributing traffic, The number of links involved in the transmission The index coefficient of .

5. The cross-departmental collaborative multi-link communication method according to claim 4, characterized in that: The real-time monitoring of link failures and switching of data streams to backup links according to the multipath TCP protocol for failed links are as follows: Using monitoring equipment to continuously collect link performance indicators and transmit the link performance indicators to a central data processing unit; Set the fault judgment threshold according to the link performance indicators; According to the link performance indicators collected in real time, the system compares them with the set fault judgment threshold, performs fault detection, and marks the faulty link; For failed links, MPTCP is used to switch to backup links. After the backup link is activated, the traffic distribution data of each link is recalculated according to the new network topology, and the data flow distribution is adjusted according to the real-time situation of the link.

6. The cross-departmental collaborative multi-link communication method according to claim 5, characterized in that: The updating of the cross-department network topology map to locate and restore the fault is carried out in the following specific steps: Update the cross-department network topology map based on the links marked as faulty; Update the graphical display of cross-department network topology through visualization; Generate fault alarms to notify network management personnel and automated fault handling units to perform fault repairs; After the link failure is restored, the restored link is reintroduced into the network and data flow distribution is adjusted according to the link performance status.

7. An inter-departmental collaborative multi-link communication system, based on the inter-departmental collaborative multi-link communication method according to any one of claims 1 to 6, characterized in that: Including, data acquisition module, model building module, link configuration module and fault handling module; The data collection module is used to collect network performance data from links of each department through monitoring equipment and construct a cross-department network topology map; The model building module is used to generate a cross-department link scheduling strategy based on the cross-department network topology diagram using a graph neural network algorithm for training; The link configuration module is used to perform link redundancy configuration of the multi-path TCP protocol according to the cross-department link scheduling strategy and optimize the link data flow distribution; The fault processing module is used to monitor link failures in real time, switch data streams to backup links according to the multi-path TCP protocol for failed links, and update the cross-department network topology map to locate and recover from the failures.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the cross-department collaborative multi-link communication method described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the cross-department collaborative multi-link communication method described in any one of claims 1 to 6 are implemented.

Images