Switch gateway control method and system based on artificial intelligence

Abstract

The present invention provides a switch gateway control method and system based on artificial intelligence, which constructs an adjustable network flow model by simulating and parsing the target switch gateway node and simulated data packets in the candidate network flow data. The network flow model can not only accurately reflect the data transmission path in the network and the network segment where the target switch gateway node is located, but also supports the editing and coverage of the target switch gateway node or the simulated data packet to generate iterative network flow data. Furthermore, by simulating and analyzing the flow of the network flow model, predicting the network flow trend, and dynamically optimizing the data packet transmission path, efficient control and optimization of the target switch gateway node is achieved, thereby significantly improving the intelligent level of network management, enhancing the ability to regulate network traffic, and ensuring the stability and efficiency of network operation.

Description

Technical Field

[0001] The present invention relates to the field of communication technology, and in particular to an artificial intelligence-based switch gateway control method and system. Background Art

[0002] In today's era of rapid development of information technology, the network has become an indispensable part of people's daily life and work. As a key node in the network, the performance and stability of the switch gateway directly affect the operating efficiency and quality of the entire network. However, traditional switch gateway control methods often rely on manual configuration and static rules, which are difficult to adapt to the dynamic changes and complex requirements of the network environment.

[0003] With the continuous development of artificial intelligence technology, it has become a trend to apply it to switch gateway control. Artificial intelligence technology can achieve intelligent prediction and optimization of network traffic through deep learning and analysis of network flow data, thereby improving the control efficiency and accuracy of switch gateways. However, the existing switch gateway control methods based on artificial intelligence still have some shortcomings. For example, these methods often only focus on the overall trend of network traffic, but ignore the transmission path of specific data packets in the network and the status of the target switch gateway node, resulting in insufficient precision and accuracy in the control effect. Summary of the invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, an embodiment of the present invention provides a switch gateway control method based on artificial intelligence, the method comprising:

[0005] Acquire candidate network flow state data including a target switch gateway node and a simulated data message, and perform simulated data message parsing on the candidate network flow state data to generate a transmission path of the simulated data message in the candidate network flow state data;

[0006] According to the transmission path of the simulated data message in the candidate network flow state data, remove the simulated data message in the candidate network flow state data, and generate reference network flow state data corresponding to the candidate network flow state data;

[0007] Performing network segment detection on the target switch gateway node in the reference network flow state data to generate a network segment where the target switch gateway node in the reference network flow state data is located;

[0008] Generate a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data, and the network segment where the target switch gateway node is located; wherein the simulated data message and the target switch gateway node in the network flow model are in an adjustable state, and the network flow model is used to generate iterative network flow data after editing and overwriting the target switch gateway node or the simulated data message;

[0009] Traffic simulation and analysis are performed on the network flow model corresponding to the candidate network flow data to predict the corresponding network traffic trend. Based on the network traffic trend prediction result, the data packet transmission path in the network flow model is dynamically optimized to achieve control optimization of the target switch gateway node.

[0010] On the other hand, an embodiment of the present invention also provides an artificial intelligence-based switch gateway control system, including a processor and a machine-readable storage medium, wherein the machine-readable storage medium is connected to the processor, the machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0011] Based on the above aspects, the embodiment of the present application constructs an adjustable network flow model by simulating and parsing the target switch gateway node and simulated data packets in the candidate network flow data. The network flow model can not only accurately reflect the data transmission path in the network and the network segment where the target switch gateway node is located, but also supports the editing and overwriting of the target switch gateway node or the simulated data packet to generate iterative network flow data. Furthermore, by simulating and analyzing the flow of the network flow model, predicting the network flow trend, and dynamically optimizing the data packet transmission path, efficient control and optimization of the target switch gateway node is achieved, thereby significantly improving the intelligent level of network management, enhancing the ability to regulate network traffic, and ensuring the stability and efficiency of network operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic diagram of the execution flow of the switch gateway control method based on artificial intelligence provided by an embodiment of the present invention.

[0013] Figure 2 It is a schematic diagram of the hardware architecture of the switch gateway control system based on artificial intelligence provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0014] The present invention will be described in detail below with reference to the accompanying drawings. Figure 1The figure is a flow chart of a switch gateway control method based on artificial intelligence provided by an embodiment of the present invention. The switch gateway control method based on artificial intelligence is introduced in detail below.

[0015] Step S110, obtaining candidate network flow data including a target switch gateway node and a simulated data message, and performing simulated data message parsing on the candidate network flow data to generate a transmission path of the simulated data message in the candidate network flow data.

[0016] In this embodiment, in a large enterprise network environment, the server manages network flow data. The enterprise network structure is complex, including subnets of multiple departments, such as R&D subnet, sales subnet, financial subnet, etc., which are interconnected through switch gateway nodes to achieve data interaction.

[0017] The server first obtains candidate network flow data including the target switch gateway node and simulated data packets. The target switch gateway node may be a key node connecting the R&D subnet and the enterprise core network. This node is responsible for forwarding a large amount of data traffic generated by the R&D department, including code transmission, test data feedback, etc. The simulated data packets simulate various data interactions generated by R&D personnel in their daily work, such as code submission messages sent from the R&D personnel's work computers to the code warehouse server, or test result messages fed back to the R&D personnel from the test server.

[0018] In order to perform simulated data message parsing on the candidate network flow data, the server will perform a series of operations. The server first performs message parsing on the candidate network flow data. During this process, the server will check each data unit and identify the simulated data message according to the specific protocol format. For common network protocols, such as TCP / IP protocol, the server will distinguish different types of messages based on the identification field of the protocol header. For example, by checking the protocol number field in the IP protocol header, if it is 6, it means that this is a TCP message, and if it is 17, it means that it is a UDP message. After determining that it is a TCP message, the server further parses the source port, destination port, sequence number, acknowledgment number and other information in the TCP header to accurately determine that this is a simulated data message.

[0019] Next, the server analyzes the transmission trajectory of the simulated data message. Taking the example of a code submission message sent by a developer to the code repository server, the server queries the routing table and forwarding table information in network devices (such as routers and switches) to determine that the message starts from the developer's work computer, first reaches the access switch of the subnet, then passes through the target switch gateway node, and then passes through a series of switches and routers in the enterprise core network, and finally reaches the code repository server. This complete path of all network nodes from the source to the destination is the transmission path of the simulated data message in the candidate network flow data. Throughout the process, the server records the relevant information of each node in detail, such as the node's IP address, MAC address, port number, etc., in order to accurately describe the transmission path.

[0020] Step S120: removing the simulated data packets in the candidate network flow data according to the transmission path of the simulated data packets in the candidate network flow data, and generating reference network flow data corresponding to the candidate network flow data.

[0021] In this embodiment, the server removes the simulated data packets in the candidate network flow data according to the transmission path of the previously generated simulated data packets in the candidate network flow data. In an enterprise network environment, this process is similar to clearing specific data records in network flow records.

[0022] The server first removes the simulated data packets from the candidate network flow data according to the transmission path of the simulated data packets in the candidate network flow data. Since the server already knows the transmission path of the simulated data packets, it can locate the records related to these simulated data packets in the database or cache storing the network flow data and delete them. For example, in a database storing network traffic logs, the server writes SQL query statements to delete the corresponding records based on the source IP address, destination IP address, port number, and transmission time of the simulated data packets, thereby generating network flow data without packets.

[0023] Then, the server optimizes the network segment from which the simulated data packets are removed from the network flow data without packets to generate reference network flow data corresponding to the candidate network flow data. In the enterprise network, it is assumed that the network segment from which the simulated data packets are removed is part of the link connecting the R&D subnet and the enterprise core network. The server first determines the adjacent network segments of this network segment, for example, other links within the R&D subnet and other links in the enterprise core network connected to the target switch gateway node are adjacent network segments.

[0024] The server obtains the flow data characteristics of adjacent network segments. This includes the traffic load of adjacent network segments. For example, by querying the port traffic statistics of network devices, it is found that a certain link in the R&D subnet has a heavy traffic load, mainly because R&D personnel frequently download large development toolkits; it also includes the type distribution of traffic, such as a certain link is mainly video conferencing traffic (which may be generated when the R&D team collaborates remotely), and the network delay, which is obtained by sending probe messages to devices in the network and measuring the round-trip time.

[0025] Based on these flow data characteristics, the server reconstructs the network segments that exclude the simulated data packets in the candidate network flow data. For example, if the adjacent network segment has a heavy traffic load and a large amount of burst traffic, the server may adjust the cache strategy of the target switch gateway node in this network segment and increase the cache size to cope with the burst traffic. At the same time, if it is found that the video conferencing traffic in the adjacent network segment has high requirements for network latency, the server may optimize the routing strategy and assign a higher priority to the video conferencing traffic to avoid the impact of network congestion on it. After such optimization operations, the reference network flow data corresponding to the candidate network flow data is generated.

[0026] Step S130 , performing network segment detection on the target switch gateway node in the reference network flow data, and generating a network segment where the target switch gateway node in the reference network flow data is located.

[0027] In this embodiment, the server performs network segment detection on the target switch gateway node in the reference network flow data to determine the network segment where the target switch gateway node is located.

[0028] First, the server obtains the member construction feature of the candidate network flow state data, which indicates the target switch gateway node in the candidate network flow state data. In an enterprise network environment, the member construction feature may include information such as the device model of the target switch gateway node, the number and type of connected ports, and the physical connection relationship with other network devices (such as other switches, routers, servers, etc.). For example, the target switch gateway node is a specific model of enterprise-class switch with 24 Gigabit Ethernet ports, of which ports 1 to 10 are connected to devices in the R&D subnet, ports 11 to 20 are connected to other switches in the enterprise core network, and ports 21 to 24 are reserved for expansion or management purposes.

[0029] Then, based on this member construction feature, the server searches for the target switch gateway node in the reference network flow data. In the database or data structure storing the reference network flow data, the server finds all records related to the target switch gateway node by matching the device model, port connection relationship and other information of the target switch gateway node, thereby generating the node mapping information of the target switch gateway node in the reference network flow data. This node mapping information describes in detail the association relationship between the target switch gateway node and other network elements (such as other switches, servers, links, etc.). For example, the node mapping information indicates that the link bandwidth between the target switch gateway node and some key servers in the R&D subnet is 1Gbps, and the link delay between the target switch gateway node and another core switch in the enterprise core network is 1ms, etc.

[0030] Then, the server generates the bounded domain of the target switch gateway node based on the node mapping information of the target switch gateway node in the reference network flow data. This bounded domain is used to describe the network segment where the target switch gateway node in the reference network flow data is located. In the enterprise network, the server determines the devices directly connected to the target switch gateway node and the devices reachable through one hop or multiple hops based on the node mapping information, and defines the area where these devices are located as the bounded domain. For example, the bounded domain includes the target switch gateway node, 10 servers in the R&D subnet directly connected to it, two routers in the enterprise core network reachable through one hop, and all links connecting these devices.

[0031] After searching for the target switch gateway node in the reference network flow data based on the member construction characteristics and generating the node mapping information of the target switch gateway node in the reference network flow data, the server will also analyze the target switch gateway node to generate the penetration area of ​​the target switch gateway node in the reference network flow data. The penetration area represents the scope of influence of the target switch gateway node on the surrounding network environment. For example, by analyzing factors such as the traffic forwarding mode and broadcast domain range of the target switch gateway node, the server determines that the network devices and links within a certain range with the target switch gateway node as the center are affected by the traffic forwarding and control strategy of the target switch gateway node. This range is the penetration area.

[0032] Finally, based on the node mapping information of the target switch gateway node in the reference network flow data and the penetration area of ​​the target switch gateway node in the reference network flow data, the bounding domain of the target switch gateway node is generated. This makes the bounding domain more accurately reflect the actual location and influence range of the target switch gateway node in the network, providing a more accurate basis for subsequent operations.

[0033] Step S140, generating a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data, and the network segment where the target switch gateway node is located. The simulated data message and the target switch gateway node in the network flow model are in an adjustable state, and the network flow model is used to generate iterative network flow data after editing and overwriting the target switch gateway node or the simulated data message.

[0034] In this embodiment, the server generates a network flow model corresponding to the candidate network flow data based on the transmission path of the simulated data message in the candidate network flow data, the reference network flow data, and the network segment where the target switch gateway node is located.

[0035] First, the server converts the simulated data packets in the candidate network flow data according to the transmission path of the simulated data packets in the candidate network flow data, and generates an adjustable message corresponding to the simulated data packet. In the enterprise network environment, taking the code submission message of the R&D personnel as an example, the server converts the original code submission message. The original message contains information such as the source IP address (the IP of the R&D personnel's work computer), the destination IP address (the IP of the code warehouse server), the port number, the protocol type, etc. During the conversion process, in addition to retaining these key information, the server also adds some adjustable parameters. For example, an adjustable traffic weight parameter is added to simulate different traffic priority situations in the network flow model; an adjustable transmission delay parameter is also added to simulate network congestion or transmission delay changes in different network environments.

[0036] Next, based on the network segment where the target switch gateway node is located, the server obtains the identification map corresponding to the target switch gateway node. The identification map is a detailed description of the network segment where the target switch gateway node is located. In an enterprise network, this identification map may be presented in the form of a graphic or data structure. It contains detailed information about all devices (such as servers, switches, routers, etc.) in the network segment where the target switch gateway node is located, including the device's IP address, MAC address, device type, connection relationship, etc., as well as link information, such as link bandwidth, link delay, etc. For example, the identification map lists in detail that the link bandwidth between the target switch gateway node and the server in the R&D subnet is 1Gbps, and the link delay between the target switch gateway node and the router in the enterprise core network is 1ms, and identifies the hierarchical relationship and functional role of each device in the network segment.

[0037] Then, the server fuses the adjustable message, the reference network flow data and the identification map to generate a network flow model corresponding to the candidate network flow data. During the fusion process, the server integrates the information of the adjustable message with the network topology structure, device configuration information, etc. in the reference network flow data. For example, the source IP address and destination IP address in the adjustable message are matched with the network topology structure in the reference network flow data to determine the actual transmission path in the network. At the same time, guided by the identification map, the position and relationship of each element in the network flow model are determined. For example, according to the device connection relationship in the identification map, the transmission path of the adjustable message is associated with the actual link in the network, and according to the device function role in the identification map, the processing method of the adjustable message on different devices is set. The simulated data message and the target switch gateway node in the network flow model generated in this way are in an adjustable state, and the network flow model can be used to generate iterative network flow data after editing and overwriting the target switch gateway node or the simulated data message.

[0038] After generating the defined domain of the target switch gateway node based on the node mapping information of the target switch gateway node in the reference network flow data, the server divides the reference network flow data based on the defined domain to generate an identification map of the target switch gateway node. For example, the server extracts the devices and links in the defined domain from the reference network flow data separately, classifies and organizes them according to factors such as the type of device and connection relationship, and constructs an identification map. This identification map describes in detail the network segment where the target switch gateway node is located in the reference network flow data, providing an accurate basis for subsequent fusion operations.

[0039] After dividing the reference network flow data based on the defined domain and generating the identification map of the target switch gateway node, the server analyzes the connection status of the target switch gateway node based on the identification map and generates the key connection points in the identification map. In an enterprise network, the connection status includes factors such as the bandwidth utilization of the link, the packet loss rate of the link, and the stability of the connection. The server determines the key connection points by comparing the changes in the connection status of each connection point. For example, if it is found that the bandwidth utilization of the link connecting the target switch gateway node and a key server in the R&D subnet fluctuates greatly over a period of time, and the connection status change exceeds the set value compared with other connection points, then this connection point is determined as a key connection point.

[0040] Next, link extension is performed based on the key connection points to generate a connection domain of the target switch gateway node. The server extends the key connection points to generate multiple extension points. An extension point is a connection point whose link status correlation with the key connection point is greater than the set correlation. For example, for the above-mentioned connection point with large fluctuations in bandwidth utilization, the server searches for other links and devices related to it and finds that the devices connected to some ports on another switch connected to the link have similar traffic patterns to the link. The connection points corresponding to these ports are determined as extension points. Then, the server links the key connection points with multiple extension points to generate a connection domain of the target switch gateway node. This connection domain is used to describe the connection status of the target switch gateway node.

[0041] Finally, based on the connection domain, the target identification map of the target switch gateway node is generated. The target identification map is used to describe the network segment where the target switch gateway node is located in the reference network flow state data. Based on the original identification map, the target identification map further highlights the connection status of the target switch gateway node, providing more comprehensive information for generating the network flow state model.

[0042] Based on the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the target identification map, the network flow model corresponding to the candidate network flow data is generated. In this process, the server comprehensively integrates the transmission path information of the simulated data message, the network structure information in the reference network flow data and the target switch gateway node connection status information in the target identification map to construct a complete network flow model. The various elements in this model are interrelated and adjustable, providing a basis for subsequent traffic simulation and analysis.

[0043] Step S150, perform traffic simulation and analysis on the network flow model corresponding to the candidate network flow data, predict the corresponding network traffic trend, and dynamically optimize the data packet transmission path in the network flow model based on the network traffic trend prediction result to achieve control optimization of the target switch gateway node.

[0044] In this embodiment, the server performs traffic simulation and analysis on the network flow model corresponding to the candidate network flow data to predict the corresponding network traffic trend, and dynamically optimizes the data packet transmission path in the network flow model based on the network traffic trend prediction results, thereby achieving control optimization of the target switch gateway node.

[0045] In an enterprise network environment, the server first simulates the traffic of the network flow model. The network flow model contains information such as adjustable messages, target switch gateway nodes, and network topology. The server simulates the transmission process of data messages in the network according to the settings in the network flow model. For example, for the code submission message of the R&D personnel (adjustable message in the network flow model), the server simulates the process of this message starting from the R&D personnel's work computer, passing through the target switch gateway node, and finally reaching the code warehouse server according to the network topology in the model (including the target switch gateway node and the connection relationship between other switches, routers, servers, etc.). During the simulation process, the server considers various parameters in the model, such as the traffic weight of the adjustable message, transmission delay and other parameters, as well as factors such as link bandwidth and device processing capacity in the network topology.

[0046] Based on the traffic simulation, the server performs traffic analysis to predict the corresponding network traffic trend. The server predicts the network traffic trend in the future by analyzing factors such as the changes in traffic load of each network node, the changes in link utilization, the location and frequency of network congestion during the simulation process. For example, the server finds that as the working hours of R&D personnel go by, there will be a peak in the traffic from the R&D subnet to the code warehouse server between 10 am and 12 pm every day, causing the link utilization between the target switch gateway node and the code warehouse server to reach more than 80%, and there is a trend of gradual increase, which may cause network congestion in the future.

[0047] According to the network traffic trend prediction results, the server dynamically optimizes the data message transmission path in the network flow model to achieve control optimization of the target switch gateway node. In response to the above predicted network congestion, the server adjusts the data message transmission path in the network flow model. For example, the server can modify the routing strategy of the adjustable message (code submission message) so that it avoids the high-load link between the target switch gateway node and the code warehouse server during peak hours and chooses to transmit through other links in the enterprise core network. This can not only avoid congestion of the link where the target switch gateway node is located, but also improve the transmission efficiency of the entire network and achieve control optimization of the target switch gateway node. At the same time, the server can also adjust the traffic forwarding strategy of the target switch gateway node based on other network traffic trend prediction results, such as the low utilization of certain links, and guide part of the traffic to these low-utilization links to achieve balanced utilization of network resources.

[0048] Based on the above steps, the embodiment of the present application constructs an adjustable network flow model by simulating and parsing the target switch gateway node and simulated data packets in the candidate network flow data. The network flow model can not only accurately reflect the data transmission path in the network and the network segment where the target switch gateway node is located, but also supports the editing and overwriting of the target switch gateway node or the simulated data packet to generate iterative network flow data. Furthermore, by simulating and analyzing the flow of the network flow model, predicting the network flow trend, and dynamically optimizing the data packet transmission path, efficient control and optimization of the target switch gateway node is achieved, thereby significantly improving the intelligent level of network management, enhancing the ability to regulate network traffic, and ensuring the stability and efficiency of network operation.

[0049] In a possible implementation, step S110 includes:

[0050] Step S111, performing message parsing on the candidate network flow data to generate simulated data messages of the candidate network flow data.

[0051] Step S112, performing transmission trajectory analysis on the simulated data message to generate a transmission path of the simulated data message in the candidate network flow data.

[0052] In this embodiment, the server first performs message parsing on the candidate network flow data to generate a simulated data message. The enterprise network contains many devices and complex interactive data, such as data exchanges between various devices in the R&D subnet and devices in other department subnets. The server checks each data unit in the network flow data and identifies the simulated data message based on the relevant knowledge of the network protocol. For data based on the TCP / IP protocol, the server checks the protocol number and other identification fields in the IP protocol header. If it is 6 (indicating TCP) or 17 (indicating UDP), the corresponding protocol header information is further parsed, such as the source port, destination port, sequence number, acknowledgment number and other contents in the TCP header. For example, when it is found that the source port of a data packet is a random high port, the destination port is 80 and the protocol number is 6, it can be determined that this is a simulated data message sent from a source end to a network server (such as a Web server within the enterprise).

[0053] After that, the server analyzes the transmission trajectory of the simulated data message to generate its transmission path in the candidate network flow data. Taking the example of R&D personnel sending code from their work computers to the code warehouse server, the server queries the routing table and forwarding table information in the network equipment (routers, switches). Starting from the R&D personnel's work computer, the data packet first reaches the access switch of the subnet where it is located, which is determined based on the topology of the subnet and the MAC address learning function of the access switch. Then, by querying the forwarding table of the target switch gateway node (the key node connecting the R&D subnet and the enterprise core network), it is found that the data packet will pass through this node. Then, in the enterprise core network, according to the routing table of the core switch, the data packet will pass through a series of intermediate switches and routers and finally reach the code warehouse server. The server records in detail the information of all network nodes passed from the source to the destination, including the node's IP address, MAC address, port number, etc., so that the transmission path of the simulated data message in the candidate network flow data is completely determined.

[0054] In a possible implementation, step S120 includes:

[0055] Step S121, according to the transmission path of the simulated data message in the candidate network flow data, the simulated data message in the candidate network flow data is removed to generate message-free network flow data.

[0056] Step S122, optimizing the network segment in the network flow data without packets, excluding the simulated data packets, and generating reference network flow data corresponding to the candidate network flow data.

[0057] In a possible implementation, step S122 includes:

[0058] Step S1221, in the message-free network flow data, determine the adjacent network segment of the network segment from which the simulated data message is removed, and obtain the flow data characteristics of the adjacent network segment.

[0059] Step S1222: reconstruct the network segment of the candidate network flow data excluding the simulated data message according to the flow data characteristics, and generate reference network flow data corresponding to the candidate network flow data.

[0060] In this embodiment, in an enterprise network environment, the server removes the simulated data packets in the candidate network flow data according to the transmission path of the simulated data packets in the candidate network flow data to generate reference network flow data.

[0061] First, based on the transmission path of the simulated data message in the candidate network flow data, the simulated data message in the candidate network flow data is eliminated to generate message-free network flow data. In the enterprise network, the server has already clarified the transmission path of the simulated data message (such as the message that the R&D personnel send the code to the code warehouse server), which involves multiple network nodes and links. The server may store network flow data in a database or cache. It will locate the records related to these simulated data messages in the storage structure based on the key information such as the source IP address, destination IP address, port number and transmission time of the simulated data message in the transmission path. Then, the server performs a deletion operation, just like deleting a specific record from a log file. For example, by writing a specific query statement in the database, all records related to the simulated data message are deleted to obtain message-free network flow data.

[0062] Next, the network segment in which the simulated data packets are removed from the packet-free network flow data is optimized to generate reference network flow data corresponding to the candidate network flow data.

[0063] In the flow data of the network without packets, the adjacent network segments of the network segment where the simulated data packets are removed are determined, and the flow data characteristics of the adjacent network segments are obtained. Taking the network segment connecting the R&D subnet and the enterprise core network mentioned above (here, the network segment where the simulated data packets are removed) as an example, its adjacent network segments include other links inside the R&D subnet except for the part directly related to the target transmission path, and other links in the enterprise core network connected to the target switch gateway node. The server obtains the flow data characteristics of the adjacent network segments, and this process involves multiple inspections. For the traffic load, the server queries the port traffic statistics of network devices (such as switches and routers). For example, it is found that the traffic load of a certain link inside the R&D subnet is heavy because the R&D personnel frequently download large development toolkits, causing the link to run at close to full load in certain periods. For the traffic type distribution, the server analyzes the data type transmitted by the port. For example, it is found that a certain link is mainly video conferencing traffic, which may be generated when the R&D team collaborates remotely. At the same time, the server also measures the network delay, and obtains the accurate delay value by sending detection packets to the devices in the network and measuring the round-trip time.

[0064] Based on these flow data characteristics, the network segments from which the simulated data packets are removed from the candidate network flow data are reconstructed to generate reference network flow data corresponding to the candidate network flow data. For example, if the traffic load of the adjacent network segment is heavy and there is a large amount of burst traffic, and the current network segment from which the simulated data packets are removed was previously configured according to the normal traffic mode. The server will adjust the cache strategy of the target switch gateway node in this network segment and increase the cache size to better cope with burst traffic. If it is found that the adjacent network segment has delay-sensitive video conferencing traffic, the server will optimize the routing strategy. For example, the priority of video conferencing traffic is increased, and its transmission is prioritized when the network is congested to avoid the traffic being affected. By reconstructing the network segment in this way, reference network flow data that meets the requirements is generated.

[0065] In a possible implementation, step S130 includes:

[0066] Step S131, obtaining a member construction feature of the candidate network flow state data, wherein the member construction feature indicates the target switch gateway node in the candidate network flow state data.

[0067] Step S132: searching the target switch gateway node in the reference network flow state data according to the member structure feature, and generating node mapping information of the target switch gateway node in the reference network flow state data.

[0068] Step S133, generating a bounding domain of the target switch gateway node according to the node mapping information of the target switch gateway node in the reference network flow data, wherein the bounding domain is used to describe the network segment where the target switch gateway node in the reference network flow data is located.

[0069] Step S140 includes:

[0070] A network flow model corresponding to the candidate network flow data is generated according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the defined domain.

[0071] In this embodiment, in an enterprise network environment, the server performs an operation of performing network segment detection on a target switch gateway node in the reference network flow data.

[0072] First, the member construction features of the candidate network flow state data are obtained. These member construction features can indicate the target switch gateway node in the candidate network flow state data. In the enterprise network, this target switch gateway node may be a key node connecting the R&D subnet and the enterprise core network. The member construction features contain many aspects of information, such as the device model of the target switch gateway node, which may be an enterprise-level switch produced by a specific manufacturer with specific hardware specifications and functional characteristics. The number and type of its connection ports are also part of the member construction features, such as having 24 Gigabit Ethernet ports. In addition, it also includes the physical connection relationship with other network devices (such as other switches, routers, servers, etc.), such as ports 1-10 connect to devices in the R&D subnet, ports 11-20 connect to other switches in the enterprise core network, and ports 21-24 are reserved for expansion or management purposes. These member construction features are stored in a storage medium accessible to the server in the form of a system record or configuration file, and the server obtains the complete member construction features by reading this information.

[0073] Next, based on these member construction features, the target switch gateway node in the reference network flow state data is searched, thereby generating the node mapping information of the target switch gateway node in the reference network flow state data. The server searches in the database or data structure storing the reference network flow state data. Since the member construction features such as the device model and port connection relationship of the target switch gateway node are already known, the server can find all records related to the target switch gateway node by matching these features. For example, the database table is searched for records whose device model field matches the target switch model and whose connection relationship field matches the known connection relationship. Through this search process, the generated node mapping information describes in detail the association relationship between the target switch gateway node and other network elements (such as other switches, servers, links, etc.). For example, the node mapping information indicates that the link bandwidth between the target switch gateway node and some key servers in the R&D subnet is 1Gbps, and the link delay between the target switch gateway node and another core switch in the enterprise core network is 1ms, etc.

[0074] Then, based on the node mapping information of the target switch gateway node in the reference network flow data, the defined domain of the target switch gateway node is generated. This defined domain is used to describe the network segment where the target switch gateway node in the reference network flow data is located. The server determines the devices directly connected to the target switch gateway node and the devices reachable through one hop or multiple hops based on the node mapping information, and defines the area where these devices are located as the defined domain. In an enterprise network, the defined domain may include the target switch gateway node, 10 servers in the R&D subnet directly connected to it, two routers in the enterprise core network reachable through one hop, and all links connecting these devices. The server accurately determines the scope of the defined domain by analyzing the connection relationship and distance information (such as the number of hops) in the node mapping information.

[0075] Afterwards, a network flow model corresponding to the candidate network flow data is generated based on the transmission path of the simulated data message in the candidate network flow data, the reference network flow data, and the network segment (i.e., the defined domain) where the target switch gateway node is located.

[0076] The server first converts the simulated data packets in the candidate network flow data according to the transmission path of the simulated data packets in the candidate network flow data, and generates an adjustable message corresponding to the simulated data packets. Taking the simulated data packets of the R&D personnel sending the code to the code warehouse server as an example, the original message contains the source IP address (the IP of the R&D personnel's work computer), the destination IP address (the IP of the code warehouse server), the port number, the protocol type and other information. When converting, in addition to retaining these key information, the server also adds adjustable parameters, such as adjustable traffic weight parameters (used to simulate different traffic priority situations) and adjustable transmission delay parameters (used to simulate network congestion or transmission delay changes in different network environments).

[0077] Next, based on the network segment (defined domain) where the target switch gateway node is located, the server obtains the identification map corresponding to the target switch gateway node. This identification map is a detailed description of the network segment where the target switch gateway node is located, which may exist in the form of a graphic or data structure. It contains detailed information on all devices (such as servers, switches, routers, etc.) within the defined domain, including the device's IP address, MAC address, device type, connection relationship, etc., as well as link information, such as link bandwidth, link delay, etc. For example, the identification map lists in detail that the link bandwidth between the target switch gateway node and the server in the R&D subnet is 1Gbps, and the link delay between the target switch gateway node and the router in the enterprise core network is 1ms, and identifies the hierarchical relationship and functional role of each device in the network segment.

[0078] Finally, the server fuses the adjustable message, the reference network flow data and the identification map to generate a network flow model corresponding to the candidate network flow data. During the fusion process, the server integrates the information of the adjustable message with the network topology structure, device configuration information, etc. in the reference network flow data. For example, the source IP address and destination IP address in the adjustable message are matched with the network topology structure in the reference network flow data to determine the actual transmission path in the network. At the same time, guided by the identification map, the position and relationship of each element in the network flow model are determined. For example, according to the device connection relationship in the identification map, the transmission path of the adjustable message is associated with the actual link in the network, and according to the device function role in the identification map, the processing method of the adjustable message on different devices is set. The simulated data message and the target switch gateway node in the network flow model generated in this way are in an adjustable state, and the network flow model can be used to generate iterative network flow data after editing and overwriting the target switch gateway node or the simulated data message.

[0079] In a possible implementation, after step S132, the method further includes:

[0080] The target switch gateway node is analyzed to generate a penetration area of ​​the target switch gateway node in the reference network flow state data.

[0081] Step S133 includes:

[0082] The defined domain of the target switch gateway node is generated according to the node mapping information of the target switch gateway node in the reference network flow state data and the penetration area of ​​the target switch gateway node in the reference network flow state data.

[0083] In a possible implementation manner, after step S133, the method further includes:

[0084] The reference network flow state data is divided based on the defined domain to generate an identification map of the target switch gateway node.

[0085] The identification map is used to describe the network segment where the target switch gateway node is located in the reference network flow data.

[0086] The step of generating a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the defined domain comprises:

[0087] A network flow model corresponding to the candidate network flow data is generated based on the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the identification map.

[0088] In this embodiment, after searching for the target switch gateway node in the reference network flow data based on the member construction characteristics and generating node mapping information, the server starts to analyze the target switch gateway node to generate its penetration area in the reference network flow data.

[0089] In an enterprise network, the target switch gateway node (for example, the node connecting the R&D subnet and the enterprise core network) is a key hub for data transmission. The server analyzes the working mode of the target switch gateway node, including its traffic forwarding rules. For example, the node may determine the forwarding path of the data message based on information such as the source IP address, destination IP address, or port number. If it forwards the data of the R&D subnet to different areas of the enterprise core network according to a specific IP address range, then these related areas are affected by its traffic forwarding strategy.

[0090] The server also checks the broadcast domain range of the target switch gateway node. In this broadcast domain, broadcast messages are forwarded to all connected devices. For example, when a network device sends an ARP (Address Resolution Protocol) request in this broadcast domain, the target switch gateway node broadcasts this request to all connected devices. Based on these traffic forwarding patterns and broadcast domain ranges, the server determines that the network devices and links within a certain range centered on the target switch gateway node are affected by the traffic forwarding and control policies of the target switch gateway node. This range is the penetration area. For example, servers in the R&D subnet directly connected to the target switch gateway node, some routers in the enterprise core network that can be reached through this node, and their connecting links are all within this penetration area.

[0091] Next, a bounded domain of the target switch gateway node is generated according to the node mapping information of the target switch gateway node in the reference network flow state data and the penetration area of ​​the target switch gateway node in the reference network flow state data.

[0092] The node mapping information describes in detail the relationship between the target switch gateway node and other network elements (such as other switches, servers, links, etc.), such as which devices it is directly connected to and the link bandwidth, latency, etc. The penetration area indicates the scope of influence of the target switch gateway node on the surrounding network environment.

[0093] The server combines these two pieces of information to determine the delimited domain. For example, the node mapping information shows that the target switch gateway node is directly connected to certain servers in the R&D subnet, and the penetration area indicates that the data sent by these servers will affect specific routers in the enterprise core network after being forwarded by the target switch gateway node. Then the delimited domain includes not only these directly connected servers and the target switch gateway node itself, but also the links between these servers and the target switch gateway node, as well as the specific routers in the enterprise core network that are affected and their related links. In this way, the delimited domain accurately describes the network segment where the target switch gateway node in the reference network flow data is located.

[0094] After generating a defined domain based on the node mapping information of the target switch gateway node in the reference network flow data, the server divides the reference network flow data based on the defined domain to generate an identification map of the target switch gateway node.

[0095] The server first identifies all network elements contained in the defined domain, including devices (such as switches, servers, routers, etc.) and links. For devices, the server extracts their detailed information, such as the device's IP address, MAC address, device type, etc. For links, the server obtains information such as link bandwidth and link delay.

[0096] For example, within the defined domain, there is a link connecting the R&D subnet and the enterprise core network. The bandwidth of this link obtained by the server is 1Gbps and the latency is 1ms. For the servers in the R&D subnet, the server records its IP address, MAC address, whether it is a file server or an application server, and other type information.

[0097] The server then constructs an identification graph based on this information. This identification graph describes the network segment where the target switch gateway node is located in the reference network flow data in an intuitive way. In the identification graph, devices are represented as nodes, links are represented as lines connecting nodes, and corresponding information such as device type, link bandwidth, etc. are marked on the nodes and lines. For example, the target switch gateway node is in the center of the graph, the link connecting it is marked with the bandwidth value, and the connected server node is marked with its function type and other information.

[0098] Next, based on the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the identification map, a network flow model corresponding to the candidate network flow data is generated.

[0099] The server converts the simulated data packets in the candidate network flow data according to the transmission path of the simulated data packets in the candidate network flow data, and generates an adjustable message corresponding to the simulated data packets. Taking the simulated data packets sent by R&D personnel to the code warehouse server as an example, the original message contains information such as the source IP address, destination IP address, port number, protocol type, etc. The server retains this key information during conversion and adds adjustable parameters, such as adjustable traffic weight parameters (used to simulate different traffic priority situations) and adjustable transmission delay parameters (used to simulate network congestion or transmission delay changes in different network environments).

[0100] Next, the server uses the network topology, device configuration and other information in the reference network flow data. For example, the switch configuration information (such as VLAN settings) and the router routing table information in the reference network flow data reflect the basic architecture and operation rules of the network.

[0101] Then, it is combined with the identification map of the target switch gateway node. The identification map describes in detail the equipment and link conditions of the network segment where the target switch gateway node is located. The server integrates the information of the adjustable message with the network topology and device configuration information in the reference network flow data, and determines the position and relationship of each element in the network flow model with the guidance of the identification map.

[0102] For example, the source IP address and destination IP address in the adjustable message are matched with the network topology in the reference network flow data to determine the actual transmission path in the network. According to the device connection relationship in the identification map, the transmission path of the adjustable message is associated with the actual link in the network, and according to the device function role in the identification map, the processing method of the adjustable message on different devices is set. The simulated data message and the target switch gateway node in the network flow model generated in this way are in an adjustable state. The network flow model can be used to generate iterative network flow data after editing and overwriting the target switch gateway node or the simulated data message.

[0103] In a possible implementation manner, after dividing the reference network flow state data based on the defined domain and generating an identification map of the target switch gateway node, the method further includes:

[0104] According to the identification map, the connection status of the target switch gateway node is analyzed to generate a key connection point in the identification map, wherein the key connection point is a connection point whose connection status changes by more than a set value compared with other connection points.

[0105] Link expansion is performed according to the key connection point to generate a connection domain of the target switch gateway node, where the connection domain is used to describe the connection status of the target switch gateway node.

[0106] A target identification map of the target switch gateway node is generated according to the connection domain, and the target identification map is used to describe the network segment where the target switch gateway node is located in the reference network flow state data.

[0107] The step of generating a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the identification graph comprises:

[0108] A network flow model corresponding to the candidate network flow data is generated based on the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the target identification map.

[0109] The link expansion is performed according to the key connection point to generate the connection domain of the target switch gateway node, including:

[0110] The key connection point is extended to generate a plurality of extension points, wherein the extension point is a connection point whose link state correlation with the key connection point is greater than a set correlation.

[0111] The key connection point is linked to the multiple extension points to generate a connection domain of the target switch gateway node.

[0112] In this embodiment, after dividing the reference network flow data based on the defined domain and generating an identification map of the target switch gateway node, the server analyzes the connection status of the target switch gateway node based on the identification map to generate key connection points in the identification map.

[0113] In an enterprise network, the connection status covers many factors. For example, the server will check the bandwidth utilization of the link. For the link connecting the target switch gateway node and a server in the R&D subnet, the average bandwidth utilization of the link within a certain period of time (such as the past hour) is calculated by querying the port traffic statistics of the network device (such as the switch). At the same time, the packet loss rate of the link is considered, which is obtained through the statistics of the transmitted data packets by the network device, as well as the stability of the connection, which can be determined by monitoring the number of link connection interruptions or the results of the link quality detection tool.

[0114] The server will compare the connection status changes of each connection point in the identification map. Assume that in the identification map, the target switch gateway node is connected to multiple devices, and the bandwidth utilization of the link connected to a test server rises sharply from 30% to 80% in a short period of time (such as 10 minutes), while the bandwidth utilization of other links changes relatively little, and the packet loss rate of this link also increases significantly, and the connection stability fluctuates. Compared with other connection points, the connection status changes exceed the set value (for example, the connection status indicator change is set to exceed the set value if it exceeds 30%), then this connection point (the connection point between the target switch gateway node and the test server) is determined as a critical connection point.

[0115] Then, link extension is performed based on the key connection points to generate the connection domain of the target switch gateway node. First, the key connection points are extended to generate multiple extension points. The server determines the extension point by analyzing factors such as network traffic characteristics and device association related to the key connection points. Taking the previously determined key connection point between the target switch gateway node and the test server as an example, the server checks the traffic patterns of other devices or links connected to the test server, the communication relationship between devices, etc. If another device in the same subnet as the test server is found, it has frequent traffic interactions with the test server, and shares some network resources with the link of the key connection point during traffic transmission (such as sharing a physical link or passing through the same switch port), this indicates that the link state association between the device and the key connection point is greater than the set association (for example, by setting the association to share a certain proportion of network resources or have a specific frequency of traffic interaction), then the connection point corresponding to this device is determined as an extension point. In this way, the server can find multiple extension points.

[0116] After determining the extension points, the server links the key connection points with multiple extension points to generate the connection domain of the target switch gateway node. For example, the target switch gateway node is logically connected to the key connection points of the test server and the extension points corresponding to other devices associated with the test server. This connection is not only the connection concept of the physical link, but also includes the logical association relationship in the network flow model. The connection domain covers the key connection points and their extension points and the relationship between them, thereby describing the connection status of the target switch gateway node. In this connection domain, factors such as the flow direction of network traffic and the mutual influence between devices are comprehensively considered. For example, traffic changes at key connection points may affect the network performance of other related devices through extension points.

[0117] Based on the connection domain, a target identification map of the target switch gateway node is generated. This target identification map is used to describe the network segment where the target switch gateway node is located in the reference network flow data.

[0118] Based on the original identification map, the server reorganizes and marks in detail the connection relationship of the target switch gateway node according to the connection domain. For the key connection points and extension points in the connection domain, the connection relationship between them and related traffic characteristics and other information are highlighted in the target identification map. For example, the link between the key connection point and the extension point is marked with a special color or a bold line to indicate their importance in the connection situation. At the same time, the device information related to these connection points is updated, such as the role of the device in the connection domain (whether it is the source or destination of the traffic, etc.), the traffic processing capacity of the device, and other information. In this way, the target identification map more accurately describes the network segment where the target switch gateway node is located in the reference network flow data. Compared with the original identification map, it focuses more on showing the connection status of the target switch gateway node and its impact on the surrounding network environment.

[0119] Based on the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the target identification map, a network flow model corresponding to the candidate network flow data is generated.

[0120] The server converts the simulated data packets in the candidate network flow data according to the transmission path of the simulated data packets in the candidate network flow data, and generates an adjustable message corresponding to the simulated data packets. Taking the simulated data packets of the R&D personnel sending the code to the code warehouse server as an example, the source IP address (the IP of the R&D personnel's work computer), destination IP address (the IP of the code warehouse server), port number, protocol type and other information in the original message are retained, and adjustable parameters are added, such as adjustable traffic weight parameters (used to simulate different traffic priority situations) and adjustable transmission delay parameters (used to simulate network congestion or transmission delay changes in different network environments).

[0121] Next, the server uses the network topology, device configuration and other information in the reference network flow data. For example, the VLAN (virtual local area network) settings of the switch and the routing table information of the router in the reference network flow data reflect the basic architecture and operation rules of the network.

[0122] Then, combined with the target identification map. The target identification map describes in detail the network segment and connection status of the target switch gateway node in the reference network flow data. The server integrates the information of the adjustable message with the network topology and device configuration information in the reference network flow data, and determines the position and relationship of each element in the network flow model with the target identification map as a guide.

[0123] For example, the source IP address and destination IP address in the adjustable message are matched with the network topology in the reference network flow data to determine the actual transmission path in the network. According to the device connection relationship in the target identification map, the transmission path of the adjustable message is associated with the actual link in the network, and according to the device function role in the target identification map, the processing method of the adjustable message on different devices is set. The simulated data message and the target switch gateway node in the network flow model generated in this way are in an adjustable state, and the network flow model can be used to generate iterative network flow data after editing and overwriting the target switch gateway node or the simulated data message.

[0124] In a possible implementation, step S140 includes:

[0125] Step S141, converting the simulated data message in the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, and generating an adjustable message corresponding to the simulated data message.

[0126] Step S142: acquiring an identification map corresponding to the target switch gateway node according to the network segment where the target switch gateway node is located.

[0127] Step S143, fusing the adjustable message, the reference network flow data and the identification graph to generate a network flow model corresponding to the candidate network flow data.

[0128] In this embodiment, a simulated data message of a test report sent from the R&D department to the test department in the enterprise network is taken as an example. This simulated data message has a specific transmission path in the candidate network flow data, starting from a server in the R&D department, passing through the target switch gateway node (assuming that it is a node connecting the R&D subnet and the enterprise core network), and then reaching the receiving server of the test department.

[0129] During the conversion process, the server first obtains various information of the original simulated data message. The original message contains the source IP address (the IP of the R&D department server), the destination IP address (the IP of the test department receiving server), the port number (such as the source port is a random high port, the destination port is a specific service port), the protocol type (such as TCP protocol), etc. Then, add adjustable parameters on this basis. For example, add an adjustable traffic weight parameter, which can be used to simulate the traffic priority under different network conditions or business needs. If the R&D department has an urgent test report that needs to be sent first, the value of this traffic weight parameter can be increased. At the same time, add an adjustable transmission delay parameter to simulate network congestion or transmission delay changes under different network environments. For example, during network peak hours, the transmission delay parameter may be increased to reflect the actual delay that may occur. In this way, the original simulated data message is converted into an adjustable message.

[0130] The network segment where the target switch gateway node is located contains multiple network elements, such as switches directly connected to it, servers, and links connecting these devices, etc. The identification map is a detailed description of this network segment.

[0131] In an enterprise network, assume that the target switch gateway node is connected to multiple servers in the R&D subnet and some devices in the test subnet. The identification map obtained by the server will list the relevant information of these devices in detail. For devices, the map contains information such as the device's IP address, MAC address, device type (server or switch, etc.). For example, the IP address of a server in the R&D subnet is 192.168.1.10, the MAC address is 00 - 11 - 22 - 33 - 44 - 55, and the type is a file server. For links, the map contains information such as link bandwidth and link delay. For example, the link bandwidth connecting the target switch gateway node and a server in the R&D subnet is 1Gbps, and the link delay is 1ms. In addition, the identification map will also reflect the connection relationship between devices, such as which servers are connected to the target switch gateway node through which ports. This information is presented in an intuitive graphical or structured data form. The server obtains this identification map by querying the stored network topology and configuration information.

[0132] Then, the adjustable message, the reference network flow data and the identification map are fused to generate a network flow model corresponding to the candidate network flow data.

[0133] First, the server integrates the information of the adjustable message with the network topology and device configuration information in the reference network flow data. The reference network flow data contains the basic architecture information of the entire enterprise network, such as the VLAN configuration of the switch and the routing table of the router. Taking the source IP address and destination IP address in the adjustable message as an example, the server will determine the actual transmission path in the network based on the network topology in the reference network flow data. If the routing table in the reference network flow data shows that it is necessary to pass through certain specific switches and routers from the source IP address to the destination IP address, then this path will be determined, and combined with the traffic weight parameters and transmission delay parameters in the adjustable message, the actual data transmission under this network topology will be simulated.

[0134] At the same time, guided by the identification map, determine the position and relationship of each element in the network flow model. For example, according to the device connection relationship in the identification map, the transmission path of the adjustable message is associated with the actual link in the network. If the identification map shows that there is a specific link from the R&D department server to the target switch gateway node, then the transmission path of the adjustable message will correspond to this actual link. And according to the device function role in the identification map, set the processing method of the adjustable message on different devices. If a device in the identification map is marked as having a firewall function, then the adjustable message will be processed according to the rules of the firewall when passing through this device, such as port scanning or protocol filtering.

[0135] Through this fusion process, the simulated data packets and target switch gateway nodes in the generated network flow model are in an adjustable state. This means that in this model, the target switch gateway node can be configured and adjusted according to different needs, such as changing its port rate or routing strategy, etc., and the parameters of the simulated data packets can also be modified, such as changing the traffic weight or transmission delay, etc., and then iterative network flow data is generated for further network analysis and optimization.

[0136] Figure 2 The hardware structure of the switch gateway control system 100 based on artificial intelligence for implementing the switch gateway control method based on artificial intelligence provided by the embodiment of the present invention is shown as follows: Figure 2 As shown, the artificial intelligence-based switch gateway control system 100 may include a processor 110 , a machine-readable storage medium 120 , a bus 130 , and a communication unit 140 .

[0137] The machine-readable storage medium 120 may store data and / or instructions. In some embodiments, the machine-readable storage medium 120 may store data obtained from an external terminal. In some embodiments, the machine-readable storage medium 120 may store data and / or instructions that the artificial intelligence-based switch gateway control system 100 uses to execute or use to complete the exemplary method described in the present invention.

[0138] In the specific implementation process, one or more processors 110 execute computer executable instructions stored in the machine-readable storage medium 120, so that the processor 110 can execute the switch gateway control method based on artificial intelligence in the above method embodiment. The processor 110, the machine-readable storage medium 120 and the communication unit 140 are connected through the bus 130, and the processor 110 can be used to control the sending and receiving actions of the communication unit 140.

[0139] The specific implementation process of the processor 110 can refer to the various method embodiments executed by the above-mentioned artificial intelligence-based switch gateway control system 100. The implementation principles and technical effects are similar, and this embodiment will not be repeated here.

[0140] In addition, an embodiment of the present invention further provides a readable storage medium, in which computer executable instructions are preset. When a processor executes the computer executable instructions, the above-mentioned switch gateway control method based on artificial intelligence is implemented.

[0141] It should be noted that in order to simplify the description of the present invention and thus help understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, various features are sometimes combined into one embodiment, drawing or description thereof.

Claims

1. A switch gateway control method based on artificial intelligence, characterized in that: The method comprises: Acquire candidate network flow state data including a target switch gateway node and a simulated data message, and perform simulated data message parsing on the candidate network flow state data to generate a transmission path of the simulated data message in the candidate network flow state data; According to the transmission path of the simulated data message in the candidate network flow state data, remove the simulated data message in the candidate network flow state data, and generate reference network flow state data corresponding to the candidate network flow state data; Performing network segment detection on the target switch gateway node in the reference network flow state data to generate a network segment where the target switch gateway node in the reference network flow state data is located; Generate a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data, and the network segment where the target switch gateway node is located; wherein the simulated data message and the target switch gateway node in the network flow model are in an adjustable state, and the network flow model is used to generate iterative network flow data after editing and overwriting the target switch gateway node or the simulated data message; Traffic simulation and analysis are performed on the network flow model corresponding to the candidate network flow data to predict the corresponding network traffic trend. Based on the network traffic trend prediction result, the data packet transmission path in the network flow model is dynamically optimized to achieve control optimization of the target switch gateway node.

2. The switch gateway control method based on artificial intelligence according to claim 1 is characterized in that: The step of parsing the candidate network flow state data as a simulated data message to generate a transmission path of the simulated data message in the candidate network flow state data includes: Performing message parsing on the candidate network flow state data to generate simulated data messages of the candidate network flow state data; The transmission trajectory of the simulated data message is parsed to generate a transmission path of the simulated data message in the candidate network flow data.

3. The switch gateway control method based on artificial intelligence according to claim 1 is characterized in that: The removing the simulated data packets in the candidate network flow state data according to the transmission path of the simulated data packets in the candidate network flow state data to generate reference network flow state data corresponding to the candidate network flow state data includes: According to the transmission path of the simulated data message in the candidate network flow state data, the simulated data message in the candidate network flow state data is eliminated to generate message-free network flow state data; The network segment in which the simulated data message is removed from the message-free network flow data is optimized to generate reference network flow data corresponding to the candidate network flow data.

4. The switch gateway control method based on artificial intelligence according to claim 3 is characterized in that: The step of optimizing the network segment where the simulated data packets are removed from the network flow data without packets to generate reference network flow data corresponding to the candidate network flow data includes: In the message-free network flow data, determine the adjacent network segment of the network segment from which the simulated data message is removed, and obtain the flow data characteristics of the adjacent network segment; According to the flow data characteristics, the network segment of the candidate network flow data excluding the simulated data message is reconstructed to generate reference network flow data corresponding to the candidate network flow data.

5. The switch gateway control method based on artificial intelligence according to claim 1 is characterized in that: The performing network segment detection on the target switch gateway node in the reference network flow state data to generate the network segment where the target switch gateway node in the reference network flow state data is located includes: Acquire a member construction feature of the candidate network flow state data, wherein the member construction feature indicates the target switch gateway node in the candidate network flow state data; Searching the target switch gateway node in the reference network flow state data according to the member structure feature, and generating node mapping information of the target switch gateway node in the reference network flow state data; Generate a bounding domain of the target switch gateway node according to the node mapping information of the target switch gateway node in the reference network flow state data; wherein the bounding domain is used to describe the network segment where the target switch gateway node in the reference network flow state data is located; The generating a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the network segment where the target switch gateway node is located includes: A network flow model corresponding to the candidate network flow data is generated according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the defined domain.

6. The switch gateway control method based on artificial intelligence according to claim 5 is characterized in that: After searching the target switch gateway node in the reference network flow state data according to the member construction feature and generating node mapping information of the target switch gateway node in the reference network flow state data, the method further includes: Analyze the target switch gateway node to generate a penetration area of ​​the target switch gateway node in the reference network flow state data; The step of generating the bounded domain of the target switch gateway node according to the node mapping information of the target switch gateway node in the reference network flow state data includes: The defined domain of the target switch gateway node is generated according to the node mapping information of the target switch gateway node in the reference network flow state data and the penetration area of ​​the target switch gateway node in the reference network flow state data.

7. The switch gateway control method based on artificial intelligence according to claim 5 is characterized in that: After generating the defined domain of the target switch gateway node according to the node mapping information of the target switch gateway node in the reference network flow state data, the method further includes: Taking the defined domain as a reference, dividing the reference network flow state data to generate an identification map of the target switch gateway node; The identification map is used to describe the network segment where the target switch gateway node is located in the reference network flow state data; The step of generating a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the defined domain comprises: A network flow model corresponding to the candidate network flow data is generated based on the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the identification map.

8. The switch gateway control method based on artificial intelligence according to claim 7 is characterized in that: After dividing the reference network flow state data based on the defined domain and generating an identification map of the target switch gateway node, the method further includes: According to the identification map, the connection state of the target switch gateway node is analyzed to generate a key connection point in the identification map; wherein the key connection point is a connection point whose connection state changes more than a set value compared with other connection points; Perform link expansion according to the key connection point to generate a connection domain of the target switch gateway node, where the connection domain is used to describe the connection status of the target switch gateway node; Generate a target identification map of the target switch gateway node according to the connection domain, wherein the target identification map is used to describe the network segment where the target switch gateway node is located in the reference network flow state data; The step of generating a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the identification graph comprises: Generate a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the target identification map; The link expansion is performed according to the key connection point to generate the connection domain of the target switch gateway node, including: Extending the key connection point to generate a plurality of extension points; wherein the extension point is a connection point whose link state correlation with the key connection point is greater than a set correlation; The key connection point is linked to the multiple extension points to generate a connection domain of the target switch gateway node.

9. The switch gateway control method based on artificial intelligence according to claim 1, characterized in that: The generating a network flow model corresponding to the candidate network flow data according to the transmission path of the simulated data message in the candidate network flow data, the reference network flow data and the network segment where the target switch gateway node is located includes: According to the transmission path of the simulated data message in the candidate network flow state data, convert the simulated data message in the candidate network flow state data to generate an adjustable message corresponding to the simulated data message; According to the network segment where the target switch gateway node is located, obtaining an identification map corresponding to the target switch gateway node; The adjustable message, the reference network flow state data and the identification map are fused to generate a network flow state model corresponding to the candidate network flow state data.

10. A switch gateway control system based on artificial intelligence, characterized in that: The artificial intelligence-based switch gateway control system includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the artificial intelligence-based switch gateway control method described in any one of claims 1 to 9 above.

Images