Network performance analysis method based on actual switch model

Abstract

The invention discloses a network performance analysis method based on an actual switch model, which solves the problems of single analysis index and incapability of dealing with link dynamic generation and task actual transmission description, and comprises the following steps of: inputting an initial position of a network node to generate a time-varying topology; generating a node actual port routing table and a task route; constructing a single-node single-task actual forwarding model; forming a multi-node multi-task actual forwarding model; constructing a network performance multi-angle index system; and network performance analysis based on an actual switch is realized. According to the method, a node actual port routing table is designed, a multi-node multi-task actual forwarding model is constructed and formed, a task real-time distribution condition is obtained, and a network performance multi-angle index system is combined to output a network performance index change trend. According to the invention, the adaptive dynamic topology description of the network is guaranteed, and the accuracy of flow distribution management and the reliability of performance analysis are improved. The method is suitable for complex dynamic networks such as a vehicle network and a satellite network in a 6G scene.

Description

Technical Field

[0001] The present invention belongs to the field of network technology, and in particular to network performance analysis, including dynamic network task data packet forwarding and transmission, and specifically is a network performance analysis method based on an actual switch model, which is used for network performance analysis and management. Background Art

[0002] A network is a collection of multiple interconnected computers, devices, and communication systems that are designed to transmit and share data. The rapid development of network technology has enabled information to spread rapidly around the world. In particular, with the advancement of 6G technology, the network will further achieve the goals of ultra-high bandwidth, low latency, and a large number of connections, thereby changing the way people live and work more efficiently. In the context of the Internet of Things and the Internet of Vehicles, the application of networks has extended to smart homes, smart transportation, autonomous driving, and other fields, promoting the progress and innovation of science and technology. Through the network, not only can data be transmitted instantly, but also different types of devices can be connected. The application of 6G technology will enhance the intelligent characteristics of the network and improve people's quality of life and social operation efficiency. However, with the enrichment of network types and the gradual expansion of the scope of application, network security issues have become increasingly prominent. Threats such as cyber attacks, data leaks, and malware have affected the security and reliability of the system. In addition, the damage to network infrastructure caused by natural and man-made disasters may lead to interruptions in information transmission, resulting in data loss. In this context, the performance analysis of the network has become an important research direction. In particular, in the 6G scenario, how to cope with more complex network requirements and service standards will be a key challenge. The network load, packet loss rate, etc. affect the quality of service (QoS) and data transmission efficiency. When the network traffic exceeds its processing capacity, congestion will occur, resulting in reduced transmission efficiency and increased load. Therefore, understanding the performance of the network in different situations is crucial to optimizing the network architecture and improving service quality.

[0003] One of the typical applications of the network, the Internet of Vehicles (IoV), is an important part of the intelligent transportation system. IoV uses vehicles with interconnected interface devices as network nodes. According to different communication objects, the communication links in IoV include vehicle-to-vehicle (V2V) communication and vehicle-to-infrastructure (V2I) communication. Among them, V2V communication promotes the information exchange process between vehicles within the communication range. V2I communication realizes the information interaction between vehicles and road side units (RSU). Vehicles in IoV can autonomously collect information and transmit information through communication links to realize data interaction between vehicles. Each node in the vehicle cluster follows autonomous rules, and the cluster individuals collaborate based on the communication protocol, have good self-organization ability, and can collaboratively process tasks. But at the same time, environmental interference, vehicle density and distribution, high-speed mobility, etc. also bring challenges to the scheduling, optimization and decision-making of IoV. How to optimize the construction of IoV is the key issue of current IoV research.

[0004] To optimize the IoV construction problem, we first need to conduct a comprehensive analysis of the performance of IoV and understand the impact of different network parameters on network performance. Among the factors that affect the performance of IoV, the most important is task traffic. With the introduction of 6G technology in IoV, its distribution management has become more complicated. Due to factors such as the high-speed mobility and uneven distribution of vehicles, the topology in IoV changes very quickly, which makes the forwarding and transmission of tasks highly uncertain. When the vehicle density is high, the task traffic in the network rises rapidly, which may cause congestion in the node cache area. Therefore, it is of great practical significance to study the task transmission in IoV and explore the real-time distribution of traffic for network analysis. Traditional wireless network performance analysis problems are mainly aimed at static networks, and the impact of mobility on network layout and node location distribution is not considered. Therefore, many researchers have carried out performance analysis on dynamic networks to adapt to the changing environment and analysis needs.

[0005] In response to the performance analysis problem of IoV, Huang et al. proposed a real vehicle-mounted wireless network model in a paper published at the IEEE Global Communications Conference in 2015. The model uses Euclidean planes to represent the actual geometric structure of urban areas, thereby analyzing interference relationships and calculating asymptotic capacity. It can perform more realistic capacity analysis in vehicle networks and provide important theoretical support for dynamic network performance analysis. In a paper published in IEEE Transactions on Wireless Communications in 2016, Kwon et al. modeled the network as geometric elements of lines and points for a one-dimensional V2V network whose topology changes over time, and used geometric probability to analyze the connectivity and capacity of the network. Through mathematical modeling, it was found that the network capacity increases in a specific way, which verifies the validity of the analysis results and reveals the law of network capacity changing with topology, thereby providing a theoretical basis for optimizing network design and improving communication efficiency. In a paper published in IEEE Transactions on Intelligent Transportation Systems in 2017, Chen et al. analyzed the achievable throughput of a vehicle-mounted network with limited traffic density under a cooperative communication strategy. By obtaining a closed-form expression for the achievable throughput, the relationship between the throughput and its main performance influencing parameters, such as communication transmission rate and vehicle density, was revealed, providing a quantitative basis for understanding and optimizing the performance of vehicle-mounted networks. The above studies obtained the solution expression of network capacity through mathematical models, but none of them took into account the queuing, caching, and port forwarding processes of tasks on nodes in the actual network.

[0006] Sarvade et al., in a paper published at the International Conference on Advances in Computing, Communications and Informatics in 2017, used real traffic pattern scenarios to compare IEEE standard 802.11ac with previous MAC protocols, using throughput, jitter, and end-to-end delay as evaluation parameters, SUMO was used to generate real traffic patterns, and NS-3 was used to evaluate the performance of these protocols. By using real traffic pattern scenarios, more practical and accurate performance evaluations can be provided. Malnar et al., in a paper published at Promet-Traffic & Transportation in 2020, developed an NS-3-based vehicle network framework for analyzing throughput and packet loss performance indicators, compared several topology-based routing protocols, and proposed an expected transmission number metric to improve performance, providing a new theoretical basis and practical method for optimizing the throughput and packet loss rate of vehicle networks. In the 2023 Sensors paper published by Park et al., a domain-based vehicle network architecture and a region-based vehicle network architecture were designed for autonomous driving. The end-to-end delay and the total length and weight of the wiring harness were used as performance evaluation indicators, and the OMNET++ network simulator was used to analyze and verify the superiority of the region-based vehicle network architecture and its applicability as a vehicle architecture. However, these research works target relatively single performance indicators and are based on existing mainstream network simulation tools, such as OMNET++ and NS-3. They cannot solve the problem of dynamic link generation to depict the time-varying characteristics of the network and analyze the real-time distribution of tasks.

[0007] Through the above analysis, the problems and defects of the existing technology are as follows: Most of the existing IoV performance analysis technologies only consider a single indicator, and the research on multiple indicators is mostly aimed at network capacity and packet loss rate. Due to the high-speed mobility of vehicles in IoV and the diversity of user tasks, the topological connection relationship is changing rapidly. Especially in the 6G scenario, the density of nodes and links has increased significantly. The reasons that affect the performance of IoV are not only the vehicle density, which is a factor from the overall perspective of the network, but also the node capabilities, especially the cache capability. On the other hand, common simulation tools such as OMNET++ and NS-3 cannot solve the problem of dynamic link generation, that is, depicting the time-varying characteristics of links. The link construction, interruption and recovery between nodes in IoV change dynamically over time, and the performance of individual nodes is also closely related to network performance. Based on actual switches, the packet forwarding on the node needs to be based on the port routing forwarding table on the node and go through a series of complex processes. In addition, tools such as NS-3 do not support indicators other than the number of hops, and there is no suitable method to efficiently calculate key performance such as node load rate, and it is impossible to analyze dynamic network performance from multiple indicators and multiple factors.

[0008] It can be seen from the IoV performance analysis that the current network performance analysis technology has relatively simple analysis indicators. Common mainstream simulation platforms are unable to cope with the generation of dynamic links and obtain the actual task distribution in real time. It is even more difficult to cope with or connect network applications in 6G scenarios.

[0009] The difficulty in solving the above problems and defects is: it is also quite difficult to describe the time-varying characteristics of highly dynamic network topologies and how to design data packet queuing and forwarding strategy models that are close to real switches, and to build a multi-angle performance indicator system to ensure the real-time robustness of the network. Summary of the invention

[0010] In view of the problems and shortcomings of the prior art, the present invention provides a network performance analysis method based on an actual switch model, which can dynamically adapt to the real-time queuing and forwarding of network tasks and has a multi-angle performance indicator system.

[0011] The present invention is a network performance analysis method based on an actual switch model, characterized in that it comprises the following steps:

[0012] Step 1: Input the initial positions of network nodes to generate time-varying topology: Design a network architecture containing different types of entities, characterize the mobility model of moving entities in the network, determine the boundary of the network model, and abstract different types of entities as nodes in the network. Input their initial positions, calculate the transmission link rate based on the channel model between nodes, and obtain the time-varying topology of the network.

[0013] Step 2: Generate the actual port routing table of network nodes and user task routing: traverse the time-varying network topology nodes, use the routing algorithm to generate a common route with a certain node as the source node and the rest of the network nodes as the destination nodes, and convert all the common routes into port routes of the actual switch model, generate the actual port routing table of each node in the time-varying network, and generate the actual port routing of each user task at each moment according to the actual port routing table, the source node and the destination node of the user task;

[0014] Step three, construct an actual data packet forwarding model for a single node and a single task: based on the actual port routing table of the node generated by the network and the user task routing, model the transmission of a single node and a single task in the network, the node input inlet traffic of the data packet forwarding model, compare the sum of the inlet traffic and the node cache queue length with the node cache capacity, determine whether the task has packet loss, and output the number of node packet losses; compare the sum of the node inlet traffic and the cache queue length, the forwarding capacity and the egress queue length, output the node egress traffic and the updated cache queue length; compare the node egress traffic with the actual time-varying egress link rate, output the inlet traffic of the next hop node of the task and the number of link packet losses, complete the construction of the actual data packet forwarding model for a single node and a single task, obtain the actual distribution of a single task, and realize the data transmission of a single task on a single node based on the actual switch model;

[0015] Step 4: Form a multi-node multi-task actual forwarding model: Based on the actual data packet forwarding model of a single node and a single task, for multi-service situations, the traffic of different ports on the topological node is superimposed, and the nodes of the model sort the data packets in the cache queue according to the task priority, and allocate the number of forwarding packets and the number of cache packets to each task, compare and calculate the export traffic of each task, and form a multi-node multi-task forwarding model based on the actual switch, realize multi-service data packet forwarding, and obtain the actual task distribution of the network;

[0016] Step 5: Construct a multi-angle indicator system for network performance: Based on the actual forwarding model of multiple nodes and multiple tasks, extract the parameters related to network performance indicators, form multiple core indicators for evaluation system construction, and from the perspective of user services, set the packet loss rate and service success arrival rate as core indicators. From the perspective of network structure, set the node load rate, link load rate and the total actual network service flow as core indicators to complete the construction of the multi-angle indicator system for network performance.

[0017] Step 6: Implement network performance analysis based on actual switches: Based on the multi-angle indicator system of network performance, input the initial position of the network node to generate a time-varying topology, obtain the real-time distribution of tasks according to the node port routing table based on the actual switch and the multi-node multi-task forwarding model, output the network packet loss rate, service success arrival rate, node load rate, link load rate and the change trend of the actual network service total traffic performance indicators over time under different conditions, and implement real-time network performance analysis based on actual switches.

[0018] The present invention solves the problem of real-time forwarding transmission characterization of dynamic network adaptive tasks including IoV based on an actual switch model, and solves the problem of constructing a multi-angle network performance indicator system.

[0019] Compared with the prior art, the present invention has the following advantages:

[0020] Ensuring the adaptive dynamic topology characterization of the network: The present invention aims at the problem that the mainstream simulation tools used in the existing network performance analysis technology cannot solve the problem of dynamic link generation. By characterizing the movement model of the moving nodes and the transmission channel model between the nodes, the links are dynamically generated and interrupted, and the network topology that changes with time is obtained in real time. By depicting the time-varying characteristics of the link, the model is more in line with the actual network environment, and can effectively adapt to the changes in the network topology caused by the changes in the node position or environment, ensuring the adaptive dynamic topology characterization of the network.

[0021] Improved accuracy of task traffic distribution management: To address the problem of inaccurate task forwarding and traffic distribution simulation in existing networks, the present invention generates a network node port routing table based on an actual switch model, constructs an actual task data packet forwarding model, simulates network traffic transmission conditions close to those of a real switch, obtains the actual distribution of network tasks, provides an efficient task forwarding and transmission strategy, and improves the accuracy and efficiency of network task traffic distribution management.

[0022] The reliability of network performance analysis is guaranteed: In view of the problems in existing network performance analysis technologies that the analysis indicators are relatively single and lack a multi-dimensional performance evaluation system, the present invention constructs a network performance evaluation system covering multiple indicators based on the task forwarding transmission results and from the perspective of user services and network structure. Combined with the real-time distribution of tasks, the system outputs the changing trends of multiple performance indicators of the network under different conditions over time, thereby ensuring the rationality and correctness of the proposed model, ensuring the reliability of network performance analysis, and laying the foundation for the expansion and application of 6G networks in the future. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 It is a flowchart of the overall process of the present invention;

[0024] Figure 2It is a flowchart of the actual port routing table of a network node and a user task routing process generated by the present invention;

[0025] Figure 3 This is an example diagram of the network task port routing conversion of the present invention;

[0026] Figure 4 It is a data packet forwarding model diagram of a single node and a single task of the present invention;

[0027] Figure 5 It is a flow chart of a single-node single-task data packet forwarding model of the present invention;

[0028] Figure 6 It is a diagram of a multi-node multi-task data packet forwarding model of the present invention;

[0029] Figure 7 It is the network simulation scenario model diagram in Examples 9-11;

[0030] FIG8 is a simulation diagram of the IoV network performance under different QoS in Example 9, wherein FIG8(a) is a packet loss rate change curve, FIG8(b) is a service success arrival rate change curve, FIG8(c) is a link load rate change curve, FIG8(d) is a network actual service total flow change curve, and FIG8(e) is a node load rate change curve;

[0031] FIG9 is a simulation diagram of the IoV network performance under different node cache capabilities in Example 10, wherein FIG9(a) is a packet loss rate change curve, FIG9(b) is a service success arrival rate change curve, FIG9(c) is a link load rate change curve, FIG9(d) is a network actual service total flow change curve, and FIG9(e) is a node load rate change curve;

[0032] Figure 10 is a simulation diagram of the IoV network performance under different vehicle densities in Example 11, wherein Figure 10(a) is a packet loss rate change curve, Figure 10(b) is a service success arrival rate change curve, Figure 10(c) is a link load rate change curve, Figure 10(d) is a network actual service total traffic change curve, and Figure 10(e) is a node load rate change curve. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0034] Example 1

[0035] At present, the rapid development of technologies such as wireless communication and the Internet of Things has enabled the network to play an important role in various fields. One of the typical applications of the network, the Internet of Vehicles (IoV), is an important part of the intelligent transportation system and is widely used in various scenarios such as 6G. Vehicles can collaboratively process tasks through data interaction. Since the environment in which the IoV is located changes rapidly and the movement and distribution of vehicle nodes will affect the user's service experience, it is of great practical significance to analyze the transmission of task traffic in the IoV and explore the performance of the network under different conditions. Most of the existing network performance analysis methods only focus on a single indicator, while research on multiple indicators usually focuses on network capacity and packet loss rate. In addition, common simulation tools such as OMNET++ and NS-3 cannot solve the problem of dynamic link generation, that is, they cannot effectively describe the time-varying characteristics of the link, and cannot perform adaptive dynamic topology characterization of the network to perform actual task traffic distribution analysis. At the same time, tools such as NS-3 do not support other performance indicators except the number of hops, and lack methods for efficiently calculating key performance indicators such as node load rate, and cannot evaluate network performance from multiple perspectives. Since 6G brings higher connection density and lower latency, and at the same time puts forward higher requirements for real-time analysis and optimization of network performance, the above problems are particularly prominent in the 6G scenario. In response to these problems, the present invention conducts research and proposes a network performance analysis method based on an actual switch model.

[0036] The present invention is a network performance analysis method based on an actual switch model. Based on the time-varying topology of a dynamic network, the present invention constructs a task data packet forwarding model based on an actual switch model, obtains the actual distribution of tasks, establishes a multi-angle index system for network performance, and realizes network performance analysis. Figure 1 , Figure 1 The overall flow chart of the network performance analysis method based on the actual switch model of the present invention includes the following steps:

[0037] Step 1: Input the initial position of the network node to generate the time-varying topology: Design a network architecture containing different types of entities, characterize the mobility model of the moving entities in the network, determine the boundary of the network model, and abstract different types of entities as nodes in the network. Input their initial positions, calculate the transmission link rate based on the channel model between nodes, and obtain the time-varying topology of the network. Taking the vehicle network as an example, design an IoV network architecture containing vehicles and roadside units (RSUs). By characterizing the vehicle motion model and referring to the urban case described in 3GPP TR 36.885, determine the boundary of the network model. Vehicles and RSUs are abstracted as nodes in the network, input their initial positions, calculate the transmission link rate based on the vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) channel models, and obtain the network time-varying topology.

[0038] Step 2, generate the actual port routing table of network nodes and user task routing: traverse the time-varying network topology nodes, use routing algorithms such as Dijkstra's shortest path algorithm, Floyd's algorithm, Bellman-Ford algorithm, etc., generate a common route with a certain node as the source node and the rest of the network nodes as the destination nodes, and convert all the common routes into port routes of the actual switch model to generate the actual port routing table of each node in the time-varying network. Divide the network into several grid spaces, count the nodes that initiate tasks in each grid, generate the destination of each node that initiates the task, and generate the actual port routing of each user task at each moment according to the actual port routing table, the source node and the destination node of the user task. The present invention is different from the conventional network performance analysis technology in which tasks are forwarded by the generated node-based common routing path. The node actual port routing table constructed by the present invention simulates the task forwarding situation based on the node port close to the real switch, which is more in line with the actual network situation.

[0039] Step 3: Constructing a single-node single-task actual data packet forwarding model: Based on the actual port routing table of the node generated by the network and the user task routing, the present invention models the transmission of a single node single task in the network, see Figure 4 and Figure 5 , Figure 4 This is a single-node single-task data packet forwarding model diagram of the present invention, Figure 5The present invention is a flow chart of the data packet forwarding model of a single node and a single task. The node input inlet traffic of the data packet forwarding model first compares the sum of the inlet traffic and the node cache queue length with the size of the node cache capacity, determines whether the task has packet loss, and outputs the number of node packet loss; on this basis, compares the sum of the node inlet traffic and the cache queue length, the forwarding capacity and the size of the outlet queue length, outputs the node outlet traffic and the updated cache queue length; finally, compares the size of the node outlet traffic and the actual time-varying outlet link rate, outputs the inlet traffic of the next hop node of the task and the number of link packet loss, completes the construction of the actual data packet forwarding model of the single node and the single task, obtains the actual distribution of the single task, and realizes the data transmission of the single task on the single node based on the actual switch model. The input of the actual data packet forwarding model of the overall single node and the single task of the present invention is the node inlet traffic, and the output is the node outlet traffic, the node updated cache queue length, the number of node packet loss, the number of link packet loss and the inlet traffic of the next hop node of the task. The actual data packet forwarding model of the single node and the single task constructed by the present invention simulates the single task traffic transmission situation close to the real switch, and lays the foundation for the formation of the multi-node multi-task actual forwarding model. In addition, the actual data packet forwarding model of a single node and a single task of the present invention records parameters other than output parameters, such as the number of link load data packets, the number of node load data packets, etc., and provides support for the definition and design of various network performance indicators required for different network application scenarios, and can be expanded and designed according to the actual application scenarios of the network.

[0040] Step 4: Form a multi-node multi-task actual forwarding model: Based on the actual data packet forwarding model of a single node and a single task, for multi-service situations, when the single-service traffic is expanded to multi-service traffic, the traffic of different ports on the topology node is superimposed, see Figure 6 , Figure 6 This is a multi-node multi-task data packet forwarding model diagram of the present invention. The nodes of the model sort the data packets in the cache queue according to the task priority, and allocate the number of forwarding packets and the number of cache packets for each task, compare and calculate the export traffic of each task, and form a multi-node multi-task forwarding model based on an actual switch, realize multi-service data packet forwarding, and obtain the actual task distribution of the network. The multi-node multi-task actual forwarding model formed by the present invention simulates the multi-task traffic transmission situation close to that of a real switch, and provides support for the construction of a network performance indicator system.

[0041] Step 5, construct a multi-angle index system for network performance: Based on the actual forwarding model of multiple nodes and multiple tasks, extract the relevant parameters of network performance indicators, and form multiple core indicators for evaluation system construction. On the one hand, since the main service object of the network is the user, from the perspective of user business, the packet loss rate and the business success arrival rate are set as core indicators; from the perspective of network structure, since the load rates of nodes and links are different when the business traffic is transmitted in the network, the two will indirectly affect the user's service experience. Therefore, the present invention also sets the node load rate, link load rate and the actual total traffic of the network as core indicators to complete the construction of the multi-angle index system for network performance. The present invention starts from different angles, constructs a multi-angle index system for network performance, effectively reflects the congestion of the network, and conducts multi-angle and multi-faceted performance evaluation of the network.

[0042] Step 6: Implement network performance analysis based on actual switches: Based on the multi-angle indicator system of network performance, input the initial position of the network node to generate a time-varying topology, obtain the real-time distribution of tasks according to the node port routing table based on the actual switch and the multi-node multi-task forwarding model, output the network packet loss rate, service success arrival rate, node load rate, link load rate and the change trend of the actual network service total traffic performance indicators over time under different conditions, and implement real-time network performance analysis based on actual switches.

[0043] The present invention is an overall technical solution. First, by obtaining the initial position information of the network nodes, the network topology is calculated in real time according to the node motion model and the channel model. Then, the ordinary task routing generated by the routing algorithm is converted into the actual port routing. Then, a single-node single-task data packet forwarding model based on the actual switch model is constructed. By superimposing the traffic of different ports on the topological nodes, a multi-node multi-task actual forwarding model is formed. Finally, a multi-angle index system of network performance is constructed from the two perspectives of user service and network structure to realize network performance analysis based on the actual switch model. The present invention combines the real-time distribution of tasks and outputs the changing trends of multiple performance indicators of the network under different conditions over time, thereby ensuring the rationality and correctness of the proposed model and the reliability of network performance evaluation.

[0044] The present invention solves the problem of real-time forwarding transmission characterization of dynamic network adaptive tasks including IoV based on actual switch models, calculates real-time network topology based on node motion models and channel models, constructs a data packet forwarding model close to a real switch, and realizes adaptive dynamic transmission characterization of networks with changing tasks at any time. At the same time, it solves the problem of building a multi-angle network performance indicator system. Based on the real-time distribution of tasks, it designs multiple performance indicators from two aspects: user services and network structure, and performs multi-angle reliable performance analysis of the network under different conditions such as different node densities and different user QoS.

[0045] The present invention introduces the actual switch model into the dynamic network performance analysis, and makes the model more suitable for the actual network environment by describing the time-varying characteristics of the link, ensuring the adaptive dynamic topology characterization of the network. At the same time, it constructs a network node port routing table and a task data packet forwarding transmission model based on the actual switch model, simulates the user task forwarding transmission process close to the real network environment, obtains the real-time distribution of traffic, improves the accuracy and efficiency of network task traffic distribution management, and the model is easy to expand, providing support for network applications and analysis for 6G scenarios. When analyzing network performance, based on the actual data packet forwarding model, from the perspective of user services and network structure, a network performance evaluation system covering multiple indicators is constructed, and the reliability of network performance analysis is guaranteed by outputting the changing trends of multiple performance indicators of the network under different conditions over time.

[0046] Example 2

[0047] The network performance analysis method based on the actual switch model is the same as that in Example 1. The actual port routing table of the network node and the user task routing described in step 2 of the present invention are generated. Figure 2 , Figure 2 The present invention generates a flow chart of the actual port routing table of a network node and a user task routing; the present invention converts the common routing generated by the node based on the routing algorithm into the port routing of the actual switch, and further generates the port routing of each task at each moment, including the following steps:

[0048] 2.1 Generate common routes for network nodes: Before generating the actual port routing table for the network nodes, for each topological node in the time-varying network, use routing algorithms such as Dijkstra's shortest path algorithm, Floyd's algorithm, Bellman-Ford algorithm, etc. to generate common routes with each topological node as the source node and the rest of the network nodes as the destination nodes.

[0049] 2.2 Generate the actual port routing table of the network nodes: Based on the ordinary routes generated by each node and the remaining nodes, convert them into the port routing form of the actual switch. For each ordinary routing path, traverse the adjacent node pairs, convert the connection relationship between the adjacent nodes in the route into the corresponding node port connection relationship, and obtain the actual port routing form between each pair of adjacent nodes. Further traverse all nodes in the network to generate the actual port routing table of the network nodes; the actual port routing table of the network nodes constructed by the present invention simulates the node status close to the real switch, and provides support for the port routing generation of user tasks.

[0050] 2.3 Generate user task port routing: Based on the actual port routing table of network nodes, the network is divided into several grid spaces, and the nodes that initiate tasks in each grid are counted. On this basis, the destination of each task initiating node is randomly selected, and the actual port routing of each user task at each moment is generated according to the source node and destination node of the user task; taking a network with 6 nodes as an example, see Figure 3 , Figure 3 This is an example diagram of the network task port routing conversion of the present invention. The source node of task 1 is node 1, and the destination node is node 6. The ordinary route of task 1 generated by Dijkstra's shortest path algorithm is 1->2->4->6. According to the port routing table of each node and the node connection relationship, the corresponding actual port routing form is converted into '1 2'—>'2 1'—>'2 4'—>'4 2'—>'4 6'—>'6 4'.

[0051] The present invention introduces the port routing model of the actual switch, converts the common routing obtained by using the routing algorithm into the port routing form, generates the actual port routing table of the node, then counts the nodes that initiate the task in the network, and determines the destination of each node that initiates the task, and generates the port routing of each user task at each moment based on the actual port routing table, simulates the forwarding strategy based on port routing close to the real switch, and lays the foundation for the subsequent construction of the actual data packet forwarding model of the task to obtain the real-time distribution of the task.

[0052] Example 3

[0053] The network performance analysis method based on the actual switch model is the same as that in Example 1-2. The actual data packet forwarding model of a single node and a single task described in step 3 of the present invention is constructed. Figure 4 , the node is modeled as a model close to the actual switch. The node has a cache queue and an egress queue. After the task arrives at the node, it first enters the cache queue, and then based on the user task port routing, it is further pushed into the corresponding egress queue to complete the forwarding process; the present invention models the transmission of a single node and a single task in the network based on the actual node port routing table generated by the network and the user task routing, and obtains the actual distribution of the task, see Figure 5 , including the following steps:

[0054] 3.1 Node parameter data acquisition: Before forwarding a single-task data packet based on the actual switch model, the parameter sizes of the node where the current task is located are clarified, and the node parameter data are obtained, including: λ is the input inlet flow of the node, μ is the egress flow of the node egress queue, that is, the node forwarding flow, C is the node cache capacity, also known as the packet buffer size, L is the actual cache queue length of the node, F is the node forwarding capacity, R is the egress queue length of the node, and R' is the actual time-varying link rate of the node egress.

[0055] 3.2 Construct an actual data packet forwarding model for a single node and a single task: Based on the user task port routing generated by the network, when a node forwards a task, it takes out the task from the cache queue and pushes it into the egress queue of a different port. For the data packet forwarding of a single node and a single task, there are two situations in which the sum of the node inlet traffic and the cache queue length is greater than or less than or equal to the node cache capacity, that is, there are two situations in which the sum of the node inlet traffic and the node cache queue length is less than or equal to the node cache capacity, and the sum of the node inlet traffic and the node cache queue length is greater than the node cache capacity. On this basis, the number of packet losses at the node is calculated, the sum of the node inlet traffic and the cache queue length, the forwarding capacity and the egress queue length are compared, and the node egress traffic and the updated cache queue length are output.

[0056] 3.3 Real-time update of the next-hop node ingress traffic and link packet loss number: Based on the actual data packet forwarding model of a single node and a single task, after obtaining μ, the ingress traffic of the next-hop node of the task is calculated. Since the node switch cannot perceive the link interruption or rate drop caused by environmental changes, the node exit queue length R remains unchanged. The ingress traffic of the next-hop node is related to the actual time-varying link rate R' of the exit. Combined with the time-varying network topology, there are two situations: μ is less than or equal to the actual time-varying link rate, and μ is greater than the actual time-varying link rate. The ingress traffic and link packet loss number of the next-hop node of the task are updated in real time. Finally, the node exit traffic, the node updated cache queue length, the node packet loss number, the link packet loss number and the ingress traffic of the next-hop node of the task are output to obtain the actual distribution of the task and realize the data transmission of a single task on a single node based on the actual switch model.

[0057] The present invention constructs a data packet forwarding model based on an actual switch, first obtains node parameter data, then the node input inlet traffic, and determines the node outlet traffic, the number of node packet losses, and the updated cache queue length according to the node inlet traffic, cache queue length, node cache capacity, node forwarding capacity, and the size of the outlet queue length. Further, the obtained node outlet traffic is compared with the real-time actual outlet link rate to obtain the inlet traffic and link packet losses of the next hop node of the current task; the actual data packet forwarding model of a single node and a single task constructed by the present invention simulates the forwarding transmission of a single task in a network close to a real switch, improves the accuracy and efficiency of the single task traffic distribution management in the network, and lays the foundation for the formation of a multi-node multi-task actual forwarding model. In addition, the actual data packet forwarding model of a single node and a single task constructed by the present invention records parameters such as the number of link load packets and the number of node load packets other than the output parameters, which provides support for the definition and design of various network performance indicators required for different network application scenarios, and can be expanded according to actual application scenarios such as future 6G networks.

[0058] Example 4

[0059] The network performance analysis method based on the actual switch model is the same as that in Embodiment 1-3. The sum of the node inlet traffic and the cache queue length described in step 3 of the present invention may be greater than or less than or equal to the node cache capacity. Figure 5 , including the following steps:

[0060] 3.2.1 The sum of the node inlet traffic and the cache queue length is less than or equal to the node cache capacity λ+L≤C: There is no packet loss on the task at the node, and the node egress traffic is constrained by the node forwarding capacity and the egress queue length, which is μ=min{λ+L,F,R}. The node cache queue length is further updated to L'=λ+L-μ.

[0061] 3.2.2 The sum of the node inlet traffic and the cache queue length is greater than the node cache capacity λ+L>C: The task loses packets on the node, and the number of packet losses is Loss=λ+LC. In this case, after the task data packet enters the node switch, packet loss occurs first. Further, the node cache queue is full, so the node's egress traffic is μ=min{C,F,R}, and the node cache queue length is further updated to L'=λ+L-Loss-μ.

[0062] The present invention builds a data packet forwarding model based on an actual switch, and for the packet loss and forwarding process of a single task on a node, determines the number of packet losses at the node by comparing the sum of the node inlet flow and the cache queue length with the node cache capacity. On this basis, the node inlet flow and the sum of the cache queue length, the node cache capacity, and the size of the egress queue length are compared to determine the node egress flow and the updated cache queue length, and simulates the packet loss, forwarding, and real-time update of the node cache queue length of a single task on a node in a network based on a real switch, providing a basis for obtaining the inlet flow of the next-hop node of the task, and improving the accuracy and efficiency of the management of packet loss and forwarding of a single task in the network.

[0063] Example 5

[0064] The network performance analysis method based on the actual switch model is the same as that in Example 1-4. The next-hop node inlet flow and link packet loss number described in step 3 of the present invention are updated in real time. For the node egress flow, there are two situations: greater than or less than or equal to the actual time-varying link rate. See Figure 5 , including the following steps:

[0065] 3.3.1 The node egress flow is less than or equal to the actual time-varying link rate μ≤R': The number of packet losses on the transmission link is 0, so the ingress flow of the next-hop node of the task is λ=μ.

[0066] 3.3.2 The node egress flow is greater than the actual time-varying link rate μ>R': Packet loss occurs on the transmission link, and the number of link packet losses is Loss'=μ-R', so the ingress flow of the next-hop node of the task is λ=R'.

[0067] The present invention builds a data packet forwarding model based on an actual switch, and for the packet loss and forwarding process on the link between the node where a single task is located and its next-hop node, determines the number of link packet losses by comparing the node export traffic and the actual export time-varying link rate, and obtains the inlet traffic of the next-hop node of the current task. It simulates the packet loss and forwarding of a single task on the link in a network based on a real switch, as well as the real-time update of the inlet traffic of its next-hop node, thereby improving the accuracy and efficiency of the single-task packet loss and forwarding management in the network.

[0068] Example 6

[0069] The network performance analysis method based on the actual switch model is the same as that in Embodiment 1-5. The multi-node multi-task actual forwarding model is formed in step 4 of the present invention. Figure 6 The present invention aims at multi-service situations, superimposes the traffic of different ports on the topological node, determines the egress traffic of each task by sorting the tasks and allocating the number of cache forwarding packets, and includes the following steps:

[0070] 4.1 Sorting of multi-task packets with different priorities: Based on the port routing of user tasks, multiple tasks on different ports of topological nodes are determined, and packets in the cache queue are sorted according to the task priority. Nodes sort the tasks in the cache queue according to the principle that high-priority tasks are ranked first, medium-priority tasks are ranked second, and low-priority tasks are ranked last. Tasks of the same priority are randomly sorted. Take three nodes and two tasks as an example, see Figure 6 , the forwarding capacity F and cache capacity C of node 1 are 300 packet / s and 200 packet respectively, and the length of the four exit queues is 500 packet / s. At a certain moment in the network, node 1 has two tasks at the same time. Task 1 has 100 cached data packets on the node and is a high-priority task. Task 2 is a newly entered low-priority task with 200 data packets. According to the port routing table of node 1 and the destinations of task 1 and task 2, the next hop nodes of the two tasks are node 2 and node 3 respectively. At this time, node 1 will sort the two tasks. According to the task priority, the sorting result is task 1 in front and task 2 in the back.

[0071] 4.2 Allocation of forwarding packets and cached packets for multiple tasks: Based on the sorting results of packets with different priorities of multiple tasks, the node allocates forwarding packets and cached packets to each task. According to the sorting order of the cache queue, the tasks at the front of the queue are preferentially allocated with the same number of forwarding packets and cached packets as their packets, until the forwarding capacity and cache capacity of the node are allocated. The node takes out the packets from the cache queue in turn according to the allocation results, and pushes the tasks into the egress queues of different ports based on the port routing of the tasks. When there are multiple tasks in the egress queue of the port at the same time, the node allocates the egress queue length for each task according to the same principle as allocating the number of forwarding packets and cached packets; see Figure 6 According to the above allocation principle, node 1 first allocates the number of forwarding packets and cache packets for task 1. Since the number of data packets of task 1 is 100 packets, the number of forwarding and cache packets of task 1 are both 100 packets. Since the forwarding capacity F and cache capacity C of node 1 are 300 packet / s and 200 packets respectively, the number of data packets of task 2 is 200 packets, so the number of forwarding and cache packets of task 2 is allocated to 200 packets and 100 packets. Based on the port routing of the task, task 1 is pushed into the egress queue of the port connecting node 1 and node 2, and the length of the egress queue of task 1 is allocated to 100 packets. Similarly, task 2 is pushed into the egress queue of the port connecting node 1 and node 3, and the length of the egress queue of task 2 is allocated to 200 packets.

[0072] 4.3 Form an actual forwarding model for multiple nodes and multiple tasks: Based on the results of the number of forwarding packets and the number of cached packets allocated for multiple tasks, compare and calculate the export traffic of each task according to the actual data packet forwarding model of a single node and a single task, and realize the data packet forwarding of multiple tasks at a single node. When multiple nodes in the network carry multiple tasks, import the data packet of each task into the corresponding forwarding port, and then calculate the export traffic and packet loss information of each node and each task in the multi-node multi-task, complete the real-time forwarding of tasks along the port routing, and finally realize the data transmission of multiple nodes and multiple tasks based on the actual switch model.

[0073] The present invention constructs a data packet forwarding model based on an actual switch, and based on the actual data packet forwarding model of a single node and a single task, for multi-service situations, by superimposing multiple tasks of different ports on a topological node, the node sorts the data packets in the cache queue according to the task priority, and allocates the number of forwarding packets and the number of cache packets to each task, and according to the single-node single-task data packet forwarding rule, the real-time forwarding effect of multiple tasks following port routing is achieved; the multi-node multi-task actual forwarding model formed by the present invention simulates the forwarding transmission of multiple tasks in a network close to a real switch, obtains the actual multi-user task traffic distribution result, improves the accuracy and efficiency of the overall task traffic distribution management in the network including IoV, and lays a solid foundation for the construction of the subsequent network performance indicator system.

[0074] Example 7

[0075] The network performance analysis method based on the actual switch model is the same as that in Embodiments 1-6. The construction of the multi-angle index system for network performance described in step 5 of the present invention is based on the multi-node multi-task actual forwarding model, extracting network performance index-related parameters, and processing them to form multiple core indicators for evaluation system construction; it includes the following steps:

[0076] 5.1 Setting user service performance indicators: Since the main service object of the network is the user, and the network performance will affect the user's service experience, and the performance of users with different priorities is different, therefore, from the perspective of user services, the packet loss rate and service success arrival rate are set as core indicators. Among them, the packet loss rate is the key to detecting whether the network can transmit reliably and is one of the important indicators for measuring network performance. When the total number of network services is n, the set of packet loss and total number of packets of network services are {Loss 1 ,…,Loss n}, When , the overall network packet loss rate loss_rate is:

[0077]

[0078] For the service arrival rate, a series of network problems that occur during data packet transmission will cause some services to lose packets or accumulate on the node, reducing the success rate of the service reaching the destination. Therefore, the service arrival rate is also an important parameter of network performance. This parameter reflects the stability of network transmission. The service arrival rate arrive_rate is defined as the ratio of the number of service packets arriving at the destination to the total number of packets when the service is initiated:

[0079]

[0080] Among them, γ i represents the total number of packets arriving at the destination for service i, ρi represents the total number of packets initiated by service i, and M represents the total number of services generated by the network. A high packet loss rate can reflect that the network is currently abnormal, while a low service success arrival rate indicates that most tasks are congested on network nodes or packet loss occurs, which will affect the transmission performance of the network. A low service success arrival rate may even cause the entire network to fail to work properly. Therefore, evaluating the network based on the packet loss rate and service success arrival rate can improve the reliability of network analysis.

[0081] 5.2 Setting network structure performance indicators: From the perspective of network structure, when the service is transmitted in the network, the load rates of nodes and links are different, and the two will indirectly affect the user's service experience. The actual total traffic of the network reflects the load of all nodes in the network on the service. Therefore, setting the node load rate, link load rate and the actual total traffic of the network as core indicators can effectively reflect the congestion of the network and conduct performance evaluation. Among them, the node load rate S_node is defined as the ratio of the sum of the node egress traffic and the updated cache queue length to the actual cache capacity of the node, μ i +L' i represents the sum of the egress traffic of node i and the updated cache queue length, C i represents the cache capacity of node i, and N represents the total number of nodes in the network:

[0082]

[0083] The link load rate is the ratio of the number of packets actually transmitted on the link to the theoretical capacity of the link. This indicator intuitively shows the utilization of the network link. i ' ,j is the number of packets actually transmitted on link i→j, Ω i,j represents the theoretical capacity of link i→j, E represents the total number of network links, and the expression of link load rate S_link is as follows:

[0084]

[0085] The actual total traffic flow of the network is the sum of the total number of forwarded service packets and the total number of cached service packets in the network, that is, the sum of the node egress traffic and the updated cache queue length. This indicator reflects the total traffic flow actually carried by the network. i +L' i It represents the sum of the egress traffic of node i and the updated cache queue length:

[0086] sumflow=∑μ i +L i (5)

[0087] When there is no packet loss at the node, the smaller the node load rate is, the services are forwarded to other nodes in time, the node utilization rate is high, and the overall network performance is good. However, when the network changes or is disturbed, some services may not be forwarded in time due to link interruption, resulting in an increase in the node cache queue length and an increase in the node load rate. The link load rate indicator intuitively reflects the utilization of network links. When there is no packet loss, the higher the link load rate is, the higher the utilization rate of the network link is, and the better the network performance is. However, when the network is disturbed and some services are lost, the load of some links will be reduced, resulting in a decrease in network performance. For the total actual network service flow, when the network is congested or disturbed, the total actual network service flow will increase rapidly in a short period of time, which effectively reflects the congestion of the network. Therefore, evaluating the network from the perspective of node load rate, link load rate, and total actual network service flow can improve the reliability of network analysis.

[0088] Based on the above five indicators, the present invention completes the construction of a multi-angle indicator system for network performance from the perspective of multiple indicators, effectively reflects the real-time situation of the network, and performs performance evaluation.

[0089] The present invention constructs a network performance evaluation system covering multiple indicators based on the real-time forwarding transmission results of the tasks. From the perspective of user services, the packet loss rate and the service success arrival rate are set as core indicators. From the perspective of network structure, the node load rate, link load rate and the actual total service flow of the network are set as core indicators. By outputting the real-time change trends of multiple performance indicators of the network under different conditions, the rationality and correctness of the proposed model are ensured, the network is evaluated from multiple angles and in all directions, and the reliability of network performance analysis is guaranteed.

[0090] Example 8

[0091] The network performance analysis method based on the actual switch model is the same as that in Embodiments 1-7. The implementation of the network performance analysis based on the actual switch described in step 6 of the present invention includes the following steps:

[0092] 6.1 Input the initial position of the network node to generate the time-varying topology: Input the initial position of the network node, calculate the real-time position of the node and the time-varying link transmission rate according to the node motion model and the node communication channel model, and obtain the network time-varying topology.

[0093] 6.2 Actual port routing table of network nodes and generation of user task routing: traverse the time-varying network topology nodes, apply routing algorithms, generate common routes, and convert all common routes into port routes, generate the port routing table of each node in the time-varying network, divide the network into several grid spaces, count the nodes that initiate tasks in each grid, and generate the port routing of each user task at each moment according to the source node and destination node of the user task.

[0094] 6.3 Obtain the real-time distribution of user tasks based on the multi-node multi-task actual forwarding model: Based on the user task port routing generated by the network, according to the actual data packet forwarding model of a single node and a single task, the model's node input inlet traffic, output node outlet traffic, updated cache queue length, node packet loss number, link packet loss number and the inlet traffic of the next hop node of the task. For multi-business situations, the traffic of different ports on the topological nodes is superimposed. The nodes of the model sort the data packets in the cache queue according to priority, and allocate the number of forwarding packets and the number of cache packets to each task, forming a multi-node multi-task actual forwarding model to obtain the real-time distribution of user tasks.

[0095] 6.4 Implement network performance analysis based on actual switches: Based on the real-time distribution of user tasks, output the change trend of each performance indicator of the network over time under different conditions such as different node density and different node cache capacity, and implement network performance analysis based on actual switches.

[0096] The present invention obtains the initial position information of network nodes, calculates the network topology in real time according to the node motion model and the channel model, and depicts the time-varying characteristics of the link to make the model more suitable for the actual network, thereby ensuring the adaptive dynamic topology characterization of the network; constructs the actual port routing table of the node to generate the user task port routing, and then constructs the actual data packet forwarding model of the single node and single task, and forms a multi-node and multi-task actual forwarding model by superimposing the traffic of different ports on the topological nodes, simulating the real-time distribution of traffic in the network, and improving the accuracy and efficiency of network task traffic distribution management; finally, constructs a multi-angle indicator system for network performance, explores the real-time change trends of multiple indicators of the network under different conditions, realizes network performance analysis based on actual switches, ensures the reliability of network performance analysis, and provides support for the expansion and application of 6G networks in the future.

[0097] A more detailed example is given below, and the present invention is further verified by combining simulation and experimental results.

[0098] Example 9

[0099] The network performance analysis method based on the actual switch model is the same as that of Examples 1-8. In order to illustrate the effectiveness of the model proposed in the present invention, a number of simulation experiments were conducted to explore the changes in performance indicators of the network of the present invention under different conditions.

[0100] Simulation conditions: The simulation platform uses PyCharm, and the simulation computer CPU is Intel i5-1240P.

[0101] Simulation scenarios and parameters: For simulation scenarios, see Figure 7 , Figure 7This is the network simulation scenario model diagram. The simulation experiment sets up 20 vehicles and 4 RSUs. The simulation sets up 300 time slots in total. The first 100 time slots are red lights on the upper and lower lanes, and the left and right vehicles pass. The next 50 time slots are yellow lights on the left and right lanes. The last 150 time slots are red lights on the left and right lanes, and the upper and lower vehicles are moving. Since the vehicle speed in the urban scene is low, the impact of the vehicle acceleration and deceleration process on the topology change can be ignored. Therefore, in order to simplify the system modeling, the simulation sets the vehicles to move in a uniform straight line along their respective driving directions. The length of each time slot is 100ms. Each time slot has a fixed number of vehicles and RSU generation tasks. The maximum packet forwarding capacity of the vehicle node is 10Gbps. The other simulation parameters are shown in Table 1.

[0102] Table 1 Simulation parameters

[0103]

[0104] Simulation content: A simulation experiment is conducted on the changes in the network performance indicators of the present invention under different QoS. The simulation results are shown in FIG8 .

[0105] Simulation results and analysis: See Figure 8, which is a simulation diagram of IoV network performance under different QoS, where Figure 8(a) is a packet loss rate change curve, Figure 8(b) is a service success arrival rate change curve, Figure 8(c) is a link load rate change curve, Figure 8(d) is a network actual service total traffic change curve, and Figure 8(e) is a node load rate change curve. The horizontal axis is the time slot length in units of timeslots. The QoS in the simulation diagram represents the traffic size of the task initiated at each moment.

[0106] As shown in Figure 8(a) and Figure 8(b), the network traffic and task arrival rate initially show an upward trend, and then change dynamically. This is because under the constraints of task QoS and link rate, tasks cannot usually be sent all at once. Therefore, when a task is just initiated, the traffic in the network will continue to rise. Then, since new tasks are pouring into the network at every moment, the traffic distribution on nodes and links is in a dynamic change process, and task transmission will continue for a period of time. When the QoS is 20Mbps, the average packet loss rate is about 3%, and the task arrival rate is about 80%. As the QoS continues to increase, the packet loss rate gradually increases, and the task arrival rate continues to decrease. This is because the increase in network inlet traffic will cause the node's cache queue to grow. When the node's cache queue is full, packet loss occurs. And the larger the QoS, the easier it is for the cache queue to be full. In addition, under the same QoS, the reason why the task arrival rate gradually decreases is that as time changes, the movement of the vehicle causes the link to be interrupted. When the port routing table is not updated, most of the packets cannot be transmitted according to the original route, so they accumulate on the node. At the same time, since tasks are initiated on the network at every moment, the increase in time will also lead to an increase in network traffic, so the task arrival rate decreases.

[0107] Figure 8(c)-Figure 8(e) The link load rate, the total network traffic, and the node load rate are simulated as the user task QoS changes. When the QoS is small, the link load rate and the total network traffic increase with the increase of QoS. This is because when the QoS is small, the network packet loss is small, the total traffic increases with the increase of the inlet QoS, and at the same time, the number of node forwarding packets increases with the increase of the total number of packets, so the link load rate increases. However, when the QoS is large, the link load rate and the total network traffic are very small. The reason is that the network packet loss rate is high at this time, and there is not much difference between the number of packets that the node can forward after the packet loss and the total number of packets carried by the network.

[0108] For the node load rate, when QoS is large, the change trend is similar to the other two indicators. The reason is the same as the above analysis. At this time, both the vehicle and the RSU have a large degree of packet loss. When QoS is small, the change trend in the early stage is similar to the other two indicators. But in the later stage, because the vehicle's cache capacity is smaller than that of the RSU, the cache queue of the vehicle node is easily filled. At this time, most of the packet loss in the network occurs on the vehicle node, and the RSU is little affected. Therefore, the increase in total network traffic is due to the RSU, and the increase in vehicle packet loss causes the overall node load rate of the network with QoS = 60Mbps to be lower than when QoS = 20Mbps.

[0109] By exploring the changes in network performance indicators under different QoS, the present invention can observe the impact of different QoS on network performance, verify the effectiveness and feasibility of the model of the present invention, ensure the reliability of network performance analysis, and provide theoretical support and experimental basis for subsequent practical applications. In addition, different QoS configuration results provide support for network configuration optimization, which can help identify potential network problems in advance and formulate corresponding risk management strategies, thereby enhancing the reliability and stability of the network.

[0110] Example 10

[0111] The network performance analysis method based on the actual switch model is the same as that in Examples 1-8, and the simulation conditions, simulation scenarios and parameters are the same as those in Example 9.

[0112] Simulation content: A simulation experiment is conducted on the changes in the network performance indicators of the present invention under different node cache capabilities. The simulation results are shown in FIG9 .

[0113] Simulation results and analysis: See Figure 9, which is a simulation diagram of the IoV network performance under different node cache capabilities, where Figure 9(a) is a packet loss rate change curve, Figure 9(b) is a service success arrival rate change curve, Figure 9(c) is a link load rate change curve, Figure 9(d) is a network actual service total traffic change curve, and Figure 9(e) is a node load rate change curve. The horizontal axes are all time slot lengths in units. The QoS in the simulation diagram represents the traffic size of the task initiated at each moment, and C is the cache capacity of the vehicle node and RSU node.

[0114] The initial cache capacities of the vehicle and RSU are 100Mb and 500Mb respectively. The simulation experiment of the present invention reduces the two by half and combines the QoS observation results. As shown in Figure 9 (a) and Figure 9 (b), from the perspective of packet loss rate and task arrival rate, when the QoS is small, the impact of halving the cache capacity on the two indicators is small, while the opposite is true when the QoS is large. This is because when the network traffic is small, the lower cache capacity can effectively carry the task packets. When the traffic is high, the increase in the total number of packets will quickly fill the smaller cache queue, so the packet loss rate increases significantly, while the task arrival rate decreases significantly.

[0115] According to Figure 9(c), in terms of link load rate, halving the cache capacity under the same QoS will cause this indicator to decrease. This is because the reduction in cache capacity reduces the number of packets stored on the node, so the number of packets that the node can forward is correspondingly reduced. In addition, the link load rate of QoS = 20Mbps is smaller than that of no change when the cache capacity changes, but larger when QoS = 100Mbps, and the reason is similar to the analysis of packet loss rate change. According to Figures 9(d) and 9(e), from the perspective of total network traffic and node load rate, when QoS is the same, the difference in total network traffic before and after the cache capacity changes shows a trend of increasing over time. Moreover, the larger the QoS, the more obvious the gap. The reason is that halving the cache capacity will cause a large number of data packets to be lost, the total number of packets carried by the network will decrease, and the downward trend will become more obvious as the traffic increases. The reason why the node load rate of QoS = 20Mbps shows this change trend is similar to the analysis of node load rate change in Example 9. When QoS = 100Mbps, the total network traffic decreases after the cache capacity is halved, but the node load rate is almost the same as before. This is because when the cache capacity is large, although there are more data packets, they are relatively evenly distributed on each node. When the cache capacity is reduced, the data packets carried by the vehicle nodes are larger, so that the overall node load rate of the network will not decrease significantly compared to before, which verifies the reliability of the model of the present invention.

[0116] By exploring the changes in network performance indicators under different node cache capabilities, the present invention can observe the impact of different cache capabilities on network performance, verify the effectiveness and adaptability of the model of the present invention, and provide theoretical support and experimental basis for cache optimization in subsequent practical applications. In addition, the simulation results of different node cache capabilities provide strong support for network performance optimization, can help identify potential transmission bottlenecks, optimize data processing and forwarding strategies, and by analyzing the impact of cache capabilities on key performance indicators such as packet loss rate, a network cache management plan can be formulated in advance to reduce congestion risks and improve the stability and reliability of the overall network in various scenarios such as 6G.

[0117] Embodiment 11

[0118] The network performance analysis method based on the actual switch model is the same as that in Examples 1-8, and the simulation conditions, simulation scenarios and parameters are the same as those in Example 9.

[0119] Simulation content: A simulation experiment was conducted on the changes in the network performance indicators of the present invention under different vehicle densities. The simulation results are shown in FIG10 .

[0120] Simulation results and analysis: See Figure 10, which is a simulation diagram of the IoV network performance under different vehicle densities, where Figure 10(a) is a packet loss rate change curve, Figure 10(b) is a service success arrival rate change curve, Figure 10(c) is a link load rate change curve, Figure 10(d) is a network actual service total traffic change curve, and Figure 10(e) is a node load rate change curve. The horizontal axis is the time slot length in units of pieces. The QoS in the simulation diagram represents the traffic size of the task initiated at each moment, and Vehicle is the number of vehicle nodes.

[0121] As shown in Figure 10(a) and Figure 10(b), when the number of vehicles increases and the vehicle density increases, the number of nodes initiating tasks increases accordingly, resulting in a larger total QoS of the network, which in turn increases the packet loss rate and reduces the task arrival rate. However, when QoS = 60Mbps, the increase in the number of vehicles has little effect on these two indicators. This is because when the inlet QoS is small, the network can effectively carry an increase in the number of nodes initiating tasks. When the inlet QoS is large, the increase in the total number of packets will quickly consume the limited cache resources in the network. Therefore, the packet loss rate increases significantly, while the task arrival rate decreases significantly.

[0122] In terms of link load rate, as shown in Figure 10(c), before about 150 time slots, the link load rate is lower when there are more vehicles than when there are fewer vehicles under the same QoS, while after 150 time slots, the trend is the opposite. This is because when the vehicle density increases, the number of links in the network topology increases accordingly, so the link load rate decreases; and as time goes by, the impact of the increase in the total number of network packets exceeds the impact of the increase in the number of links, resulting in an increase in the number of packets forwarded by the node, so the link load rate increases compared to when the vehicle density is small.

[0123] according to Figure 10(d)-Figure 10(e) , for the total network traffic index, the index value of QoS = 60Mbps and Vehicle = 32 is higher than that of QoS = 180Mbps and Vehicle = 20 in the early stage, and is almost the same in the later stage. The reason is that in the early stage of the network, the network with a high vehicle density has sufficient node caches and links to store and forward data packets. But as time goes by, when the node cache tends to be stable, the gap between the two becomes smaller due to packet loss. The reason for the trend of node load rate changes is similar to the analysis of node load rate changes in Example 9. When QoS = 60Mbps, the index shows a downward trend in the later stage as the number of vehicles increases. According to Figure 10 (d), at this time, the total network traffic gap is small and the number of nodes is large, so the node load rate decreases.

[0124] By exploring the changes in network performance indicators under different vehicle densities, the present invention can observe the impact of vehicle density on network performance, verify the effectiveness and adaptability of the present invention in a dynamic traffic environment, and provide theoretical support and experimental basis for subsequent practical applications. In addition, the simulation results of different vehicle densities provide an important basis for network optimization, which can help identify possible congestion problems under high-density conditions and formulate corresponding network management strategies, thereby improving the efficiency and reliability of data transmission. By analyzing the specific impact of vehicle density on network performance, the network architecture can be optimized to ensure stable and efficient communication services under different traffic conditions.

[0125] In summary, the present invention belongs to the field of network technology, and in particular to network performance analysis. It discloses a network performance analysis method based on an actual switch model, which solves the problem that the current network performance analysis technology has relatively single analysis indicators, and common mainstream simulation platforms are unable to cope with the generation of dynamic links and obtain the actual task distribution in real time. Its implementation includes: inputting the initial position of the network node to generate a time-varying topology; generating the actual port routing table of the network node and the user task routing; constructing an actual data packet forwarding model of a single node and a single task; forming a multi-node and multi-task actual forwarding model; constructing a multi-angle indicator system for network performance; and realizing network performance analysis based on an actual switch. First, the network architecture is designed, and the dynamic network topology at each moment is obtained through the initial position of the node and its motion model and channel model. Then, the nodes in the dynamic network topology are converted into actual port routing tables according to the common routes generated by the routing algorithm. On this basis, the user task port routing is obtained. Then, the actual data packet forwarding model of a single node and a single task is constructed based on the actual switch, and it is expanded to form a multi-node multi-task actual forwarding model, and the number of packet losses and export traffic of each task on the node are obtained, and the real-time distribution of user tasks in the network is obtained. Then, based on the data packet forwarding model, the parameters related to the formation of indicators are extracted, and a multi-angle indicator system of network performance is constructed from the two perspectives of user business and network structure. Finally, based on the multi-angle indicator system of network performance, combined with the real-time distribution of tasks, the trend of network performance indicators over time under different conditions is output, and network performance analysis based on actual switches is realized. The network performance analysis method based on the actual switch model proposed in the present invention ensures the adaptive dynamic topology characterization of the network, improves the accuracy of task traffic distribution management, and ensures the reliability of network performance analysis. It is suitable for complex dynamic networks such as vehicle networks, drone networks, and satellite networks in 6G scenarios.

Claims

1. A network performance analysis method based on an actual switch model, characterized in that: Based on the time-varying topology of the dynamic network, we build a node port routing table and task forwarding model based on the actual switch model, obtain the real-time distribution of tasks, establish a multi-angle indicator system for network performance, and realize real-time network performance analysis, including the following steps: Step 1: Input the initial positions of network nodes to generate time-varying topology: Design a network architecture containing different types of entities, characterize the mobility model of moving entities in the network, determine the boundary of the network model, and abstract different types of entities as nodes in the network. Input their initial positions, calculate the transmission link rate based on the channel model between nodes, and obtain the time-varying topology of the network. Step 2: Generate the actual port routing table of network nodes and user task routing: traverse the time-varying network topology nodes, use the routing algorithm to generate a common route with a certain node as the source node and the rest of the network nodes as the destination nodes, and convert all the common routes into port routes of the actual switch model, generate the actual port routing table of each node in the time-varying network, and generate the actual port routing of each user task at each moment according to the actual port routing table, the source node and the destination node of the user task; Step three, construct an actual data packet forwarding model for a single node and a single task: based on the actual port routing table of the node generated by the network and the user task routing, model the transmission of a single node and a single task in the network, the node input inlet traffic of the data packet forwarding model, compare the sum of the inlet traffic and the node cache queue length with the node cache capacity, determine whether the task has packet loss, and output the number of node packet losses; compare the sum of the node inlet traffic and the cache queue length, the forwarding capacity and the egress queue length, output the node egress traffic and the updated cache queue length; compare the node egress traffic with the actual time-varying egress link rate, output the inlet traffic of the next hop node of the task and the number of link packet losses, complete the construction of the actual data packet forwarding model for a single node and a single task, obtain the actual distribution of a single task, and realize the data transmission of a single task on a single node based on the actual switch model; Step 4: Form a multi-node multi-task actual forwarding model: Based on the actual data packet forwarding model of a single node and a single task, for multi-service situations, the traffic of different ports on the topological node is superimposed, and the nodes of the model sort the data packets in the cache queue according to the task priority, and allocate the number of forwarding packets and the number of cache packets to each task, compare and calculate the export traffic of each task, and form a multi-node multi-task forwarding model based on the actual switch, realize multi-service data packet forwarding, and obtain the actual task distribution of the network; Step 5: Construct a multi-angle indicator system for network performance: Based on the actual forwarding model of multiple nodes and multiple tasks, extract the parameters related to network performance indicators, form multiple core indicators for evaluation system construction, and from the perspective of user services, set the packet loss rate and service success arrival rate as core indicators. From the perspective of network structure, set the node load rate, link load rate and the total actual network service flow as core indicators to complete the construction of the multi-angle indicator system for network performance. Step 6: Implement network performance analysis based on actual switches: Based on the multi-angle indicator system of network performance, input the initial position of the network node to generate a time-varying topology, obtain the real-time distribution of tasks according to the node port routing table based on the actual switch and the multi-node multi-task forwarding model, output the network packet loss rate, service success arrival rate, node load rate, link load rate and the change trend of the actual network service total traffic performance indicators over time under different conditions, and implement real-time network performance analysis based on actual switches.

2. The network performance analysis method based on the actual switch model as claimed in claim 1, characterized in that: The generation of the actual port routing table of the network node and the user task routing described in step 2 is to convert the ordinary routing generated based on the routing algorithm into the port routing of the actual switch, and further generate the port routing of each task at each moment; it includes the following steps: 2.1 Generate common routes for network nodes: Before generating the actual port routing table for the network nodes, for each topological node in the time-varying network, a routing algorithm such as Dijkstra's shortest path is used to generate common routes with each topological node as the source node and the rest of the network nodes as the destination nodes; 2.2 Generate the actual port routing table of network nodes: Based on the common routes generated by each node and the other nodes, convert them into the port routing form of the actual switch. For each common routing path, traverse the adjacent node pairs, convert the connection relationship between the adjacent nodes in the route into the corresponding node port connection relationship, and obtain the actual port routing form between each pair of adjacent nodes. Further traverse all nodes in the network to generate the actual port routing table of the network nodes; 2.3 Generate user task port routing: Based on the actual port routing table of network nodes, divide the network into several grid spaces, count the nodes that initiate tasks in each grid, and on this basis, determine the destination of each node that initiates the task. According to the source node and destination node of the user task, generate the actual port routing of each user task at each moment.

3. The network performance analysis method based on the actual switch model as claimed in claim 1, characterized in that: The actual data packet forwarding model of a single node and a single task described in step 3 is constructed based on the actual port routing table of the node generated by the network and the user task routing, and the transmission of a single node and a single task in the network is modeled to obtain the actual distribution of the task; The following steps are included: 3.1 Node parameter data acquisition: Before forwarding a single-task data packet based on the actual switch model, the parameter sizes of the node where the current task is located are clarified, and the node parameter data are obtained, including: λ is the input inlet flow of the node, μ is the egress flow of the node egress queue, i.e., the node forwarding flow, C is the node cache capacity, L is the actual cache queue length of the node, F is the node forwarding capacity, R is the egress queue length of the node, and R' is the actual time-varying link rate of the node egress; 3.2 Constructing the actual data packet forwarding model of a single node and a single task: Based on the user task port routing generated by the network, when the node forwards the task, it takes out the task from the cache queue and pushes it into the egress queue of different ports. For the data packet forwarding of a single node and a single task, there are two situations: the sum of the node inlet traffic and the cache queue length is greater than or less than or equal to the node cache capacity, that is, there are two situations: the sum of the node inlet traffic and the node cache queue length is less than or equal to the node cache capacity, and the sum of the node inlet traffic and the node cache queue length is greater than the node cache capacity. Calculate the number of packet losses at the node, compare the sum of the node inlet traffic and the cache queue length, the forwarding capacity and the egress queue length, output the node egress traffic and the updated cache queue length; 3.3 Real-time update of the next-hop node ingress traffic and link packet loss number: Based on the actual data packet forwarding model of a single node and a single task, after obtaining μ, the ingress traffic of the next-hop node of the task is calculated. Since the node switch cannot perceive the link interruption or rate drop caused by environmental changes, the node exit queue length R remains unchanged. The ingress traffic of the next-hop node is related to the actual time-varying link rate R' of the exit. Combined with the time-varying network topology, there are two situations: μ is less than or equal to the actual time-varying link rate, and μ is greater than the actual time-varying link rate. The ingress traffic and link packet loss number of the next-hop node of the task are updated in real time. Finally, the node exit traffic, the node updated cache queue length, the node packet loss number, the link packet loss number and the ingress traffic of the next-hop node of the task are output to obtain the actual distribution of the task and realize the data transmission of a single task on a single node based on the actual switch model.

4. The network performance analysis method based on the actual switch model as claimed in claim 3, characterized in that: In step 3, the sum of the node ingress traffic and the cache queue length may be greater than or less than or equal to the node cache capacity. The steps include: 3.2.1 The sum of the node inlet traffic and cache queue length is less than or equal to the node cache capacity λ+L≤C: The task has no packet loss on the node, and the node egress traffic is constrained by the node forwarding capacity and the egress queue length, which is μ=min{λ+L,F,R}. The node cache queue length is further updated to L'=λ+L-μ; 3.2.2 The sum of the node inlet traffic and the cache queue length is greater than the node cache capacity λ+L>C: The task loses packets on the node, and the number of packet losses is Loss=λ+LC. In this case, after the task data packet enters the node switch, packet loss occurs first. Further, the node cache queue is full, so the node's egress traffic is μ=min{C,F,R}, and the node cache queue length is further updated to L'=λ+L-Loss-μ.

5. The network performance analysis method based on the actual switch model as claimed in claim 3, characterized in that: The next hop node ingress traffic and link packet loss number described in step 3 are updated in real time; including the following steps: 3.3.1 The node egress flow is less than or equal to the actual time-varying link rate μ≤R': The number of packet losses on the transmission link is 0, so the ingress flow of the next hop node of the task is λ=μ; 3.3.2 The node egress flow is greater than the actual time-varying link rate μ>R': Packet loss occurs on the transmission link, and the number of link packet losses is Loss'=μ-R', so the ingress flow of the next-hop node of the task is λ=R'.

6. The network performance analysis method based on the actual switch model as claimed in claim 1, characterized in that: The formation of the multi-node multi-task actual forwarding model described in step 4 is to superimpose the traffic of different ports on the topological node for multiple services and determine the egress traffic of each task; it includes the following steps: 4.1 Sorting of data packets with different priorities for multiple tasks: Based on the port routing of user tasks, multiple tasks on different ports of topological nodes are determined, and data packets in the cache queue are sorted according to the task priority. The nodes sort the tasks in the cache queue according to the principle that high-priority tasks are placed first, medium-priority tasks are placed second, and low-priority tasks are placed last. Tasks with the same priority are sorted randomly. 4.2 Allocation of forwarding packets and cached packets for multiple tasks: Based on the sorting results of different priority packets of multiple tasks, the node allocates forwarding packets and cached packets to each task. According to the sorting order of the cache queue, the tasks at the front of the queue are preferentially allocated with the same number of forwarding packets and cached packets as their packets, until the forwarding capacity and cache capacity of the node are allocated. The node takes out the packets from the cache queue in turn according to the allocation results, and pushes the tasks into the egress queues of different ports based on the port routing of the tasks. When the egress queue of a port has multiple tasks at the same time, the node allocates the egress queue length for each task according to the same principle as allocating the number of forwarding packets and cached packets. 4.3 Form an actual forwarding model for multiple nodes and multiple tasks: Based on the results of the number of forwarding packets and the number of cached packets allocated for multiple tasks, compare and calculate the export traffic of each task according to the actual data packet forwarding model of a single node and a single task, and realize the data packet forwarding of multiple tasks at a single node. When multiple nodes in the network carry multiple tasks, import the data packet of each task into the corresponding forwarding port, and then calculate the export traffic and packet loss information of each node and each task in the multi-node multi-task, complete the real-time forwarding of tasks along the port routing, and finally realize the data transmission of multiple nodes and multiple tasks based on the actual switch model.

7. The network performance analysis method based on the actual switch model as claimed in claim 1, characterized in that: The construction of the multi-angle index system for network performance described in step 5 is based on the actual forwarding model of multiple nodes and multiple tasks, extracting parameters related to network performance indicators, and processing them to form multiple core indicators for evaluation system construction; it includes the following steps: 5.1 Setting user service performance indicators: Since the main service object of the network is the user, and the network performance will affect the user's service experience, and the performance of users with different priorities is different, therefore, from the perspective of user services, the packet loss rate and service success arrival rate are set as core indicators. Among them, the packet loss rate is the key to detecting whether the network can transmit reliably and is one of the important indicators for measuring network performance. When the total number of network services is n, the set of packet loss and total number of packets of network services are {Loss1,…,Loss n }, When , the overall network packet loss rate loss_rate is: For the service arrival rate, a series of network problems that occur during data packet transmission will cause some services to lose packets or accumulate on the node, reducing the success rate of the service reaching the destination. Therefore, the service arrival rate is also an important parameter of network performance. This parameter reflects the stability of network transmission. The service arrival rate arrive_rate is defined as the ratio of the number of service packets arriving at the destination to the total number of packets when the service is initiated: Among them, γ i represents the total number of packets arriving at the destination for service i, ρ i represents the total number of packets initiated by service i, and M represents the total number of services generated by the network; 5.2 Setting network structure performance indicators: From the perspective of network structure, when the service is transmitted in the network, the load rates of nodes and links are different, and the two will indirectly affect the user's service experience. The actual total traffic of the network reflects the load of all nodes in the network on the service. Therefore, setting the node load rate, link load rate and the actual total traffic of the network as core indicators can effectively reflect the congestion of the network and conduct performance evaluation. Among them, the node load rate S_node is defined as the ratio of the sum of the node egress traffic and the updated cache queue length to the actual cache capacity of the node, μ i +L' i represents the sum of the egress traffic of node i and the updated cache queue length, C i represents the cache capacity of node i, and N represents the total number of nodes in the network: The link load rate is the ratio of the number of packets actually transmitted on the link to the theoretical capacity of the link. This indicator intuitively shows the utilization of the network link. i,j is the number of packets actually transmitted on link i→j, Ω i,j represents the theoretical capacity of link i→j, E represents the total number of network links, and the expression of link load rate S_link is as follows: The actual total traffic flow of the network is the sum of the total number of forwarded service packets and the total number of cached service packets in the network, that is, the sum of the node egress traffic and the updated cache queue length. This indicator reflects the total traffic flow actually carried by the network. i +L' i It represents the sum of the egress traffic of node i and the updated cache queue length: sumflow=Sm i +L i Based on the above five indicators, the construction of a multi-angle indicator system for network performance is completed from the perspective of multiple indicators, which can effectively reflect the real-time situation of the network and conduct performance evaluation.

8. The network performance analysis method based on the actual switch model according to claim 1, characterized in that: The network performance analysis based on the actual switch described in step 6 includes the following steps: 6.1 Input the initial position of the network node to generate the time-varying topology: Input the initial position of the network node, calculate the real-time position of the node and the time-varying link transmission rate according to the node motion model and the node communication channel model, and obtain the network time-varying topology; 6.2 Actual port routing table of network nodes and generation of user task routing: traverse the time-varying network topology nodes, apply routing algorithms, generate common routing, and convert all common routing into port routing, generate the port routing table of each node of the time-varying network, divide the network into several grid spaces, count the nodes that initiate tasks in each grid, and generate the port routing of each user task at each moment according to the source node and destination node of the user task; 6.3 Obtaining the real-time distribution of user tasks based on the multi-node multi-task actual forwarding model: Based on the user task port routing generated by the network, according to the actual data packet forwarding model of a single node and a single task, the model's node input inlet traffic, output node egress traffic, updated cache queue length, node packet loss number, link packet loss number, and the inlet traffic of the next hop node of the task, for multi-service situations, superimpose the traffic of different ports on the topological nodes, the model's nodes sort the data packets in the cache queue according to priority, and allocate the number of forwarding packets and cache packets to each task, forming a multi-node multi-task actual forwarding model to obtain the real-time distribution of user tasks; 6.4 Implement network performance analysis based on actual switches: Based on the real-time distribution of user tasks and combined with the multi-angle indicator system of network performance, output the change trend of each performance indicator of the network under conditions such as different node densities over time to implement network performance analysis based on actual switches.

Images