Research On Internet Router-Level Topology Measurement And Identification Method

In recent years,with the rapid development of network technology,the scale of the network is also expanding,and the network structure has become extremely complex.In order to better monitor the use of the network and optimize the network reasonably,network administrators must use large-scale network detection methods to clearly understand the scale of the Internet and the hierarchy of each part.Therefore,the research of network detection methods and the research of detection node deployment have become hot topics in network research in recent years.Traditional network detection research is currently mainly focused on the detection of network topology and the measurement of network performance parameters such as packet loss rate and delay.For network topology detection,more random detection methods are used to obtain the node information and basic structure of the network.However,the deployment of network nodes needs to consume a certain amount of resources.Random detection methods want to cover more network nodes,usually large-scale detection is required,the overhead is large and the coverage rate of network nodes is low.Therefore,the work of this article focuses on how to detect the network to achieve a higher coverage of network nodes and how to deploy the corresponding detection nodes.Choosing a suitable detection method helps to obtain a more complete network structure with less detection overhead.Choosing a suitable detection node deployment for multi-source detection can detect a larger network topology range while deploying fewer nodes.Therefore,the main work of this article can be summarized into the following two parts:(1)Aiming at the deployment of network nodes,a detection node deployment method based on the importance of nodes is proposed.This method utilizes some common characteristics in complex network topology,estimates and restores the complete network topology by using the known network topology,and calculates the importance of each node to determine the influence on the network based on the estimated complete network topology Larger nodes,through multiple statistics,and then identify the network detection nodes that should be deployed.This method uses NS3 simulation tool to generate router-level topology and test it,which verifies the effectiveness of the method.(2)Aiming at the problem of unsatisfactory coverage of nodes in the network detected by the prior art,an efficient network topology detection method based on gradient guidance is proposed.Thesis uses neural network to analyze the mapping relationship between the destination node and the passing node obtained in the initial network detection,and designs a method that uses the gradient calculation of the neural network and generates the destination IP for the next detection.This method is tested in the NS3 simulation tool and the real network in Japan.Compared with the random detection method,it can achieve more network node coverage.