Research On Network Topology Inference Technology Based On Network Delay

Today the size and the complexity of network is growing.In the research of the Interner,the accurate and timely topology inference of the Internet router level is one of the mainunresolved issues. The network topology inference based on network tomography is apotentially attractive approach.Network tomography (Network Tomography,NT) technology which is based on theconcept proposed to medical perspective is a network measurement technology using newthought, it is based on end to end measurement,it infers the internal characteristics of thenetwork by inversion, and uses the principle of statistical reasoning and inference toreconstrust topology. network tomography technique does not require the cooperation ofNodes within the network.,only needs to select a group receiving nodes to complete theinternal characteristics of the network inference. We can use network tomography technologyto solve the problem of non-cooperation within the network nodes. However, the restriction ofcurrent network tomography is not inferring scale network topology，because a large scalenetwork topology inference required measurement engineering too difficult to achieve. Thispaper studies the network topology inference techniques based on delay characteristics,describes a leaf node based on the depth-first search (DFS) sequence topology inferencealgorithm.First, identify the DFS sequence using a recursive bisection leaf node, then use theleaf nodes DFS sequences inferred network topology. The algorithm can reconstrust thelogical tree of network topology accurately and efficiently.A large number of simulationexperiments show that this method requires the number of pairs of probe packets of less than15%currently used clustering algorithm.the network topology can be infered more accuratelyand faster. Paper details are as follows:The first three chapters introduce the research background, research status and researchcontents, and describes the relevant content of the research techniques used in networkmeasurement and network tomography, and the more mature of the existing network topologyinference algorithm conducted a summary and conclusion.The fourth chapter is the core content of this article. Sequence DFS leaf node networktopology inference algorithm is introduced by this chapter. This chapter has a detaileddescription of the algorithm used to infer the network topology of the entire process.The fifth chapter is introduce the simulation using NS2simulation platform. Through theexperiments we know that based on the sequence of leaf nodes DFS network topologyinference algorithm proposed in Chapter IV were effectively.The experiments have provedthe efficiency and feasibility of the algorithm.