Research On Internet Topology Change Detection And Reconstruction Methods

Internet topology is an important tool for reflecting the physical connectivity of real devices in a network,and is the basis for further research work such as network congestion control,network fault location,network load balancing,and network performance evaluation.Obtaining accurate network topology has important practical significance.Existing research has proposed a wealth of Internet topology identification methods,and most of the existing methods identify the topology structure of a certain historical state based on the information of the network over a period of time.However,in actual network environments,the state of network devices may change due to network attacks,device failures,and other reasons,resulting in changes in the network topology structure.Therefore,historical topology information will soon lose its validity in practice.In response to the above issues,this thesis studies the perception and reconstruction of internet topology changes.Based on the use of alias resolution under constraint conditions to obtain router level topology that conforms to the real internet structure,tomography is used to perceive dynamic topology changes and reconstruct the topology structure,thereby achieving the recognition and tracking of dynamic network topology.The main contributions are summarized as follows:(1)A network topology identification method with prior knowledge constraints is proposed.Using alias resolution rules based on analytical inference,this method considers several key structures that exist in real network topologies,analyzes the key structures,and maps them to the set of path fragments in Traceroute detection information.Then,during the alias resolution process,new rules are formulated to process the set of path fragments corresponding to the key structures.This method can make the network topology identified by alias resolution more consistent with the real network topology.(2)An efficient reconstruction method of dynamic network topology based on tomography is proposed.Based on the network topology obtained from historical information,this method senses the impact of network changes on partial network topology through end-to-end measurement.Collect end-to-end delay information through targeted detection,determine the specific location of network topology branches affected by network changes,and reinsert the affected network topology branches into the network topology,thereby restoring the accurately changed network topology.Through simulation verification,it can be proved that this method can quickly reconstruct the dynamically changed network topology using a small amount of re detection.