The Network Topology Identification Based On Multi-source Information Fusion

With the gradual popularization of the network, a various of industries have already made vigorous development by virtue of the advantage of the network, and network itself has also increasingly become a part of everyone’s daily life. Autonomous System level network topology refers to the autonomous system as node, and display the interconnection and routing update strategy between autonomous systems.Identifying an AS-level network topology can help manage and optimize the network more effectively, which makes the network topology identification technology necessary for the fundamental research on the Internet. Using the BGP information to identify an AS-level network needs to deploy a sufficient number of BGP feeders,which can help achieve the complete AS path information. However, due to the fall short of BGP feeders deployed around the world, which makes the available BGP routing information very limited, there are a lot of connections that can’t be found.Besides, BGP information is a kind of control layer information. It means that BGP information is restricted by related routing strategy and its feeders are usually deployed in the upper position of the network, which makes a lot of connections in the lower position not to be identified. To solve the above problems, this thesis mainly consists of the following two research contents:(1)Putting forward a layout method for feeders: This methodology can help effectively determine how many new feeders are needed to be deployed in order to collect more complete network topology data, and then rank those deployed feeders according to their coverage rate on the non-stub ASes, which can help determine which feeders should be add to the route collectors first with the limited resources in order to maximize the non-stub AS coverage. No matter deploying BGP feeders or Traceroute monitors, it is both possible to collect routing information by covering the non-stub AS,thus this methodology is suitable for both of them.(2)Putting forward an AS-level topology identification method based on multi-source fusion: Topology identification method based on Traceroute belongs to the detection method of data level, which mainly reflect the link situation through which the data will pass actually in the transmission process. Its monitors are usually deployed in the lower position of the network, which can effectively solve the problem caused byBGP feeders that are located in the upper position of the network. However, topology identification method based on Traceroute can only achieve IP address information, in order to get the corresponding AS data, this thesis adopts a methodology based on dynamic programming and iteration to solve the IP-to-AS mapping problem, which can reduce the false matching rate greatly. Through merging Traceroute information into BGP information, then fusing IRR database information as a supplement, can get a more complete topology structure in the end.In this thesis, by comparing the non-stub AS coverage rate of the existing feeders with the non-stub AS coverage rate of the additionally deployed feeders, the result shows the feasibility of the layout method for BGP feeder or Traceroute monitor. Then through comparing the AS connection relationships identified by the existing BGP information with the AS connection relationships identified by the identification method combined with multi-source, the result shows the effectiveness of this new method.