Research On Key Technologies Of IPv6 Network Topology Mearsurement

The evolution of the Internet to the next-generation Internet based on IPv6 has become a universal consensus.At present,the deployment and use of IPv6 is growing rapidly.IPv6 network topology measurement has attracted more and more attention from the measurement community and researchers.Through in-depth understanding and analysis of IPv6 networks Topology helps to observe network routing behavior,discover network bottlenecks,optimize network configuration,and improve network security.However,due to IPv6's huge address space,complicated address planning,diverse address space division policies,and low actual address utilization,these issues are different from IPv4,making IPv6 network topology measurement a huge challenge.In order to completely detect the IPv6 network topology,the two dimensions of measurement target and measurement method need to be considered.In view of the huge address space of IPv6,the target selection of IPv6 network topology measurement adopts the commonly used methods in IPv4.For example,it is impossible to detect short fixed-length prefixes in the BGP routing table such as random addresses in /48.On the other hand,the huge address space of IPv6 and the rapid growth of deployment and usage trends make it more difficult to complete and efficiently obtain a large-scale interface-level IPv6 network topology than IPv4.In summary,this article focuses on IPv6 network topology measurement,mainly including the following aspects of work:(1)Based on the IPv6 Hitlists,an IPv6 network topology measurement target selection technique is proposed to improve the effectiveness and completeness of IPv6 network topology measurement.First collect and fuse IPv6 Hitlists from different sources,including about 17 M addresses,then analyzed the characteristics of IPv6 Hitlists,and found that fused Hitlists showed high aggregation,low density and unpredictability of interface identification,and then proposed The prediction algorithm for IPv6 Hitlists,compared with Entropy/IP prediction,the prediction accuracy is about 1.4 to 5.5 times,and found that the results of the two are highly complementary.Finally,a comprehensive scheme for selecting IPv6 network topology measurement targets is given,compared with uniform random sampling.The method significantly improves the completeness of topology discovery,and the new topology discovery rate exceeds 94%.(2)A cross-platform solution combining the advantages of different measurement platforms of VPS and Looking Glass is proposed to efficiently and completely mearsure specific IPv6 networks.First explored the automatic deployment and collection of measurement points on different platforms,deploying 37 VPS vantage points in different computer rooms,and collecting 1111 Looking Glass vantage points contained in 145 different sites,and then design and implement based on VPS and Looking Glass' s cross-platform IPv6 network topology measurement system,propose principles for system design and implementation from vantage points,measurement itself,measurement strategies,and user use,and discuss the measurement's advantages and disadvantages of the two platforms.Experimental comparison found that Looking Glass is significantly more complete in topology measurement than VPS overall.Finally,a cross-platform IPv6 network topology measurement collaboration algorithm is proposed in conjunction with the evaluation method of node importance in complex networks.Compared with CAIDA Ark,the topology integrity is improved.The number of discovered nodes and links is 1.5 to 3 times,and the new topology discovery rate exceeds 79%.