摘要
以建立中国IP级网络拓扑图景为目标,首先评价了传统的目标抽样法,提出了3项改进技术,并测量了中国网络,然后运用复杂网络研究方法提取并比较分析了拓扑特征,最后采用IP2AS技术统计了自治域规模。结果表明,传统目标抽样法将丢失大量信息,改进技术在保持完整性的同时降低约一半负载。获得新的中国拓扑规模5倍于CAIDA的skitter。新拓扑与skitter拓扑相比,呈现更显著的非相称性、更弱的聚集、更短的距离、更不均衡的节点流量分布。发现了中国自治域规模的重尾分布。
To build the prospect of Chinese IP-level topology, the traditional destination sampling method was evaluated; three new improved techniques were developed; and Chinese network was measured. With the methodology of complex network, the topology characteristics were extracted and compared. The sizes of autonomous systems (AS) were investigated with the IP2AS technique. The results showe that lots of information is missed by the traditional method; the load is reduced to about a half by the new techniques while the completeness is retained; and the size of new discovered Chinese topology is 5 times as many as that discovered by CAIDA's skitter. Compared to skitter's topology, the more complete topology showe more observable dissortativity, weaker clustering, shorter distance and more disequilibrium of node traffic distribution. The heavy-tailed distribution of AS sizes is discovered.
出处
《通信学报》
EI
CSCD
北大核心
2007年第12期96-101,共6页
Journal on Communications
基金
国家自然科学基金资助项目(60203021)~~
关键词
拓扑测量
拓扑建模
复杂网络
自治域规模
topology measurement
topology modeling
complex network
autonomous system size