A Power-Law Approach on Router-Level Internet Macroscopic Topology Modeling

摘要

With the giant samples of router-level Internet macroscopic topology, obvious power-law properties were found after approaches such as frequency-degree power-law, degree-rank power-law and CCDF(d)-degree power-law distributions were studied and applied. The frequency-degree power-law relationship is found to be with a power exponent of 2.1406. The degree-rank power-law, however, is found to have two phases with power-exponents of 0.29981 and 0.84639 respectively. And the CCDF(d)-degree power-law property was proved not to exist in the measured Internet topology after compared with the Weibull distribution. With these, especially the two-phase degree-rank power-law power-exponents, we began to construct an Internet topology model (IBA model) by improving and optimizing the traditional BA model to fit the measured samples. Algorithm for the IBA model was finally given.