中文摘要：

A surge number of models has been proposed to model the Internet in the past decades.However,the issue on which models are better to model the Internet has still remained a problem.By analysing the evolving dynamics of the Internet,we suggest that at the autonomous system (AS) level,a suitable Internet model,should at least be heterogeneous and have a linearly growing mechanism.More importantly,we show that the roles of topological characteristics in evaluating and differentiating Internet models are apparently over-estimated from an engineering perspective.Also,we find that an assortative network is not necessarily more robust than a disassortative network and that a smaller average shortest path length does not necessarily mean a higher robustness,which is different from the previous observations.Our analytic results are helpful not only for the Internet,but also for other general complex networks.

英文摘要：

A surge number of models has been proposed to model the Internet in the past decades. However, the issue on which models are better to model the Internet has still remained a problem. By analysing the evolving dynamics of the Internet, we suggest that at the autonomous system （AS） level, a suitable Internet model, should at least be heterogeneous and have a linearly growing mechanism. More importantly, we show that the roles of topological characteristics in evaluating and differentiating Internet models are apparently over-estimated from an engineering perspective. Also, we find that an assortative network is not necessarily more robust than a disassortative network and that a smaller average shortest path length does not necessarily mean a higher robustness, which is different from the previous observations. Our analytic results are helpful not only for the Internet, but also for other general complex networks.