Abstract: The explosive applications of Internet stimulate the research on the structure evolution of networks involved. The available methods only add entities and connections to the existing networks, which is totally unrealistic since the real Internet is dynamic due to the new equipments deletion and add. To attack this issue, it presents an algorithm that can take into consideration the dynamics in Internet via a preferential tendency probability of topology, in which time series, nodes life are included. The experimental results demonstrate that the algorithm is more realistic than traditional ones, particularly in the dynamic evolution of the Internet.

Key words: Internet topology, non-monotonic growth, preferential tendency probability, topology model

摘要： 互联网拓扑模型的构造，是网络进一步应用和发展的重要研究基础。在现有建模方法中，构成拓扑的节点和连接均为单调增长，显然无法体现网络的真实演化。通过分析长时间跨度的网络实测数据，提取出网络拓扑中节点和连接的消亡与新生规律。在此基础上，进一步修正了拓扑增长的优先倾向概率，并给出一种新的非单调增长的互联网拓扑模型。实验结果说明，在多项重要拓扑特征量方面，该模型均可以很好地体现互联网的动态演化过程。

关键词: 互联网拓扑, 非单调增长, 优先倾向概率, 拓扑模型