【摘要】 结合复杂网络理论与CAIDA授权的关于Internet网络的真实海量数据,从复杂网络理论角度对真实的Internet数据进行分析与研究.首先借助物理学和生物学研究的方法,将Internet网络视为具有生命涨落特征的活体系统,形式化定义了Internet物理特征量———访问直径.然后根据目标复杂系统涨落演化特点,提出了3种基于Logistic模型的、以带衰减因子的正余弦函数组合模拟振荡涨落的数学模型.使用浮点型遗传算法分别进行拟合实验,并通过实验结果对上述3种模型进行优选.最终优选模型的拟合准确度为97.87%,预测准确度为97.47%,准确度高,符合Internet网络真实数据变化情况.文中使用模型对较远未来网络情况进行了预测,并得出结论:从现在开始至2011年12月,将是Internet网络高速发展时期,之后发展速度变缓,并于2021年10月左右趋于稳定,此时Internet网络访问直径为10.2073跳.最后,应用文中模型重点预测出了2008年8月北京奥运期间Internet网络访问直径为10.7726跳,并得出奥运期间Internet网络效率较高的结论.更多还原

【Abstract】 Based on the theory of complex networks and the giant data samples authorized by CAIDA, a research of real Internet samples is performed in view of complex networks theory. Firstly, Internet is regarded as an alive system with fluctuation property according to methods from Physics and Biology, and a formalized definition-Internet Traveling Diameter(ITD)- is put foreword. Secondly, according to features of real samples, three models simulating the development of ITD are put foreword. All three models include two parts, one part is Logistic function, which simulates the basic developing trace of ITD, and the other part is composed of sine and cosine functions, which simulates the fluctuations during the ITD’s development. Then Experiments are performed by Float-point GA to train and learn the parameters of the three models, and the most optimized model is selected with a fitting accuracy of 97.87% and a forecast accuracy of 97.47%. Finally, a forecast of Internet development is performed and a conclusion is made that Internet would develop at top speed from now on to Dec. 2011, and after that the speed would slow down and become stable in Oct. 2021 when ITD would be 10.2073 hops. What’s more, a forecast of Internet in Beijing Olympic Games in Aug. 2008 is performed and results are yielded that ITD would be 10.7726 hops and Internet validity would be in a high state at that time.更多还原