基于遗传程序设计的Internet业务流量分析与预测

A Real-Time Monitor System of Network Performance Based on the Traffic Engineering

由于现代网络信息传输的高突发性和高随机性,现有基于时间序列的分析与预测方法(如统计建模、回归分析、灰色建模、神经网络建模等)存在很大局限性。在指出了现有方法的局限性的基础上,为了着重解决Internet业务流量的建模与预测自动化问题,从网络行为具有自相似的特性出发,提出了对Internet业务流量时间序列的建模与预测采用遗传程序设计的自动化演化建模方法。

Because of high dynamics and high stochastic probability that deliver in modern network information transmission, current analysis and estimating methods according to time series of the row are still difficult to simulate the complicated Internet behavior at system level. Based on the similarity of network behavior, the genetic program design method is applied to establish an Internet business flow model. This method can easily search the aim function and optimize the parameters under the given model aim and function sets. This method also can predict automatically and intelligently, and effectively save the time of Internet business flow. The efficiency and accuracy of prediction are also enhanced.