摘要
针对流量行为分析对自动化、智能化和预测精度的进一步需求和存在的不足 ,提出了基于遗传程序设计(GP)的IP业务流量长期预测算法 .论文首先给出了GP的定义和GP个体的二叉树表示 ,然后重点论述了GP的演化算法设计流程、数据样本获取、GP的演化建模和实验 .实验结果表明 ,通过对Web业务流量进行GP建模实验 ,搜索到了两个较好的业务流量预测模型 ,模型所对应的预测曲线与原始数据曲线拟合得较好 ,说明采用GP建模是有效的 ,能反映业务流量的变化规律 .
To improve the automation, intelligence and precision of IP network traffic prediction, this paper proposes a genetic programming (GP)-based modeling algorithm to predict the long-range prediction of IP network traffic. This paper first gives the definition of GP, then focuses on GP individuals’ description structure, design flow chart of GP evolutional algorithm, get data sample acquization, and GP evolutional modeling and experiment respectively. By experimenting on Web traffic where GP is employed to optimize the model structure, we get two vigorous traffic prediction models. It is shown that the prediction curves of the models fit the original data’s curve very well, which explains that modeling on the basis of GP is valid and can reflect the change law of IP network traffic.
出处
《计算机学报》
EI
CSCD
北大核心
2003年第12期1786-1790,共5页
Chinese Journal of Computers
基金
国家自然科学基金 (60 1 32 0 30
699830 0 5)资助