基于稳态过程的多重分形Web日志仿真生成算法

发布时间：2018-04-18 16:42

本文选题：稳态过程 + 多重分形　；参考：《计算机应用》2017年02期

【摘要】：运行在服务器集群的软件系统需要Web日志的大规模数据集以满足性能测试的需求,但现有仿真生成算法因模型单一而无法满足要求。针对此问题,提出一种基于alpha稳态过程的多分形Web日志的仿真生成算法。首先,在长相关尺度(LRD)下采用alpha稳态过程来描述Web日志的自相似性;其次,在短相关尺度(RSD)下采用二项式b模型描述Web日志的多重分形性;最后,将长相关模型和短相关模型融合于改进的ON/OFF框架中。与单一的模型相比,新算法的参数物理意义明确,具有良好的自相似性和多分形性。实验结果表明,该算法能够较准确地模拟真实Web日志,可以有效地应用于Web日志大规模数据集的仿真生成。
[Abstract]:The software system running in the server cluster needs a large data set of Web logs to meet the requirements of performance testing, but the existing simulation generation algorithms can not meet the requirements because of the single model.To solve this problem, a simulation algorithm for generating multifractal Web logs based on alpha steady-state process is proposed.Firstly, the self-similarity of Web logs is described by alpha steady-state process under long correlation scale. Secondly, the binomial b model is used to describe the multifractal properties of Web logs.Long correlation model and short correlation model are integrated into the improved ON/OFF framework.Compared with the single model, the new algorithm has clear physical meaning of parameters, good self-similarity and multifractal.The experimental results show that the algorithm can simulate the real Web log accurately and can be used to generate large scale data sets of Web logs effectively.
【作者单位】：福建师范大学软件学院;福建省公共服务大数据挖掘与应用工程技术研究中心;
【基金】：福建省高校产学合作项目(2016H6007)~~
【分类号】：TP393.09;TP301.6

【参考文献】