基于多尺度小波变化的自相似网络流量预测  

Prediction of self-similar network traffic based on multi-scale DWT analysis

作　　者：何凯霖[1] 丁晓峰[1] 

机构地区：[1]电子科技大学成都学院计算机系,四川成都611731

出　　处：《计算机工程与设计》2015年第4期915-919,961,共6页Computer Engineering and Design

摘　　要：针对现有自相似网络的长相关性不稳定问题,以离散小波变换(DWT)多尺度分析为依据建立网络流模型并进行流量预测。把利用该模型计算得到的预测流量相似性程度值Hurst引入路由器的主动队列管理算法(AQM),采用AQM算法形成的严谨响应方案对异常估计流量进行处理,提高动态流量控制的有效性,实现网络协议的优化。仿真结果表明,该方案可以保证拥塞控制的实现,具有稳定性和可扩展性。For the problem of correlation length instability of current self-similar network,based on the DWT multi-scale analysis,the network flow model was established and flow projections were conducted.Predictive value Hurst of the degree of similarity between the flows was calculated using the model and introduced into AQM.For the anomalous estimated flow,stringent response scheme formed by AQM algorithm was adopted,which improved the effectiveness of dynamic flow control,network protocols were improved.Simulation results show that this scheme can guarantee the realization of congestion control with the stability and scalability.

关 键 词：离散小波变换 主动队列管理算法 流量控制 自相似网络 协议优化 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]