一种基于粒子群的实际业务流预测算法  

An Actual Traffic Prediction Algorithm Based on Particle Swarm

作　　者：李文泽[1] 盛光磊[2] 

机构地区：[1]河南广播电视大学,河南郑州450008 [2]郑州升达经贸管理学院,河南郑州451191

出　　处：《微电子学与计算机》2014年第1期56-59,共4页Microelectronics & Computer

基　　金：国家自然科学基金项目(10901144)

摘　　要：为了解决因业务流迅速增长而造成的网络拥塞现象,提出了一种新的业务流预测算法PSATP(Particle Swarm-ARIMA model based Traffic Prediction).该算法首先利用ARIMA模型和Kalman滤波建立预测方法,并且结合粒子群进行优化,使预测精度得以提高.同时,以实际数据进行仿真实验,深入研究了时延、丢包率与缓冲区、利用率之间的关系.相比于其他算法,仿真结果表明PSATP算法具有较好性能.In order to mitigate network congestion caused by the rapid growth of traffic, a novel traffic prediction algorithm PSATP （Particle Swarm--ARIMA model based Traffic Prediction） is proposed. In this algorithm, the prediction method is presented by ARIMA model and Kalman filter at first, and the prediction accuracy is improved with particle swarm optimization. Then, a simulation with actual data was conducted to study the relationship between delay, dropping rate buffer and utilization rate. The simulation results show that, compared to other algorithm, PSATP has better performance.

关 键 词：拥塞 预测 精度 粒子群 ARIMA 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]