网络时延的混沌特性分析及预测  

Chaotic Analysis and Prediction of Internet Time-Delay

作　　者：曹庆璞[1] 董淑福[1] 罗赟骞[1] 

机构地区：[1]空军工程大学电讯工程学院,陕西西安710077

出　　处：《计算机技术与发展》2010年第4期43-46,共4页Computer Technology and Development

基　　金：国家综合业务网理论及关键技术开放基金(ISN-9-08)

摘　　要：针对互联网的传输时延处于不确定状态的问题,从网络时延序列表现出的性质出发,将混沌理论引入对网络时延的特性分析。通过对实际网络时延数据的处理,计算验证了网络时延具有混沌特性。并提出了采用混沌—支持向量机回归模型对互联网络时延进行预测的方法,通过实例分析对该方法进行了验证。分析结果证明了混沌—支持向量机模型可以对网络时延的变化趋势进行较为准确的预测,和BP神经网络回归预测相比,混沌—支持向量机对时延的短期预测具有更高精度。Aiming at the problem of indeterminacy of Internet time- delay, based on its character, the chaos theory was implemented. By analyzing the actual time- delay data, the validity of its chaotic character was demonstrated. Then a method of time- delay prediction using the chaos-support vector rnachine（SVM） model was put forward, and the feasibility of the method was validated by practical application. Results of the analysis indicated that the chaos- SVM model could predict the changing trend of network time - delay accurately. Compared to the BP neural network prediction, short - term prediction using the chaos- SVM model has a higher accurecy.

关 键 词：网络管理 时延 预测 混沌 支持向量机 

分 类 号：TP393.06[自动化与计算机技术—计算机应用技术]