基于EKF的神经网络在网络流量预测中的应用

在网络资源有限的情况下,建立合理的网络流量预测模型,并根据其预测结果及时做出控制决策或调整措施,对网络性能和服务质量的提高均有重要意义.根据网络流量的时变、非线性特点建立一个时间相关的流量预测模型,预测和分析网络流量状况,并利用人工神经网络在非线性建模方面的优势,给出了基于EKF算法的前馈神经网络的结构设计及学习算法.最后在Matlab环境下使用该预测模型对网络流量进行了仿真,结果表明该模型具有较好的自适应性和较高的预测精度.

Network Traffic Prediction Based on EKF Neural Network

As the resource of network is limited,it is significant to improve the network performance and quality of service by establishing a reasonable model of network traffic,and make timely decision or adjust in accordance with the result of prediction.The traffic of network is a time-varying and non-linear system,and the Artificial Neural Networks（ANN） have advantages to handle non-linear system.A model of time-dependent ANN is established to predict the network traffic and give the design of structure and learning algorithm based on Extended Kalman Filter（EKF）algorithm. The result of MATLAB simulation indicates that the model has good adaptability and high accuracy of prediction.