一种基于BP神经网络预测的RED改进算法  

An Improved RED Algorithm Based on BP Neural Network Prediction

作　　者：冯思楠 康巧燕[1] 王建峰[1] 李静 于必成 FENG Sinan;KANG Qiaoyan;WANG Jianfeng;LI Jing;YU Bicheng(Institute of Information and Navigation,AirForce Engineering University,Xi’an 710003,China)

机构地区：[1]空军工程大学信息与导航学院,陕西西安710003

出　　处：《计算机与网络》2019年第3期68-71,共4页Computer & Network

基　　金：国家自然科学基金资助项目(61201209)

摘　　要：主动队列管理技术是网络拥塞控制的重要技术,随机早期检测算法作为一种典型的主动队列管理机制,在拥塞控制中发挥着一定作用,针对该算法参数固定、难以适应复杂网络环境的问题利用BP神经网络算法进行流量预测,与随机早期检测算法结合,设计了一种基于BP神经网络预测的网络拥塞控制算法,及时有效地获取网络状态信息为拥塞控制提供应对时间,并通过Matlab仿真,检测新算法性能。Active queue management technology is an important technology of network congestion control,random early detection algorithm as a kind of typical active queue management mechanism,play a certain role in congestion control,according to the algorithm parameters are fixed,it is difficult to adapt to complex network environment,this paper use BP neural network traffic prediction algorithm,combined with random early detection algorithm,we design a network congestion control algorithm based on BP neural network prediction,timely and effective network status information provides the response time for congestion control,and through Matlab simulation,test the new algorithm performance.

关 键 词：队列管理 随机早期检测算法 BP神经网络 MATLAB 

分 类 号：TP31[自动化与计算机技术—计算机软件与理论]