摘要
主动队列管理(AQM)PID(Proportional integral derivative)算法的控制效果取决于比例、积分及微分系数的整定,但传统整定往往基于试凑方法和经验知识。根据Ad Hoc网络参量时变的特点,推导Ad Hoc网络的TCP/AQM模型,利用遗传算法动态调整RBF(Radial Basis Function)神经网络PID控制器系数,提出基于遗传算法的RBF神经网络PID-AQM。仿真表明,相较RBF-PID,新算法在信道状态复杂的Ad Hoc网络健壮性更好,并具有较好的队列控制效果。
The control effect of active queue management (AQM) PID algorithm depends on the tuning of the coefficients of proportional, integral and differential, bilt traditional tuning is usually based on experiences and trial methods. According to the time-varying characteristic of Ad Hoc network parameter, in this paper we deduce the TCP/AQM model of Ad Hoc network. Then we use genetic algorithm to dynamically adjust the PID controller coefficients in RBF neural network. Finally, we propose a genetic algorithm-based RBF neural network PID-AQM. Simulations demonstrate that compared with RBF-PID, the new algorithm has better robustness in Ad Hoc network with complicated channel condition and has better queue management effect as well.
出处
《计算机应用与软件》
CSCD
北大核心
2012年第7期29-32,39,共5页
Computer Applications and Software
基金
国家自然科学基金项目(60874021)
江苏省现代教育技术研究课题(2011-R-19876)