摘要
为了提高随机指数标记算法(REM)的响应能力和适应性,提出了一种基于自适应神经元的REM算法(ANREM)。采用具有比例积分微分(PID)控制属性的加强型价格来检测和控制网络拥塞。利用神经元构造PID价格,并设计控制参数的自适应调整机制,以增强算法在动态环境中的适应性。在NS2仿真平台中,将AN-REM与REM及其改进方法进行对比实验。结果表明,ANREM提高了队列长度的响应能力,增强了主动队列管理算法的适应性和鲁棒性。
To improve the responsiveness and adaptability of random exponential marking (REM) , this paper proposed a no- vel REM algorithm named ANREM based on adaptive neuron. ANREM employed an enhanced price with proportional-integral- derivative (PID) control property to detect and control network congestion. Introduced an adaptive neuron to design the PID- type price, in which adjusted the control parameters online to improve the adaptability in dynamic network scenarios. Conduc- ted simulation experiments in the NS2 platform to compare the performance of ANREM with those of REM and its variants. The results demonstrate that ANREM can enhance the responsiveness of queue length and improve the adaptability and robustness for active queue management.
出处
《计算机应用研究》
CSCD
北大核心
2011年第1期268-270,274,共4页
Application Research of Computers
基金
国家创新研究群体科学基金资助项目(60721062)
国家"863"计划资助项目(2007AA041301)
关键词
拥塞控制
主动队列管理
随机指数标记
自适应神经元
congestion control
active queue management (AQM)
random exponential marking (REM)
adaptive neuron