摘要
在存在用户数据报协议业务流的情况下,为了有效地保护传输控制协议(TCP)流,并保证业务流的公平性,提出了一种自适应的流随机早期检测(FRED)缓存管理算法AFRED(Adaptive FRED).新算法通过对分组丢弃门限和丢弃概率的动态调节,保护了TCP流,提高了系统的公平性和带宽利用的有效性.分析和仿真表明,算法的复杂度与FRED相近,但是能获得更好的性能.
An adaptive fair queue management algorithm-adaptive flow random early detection （AFRED） is proposed to protect transfor control protocal（TCP） flows in the presence of user data protocal （UDP） flows and to provide flow fairness. By dynamically varying the threshold of each flow, that could be allowed into the buffer and the drop probability, the algorithm can protect TCP flows effectively and improve the fairness of the system and the efficiency of the bandwidth utilization. The analysis and simulations indicate that the implementation complexity of AFRED is comparable to flow random early detection （FRED）, but its performance is better than FRED.
出处
《北京邮电大学学报》
EI
CAS
CSCD
北大核心
2006年第1期88-91,共4页
Journal of Beijing University of Posts and Telecommunications
关键词
随机早期检测
流随机早期检测
缓存管理
random early detection
flow random early detection
buffer management
