Intelligent Parameter Tuning Using Deep Q-Network for RED Algorithm in Adaptive Queue Management Systems<bold> </bold>

被引：0

作者：

Basheer, Ayman ^{[1
]}

Hassan, Hassan Jaleel ^{[1
]}

Muttasher, Gaida ^{[1
]}

机构：

[1] Univ Technol Baghdad, Dept Comp Engn, Baghdad, Iraq

来源：

MICRO-ELECTRONICS AND TELECOMMUNICATION ENGINEERING, ICMETE 2021 | 2022年 / 373卷

关键词：

Adaptive queue management; AQM; RED; Network congestion; Network traffic management; Deep Q-network; Reinforcement learning<bold>; </bold>;

D O I：

10.1007/978-981-16-8721-1_42

中图分类号：

TP39 [计算机的应用];

学科分类号：

081203 ; 0835 ;

摘要：

Network traffic is growing with every passing day, making it more critical than ever to deal with the exploding amounts of Internet traffic and reduce delays; thus, self-learning network management systems are a must for efficient network management; such systems are called active queue management systems or AQM. This paper discusses the use of machine learning to auto-tune the parameters of AQM algorithms by training a deep reinforcement learning model to balance the queuing delay and through putto get the maximum possible network score which is also known as the reward system. Deep Q-Network (DQN) is used as the foundation to control and auto-tune the parameters of the RED algorithm. Results from the NS3 simulation suggest that the DQN algorithm has better network reliability and is thus preferable to the RED active queue control algorithm.<bold> </bold>

引用

页码：439 / 446

页数：8