LFQ: Online Learning of Per-flow Queuing Policies using Deep Reinforcement Learning

被引:2
作者
Bachl, Maximilian [1 ]
Fabini, Joachim [1 ]
Zseby, Tanja [1 ]
机构
[1] Tech Univ Wien, Vienna, Austria
来源
PROCEEDINGS OF THE 2020 IEEE 45TH CONFERENCE ON LOCAL COMPUTER NETWORKS (LCN 2020) | 2020年
关键词
D O I
10.1109/LCN48667.2020.9314771
中图分类号
TP3 [计算技术、计算机技术];
学科分类号
0812 ;
摘要
The increasing number of different, incompatible congestion control algorithms has led to an increased deployment of fair queuing. Fair queuing isolates each network flow and can thus guarantee fairness for each flow even if the flows' congestion controls are not inherently fair. So far, each queue in the fair queuing system either has a fixed, static maximum size or is managed by an Active Queue Management (AQM) algorithm like CoDel. In this paper we design an AQM mechanism (Learning Fair Qdisc (LFQ)) that dynamically learns the optimal buffer size for each flow according to a specified reward function online. We show that our Deep Learning based algorithm can dynamically assign the optimal queue size to each flow depending on its congestion control, delay and bandwidth. Comparing to competing fair AQM schedulers, it provides significantly smaller queues while achieving the same or higher throughput.
引用
收藏
页码:417 / 420
页数:4
相关论文
共 13 条
  • [1] Bachl M., 2019, WORKSH BUFF SIZ ACM WORKSH BUFF SIZ ACM
  • [2] Bachl M, 2019, IEEE ICC
  • [3] Bisoy S. K., 2017, ANTS
  • [4] Bless R., 2018, IFIP NETWORKING
  • [5] BBR: Congestion-Based Congestion Control
    Cardwell, Neal
    Cheng, Yuchung
    Gunn, C. Stephen
    Yeganeh, Soheil Hassas
    Jacobson, Van
    [J]. COMMUNICATIONS OF THE ACM, 2017, 60 (02) : 58 - 66
  • [6] Dong M, 2018, PROCEEDINGS OF THE 15TH USENIX SYMPOSIUM ON NETWORKED SYSTEMS DESIGN AND IMPLEMENTATION (NSDI'18), P343
  • [7] Dumazet Eric., 2013, Pkt_sched: Fq: Fair Queue Packet Scheduler
  • [8] Fejes F., 2019, WORKSH BUFF SIZ ACM WORKSH BUFF SIZ ACM
  • [9] Kim M, 2019, DEEP REINFORCEMENT L
  • [10] Lin X., 2015, CHINACOM