SINET: Enabling scalable network routing with deep reinforcement learning on partial nodes

Penghao Sun; Junfei Li; Zehua Guo; Yang Xu; Julong Lan; Yuxiang Hu

doi:10.1145/3342280.3342317

SINET: Enabling scalable network routing with deep reinforcement learning on partial nodes

Penghao Sun
, Junfei Li
, Zehua Guo
, Yang Xu
, Julong Lan
, Yuxiang Hu

School of Automation

National Digital Switching System Engineering and Technological R and D Center
Peng Cheng Laboratory
Fudan University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

32 Citations (Scopus)

Abstract

In this paper, we propose SINET, a scalable and intelligent network control framework for routing optimization. SINET uses the idea of partial control to collect network information from critical nodes and uses Deep Reinforcement Learning (DRL) to dynamically optimizes routing policies based on the collected network information. Simulation results show that SINET can reduce the average flow completion time and exhibit better robustness against minor topology changes, compared to existing DRL-based schemes.

Original language	English
Title of host publication	SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019
Publisher	Association for Computing Machinery
Pages	88-89
Number of pages	2
ISBN (Electronic)	9781450368865
DOIs	https://doi.org/10.1145/3342280.3342317
Publication status	Published - 19 Aug 2019
Event	2019 ACM SIGCOMM Conference Posters and Demos,SIGCOMM Posters and Demos 2019, Part of SIGCOMM 2019 - Beijing, China Duration: 19 Aug 2019 → 23 Aug 2019

Publication series

Name	SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019

Conference

Conference	2019 ACM SIGCOMM Conference Posters and Demos,SIGCOMM Posters and Demos 2019, Part of SIGCOMM 2019
Country/Territory	China
City	Beijing
Period	19/08/19 → 23/08/19

Keywords

Deep reinforcement learning
Pinning control
Routing optimization
Software-defined networking

Access to Document

10.1145/3342280.3342317

Cite this

Sun, P., Li, J., Guo, Z., Xu, Y., Lan, J., & Hu, Y. (2019). SINET: Enabling scalable network routing with deep reinforcement learning on partial nodes. In SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019 (pp. 88-89). (SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019). Association for Computing Machinery. https://doi.org/10.1145/3342280.3342317

Sun, Penghao ; Li, Junfei ; Guo, Zehua et al. / SINET : Enabling scalable network routing with deep reinforcement learning on partial nodes. SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019. Association for Computing Machinery, 2019. pp. 88-89 (SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019).

@inproceedings{c40c4e0f67504c3d9f187258954b94b2,

title = "SINET: Enabling scalable network routing with deep reinforcement learning on partial nodes",

abstract = "In this paper, we propose SINET, a scalable and intelligent network control framework for routing optimization. SINET uses the idea of partial control to collect network information from critical nodes and uses Deep Reinforcement Learning (DRL) to dynamically optimizes routing policies based on the collected network information. Simulation results show that SINET can reduce the average flow completion time and exhibit better robustness against minor topology changes, compared to existing DRL-based schemes.",

keywords = "Deep reinforcement learning, Pinning control, Routing optimization, Software-defined networking",

author = "Penghao Sun and Junfei Li and Zehua Guo and Yang Xu and Julong Lan and Yuxiang Hu",

note = "Publisher Copyright: {\textcopyright} 2019 Association for Computing Machinery.; 2019 ACM SIGCOMM Conference Posters and Demos,SIGCOMM Posters and Demos 2019, Part of SIGCOMM 2019 ; Conference date: 19-08-2019 Through 23-08-2019",

year = "2019",

month = aug,

day = "19",

doi = "10.1145/3342280.3342317",

language = "English",

series = "SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019",

publisher = "Association for Computing Machinery",

pages = "88--89",

booktitle = "SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019",

}

Sun, P, Li, J, Guo, Z, Xu, Y, Lan, J & Hu, Y 2019, SINET: Enabling scalable network routing with deep reinforcement learning on partial nodes. in SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019. SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019, Association for Computing Machinery, pp. 88-89, 2019 ACM SIGCOMM Conference Posters and Demos,SIGCOMM Posters and Demos 2019, Part of SIGCOMM 2019, Beijing, China, 19/08/19. https://doi.org/10.1145/3342280.3342317

SINET: Enabling scalable network routing with deep reinforcement learning on partial nodes. / Sun, Penghao; Li, Junfei; Guo, Zehua et al.
SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019. Association for Computing Machinery, 2019. p. 88-89 (SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - SINET

T2 - 2019 ACM SIGCOMM Conference Posters and Demos,SIGCOMM Posters and Demos 2019, Part of SIGCOMM 2019

AU - Sun, Penghao

AU - Li, Junfei

AU - Guo, Zehua

AU - Xu, Yang

AU - Lan, Julong

AU - Hu, Yuxiang

PY - 2019/8/19

Y1 - 2019/8/19

N2 - In this paper, we propose SINET, a scalable and intelligent network control framework for routing optimization. SINET uses the idea of partial control to collect network information from critical nodes and uses Deep Reinforcement Learning (DRL) to dynamically optimizes routing policies based on the collected network information. Simulation results show that SINET can reduce the average flow completion time and exhibit better robustness against minor topology changes, compared to existing DRL-based schemes.

AB - In this paper, we propose SINET, a scalable and intelligent network control framework for routing optimization. SINET uses the idea of partial control to collect network information from critical nodes and uses Deep Reinforcement Learning (DRL) to dynamically optimizes routing policies based on the collected network information. Simulation results show that SINET can reduce the average flow completion time and exhibit better robustness against minor topology changes, compared to existing DRL-based schemes.

KW - Deep reinforcement learning

KW - Pinning control

KW - Routing optimization

KW - Software-defined networking

UR - https://www.scopus.com/pages/publications/85071915168

U2 - 10.1145/3342280.3342317

DO - 10.1145/3342280.3342317

M3 - Conference contribution

AN - SCOPUS:85071915168

T3 - SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019

SP - 88

EP - 89

BT - SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019

PB - Association for Computing Machinery

Y2 - 19 August 2019 through 23 August 2019

ER -

Sun P, Li J, Guo Z, Xu Y, Lan J, Hu Y. SINET: Enabling scalable network routing with deep reinforcement learning on partial nodes. In SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019. Association for Computing Machinery. 2019. p. 88-89. (SIGCOMM 2019 - Proceedings of the 2019 ACM SIGCOMM Conference Posters and Demos, Part of SIGCOMM 2019). doi: 10.1145/3342280.3342317

SINET: Enabling scalable network routing with deep reinforcement learning on partial nodes

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this