Design and Implementation: Adaptive Active Queue Management Algorithm Based on Neural Network

Author

Zhenwei Meng ; Junlong Qiao ; Lijun Zhang

Author_Institution

Sch. of Comput. Sci. & Eng., Beihang Univ., Beijing, China

fYear

2014

fDate

15-16 Nov. 2014

Firstpage

104

Lastpage

108

Abstract

The packet queue management mechanism deployed on network intermediate device is an important means to prevent congestion. Existing queuing algorithms still need to be improved in control effect, response speed and stability, etc. In this paper, we present an adaptive active queue management algorithm GRPID based on classic PID controller, and incorporate the advantages of RBF neural network then validate the algorithm in NS2 Network simulator. In addition, we implement the algorithm in the Linux kernel and conduct performance measurement test.

Keywords

computer networks; radial basis function networks; three-term control; transport protocols; GRPID; Linux kernel; NS2 network simulator; PID controller; RBF neural network; TCP/IP; adaptive active queue management algorithm; congestion; network intermediate device; neural network; packet queue management mechanism; performance measurement test; response speed; stability; Adaptive systems; Algorithm design and analysis; Delays; Educational institutions; Kernel; Linux; Neural networks; AQM; Congestion control; GRPID algorithm; Linux kernel; NS2;

fLanguage

English

Publisher

ieee

Conference_Titel

Computational Intelligence and Security (CIS), 2014 Tenth International Conference on

Conference_Location

Kunming

Print_ISBN

978-1-4799-7433-7

Type

conf

DOI

10.1109/CIS.2014.104

Filename

7016862