COMMUNICATIONS
T-Comm Tом 10. #1-2016
24
EXPERIMENTAL RESEARCH OF TRAFFIC CONTROL APPLICATION
IN LINUX-ROUTER QUEUE DISCIPLINE BASED ON FUZZY LOGIC
Maslennikov Andrey Gennad'evich,
external PhD student, faculty "Communication networks and switching systems",
Moscow Technical University of Communications and Informatics, Moscow, Russia,
amasl2048@gmail.com
Abstract
In the article is investigated fuzzy logic controller (FLC) application for queue discipline in Linux-router.
Fuzzy logic con-
troller is widely used for control complicated processes with non-linear dynamics and well fit to traffic processing tasks in
routers' queue in network overload mode. Due to FLC it is possible to prevent overload beginning
and automatically select
proper value of drop or marking probability for packets in a router, in order to control incoming
traffic intensity with a
big quantity of TCP-sessions and limited data transmission channel rate. The possibility of FLC to hold queue length on
reference level will allow to planning of packets delay in a queue and decrease jitter. Joint usage
of FLC method with explic-
it congestion notification (ECN) technique allows to decrease packets loss to minimal values due to marking of packets
instead of packets drop. Open system application based on Linux operating system gives
possibility to implement own
queue discipline in a router and take advantage of existing in Linux IP traffic control technologies. FLC is loading to a router
as a developed software module for Linux kernel. Up to 100 simultaneously TCP-sessions plus
UDP traffic with constant
bit rate are transmitted over bottleneck link. The developed FLC method showed ability to hold queue length near refer-
ence level and prevent queue overloading or underloading despite of link overload.
Do'stlaringiz bilan baham: 
