Research On Active Queue Management Algorithm Based On Fuzzy Control In Wireless Network

The wireless network congestion is deteriorating with the development of wireless network and the surge in the number of users. Active Queue Management (AQM) has become an effective way to relieve wireless networks congestion. As a great change of end-to-end congestion control, it aims to modulate packet loss rate dynamically and take place of "Tail-Drop" of the traditional technology. Currently research on AQM and derived algorithms based on fuzzy control theory has become a hot point.In this thesis, an improved algorithm E-IMBA based on Intelligent Management Buffer Algorithm (IMBA) is proposed. The algorithm makes full use of fuzzy logic theory and divides four congestion states by the occupancy of queue buffer. Threshold is set in the queue number of serious congestion state in the system, and different strategies are used to joint both global and local loss algorithms to control packet loss rate. The simulation result shows that proper threshold can reduce the global repeated delivery rate effectively with the occupy rate of buffer maintaining at the same level.A novel F-PERED algorithm based on fuzzy theory and pre-estimated RED is also presented. To avoid the drawbacks that PERED does not distinguish the traffic flows, this algorithm amends packet loss rate in it by introducing the load-factor to obtain load control gain. The simulation result indicates that F-PERED further decreases or increases the loss rate after taking load-factor into account, which can relieve congestion rapidly by the differences between flows. It proves to have excellent self-adaptation and fairness.