Using fuzzy-PI controller in active queue management

In this paper we propose using fuzzy logic to improve the performance of PI controller in the design of AQM (active queue management). With the introduction of integral factor in PI controller, the steady state error in proportional controller (such as RED) is eliminated. However, the response speed is slowed down. We design a fuzzy-PI (FPl) controller to solve this problem. FPI controller combines the advantages of fuzzy control while maintaining the simplicity and robustness of a conventional PI controller. The performance of FPI is verified and compared with Pl controller using ns-2 simulations. It is suggested that FPI is superior to Pl in response speed. Thus FPI is more robust in the presence of disturbance caused by the non-responsive UDP flows and the changing number of active flows.