A genetic algorithm for the design of a fuzzy controller for active queue management

摘要

Active queue management (AQM) policies are thoseudpolicies of router queue management that allow for the detection of network congestion, the notification of such occurrences to theudhosts on the network borders, and the adoption of a suitable controludpolicy. This paper proposes the adoption of a fuzzy proportionaludintegral (FPI) controller as an active queue manager for Internetudrouters. The analytical design of the proposed FPI controller isudcarried out in analogy with a proportional integral (PI) controller,udwhich recently has been proposed for AQM. A genetic algorithm isudproposed for tuning of the FPI controller parameters with respectudto optimal disturbance rejection. In the paper the FPI controlleruddesign metodology is described and the results of the comparisonudwith random early detection (RED), tail drop, and PI controllerudare presented.