Prediction-based Active Queue Management in the Internet

Random early detection (RED) is the most popular active queue management algorithm that is used by the Internet routers. This paper proposes a neuro-fuzzy controller which enhances the network performance by dynamically tuning of RED's maxp parameter. The controller first learns the network behavior against maxp variations and then adjusts maxp. Simulation results in ns-2 environment show that, the proposed learning RED, called LRED, keeps queue length and queuing delay in a pre-determined level and outperforms RED in terms of queue length and stability.

---

W artykule zaprezentowano sterownik neuro-fuzzy który poprawia dynamiczne strojenie system RED stosowanego do kolejkowania w Internecie. Proponowany uczący się algorytm nazwany LRED pozwala na utrzymanie długości kolejki i opóźnienia w założonych granicach.