A robust RED algorithm based on time-delayed feedback control

Random early detection (RED) is the most popular active queue management (AQM) mechanism used in routers on the Internet to allow network operators to simultaneously achieve high throughput and low average delay. However, the RED algorithm may cause heavy oscillation of average queue length and induce network instability. Based on theoretical analysis and extensive network simulations, we show that the performance of RED can be greatly enhanced by adjusting a RED parameter via delayed feedback control.