Trade-off queuing delay and link utilization for solving bufferbloat

Hao Wang

Abstract: To solve the bufferbloat problem, active queue management (AQM) has been recommended but existing AQM algorithms suffer from poor adaptability to dynamic traffics. We develop a novel adaptive AQM algorithm named TODU for a better Trade-Off between queuing Delay and link Utilization. TODU aims to maintain a stable queue length against dynamical network traffics by adapting the drop probability to balance the expected link utilization and queuing delay. We compare TODU with benchmark algorithms in Network Simulator-2 and the results demonstrate that TODU achieves a better trade-off between low queuing latency and high link utilization in a dynamical environment.