A new self-tuning Active Queue Management algorithm based on adaptive control

Abstract

Most Active Queue Management (AQM) algorithms based on control theory have difficulty in obtaining desirable performance once the network conditions or the traffic patterns change out of the presumed ones they are designed for. To address these problems, a new self-tuning AQM algorithm called STR is proposed in this paper. STR has the ability of keeping minimum variance between the instantaneous queue length of the router and the reference value by estimating the parameters of the model of controlled object online and adjusting the packet drop probability accordingly. The performance of STR is evaluated through extensive simulations. The results show that STR is robust against the great changes of the network parameters and the traffic load. © IFIP International Federation for Information Processing 2005.