Abstract: A new algorithm for AQM is proposed based on the Smith predictor and Fuzzy2PI control(SFuzzy2PI). SFuzzy2PI combines PI controller with fuzzy controller. When congestion level is high, this paper uses fuzzy controller to improve the quick convergence speed and strong robustness. And when congestion level is lower, this paper uses PI controller to improve the stability. As the network congestion control system is essentially a delay system, so it brings up Smith predictor to overcome the negative impact on stability by large delay. The network simulation (NS2) results show that the transient response and tracking capability of the SFuzzy2PI is better, which is more suitable for the changing and large delay network system.

Key words: Active Queue Management(AQM), congestion control, Smith predictor, fuzzy logic control, proportional integral control