Gain Adaptive Smith Predictor for Congestion Control in Robust Active Queue Management

Abstract

Active Queue Management (AQM) policies provide an early indication of incipient congestion to the source. In this paper, in terms of Lyapunov asymptotic stability principle, a robust AQM algorithm based on the gain adaptive Smith predictor and PI controller (GAS-PI algorithm) is designed by using the linearized TCP model with time delays. With the simple structure, good robustness and better performance, the proposed algorithm overcomes the negative impact on the queue stability caused by a large delay. It is illustrated by extensive and propound simulations that (i) the transient response of GAS-PI outperforms that of the other AQM schemes when the traffic load changes dynamically, (ii) the promising performance of GAS-PI holds up in the case of more realistic network scenarios, (iii) the better dynamic performance of GAS-PI is still maintained in a large-delay condition, (iv) the performance of GAS-PI is less sensitive to changes in system parameters, hence GAS-PI works well for a wide range of system parameters.