Robust control of congestion in computer networks: An adaptive fractional-order approach

In this study, the congestion problem of computer networks is solved by introducing an adaptive fractional-order controller. The controller is designed to enhance the efficiency of nonlinear transmission control protocol/active queue management (TCP/AQM) in these networks. External disturbances can perturb the data transmission. Moreover, factors such as undetermined link capacity can affect the congestion control problem. Hence, the proposed controller should guarantee robustness against disturbances and uncertainties. Besides, small elapsed time to reach the desired queue length as well as convergence of tracking error to zero are essential in queue management. These objectives are fulfilled by designing a fractional-order controller. High tracking capability and robustness make the controller an effective method for TCP/AQM networks. Furthermore, asymptotic stability of the network is precisely proven based on the fractional-order Lyapunov lemma. A variety of simulations are performed to examine capability of the proposed method confronting disturbances and uncertainties while providing a fast and stable response with zero tracking error. The results are also compared with the results of two recently developed control approaches for congestion problem, using performance indices, which confirm the efficiency and superiority of the introduced controller.