Sliding mode controller for the binary ABR flow control

A moving sliding surface based on RBF neural network is presented for the binary ABR flow control. Unlike the conventional sliding mode control, the moving sliding surface moves to the desired sliding surface from the initial condition and thus fast tracking can be obtained. In addition, the equivalent control law is presented to control over the oscillation of the queue length in the binary ABR flow control. The simulation results show that not only the sliding mode controller can restrain the oscillation of the queue length in the binary ABR flow control, but also the system has strong robustness.