基于PID神经元网络和内模控制的拥塞控制算法

Congestion control algorithm based on PID neural network and internal model control

针对网络系统的大时滞和非线性特性,设计了一种新的拥塞控制算法,将PID神经元网络与内模控制相结合应用于主动队列管理中,并使用Lyapunov理论证明了此算法的稳定性。NS仿真结果表明,这种算法的稳态和瞬态性能都优于PID算法,并且在参数变化和负载扰动时具有很强的鲁棒性。

This paper designed a new congestion control algorithm for large delay and nonlinear network systems. Applied pro- portional integral differential （PID） neural network controller and internal model control（IMC） in active queue management （AQM）. Proved stability by Lyapunov theory. The simulation results show that this algorithm＇ s stability and transient perform- ance are superior to PID algorithm. Moreover it possesses high robustness even when system parameter changes or network load fluctuates.