自适应卡尔曼滤波的主动队列管理算法

PID控制器通过微分环节加快了控制器的调节速度,但PID的参数是固定的,不能根据动态的网络自调整参数,故不能有效控制队列的稳定性。由于神经元网络有自适应性,提出了一种自适应卡尔曼滤波的主动队列管理算法(adaptive-KF-AQM)。它结合卡尔曼滤波和神经元网络方法,根据队列长度及其变化率来估计下一时刻的队列长度,使队列长度在期望值附近波动。仿真结果表明,该算法在队列稳定性、收敛速度、延时和链路利用率等方面都明显优于传统的PID算法。

New Adaptive Active Queue Management Algorithm with Kalman Filter

PID controller accelerates the regulation speed of the controller through differential factor.But the parameters of PID controller are fixed,they can’t be adapted with dynamic network,so the stability of the queue can’t be controlled effectively.A new adaptive active queue management(AQM) algorithm with Kalman filter was presented accor-ding to the adaptivity of the neural network.The new algorithm combines Kalman filter law with neural network,which has the merits of both.It can determinate future queue length based on queue lengths and some rates of change in the queue length.The results of simulation show that the new AQM algorithm is superior to the typical PID controller on the queue stability,time delay and link utilization.