提出了一种新颖的基于隐含类型的GM模型的预测PI算法。利用预测队列误差而非瞬时采样值来决定路由器的丢包率。该方法致力于在响应速度与增益两者之间取得一个最佳的折中。仿真结果验证了该灰色预测PI主动队列管理策略的有效性。
This paper proposes a novel predictive PI algorithm based on connotative type grey model.The predicted queue errors, while not the sample values,are utilized to determine the packet drop rate in the router.This method aims to achieve the satisfactory trade-off between fast responsiveness and low overshoot.Simulation results demonstrate the effectiveness of the grey predictive PI active queue management scheme.