基于Smith预估的模糊PID主动队列管理算法研究

Research on fuzzy PID active queue management based on smith prediction

为解决网络拥塞控制系统中由于网络大时滞对主动队列管理算法产生不利影响的问题,提出了一种基于Smith预估的模糊PID主动队列管理算法。该算法将Smith预估控制与模糊控制相结合,利用Smith预估器补偿网络时滞,同时运用模糊控制在一定程度上克服了传统Smith预估器对模型结构与参数的精确性过于敏感、鲁棒性差的缺点,使主动队列管理算法控制性能有明显提高。仿真实验结果表明,该算法在大时滞的网络环境下能很好地将路由器队列长度收敛于期望值,并能适应突发流和非弹性业务流的干扰,适用于动态变化的网络环境。

To solve the problem of the negative impact on the performance of active queue management （AQM） in network caused by large delay, a fuzzy PID AQM algorithm is proposed based on Smith predictor. This algorithm combines fuzzy PID control and Smith predictor control, the Smith predictor is utilized to compensate the time-delay, while fuzzy PID is utilized to overcome the disadvantage of highly sensitive to precise model and parameter, lack of robustness in the traditional Smith predictor to a certain extent. The performance of control in large delay network is improved. The simulation result show that the new algorithm control the size of the queue in the buffer to an expected length. It could adapt the disturbance caused by burst and nonelastic traffic, and be suitable for dynamic network environment.