一种基于卡尔曼滤波的队列长度自适应算法

An adaptive queue length algorithm based on Kalman filtering

传统主动队列管理(AQM)算法在处理传感器网络突发流时具有响应速度慢、抗网络突变性能弱的缺点。针对此问题,提出了一种新的AQM算法,算法首先将队列长度作为早期拥塞检测参量,运用卡尔曼滤波理论预测队列长度;其次根据队列长度在缓冲区的占用比来划分网络状态;最后根据不同占用比采取相应的丢包策略,自适应地调整丢包率,当出现网络突变时,加大调整幅度,使队列长度保持在理想区间。仿真实验表明:新算法能够较好地适应网络波动,提高网络服务质量(QoS),算法综合性能优于主流AQM算法。

Traditional active queue management （AQM）algorithm has shortcomings of slow response and weak performance of dealing with sudden flow in sensor networks. To solve this problem, propose a new AQM algorithm,firstly the algorithm put queue length as an early congestion detection parameters, predict the queue length by Kalman filtering theory;secondly the algorithm divides network status according to occupancy ratio of queue length in buffer zone;finally, the algorithm takes corresponding packet loss strategy according to different occupancy ratio ,adjust packet loss rate adaptively, when network mutation occurs, algorithm increase adjustment amplitude to make queue length remains at desired interval. Simulation results show that the new algorithm can adapt to network fluctuations better and improve network quality of service（QoS） , comprehensive performance of the new algorithm is better than mainstream AQM algorithms.