基于域的无响应流的控制

Domain-based Unresponsive-flow Control

现有的无响应流控制方法存在某些缺陷,如通信代价过高,或者适用面过于狭窄。在权衡通信代价和核心路由器计算量的基础上,文章提出了一种简单易实现的方法。该方法主要由两部分组成:基于域的反馈机制和丢弃数据包驱动的AIMD模拟机制。由于算法不需要实时计算RTT值,因而有着较广的应用范围。通过模拟实验,该方法在FIFO和RED队列调度模式下的有效性和稳定性得到了证实。

Methods for unresponsive flow regulation either have high communication costs or have very narrow applications. In order to solve these difficulties, a simple method is put forward in this paper, which is a balance between communication costs of network and computational costs of core touters. This method is composed of two major parts： domain-based feedback mechanism and drop-driven AIMD simulation mechanism. Since no real time RTr calculation is needed, this method has wide applications. By simulation experiments, the effectiveness and stability of this method under FIFO and RED queuing have been verified.