中文摘要：

对控制报文和网络拥塞间的平衡问题进行研究。通过一个单服务队列模型来描述拥塞控制策略,利用排队系统中的马尔可夫过程,提出一种两阈值的流量控制算法使其控制报文速率能满足最好的拥塞概率。通过分析发现排队系统中拥塞概率随缓冲区大小变化发生指数衰变,并定义该衰变指数为大偏差指数用来描述控制报文与拥塞概率间的比例。最后通过带宽共享模型,模拟并分析不同带宽情况下控制报文与拥塞概率间的最佳比例及其大偏差指数。

英文摘要：

The balance between control message and network congestion control is studied. The congestion control strategy is described with a single service queue model. A two-threshold flow control algorithm is put forward by utilizing Markov process to make the control message rate satisfy the optimal congestion probability. It is found by analysis that the congestion probability occurs exponential disintegration with the buffer size,which is defined as the large deviation index to describe the ratio of control message and congestion probability. The ratio and large deviation index in different bandwidth are simulated and analyzed with bandwidth sharing model.