Congestion and Critical Fluctuation in the Internet Information Traffics and Control of Complex Systems

在本研究中，我分析了详细的流量数据的信息流在互联网上。结果表明，有两个阶段：一个是拥塞阶段，其中的信息包很可能会丢失和重新提交。另一个是非拥塞阶段，在这个阶段中，数据包运行平稳。相位由输入的信息量决定。在信息流的临界量下，实现了1/f波动和幂律分布等临界行为。我做了一个理论公式来描述这些发现。结果表明，交通系统在干扰与非干扰边界的临界状态下具有最大的效率，并在此基础上提出了改进协议以提高信息系统效率的建议。

In this research I analyzed detail traffic data of information flow in the Internet. It is shown that there are two phases : One is the congestion phase in which information packets are likely to be lost and re-submitted. The other is the non-congestion phase in which packets run smoothly. The phase is determined by the amount of information input. At the critical amount of information flow critical behaviors such as the 1/f fluctuations and power law distributions are realized. I made a theoretical formulation describing these findings. It is shown that a traffic system has the maximum efficiency at the critical state that is the border of jamming and non-jamming.On the basis of these results, I suggested an improvement of the protocol to raise efficiency of the information system.