计算机网络的长程相关特性

针对一种计算机网络模型，利用节点排队长度累计量的均方涨落函数，研究了网络节点在时间上的长程相关特性.结果表明，随着负载的增加，网络节点数据包排队长度在时间上由自由流状态的不相关或短程相关逐渐演变为临界和拥塞时的长程相关，关联范围逐渐增大，长程关联特性开始显现.在自由流状态时，节点的不相关或短程相关，并且有一致的数值为0.5的幂指数这一典型特征.而在临界状态时，节点数据包排队长度长程相关，有大于0.5的幂指数为特征.并且随网络规模的增大，节点间的群体作用逐渐显著，幂指数呈下降趋势.关键词：计算机网络相变长程相关幂律

Long-range correlation in computer network

The long-range correlation of nodes in a computer network model is studied with the mean square fluctuation function of cumulative variable of queue lengths. It is shown that the queue lengths of the data packets of nodes change their temporal independence on or short-range correlation in the free flow state to long-range correlation in the critical and congested state with increasing system loading. The range of correlation enlarges and the collective long-range correlation emerges. In a free flow, the nodes are independent of each other or short-range correlative, and there exists a typical characteristic power exponent of 0.5. At the critical state, the nodes are long-range correlative, and there exists a typical characteristic power exponent bigger than 0.5. Moreover, the collective interaction becomes obvious and the power exponent decreases with enlarging network scale.