多重分形网络流量输入下多队列排队模型

Multi-Queuing Model Based on Inputting Multi-Fractal Network Traffic

针对目前分形网络流量输入下排队系统研究大多没有给出解析结果,且只研究单队列排队的缺点。根据实际网络的排队性能,在无约束条件下,采用了多队列多服务台成批到达的排队方式,建立模型,推导出多重分形特性网络流量输入下单队列和多队列溢出概率的解析式。并通过真实网络流量数据描绘出溢出概率与队列缓存长度的关系,结果发现该多队列排队模型下的溢出概率的解析式能够较好地拟合真实网络流量的排队性能。

Now queuing systems based on inputting multi-fractal network traffic have no parse result and only aim at single-queuing. This paper constructs a multi-queue multi-server group-by-group queuing model on no restriction and real network queuing performance, and deduces the calculating overflow probability expression based on inputting single-fractal and multi-fractal network traffic. Then the connection between overflow probability and cache size is described by real network traffic. The result shows the flow probability expression in the model can fit real network traffic queuing performance preferably.