自相似流量下的ATM交换机缓存溢出概率分析

ANALYSIS OF BUFFER OVERFLOW PROBABILITY OF ATM SWITCHES ON SELF SIMILAR TRAFFIC

近期的研究表明 ,在 WAN,L AN及 VBR业务、ISDN中传输的业务等多种通信业务中普遍存在着自相似特性 (或称为长相关 ) .对自相似流量模型下的排队分析方法目前还很不成熟 .分析并得到了在二阶渐近自相似流量下的 ATM交换机的缓存溢出概率的上下界 .从分析可以看出 ,流量的自相似特性对缓存溢出概率有着非常显著的影响 .与利用传统流量模型得到的结果相比 ,得到的缓存溢出概率随着缓存的增加 。

Recent measurement on WAN, LAN, VBR, and ISDN traffic shows that there exist intricately self similar or long range dependence characteristics in the traffic. The queuing analysis with self similar traffic is not complete until now. In this paper, the buffer overflow probability of ATM switches with second order asymptotically self similar input traffic is analyzed. The self similar behavior of traffic has serious influence on the buffer overflow probability of the switches. In contrast to typical ATM traffic models, the buffer overflow probability decreases much more slowly with buffer size.