业务量自相似特性对ATM缓存溢出概率的影响

Impact of Self Similarity of Traffic on Overflow Probability of ATM Buffers

基于在广域网（ＷＡＮ）、局域网（ＬＡＮ）、综合业务数字网（ＩＳＤＮ）中传输的业务以及可变速率（ＶＢＲ）业务等多种通信业务中普遍存在着自相似特性（长相关），叙述了自相似流量的数学定义和一些基本性质，描述了异步传输模式（ＡＴＭ）缓存的排队模型和一个具有自相似特性的业务量模型，给出了在自相似业务量下的ＡＴＭ 缓存溢出概率的下界，并与传统模型作了比较．比较结果表明，传统模型对ＡＴＭ 缓存性能的估计过于乐观．因此，在分析ＡＴＭ 网络性能的时候必需充分考虑业务量的自相似特性．

The recent measurement on wide area network (WAN), local area network (LAN) and variable bit rate (VBR) traffic shows that there exist intricately self similar or long range dependence characteristics in the traffic. This paper discussed the definition and properties of self similar, and presented a model of ATM queue system and a traffic model with self similarity. At last, the lower bound to the overflow probability of a finite ATM buffer was obtained. It was compared with the traditional model.