信息网络的时域业务模型

信息媒体基本上表现为三大类,即话音,数据与图象。对于提供综合服务的信息网络,从时域中观察这三类业务,可抽象为三类数学模型:周期性业务,随机性业务及突发性业务。而且周期性业务与随机性业务在大量统计中呈现为相同特性,称为匀流业务,可用二项分布进行数学描述。然而突发性业务与前两类业务呈现出明显不同的特性,如果简单地用二项分布描述,与实际特性会出现很大的偏差。采用两态一维马尔可夫过程进行数学描述比较符合实际。文中引用美国LANQuest LABS的测试结果总结出匀流业务与突发业务对网络的主要性能--吞吐量的影响,并有这样的结论:不管何种业务随着业务源速率增高,网络吞吐量均会降低;当业务源速率比较低时匀流业务与突发业务对网络吞吐量的影响的差别不很显著,但是业务源速率较高时,突发业务对网络吞吐量的不利影响显著大于匀流业务的影响。得出的结论对网络规划与容量分配有参考价值。

Time-domain Service Models of Information Networks

Information media are basically presented as three categories which are voice, data and image. Traffic in integrat-ed service information networks can be classified into three mathematical models: periodic, random and bursty, ifviewed from time-domain. The former two services have the same characteristics in the large counts, and thus are referred to as uniform service described with binomial distribution. The bursty service has notably differentcharacteristics and should be described with two-state and one-dimension Markov process. The effect of uni-form and bursty services on the network performance that is chiefly throughput is summarized by quoting the testresults of LANQest LABS, USA. We have reached the conclusion that no matter what the service is, network throughput degrades when the source rate increases; especially, at higher rate the adverse effect of bursty serviceis greater than that of uniform service though they have no distinctive difference at lower source rate. The conclu-sion can be used as a reference on network design and capability allocation.