【摘要】 近年来许多研究表明,网络业务本质是多分形的,呈现出自相似性与突发性,传统的基于泊松分布的模型不再适用。本文从理论分析和实际网络业务数据仿真验证两方面严格论证了网络业务流量能用α稳定分布准确建模,并与其它流量模型对比指出其优越性。更多还原

【Abstract】 Recen empirical studies indicate that the network traffic is multi-fractal in nature. Due to the self-similarity and burst of the traffic, the traditional models based on the Poisson distribution are stale in the characterization of network traffic. In this paper the anthors analyze the essence of network traffic and carry out experiments on practical traffic data sequence. These analytical results prove that traffic throughput can be accurately modeled by alpha-stable distribution. Moreover, comparison with other distribution models shows that alpha-stable distribution has more advantages in capturing nature of network traffic.更多还原