基于自相似特性系统建模的算法改进  

The modification of system modeling algorithm based on self-similar characteristics

作　　者：刘佳[1] 金德鹏[1] 曾烈光[1] 

机构地区：[1]清华大学电子工程系微波与数字通信技术国家重点实验室,北京100084

出　　处：《清华大学学报（自然科学版）》2007年第10期1650-1653,共4页Journal of Tsinghua University(Science and Technology)

基　　金：国家"八六三"高技术项目(2001AA121071)

摘　　要：针对现有基于自相似特性的系统建模算法复杂度高、可实现性差等问题,改进了线性自回归模型叠加生成自相似源的算法;利用业务的自相似特性,Taylor展开式,以及对Pollaczek-Khinchin第一变换式进行恒等变形等方法简化了系统建模模型,解决了服务时间母函数不解析,以及运算繁琐等问题。与原算法相比,改进算法大大降低了算法复杂度,增强了其可实现性。以以太网为例,对改进算法进行验证,结论与原算法一致,运算时间仅为原算法的1/14。An autoregressive process of order one （AR1） accumulative algorithm was developed to generate a self-similar source to reduce algorithm complexity in self-similar system modeling. The self-similarity characteristics, a Taylor expansion, and an identity transform of the first Pollaczek-Khinchin equation are used to simplify the non-analytic generating function for the service time and the overall algorithm. The algorithm significantly reduces the arithmetic complexity and improves realization compared with the original network resource algorithm. The modified algorithm has been validated with an Ethernet test that gave the same results in 1/14 of the calculation of time.

关 键 词：自相似 自回归 TAYLOR级数 Pollaczek—Khinchin 

分 类 号：TP393.1[自动化与计算机技术—计算机应用技术]