一种基于多重分形的软件衰退分析方法  

Multi-fractal Based Methodology for Software Aging

作　　者：徐建[1,2] 许满武[2] 严悍[1] 李千目[1,2] 

机构地区：[1]南京大学计算机软件新技术国家重点实验室,南京210093 [2]南京理工大学计算机科学与技术学院,南京210094

出　　处：《计算机科学》2010年第8期178-181,共4页Computer Science

基　　金：国家自然科学基金(90718021;60903027);江苏省自然科学基金企业博士创新项目(BK2009535);高等学校博士学科点专项科研基金(No.20093219120024)资助

摘　　要：为了提高软件衰退预测的精度,采用了多重分形分析方法,以系统资源参数时间序列为研究对象,提出了一种定性和定量相结合的分析方法,用以研究其波动规律。定性分析阶段,借鉴分形理论分析影响软件性能的系统资源参数,揭示参数的波动具有分形特性;且其多重分形谱特征能刻画系统运行过程中随时间变化的情况。定量预测阶段,提出了一种多维的Hlder指数计算方法,用于计算多个资源参数序列的Hlder指数,并采用自回归移动平均模型(ARMA)预测Hlder指数。最后进行了实证分析,结果表明,该方法具有较好的定性分析和定量预测能力。This paper discussed a multi-fractal based method to analyze the fluctuation of the parameters of system resource,and proposed a new methodology which combined qualitative analysis with quantitative analysis to predict resource consumption and the trend of software aging. Firstly,this study used fractal theory to discuss the fractal structure of the parameters of system resources that influenced the performance of software system. And the results show that the variations of the parameters are not a stochastic process,but have characteristic properties of fractal. In addition, the characteristics of the spectra can be used to analyze the changes of system parameters during the running time qualitatively. Secondly, this paper put forward a new methodology for calculating multidimensional exponent, which is applied to data of system resource usage. Thirdly, the Auto-Regressive Moving-Average （ARMA） model was adopted in order to carry out the analysis of the multidimensional exponent and build the corresponding forecast model. Finally, the experiment was taken to calculate the multidimensional exponent of parameter series related with several memory resources using the parameters data collected from a realistic software system. The results of the experiment indicate effective ability to predict software aging.

关 键 词：软件衰退 多重分形 预测 

分 类 号：TP302.7[自动化与计算机技术—计算机系统结构]