基于复合Poisson过程的软件可靠性度量  

Software reliability measurements based on compound Poisson processes

作　　者：赵向辉[1] 刘林[1] 黄雅芳[2] 张磊[1] 姚原岗[1] 

机构地区：[1]中国信息安全测评中心,北京100085 [2]清华大学软件学院,北京100084

出　　处：《清华大学学报（自然科学版）》2013年第12期1743-1749,1760,共8页Journal of Tsinghua University(Science and Technology)

基　　金：国家自然科学基金资助项目(90818021)

摘　　要：该文面向软件可靠性度量领域,探索软件在不同使用条件下的可靠性问题,提出了基于复合Poisson过程的软件可靠性度量方法。该方法是一种以软件运行和维护中带来的成本和利润损失按设定的损失等级进行强可靠性度量的方法。引入损失量作为失效严重程度的度量标准,并运用复合Poisson过程改进度量模型,使其不仅能度量软件失效总数和失效间隔时间,同时也度量平均软件失效损失量。最后,通过数学推导和实验应用,验证了本方法的正确性和有效性。A software reliability measurement method was developed based on compound Poisson processes with a strength reliability measurement to measure the cost profit losses during software Operation and maintenance according to given grades. A loss measure （or loss quantity） is introduced as a standard measurement of the severity of invalidation, with the measurement model improved by compound Poisson processes to measure invalid numbers and intervals and the average losses. This approach is accurate and has been verified mathematically and in tests.

关 键 词：软件可信 软件可靠 损失量 复合POISSON过程 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]