Abstract: Computing system performance degradation is mainly due to the loss of system resources, in order to avoid the high cost due to software aging, it is needed to study the resources consumption rule in the process of aging. Through building simulation environment, the performance degradation process is simulated and the resources data of computing system are collected. The noise data are removed using wavelet technique, the linear regression model and state-space model are established to analyze the resources consumption rule. Experimental results show significant availability for detection of usage of system resources, which can estimate the performance of computing system and offer data support for establishing software rejuvenation policy.

Key words: software aging, wavelet de-noise, state-space model, performance estimation

摘要： 计算系统性能衰退的主要原因是系统资源的损耗，为避免软件衰退导致的高成本，需要对衰退过程中的资源耗费规律进行研究。通过搭建仿真环境，模拟系统的性能衰退过程，采集衰退过程中的系统资源数据，利用小波去除噪声数据，建立资源耗费规律的一元线性回归模型和状态空间模型。实验结果表明，该方法能够有效地检测出系统资源的使用情况，对计算系统的性能做出评价，从而为软件抗衰策略的制定提供数据支持。

关键词: 软件衰退, 小波去噪, 状态空间模型, 性能评价

CLC Number: 

• TP311