Detection and Estimation of Software Aging Based on Wavelet and Statistical Analysis
Liu Feng-yu · Jisuanji gongcheng · 2006
This paper proposes a hybrid method based on wavelet analysis and known statistical techniques to analyse the usage of system resources,to detect the existence of software aging and estimate the time to exhaustion.It proposes a methodology based on Haar wavelet analysis to denoise.Then,statistical trend detection techniques are applied to analyze the preprocessed data,validate the existing of software aging and quantify the effect of aging in system resources.Periodically monitored and collected data on the attributes responsible for determining the “health” of the executing software are applied to complete the experiment.The results of the experiment indicate significant availability for on-line detection of software aging.