Analysis of Software Aging in Android

There are a number of researches on software aging in different systems, but few researches have been done in Android system. At the same time, due to the complexity of Android memory management mechanisms and lack of comprehensive documents, it is difficult to analyze corresponding memory management mechanisms to determine whether there is aging. Based on these, we focus on some aging-related phenomena by observing available memory of Android, and use experiments to progressively verify some hypotheses about aging in Android. Then, we carry out mathematical modeling of aging and predictive analysis with Markov chains. Through our experiments and modeling analysis, we observe the existence of Android aging, and find that aging is generally irreversible and that the warm rejuvenation (application restart) has little effect on the recovery or mitigation of aging problem. The current experiments are preliminary and limited to specific version of Android OS, smartphone, and applications, but the results and observations may provide a new perspective for further study on aging and rejuvenation in Android with more statistically robust experiments.