Probabilistic Model-Based Analysis to Improve Software Energy Efficiency

Software energy consumption has recently become a concern in software development. However, developers still lack knowledge about how to produce, evaluate and evolve their software considering energy consumption, which might limit their execution in some platforms and prevent users from adopting them. Towards providing more support for energy consumption analysis, we provide a set of properties to analyse software consumption considering not only energy costs but also probabilistic information. We demonstrate how to use, combine and interpret the results of analyses of these properties. We discuss experiments involving the analysis of the proposed properties in different scenarios and how, based on the results of these analyses, recommendations of possible actions to adjust energy consumption can be proposed.