Estimating Android applications' CPU energy usage via bytecode profiling

Abstract—Optimizing the energy efficiency of mobile ap-plications can greatly increase user satisfaction. However, developers lack easily applied tools for estimating the energy consumption of their applications. This paper proposes a new approach, eCalc, that is lightweight in terms of its developer requirements and provides code-level estimates of energy con-sumption. The approach achieves this using estimation tech-niques based on program analysis of the mobile application. In evaluation, eCalc is able to estimate energy consumption within 9.5 % of the ground truth for a set of mobile applications. Additionally, eCalc provides useful and meaningful feedback to the developer that helps to characterize energy consumption of the application. Keywords-Android apps, bytecode profiling, eCalc, energy estimation I.