Towards energy-aware coding practices for Android

Abstract

This paper studies how the use of different coding practices when developing Android applications influence energy consumption. We consider two common Java/Android programming practices, namely string operations and (non) cached image loading, and we show the energy profile of different coding practices for doing them. With string operations, we compare the performance of the usage of the standard String class to the usage of the StringBuilder class, while with our second practice we evaluate the benefits of image caching with asynchronous loading. We externally measure energy consumption of the example applications using the Trepn profiler application by Qualcomm. Our preliminary results show that selected coding practices do significantly affect energy consumption, in the particular cases of our practice selection, this difference varies between 20% and 50%.