Determining Per-Mode Battery Usage within Non-trivial Mobile Device Apps

The impact on battery life has become a key design criteria for mobile device applications (apps). Poor user perceptions about an app\´s energy use can now quickly lead to negative social media reviews and the resulting adverse impact son the app\´s marketability. Traditionally, assessments of an app\´s energy use have been done by simply measuring the total current draw experienced by the device while the app is on. Such approaches have become insufficient within modern devices due to the prevalence of energy management features within their hardware and operating systems (OSes) and the increased complexity of modern apps. This leads to: (i) non-stationary time domain current signals, due to the devices\´ dynamics power management, and (ii) different operational "modes" within an app having distinct energy use profiles. Within this work, an approach based on matched filters is developed to allow per-mode energy profiles to be correctly identified and characterized under(i) and (ii). This approach is then applied to quantitatively assess the energy use profile for a dual-mode multi-platform commercial Android quality of experience (QoE) assessment app across a number of mobile devices and OS variants, where it is shown that such differences can lead to significant differences in the app-level energy profiles that are produced.