Skip to main content
eScholarship
Open Access Publications from the University of California

UCLA

UCLA Electronic Theses and Dissertations bannerUCLA

Automating Personalized Battery Management on Smartphones

Abstract

The widespread use of smartphones and proliferation of mobile applications are reshaping many other areas ranging from social networking to health care Today's smartphones are much more capable than before, but mobile application are still restricted by limited resources on smartphones. The key hypothesis of this dissertation is that resource management on smartphones can be improved by adapting to usage patterns of users. We extensively studied users in the wild to characterize smartphone usage. We discovered significant diversity in smartphone usage. Along all aspects that we studied, users differ by one or more orders of magnitude. This finding suggests that resource management policies and algorithms on smartphones can become more effective if they learn and adapt to user behavior.

We developed the prototype of a system that adaptively manages battery, one of the most strained resources on smarthpones, and evaluated its performance. PowerLeash is a system that gives users control over their smartphones' battery lifetime when running background applications. With PowerLeash a user who is running power consuming background applications on her smartphone can decide how long her battery should last. PowerLeash continuously monitors the phone's battery level, the user's interactions with the phone, and progress of background applications. It builds a personalized model to estimate battery consumption based on usage and background applications progress. Using the on-line model and other information, PowerLeash dynamically adjusts the power consumption of background applications to meet the user's desired battery lifetime. We have designed PowerLeash to be easy to deploy, easy to use, and easy to incorporate in background applications. PowerLeash can run on any Android smartphone as a user level application, and relies only on information that is available to user-level processes. We present the design of PowerLeash and a detailed performance evaluation based on user studies. We use the lessons from deploying PowerLeash on volunteers smartphones to inform future iterations.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View