Modeling and Optimizing Smartphone Energy Consumption
Despite the tremendous market penetration of smartphones, their utility has been and will remain severely limited by their battery life. To extend battery life on smartphones, it is critical to understand how the battery is drained by various components and apps, and to develop new energy saving techniques to save limited battery resources for users. This thesis develops novel techniques to help app developers and tool developers to more effectively model, profile and optimize the smartphone energy drain. The first part of the thesis focuses on measuring and modeling energy drain of smartphone wireless interfaces under various signal strength. We first conduct a large scale measurement study on wireless signal strength experienced by 3785 smartphone users. We then quantify the extra energy consumption on data transfer induced by poor wireless signal strength and develop a new signal-strength-aware power model for WiFi and 3G that significantly improves the modeling accuracy over previous state of the art. The second part of the thesis focuses on profiling and optimizing apps graphics energy. Profiling or optimizing app graphics energy is challenging due to the highly asynchronous graphics rendering pipeline in which multiple system layers are involved. To solve the problem, we develop a source-code-level UI energy profiler, GfxDoctor, that exposes graphics performance energy trade-off and pinpoints energy inefficiencies in apps UI. Using GfxDoctor we detected graphics energy bugs in 8 out of 30 popular Android apps. Removing these bugs reduces the app energy drain by 46% to 90%.
Hu, Purdue University.
Off-Campus Purdue Users:
To access this dissertation, please log in to our