Smartphone Energy Drain in the Wild: Measurement, Analysis and Optimizations
Abstract
The limited battery life of modern smartphones remains a leading factor adversely affecting the user experience. To extend battery life, it is critical to understand where and how battery drains under normal usage. This thesis presents the first large-scale measurement study of smartphone energy drain in the wild, and makes contributions on measurement study methodology, analysis, and optimizations of smartphone energy drain under real world usage. Firstly, we conduct the first extensive measurement and modeling of energy drain of 1520 smartphones. We develop a hybrid power model to enable such a measurement study. Using the collected traces, we present detailed analysis of how CPU time and energy are spent. Motivated by a major finding of the measurement study, that background activities drains 24% of smartphone battery, we study how to optimize the background energy. We develop a metric to measure the usefulness of background activities and present a system called HUSH that monitors the metric and automatically identifies and suppresses useless background activities. Motivated by another major finding of the measurement study, that 12.5% of daily energy was incurred by cellular maintenance, this thesis presents, to our knowledge, the first measurement study of modem behavior on smartphones in the wild. We first present a fine-grain event-based modem power model. We further analyze modem energy on 12 real users utilizing the developed power model. Our study gives the first insight into modem behavior and energy drain in the wild under operational context and exposes opportunities for optimization of modem chipset implementation.
Degree
Ph.D.
Advisors
Hu, Purdue University.
Subject Area
Computer Engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.