Battery power is a scarce resource on phones and is one of the main factors that limit the deployment of rich new kinds of mobile applications. There is presently no comprehensive solution to this power management problem, and existing solutions are narrowly focused on specific applications or specific device components. In this research, we will develop a system called PowerMedic that makes device-wide energy decisions across applications. The approach is to implement power management as a fundamental primitive. Much like other network primitives such as send and receive, applications need only specify their requirements; they no longer need to perform low-level energy optimizations. PowerMedic takes into account the interactions between device components such as network, CPU, screen, and sensors, to optimize power consumption, while satisfying the application requirements. The runtime decisions to save power are dynamic and driven by the user's application patterns, mobility, network connectivity, and other environmental factors. For evaluation purposes, it is expected that a user study will be conducted to gather mobile usage data. The data can be used to build predictive models of user behavior, such as charging patterns, to further improve the system.
This research is relevant to a broad segment of the population, as it is projected that there will be 1 billion smartphones worldwide by 2013. PowerMedic will enable a wide range of critical applications on phones, ranging from healthcare applications to providing assistive technology for the disabled. Today, these applications have limited success on phones because of the power bottleneck. This research will also benefit the research community through the release of the PowerMedic software and the user study data.
