In his visionary 1991 article "The Computer for the 21st Century", Mark Weiser described a world in which people interact with smart information appliances, and where one would routinely have over a hundred such devices per room. He also described some of the key technologies needed to realize this vision, including devices where "a small battery will provide several days of continuous use." Since then, many of Weiser's predictions about the advent of ubiquitous computing have come to pass or have been exceeded. The capabilities of modern computing systems have far exceeded the requirements for Weiser's vision in terms of capacity and performance. However, truly ubiquitous computing still eludes us because the combination of increasing energy requirements and (relatively) slow progress in battery technology limits the lifetime of battery-powered devices. The goal of this project is to create hardware and software solutions to enable long-lifetime mobile information appliances.
The emergence of non-volatile memory technologies (e.g. magnetic random access memory) and bi-stable displays (e.g. e-Ink) significantly alter the energy signature of mobile devices by enabling near-zero quiescent power for the main memory and display. This project alters the energy signature for the microprocessor by tailoring its performance and energy to the requirements of typical user interaction patterns in the information appliance context. These patterns include viewing/reading the display (no computation), interacting with the display (minimal computation), and requesting a task that requires significant computation such as rendering a new screen. The processor architecture operates in high-performance mode with an energy signature resembling a modern embedded processor, low performance mode with an energy signature resembling a microcontroller, and ultra low power idle mode with extremely fast switching times between the different modes so as to provide the illusion of a single high-performance processor. This project creates the software interfaces necessary to exploit the low power properties of the new microprocessor thus simplifying application development for low power information appliances. The final goal of the project is to demonstrate the new concepts in a prototype e-book reader that is also a hardware testbed for the evaluation of the energy requirements of mobile information appliances.
This project will create enabling technologies for the development of mobile computing devices whose battery life can be measured in months, rather than days. The results of this project will be disseminated via research publications, by making a prototype mobile platform available to the research community, and by training undergraduate and graduate students in the design and implementation of energy-efficient mobile appliances.
Given the billions of mobile devices currently in use worldwide, a significant reduction in the energy requirements for mobile platforms could lead to a reduction in the energy consumed by the world's information technology infrastructure. Mobile devices that do not need to be re-charged on a daily basis would help us truly realize Mark Weiser's vision of ubiquitous computing.
