The last few decades have seen a tremendous amount of innovation in computer system design to the point where electronic devices have become very inexpensive. This has brought us on the verge of a new paradigm in computing where there will be hundreds of devices in a person’s environment, ranging from mobile phones to smart home devices to wearables to implantables, all interconnected. This paradigm, called the Internet of Things (IoT), brings new challenges in terms of power, cost, and security. For example, power and energy have become critical design constraints that not only affect the lifetime of an ultra-low-power (ULP) system, but also its size and weight. While many conventional techniques exist that are aimed at energy reduction or that improve energy efficiency, they do so at the cost of performance. As such, their impact is limited in circumstances where energy is very constrained or where significant degradation of performance or functionality is unacceptable. Focusing on the opposing demands to increase both energy efficiency and performance simultaneously in a world where Moore’s law scaling is decelerating, one of the underlying themes of this work has been to identify novel insights that enable new pathways to energy efficiency in computing systems while avoiding the conventional tradeoff that simply sacrifices performance and functionality for energy efficiency. To this end, this work proposes a method to analyze the behavior of an application on the gate-level netlist of a processor for all possible inputs using a novel symbolic hardware-software co-analysis methdology. Using this methodology several techniques have been proposed to optimize a given processor-application pair for power, area and security.
University of Minnesota Ph.D. dissertation. August 2019. Major: Electrical/Computer Engineering. Advisor: John Sartori. 1 computer file (PDF); xii, 101 pages.
Application-specific Design and Optimization for Ultra-Low-Power Embedded Systems.
Retrieved from the University of Minnesota Digital Conservancy,
Content distributed via the University of Minnesota's Digital Conservancy may be subject to additional license and use restrictions applied by the depositor.