CAREER: Design Optimization and Quantitative Evaluation of Wireless Embedded Sensing Systems

University Of California-Irvine, Irvine CA

Investigators

Abstract

Wireless embedded sensing systems (WESS) are enabling innovations in many disciplines while driving research and education activities. Unfortunately, even experts today cannot easily compare competing WESS designs due to the lack of commonly accepted metrics and benchmark suites, and difficulties with reproducing sensing conditions. The PI identifies fitness-based metrics for quantitative evaluation of WESSs. To maximize fitness to the diverse sensing requirements, including size, power, and timing, the PI constructs a modular, configurable, multi-microcontroller WESS architecture called pico-cluster that spans the space from ultra-compact, wirelessly-networked event detectors to high duty-cycle data acquisition units. True parallelism of a pico-cluster enhances timing fitness, while de-centralized, multi-supply, load-matching power management techniques maximize power fitness and improve data fitness. Embedded systems so optimized will achieve new levels of energy efficiency by balancing energy harvesting and consumption while delivering high-quality data in a timely manner. To enable WESS benchmarking, the PI extends the B# battery emulator to emulating environmental stimuli and non-ideal power sources with full reproducibility. To advance the field of WESS based on concrete, measurable results from actual applications, the PI assembles and distributes a comprehensive, real-world benchmark suite, emulation-based evaluation framework, and the WESS architecture openly to the research community at large. The PI evaluates WESS architectures by actual deployment with collaborators in civil and mechanical engineering, medicine, and performance art while advancing the state-of-the-art in these fields. The compelling application drivers make project courses rewarding for students and add excitement to outreach activities particularly for women and minorities.