CNS Core: Small: Expedient Computing: A Framework for Energy-Efficient and Timely Computing at the Edge

点击翻译按钮获取中文摘要

Emerging and future cyber-physical system applications demand timely processing at the network edge of large volumes of fresh data from sensor nodes, which may have limited processing ability and energy budgets. The sheer variety of envisioned applications and their requirements call for new performance metrics and strategies for balancing freshness of information, timeliness and accuracy of results, and the energy consumed in producing those results. This project proposes and develops a new holistic framework called Expedient Computing, in which results are produced as quickly as needed, with as much precision/accuracy as needed, using information that is as fresh as needed. The specific research objectives of the project include: (i) characterizing fundamental tradeoffs in expedient computing, (ii) exploration of emerging hardware designs in edge servers for expedient computing, (iii) developing an integrated optimization framework for expedient computing, and (iv) demonstrating expedient computing principles. This cross-disciplinary project will draw upon expertise in hardware architecture, network optimization and performance analysis, algorithms, and fundamental limits.Due to the inexorable desire to make the world smarter, millions upon millions of sensors are being deployed worldwide. The successful realization of envisioned applications such as connected autonomous vehicles and remote robotic surgery will critically depend on the timely extraction of actionable intelligence from sensor data. This project lays the groundwork for doing so while balancing the limitations in energy and processing capabilities and exploring the tradeoffs between them. The outcomes of this research project can offer significant performance improvements by maximizing precision, timeliness, energy efficiency and their tradeoffs in broad application domains such as edge computing, mobile sensing, and Internet of Things, and catalyze the industry toward seeking energy-efficient and timely computing technologies at the edge. Project results will be disseminated via research publications, organizing workshops and tutorials in top conferences, and through training students in this crucial cross-cutting field.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.