Many Internet-of-Things (IoT) applications such as autonomous vehicles and augmented reality require offloading tasks to a more powerful computing infrastructure--cloud. These applications require low latency and fast response time, which are often difficult to attain due to the physical distance and bandwidth limitations between users and the cloud. To cope with these limitations, edge computing has been introduced as a new paradigm that optimizes cloud computing to provide distributed computing solutions at the edge of the network, where IoT users utilize the computing resources in their vicinity (sometimes called "cloudlets"). Driven by the surging demand for edge computing, the energy consumption of clouds, cloudlets, and devices becomes increasingly important both from an environmental and cost viewpoint. Most of the existing energy consumption improvements in clouds come from improved engineering rather than improved algorithms. Such a practice fails to incorporate significant optimization opportunities available for clouds/cloudlets/devices to reduce their energy consumption. This two-year project focuses on developing resource management systems to significantly improve energy efficiency in edge computing.
The project consists of the following research thrusts: 1) understanding energy consumption in edge computing by exploring recurrent energy consumption patterns and identifying potential improvements; 2) introducing energy-aware data and job decomposition algorithms by developing mathematical and scalable computational models that seamlessly integrate multiple system objectives; 3) developing energy-aware placement of jobs to the edge components based on matching theory, graph theory, and distributed online algorithm design; and 4) evaluating the system on the National Science Foundation-funded CloudLab platform.
This project will enable more efficient use of cloud computing at the edge, while reducing energy consumption, which leads to cost reduction in edge services. It also benefits IoT users by extending battery lifetime of their smart devices and by improving their application performance. This project may lead to societal benefits such as promoting livable communities and smart cities. This research will be integrated into the into classroom teaching via a cloud computing course. Specialized outreach activities of this project are aimed at increasing participation of students from groups underrepresented in science and engineering. The project will maintain a dedicated website for dissemination of results.
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.
