Intermittent Computing: Challenges and Opportunities
ECE 595 Department Seminar Series
February 28, 2025
11:00 AM - 12:00 PM
Intermittent Computing: Challenges and Opportunities
Presenter: Arman Roohi, UIC
Abstract: The Internet of Things (IoT) is shifting from cloud-centric to data-centric approaches, with a market projected to reach $4.5 trillion by 2035. While intelligent IoT systems could reduce carbon emissions by 3% by 2030, battery dependence creates sustainability challenges, with projections of 78 million battery units disposed daily by 2025. This environmental concern necessitates exploring alternative power solutions.
Energy harvesting systems offer a sustainable alternative by capturing environmental energy, enabling device operation independent from power grids. However, these systems require intermittent execution approaches to handle inevitable power interruptions. This intermittent nature introduces unique computational challenges, particularly for advanced applications. As Convolutional Neural Networks (CNNs) move from cloud to edge computing, energy-harvesting IoT devices face the critical challenge of maintaining inference during power fluctuations. Despite their potential, existing solutions like checkpointing and task-based programming models prove inadequate for hardware-accelerated CNN inference due to performance limitations and hardware compatibility issues. These shortcomings highlight the need for innovative approaches.
This seminar examines the challenges of intermittent computing and presents our research findings on enabling efficient, reliable inference on battery-less devices. We will discuss novel approaches that overcome existing limitations, providing a pathway toward sustainable and intelligent IoT systems that can operate effectively under intermittent power conditions.
Speaker bio: Arman Roohi (SM’22) is currently an associate professor in the Department of Electrical and Computer Engineering at University of Illinois Chicago, where he is the director of the Self-sustainable Intelligent Resilient Ubiquitous Systems (SIRIUS) Laboratory. Before joining UIC, he was an assistant professor at the University of Nebraska-Lincoln (2020-2024) and a postdoctoral research fellow at the University of Texas at Austin (2019-2020). He received his PhD in Computer Engineering at the University of Central Florida in 2019. His research interests span the areas of cross-layer co-design for implementing complex machine learning tasks and secure computation, including hardware security and the security of artificial intelligence, reconfigurable and adaptive computer architectures, and beyond CMOS computing. He has completed over 100 publications on these topics, including best paper recognition, book chapters, and STEM curricular development. He received the PhD. Forum at DAC 2018 Scholarship, Frank Hubbard Engineering Endowed Scholarship in 2018, best paper recognition in IEEE Transactions on Emerging Topics in Computing in 2019, and paper of the month at IEEE Transactions on Computers in 2017.
Date posted
Feb 26, 2025
Date updated
Mar 6, 2025