UCLA Researchers Part of National Science Foundation Project to Decarbonize Infrastructure

A pair of UCLA researchers are part of a five-year, $12 million research program funded by the National Science Foundation to decarbonize the full life cycles of infrastructure by optimizing and automating carbon efficiency across multiple sectors.

Headquartered at the University of Massachusetts Amherst (UMass), the multi-institution effort aims to reduce the carbon emissions of complex computing and societal infrastructure systems from manufacturing to recycling. The NSF Expeditions in Computing for Computational Decarbonization of Societal Infrastructures at Mesoscales (CoDec) will set out to find new ways to reduce carbon emissions across multiple infrastructure domains including computing, transportation, buildings and the power grid.

Led by Mani Srivastava, a professor of electrical and computer engineering at the UCLA Samueli School of Engineering, the UCLA team will focus on novel computation, sensing and verification services for computational decarbonization, while exploring the impact of economics and human behavior. Srivastava, who also holds a faculty appointment in computer science, brings expertise in artificial intelligence, sensors and cyber-physical systems. He is joined by Deepak Rajagopal, an associate professor at the UCLA Institute of the Environment and Sustainability specializing in industrial ecology, life-cycle assessment and energy policy. 

The team will analyze computer systems’ overall carbon emissions — from raw material extraction and processing to product manufacturing, operation and maintenance, and ultimately, end-of-use. Supported with $1.46 million in funding from CoDec, the researchers will also analyze the role cross-geographical regulations can play in effectively reducing carbon emissions. 

One of the project’s core challenges is time, said Srivastava, who holds the Mukund Padmanabhan Term Chair and directs the UCLA Networked & Embedded Systems Laboratory.

“Decarbonizing complex systems means looking years, or even decades, ahead. Our end goal is to reduce emissions throughout this full life cycle of a broad range of interconnected infrastructures,” Srivastava said. “To do that, we’ll measure and track carbon emissions across various stages in these systems’ life cycle, then use AI and machine learning techniques to devise interventions that reduce carbon emissions without compromising usability, privacy, intellectual property or economics.”

The team will also bring industrial ecology and life-cycle assessment perspectives to understand the footprint of computing and societal infrastructure systems.

“Our goal is to use life-cycle assessment to identify the hotspots — key materials, processes and behavioral phenomena that contribute the most to the overall burden from computing and societal infrastructure systems,” Rajagopal said. “Then we can develop strategies from a computational, technological and behavioral perspective to minimize the environmental footprint.” 

The CoDec team also includes collaborators from Carnegie Mellon University (CMU), the Massachusetts Institute of Technology (MIT), the University of Chicago (UChicago) and the University of Wisconsin–Madison (UW-Madison). 

UMass and CMU will serve as the primary testbed sites, collecting and analyzing data from sensors, testing supply chains and manufacturing processes. Additional testing on transportation and building energy systems will be conducted at UCLA and UChicago. MIT and the UW-Madison will work on efforts related to machine learning and grid computing. The partnering universities will also focus on training and outreach efforts at the high school, undergraduate, graduate and professional levels.

CoDec is one of three Expeditions in Computing projects recently announced by the NSF’s Directorate for Computer and Information Science and Engineering. The projects are designed to maximize computational performance and reduce energy demand. They are all part of a $36 million investment by the NSF in advancing sustainability via innovations in computing.