Communications of the ACM
Is TinyML Sustainable?
Credit: Andrij Borys Associates, Shutterstock
The continued growth of carbon emissions and global waste presents a great concern for our environment, increasing calls for a more sustainable future. In response, the United Nations' (UN) 2030 Agenda for Sustainable Development established a shared framework aiming toward peace and prosperity for people and the planet. At its core are 17 Sustainable Development Goals (SDGs),48 a call to action for all countries to work toward a more environmentally, economically, and socially sustainable future.
Key Insights
Tiny machine learning (TinyML), which enables ML on microcontroller (MCU) devices, holds potential for addressing numerous UN Sustainable Development Goals, particularly those related to environmental sustainability (see Figure 1). While TinyML's operational benefits for sustainability are often highlighted, it is crucial to consider the entire life cycle of both applications and hardware to ensure a net carbon reduction. This article contributes by presenting case studies illustrating TinyML's sustainability benefits, examining the environmental impacts of TinyML at both MCU and system levels through a life cycle analysis (LCA), and identifying future research directions for sustainable TinyML.
No entries found