Open Access

The proliferation of the Internet of Things (IoT) has enriched modern life, but their increasing ubiquity raises concerns about environmental impact. To address this, comprehensive Life Cycle Assessments (LCAs) of IoT products, which have historically been manual, costly, and time-consuming, are vital. Noting the recurring nature of core components in IoT devices, such as CPUs and sensors, we propose to use graphs and machine learning to simplify and scale LCA estimations for IoT products. This paper introduces a novel approach to representing IoT devices as graphs with specific component characteristics and interconnections. Applied to a preliminary dataset of smart home IoT devices, the methodology unveils insights into structural similarities using a composite kernel approach. This initial phase lays the groundwork for the machine learning component. The integration of machine learning planned as part of ongoing research, provides a pathway for efficient and timely ecological assessments, ensuring that the rapid growth of IoT aligns with sustainable practices.