CPS: Medium: Batteryless Sensors Enabling Smart Green Infrastructure

点击翻译按钮获取中文摘要

Cities across the nation have invested significantly in making infrastructure smarter, more sustainable, and more resilient to extreme weather events. Green infrastructure (GI), a general term for planned installations of trees, plants, soils, wetlands, and other natural resource, is increasingly being installed in cities nationwide to provide resilience to flooding, sewer overflows, urban heat, air pollution, habitat loss, and coastal erosion, that are overwhelming traditional infrastructure and negatively impact urban life. Embedding smart devices in GI provides numerous benefits to a city, providing insight into the health and effectiveness of the system, and the operations and fitness of the city. However, deploying smart devices in GI is challenging because of the scale and need for long-term deployment, meaning battery powered or expensive plugged-in devices are not feasible. This project builds Smart Green Infrastructure; augmenting GI with battery-free smart devices, powered by energy harvested directly from soil, which gather data, infer, actuate, and collaborate with each other. By harvesting from freely available soil and removing batteries, these devices can last for decades. Through partnerships with organizations in Chicago, Illinois–The Nature Conservancy, the Chicago Botanic Garden, and the Lincoln Park Zoo–the project demonstrates applicability by tackling stormwater management, urban wildlife surveillance, green roofs and other real-world applications. . Beyond cities, the work in this project will enable new applications in agriculture and smart farming, water resources management, and any applications where long term, zero maintenance embedded intelligence is required. Building Smart GI presents cyber-physical systems challenges in enabling robust inference, communication, and coordination on ultra-constrained computing platforms, and despite frequent power failures and dynamic energy availability. These devices harvest energy from soil or plant organic matter using terrestrial Microbial Fuel Cells (MFCs), constructed from inert materials providing decades long lifetime. These devices use machine learning to understand their environment, enabling robust, long-term monitoring that requires no maintenance or replacement. Finally, these devices actuate to "heal" the surrounding environment, switching the MFC from energy generation, to producing a disinfectant instead with the microbial community. A high-powered edge device orchestrates the actions of the swarm of MFC powered nodes, making decisions based on network data and external factors like user tasks. Project tasks include; co-design of the Terrestrial MFC and corresponding energy model for efficient and dynamic energy harvesting and; development of a tiny, resilient computing platform that harvests soil energy and supports sensing, and actuation; creation of a framework for conducting on-device machine learning to recognize subtle environmental changes from lossy data; exploration of orchestration of a network of intermittently powered devices; and conducting a series of end-to-end deployments with Chicago institutions.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.