TinyAIoT: Energy and resource efficient artificial intelligence for modern IoT applications

About

TinyAIoT aims to adapt AI methods to the needs of modern Internet of Things (IoT) applications. Today, these are often based on microcontrollers such as the Arduino family, which can send and receive data via special network protocols – for example, the Long Range Wide Area Network (LoRaWAN) network protocol. To function over long time and distances, these devices need to be resource efficient. TinyAIoT is exploring AI-based methods to increase resource efficiency by minimizing battery consumption and the amount of data sent.

Project Partners

TinyAIoT is a joint project together with the Institute for Information systems (Prof. Fabian Gieseke) and the re:edu GmbH & Co KG. For our use cases in smart cities and agriculture 4.0, we work together with our associated partners, Stadtwerke Emsdetten GmbH, the Smart City Stabstelle of the city of Münster, Naturschutzzentrum Kreis Coesfeld e.V. and Hof Homann eG, and our subcontractors opensenselab gGmbH and Budelmann Elektronik GmbH.

Use Cases: Agriculter 4.0 and Smart City

IoT microcontrollers are usually equipped with various sensors, e.g. to measure temperature, fine dust pollution, soil moisture. Such sensor networks can already be found in a many areas such as smart cities and agriculture 4.0. Among other things, we use senseBox technology to develop a resource-efficient digital harvester assistance. Among other things, we use senseBox technology to develop a resource-efficient digital harvester aid and monitor plant growth to identify the ideal time for seeding, harrowing and harvesting.

Energy-Efficient Environmental Monitoring with TinyML

This project explores how sensor data and artificial intelligence can be combined in Internet of Things (IoT) devices to improve environmental monitoring. By using Tiny Machine Learning (TinyML), compressed neural networks can run directly on low-power microcontrollers, allowing sensor nodes to process data locally instead of transmitting large raw datasets.

Our experiments show that performing AI inference on-device and transmitting only the results can significantly reduce energy consumption—by up to five times when processing image data. These findings help guide the development of energy-efficient IoT systems for long-term environmental monitoring in remote areas.