TinyRBF: On-Device Learning for Sensor Self-Calibration in Precision Agriculture Applications

The Thesis addresses the challenges of precision agriculture by focusing on enhancing sensor accuracy and developing computational solutions for edge devices. Sensor drift, caused by environmental and operational stresses, is a significant issue that impacts irrigation, fertilization, and climate monitoring, ultimately affecting crop yield and quality. The research adopts an edge-centric approach, minimizing reliance on cloud platforms to improve data privacy, reduce latency, and support real-time decision-making. This strategy involves designing lightweight, resource-efficient algorithms that operate on energy-constrained and low-cost edge devices, commonly used in agriculture. The work emphasizes on-device learning techniques, enabling continuous adaptation to changing conditions while meeting strict power consumption requirements. A key contribution is the development of the TinyRBF model, a dynamic Radial Basis Function network tailored for edge devices. TinyRBF addresses the limitations of existing on-device learning methods by balancing prediction accuracy with memory and computational efficiency. Experimental results demonstrate its low resource usage and robust performance, making it a scalable solution for real-world agricultural applications.