TinyML-based Event Detection: An Edge-Cloud Approach for Smart Agriculture over LoRa WSNs

In modern agriculture, the capability to promptly detect and respond to specific events is crucial. This study centres on the transformative potential of TinyML for enhancing event detection in Smart Agriculture, particularly when integrated with LoRa-based Wireless Sensor Networks (WSNs). In this work, we underscore the unique advantages of utilizing TinyML at the edge-bypassing the latency and overhead associated with cloud-centric models and ensuring immediate, on-site analytical insights. Employing LoRa WSNs as the backbone provides a seamless, low-power, and expansive data communication framework. Detailed experiments demonstrate TinyML's efficacy in accurately predicting agricultural events while reducing computational and energy consumption. In conclusion, the synergy between TinyML and LoRa WSNs offers a promising approach for fine-tuned, real-time event detection, aiming for sustainable and high-yield agricultural practices.