Cloud-Oriented Meteorological Monitoring with Periodic, Event-Driven, and Hybrid Data Transmission

The growing demand for high-resolution environmental observations has increased the need for meteorological monitoring systems that provide frequent data updates while maintaining communication efficiency. This study presents a cloud-oriented hardware–software architecture that integrates environmental sensing, embedded processing, real-time cloud synchronization, long-term storage, and web-based visualization. The experimental evaluation was conducted using a stationary prototype based on a Raspberry Pi 4 and a BME680 sensor, which acquired 2432 measurements during a single controlled experiment. Periodic and event-driven transmission strategies were evaluated through offline emulation using receiver-side zero-order-hold reconstruction. The analysis also included parameter-sensitivity, equal-budget, and hybrid maximum-silence configurations. For the representative configurations, periodic transmission at 10 s intervals generated 122 records, with an MAE of 0.0143 °C and an RMSE of 0.0232 °C. Event-driven transmission with a threshold of 0.10 °C generated 43 records, reducing communication activity by 64.75%, with an MAE of 0.0397 °C and an RMSE of 0.0474 °C. Under an equal communication budget of 43 transmissions, the event-driven strategy produced lower RMSE, maximum, and P95 reconstruction errors than the corresponding periodic configuration. A hybrid configuration combining a 0.10 °C threshold with a maximum silent interval of 30 s generated 61 transmissions and provided a periodic heartbeat while achieving an MAE of 0.0305 °C. The results show that transmission parameters should be selected according to the required balance between communication activity and receiver-side reconstruction quality rather than on the assumption that one strategy is universally superior.