Frosts are an adverse climatic phenomenon that severely affects agriculture, causing irreversible damage to plants. This process leads to dehydration, deterioration of plant tissue, and, in extreme cases, total crop loss. The lack of effective prevention and response strategies exacerbates its consequences, resulting in significant economic losses, particularly among small-scale farmers. In Ecuador, the Andean region is especially vulnerable, as its agricultural production depends on stable climatic conditions to achieve adequate harvests. This study presents the development of an early warning system against frosts, based on an Internet of Things (IoT) architecture, which employs long-range Wireless Sensor Networks (WSNs) and supervised learning algorithms specifically decision trees to anticipate frost-prone conditions and issue timely alerts to farmers. This enables the adoption of preventive measures that minimize agricultural losses and strengthen the sector’s resilience to extreme weather events. The research was conducted using the Action Research methodology, which involved an analysis of the phenomenon and the requirements of the stakeholders, followed by the design of a solution that included the selection of appropriate hardware and technologies, as well as the validation of technical requirements. The system was designed and implemented in a real agricultural environment, and the results demonstrated that the proposed solution is reliable, accurate, and adaptable, enabling timely frost detection and real-time alert generation. Furthermore, its effectiveness in crop protection was evident, contributing to the development of a more efficient, sustainable, and technologically advanced agriculture.
Electromagnetic underwater communications are strongly limited by water conductivity and depth, particularly at ISM frequencies above 900 MHz. Although LoRa technology has been widely adopted in low-power wireless sensor networks, experimental studies of LoRa-based underwater-to-overwater (UW2OW) links at 915 MHz in real freshwater environments remain scarce. This work presents an experimental evaluation of a 915 MHz LoRa UW2OW communication link conducted in three freshwater scenarios with different conductivity conditions: a controlled swimming pool and two natural lakes. The experimental campaign analyzes the influence of water depth and horizontal distance on key performance metrics, including received signal strength indicator (RSSI), signal-to-noise ratio (SNR), packet loss rate, and end-to-end latency. Measurements were carried out at depths between 0.25 m and 0.8 m. Multiple spreading factors were evaluated in the pool scenario, while SF12 was selected for the natural-lake experiments. The results demonstrate that short-range UW2OW communication is feasible, achieving effective distances of up to 13 m at shallow depths. However, increased depth and higher water conductivity lead to significant performance degradation, with packet loss rates exceeding 80% at longer distances. The obtained results provide empirical insights and practical design criteria for short-range freshwater underwater wireless sensor networks based on LoRa technology.