A system for efficient environmental monitoring and detection of forest fires through LoRa IoT and Artificial Intelligence
Loading...
Files
Full Text E-thesis
Date
2025
Authors
Uremek, Ipek
Journal Title
Journal ISSN
Volume Title
Publisher
University College Cork
Published Version
Abstract
Forest fires are among the most pressing global challenges, causing devastating environmental, economic, and social consequences. With climate change intensifying the frequency and intensity of wildfires, the need for innovative, efficient, and scalable early warning systems has become more critical than ever. This thesis presents a versatile, cost-effective software and hardware platform for forest fire detection and prevention, leveraging advancements in low-power IoT protocols and Artificial Intelligence (AI) to address this global issue. The proposed system integrates a network of low-cost, LoRa-enabled sensors to monitor key environmental parameters such as temperature, humidity, and atmospheric pressure across vast forest ecosystems. LoRa (Long Range) technology provides an ideal communication solution due to its ability to transmit data over long distances with minimal energy consumption, ensuring the sustainability of the system in remote and resource-constrained areas. Central to this system is the collaborative interaction between a localized model and a global model. The localized model operates at the node level, performing anomaly detection under resource constraints. Meanwhile, the global model, represented by meteorological centers, enhances predictions by integrating high-quality environmental datasets. ARIMA is employed for time-series forecasting, analyzing environmental trends, and enabling better model interplay, while decision tree algorithms are utilized for fire risk assessment, providing critical insights into potential fire occurrences. The system’s protocol integrates these components efficiently. LoRa-enabled sensors collect real-time environmental data, including temperature, humidity, and air pressure, and transmit it via LoRaWAN. At the node level, localized models perform initial anomaly detection, generating alerts and probability scores. These, along with raw sensor data, are sent to a central system, where a collaborative framework refines the analysis for improved fire prediction. To enhance reliability, the system incorporates a feedback mechanism that compares sensor readings with meteorological data from the Irish Meteorological Service, mitigating issues caused by noisy or incomplete data. This hybrid approach, combining low-power IoT protocols with machine learning enhances wildfire detection accuracy while maintaining energy efficiency and reducing operational costs.
Description
Keywords
Machine learning , IoT (Internet of Things) , Forest fire detection , Edge computing , Wireless sensor network (WSN)
Citation
Uremek, I. 2025. A system for efficient environmental monitoring and detection of forest fires through LoRa IoT and Artificial Intelligence. MRes Thesis, University College Cork.