An innovative IoT system for wildfire detection

Europe has witnessed a significant increase in the number and ferocity of so-called ‘mega-fires’, a phenomenon linked with climate change. Edge/IoT devices, coupled with AI/ML, can play an important role in preventing and fighting wildfires. Information gathered from environmental sensors deployed in the forest not only offers better monitoring but also helps to predict, detect, and manage wildfires. By using a traditional cloud-centric model, near-real-time analytics on the behavior and spread of wildfires cannot be achieved effectively due to the large amount of information to be transmitted. Improving the data processing capabilities of edge applications that are closer to the response teams deployed on the ground can provide a powerful tool for real-time assessment of wildfires.

We have developed an innovative sensor which is capable of continuous monitoring of IR temperature with a 768-pixel image for flame detection. It is able to capture a 3MP RGB image under flame trigger and report data on CO2, PM1, PM2.5 and PM10 concentrations and the air quality index. Data are transmitted by NB-IoT LTE and CAT-M2 bands to a cloud server for alarm verification and fire time evolution. Edge AI algorithms are used to detect the onset of the flame. Field tests show the ability to detect a flame at a 90m distance with approximately 50 x 50 x 50 cm flame dimensions. The system has been designed to run with ultra-low power processors and electronics with a battery power supply lasting 6 months. A low-cost design for industrial production was also considered for the potential of large-scale deployments.