IoT-based indoor environmental monitoring and CO2 prediction using FIWARE and Sigfox for smart laboratory environments
Takeshi Tanaka, Tomoo Sigehuzi, Masayuki Yamauchi, Shuji Oji, Masaya Watanabe, Satoru Tada, Koji Kakugawa, Katia Vutova
This paper presents an Internet of Things–based system for continuous monitoring of carbon dioxide (CO2), temperature, relative humidity, atmospheric pressure, and illuminance in a laboratory, together with a 15-min-ahead warning function for elevated CO2 conditions. Sigfox-enabled sensors transmit measurements to ThingSpeak, while FIWARE Orion, QuantumLeap, CrateDB, and Grafana provide context management, time-series storage, and visualization. MATLAB retrieves the measurements, generates lagged CO2 and recent-change features, and applies regression and ensemble classification models. Because high-CO2 observations were infrequent, cost-sensitive learning and decision-threshold adjustment were employed. On the evaluated dataset, the threshold classifier achieved an accuracy of 62.63%, a precision of 0.09, a recall of 0.33, and an area under the receiver operating characteristic curve of 0.61. These results demonstrate the feasibility of the proposed end-toend monitoring architecture, although the model exhibited only modest predictive discrimination. Consequently, larger and more diverse datasets are required before the model can be considered for automatic ventilation control.
Cite this article as:
Tanaka T., Sigehuzi T., Yamauchi ., Oji S., Watanabe M., Tada S., Kakugawa K., Vutova K., IoT-based indoor environmental monitoring and CO2 prediction using FIWARE and Sigfox for smart laboratory environments. Electrotechnica & Electronica (Е+Е), Vol. 61 (1-2), 2026, pp.28-31, ISSN: 0861-4717 (Print), 2603-5421 (Online)
