IoT-Based Short-Term Water Level Monitoring and Prediction Using Random Forest with MQTT QoS Evaluation
International Journal of Electrical and Electronics Research · 30.06.2026 · 618
DOI: 10.37391/ijeer.140237 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Electrical and Electronics Research, online erschienen am 30.06.2026. Verfasst von Satryo Budi Utomo. Die Arbeit nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Flood-related water level variations are among the most frequent hydrometeorological challenges with high frequency and impact in Indonesia, so an early warning system capable of providing fast, accurate, and real-time information is needed. An Internet of Things (IoT)-based experimental framework is proposed for short-term water level monitoring and prediction. Ultrasonic sensors are used to continuously measure water surface elevation, while NodeMCU functions as the main controller to process data and send information to the MQTT broker using a publish-subscribe architecture. The Random Forest model is trained using sequential water level data (time-series) and evaluated using MSE, RMSE, MAE, and R² parameters. The results show that the Random Forest model is able to produce high prediction accuracy with an R² value of 0.995488, indicating the model's strong ability to follow the temporal pattern of water level changes. In addition, the performance of the MQTT protocol is analyzed through QoS parameters including throughput, delay, jitter, packet loss, and Round-Trip Time (RTT). The test results show a packet loss value of 0.6% on the publisher-broker path and 1% on the broker-subscriber path, which is categorized as very good for IoT communication. Delay and jitter values are also in a stable range, thus supporting a decision support system for early warning applications. The integration of machine learning methods and the MQTT protocol in this study offers an experimentally evaluated and lightweight implementation framework for an IoT-based short-term water level monitoring and early warning support system. These findings indicate that the combination of Random Forest and MQTT has the potential to be an effective approach to support local monitoring and early warning decision-making through accurate short-term predictions and reliable data transmission. The contribution of this study is experimental, integrative, and evaluative in nature, focusing on system-level performance assessment rather than the development of new algorithms or theoretical models.
Zitieren
APA
Utomo, S. B. (2026). IoT-Based Short-Term Water Level Monitoring and Prediction Using Random Forest with MQTT QoS Evaluation. International Journal of Electrical and Electronics Research, 618. https://doi.org/10.37391/ijeer.140237
Vancouver
Utomo SB. IoT-Based Short-Term Water Level Monitoring and Prediction Using Random Forest with MQTT QoS Evaluation. International Journal of Electrical and Electronics Research. 2026:618. doi:10.37391/ijeer.140237
BibTeX
@article{Utomo2026,
author = {Utomo, Satryo Budi},
title = {IoT-Based Short-Term Water Level Monitoring and Prediction Using Random Forest with MQTT QoS Evaluation},
journal = {International Journal of Electrical and Electronics Research},
year = {2026},
pages = {618},
doi = {10.37391/ijeer.140237}
}
EUVIMED