A Low-Cost Smart Laboratory Monitoring System Using IoT Sensors and MQTT Communication with Digital Twin-Oriented Visualization
Smart laboratory monitoring requires continuous observation of indoor environmental conditions and space usage while maintaining user privacy and using an affordable sensing infrastructure. Conventional indoor monitoring systems may present environmental data through separate or limited visualization interfaces, while camera-based occupancy monitoring can introduce privacy concerns in laboratory and educational environments. To address these limitations, this study presents a low-cost and privacy-conscious smart laboratory monitoring system that combines IoT-based environmental sensing, MQTT communication, machine learning analysis, occupancy monitoring, and an Omniverse-based visualization layer. The proposed system uses an ESP32 development board as the edge-processing unit and integrates an SHT31 temperature and humidity sensor, an LDR light sensor, a magnetic door sensor, and dual VL53L0X Time-of-Flight sensors for entry/exit observation. The collected data were transmitted through a HiveMQ MQTT Broker and processed for environmental forecasting, light-state classification, anomaly detection, occupancy analysis, and digital twin-oriented visualization. During the monitoring period, 45,091 timestamped records were collected between 2026-05-21 and 2026-06-03, and 42,110 model-ready samples were used after feature generation and preprocessing. Multi-horizon forecasting produced high short-term R² values for temperature and relative humidity, although a persistence-based naive baseline remained a strong reference. Light-state classification was evaluated using Logistic Regression, Decision Tree, and K-NN models, with Logistic Regression achieving the highest F1-score of 0.849. Using the predefined upper 0.5% anomaly-score threshold, 211 records were flagged as unusual. Occupancy-related analysis showed that laboratory activity was limited and mainly concentrated on weekday daytime hours; the Random Forest occupancy-profile model achieved a lower MAE than the naive global-mean baseline, although instantaneous prediction performance remained limited. The Omniverse environment was implemented as an interface-level monitoring and visualization layer, while the smart switch pathway was retained as a prototype-level extension for future physical validation. Overall, the proposed system demonstrates a modular, affordable, and camera-free framework for smart laboratory monitoring and provides a scalable foundation for future digital twin-oriented and energy-aware control applications.
Authors
- Baykal Sarioglu (ORCID: https://orcid.org/0000-0002-7433-3823)
- Yavuz Selim Kızılgül (ORCID: https://orcid.org/0009-0009-1116-0484)
- Onur Mert Çakıcı (ORCID: https://orcid.org/0009-0004-3172-1924)
Institutions
- Istanbul Bilgi University (TR)
Publication Details
- Journal
- Black Sea Journal of Engineering and Science
- Published
- 2026-11-15
- DOI
- https://doi.org/10.34248/bsengineering.1969160
- Primary Topic
- Arduino and IoT Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00