Personal VOCs Exposure with a Sensor Network Based on Low-Cost Gas Sensor, and Machine Learning Enabled Indoor...

Area of Science:

• Environmental Science
• Sensor Technology
• Machine Learning

Background:

• Indoor environments with poor air exchange are susceptible to harmful volatile compound (VOC) contamination.
• Monitoring chemical distribution indoors is crucial for risk reduction.

Purpose of the Study:

• To develop and evaluate a novel monitoring system for indoor chemical distribution.
• To achieve accurate localization of volatile organic compound sources using a Wireless Sensor Network (WSN).

Main Methods:

• A machine learning approach was employed to process data from low-cost wearable VOC sensors within a WSN.
• Localization of mobile sensor units was achieved by analyzing Received Signal Strength Indicators (RSSIs) using machine learning algorithms.
• A commercial metal oxide semiconductor gas sensor was utilized for mapping ethanol distribution.

Main Results:

• The WSN achieved a localization accuracy greater than 99% in a 120 m² indoor area.
• The sensor signal strongly correlated with actual ethanol concentrations measured by a PhotoIonization Detector (PID).
• Simultaneous detection and localization of a VOC source were successfully demonstrated.

Conclusions:

• The developed WSN system effectively monitors and localizes indoor volatile organic compound sources.
• Machine learning-based RSSI analysis provides a highly accurate method for indoor localization.
• This technology offers a promising solution for indoor air quality management and chemical risk assessment.