Sensor Fingerprinting–Based Gas Identification Using Artificial Intelligence

Abstract

Accurate identification of hazardous gases remains a critical challenge in environmental monitoring due to the limitations of single-sensor and threshold-based systems. This work presents an intelligent gas identification approach based on sensor fingerprinting and embedded artificial intelligence. A multi-sensor array comprising MQ-series sensors is used to capture distinct response patterns generated by different gases. These patterns are preprocessed and analyzed using a lightweight TinyML model deployed on an ESP32 microcontroller for on-device classification. The system enables real-time detection, local visualization, and wireless transmission of gas data for remote monitoring. An integrated alert mechanism enhances safety by providing immediate warnings when abnormal conditions are detected. The proposed solution offers a compact, low-cost, and scalable framework suitable for smart environments, industrial safety, and IoT-based monitoring applications.