AI-Enhanced Low-Power Gas Sensor System for Addressing Measurement Challenges of Dynamic Range in Portable Applications

This work presents an AI-enhanced gas sensor system optimized for portable applications, addressing the challenges of low-power consumption and wide dynamic range (WDR). By integrating a 1-D convolutional neural network (1D-CNN) for gas concentration prediction and a gas exposure detection algorithm, the proposed system enhances prediction accuracy, robustness, and energy efficiency. The system detects gas exposure in real-time, reducing unnecessary computations and improving overall efficiency. Experimental results demonstrate a quantification average error of only 2.56% and a power consumption reduction of 73.66% compared to conventional systems. These improvements make the proposed sensor system suitable for battery-operated portable devices, offering a practical solution for accurate real-time gas monitoring under varying environmental conditions. The proposed approach thus meets the needs of portable sensing technologies, combining high performance with low-power requirements for effective and reliable gas detection.