E-nose system based on ultra-low power single micro-LED gas sensor and deep learning

Electronic nose (e-nose) technology utilizing chemoresistive sensors has gained substantial attention across various domains such as smart factory and personal health monitoring. The primary challenge associated with chemoresistive sensors lies in their susceptibility to cross-reactivity towards different gas species. To address this issue, we propose an innovative approach based on a photoactivated gas sensor integrated with micro-LED. By applying a pseudo-random voltage input to the LED, transient sensor responses for various gases are induced. Subsequently, we employ a deep learning to analyze the complex transient signals obtained, facilitating prediction of gas species and concentrations. In particular, the proposed e-nose system achieves high accuracy of gas type determination and concentration prediction for various target gases with only 0.5 mW by using only single sensor. Since it consumes less than one hundredth of the power of the existing e-nose systems, it has the advantage of being able to operate for a long time in battery-powered situation, and this advantage is expected to be utilized in combination with internet-of-things (IoT) technologies.