A High Performance, Low Cost, and Compact E-nose System for Pollutants Detection

Detection and quantification of contaminants is of great significance for environmental monitor to both military and air space traveling. Although many existing sniffing devices may be used for pollutants detection, the size, cost, weight, and portability are primary concern in the development of a field deployable sniffing devices. Herein we propose to build high performance, low cost, compact, and reliable e-nose devices for contaminants detection and quantification. The system has three key features. First, we have identified an e-nose chip family, called “MEMS conductometric gas sensor”, developed by Boston Microsystems. In addition to small size and low cost, the chip is very reliable, selectivity and sensitivity for specific contaminant components based on it’s unique Micro-Electro-Mechanical Systems (MEMS) technology on MOS micro¬hotplate arrays. Second, we propose to use Support Vector Machines (SVM) to improve the accuracy of contaminants classification and nonlinear unmixing for accurate contaminant components concentration estimation in a mixture. SVM has several attractive advantages and better classification performance than most other classifiers. Most existing unmixing methods are linear in nature. However, our proposed unmixing is a nonlinear approach, which addresses the nonlinear relationship between e-nose responses and different contaminant compounds. Finally, we will use a PDA to collect the e-nose sensor reading and analyze the data. We believe that our proposed system can find many applications.