Design and implementation of an electronic nose system for real-time detection of marijuana

Abstract

Rapid and sensitive identification of illicit drugs has been a challenge and requires new methods. This study proposes a newly developed electronic nose (e-nose) based on commercially available gas sensors to provide a nondestructive, rapid, low cost, and portable solution for in situ detection for marijuana samples. Samples of seized marijuana, pseudo-narcotic marijuana, and cigarettes were analyzed. Principal component analysis (PCA), soft independent modeling of class analogies (SIMCA), and successive projection algorithm-linear discriminant analysis (SPA-LDA) were used for exploratory analysis, classification, and the reduction of variables and classification, respectively. The proposed e-nose system achieved 100% sensitivity, specificity, and accuracy for classifying the samples by SPA-LDA. This approach reduced the number of variables for classification from 355 to 10. The system further provided real-time detection from an internet-of-things (IoT) architecture. The proposed device showed decentralized, rapid, and accurate measurements and therefore may be an alternative for sniffer dogs or the current in situ screening methods. This device can further be expanded to detect other classes of illicit drugs.