oa Sick Building Syndrome And Detection Of Volatile Organic Compounds With An Electronic Nose

Background: Sick building syndrome (SBS) describes a situation in which building occupants experience mild to severe health problems for no perceptible reason. Indoor volatile organic compounds (VOCs), namely benzene (C6H6) and formaldehyde (CH2O) are considered as potential contributors to the SBS condition. Based on sufficient evidence of carcinogenicity from studies of human cancer and their exposure to benzene and formaldehyde, the International Agency for Research on Cancer (IARC) of the World Health Organization (WHO) listed them as human carcinogens. Benzene and formaldehyde exposure cause acute lymphocytic leukaemia and nasopharyngeal cancer. Spectro fluorimetry and gas chromatography are possible solutions to detect these VOCs but cannot be frequently used due to high cost and long processing time. Objective: We introduce a microcontroller (MCU) based electronic nose to identify benzene and formaldehyde. Our proposed electronic nose contains a 4 x 4 tin-oxide gas sensor array and a radio frequency module for data exchange with a remote monitoring system. Method: We characterize our electronic nose system in the laboratory as shown in fig. 1. Bio-inspired coding schemes are used to identify the signatures of the VOCs. These schemes map the response vector of the sensor array into a temporal sequence. Experimental data is distributed into two sets, namely training and a testing data set. From the available training data, we build two libraries, namely a spike rank and a spike distance library. The spike rank library contains the temporal sequences of the sensors corresponding to benzene and formaldehyde exposure. In the spike distance library, we store the minimum spike distance for these VOCs which are extracted from the difference of each spike time and minimum spike time in the training data sequences Results: We compare each spike sequence from the testing data set with the reference sequences in the spike rank library. Fig. 2 shows the training spike sequences of benzene and formaldehyde. 95.833% of samples from the testing data set are correctly matched with these reference spikes and remaining samples are correctly identified with the spike distance library Conclusion: We introduce a low cost and compact portable solution for the detection of the carcinogenic odors of benzene and formaldehyde. Hardware friendly identification algorithms are used to reduce system complexity.