Gas identification with spike codes in wireless electronic nose: A potential application for smart green buildings

Recently, building related illness and sick building syndrome have appeared as growing concerns for building residents. Ambient assisted solutions can be opted for in monitoring air quality in indoor environments by rapidly identifying health endangering gases. Industrial solutions are not appropriate for such a purpose because these incur high cost and long analysis time. In this paper, we present a wireless electronic nose system, containing commercially available gas sensors, to identify toxic gases in the indoor environment. Rapid identification with a reduced computational power and memory requirement is the major challenge to adopting a wireless electronic nose as an ambient assisted solution. Recently, logarithmic time encoding model based spike latency coding schemes have been used for hardware friendly implementation. However, these involve regression operation and a large memory requirement. In this paper, we use transient features to form spike codes instead of the logarithmic time encoding model, and as a result, we not only eliminate the requirement of regression but also achieve rapid identification with reduced memory size. A confidence coefficient is defined to examine the correctness of our approach, and if its value is below a certain threshold then a new sample can be collected for the classification decision. As a case study, data of five gases, namely carbon dioxide, chlorine, nitrogen dioxide, propane, and sulphur dioxide, is acquired in the laboratory environment and used to evaluate the performance of our approach.