IPNOSE: A PORTABLE ELECTRONIC NOSE BASED ON EMBEDDED TECHNOLOGY FOR INTENSIVE COMPUTATION AND TIME DEPENDENT SIGNAL PROCESSING

Recent developments in computer technology have made possible lowcost integration of powerful computers in small volumes. Here we suggest the integration of a small form factor computer for an electronic nose system. This concept allows us to seamlessly implement arbitrary temperature modulation for tin-oxide sensors, remote connectivity, large data storage, and complex signal processing. Gas sensors used in electronic noses are based on broad selectivity profiles, mimicking the responses of olfactory receptors in the biological olfactory system [1]. The basic building blocks of a generic electronic-nose systems include sample delivery, sensor chamber, signal transduction and acquisition, data preprocessing, feature extraction and feature classification. In conventional systems, the processing module is a personal computer separated from the remaining parts of the system. This module is responsible for data preprocessing, feature extraction and classification. Significant efforts are required to improve the overall performance of the instrument, and every component mush be given careful consideration. An alternative and effective approach to large sensor arrays is to modulate the operating temperature of a small number of tin oxide sensors. Excitation of the sensor at multiple temperatures during exposure to the analytes is asymptotically equivalent to having a large number of virtual sensors with different selectivities [2-4]. However, this solution is not convenient for portable systems, as it requires large memory allocation for each temperature-modulated waveform. In addition, feature extraction becomes more involved than in isothermal DC measurements, increasing the computational load of the computer system. In terms of signal-processing requirements, all the available options summarized in Table 1 have to be analyzed when designing a temperature-modulated electronic nose system. Considering recent trends in portable computing, the use of embedded systems at reasonably low cost and size is the most effective solution. This type of systems can be applied not only for portable devices but also for at-line analyzers, significantly improving the capabilities of the instrument. In this article we present a portable