Research Article Artificial olfactory system with fault-tolerant sensor array Nasser Lotfivand a,n , Vida Abdolzadeh b , Mohd Nizar Hamidon c a Department of Electronic Engineering, Tabriz Branch, Islamic Azad University, Tabriz, Iran b Young Researchers and Elite club, Tabriz Branch, Islamic Azad University,Tabriz, Iran c Institute of Advanced Technology, University Putra Malaysia,43400 Serdang, Selangor, Malaysia article info Article history: Received 24 March 2015 Received in revised form 21 January 2016 Accepted 13 March 2016 This paper was recommended for publica- tion by Dr. Ahmad B. Rad. Keywords: Gas sensor array architecture Artificial olfactory system Electronic nose Fault-tolerant sensor array abstract Numerous applications of artificial olfaction resulting from research in many branches of sciences have caused considerable interest in the enhancement of these systems. In this paper, we offer an architecture which is suitable for critical applications, such as medical diagnosis, where reliability and precision are deemed important. The proposed architecture is able to tolerate failures in the sensors of the array. In this study, the discriminating ability of the proposed architecture in detecting complex odors, as well as the performance of the proposed architecture in encountering sensor failure, were investigated and compared with the generic architecture. The results demonstrated that by applying the proposed architecture in the artificial olfactory system, the performance of system in the healthy mode was identical to the classic structure. However, in the faulty situation, the proposed architecture implied high identification ability of odor samples, while the generic architecture showed very poor performance in the same situation. Based on the results, it was possible to achieve high odor identification through the developed artificial olfactory system using the proposed architecture. & 2016 ISA. Published by Elsevier Ltd. All rights reserved. 1. Introduction Various investigations have been reported regarding the enhancement of the Artificial Olfactory System (AOS). In 1994, Gardner and Bartlett offered the following definition for the electronic nose [1]: “An electronic nose is an instrument, which comprises an array of electronic chemical sensors with partial specificity and an appropriate pattern-recognition system, capable of recognizing simple or complex odors.” Based on the above definition, an AOS is comprised of two main partitions: an array of gas sensors, and a pattern recognition unit. The sensor array acts like olfactory receptors in biological olfactory system. To achieve high classification and recognition rates, the sensor array is generally composed of several sensors with dif- ferent selectivity patterns which are chosen based on the sensi- tivity to a particular gas intended for detection [2]. An essential concept in designing a sensor array is the different sensitivity profile of each sensor over the range of odors expected as the target application. Therefore, the sensor array for different odors generates unique response patterns similar to fingerprints. The outputs of the sensor array are utilized for signal processing and odor classification units. For gas identification, diverse methods and techniques could be applied at these phases. Sensor technol- ogy and signal processing categories are the subject of the majority of reports concerning the development of artificial olfaction [1,3,4]. Nowadays, artificial olfactory systems are used in broad fields such as food [5,6], beverage[7,8], environment [9,10], medicine [11], etc. Most researchers have focused on the gas sensors of the system, and the development of signal processing and pattern recognition techniques. At the same time, various attempts have investigated the gas sensors by applying the different types of sensors, such as conductive polymers [12–14], metal-oxide [15– 17], surface acoustic wave [18–20], quartz crystal microbalance [21,22], and developing the structure of the sensor [23–26]. In terms of improving signal processing, various efforts have been reported on signal processing, feature extraction and classification techniques [4,27,28]. On the other hand, for critical applications, such as medical diagnosis, reliability and precision are important. A study on the aging of gas sensors demonstrated that, due to gas exposure, after ten months, the fractional change in resistance had decreased due to the aging issue, which limits the operational life of the sensor [29]. Furthermore, devices for medical applications must have a high level of reliability. Faults in medical equipment have been very costly in terms of deaths and injuries [30]. A report from the US Food and Drug Administration (FDA) documented that about 44% of the quality-related problems were the result of errors that could have been prevented through effective design controls [31]. Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/isatrans ISA Transactions http://dx.doi.org/10.1016/j.isatra.2016.03.012 0019-0578/& 2016 ISA. Published by Elsevier Ltd. All rights reserved. n Corresponding author. E-mail addresses: Lotfivand@Iaut.ac.ir (N. Lotfivand), Vida.Abdolzadeh@yahoo.com (V. Abdolzadeh), mnh@upm.edu.my (M.N. Hamidon). Please cite this article as: Lotfivand N, et al. Artificial olfactory system with fault-tolerant sensor array. ISA Transactions (2016), http: //dx.doi.org/10.1016/j.isatra.2016.03.012i ISA Transactions ∎ (∎∎∎∎) ∎∎∎–∎∎∎