CMOS芯片
传感器阵列
炸薯条
电子鼻
超大规模集成
计算机科学
电子工程
计算机硬件
电气工程
工程类
人工智能
机器学习
作者
Kwan Ting Ng,Farid Boussaïd,Amine Bermak
出处
期刊:IEEE Transactions on Circuits and Systems I-regular Papers
[Institute of Electrical and Electronics Engineers]
日期:2011-05-13
卷期号:58 (7): 1569-1580
被引量:94
标识
DOI:10.1109/tcsi.2011.2143090
摘要
This paper presents a CMOS single-chip gas recognition circuit, which encodes sensor array outputs into a unique sequence of spikes with the firing delay mapping the strength of the stimulation across the array. The proposed gas recognition circuit examines the generated spike pattern of relative excitations across the population of sensors and looks for a match within a library of 2-D spatio-temporal spike signatures. Each signature is drift insensitive, concentration invariant and is also a unique characteristic of the target gas. This VLSI friendly approach relies on a simple spatio-temporal code matching instead of existing computationally expensive pattern matching statistical techniques. In addition, it relies on a novel sensor calibration technique that does not require control or prior knowledge of the gas concentration. The proposed gas recognition circuit was implemented in a 0.35 μm CMOS process and characterized using an in-house fabricated 4 × 4 tin oxide gas sensor array. Experimental results show a correct detection rate of 94.9% when the gas sensor array is exposed to propane, ethanol and carbon monoxide.
科研通智能强力驱动
Strongly Powered by AbleSci AI