Programmable Neuron Array Based on a 2-Transistor Multiplier Using Organic Floating-Gate for Intelligent Sensors

• Published in: IEEE journal on emerging and selected topics in circuits and systems 2017-03, Vol.7 (1), p.81-91
• Main Authors: Mahfuzul Islam, A. K. M., Hamamatsu, Masamune, Yokota, Tomoyuki, Sunghoon Lee, Yukita, Wakako, Takamiya, Makoto, Someya, Takao, Sakurai, Takayasu
• Format: Article
• Language: English
• Subjects: analog filter; artificial neuron; CMOS; Computer architecture; Computer memory; Data processing; flexible sensor sheet; floating-gate memory; local signal processing; Monitoring; Neurons; OFETs; Organic thin film transistor; pressure sensor; Rubber; Semiconductor devices; Sensor arrays; Threshold voltage; Transistors; Virtual memory systems
• ISSN: 2156-3357; 2156-3365
• DOI: 10.1109/JETCAS.2016.2628411

Artificial neurons are introduced for local data processing in a large area sensor device, thereby reducing multiple sensor outputs to 1-bit digital output. To reduce the area overhead and layout complexity due to the artificial neuron array, a 2-transistor multiplier using organic floating-gate pFET is proposed. Each artificial neuron has multiple organic floating-gate pFETs embedded in it. A virtual scalable floating-gate based memory architecture is realized for selective programming of each floating-gate pFET. Threshold voltage tuning is utilized to realize both of the positive and the negative weight values for a neuron. A flexible interface pressure monitoring device with pressure-sensitive rubber sheet is developed for pressure ulcer prevention to show the feasibility of our concept. A 50 mm 2 organic sheet integrating 2×2 neuron array is fabricated with 2-V organic CMOS transistors. Selective floating-gate programming and neuron operation are successfully demonstrated.