A Memristor-Equivalent Analog Integrated Neural Network for Air Quality Classification

This work presents a pure analog IC implementation of an artificial neural network with application in air quality. The high-level architecture consists of a pure analog memristor-equivalent model, operational amplifiers, and passive components. Additionally, the model comprises an operational transconductance amplifier, a differential voltage current conveyor, and passive components. This implementation has several advantages as it can handle a large number of features, has low power consumption, and high processing speed without increased complexity. Furthermore, a comparison is presented both with the software-equivalent and with other classifiers. The architecture was designed and simulated in the TSMC 90nm CMOS process using the Cadence IC Suite.